Datasets:
ccdv
/
arxiv-summarization

Dataset card Data Studio Files Community
7
article
stringlengths
0
745k
abstract
stringlengths
0
92.7k
the studies on the dynamical properties of classical atomic models , in which only coulomb forces are taken into account , have a long history , going back to nicholson , bohr and langmuir ( see refs . ) . the main result found in such old works ( see the work of nicholson for the case of beryllium , at those times called nebulium ) is that there exist periodic orbits having `` normal mode frequencies '' whose ratios are near to those of the observed spectral lines . among the recent works on this subject a particularly relevant place is taken by the papers of de luca @xcite on the helium atom . this author went beyond the coulomb approximation , taking retardation of the electromagnetic forces into account . this was obtained through a perturbation scheme which , truncated in a suitable way , leads to the conservative darwin lagrangian that one finds illustrated in common textbooks@xcite . in such a way the scale invariance of the purely coulomb model was eliminated , and the result of nicholson could be improved , showing that for the helium atom there exist periodic orbits leading to dynamical frequencies which themselves , and not only their ratios , have a rather good agreement with the empirical spectral frequencies . on the other hand , it was known since the time of nicholson , and confirmed for example by poirier @xcite , that all such periodic orbits are unstable . more recently yamamoto and kaneko@xcite showed , for the case of helium , that the instability of the periodic orbits actually corresponds to a much more general and acute form of instability . indeed they found that the vast majority of initial data with negative energy lead to the _ autoionization _ of the atom , namely , to motions in which one of the electrons is expelled . one can see that this occurs also for a generic atom , and even in a stronger way , inasmuch as one meets with motions in which all electrons but one are expelled . so , the whole classical theory of atomic models seems to be plagued by a general failure , for which no remedy is known . one usually takes the pragmatic attitude of just forgetting the autoionizing motions . one should however indicate an internal dynamical justification of a more general character for such a selection of the initial data . moreover , one might hope that such a general prescription would also automatically explain why the system chooses to fall ( through some peculiar kind of dissipation ) on the physically interesting periodic orbits , for which an agreement betweeen dynamical and spectral frequencies is found . obviously , satisfying the latter point requires to abandon a purely lagrangian , and thus conservative , approach to the problem , by taking electromagnetic radiation into account . this is precisely what we do in the present paper , where we take into account the emission of radiation due to the accelerations of the electrons in the simplest possible way , namely , by adding to the coulomb forces the radiation reaction ones . so , in a sense our model is complementary to that of de luca , who introduces only the conservative terms produced by the darwin approximation scheme , and neglects dissipation . we leave for future work the study of a model which takes both features into account . it will be shown here that the modification of the coulomb model that takes radiation into account , in the first place gives a model in which the autoionization problem no more arises . this is due to the fact , first pointed out by dirac in the general context of classical electrodynamics of point particles@xcite , that most initial data in the phase space suited to the model lead to motions in which an infinite amount of energy is emitted , so that the definition of the model has to be complemented by the explicit prescription that such initial data have to be discarded . the remaining initial data ( constituting a set that will be called here the _ dirac _ or _ physical manifold _ ) will be shown to lead to motions in which the phenomenon of autoionization no more occurs . so , the elimination of the initial points in phase space leading to autoionizing solutions appears no more as a special trick to be strangely introduced _ ad hoc _ , but rather as a particular case of a completely general prescription that , following dirac , has always to be introduced when dealing with classical models of matter radiation interaction involving point particles . then we study the dynamics on the dirac physical manifold , in which the dirac precription turns out to introduce a dissipation af a peculiar type . we show that the dirac prscription allows one to find ( and in a easy way ) all periodic orbits , proving furthermore that they form in phase space a manifold . this is just the _ zero dipole manifold _ , which is constituted of phase space points leading to motions that do not radiate energy away in the dipole approximation . a major aim of this paper is to study such a manifold . we first show that it is an attractor , for initial states with negative energy . then we study , by numerical methods , the invariant measure naturally induced on it by the time flow . this amounts to studying the _ formation process _ of the helium atom , namely , motions having initial data with the two electrons coming from infinity that are then captured by the nucleus , and finally fall on the zero dipole manifold , where emission of energy comes to an end . from the point of view of the theory of dynamical systems , an interesting result we find is that , while the attractor ( the zero dipole manifold ) is really simple as a manifold ( being just a portion of a linear subspace in the system s phase space ) , it is the invariant measure induced on it by the time evolution that is very peculiar . indeed it is presumably singular with respect to the restriction of the lebesgue measure , possibly having a fractal structure . in section 2 the model is introduced , and a preliminary analytical discussion of its properties is given . in section 3 the numerical method for obtaining the invariant measure is described , and the numerical results are presented . the conclusions follow . let us recall that , in the case of a single point charge , the emission of radiation is taken into account without introducing the infinitely many degrees of freedom of the electromagnetic field , through the expedient of adding in the equation of motion for the particle an `` effective force '' . this force , which is traditionally called _ radiation reaction force _ and denoted by @xmath0 , is given by @xmath1 where @xmath2 is the charge of the particle , @xmath3 the speed of light , and @xmath4 the position vector of the particle . the procedure which leads to such a force goes back to planck , lorentz and abraham , and was given a final form in the work of dirac @xcite of the year 1938 , in which an extension to the relativistic case was performed in an extremely elegant way . in its most elementary form , which is illustrated in standard textbooks ( see refs . ) , the procedure amounts to requiring that the effective force @xmath0 produces an energy loss consistent with the power radiated away by an accelerating particle according to the larmor formula , namely , @xmath5 . as is well known , this force can also be interpreted as being produced by the regular part of the self field of the particle , the divergent part having been reabsorbed through mass renormalization . we illustrate now , in the simple case of the free particle in the nonrelativistic approximation , how the `` runaway '' solutions then show up . in terms of the particle s acceleration @xmath6 , the equation of motion for the free particle takes the form @xmath7 with @xmath8 ( @xmath9 being the particle s mass ) , with solutions @xmath10 depending parametrically on the initial acceleration @xmath11 . so , for generic initial data the solution exponentially explodes , i.e. , presents _ runaway _ character , a fact which makes no sense for a free particle . thus dirac @xcite introduces the explicit prescription that , for the case of a free particle , only the solutions with vanishing initial acceleration , @xmath12 , be retained . now , the phase space @xmath13 mathematically suited to the considered third order equation of motion , is the vector space @xmath14 , a point of which is defined by the coordinates @xmath15 of position , velocity and acceleration . in such an ambient phase space @xmath13 , the physically meaningful phase space is thus the subset characterized by motions of nonrunaway type . this we will call `` _ _ dirac _ or _ physical manifold _ _ '' , and coincides with the hyperplane @xmath16 . more in general , in the presence of an external force vanishing at infinity , one should assume that for scattering states , in which the particle eventually behaves as a free one , the physical or nonrunaway manifold be defined by the asymptotic condition @xmath17 prescriptions of such a type may be formulated in a different way , which captures another side of the problem . one makes reference to the total energy emitted by the particle , namely , the quantity @xmath18 and a physically natural requirement is then to restrict oneself to motions for which the amount of energy radiated away during the whole motion is finite , i.e. , one has @xmath19 . for smooth motions , such as those we are considering which are solutions of an ordinary differential equation , the latter condition implies the asymptotic condition ( [ condizione ] ) . thus condition ( [ condizione ] ) turns out to be justified also for motions not having a scattering character ( as the bound states ) , for which the argument referring to the eventual free type motion of the particle does not apply . the asymptotic condition ( [ condizione ] ) will be referred to as the _ dirac prescription_. due to the global character ( with respect to time ) of the latter , the dynamics on the dirac physical manifold thus acquires quite peculiar fetures @xcite ; for example , it can be folded @xcite . we will give arguments indicating that , for the helium atom model with radiation reaction , the analog of the dirac prescription automatically eliminates the autoionization problem , and thus overcomes the main inconsistency of all classical models which neglect radiation . in the case of several particles one can operate as for just one particle , by introducing a suitable radiation reaction force acting on each particle , in such a way that the power dissipated by the whole system be equal to that emitted in the dipole approximation ( see sec . 9.2 ) . in the case of two particles the radiated power is @xmath20 , where @xmath21 , @xmath22 are the position vectors of the particles ( here , the position vectors of the two electrons with respect to the nucleus , assumed to be fixed at the origin of an inertial frame ) . this gives for the effective forces @xmath23 and @xmath24 acting on each electron the same expression , namely , @xmath25 + @xmath26 . thus the model is described by the system of equations @xmath27 ( in the gauss system , with @xmath28 given by ( [ epsilon ] ) ) . one obtains the same expression for the radiative force acting on each particle , also if one considers it as due to the sum of the ( regular part of the ) self field , and of the retarded field produced by the other charge , expanded with respect to the distance . one can check this fact through computations which are completely standard , though a little cumbersome and not particularly illuminating . so we omit them . the only point worth of mention is that the expansion in the distance should be performed up to third order and not just to the second one as one finds in the textbooks ; this is indeed the point which is responsible for the appearing of the relevant nonconservative term proportional to the third derivative of the center of mass . system ( [ eq:1 ] ) , when expressed in terms of the center of mass @xmath29 and of the relative position vector @xmath30 , defined by @xmath31 takes the form @xmath32 in which there appears only one third derivative term , @xmath33 , which enters just the equation for the center of mass . in conclusion , when radiation reaction is taken into account , the equations of motion in the variables @xmath29 , @xmath30 are exactly the same as for the purely coulomb model , with the only addition of the radiation reaction force in the equation for the center of mass . so the ambient phase space @xmath34 suited for the motions of the two point electrons is the vector space @xmath35 , a point of which is defined by the coordinates @xmath36 now , the radiation reaction term acts in the equation for the center of mass just as in the case of a single particle . thus , in the spirit of the classical work of dirac , we will define our model by complementing the system of equations ( [ eq:1 ] ) or ( [ eq:2 ] ) through the dirac prescription : _ in the ambient phase space @xmath37 the only admitted points are those leading to motions that satisfy the asymptotic condition _ @xmath38 as in the case of the single particle , this corresponds to the physical condition that the amount of energy radiated away during the whole motion be finite . this prescription implicitly introduces a selection among the allowed initial data , and the subset of the allowed points in the ambient phase space @xmath39 will be called the _ dirac _ or _ physical manifold_. we will show that this restriction eliminates the autoionization problem , and makes the zero dipole manifold become attractive for initial data of negative energy . one immediately sees that there exists a particular invariant submanifold of the dirac physical manifold , which will play a fundamental role in this work . it is the _ zero dipole manifold _ , defined by the condition @xmath40 for all times , which corresponds to the hyperplane @xmath41 invariance is immediately checked . one also easily checks that such a manifold is composed of orbits which are solutions of the purely coulomb model in the unknown @xmath42 with a suitable charge , and are thus periodic in the case of negative energy . we will make use of an energy theorem . this is immediately established by multiplying , as usual , equations by @xmath43 and @xmath44 respectively , adding them and using the leibniz formula for the product @xmath45 . this gives @xmath46 where @xmath47 is the kinetic energy , @xmath48 the potential energy corresponding to the coulomb forces , and @xmath49 the so called schott term . so , the quantity @xmath50 may be simply called the _ energy _ of the system , ( while the quantity @xmath51 may be called the _ mechanical energy _ ) , and @xmath52 turns out to be a non increasing function of time . obviously the two energies coincide , @xmath53 , on the zero dipole manifold . from the energy theorem one immediately gets that periodic orbits necessarily belong to the zero dipole manifold . indeed , just integrating the energy equation over a period , a periodic orbit is seen to necessarily have vanishing center of mass acceleration , and consequently is seen to belong to the zero dipole manifold . in addition , the portion of the zero dipole manifold with negative energy is wholly foliated by periodic orbits which , as already mentioned , are just the periodic orbits for the two body coulomb problem with a suitable charge . using the energy theorem one can also make more precise the notion of an autoionizing motion . we recall that the instability problem that plagues the purely coulomb model is that the vast majority of initial states with negative energy lead to motions in which one of the electrons is expelled to infinity , so that the atom is unstable . we show instead that the analogous situation does not occur in the model with radiation reaction , for initial states with negative energy . indeed , if autoionization occurs with the two electrons both escaping to infinity , then both the potential and the schott term finally vanish , so that the energy finally becomes positive , i.e. it _ has increased _ , against the energy theorem . if instead autoionization occurs with only one electron , say the first one , escaping to infinity ( in a nonrunaway fashion ) , then the equation of motion for @xmath22 reduces to @xmath54 which is the equation for just one electron with radiation reaction , in the external field of the nucleus . on the other hand , as shown in papers and , starting from initial data with negative energy , equation ( [ singolo ] ) admits only runaway solutions . furthermore , the zero dipole manifold is an attractor for initial states with negative energy , as a consequence of the asymptotic dirac condition ( [ eq:3 ] ) . this can be seen by rewriting the equation for the center of mass in the integro differential form @xmath55 where @xmath56 is the function defined by the last two terms at the right hand side of the second equation in ( [ eq:2 ] ) . so , from the de lhopital rule the property @xmath57 for @xmath58 follows . finally , the result @xmath59 for @xmath58 follows from the form of the function @xmath56 , using the previously proven fact that for negative energies the motion of the center of mass is bounded . in conclusion , following the dirac prescriprion of restricting the ambient phase space to the dirac physical manifold , on the one hand the difficulty of the generic autoionization is eliminated , and on the other hand our dynamical system presents a _ dissipative _ character , because all initial states with negative energy are definitively attracted to the invariant zero dipole manifold , corresponding to motions that do not radiate energy away ( in the dipole approximation ) . however , in the next section we will show that the electrodynamical dissipation due to radiation reaction has a peculiar character with respect to that of the familiar dissipative systems such as the lorenz one . it is simple to check that the lebesgue measure in phase space , restricted to the zero dipole manifold , is invariant under the flow . however , other , actually infinitely many , invariant measures exist , this being a general property of any continuous group @xmath60 of diffeomorfisms on a smooth manifold @xmath61 . indeed , following krylov and bogolyubov @xcite , for any given measure @xmath62 with @xmath63 , a corresponding invariant measure @xmath64 can be constructed by defining @xmath65 the intuitive meaning of the measure @xmath64 is the fraction of the initial points of phase space ( chosen according the given probability @xmath62 ) which fall on any given set @xmath66 . so , if we choose for @xmath62 the restriction of the lebesgue measure to the dirac physical manifold , and take initial data with the two electrons coming freely from infinity , then from the physical point of view the measure @xmath67 will describe the state of the atom at the end of the formation process . our aim is now to find a numerical scheme for constructing the measure @xmath67 . for the sake of simplicity , in the implementation we will restrict ourselves to the case of planar motions . so , the phase space corresponding to equations has dimension 10 , and the dirac physical manifold has at most dimension 8 , while the zero dipole manifold has dimension 4 . it goes without saying that the motions are to be obtained by integrating the system backward in time . indeed the dirac physical manifold is defined only implicitly , through the asymptotic condition ( [ eq:3 ] ) . thus one concretely has to operate in the ambient phase space @xmath39 , in which the dirac manifold has vanishing measure . besides , even if one were able to take initial data on the dirac manifold , by integrating forward in time numerical errors would make the orbits escape from it , and in a runaway fashion . such problems are overcome by the standard procedure of integrating the equations of motion backward in time . in such a way one is guaranteed to approach ( in a time of the order @xmath68 ) the dirac manifold . by the way , the integration beackward in time is just the one needed in order to compute the asymptotic measure @xmath67 according to the definition . from the numerical point of view , the only practical way in which a measure can be described , is to compute its density @xmath69 . here we denote by @xmath70 the coordinates of a point in the ambient phase space @xmath34 . so we implemented the following algorithm . taking @xmath70 on the zero dipole manifold as initial datum , we follow , backward in time , the time evolution of a small hypercube of equal sides @xmath71 ( @xmath72 ) , having @xmath70 as one of its vertices , i.e. , we follow the evolution of the points @xmath73 , and define , as usual , the side of the evolved hypercube by @xmath74 . notice that the hypercube does not belong to the zero dipole manifold ( not even to the dirac one ) ; however , in our specific case , the integration of the equations of motion , backward in time , insures that the trajectories fall on the dirac manifold after a very short transient ( of the order @xmath68 ) , because the runaway component of the motion is exponentially damped as one proceeds backward in time . so , the hypercube becomes squeezed on the physical manifold , and one of its 8dimensional faces becomes tangent to the manifold . to get the density @xmath69 , it is then sufficient to compute the area of the latter face , divided by the initial area , and then calculate its time average along the trajectory of @xmath70 ( which obviously belongs to the zero dipole manifold for all times ) . we chose to numerically integrate the system with a fourth order runge kutta method and autoadaptation of the time grid . we take units with @xmath75 and @xmath76 , completing them by taking as unit of length the bohr radius @xmath77 , as usually done in atomic physics . in this way @xmath28 would take the value @xmath78 . however , as we are only interested in the qualitative behaviour of the system , in order to reduce the computational time we decided to take instead @xmath79 . the autoadaptation of the time grid was implemented by choosing the integration step @xmath80 according to @xmath81 where @xmath82 , while @xmath83 , @xmath84 and @xmath85 are the distances of the first and second electron from the nucleus and their mutual distance respectively , while @xmath86 and @xmath87 are factors introduced to avoid that the integration step becomes too small or too large . the sides @xmath71 of the initial hypercubes were all taken equal to @xmath88 . once the trajectories are known ( numerically ) , there remains the problem of how to identify the face of the hypercube lying on the dirac manifold ( which is utterly unknown ) . in order to solve this problem , one can start from the fact that all sides @xmath89 are essentially contained in the tangent plane to the manifold . then this plane can be found by determining the 8dimensional hyperplane to which the vectors @xmath89 are closer . as an hyperplane is defined by two independent unit vectors orthogonal to it , say @xmath90 and @xmath91 , then the tangent plane will be defined by the unit vectors which minimize @xmath92 $ ] , i. e. , the sum of the distances of the points @xmath89 from the plane . in other terms one has to minimize the function @xmath93 where the symmetric matrix @xmath66 is defined by @xmath94^t\cdot [ \delta { { \mathbf { x } } } _ 1,\ldots,\delta { { \mathbf { x } } } _ { 10 } ] \ .\ ] ] as is well known , the two independent unit vectors which minimize @xmath95 are the eigenvectors corresponding to the two smallest eigenvalues @xmath96 , @xmath97 of @xmath66 . with a little reflection , one also gets that the area of the tangent face is nothing but the square root @xmath98 of the product of the remaining 8 eigenvalues @xmath99 , @xmath100 of the matrix @xmath66 . in summary , the numerical procedure consists in taking for @xmath69 the time average of @xmath101 . notice that , as occurs in the computation of the lyapunov exponents , the sides @xmath102 can grow too much during the evolution . if , for some @xmath103 , @xmath71 becomes larger than @xmath104 , we renormalize the cube choosing a new one with sides directed as the eigenvectors of @xmath66 and again of size @xmath88 , and the average is then accordingly modified . just to give an indication of the form of the natural measure , in the zero dipole manifold we chose @xmath105 initial points @xmath70 uniformly distributed in the cube @xmath106\times[0.5,4.0 ] \\ \dot { { { \mathbf { r } } } } & \in & [ -1.5,1.5]\times[-1.5,1.5 ] \ , \end{array } \right.\ ] ] with the condition that they have negative energy . we checked that averaging over a period is enough to insure that the time average is stabilized . invariant density @xmath107 on the invariant zero dipole manifold , projected on a rectangle of the ( @xmath108 ) plane , of energy and angular momentum.,scaledwidth=50.0% ] same as figure [ fig:1 ] , with the values of @xmath107 represented in a gray decimal logarithmic scale , covering a thousand orders of magnitude.,scaledwidth=50.0% ] same as figure [ fig:1 ] , restricted to a smaller portion of the original rectangle , with the ordinates in a decimal logarithmic scale as in figure [ fig:2 ] .,scaledwidth=40.0% ] now , the zero dipole manifold is 4dimensional , so that in principle @xmath109 depends on four variables ( i.e. , position and velocity ) . recall however that on such a manifold the motion is a purely coulomb one , so that it is integrable . then , thinking in terms of action angle variables , the interesting ones are just the two actions , which in our case are @xmath110 , @xmath111 being the ( mechanical ) energy , and the angular momentum @xmath112 , whereas the angles , i.e. , the orientation of the orbit and the starting point of the motion on the orbit , are irrelevant . so we chose to integrate away the two angles from the density @xmath109 , and to analyze the corresponding reduced density , @xmath113 ( by abuse of language still denoting it by @xmath109 ) . the result is shown in figures [ fig:1][fig:3 ] . in figure [ fig:1 ] the reduced density @xmath109 is reported versus @xmath114 and @xmath112 . while a priori the measure could be spread over the whole rectangle , the figure shows that it is actually concentrated in an extremely small area . this is more clearly exhibited in figure [ fig:2 ] , which still covers the whole rectangle , but gives the density in a decimal lograthmic gray scale , which covers a thousand orders of magnitude . figure [ fig:3 ] is instead the same as figure [ fig:1 ] , restricted to a smaller portion of the rectangle , still with the ordinates in a decimal logarithmic scale . it shows that the function @xmath115 has a very complex structure , with values ranging over a thousand orders of magnitude in an apparently nonsmooth way . this fact might suggest that the measure is not absolutely continuous with respect to the restriction of the lebesgue one , possibly having some kind of fractal structure , but we leave this problem for possible future work . in any case , we have shown that the peculiar character of dissipation entering electrodynamics of point particles restricted to the dirac physical manifold , makes the final dynamics quite different from that of more familiar dissipative systems such as the lorenz one . indeed in the latter case it is the invariant attractor that has a _ strange _ character , while the natural invariant measure induced on by the dynamics is somehow trivial , as corresponding to that of a hyperbolic system . in the former case , instead , the invariant manifold is trivial ( being just the hyperplane @xmath116 in the ambient phase space ) , while it is the invariant measure which apparently has a strange character . we investigated the main modifications that are introduced in the coulomb helium atom model when the radiation reaction forces are taken into account in the simplest possible way , and the asymptotic dirac prescription is consequently introduced , according to which the phase space has to be restricted to the submanifold of points leading to motions with a finite emitted energy . the first qualitative result we found is that the problem of the breakdown of the atom ( corresponding to autoionization for generic initial data with negative energy ) , which makes the coulomb model inconsistent , is now eliminated . the second one is the existence of an invariant manifold of stable periodic orbits , that moreover is attractive . such a manifold is the zero dipole one , on which the system does not radiate energy away , in the dipole approximation . finally , the invariant measure naturally induced by the time flow on the zero dipole manifold was studied numerically . we showed that such a measure is far from trivial , being presumably non absolutely continuous with respect to the restriction of the lebesgue measure , possibly with some kind of fractal structure . this is at variance with the known examples of dissipative systems , for which the measure is trivial , while it is the attractor that is strange ( i.e. a fractal set ) . this is apparently a consequence of the peculiar character of the radiation reaction force , which makes the system dissipative when restricted to the dirac physical manifold , while making it expansive in the ambient phase space . a very interesting problem that remains open is that of comparing the special periodic orbits which are selected on the zero dipole manifold as the most probable ones according to the natural measure , with those , apparently of special physical relevance , studied by de luca . we leave this point for future work . 99 j. w. nicholson , mon . not . . soc . * 72 * , 49 ( 1911 ) . n. bohr , phil . mag . * 26 * , 477 ( 1913 ) . i. langmuir , phys . rev . * 17 * , 339 ( 1921 ) . m. poirier , phys . a * 40 * , 3498 ( 1989 ) . t. yamamoto , k. kaneko , phys . lett . * 70 * , 1928 ( 1993 ) . t. yamamoto , k. kaneko , prog . . phys . * 100 * , 1089 ( 1998 ) . j. de luca , phys . rev . lett . * 80 * , 680 ( 1998 ) . j. de luca , phys . e * 58 * , 5727 ( 1998 ) . lorenz , j. of atm . sci . * 20 * , 130 ( 1963 ) . landau , e.m . lifshitz , _ the classical theory of fields _ ( pergamon press , oxford 1962 ) . jackson , _ classical electrodynamics _ ( wiley , new york 1975 ) . p.a.m . dirac , proc . a * 167 * , 148 ( 1938 ) . lanford iii , _ qualitative and statistical theory of dissipative systems _ , in _ statistical mechanics _ , g. gallavotti ed . , cime summer school * 71 * , 24 ( springer , berlin 2011 ) . w. pauli , _ pauli lectures on physics - vol 1 _ ( mit press , cambridge 1973 ) . w. heitler , _ the quantum theory of radiation _ ( clarendon press , oxford 1966 ) . hale , a.p . stokes , j. math . phys . * 3 * , 70 ( 1962 ) . r. haag , z. naturforsch . a * 10 * , 752 ( 1955 ) . a. carati , p. delzanno , l. galgani , j. sassarini , nonlinearity * 8 * , 65 ( 1995 ) . a. carati , j. phys . a * 34 * , 5937 ( 2001 ) . m. marino , j. phys . a * 36 * , 11247 ( 2003 ) . krylov , n.n . bogolyubov , annals of math . * 38 * , 65 ( 1937 ) .
we study a classical model of helium atom in which , in addition to the coulomb forces , the radiation reaction forces are taken into account . this modification brings in the model a new qualitative feature of a global character . indeed , as pointed out by dirac , in any model of classical electrodynamics of point particles involving radiation reaction one has to eliminate , from the a priori conceivable solutions of the problem , those corresponding to the emission of an infinite amount of energy . we show that the dirac prescription solves a problem of inconsistency plaguing all available models which neglect radiation reaction , namely , the fact that in all such models most initial data lead to a spontaneous breakdown of the atom . a further modification is that the system thus acquires a peculiar form of dissipation . in particular , this makes attractive an invariant manifold of special physical interest , the zero dipole manifold , that corresponds to motions in which no energy is radiated away ( in the dipole approximation ) . we finally study numerically the invariant measure naturally induced by the time evolution on such a manifold , and this corresponds to studying the formation process of the atom . indications are given that such a measure may be singular with respect to that of lebesgue . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ at variance with the existing works on the dynamical properties of classical atomic systems ( see for example refs . ) , in this paper we take into account , for the case of the helium atom , the effects on the dynamics due to the radiation emitted by the electrons during their motion . this introduces a qualitative change in the dynamics , with respect to the purely coulomb model , because the model now corresponds to a non conservative dynamical system . this leads to the appearing of an `` attractive '' manifold of stable periodics orbits , where attractive is to be understood in a proper sense which will be explained below . the aim of this paper is to study the properties of such a manifold , through analytical and numerical methods . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
the spectrum of one - particle excitations of a system of correlated electrons in a solid is a fundamental question in condensed - matter physics . the theoretical understanding of the excitation spectrum poses a long - standing and not yet generally solved problem . within the independent - electron approximation the spectrum is simply given in terms of the one - particle eigenenergies of the hamiltonian . analogously , it is widely accepted to interpret a measured photo emission spectrum by referring to the results of band - structure calculations that are based on density functional theory ( dft ) and the local density approximation ( lsda ) @xcite . such an interpretation is questionable since there is actually no direct correspondence between the kohn - sham eigenenergies and the one - particle excitations of the system @xcite . for an in principle correct description of the excitation energies , the local lsda exchange - correlation potential has to be supplemented by the non - local , complex and energy - dependent self - energy which leads to the dyson equation @xcite instead of the schrdinger- or dirac - type equation in the non- or fully relativistic kohn - sham scheme . to overcome these deficiencies we have developed a generalized approach by accounting properly for electronic correlations beyond the lsda . for this purpose a general non - local , site - diagonal , complex and energy - dependent self - energy @xmath0 @xcite , has been included self - consistently in the fully relativistic korringa - kohn - rostoker multiple scattering theory ( sprkkr ) @xcite . within this scheme the self - energy is determined by a self - consistent dynamical mean field theory ( dmft ) calculation . this method is straightforwardly applicable for electronic structure calculations of 3d- and 2d - systems with perfect lateral translational invariance and arbitrary number of atoms per unit cell . furthermore , the calculational scheme has been extended to the case of disordered alloys @xcite , to include many - body correlation effects in the electronic structure and photo emission calculations of these class of complex compound materials . provided that the self - energy is known , one can deduce a pes / ipe `` raw spectrum '' from the solution of the dyson equation . to achieve a reliable interpretation of experiments , however , it is inevitable to deal with so - called matrix - element effects which considerably modify and distort the raw spectrum . above all , the wave - vector and energy dependence of the transition - matrix elements has to be accounted for . these dependencies are known to be important and actually can not be neglected . they result from strong multiple - scattering processes which dominate the electron dynamics in the low - energy regime of typically 1 - 200 ev @xcite . the transition - matrix elements also include the effects of selection rules which are not accounted for in the raw spectrum . strictly speaking , it can be stated that the main task of a theory of photo emission is to close the gap between the raw spectrum obtained by lsda or lsda+dmft electronic - structure calculations and the experiment . the most successful theoretical approach is the so - called one - step model of photo emission as originally proposed by pendry and co - workers @xcite . in the following a short overwiew will be given on the recent extensions of the one - step model which are connected with correlation effects , disordered alloys , phonon related features and high excitation energies far beyond the ultraviolet regime . the main idea of the one - step model is to describe the actual excitation process , the transport of the photoelectron to the crystal surface as well as the escape into the vacuum @xcite as a single quantum - mechanically coherent process including all multiple - scattering events . within this model self - energy corrections , which give rise to damping in the quasi - particle spectrum , are properly included in both the initial and the final states . this for example allows for transitions into evanescent band gap states decaying exponentially into the solid . similarly the assumption of a finite lifetime for the initial states gives the opportunity to calculate photo emission intensities from surface states and resonances . treating the initial and final states within the fully relativistic version of layer korringa - kohn - rostoker ( kkr ) theory @xcite , it is a straight forward task to describe complex layered structures like thin films and multilayers within the photo emission theory . furthermore , the surface described by a barrier potential can be easily included into the multiple - scattering formalism as an additional layer . a realistic surface barrier model which shows the correct asymptotic behavior has been introduced , for example , by rundgren and malmstrm @xcite . we start our considerations by a discussion of pendry s formula for the photocurrent which defines the one - step model of pes @xcite : @xmath1 the expression can be derived from fermi s golden rule for the transition probability per unit time @xcite , with @xmath2 denoting the elastic part of the photocurrent . vertex renormalizations are neglected . this excludes inelastic energy losses and corresponding quantum - mechanical interference terms @xcite . furthermore , the interaction of the outgoing photoelectron with the rest system is not taken into account . this `` sudden approximation '' is expected to be justified for not too small photon energies . we consider an energy- , angle- and spin - resolved photo emission experiment . the state of the photoelectron at the detector is written as @xmath3 , where @xmath4 is the component of the wave vector parallel to the surface , and @xmath5 is the kinetic energy of the photoelectron . the spin character of the photoelectron is implicitly included in @xmath3 which is understood as a four - component dirac spinor . the advanced green function @xmath6 in eq . ( [ eq : pendry ] ) characterizes the scattering properties of the material at the final - state energy @xmath7 . via @xmath8 all multiple - scattering corrections are formally included . for an appropriate description of the photo emission process we have to model the correct asymptotic behavior of @xmath9 beyond the crystal surface , introducing a single outgoing plane wave characterized by @xmath5 and @xmath10 . furthermore , the damping of the final state due to the imaginary part of the inner potential @xmath11 must be taken into account . we thus construct the final state within spin - polarized low - energy electron diffraction ( spleed ) theory considering a single plane wave @xmath12 advancing onto the crystal surface . using the standard layer - kkr method @xcite generalized for the relativistic case @xcite , we first obtain the spleed state @xmath13 . the final state is then given as the time - reversed spleed state ( @xmath14 is the relativistic time inversion operator ) . many - body effects are included phenomenologically in the spleed calculation , by using a parametrized , weakly energy - dependent and complex inner potential @xmath15 @xcite . this generalized inner potential accounts for inelastic corrections to the elastic photocurrent @xcite as well as the actual ( real ) inner potential , which serves as a reference energy inside the solid with respect to the vacuum level @xcite . due to the finite imaginary part @xmath16 , the flux of elastically scattered electrons is continuously reduced , and thus the amplitude of the high - energy wave field @xmath9 can be neglected beyond a certain distance from the surface . the practical calculation starts with the dirac hamiltonian @xmath17 which one has to consider in the framework of dft @xcite . in atomic units ( @xmath18 ) one has : @xmath19 @xmath20 denotes the ( effective ) spin - independent potential , and @xmath21 is the ( effective ) magnetic field given by @xcite : @xmath22 @xmath23 the constant unit vector @xmath24 determines the spatial direction of the ( uniform ) magnetization as well as the spin quantization axis . @xmath25 denotes the usual @xmath26 dirac matrix with the nonzero diagonal elements @xmath27 and @xmath28 , and the vector @xmath29 is given by its components @xmath30 ( @xmath31 ) in terms of the @xmath32 pauli - matrices @xmath33 . the `` low - energy '' propagator @xmath34 in eq . ( [ eq : pendry ] ) , i. e. the one - electron retarded green function for the initial state in the operator representation , yields the `` raw spectrum '' . it is directly related to the `` bare '' photocurrent and thereby represents the central physical quantity within the one - step model . @xmath35 is to be evaluated at the initial - state energy @xmath36 , where @xmath37 is the photon energy . in the relativistic case @xmath34 is described by a @xmath26 green matrix which has to be obtained for a semi - infinite stack of layers . this quantity is defined by the following equation : @xmath38 g_{1}^{+}({\bf r},{\bf r}',e_{i})=\delta({\bf r}-{\bf r}')~. \label{eq : eomgr}\end{aligned}\ ] ] in order to account for strong electronic correlations beyond the lsda - scheme one has to introduce a non - local , energy and spin - dependent potential v. this quantity can be defined in the following way : @xmath39 with @xmath40 and @xmath41 the resulting integro - differential equation for the initial - state one - electron retarded green function takes the form : @xmath42 g_1^{+}({\bf r},{\bf r}',e ) \nonumber \\ & + & \hspace*{-0.5 cm } \int v({\bf r},{\bf r}'',e)~g_1^{+}({\bf r''},{\bf r}',e ) d { \bf r}''= \delta({\bf r}-{\bf r}')~. \label{eq : dirac}\end{aligned}\ ] ] according to the lsda+dmft approach realized in the framework of the fully relativistic korringa - kohn - rostoker multiple scattering theory ( spr - kkr ) @xcite we use a self - energy @xmath43 calculated self - consistently using dynamical mean - field theory ( dmft ) @xcite . the explicit form in relativistic notation is given by : @xmath44 the spin - orbit quantum number @xmath45 and magnetic quantum number @xmath46 were combined in the symbol @xmath47 . + for disordered alloys the coherent potential approximation ( cpa ) scheme can straightforwardly be extended to include the many - body correlation effects for disordered alloys @xcite . according to the lsda+dmft approach realized in the framework of the fully relativistic spr - kkr multiple scattering theory we use a self - energy @xmath48 calculated self - consistently using dynamical mean field theory @xcite for the component @xmath49 of an alloy . within the kkr approach the local multi - orbital and energy dependent self - energy @xmath50 is directly included into the single - site matrix @xmath51 for the component @xmath49 , respectively by solving the corresponding dirac eq . ( [ eq : dirac ] ) . consequently , all the relevant physical quantities connected with the green s function and used for photo emission calculations , as for example , the matrix elements or scattering matrices like @xmath52 contain the electronic correlations beyond the lsda scheme . a detailed description of the generalizated one - step model for disordered magnetic alloys can be found in @xcite . here we present the expression for the cpa photocurrent only . following durham _ @xcite we have to perform the configurational average for the various contributions to the photocurrent : @xmath53 when dealing with disorder in the alloys , an additional contribution @xmath54 , the so called incoherent term appears . this contribution to the alloy photocurrent appears because the spectral function of an disordered alloy @xcite is a non single - site quantity . in fact this contribution is closely connected with the presence of the irregular wave functions well - known from the spherical representation of the green function g@xmath55 . within the next section we present some selected results from photo emission investigations on various systems to demonstrate the great flexibility of the method described above . we start with a photo emission study on mgo / fe(001 ) performed for photon energies in the hard x - ray regime . recently , the one - step model has been extended to the soft x - ray regime @xcite . starting with a new ansatz for the final state which generalizes the so called single - scatter approach , it becomes possible to account for the lattice for the complete photo emission calculation . this method is applicable also for higher photon energies ranging from 1 kev up to more or less than 10 kev . increasing the photon energy means increasing the bulk sensitivity of the corresponding photo emission data . we demonstrate this effect by comparing spectra obtained by photoelectron excitation with 1 kev and 6 kev radiation from the clean fe(001 ) surface and from the overlayer system mgo / fe(100 ) with 8 ml mgo on fe . 1a presents the spectrum for the clean fe(001 ) surface and a photon energy of 1 kev . as expected for this photon energy regime , the bulk sensitivity is enhanced and the surface emission is reduced to a negligible extent . also the relative intensity fraction of the sp - bands is obviously increased . going to a much higher photon energy of 6 kev the d - band intensity is strongly reduced when compared with sp - band related features . this is shown in fig . 1b . in a next step we put 8 ml of mgo on the fe(001 ) surface and repeat our calculations for both photon energies . this is shown in the lower panel of fig . 1 . at 1 kev mostly the mgo bulk band - structure is visible in fig . this is due to the thickness of the mgo film which consists of 8 ml . increasing the photon energy to 6 kev we expect due to the much larger mean free path length of the photo electron that fe - related features will recover , namely the fe sp - states . this is clearly demonstrated by fig . 1d which represents the corresponding photo emission spectrum calculated for the 8ml mgo / fe(001 ) system at a photon energy of 6 kev . besides the effect that more than one brillouin zone is visible for a fixed escape angle regime due to the very high photon energy it is undoubtly observable that nearly all mgo related features are vanished in the intensity plot . this , for example , should give access to buried interfaces . [ haxpes ] [ figure1 ] + the following examples concerns the ferromagnetic transition metal systems ni and fe as prototype materials to study electronic correlations and magnetism beyond the lsda scheme . in fig . 2 we present a comparison between experimental photo emission data @xcite and calculated spectra using different theoretical approaches @xcite . in the upper row spin - integrated arups measurements from ni(011 ) along @xmath56 for different angles of emission are shown . the dotted lines represent the experimental data , whereas the solid lines denote the single - particle approach to the measured spectral function . obviously , the lsda - based calculation completely fails to describe the experimental data . the energetic positions of the theoretical peaks deviate strongly from the measured ones . furthermore , the complicated intensity distributions that appear for higher angles of emission could not be reproduced by the lsda - based calculations at all . in contrast , the non self - consistent quasi - particle calculation provides a significant improvement when compared to the measured spectra . for the complete variety of emission angles the energetic peak positions coincidence with the experiment within about 0.1 ev . only the overall shape of the measured spectral intensities deviate from the calculations because of the neglection of multiple scattering and surface - related effects . in the experiment the different peaks seem to be more broadened and the spectral weight especially for nearly normal emission is shifted by about 0.1 ev to higher binding energies . in addition it seems that for very high emission angles like 60@xmath57 an even more complicated peak structure is hidden due to limited experimental resolution . an additional spin - analysis is therefore highly desirable for these experiments . [ figure2 ] for three different angles of emission . upper row : comparison between lsda - based calculation and experiment @xcite ; middle row : comparison between experiment and non self - consistent quasi - particle calculations neglecting matrix element and surface effects @xcite ; lower row : spin - integrated lsda+dmft spectra including photo emission matrix elements ( this work ) . theory : solid red line , experiment : black dots.,scaledwidth=45.0% ] [ figure3 ] n direction of the bulk brillouin zone . the curves are labeled by the wave vectors in units of @xmath58n=1.55 @xmath59 . ( b ) corresponding one - step model calculations based on the lsda+dmft method which include correlations , matrix elements and surface effects.,scaledwidth=45.0% ] within our work we could go far beyond previous theoretical studies by combining a sophisticated many - body approach as the self - consistent lsda+dmft method with the one - step based calculation of the corresponding spectral function . the self - energy within the dmft has been applied only for d - states and can be calculated in terms of two parameters - the averaged screened coulomb interaction @xmath60 and the exchange interaction @xmath61 . the screening of the exchange interaction is usually small and the value of @xmath61 can be calculated directly and is approximately equal to @xmath62 ev for all @xmath63 elements . this value has been adopted for all of our calculations presented here . for @xmath60 we used a value of @xmath64 ev . however , our test calculations show that a somewhat different choice of @xmath60 does not substantially changes features and trends in the calculated spectra . the intensity distributions resulting from the corresponding photo emission calculation are shown in the lower row of fig . 2 . a first inspection reveals very satisfying quantitative agreement between experiment and theory for all emission angles . let us concentrate first on the excitation spectrum calculated for @xmath65 . the spin - integrated spectrum exhibits a pronounced double - peak structure with binding energies of 0.1 ev and 0.3 ev . the second peak is slightly reduced in intensity which is also in accordance with the experimental findings . furthermore , the width of the spectral distribution is quantitatively reproduced . the calculated binding energies are related to the real part of the self - energy that corrects the peak positions due to dynamical renormalization procedure of the quasiparticles which is missing in a typical lsda - based calculations . the relative intensities and the widths of the different peaks , on the other hand , must be attributed to the matrix - element effects which enter our calculations from the very beginning via the one - step model of photo emission . the double - peak structure originates from excitation of the spin - split d - bands in combination with a significant amount of surface - state emission @xcite . the two spectra calculated for high angles of emission show the more broadened spectral distributions observable from the experimental data . an explanation can be given in terms of matrix - element effects , due to the dominating dipole selection rules . the spin - resolved spectra reveal a variety of d - band excitations in both spin channels , which in consequence lead to the complicated shape of the spectral distributions hardly to be identified in the spin - integrated mode . our second example within this section concerns a spectroscopic study on ferromagnetic fe @xcite . figures 3a and 3b display a comparison between spin - integrated arpes data and theoretical lsda+dmft based one - step photo emission calculations of fe(110 ) along the @xmath58n direction of the bulk brillouin zone ( bz ) with p - polarized radiation . in our lsda+dmft investigation underlying the arpes calculations we use for the averaged on - site coulomb interaction @xmath60 a value @xmath60=1.5 ev which lies between the experimental value @xmath661 ev @xcite and a value @xmath662 ev derived from theoretical studies @xcite . the @xmath67 values were calculated from the used photon energies ranging from 25 to 100 ev . near the @xmath58 point ( [email protected] @xmath58n ) , the intense peak close to the fermi level corresponds to a @xmath69 minority surface resonance , as indicated on top of fig . 3 . experimentally , its @xmath70 bulk component crosses the fermi level at @xmath710.33 @xmath58n , leading to a reversal of the measured spin - polarization and to a strong reduction of the intensity at @xmath720.68 @xmath58n in the minority channel . the peak at the binding energy [email protected] ev , visible mainly for p - polarization in a large range of wave vectors between @xmath58 and n , can be assigned to almost degenerate @xmath73 bulk - like majority states . a @xmath74 feature at [email protected] ev dominates the spectrum at the @xmath58-point . depending on the polarization its degenerate @xmath75 states form a shoulder around the same be . the broad feature around 2.2 ev , visible at various @xmath67-points , but not at the n - point , is related to a majority @xmath76 surface state . around the n - point ( [email protected] ) and at be@xmath783 ev we observe a @xmath79 band having strong sp character . the pronounced difference between its theoretical and experimental intensity distributions can be attributed to the fact that in the present calculations only local coulomb repulsion between d electrons is considered , without additional lifetime effects for the sp bands . finally , we notice that the background intensity of the spectrum at @xmath67=0.66 @xmath58n , corresponding to a photon energy of 55 ev , is strongly increasing by the appearance of the fe 3p resonance . in conclusion , we have presented spectral function calculations of ferromagnetic ni and fe , which coherently combine an improved description of electronic correlations with multiple - scattering , surface emission , dipole selection rules and other matrix - element related effects that lead to a modification of the relative photo emission intensities . as has been demonstrated , this approach allows on the one hand side a detailed and reliable interpretation of high - resolution angle - resolved photo emission spectra of 3d - ferromagnets . on the other hand , it also allows for a very stringent test of new developments in the field of dmft and similar many - body techniques . [ figure4 ] pd@xmath80(001 ) alloy surfaces as a function of the concentration x for a fixed photon energy of @xmath81=40.0 ev along @xmath58x in normal emission . experimental data shown in the left panel calculated spectra presented in the right panel . depending on the concentration @xmath82 a pronounced shift in spectral weight towards the fermi level is visible.,title="fig:",scaledwidth=23.0% ] pd@xmath80(001 ) alloy surfaces as a function of the concentration x for a fixed photon energy of @xmath81=40.0 ev along @xmath58x in normal emission . experimental data shown in the left panel calculated spectra presented in the right panel . depending on the concentration @xmath82 a pronounced shift in spectral weight towards the fermi level is visible.,title="fig:",scaledwidth=23.0% ] in this section we want to study the alloying effect in combination with electronic correlations . 4 shows a series of spectra as a function of the concentration @xmath82 calculated for a photon energy @xmath81=40 ev with linear polarized light . the experimental data have been presented in the left panel and the corresponding lsda+dmft - based photo emission calculations are presented in the right one . our theoretical analysis shows that starting from the pure ni case , the agreement is fully quantitative with deviations less than 0.1 ev binding energy , as expected . going to the ni@xmath83pd@xmath84 alloy the agreement is on the same level of accuracy concerning the range of binding energies between the fermi energy and 2 ev binding energy . inspecting the density of states ( dos ) for the ni@xmath83pd@xmath84 alloy this fact becomes explainable , because this energy interval represents the ni - dominated region . the pd - states start to appear at about 2 ev below @xmath85 besides of the small dip at the fermi level . for higher binding energies the agreement is also very good , although a bit more structure is observable in the theory , especially around 3.5 ev . an explanation for this behavior can be found in terms of lifetime effects , but it should be mentioned here that the experimental intensity background consisting of secondary electrons was not considered in the theoretical analysis . from the ni@xmath86pd@xmath87 alloy system it becomes clearly visible that the deviation between theory and experiment is mainly introduced for increasing concentration of pd . the same argument holds for the ni@xmath88pd@xmath88 and ni@xmath87pd@xmath86 alloys shown next in the series . in addition , the spectra of ni@xmath87pd@xmath86 reveal some deviations near the fermi level . also , the spectral intensity of the ni surface resonance , that appears at about 0.5 ev binding energy is underestimated in the calculation when compared to the experiment . our spectroscopical analysis has clearly demonstrated that the electronic properties of the ni@xmath89pd@xmath80 alloy system depend very sensitively on the interplay of alloying and electronic correlation . a description within the lsda approach in combination with a cpa method results in a qualitative description of the electronic structure of ni@xmath89pd@xmath80 only . [ figure5 ] the second example concerns more surface sensitive features , namely the spin - orbit - splitting observable for surface states from high z - materials like au @xcite . here we want to study the interplay between spin - orbit interaction and correlation using as a prototype material the magnetic au@xmath90ni@xmath80 alloy . we are able to tune the magnetic moment of the alloy as a function of the concentration @xmath82 and in consequence we can study both interactions on the same level of accuracy . 5 shows a series of spin - resolved photo emission spectra from the au@xmath90ni@xmath80(111)-surface . for a very low ni concentration of about 5@xmath91 the magnetic moment is nearly zero and we observe the pure spin - orbit split shockely surface state , well known from literature @xcite . with increasing the ni concentration the magnetic moment increases and the magnetic exchange interaction starts to compete with the spin - orbit splitting . exchange causes a splitting in energy , whereas spin - orbit interaction causes a splitting in k@xmath92 . for ni concentrations which lead to an exchange splitting comparable with the spin - orbit one we find the interesting result that the two parabolas seem to shift both in energy and k@xmath92 with the additional effect of structural distortion . for larger magnetic moments the exchange interaction starts to win against the spin - orbit interaction and one can observe as usual the exchange - split shockely surface state which has to be expected on the ni(111 ) surface . these two examples may illustrate that the use of the cpa alloy theory self - consistently combined with a dmft+lsda approach serves as a powerful tool for electronic structure calculations , whereas the application of the fully relativistic one - step model of photo emission , which takes into account chemical disorder and electronic correlation on equal footing guarantees a quantitative analysis of the corresponding spectroscopical data . nowadays high resolution photo emission measurements allow to investigate in very detail the electronic structure of materials close to the fermi level . this gives access to energy bands modified by electron - phonon interaction , and therefore serves as a very important spectroscopical tool to characterize transport properties or , for example , properties of super - conducting materials @xcite . modifications of the electronic structure generated by electron - phonon interaction are typically not accounted for in lsda - based band - structure calculations . the theoretical treatment of electron - phonon effects requires the use of many - body techniques which originally had been developed to give a quantitative description of strong electronic correlation effects . the self - energy approach often applied in many - body calculations can be mapped to the case of electron - phonon related correlations . therefore , having the electronic structure obtained by an ab - initio calculation , the modifications due to electron - phonon interaction can be described via an electron - phonon self - energy @xmath93 @xcite that appears in the momentum - resolved green s function operator @xmath94 the electron - phonon self - energy is derived using many - body perturbation theory within second order . it follows @xcite : @xmath95 where @xmath96 represents the self - energy obtained within the einstein model @xcite . @xmath97 denote the momentum - resolved eliashberg function : @xmath98 with the electron - phonon matrix element @xmath99 herein , @xmath100 stand for the wave vector and band index , respectively . @xmath101 is the phononic wave vector and @xmath102 denotes the number of phonon modes . @xmath103 defines the polarisation of a phonon mode and @xmath104 numbers the atoms within the unit cell . to obtain explicitly the electron - phonon self - energy , we focus our attention on the eliashberg function as the main quantity determined through all coupled electronic and phononic states of the system . this function can be reformulated in terms of the green s function by use of the identity ( see e.g. butler ( 1985 ) ) . after some algebra the eliashberg function follows as the sum over all phonon states : @xmath105~ , \nonumber\end{aligned}\ ] ] with @xmath106 according to gyrffy and coworkers @xcite the green s function @xmath107 can be expressed in terms of the scattering path operator at the fermi level @xmath108 and is therefore accsessible in terms of multiple scattering theory . [ figure6 ] for pb(110 ) for @xmath109 . lower panel : corresponding k - dependent self - energy , real part shown on left , imaginary part shown on the right.,title="fig:",scaledwidth=28.0% ] + for pb(110 ) for @xmath109 . lower panel : corresponding k - dependent self - energy , real part shown on left , imaginary part shown on the right.,title="fig:",scaledwidth=23.0% ] for pb(110 ) for @xmath109 . lower panel : corresponding k - dependent self - energy , real part shown on left , imaginary part shown on the right.,title="fig:",scaledwidth=23.0% ] the complex self - energy contributes due to electron - electron and electron - impurity scattering , as well as due to electron - phonon scattering described by the electron - phonon self - energy @xmath110 . the real and imaginary parts of @xmath110 are responsible for band renormalisations around @xmath85 . these quantities are determined by electron - phonon scattering induced quasiparticle lifetimes , measurable in photo emission experiments . to demonstrate the effect of electron - phonon interactions on the electronic quasi - particle states near @xmath85 we present photo emission calculations for pb , which serves as a prototype system and compare our theoretical results to corresponding experimental data @xcite . pb is of special interest because of the presence of strong electron - phonon interactions which in particular lead to the superconducting state at low temperatures @xcite . the calculations have been performed for energies close to the fermi level . the wave vectors in the bz ( @xmath109 ) correspond to the experimental geometry @xcite . [ figure6 ] shows a comparison between the experimental and theoretical real and imaginary parts of the self - energy @xmath111 . obviously , the agreement is very satisfying . [ figure7 ] in general , it is possible to deduce a phononic self - energy from measured photo emission spectra . the imaginary part of the self - energy can be estimated from the full width at half maximum of the energy distribution curves fitted by a lorenzian . the real part is than obtained by a kramers - kronnig transformation . however , in cases where for example several transitions appear close to the fermi level or even additional surface contributions to the photocurrent take place , this analysis might be questionable . therefore , it seems to be quite reasonable to make a direct comparison between angle resolved photo emission calculations and corresponding experimental data . for this purpose we developed a scheme that allows to include directly a @xmath112-dependent self - energy into the one - step model of photo emission . in fig . 7 we present photo emission spectra calculated for the two cases without ( left pannel ) and with ( middle panel ) electron - phonon interactions and compare them to the correponding measurements @xcite . the contour plot has been calculated for hei light along @xmath113 . more precisely , bands forming the locally tubular fermi surface of pb(110 ) crosses e@xmath114 at about @xmath115 off - normal emission along @xmath113 . in the middle pannel of fig . 7 a clear renormalisation of the bare band ( often called _ kink _ ) at @xmath116mev binding energy is observable . this effect appears just where the real part of the self - energy reaches its maximum . futhermore , an additional broadening of the spectral features due to the imaginary part of the phononic self - energy is visible in the calculated spectra . this , of course , completes a lifetime analyses of spectral features in a quantitative sense because the imaginary part of the phononic self - energy can be identified as the last contribution that has to be considered in addition to impurity scattering and electronic correlation related broadening mechanisms of photo emission intensities . we have shown several examples of photo emission calculations which reflect the wide range of applicability of the one - step model of photo emission . in its latest version that is directly combined with the sprkkr program package quantitative spectroscopical investigations may range from 10 ev up to 10 kev concerning the photon energy . this gives not only direct accsess to the uv regime but also to the so called haxpes - regime of angle resolved photo emission . due to the lsda+dmft method the range of applicability has been greatly enhanced ranging now from simple paramagnetic metals to complex layered compounds inhibiting strong electronic correlations . the recent extent of the one - step model to disorderd magnetic alloys round up the range of applicability to a variety of complex 3d- as well as 2d - materials and allows investigations on surface related features for low photon energies as well as bulk like analyses for higher photon energies . as it holds for photocurrent calculations in the haxpes regime phonon related effects are also very important when analysing the intensity distributions which are obtained for energies around the fermi level . the second point has been addressed in the last section in which we demonstrated how to introduce in a quatitative sense a k - dependent phononic self - energy that allows for spectroscopical calculations from kink - like structures . therefore this ansatz enlarges in combination with the lsda+dmft method again the range of applicability to systems like high t@xmath117 cuprats or other highly correlated classes of materials . financial support by the deutsche forschungsgemeinschaft through for 1346 , eb-154/18,eb-154/23 and mi-1327/1 is gratefully acknowledged . we would like to thank a. eiguren and c. ambrosch - draxl for giving access to the program calculating electron - phonon self - energy . j. braun , in _ new developments in up- and xp - spectroscopy from ferromagnetic materials , in band - ferromagnetism _ , _ lecture notes in physics _ , edited by k. baberschke and et al ( springer verlag , berlin , 2001 ) , p. 341 . j. snchez - barriga , j. fink , v. boni , i. di marco , j. braun , j. minr , a. varykhalov , o. rader , v. bellini , f. manghi , h. ebert , m. i. katsnelson , a. i. lichtenstein , o. eriksson , w. eberhardt , and h. a. drr , phys . lett . * 103 * , 267203 ( 2009 ) .
various technical developments enlarged the potential of angle - resolved photo emission ( arpes ) tremendously during the last one or two decades . in particular improved momentum and energy resolution as well as the use of photon energies from few ev up to several kev makes arpes a rather unique tool to investigate the electronic properties of solids and surfaces . obviously , this rises the need for a corresponding theoretical formalism that allows to accompany experimental arpes studies in an adequate way . as will be demonstrated by several examples this goal could be achieved by various recent developments on the basis of the one - step model of photo emission : the spin - orbit induced rashba - splitting of shockley - type surface states is discussed using a fully relativistic description . the impact of chemical disorder within surface layers can be handled by means of the coherent potential approximation ( cpa ) alloy theory . calculating phonon properties together with the corresponding electron - phonon self - energy allows a direct comparison with features in the arpes spectra caused by electron - phonon interaction . the same holds for the influence of electronic correlation effects . these are accounted for by means of the dynamical mean field theory ( dmft ) that removes the most serious short comings of standard calculations based on the standard local density approximation ( lda ) . the combination of this approach with the cpa allows the investigation of correlated transition metal alloys . finally , accounting for the photon momentum and going beyond the single scatter approximation for the final state allows to deal quantitatively with arpes in the high - energy regime ( haxpes ) that reduces the influence of the surface on the spectra and probing primarily the bulk electronic structure this way . corresponding calculations of arpes spectra , however , have to deal with thermal vibrations in an adequate way . for this , a new scheme is suggested that makes use of the cpa .
various astrophysical and cosmological observations provide convincing evidences for the existence of dark matter ( dm ) . dark matter distribution spans in wide range of scales from galaxy to clusters of galaxies and the large scale structure of the universe . recently , anomalous x - ray line emissions have been observed from galaxy clusters and also in the andromeda galaxy @xcite . while those might be a result of systematic effects , it would be interesting if the line came from the new source of astrophysical phenomena or from new physics . it was suggested that the signal might come from decaying dark matter with the mass and lifetime , @xmath2 assuming that they are the dominant component of dark matter . some theoretically interesting particle models have been suggested such as sterile neutrino @xcite , exciting dark matter @xcite , millicharged dark matter @xcite , axion like particle @xcite , in the effective theory @xcite . in this paper , we study the warm dark matter axino in a r - parity violating supersymmetric model . with bilinear r - parity breakings , neutralinos mix with neutrinos and thus the axino can decay into photon and neutrino . we find that the axino mass with @xmath3 kev can have the proper lifetime and relic density for the x - ray line emission . in this scenario , as an interesting consequence , the upper bound on the neutrino mass imposes that the bino mass is lighter than about @xmath4 gev . in section [ axino ] we introduce the model of axino dark matter and in section [ rparity ] we consider the r - parity violation and decay of axinos . we summarize in section [ conclusion ] . the strong cp problem and the hierarchy problem in the standard model can be naturally solved in the supersymmeric axion model @xcite . if axino , the fermionic superpartner of axion , is the lightest supersymmetric particle ( lsp ) , then it is a good candidate of dark matter @xcite . the effective operator of the axino can be derived by the supersymmetric transformation of the axion interactions and is given by @xmath5\tilde{g}^b g^b_{\mu\nu } + i\frac{\alpha_yc_{ayy}}{16\pi { f_{a}}}\overline{{{\tilde{a } } } } \gamma_5[\gamma^\mu,\gamma^\nu ] \tilde{y}y_{\mu\nu } , \label{eq : laxino}\ ] ] where @xmath6 is the peccei - quinn breaking scale , and @xmath7 , @xmath8 , @xmath9 and @xmath10 , @xmath11 , @xmath12 are the gauge couplings , gaugino fields and the field strength for @xmath13 and @xmath14 gauge groups respectively . the mass of axino is expected to be of the order of gravitino mass , but it can be much smaller @xcite , or much larger @xcite than the typical supersymmetric particle mass scale , depending on the specific models @xcite . here we take the light axino mass of the order of kev as dark matter component . the primordial axinos are generated from the thermal plasma during reheating after the primordial inflation . if the reheating temperature is lower than the decoupling temperature @xcite @xmath15 the axinos can not reach the thermal equilibrium . then axinos are generated through scatterings and decay of heavy particles in the thermal plasma , and the amount could be abundant enough for axino to be the dominant dark matter component @xcite . the abundance of thermally produced axinos depends on the reheating temperature for the ksvz axion model @xcite . the axino number density to entropy density ratio is estimated as @xcite @xmath16 with this , the relic density of non - relativistic axino at present is given by @xmath17 we can find that @xmath18 axino can be a natural candidate for warm dark matter when the reheating temperature is around @xmath19 and peccei - quinn scale @xmath20 . such thermally produced kev axino is a warm dark matter candidate and may solve various problems at the small scale in cold dark matter model @xcite . this range of reheating temperature is free from the gravitino problem @xcite . we consider the bilinear type r - parity violation with the usual @xmath21 -term superpotential in the minimal supersymmetric standard model @xmath22 where @xmath23 and @xmath24 are chiral super fields of the lepton doublet and up - type higgs doublet and @xmath25 parameterizes the size of the r - parity violation . by redefining the @xmath23 and @xmath26 , we can eliminate the r - parity violating term in eq . ( [ wp ] ) , then r - parity violating effect appears only in the scalar potential @xcite , @xmath27 where the coefficients are @xmath28 and @xmath29 . from this scalar potential , the sneutrinos obtain non - zero vacuum expectation values ( vevs ) @xmath30 where @xmath31 and @xmath32 and @xmath33 is the sneutrino mass . since the non - zero vevs of sneutrinos induce mixings between leptons and gauginos , the neutrinos mix with neutralinos and can obtain mass at the tree level as @xcite @xmath34 where @xmath35 parameterizes the r - parity breaking given by @xmath36 the upper bound of the neutrino mass @xmath37 constrains the r - parity breaking parameters . the baryon asymmetry can be erased in the early universe by the @xmath38 violating interactions @xcite . however if one of the lepton flavors of the r - parity violating couplings is preserved and the slepton mixing is small enough , the problem can be avoided @xcite . due to the r - parity violation , the stability of lsp is not guaranteed anymore @xcite . for bilinear r - parity breakings , the light axino decays dominantly into photon and neutrino and the decay rate is given by @xcite @xmath39 where the photino - zino mixing is given by @xmath40 with the neutralino mixing matrix @xmath41 . in the case of @xmath42 , it is simplified to be @xmath43 although axino can also decay to three neutrinos mediated by z - boson , this mode is highly suppressed and negligible . [ cols="^ " , ] the relatively large @xmath35 can be inconsistent with the mass of neutrino from eq . ( [ mnu ] ) . in fig . [ fig : xi ] , we show the contour plot of @xmath44 in the plane of @xmath45 . to explain the neutrino mass @xmath46 with @xmath47 and @xmath48 , the value in the contour need to be smaller than @xmath49 . that can be obtained when @xmath50 with slight dependence on @xmath51 . an interesting point to note is that the value of @xmath52 implies light bino @xmath53 for @xmath54 from the upper bound for neutrino mass . axino is a good candidate for dark matter . when its mass is around @xmath55 and r - parity is broken bilinearly , the axino decays into photon and neutrino . we studied this decaying axino warm dark matter in the light of the recent observation of x - ray line emission from the center of galaxy clusters and andromeda galaxy observed by xmm - newton . we find that the decaying axino can naturally explain the x - ray signal with / without additional component of dark matter . we note that the neutrino mass bound implies that the bino mass is less than about @xmath4 gev . as this work was being submitted , a paper @xcite appeared that also discusses decaying axino dark matter as a source of the x - ray line signal . k .- y.c . would like to acknowledge the hospitality of hokkaido university during his visit , where a part of this work was carried . k .- y.c . was supported by the basic science research program through the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology grant no . 2011 - 0011083 . # 1#2#3phys . rep . * # 1 * # 2 ( # 3 ) # 1#2#3rev . phys . * # 1 * # 2 ( # 3 ) # 1#2#3annu . nucl . part . sci . * # 1 * # 2 ( # 3 ) # 1#2#3nucl . * b#1 * # 2 ( # 3 ) # 1#2#3phys . * b#1 * # 2 ( # 3 ) # 1#2#3phys . * d#1 * , # 2 ( # 3 ) # 1#2#3phys . lett . * # 1 * # 2 ( # 3 ) # 1#2#3jhep * # 1 * # 2 ( # 3 ) # 1#2#3jcap * # 1 * # 2 ( # 3 ) # 1#2#3z . phys . * # 1 * # 2 ( # 3 ) # 1#2#3euro . j. * # 1 * # 2 ( # 3 ) # 1#2#3int . j. mod . * # 1 * # 2 ( # 3 ) # 1#2#3mod . # 1 * # 2 ( # 3 ) # 1#2#3astrophys . j. * # 1 * # 2 ( # 3 ) # 1#2#3nature * # 1 * # 2 ( # 3 ) # 1#2#3sov . j. nucl . # 1 * # 2 ( # 3 ) # 1#2#3astrophys . j. * # 1 * # 2 ( # 3 ) # 1#2#3int . j. mod . phys . * # 1 * # 2 ( # 3 ) # 1#2#3astropart . * b#1 * , # 2 ( # 3 ) # 1#2#3mon . not . r. astron . soc . * # 1 * # 2 ( # 3 ) # 1#2#3nature ( london ) * # 1 * # 2 ( # 3 ) # 1#2#3astrophys . j. supp . * # 1 * # 2 ( # 3 ) # 1#2#3aip conf . # 1 * # 2 ( # 3 ) l. covi , j. e. kim and l. roszkowski , phys . lett . * 82 * , 4180 ( 1999 ) . l. covi , h. b. kim , j. e. kim and l. roszkowski , . k. -y . choi , l. covi , j. e. kim and l. roszkowski , jhep * 1204 * 106 ( 2012 ) . k. -y . choi , j. e. kim and l. roszkowski , j. korean phys . soc . * 63 * 1685 ( 2013 ) . a. a. klypin , a. v. kravtsov , o. valenzuela and f. prada , astrophys . j. * 522 * , 82 ( 1999 ) ; + b. moore , s. ghigna , f. governato , g. lake , t. r. quinn , j. stadel and p. tozzi , astrophys . j. * 524 * , l19 ( 1999 ) . b. moore , nature * 370 * , 629 ( 1994 ) ; + r. a. flores and j. r. primack , astrophys . j. * 427 * , l1 ( 1994 ) ; + w. j. g. de blok and s. s. mcgaugh , mon . not . . soc . * 290 * , 533 ( 1997 ) . k. jedamzik , phys . d * 70 * 063524 ( 2004 ) . m. kawasaki , k. kohri and t. moroi , phys . b * 625 * , 7 ( 2005 ) ; + m. kawasaki , k. kohri and t. moroi , phys . d * 71 * , 083502 ( 2005 ) . r. barbier , c. berat , m. besancon , m. chemtob , a. deandrea , e. dudas , p. fayet and s. lavignac _ et al . _ , phys . * 420 * 1 ( 2005 ) . m. hirsch and j. w. f. valle , new j. phys . * 6 * 76 ( 2004 ) . m. hirsch , m. a. diaz , w. porod , j. c. romao and j. w. f. valle , phys . d * 62 * ( 2000 ) 113008 [ erratum - ibid . d * 65 * ( 2002 ) 119901 ] [ hep - ph/0004115 ] . b. a. campbell , s. davidson , j. r. ellis and k. a. olive , phys . b * 256 * ( 1991 ) 484 . w. fischler , g. f. giudice , r. g. leigh and s. paban , phys . b * 258 * ( 1991 ) 45 . h. k. dreiner and g. g. ross , nucl . b * 410 * ( 1993 ) 188 [ hep - ph/9207221 ] . m. endo , k. hamaguchi and s. iwamoto , jcap * 1002 * ( 2010 ) 032 [ arxiv:0912.0585 [ hep - ph ] ] . d. hooper and l. -t . wang , phys . d * 70 * 063506 ( 2004 ) . e. j. chun and h. b. kim , jhep * 0610 * 082 ( 2006 ) . m. endo , k. hamaguchi , s. p. liew , k. mukaida and k. nakayama , phys . b * 721 * 111 ( 2013 ) . l. covi and j. e. kim , new j. phys . * 11 * ( 2009 ) 105003 [ arxiv:0902.0769 [ astro-ph.co ] ] . k. -y . choi , d. restrepo , c. e. yaguna and o. zapata , jcap * 1010 * ( 2010 ) 033 [ arxiv:1007.1728 [ hep - ph ] ] . s. bobrovskyi , w. buchmuller , j. hajer and j. schmidt , jhep * 1010 * ( 2010 ) 061 [ arxiv:1007.5007 [ hep - ph ] ] . bae , j .- h . huh and j. e. kim , . k. kong , j. -c . park and s. c. park , arxiv:1403.1536 [ hep - ph ] .
we consider axino warm dark matter in a supersymmetric axion model with r - parity violation . in this scenario , axino with the mass @xmath0 kev can decay into photon and neutrino resulting in the x - ray line signal at @xmath1 , which might be the origin of unidentified x - ray emissions from galaxy clusters and andromeda galaxy detected by the xmm - newton x - ray observatory .
the physics of compact objects is entering a particularly exciting phase . new instruments , including x - ray and gamma - ray satellites and the new neutrino observatories sno and super - kamiokande , can now yield unprecedented observations of neutron stars and black holes . perhaps most excitingly , the new gravitational wave detectors ligo , tama , geo and virgo promise to open a gravitational wave window to the universe and make gravitational wave astronomy a reality ( see , e.g. , @xcite ) . simultaneously , the availability of computational resources at modern supercomputers makes the simulation of realistic astrophysical scenarios involving relativistic compact objects feasable . several groups , including two `` grand challenge alliances '' @xcite , have launched efforts to construct numerical codes capable of solving einstein s equations with or without matter sources in three dimensions , and simulating the merger of black hole or neutron star binaries ( see , e.g. , @xcite ) . numerical simulations will be necessary to predict the gravitational wave form from such processes to increase the likelihood of detections and , ultimately , to extract physical information from observations . even though much progress has recently been made ( see @xcite for the most recent developments ) , significant obstacles still remain . in this contribution , we report on our systematic approach towards constructing such a numerical code . we first review our formulation of einstein s equations and describe several tests , both for vacuum spacetimes and analytical matter sources , which demonstrate its advantages for numerical integrations . we then outline our implementation of hydrodynamics , and present promising numerical test results . we adopt geometrized units ( g = c = 1 ) and the convention that greek indices run from 0 to 3 , while latin indices only run from 1 to 3 . most numerical implementations of einstein s equations adopt a cauchy formulation based on the 3 + 1 formulation by arnowitt , deser and misner ( @xcite , see , e.g. , @xcite for an alternative characteristic formulation ) . however , a straightforward implementation of these `` undressed '' adm equations tends to develop instabilities even for the evolution of small amplitude gravitational waves on a flat background ( compare @xcite ) . following shibata and nakamura @xcite and the spirit of many earlier one and two - dimensional codes @xcite , we have recently developed a modification of the adm equations which proves to be much more suitable for numerical implementations @xcite . more specifically , we modify the adm equations in two ways . first , we split the spatial metric @xmath0 into a conformal factor @xmath1 and a conformally related metric @xmath2 according to @xmath3 and evolve @xmath4 and @xmath2 separately . we choose @xmath4 such that @xmath5 , and similarly split the extrinsic curvature into its trace and its trace - free part . this split separates the `` radiative '' variables from the `` non - radiative '' variables in the spirit of the `` york - lichnerowicz '' decomposition @xcite . in the second stage we introduce `` conformal connection functions '' @xmath6 as independent functions ( compare @xcite ) . some of the mixed second derivatives of @xmath2 in the conformal , spatial ricci tensor @xmath7 can now be written as first derivatives of the @xmath8 . as a result , @xmath7 becomes a manifestly elliptic operator on the metric @xmath9 . the analogous technique was used for the four - dimensional ricci tensor @xmath10 as early as in the 1920 s to make einstein s equations manifestly hyperbolic @xcite . for more details of our formulation , including an evolution equation for the @xmath11 , the reader is referred to @xcite ( see also the mathematical analysis in @xcite , and further numerical applications in @xcite ) . in @xcite , we tested this form of einstein s equations for small amplitude gravitational waves , and found that it performs far better than a similar implementation of the original adm formulation . in particular , we found that we could evolve such waves with harmonic slicing without encountering growing instabilities , whereas the original adm code crashed after about 35 light - crossing times for the same initial data . in @xcite , we inserted analytic matter sources on the right hand side of einstein s equations and evolved the fields in their presence . this approach allows us to study the numerical properties of the field evolution in the presence of highly relativistic matter sources without having to solve the equations of hydrodynamics : `` hydro - without - hydro '' . we inserted the oppenheimer - snyder solution for a relativistic , static star to test the long term stability of the evolution code , and the oppenheimer - volkov solution for the collapse of a sphere of dust to a schwarzschild black hole . these simulations focus on the highly relativistic , longitudinal fields , and complement our earlier tests involving dynamical transverse fields in @xcite . with the code having passed these tests , we are now implementing both collisionless matter , which will be described elsewhere , and hydrodynamics , as described below , to evolve the matter self - consistently with the fields . for a perfect fluid , the stress - energy tensor can be written @xmath12 where @xmath13 is the rest mass density , @xmath14 the specific internal density , @xmath15 the pressure , @xmath16 the fluid four velocity , and @xmath17 the four dimensional spacetime metric . we construct constant entropy initial data with a polytropic equation of state @xmath18 where @xmath19 and @xmath20 is the polytropic index , and where we assume the polytropic constant @xmath21 to be unity without loss of generality . during the evolution , we adopt the gamma - law relation appropriate for adiabatic flow , @xmath22 following @xcite , we write the equation of continuity @xmath23 and the equations of motion @xmath24 in the form @xmath25 here we have defined the auxiliary quantities @xmath26 similar equations have been used by many other groups ( e.g. @xcite and references therein ) . we integrate equations ( [ e : twb:1 ] ) - ( [ e : twb:2 ] ) with an artificial viscosity scheme suggested in @xcite ( see @xcite for implementations of more elaborate shock capturing schemes ) . given @xmath27 , @xmath28 and @xmath29 on a new timelevel , @xmath30 can be found iteratively from the normalization relation @xmath31 , which yields @xmath32 the matter sources for the right hand sides of einstein s equations can then be constructed from these variables . as a first test of our implementation of hydrodynamics , we adopt the oppenheimer - volkov solution describing equilibrium neutron stars in spherical symmetry as initial data and evolve these dynamically . constructing a sequence of such oppenheimer - volkov solutions for increasing central rest mass densities @xmath33 , the adm mass @xmath34 of the star takes a maximum @xmath35 at @xmath36 ( see the left panel in figure 1 ) . for central densities smaller than @xmath36 , the star is in stable equilibrium , while for @xmath37 it is unstable and will collapse to a black hole . we adopt a polytropic equation of state with @xmath38 ( @xmath39 ) , for which @xmath40 and @xmath41 when @xmath42 . we choose as initial data the two configurations marked with filled circles in figure 1 ; a stable configuration with @xmath43 , @xmath44 and @xmath45 , and an unstable configuration with @xmath46 , @xmath47 and @xmath48 . in the right panel of figure 1 , we show the central density as a function of time for the two initial configurations . we performed these runs on quite modest numerical grids with @xmath49 gridpoints using cartesian coordinates and imposing outgoing wave boundary conditions at @xmath50 . we employ harmonic slicing and zero shift . as expected , the unstable configuration soon collapses , while the stable configuration remains stable for several dynamical timescales ( the period of the fundamental radial oscillation for a @xmath51 configuration is approximately @xmath52 , compare @xcite ) . ultimately , accumulation of numerical error causes this configuration to collapse too , but this can be delayed by increasing the grid resolution . in figure 2 we show density contours at the beginning and towards the end of the simulation for the unstable configuration with @xmath53 initially . we also include arrows indicating the fluid flow @xmath29 . the star is rapidly contracting and collapsing to a black hole . up to the late stage of the collapse , the mass @xmath34 is conserved to about 5 % . we can follow the collapse to about a 18 fold increase of the central density . by this time , the central lapse has decreased from 0.4 initially to about 0.03 . we report on our systematic approach towards constructing a fully relativistic hydrodynamics code in three spatial dimensions . as part of this program , we have developed a new formulation of einstein s equations which in several tests and applications has proved to be much more suitable for numerical implementations than the traditional adm formulation @xcite . mathematical properties of this formulation have been analyzed in @xcite . we have studied the evolution of small amplitude gravitational waves to test dynamical transverse fields and have inserted analytical solutions as matter sources ( hydro - without - hydro ) to test highly relativistic longitudinal fields . we outline our implementation of the relativistic equations of hydrodynamics and present preliminary test results for spherical neutron stars in hydrostatic equilibrium ( oppenheimer - volkov stars ) . as expected , we find that stable stars remain stable for several dynamical timescales , while unstable stars soon collapse to black holes . we conclude that our method seems like a very promising approach towards simulating the final plunge and merger of binary neutron stars . on a more speculative note , we point out that fully self - consistent hydrodynamics may not be a feasable approach towards simulating the coalescence and gravitational wave emission from neutron stars in the intermediate inspiral phase . in this epoch , the stars are very close and interact through a strongly relativistic tidal field , but reside outside the innermost stable circular orbit and hence move on a nearly circular orbit . simulating the slow inspiral would require evolving the stars for hundreds or thousands of orbits , which is presently impossible . it may be possible , however , to insert known quasi - equilibrium binary configurations ( e.g. @xcite ) into the field evolution code to get the transverse wave components approximately . decreasing the orbital separation ( and increasing the binding energy ) at the rate consistent with the computed outflow of gravitational - wave energy would generate an approximate strong - field wave inspiral pattern . such a `` hydro - without - hydro '' calculation may yield an approximate gravitational waveform from inspiraling neutron stars , without having to couple the matter and field integrations . calculations were performed on sgi cray origin2000 computer systems at the national center for supercomputing applications , university of illinois at urbana - champaign . this work was supported by nsf grants ast 96 - 18524 and phy 99 - 02833 and nasa grant nag 5 - 7152 at illinois . for example : j. m. bardeen and t. piran , _ phys . rep . _ * 96 * , 205 ( 1983 ) ; c. r. evans , phd thesis , university of texas at austin ( 1984 ) ; a. m. abrahams and c. r. evans , _ phys . rev . d _ * 37 * , 318 ( 1988 ) ; a. m. abrahams , g. b. cook , s. l. shapiro and s. a. teukolsky , _ phys . d _ * 49 * , 5153 ( 1994 ) .
we report on our numerical implementation of fully relativistic hydrodynamics coupled to einstein s field equations in three spatial dimensions . we briefly review several steps in our code development , including our recasting of einstein s equations and several tests which demonstrate its advantages for numerical integrations . we outline our implementation of relativistic hydrodynamics , and present numerical results for the evolution of both stable and unstable oppenheimer - volkov equilibrium stars , which represent a very promising first test of our code .
if one assumes that the graviton background exists with the planckian spectrum and an effective temperature @xmath0 , which we will consider in a flat space - time , an energy of any photon decreases with a distance @xmath1 so a redshift @xmath2 is equal to @xcite : @xmath3 here @xmath4 where the hubble constant @xmath5 @xmath6 is an average graviton energy , @xmath7 is the stephan - boltzmann constant , @xmath8 is a new constant . it is necessary to have the following value of this constant : @xmath9 i.e. gravitons should be super - strong interacting particles . in this approach , the newton constant @xmath10 is connected with @xmath11 that gives the following value of it @xcite : @xmath12 by @xmath13 additional photon flux s average energy losses on a way @xmath14 due to rejection of a part of photons from a source - observer direction should be proportional to @xmath15 where the relaxation factor @xmath16 is equal to : @xmath17 as it is shown in the next section . it is shown here how to find the value of relaxation factor @xmath16 which was used in author s paper @xcite . let us assume that by non - forehead collisions of a graviton with a photon , the latter leaves a photon flux detected by a remote observer ( an assumption of narrow beam of rays ) . so as both particles have velocities @xmath18 a cross - section of interaction , which is `` visible '' under an angle @xmath19 ( see fig . 1 ) , will be equal to @xmath20 if @xmath21 is a cross - section by forehead collisions . the function @xmath22 allows to take into account both front and back hemispheres for riding gravitons . additionally , a graviton flux , which falls on a picked out area ( cross - section ) , depends on the angle @xmath23 we have for the ratio of fluxes : @xmath24 where @xmath25 and @xmath26 are the fluxes which fall on @xmath21 under the angle @xmath19 and normally , @xmath27 is a square of side surface of a truncated cone with a base @xmath21 ( see fig . = 12.98 cm finally , we get for the factor @xmath28 @xmath29 by @xmath30 a formed cone contains self - intersections , and it is @xmath31 . by @xmath32 we have @xmath33 . after computation of simple integrals , we get : @xmath34 this additional relaxation of any photonic flux due to non - forehead collisions of gravitons with photons leads in this model to the luminosity distance @xmath35 @xmath36 = 12.98 cm the model may be compared with supernova data by riess et al . @xcite if one introduces distance moduli @xmath37 , where @xmath38 is a constant ( it is a single free parameter to fit the data ) ; @xmath39 is the luminosity distance in units of @xmath40 . in figure 2 , taken from my paper @xcite , the function @xmath41 is shown with @xmath42 to fit observations for low redshifts ; observational data are taken from table 5 of @xcite . the predictions fit observations very well for roughly @xmath43 . given this concordance between the theory _ without any kinematics _ and observations for low redshifts , we can think that any expansion of the universe _ is not necessary _ for higher redshifts , too . there are discrepancies between predicted and observed values of @xmath41 for higher @xmath2 . it would be explained in the model as a result of deformation of sn spectra due to a discrete character of photon energy losses . today , a theory of this effect does not exist , and its origin may be explained only qualitatively @xcite . a main goal of any physical model is not to fit observational data but to understand them to have a possibility to build an adequate picture of the nature . there are a few facts - the pioneer 10 anomaly @xcite , an existence of redshifts and sne 1a specific dimming , - which may be explained from one point of view in the considered approach . this approach needs an existence of the graviton background with very unexpected properties . today , there is not any model in physics of particles dealing with _ an external sea of particles _ with a fixed spectrum . our understanding of cosmology , gravitational physics and physics of particles would be essentially changed if this background really exists . references m.a.ivanov , general relativity and gravitation , * 33 * , 479 ( 2001 ) ; erratum : * 35 * , 939 ( 2003 ) ; [ astro - ph/0005084 v2 ] . m.a.ivanov . screening the graviton background , graviton pairing , and newtonian gravity [ gr - qc/0207006 ] . a.g . riess et al . type ia supernova discoveries at @xmath44 from the hubble space telescope ... , [ astro - ph/0402512 ] ( to appear in apj , 2004 ) . ivanov m.a . contribution to qels95 , may 21 - 26 , 1995 , baltimore , usa ; contribution to eqec96 , sept . 8 - 13 , 1996 , hamburg , germany . another possible interpretation of sn 1a data - without kinematics : comments on the paper astro - ph/0402512 by a. riess et al . [ astro - ph/0403130 ] . j.d.anderson et al . lett . , 1998 , v.81 , p. 2858 ; phys . rev . * d65 * ( 2002 ) 082004 . [ gr - qc/0104064 v4 ]
if gravitons are super - strong interacting particles which fulfill a flat non - expanding universe , we would have another possibility to explain cosmological redshifts - in a frame of non - kinematic model . it is shown by the author that in this case sne 1a data may be understood without any dark energy and dark matter . a value of relaxation factor is found in this paper . in this approach , we have newton s law of gravity as a simplest consequence , and the connection between newton s and hubble s constants . a value of the latter may be theoretically predicted .
the gelation transition transforms a viscous liquid ( sol ) into an elastic disordered solid ( gel ) . in general this process is due to the formation of a macroscopic molecule , due to the bonding of multifunctional monomers in solution , which makes the system able to bear stress . the extent of the gelation process may be measured by the monomer volume fraction @xmath0 , defined as @xmath1 , where @xmath2 is the number of monomers , @xmath3 is the single monomer volume and @xmath4 is the total system volume . on the static point of view , the sol - gel transition was interpreted @xcite in terms of the appearance of a percolating cluster of monomers linked by bonds @xcite , and experimental measurements of the geometric properties of gels have confirmed this correspondence ( for a review see stauffer et al.@xcite and references therein ) . complex dynamical behaviours are observed in gelling systems already in the sol phase . for example , light scattering measurements show non - exponential decay of the intermediate scattering function , @xmath5 , in both permanent @xcite and thermoreversible physical gels @xcite . in particular power laws are observed at intermediate times , followed , at long times , by stretched exponential decays , @xmath6 , with @xmath7 . in the gel phase , where ergodicity is broken , only the power law decay survives . usually the onset of stretched exponential decays ( present also in other complex systems , as spin glasses and glassy systems ) is associated to the widening of relaxation times , which in gelling systems is due to the presence of a broad cluster size distribution close to the gelation threshold . however general predictions which connect this kind of relaxation to percolation theory are not easily feasible . in this paper , assuming `` simple '' behaviours for the relaxation of clusters with given size , we show how the `` complex '' phenomenology of the relaxation in permanent gels may be obtained from the superposition of the behaviours of clusters with different sizes . in particular we are able to predict the behaviours of the self intermediate scattering function , of the dynamical susceptibility , of the van - hove function , and of the non - gaussian parameter , and to connect these behaviours to the cluster size distribution and the critical exponents of percolation . then we check the theoretical predictions in a specific model for permanent gels , studied using molecular dynamics simulations . the paper is organized as follows . in sect.[sect2 ] the analytical results are briefly summarized , and in sect.[sect3 ] they are compared with the data obtained by molecular dynamics simulations of a model for permanent gels , where bonds between monomers are described by a fene potential @xcite . in sect.[sect4 ] concluding remarks are discussed . finally , in [ appendix1 ] , [ appendix2 ] and [ appendix3 ] the calculations are presented in details . in this section we summarize our calculations , which will be show in details in appendices . we consider a system of randomly distributed monomers with a fixed volume fraction , @xmath0 . at time @xmath8 permanent bonds are introduced at random between monomers at a distance @xmath9 , where @xmath10 is suitably chosen . for a particular model see the fene model @xcite ( sect . [ sect4 ] ) , however the following arguments are independent on the details of the model . following the percolation approach @xcite , we identify the gel phase as the state where a percolating cluster is present . we denote by @xmath11 the volume fraction of the percolation threshold . in our calculations we use some results from percolation theory @xcite : in particular , in the sol phase , near the threshold , the cluster size distribution is given by @xmath12 ( where @xmath13 is a cut - off value given by @xmath14 , @xmath15 is the fractal dimension , and @xmath16 is the connectedness length which diverges at the threshold with the exponent @xmath17 ) ; in the gel phase , near the threshold , @xmath18 is put equal to @xmath19 , where @xmath20 is the density of particles in the percolating cluster of mass @xmath21 , @xmath22 is the spatial dimension and @xmath23 is a constant . moreover we assume that the relaxation time of clusters increases as a power law of the size @xcite , @xmath24 . with these assumptions , in the hypothesis of simple behaviour for single cluster ( exponential relaxation , simple diffusion , etc . ) , we obtain all the complex phenomenology observed experimentally and numerically near the threshold in gelling systems . in particular we are able to predict the behaviours of the self intermediate scattering function , of the dynamical susceptibility , of the van - hove function , and of the non - gaussian parameter . we first consider the self intermediate scattering functions ( isf ) : @xmath25\ ] ] where @xmath26 is the thermal average over a fixed bond configuration , @xmath27 $ ] is the average over independent bond configurations of the system , @xmath28 and @xmath2 is the number of particles . in the following we fix the wave vector @xmath29 and @xmath30 , with @xmath31 the linear system dimension . in terms of the contributions due to different clusters , @xmath32 can be written as @xmath33 , \label{eq5}\ ] ] where @xmath34 is the cluster size distribution ( @xmath35 gives the number of clusters of size @xmath36 ) and @xmath37 , where @xmath38 is the self isf limited to a given cluster @xmath39 of size @xmath36 , and @xmath40 is the average over all clusters of given size @xmath36 . by replacing the sum with the integral , the self isf for a given bond configuration becomes : @xmath41 by assuming @xmath42 the integral in eq.([integral ] ) gives in the thermodynamic limit the following predictions for the time dependence of @xmath32 in a permanent gel : 1 . at the gelation threshold ( @xmath43 ) @xmath44 where @xmath45 is the @xmath46-function with @xmath47 . 2 . in the sol phase ( @xmath48 ) @xmath49 where @xmath50 , @xmath51 , and @xmath52 , and @xmath53 . this approximated form , obtained in the long time limit , coincides with that suggested by ogielski@xcite as fitting function for the time dependent order parameter in spin glasses , and it is in agreement with experimental @xcite and numerical @xcite findings in gelling systems . 3 . in the gel phase ( @xmath54 ) @xmath55 where @xmath50 , and @xmath51 are the same exponents obtained in the sol phase . the plateau value , @xmath20 , gives the density of _ localized _ particles @xcite . clearly the main contribution comes from localized particles of the percolating cluster , however a small contribution may be due to particles trapped inside it . the calculations are shown in details in [ appendix1 ] . our findings given in eq.s ( [ conto_soglia ] ) , ( [ conto4 ] ) and eq.([conto_gel ] ) are in agreement with the theoretical predictions obtained in ref.@xcite in the rouse and zimm models for randomly cross - linked monomers , where @xmath56 and @xmath57 respectively . similar calculations are also done in ref.@xcite in a different context . in ref . @xcite it was studied the dynamical susceptibility , defined as the fluctuations of the self isf : @xmath58 . \label{chi4}\ ] ] in particular it was shown that , in the sol phase , in the limit of @xmath59 and @xmath60 , @xmath61 coincides with the mean cluster size , @xmath62 : _ k0_t _ 4(k , t)=s , which diverges at the threshold @xcite with the exponent @xmath63 . here we are interested in the time dependence of the dynamical susceptibility approaching the asymptotic value . we neglect the contributions due to disconnected particles at each time @xmath64 . in this way we can write @xmath61 as a superposition of the contributions due to different clusters : _ 4(k , t ) , [ eq10 ] where again @xmath38 is the self isf limited to a given cluster @xmath39 of size @xmath36 , @xmath40 is the average over all clusters of given size @xmath36 , @xmath65 is the thermal average over a fixed bond configuration , and @xmath27 $ ] is the average over independent bond configurations . the term @xmath66 in eq.([eq10 ] ) can be written as @xmath67 where we have put @xmath68 . for connected particles @xmath69 and @xmath70 , @xmath71 is finite , and in the low wave vector limit where @xmath72 we can assume @xmath73 . then , by supposing that @xmath74 , in the zero wave vector limit the dynamical susceptibility for a given bond configuration can be written as : _ t)_k0_s s^2 n(s ) ( 1- f_s(k , t)^2 ) . [ eq16 ] from this equation , using eq.([eq : exp_simple ] ) , it is direct to see that the dynamical susceptibility goes from zero ( for @xmath8 ) to the mean cluster size @xmath75 ( in the @xmath59 limit ) , since the self isf of clusters of given size , in the sol phase , goes from @xmath76 ( for @xmath8 ) to zero ( in the @xmath59 limit ) . as in the previous section we can evaluate @xmath77 in the sol phase , @xmath48 . we find that , for time long enough , @xmath61 approaches the asymptotic value in the following way : @xmath78 where @xmath50 and @xmath79 . the exponent @xmath80 is exactly the same which appears in eq.([conto4 ] ) for the decay to zero of the self isf , the relaxation time in the stretched exponential function is given by @xmath81 , and finally the power law has a positive exponent @xmath82 different from the exponent @xmath83 which appears in eq.([conto4 ] ) . the calculations are shown in details in [ appendix2 ] . the self part of the van - hove function @xcite is given by : g_self(r , t)=. [ gself0 ] if the motion of particles is diffusive with a diffusion coefficient d , @xmath84 where @xmath85 is the distance traveled by a particle in a time @xmath64 . deviations from the gaussian distribution were observed in different glassy and gelling systems @xcite . in fact the van - hove function seems fitted by a gaussian only for short distances , instead , for long distances , it is well fitted by an exponential tail that extends to larger distances for increasing times . the deviation from the gaussian distribution indicates that some particles move faster than others , due to the presence of heterogeneities . in permanent gels , heterogeneities coincide with clusters of particles connected by bonds @xcite . as matter of fact particles belonging to different clusters have a different diffusion coefficient depending on the cluster size . as a consequence it has been suggested @xcite that , in the sol phase and in the diffusive regime ( i.e. in the long time limit ) , @xmath86 is given by a superposition of gaussians @xmath87 , \label{self - van - hove}\ ] ] where @xmath88 is the diffusion coefficient of clusters of size @xmath36 and @xmath34 is the cluster size distribution . by assuming @xmath89 , and by replacing the sum with the integral in eq.([self - van - hove ] ) , predictions can be given for the dependence of @xmath86 on @xmath85 and @xmath64 . we find , in the limit @xmath90 : @xmath91 , \label{eq17}\ ] ] where @xmath92 . the calculations are shown in details in [ appendix3 ] . the non - gaussian parameter is defined as @xcite : @xmath93 - 1,\ ] ] where @xmath94 is zero if the probability distribution of the particle displacements is gaussian . using eq.([self - van - hove ] ) , in permanent gels the non - gaussian parameter is expected to tend in the long time limit to a plateau , whose value is given by @xmath95 - 1= \left[\frac{\overline { d ^2 } -\overline{d } ^2}{\overline{d } ^2}\right ] , \label{fluttd}\ ] ] where , for each bond configuration , @xmath96 is the average over the cluster distribution . from this relation it appears clear that the deviation from gaussianity ( @xmath97 ) is due to the fluctuations of the diffusion coefficient , which in turns is related to the presence of dynamical heterogeneities , i.e. groups of particles with different diffusion coefficient . in this section we check the theoretical predictions obtained in sect . [ sect2 ] in the fene model for permanent gels . we consider a @xmath98 system of @xmath2 particles interacting with a soft potential given by weeks - chandler - andersen ( wca ) potential @xcite : @xmath99 , & r_{ij}<2^{1/6}\sigma \\ 0 , & r_{ij}\ge2^{1/6}\sigma \end{array } \right.\ ] ] where @xmath100 is the distance between the particles @xmath69 and @xmath70 . after the equilibration , at a given time @xmath8 particles distant less than @xmath101 are permanently linked by adding an attractive potential : @xmath102 , & r_{ij } < r_0\\ \infty , & r_{ij}\ge r_0 \end{array } \right.\ ] ] representing a finitely extendable nonlinear elastic ( fene ) . the fene potential was firstly introduced by warner@xcite and is widely used to study linear polymers @xcite . we choose @xmath103 and @xmath104 as usual@xcite in order to avoid any bond crossing and to use an integration time step @xmath105 not too small . we have performed molecular dynamics simulations of this model @xcite : the equations of motion were solved in the canonical ensemble ( with a nos - hoover thermostat ) using the velocity - verlet algorithm @xcite with a time step @xmath106 , where @xmath107 is the standard unit time for a lennard - jones fluid and @xmath108 is the mass of particle . we use reduced units where the unit length is @xmath109 , the unit energy is @xmath110 and the boltzmann constant @xmath111 is set equal to @xmath76 . we choose periodic boundary conditions , and average all the investigated quantities over @xmath112 independent configurations of the system . the temperature is fixed at @xmath113 and the volume fraction @xmath114 ( where @xmath31 is the linear size of the simulation box in units of @xmath109 ) is varied from @xmath115 to @xmath116 . using the percolation approach , we identify the gel phase as a state where a percolating cluster is present @xcite . with a finite size scaling analysis @xcite we obtain that this transition is in the universality class of random percolation . in particular , we obtain that the cluster size distribution , @xmath117 at the gelation threshold @xmath118 , with @xmath119 ; the mean cluster size @xmath120 , with @xmath121 ; the connectedness length @xmath122 , with @xmath123 ; and the fractal dimension of large clusters is @xmath124 . in the following we fix the number of particles , @xmath125 , where the threshold is @xmath126 . for clusters of size @xmath127 , for @xmath128 , @xmath29 . the curves are exponential functions , @xmath129 . * inset * : the relaxation time , @xmath130 , as function of the cluster size @xmath36 for @xmath29 and @xmath128 . the continuous curve is a power law @xmath131 with @xmath132.,width=264 ] in the sol phase we have studied the dynamical behaviour of the clusters as a function of the size @xmath36 . in particular we have measured the self isf and the mean squared displacement of clusters , respectively @xmath133 , and latexmath:[$\lan \delta r^2(s , t)\ran\equiv \frac{1}{s}\overline{\sum_{i\in c_s}\langle @xmath96 is the average of all clusters @xmath39 with fixed size @xmath36 . after an initial transient , we find that @xmath135 for @xmath29 is well fitted by exponential tail , @xmath129 ( fig.[fig2 ] ) , with @xmath136 ( inset of fig.[fig2 ] ) and @xmath137 not depending on the volume fraction . our data furnishes@xcite @xmath138 . the mean squared displacement of clusters ( shown in the main frame of fig . [ fig2_bis ] ) , after a ballistic regime at short time , displays a diffusive behaviour . the diffusion coefficient of clusters , @xmath88 , obtained as @xmath139 , is plotted in the inset of fig . [ fig2_bis ] as a function of size @xmath36 . in agreement with the results for the relaxation time , we find that @xmath140 , with @xmath141 . , for @xmath128 , and clusters of size @xmath127 . * inset * : the diffusion coefficient , @xmath88 , as function of the cluster size @xmath36 for @xmath128 . the continuous curve is a power law @xmath142 with @xmath143.,width=264 ] in the sol phase , due to the superposition of the contributions of different clusters , the self isf is expected to follow eq.([conto4 ] ) : @xmath144 where @xmath145 , @xmath146 , and @xmath52 , with @xmath147 . @xmath32 is plotted in fig . [ fig_self ] for @xmath48 . after the initial transient , the data are well fitted by the function , eq.([conto4 ] ) ( continuous curves in figures ) . furthermore , in agreement with theoretical predictions , the relaxation time @xmath148 ( plotted in the inset of fig.[fig_self ] as a function of @xmath0 ) appears to diverge approaching the transition threshold with the exponent @xmath149 . at the threshold @xmath150 finite size effects appear . interestingly the presence of dynamical heterogeneities , i.e. groups of particles with different diffusion coefficient , is related to the breakdown of stokes - einstein relation . as shown in [ appendix1 ] , the relaxation time @xmath148 is essentially the relaxation time of the critical cluster . on the contrary the diffusion coefficient of the system , obtained as @xmath151 , is given by the average over clusters with different sizes , @xmath152 . since @xmath153 , it is clear that @xmath154 is dominated by small clusters . as a consequence , although @xmath148 diverges at the threshold , @xmath154 does not go to zero at @xmath11 ( see inset of fig.[fig_self ] ) , due to the diffusion of small clusters through the gel matrix even for @xmath155 . for @xmath29 and @xmath115 , @xmath156 , @xmath157 , @xmath158 , @xmath159 , @xmath160 ( from left to right ) as a function of time @xmath64 . the lines are fitting curves : @xmath161 . * inset * : structural relaxation time , @xmath148 ( full circles ) , compared with the inverse of the diffusion coefficient @xmath162 ( full triangles ) as a function of the volume fraction . the full line is the fitting curve : @xmath163 , with @xmath164 and @xmath150 . , width=264 ] for @xmath29 and @xmath165 @xmath166,@xmath167,@xmath168,@xmath169 . , width=264 ] for @xmath115 , @xmath156 , @xmath157 , @xmath158 ( from bottom to top ) as a function of time @xmath64 . the lines are fitting curves : @xmath170.,width=264 ] in the gel phase a detailed analysis is not possible : @xmath32 ( plotted in fig.[fig_gel ] for @xmath54 ) displays a plateau , however at long times , due to finite size effects , it relaxes to zero following an exponential function . finally in the sol phase , we have also measured the dynamical susceptibility , @xmath61 , defined as the fluctuations of the self isf , given by eq.([chi4 ] ) . in fig.[fig_chi ] , @xmath61 is plotted for @xmath29 and different volume fractions . after the initial transient , the approach to the plateau is well fitted by eq.([conto6 ] ) : @xmath171 where @xmath172 and @xmath173 . note that the relaxation time @xmath174 coincides with @xmath81 only at low volume fraction ; near the threshold @xmath174 is instead lower than @xmath81 , due to the contribution of disconnected particles at intermediate times . and time @xmath175 ( from top to bottom ) . full lines are obtained from eq.([self - van - hove]).,width=264 ] in the sol phase we have also measured the self part of the van - hove function , @xmath86 , defined by eq.([gself0 ] ) . in the long time regime , @xmath86 is fitted by a gaussian curve only for short distances , and it seems well fitted by an exponential function for long distances @xcite . in the long time regime , clusters of any size show a diffusive behaviour ( see fig . [ fig2_bis ] ) , then we have suggested @xcite that @xmath86 is given by a superposition of gaussians , eq.([self - van - hove ] ) : @xmath87 , \label{eq24}\ ] ] where @xmath88 is the diffusion coefficient of clusters of size @xmath36 , and @xmath34 is the cluster size distribution . data well agree with our hypothesis , as indicated in fig.[fig : gr ] , where we have used @xmath34 and @xmath88 measured in the simulations . as shown in the main frame of fig . [ fig_fitg ] , @xmath176 plotted as a function of @xmath177 for fixed @xmath0 and for different times after the initial transient , collapse onto a single master curve , supporting the hypothesis that the data satisfy eq.([eq24 ] ) . finally the comparison with the approximate form obtained in [ appendix3 ] for @xmath178 , eq . ( [ eq17 ] ) , @xmath179 , \label{conto : fitg}\ ] ] ( with @xmath92 , @xmath13 and @xmath180 obtained from the simulations ) gives again a good agreement for high enough @xmath85 ( fig.[fig_fitg ] and inset ) . note that the curves obtained from eq.([conto : fitg ] ) and shown in figure are numerically indistinguishable from exponential functions in the considered range @xcite ( inset of fig.[fig_fitg ] ) . in agreement with the above picture , we expect that the non - gaussian parameter tends in the long time limit to a plateau , whose value coincides with the fluctuations of the diffusion coefficient , eq.([fluttd ] ) . in the main frame of fig.[alfa ] the non - gaussian parameter is plotted for different volume fractions , and in the inset of fig.[alfa ] the plateau value is compared with the fluctuations of the diffusion coefficient . the data are in good agreement with eq.([fluttd ] ) , confirming that in permanent gels the non - gaussianity of the displacement distribution is due to the superposition of the contributions of clusters of different sizes . it is worth to notice that the main contribution to @xmath181 comes from small finite clusters . in fact , the bigger the cluster , the lower its diffusion coefficient and hence its contribution to the non - gaussian parameter . therefore , no criticality of the plateau value of the non - gaussian parameter is observed approaching the transition threshold . as a function of @xmath177 for @xmath182 and @xmath183 . the full line is obtained from eq.([conto : fitg ] ) . * inset * : the self part of the van - hove distribution for @xmath182 and time @xmath175 ( from top to bottom ) . full lines are obtained from eq.([conto : fitg ] ) with the values of @xmath13 , @xmath137 , @xmath184 and @xmath180 measured in the simulations . note that the curves shown in figure are numerically indistinguishable from exponential functions.,width=264 ] , as a function of time @xmath64 for @xmath185 , @xmath156 , @xmath157 , @xmath159 ( from bottom to top ) . * inset * asymptotic value of @xmath181 ( empty squares ) compared with the fluctuations of the diffusion coefficient given by eq.([fluttd ] ) ( full circles ) . , width=264 ] in this paper we show how the complex dynamics , such as stretched exponentials and power law behaviors , observed experimentally and numerically in gelling systems , emerges from the contribution of single clusters , which instead decay with a simple exponential . furthermore , we establish a connection between this complex behaviour and critical exponents of percolation theory . we also find , in the diffusive regime , an asymptotic form ( for long enough distances ) of the self part of the van - hove function , which deviates from the gaussian distribution , and is numerically very similar to an exponential tail , usually observed in a large variety of complex system . our finding suggest therefore that such deviation from gaussianity may be due to a general mechanism , which may be ascribed to the presence of heterogeneities . the theoretical predictions are found in agreement with numerical results , that we find in the fene model for permanent gels . we suggest that a similar analysis can be extended to systems with finite lifetime bonds , as colloidal gels , glassy systems or spin glasses , where a `` suitable '' definition of clusters is necessary . in this section we show in details the calculations which gives the predictions shown in sect . [ sect3 ] for the time dependence of the self isf in the thermodynamic limit . starting from eq.([integral ] ) , in the hypothesis that @xmath186 , and @xmath187 , the self isf for a given bond configuration becomes : @xmath188 let us consider three different cases : ( i ) @xmath43 ; ( ii ) @xmath48 ; ( iii ) @xmath54 . for @xmath189 , @xmath34 can be written as @xmath190 @xcite , where @xmath13 is a cutoff value @xcite . then we obtain : @xmath191 where @xmath192 \(i ) at the gelation threshold , @xmath43 , @xmath193 , and the integral , eq.([sol ] ) , with @xmath194 given by eq.([ts ] ) , can be evaluated exactly : @xmath195 where @xmath196 is the @xmath46-function with @xmath47 . \(ii ) in the sol phase , @xmath48 , we are able to give only approximated predictions . the function @xmath194 , given by eq.([ts ] ) , has a maximum for @xmath197 such that @xcite @xmath198 let us approximate @xmath194 with @xmath199 , where @xmath200 if @xmath201 , @xmath202 = ( 2\pi\sigma^2)^{1/2}$ ] , and @xmath203\int ds~\exp\left[-\frac{(s-\tilde{s})^2}{2\sigma^2}\right ] \nonumber \\ \propto \frac{1}{{\tilde{s}}^{(\tau-2)}}\left(-(\tau-1)+x(x+1)\frac{at}{\tilde{s}^x } \right ) ^{-1/2 } \exp\left[-\left(\frac{\tilde{s}}{s^*}+\frac{at}{\tilde{s}^x}\right)\right ] . \label{conto3}\end{aligned}\ ] ] let us consider two limit cases : ( 1 ) @xmath204 ; ( 2 ) @xmath205 . \(1 ) using eq.([tilde ] ) , eq.([conto3 ] ) can be written in the following way : @xmath206 } { { \tilde{s}}^{(\tau-2 ) } } \left(x(\tau-1)+(x+1)\frac{\tilde{s}}{s^*}\right)^{-1/2},\end{aligned}\ ] ] which in the limit @xmath204 , where @xmath207 , gives again @xmath208 with @xmath47 , in agreement with previous calculations . \(2 ) using eq.([tilde ] ) , eq.([conto3 ] ) can be written in the following way : @xmath209 } { { \tilde{s}}^{(\tau-2 ) } } \left(-(\tau-1)+x(x+1)\frac{at}{\tilde{s}^x } \right)^{-1/2}.\end{aligned}\ ] ] in the limit @xmath210 , @xmath211 , and we obtain @xmath212 where @xmath50 , @xmath51 , and @xmath213 , which diverges at the threshold as power law with the exponent @xmath53 . \(iii ) finally in the gel phase , @xmath54 , @xmath18 in eq.([eq : a1 ] ) can be written as @xcite @xmath214 , where @xmath20 is fraction of particles belonging to the percolating cluster , @xmath21 is infinite in the thermodynamic limit , @xmath22 is the spatial dimension , and @xmath23 is a constant . then , from eq . ( [ eq : a1 ] ) , we obtain : @xmath215 where @xmath216 following the same arguments as in the sol phase , the second term in eq . ( [ gel ] ) is written as : @xmath217 \nonumber \\ & \cdot & \left(-\frac{(\tau-1)(d-1)}{d } + x\left(x+1-\frac{1}{d}\right)\frac{at}{\tilde{s}^x } \right)^{-1/2 } , \label{contogel_new}\end{aligned}\ ] ] where the maximum point , @xmath197 , of @xmath194 , eq . ( [ ts_gel ] ) , is obtained from @xmath218 in the limit @xmath210 , @xmath219 , and we obtain @xmath220 where @xmath221 , @xmath222 . for a finite system however @xmath21 is finite , and the behavior of @xmath32 is given by : @xmath223 where @xmath224 is the relaxation time of the percolating cluster . in the hypothesis of the previous section ( i.e. @xmath225 and @xmath226 ) eq.([eq16 ] ) becomes _ k0_4(k , t)=_s s^2 n(s ) ( 1-e^-2t/(s ) ) , which , for @xmath48 , by replacing the sum with the integral , gives : _ k0_4(k , t)s - ds ( w(s ) ) , [ integral2 ] where w(s)-(-2)s-- . [ ts2 ] note that the function @xmath227 has a maximum for @xmath228 such that @xmath229 let us approximate @xmath227 with @xmath230 , where @xmath231 if @xmath232 , @xmath233= ( 2\pi\sigma^2)^{1/2}$ ] , and @xmath234\int ds\exp\left(-\frac{(s-\overline{s})^2}{2\sigma^2}\right ) \nonumber \\ \nonumber \\ \propto\frac{1}{\overline{s}^{(\tau-3 ) } } \left((\tau-2)-x(x+1)\frac{2at}{\overline{s}^x}\right)^{-1/2 } \exp\left[-\left(\frac{\overline{s}}{s^*}+ \frac{2at}{\overline{s}^x}\right)\right ] . \label{conto5}\end{aligned}\ ] ] the limit @xmath235 , where @xmath236 , gives @xmath237 where @xmath50 and @xmath79 . the exponent @xmath80 is exactly the same which appears in eq.([conto4 ] ) for the decay to zero of the self isf . note that the relaxation time in the stretched exponential function is given by @xmath81 . in the sol phase , where after an initial transient the system is found in a diffusive regime , we assume the validity of eq.([self - van - hove ] ) for the self part of the van - hove function . by replacing the sum with the integral , and putting @xmath238 , @xmath86 can be written as g_self(r , t ) ds ( z(s ) ) , [ gself1 ] where z(s)(1-+)s-- . [ zs ] the condition for the first derivative of @xmath239 to be zero gives : @xmath240 this equation admits a solution @xmath241 only if @xmath242 . under this condition the solution of eq.([sm ] ) is a maximum , in fact @xmath243 is always true for @xmath244 ( i.e. @xmath245 ) @xcite . in this case we can approximate @xmath239 with @xmath246 , where @xmath247 . if @xmath248 , @xmath249= ( 2\pi\sigma^2)^{1/2},$ ] and we can write : @xmath250 \int ds~\exp\left[-\frac{(s - s_m)^2}{2\sigma^2}\right ] \nonumber \\ \propto s_m^{1+a } \exp\left[-\frac{s_m}{s^*}-\frac{r^2 s^x_m}{4at}\right ] \left(a+x(x-1)\frac{r^2 s^x_m}{4at}\right)^{-1/2}. \label{gself}\end{aligned}\ ] ] let us consider two limit cases : ( i ) @xmath251 , and ( ii ) @xmath90 . from eq.([sm ] ) we obtain : @xmath252 and eq.([gself ] ) becomes : @xmath253 & r^2\ll at \\ \left(\frac{4aat}{xr^2}\right)^{(a+1)/x } \exp \left[-\frac{a'}{s^*}\left ( \frac{4at}{r^2}\right)^{1/x}\right ] & r^2\gg at , \end{array } \right.\ ] ] where @xmath254 , and @xmath255 . note that the condition @xmath256 is satisfied in the limit @xmath257 where @xmath258 , which increases with increasing @xmath137 and/or approaching the gelation threshold . in the opposite limit , @xmath259 , @xmath260 , and hence the condition @xmath256 is not satisfied . in this case this approximation is expected not to hold and a development at the first order of the taylor series of @xmath239 around @xmath261 should be more appropriate ( see below ) . in the case @xmath262 ( as in the fene model for permanent gels presented in sect . [ sect3 ] ) , @xmath239 is a monotonic decreasing function of @xmath36 : = --<0 . let us develop @xmath239 at the first order of the taylor series around the maximum @xmath263 : z(s)z(s_m)+.|_s = s_m(s - s_m ) . [ zs1 ] by replacing eq.([zs1 ] ) in eq.([gself1 ] ) , we obtain @xmath264\nonumber\\ \propto \left(-a+\frac{1}{s^*}+\frac{xr^2}{4at}\right)^{-1 } \exp\left[-\frac{r^2}{4at}\right].\end{aligned}\ ] ] this approximation is expected to hold in the limit of long distances ( @xmath90 ) , where the second order of the taylor series around @xmath263 is much smaller than the first one , and only a small interval of values of @xmath36 around @xmath263 contribute to the integral in eq.([gself1 ] ) . the research is supported by cnr - infm parallel computing initiative , s.co.p.e . and l.r . 99 p. j. flory , _ the physics of polymer chemistry _ , cornell university press ( 1954 ) . p. g. de gennes , _ scaling concepts in polymer physics _ , cornell university press ( 1993 ) . d. stauffer , a. aharony , _ introduction to percolation theory _ , taylor & francis ( 1992 ) . d. stauffer , a. coniglio , m. adam , adv . sci . * 44 * , 103 ( 1982 ) . j. e. martin , j. wilcoxon , d. adolf phys . a * 36 * , 1803 ( 1987 ) ; j. e. martin , j. p. wilcoxon , phys . lett . * 61 * , 373 ( 1988 ) ; j. e. martin , j. p. wilcoxon , j. odinek , phys . a * 43 * , 858 ( 1991 ) . f. ikkai , m. shibayama phys . lett . * 82 * , 4946 ( 1999 ) . s. z. ren , c. m. sorensen phys . lett . * 70 * , 1727 ( 1993 ) . warner , ind . fundam . * 11 * , 379 ( 1972 ) k. kremer , g. s. grest , j. chem . phys . * 92 * , 5057 ( 1990 ) and * 94 * , 4103 ( 1991 ) ( erratum ) ; g. s. grest , k. kremer macromolecules * 20 * , 1376 ( 1987 ) ; g. s. grest , k. kremer phys . rev . a * 33 * , 3628 ( 1986 ) ; m. murat , g. s. grest phys . rev . lett * 63 * , 1074 ( 1989 ) . t. abete , a. de candia , e. del gado , a. fierro , and a. coniglio , phys . rev * 98 * , 088301 ( 2007 ) . m. doi , s. f. edwards , _ the theory of polymer dynamics _ , clarendon press ( 1986 ) . as the relaxation time @xmath184 is dominated by the longest time @xmath265 , we assume @xmath266 . using that @xcite @xmath267 , @xmath268 and @xmath269 , we obtain @xmath270 , which in @xmath98 random percolation gives @xmath271 . ogielski , phys . rev . b * 32 * , 7384 ( 1985 ) . e. del gado , a. fierro , l. de arcangelis and a. coniglio , phys . rev . e * 69 * , 051103 ( 2004 ) ; e. del gado , l. de arcangelis , and a. coniglio , eur . j. * 2 * , 359 ( 2000 ) . w. peng , h.e . castillo , p.m. goldbart , and a. zippelius , phys . b * 57 * , 839 ( 1998 ) ; p.m. goldbart , _ proc . conf . on rigidity theory and applications _ ( traverse city , mi , 14 - 17 june 1998 ) ed m.f . thorpe and p.m. duxbury ( new york : kluwer / plenum ) pp 95 - 124 ; p.m. goldbart , j. phys . : condens . matter * 12 * , 6585 ( 2000 ) . k. broderix , p. m. goldbart , and a. zippelius , phys . * 79 * , 3688 ( 1997 ) ; m. kntzel , h. lwe , p. mller and a. zippelius , eur . j. e * 12 * , 325 ( 2003 ) ; h. lwe , p. mller and a. zippelius , j. phys . : condens . matter * 17 * , s1659 ( 2005 ) . f. brouers and o. sontolongo - costa , europhys . * 62 * , 808 ( 2003 ) . hansen and i. r. mcdonald , _ theory of simple liquids _ , 2nd ed . ( academic , london , 1986 ) . d. a. stariolo , g. fabricius , j. chem . phys . * 125 * , 064505 ( 2006 ) ; p.i . hurtado , l. berthier , and w. kob , phys . lett . * 98 * , 135503 ( 2007 ) ; p. chaudhuri , l. berthier , and w. kob , phys . rev . lett . * 99 * , 060604 ( 2007 ) . t. abete , a. de candia , e. del gado , a. fierro , and a. coniglio , phys . e * 78 * , 041404 ( 2008 ) . a. rahman , phys . rev . * 136 * , a405 ( 1964 ) . j.d . weeks , d. chandler , and h.c . andersen , j. chem . phys * 54 * , 5237 ( 1971 ) . s. nos , j. chem . phys * 81 * , 511 ( 1984 ) ; w. g. hoover , phys . a * 31 * , 1695 ( 1985 ) ; m. p. allen , d. j. tildesley _ computer simulation of liquids _ oxford press ( 2000 ) . the function @xmath272 $ ] can be approximated by a taylor expansion around @xmath273 . we observe that for @xmath274 , the second order term of the expansion of @xmath275 is negligible respect the first order one , indicating that the curve in eq.([conto : fitg ] ) resembles an exponential function of @xmath85 in the asymptotic regime . for finite @xmath85 it is always possible to find a range , whose amplitude depends on @xmath276 and @xmath64 , in which the first order expansion represents a good approximation . it is worth to notice that the cutoff value @xmath13 depends on the wave vector @xmath277 . in fact @xmath278 , where @xmath279 is a function which tends to a constant for small @xmath280 , whereas it behaves as @xmath281 for large values of @xmath280 . actually we should consider @xmath285 , where @xmath286 is a distribution peaked around @xmath21 , which tends to @xmath287 for @xmath288 and @xmath289 . by writing @xmath290 in the following way : @xmath291 we easily see that , for @xmath244 , @xmath290 is always less than zero , for both @xmath292 and @xmath293 .
a theoretical and numerically study of dynamical properties in the sol - gel transition is presented . in particular , the complex phenomenology observed experimentally and numerically in gelling systems is reproduced in the framework of percolation theory , under simple assumptions on the relaxation of single clusters . by neglecting the correlation between particles belonging to different clusters , the quantities of interest ( such as the self intermediate scattering function , the dynamical susceptibility , the van - hove function , and the non - gaussian parameter ) are written as superposition of those due to single clusters . connection between these behaviours and the critical exponents of percolation are given . the theoretical predictions are checked in a model for permanent gels , where bonds between monomers are described by a fene potential . the data obtained in the numerical simulations are in good agreement with the analytical predictions .
mott transitions in multiband scenarios have recently received increased attention in the condensed matter community , partly owing to the interests in materials related to the iron - based superconducting ( fesc ) materials , whose parent compounds are believed to be in close proximity to mott transitions.@xcite indeed , while most parent compounds of the fescs are poor metals , insulating behaviors have been observed in@xcite ( tl , k)fe@xmath1se@xmath2 and in@xcite la@xmath2o@xmath2fe@xmath2o(se , s)@xmath2 . given the sizable magnetic moment of approximately 0.81.0 @xmath3 in the parent compounds of the `` 122 '' family fescs,@xcite a persistent debate in the field has been whether the magnetism is best described by an itinerant spin - density - wave ( sdw ) nesting@xcite or a local - moment ( lm)@xcite picture . theoretically , _ ab initio _ lsda calculation@xcite is able to obtain the experimentally observed magnetic ordering pattern in these compounds , but the magnetic moment is overestimated to be approximately 2.0 @xmath3 , suggesting that a weak - coupling sdw picture alone may not be adequate in describing these materials . with the above considerations , it is beneficial to consider how magnetism and fermi surface nesting affect the mott transition , and vice versa , in a multiband scenario . of particular interest is the question of whether there are any signs of local - moment physics on the metallic side of the phase diagram . however , while multiband mott transitions have previously been studied via dynamical mean - field theory @xcite and slave - spin mean - field,@xcite many of these studies have been focused on paramagnetic states . in contrast , in this paper we present a slave - rotor@xcite study of an orbital - symmetric multiband mott transition in which , utilizing a natural separation of scale between the orbital - symmetric coulomb repulsion and the hund s coupling , the sdw fermi surface nesting has been explicitly taken into account . we find that for a sizable range of hund s coupling , the existence of ( nearly ) nesting fermi surfaces causes the mott transition to acquire a strong first - order character , in which the ( staggered ) magnetization jumps across the phase boundary . however , as the strength of hund s coupling further increases , the first - order transition becomes weaker and an enhancement of magnetization can be observed on the metallic side . we interpret the former scenario as one in which a sharp distinction can be drawn between the itinerant sdw nesting and the lm picture , and the latter scenario as one in which signs of lm physics begin to develop in the metallic phase . it is worth noting that in the region of parameter space in which the staggered magnetization is enhanced , the renormalized hopping parameters also develop a discernible anisotropy between the @xmath4 and @xmath5 nearest - neighbor bonds , in which the antiferromagnetic @xmath4 direction shows a larger renormalized hopping than the ferromagnetic @xmath5 direction . such anisotropy is insignificant in other regions of the parameter space we considered . in the context of fesc , such anisotropy may be associated and is in agreement with the observed resistivity anisotropy in detwined ba(fe@xmath6co@xmath1)@xmath2as@xmath2 samples,@xcite in which the in - plane resistivity in the antiferromagnetic direction is found to be smaller than that in the ferromagnetic direction a result difficult to reconcile with a simple landau fermi liquid picture.@xcite we also remark that the renormalized bandwidth we obtained in this region of enhanced magnetization is about 2.55 times narrower than that of the corresponding noninteracting tight - binding model values that are close to , albeit larger , than those reported for some fesc materials.@xcite similarly , the staggered magnetization in such region is around 1.31.5 @xmath3 , close to but slightly larger than the values observed in the fesc materials in the `` 122 '' family . together with the observed resistivity anisotropy , it is tempting to suggest that some fesc materials in the `` 122 '' family are close to this region of enhanced magnetization that we associated with remnant of lm physics in the metallic phase . a multiband mott transition can be described by a multiband hubbard model , which can be written as the sum of a noninteracting hopping hamiltonian and an on - site interaction term : @xmath7 where @xmath8 label lattice sites , @xmath9 label orbitals , and @xmath10 label spins . note that the sum over repeated spin indices is assumed in the above equations and will be assumed throughout this paper . note also that we assume @xmath11 to be orbital independent , as would be case the when all orbitals are atomic . in that limit , it can also be shown@xcite that @xmath12 , which we shall assume henceforth and thus drop the subscripts . henceforth we shall also refer to the @xmath11 term as the coulomb interaction and the @xmath13 term as the hund s coupling , even though the latter includes the contributions from co - hopping and inter - orbital coulomb interactions . for concreteness we take @xmath14 to be the orbital - symmetric two - band hamiltonian introduced by ran _ _ in ref . , whose hopping parameters and fermi surface topology are shown in fig . [ fig : h_hop ] . such hamiltonian is believed to capture certain essential physics in fescs , and in such context the two orbitals can be understood as the @xmath15 and @xmath16 orbital of the fe site , with the @xmath17 axes oriented in the fe layer and at 45@xmath18 with the fe fe bond direction . as is evident from fig . [ fig : fs ] , at half - filling ( i.e. , two electrons per site ) there is strong fermi surface nesting in @xmath14 , with nesting vector @xmath0 and @xmath19 , consistent with the sdw ordering in the@xcite `` 1111 '' and@xcite `` 122 '' families fescs . however , as explained in ref . , for topological reasons the fermi surfaces are not fully gapped even when interactions are included ; hence there is a robust metallic sdw phase in such model . in the slave - rotor formalism,@xcite the hilbert space is enlarged by decomposing the electron operator @xmath20 into an @xmath21 rotor @xmath22 and a fermionic spinon @xmath23 , such that @xmath24 . note that in such decomposition there is only one rotor per lattice site and the spin and orbital indices are carried solely by @xmath25 . such economy in the slave - rotor formalism is made possible by the orbital - symmetric structure of the hubbard hamiltonian eq . [ eq : h_hb ] , particularly in the existence of a large orbital - symmetric coulomb interaction @xmath11 . in this enlarged hilbert space , the physical subspace is given by the constraint @xmath26 , where @xmath27 is the angular momentum operator conjugate to @xmath22 . plugging the slave - rotor decomposition into the hubbard hamiltonian eq . [ eq : h_hb ] , and applying a mean - field approximation to decouple the spinon @xmath25 and the rotor @xmath28 , we obtain : @xmath29 where @xmath30 are mean - field parameters introduced to enforce the constraint on average , @xmath31 , and @xmath32 , all of which are to be determined self - consistently . in this paper we consider only translationally invariant solutions in which the rotor unit cell is unenlarged while the spinon unit cell is enlarged to include two lattice sites per cell . moreover , since the underlying hamiltonian is time - reversal invariant , we consider only solutions for which both @xmath33 and @xmath34 are real . furthermore , since the system is at half - filling , the constraint is satisfied by setting @xmath35 . in this paper , the self - consistency for @xmath34 and @xmath33 are solved by repeated iterations . following florens and georges,@xcite a second mean - field approximation is applied to the rotor hamiltonian eq . [ eq : h_theta ] , in which the rotor variable @xmath36 is replaced by a complex bosonic field @xmath37 subjected to the constraint @xmath38 , enforced on average by a second mean - field parameter @xmath39 . with such approximation the rotor hamiltonian eq . [ eq : h_theta ] reduces to a system of coupled simple harmonic oscillators . the electron quasiparticle weight in this formulation is given by @xmath40 , where @xmath41 is the dispersion of the @xmath42 boson in @xmath43 space . the system is in the metallic phase when @xmath44 , which happens when the @xmath42 boson condenses at its energy minimum . for the spinon sector , we apply mean - field factorization to the four - fermion terms that appear in the spinon hamiltonian eq . [ eq : h_f ] . specifically , given the fermi surface nesting , we consider the following sdw mean - field hamiltonian : @xmath45 where @xmath46 denotes the @xmath4 coordinate of site @xmath47 , and @xmath48 is a hermitian matrix of mean - field parameters , to be determined for given @xmath34 by minimizing @xmath49 with respect to the mean - field state obtained from @xmath50 . note that since @xmath48 carries orbital indices , the possibility of an orbital symmetry broken sdw state is included in our mean - field ansatz . since the @xmath50 breaks the symmetry between the @xmath4 and @xmath5 direction in the underlying hamiltonian eq . [ eq : h_hb ] , the self - consistent mean - field parameters @xmath33 and @xmath51 can take different values along the @xmath4 and @xmath5 direction . we have also allowed for the possibility in which @xmath33 and @xmath51 take different values along the @xmath52 and @xmath53 directions , but found no manifestation of such cases in the parameter space we considered . it is well - noted that different parts of the electron two - body interactions are treated differently in our formulation , with the coulomb interaction @xmath11 handled via a `` strong - coupling '' approach while the hund s coupling @xmath13 handled via a `` weak - coupling '' approach . while this may seem unsatisfactory at first sight , it may nonetheless capture the qualitative aspects of the system when @xmath11 is the dominant scale , which seems to hold true for 3d transition metals@xcite and has been assumed in other theoretical calculations involving fescs@xcite . in this section we present the results of the mean - field calculations outlined in sec . [ sect : formulation ] for the half - filled orbital - symmetric two - band hubbard hamiltonian eq . [ eq : h_hb ] , without worrying about the validity of the approximations made . we shall return to the issue of validity in sec . [ sect : discussions ] . to simplify notations , we measure both @xmath13 and @xmath11 in units of @xmath54 ( for comparison , note that the full bandwidth of the noninteracting hamiltonian eq . [ eq : h_hop ] is @xmath55 ) and denote @xmath56 , such that @xmath57 . we shall refer to @xmath58 as the bond renormalization factor , @xmath34 as the renormalized hopping parameter , and the noninteracting part of @xmath59 in eq . [ eq : h_f ] as the renormalized hopping hamiltonian . we also denote the orbital - dependent staggered magnetization as @xmath60 , i.e. , @xmath61 ( here @xmath62 is the number of lattice sites ) , and decompose it as @xmath63 ) , with @xmath64 being the pauli sigma matrices . ( color online ) the phase diagram of the half - filled orbital - symmetric two - band hubbard hamiltonian eq . [ eq : h_hb ] , where @xmath13 and @xmath11 are given in units of @xmath54 ( for comparison , note that the full bandwidth of the noninteracting system is @xmath55 ) . here p stands for paramagnetic metal , sdw stands for spin density wave metal , lm stands for antiferromagnetic insulator , and sl stands for spin liquid . the insulating region is indicated by the gray shade , and first - order phase boundaries are indicated by the yellow shade . for an explanation of the `` nnn decoupling '' and `` nn decoupling '' lines , see fig . [ fig : decoupling ] and the main text . ] our major result is the phase diagram shown in fig . [ fig : phases ] , in which we vary @xmath13 and @xmath11 while keeping the bare hopping parameters @xmath65 fixed . to provide further insight into the phase diagram , we also plot in fig . [ fig : j_cuts ] the quasiparticle weight @xmath66 , the bond renormalization factor @xmath58 along the @xmath4 , @xmath5 , and diagonal ( @xmath67 ) direction , and the orbital - diagonal staggered magnetization @xmath68 as function of @xmath11 for four characteristic values of @xmath13 . it should be noted that within the parameter range we considered , the orbital - dependent staggered magnetization @xmath60 is always dominated by the orbital - diagonal component @xmath68 , in agreement with the weak - coupling results in ref . . in other words , no tendency towards an orbital - selective sdw state is found in our results . because of this , henceforth we shall consider exclusively @xmath68 , and refer to it simply as the staggered magnetization . as typified by fig . [ fig : j=0.4 ] , for @xmath69 we find that the system undergoes a second - order mott transition before developing any magnetism . the character of the mott transition is essentially the same as the one considered by florens and georges in ref . . as a result , for small values of @xmath13 part of the parameter space supports a non - magnetic insulating state , i.e. , a spin liquid . interestingly , the onset of magnetism for this parameter range seems to coincide with a phase boundary ( labeled as `` nn decoupling '' in fig . [ fig : phases ] ) in which the nearest - neighbor bonds renormalize to zero . as @xmath13 increases to @xmath70 , the onset of magnetism begins to shift to the left of the mott transition . moreover , the first - order boundary in which the nearest - neighbor bonds renormalize to zero starts to coincide with the mott transition . the combined transition retains a first - order character , and the staggered magnetization exhibits a discontinuity across the transition . furthermore , another phase boundary in which the _ next_-nearest - neighbor bonds also renormalize to zero ( labeled as `` nnn decoupling '' in fig . [ fig : phases ] ) can now be seen , whose critical value @xmath71 decreases with increasing @xmath13 . note that the phase boundary of this next - nearest - neighbor decoupling traces well with the onset of a fully polarized insulating state . this situation is typified in fig . [ fig : j=0.8 ] . for clarity , we also illustrate the real - space picture of the nearest - neighbor and next - nearest - neighbor decoupled state in fig . [ fig : decoupling ] . as @xmath13 further increases to @xmath72 , the mott transition becomes even stronger in first - order character and starts to coincide with the transition in which the next - nearest - neighbor bonds decouple . the phase transition is now accompanied by a jump to full polarization across the phase boundary . this situation is typified in fig . [ fig : j=1.2 ] . however , as @xmath13 further increases to @xmath73 , the first - order character of the mott transition starts to soften up , and , as typified by fig . [ fig : j=2.2 ] , a small shoulder in which the staggered magnetization is enhanced can now be observed immediately to the left of the mott transition . moreover , as can be seen in the inset of fig . [ fig : j=2.2 ] , an anisotropy of bond renormalization factors between the @xmath4 and @xmath5 nearest - neighbor bonds has also become discernible in this region where staggered magnetization is enhanced . in contrast , such anisotropy remains insignificant in all other parts of the parameter space we considered , including regions in the metallic phase in which the staggered magnetization is not enhanced . note also that the bond renormalization factor is larger in the antiferromagnetic @xmath4 direction than in the ferromagnetic @xmath5 direction . since the results in sec . [ sect : results ] rely on various mean - field approximations , one should be cautious about the results thus obtained . in particular : 1 . the critical value of @xmath11 for the mott transition , @xmath74 , obtained in the slave - rotor approach is smaller than that obtained in the slave - spin approach.@xcite this is a known discrepancy between the two methods and should vanish in the large-@xmath75 ( number of orbitals ) limit . for the case where @xmath75 is finite the slave - spin estimate is believed to be more accurate.@xcite this , however , should not affect the qualitative statements made in this paper . 2 . in the slave - rotor formalism , all bonds that connect between the same sites are renormalized by the same factor , regardless of orbital character . this may lead to worries that some orbital selective aspects@xcite of the system are neglected . such worries are partially alleviated by the study@xcite of a similar two - band system in the slave - spin formalism , which allows for the possibility of spin - and - orbital dependent bond renormalizations and finds no evidence for an orbital selective state for @xmath76 . furthermore , we reiterate that the possibility of an orbital symmetry broken sdw state is included in our formulation via an orbital - dependent mean - field parameter @xmath48 , but that we always find the staggered magnetization of the resulting state to be dominated by the orbital - diagonal component . therefore , while orbital - selective scenarios have been proposed in the context of fescs,@xcite we have found no supporting evidence for such state in the two - band model we considered . 3 . as explained in sec . [ sect : formulation ] , the distinct treatment for the coulomb interaction and hund s coupling is justified by the assumption that @xmath11 is the dominant scale . generally , spin fluctuation is believed to play a more important role when the inter - orbital coulomb interaction @xmath77 is comparable to @xmath13 , i.e. , when @xmath78 . since @xmath74 decreases rapidly as @xmath13 increases , the phase transition starts to enter the region where @xmath78 around @xmath79 . therefore , the upper part of the phase diagram in fig . [ fig : phases ] may be stretching the limit of validity for the present approach . 4 . by treating @xmath13 via a weak - coupling approach , we have neglected the strong - coupling aspect of the hund s coupling . in particular , one should not expect the @xmath74 to remain unchanged for small @xmath13 as in fig . [ fig : phases ] . instead , one should expect @xmath74 to drop immediately as @xmath13 increase , as is found@xcite in the slave - spin approach . 5 . in similar spirit , the present approximation has neglected the weak - coupling aspects of @xmath11 . in particular , at the perturbative level both @xmath11 and @xmath13 are expected to drive the fermi surface nesting at @xmath0 . hence , the paramagnetic metal and the spin - liquid region in fig . [ fig : phases ] should , if exist at all , be greatly reduced in size . . one peculiar feature of the mean - field result presented in sec . [ sect : results ] is the existence of phases in which certain bonds in the hopping hamiltonian is renormalized to zero in the insulating phase . this seems to be a general feature of the slave - rotor method , since similar results is also found in a hubbard model with a different lattice geometry ( and which the hund s coupling is dropped).@xcite indeed , the problem can already be seen in the simplest case of a square - lattice hubbard model with non - mixing bands having only nearest - neighbor hoppings . in such case it can be checked that on the insulating side , within the slave - rotor mean field , a symmetry broken phase in which all bonds along a particular direction is renormalized to zero is energetically favorable to a state in which all bonds are renormalized by the same amount . this rotation - symmetry broken state is , in turn , less energetically favorable to a translation - and - rotation - symmetry broken state in which the bonds renormalize to a dimer pattern ( see fig . [ fig : sq ] for illustration ) . since the dimer pattern is known to be favorable in the large-@xmath75 limit,@xcite the decoupled phase in fig . [ fig : phases ] may be an artifact of the large-@xmath75 approximation . in spite of the issues mentioned above , the present study may still shed light on the question we posed in the introduction of this paper : namely , whether there are any signs of local - moment physics on the metallic side of the phase diagram . for this , observe that @xmath74 remains in the region where @xmath80 for @xmath13 in the range @xmath81 , and as we approach the phase boundary from @xmath82 the bond renormalization remains modest for much of the parameter space , such that the renormalized bandwidth @xmath83 is at least comparable to @xmath13 . in such region the approximation we took in this paper may be valid . and for @xmath84 , as we approach the phase boundary from @xmath82 there is no sign of magnetization enhancement ( which , in this formulation , will be accompanied by a sharper - than - usual drop in @xmath58 ) until the first - order phase boundary is hit . thus , our result is suggestive that in this range of @xmath13 there is no sign of local - moment physics on the metallic side . in contrast , for @xmath85 there is a sharper drop in @xmath58 as we approach the phase boundary , which suggests the possibility that in this range of @xmath13 some remnant of local - moment physics can be found . we thank fa wang , ying ran , and senthil todadri for helpful discussions . this research is supported in part by doe under grant no . de - fg02 - 03er46076 ( whk and pal ) and in part by nsf under grant no . phy05 - 51164 ( whk ) .
motivated by the debate of spin - density - wave ( sdw ) versus local - moment ( lm ) picture in the iron - based superconducting ( fesc ) materials , we consider a two - band orbital - symmetric hubbard model in which there is robust fermi surface nesting at @xmath0 . we obtain the phase diagram of such system by a mean - field slave - rotor approach , in which the fermi surface nesting and the sdw order are explicitly taken into account via a natural separation of scale between the hund s coupling and the coulomb interaction . we find that for a sizable range of hund s coupling the mott transition acquires a strong first - order character , but there also exists a small range of stronger hund s coupling in which an enhancement of magnetization can be observed on the sdw side . we interpret the former scenario as one in which a sharp distinction can be drawn between lm and the sdw picture , and the latter scenario as one in which signs of lm physics begin to develop in the metallic phase . it is tempting to suggest that some fesc materials are in the vicinity of the latter scenario .
gamma ray bursts ( grbs ) are thought to be driven by mass accretion onto compact objects @xcite . the enabling cooling mechanism allowing accretion is neutrino emission , raising the usual eddington rate for photons by sixteen orders of magnitude @xcite . this applies both to long and short grbs , perhaps due to the collapse of massive rotating stars @xcite and compact binary mergers @xcite , respectively , or magnetized neutron stars @xcite . the prompt gamma - ray emission originates at @xmath0 cm from this source , possibly from internal shocks in the relativistic outflow generated by the engine @xcite . the engine itself is hidden from view due to the high opacities , and can be potentially probed directly only through gravitational waves and neutrinos . the observed time series in grb prompt emission show diversity between events ( see , e.g. , * ? ? ? * ) : some have a single peak , others multiple emission episodes , with correlations between the fluence of the active period and the length of the quiescent interval preceding it @xcite . on top of this , rapid ( ms ) fluctuations are routinely observed . fourier analysis of the high - energy light curves of the prompt emission in grbs in the source frame reveals power - law spectra @xcite , with index @xmath1 and a break at @xmath2 hz . variability is likely due to a combination of several effects , allowing in principle an additional way to probe the central engine indirectly . some are probably intrinsic to the progenitor : the distribution of angular momentum with radius inside the star in the case of a collapsar may lead to distinct episodes of energy release @xcite ; the fall back at late times of material stripped from a tidally disrupted neutron star is capable of powering secondary accretion episodes @xcite ; hydrodynamical or magnetic instabilities in the accretion disk may result in intermittent accretion @xcite . others can come from the relativistic outflow : the interaction of a jet with high lorentz factor with the stellar envelope before breakout can lead to irregularities and shocking @xcite ; the outcome of internal shocks between shells in the flow depends on the variation in mass and energy upon ejection @xcite . an additional factor , upon which we focus here , is related to the variability present in the accretion disk as a result of turbulent motions . the dissipation is related to the local hydrodynamical variables , and as these vary in time , so will the energy output . the magnetorotational instability ( mri ) @xcite is a possible mechanism that will allow for the transport of angular momentum in accretion disks with differential rotation . its behavior under the physical conditions in neutrino cooled disks , which are similar to those occurring in supernovae , but different from the usual ones present in x - ray binaries and agn , has come under scrutiny more recently @xcite . one such difference lies in the sensitivity of the cooling rate to temperature and is at the heart of this work . whereas , for example , the photon bremsstrahlung emissivity scales as @xmath3 in the optically thin limit , for neutrinos @xmath4 , where @xmath5 depending on the cooling process . further , while photon - cooled disks are typically optically thick , @xmath6 , for a wide range of relevant parameters their neutrino - cooled counterparts are optically thin , @xmath7 , tightly coupling the local conditions to the emitted luminosity . in this _ letter _ , we characterize the local variability in neutrino cooled disks through shearing box mhd simulations ( [ sec : shear ] ) , and then use the results of global disk simulations to scale the results for the central engine as a whole ( [ sec : global ] ) . prospects for placing constraints on grbs are discussed in [ sec : prospects ] . the local flow in the disk is modeled by the shearing box approximation @xcite : a rectangular cartesian coordinate system represents a local neighborhood inside the disk , at an arbitrary orbital radius @xmath8 with dimensions which are much smaller than @xmath8 . the @xmath9 , @xmath10 and @xmath11 axes represent the radial , azimuthal and vertical directions in the disk , respectively . the radial component of the central object s gravity is included , and differential rotation is replaced by a keplerian shear flow along the @xmath10 direction . periodic boundary conditions are set in @xmath10 and @xmath11 , while in @xmath9 these are `` shearing periodic '' : upon crossing a radial boundary , a fluid element is displaced along @xmath10 by an amount given by the shear value . the eulerian zeus-3d code [ @xcite ] is used to solve the equations of ideal mhd . these are : @xmath12 @xmath13 @xmath14 @xmath15 and an ideal gas equation of state with @xmath16 where @xmath17 , @xmath18 and @xmath19 are the gas velocity , density and pressure , respectively , @xmath20 is the magnetic field , @xmath21 is the disk angular frequency , @xmath22 is the internal energy density , and @xmath23 is a unitary radial vector . the second term on the right - hand side of eq.([eq : energy ] ) is added to model neutrino cooling processes , with the nature of the particular emission mechanism fixing the value of @xmath24 . in grb central engines , the dominant processes are @xmath25 pair annihilation @xcite , with @xmath26 , and @xmath25 capture onto free neutrons and protons @xcite , with @xmath27 . a fraction @xmath28 of the initial internal energy of the fluid is removed over 100 orbital periods at @xmath8 , the duration of the simulation . we used @xmath29 , thus the cooling time is much longer than the dynamical time . initially the density , internal energy and pressure in the box are uniform , and the magnetic field is vertical with non - zero net magnetic flux across the box . the ratio of gas to magnetic pressure is @xmath30 . the velocity field is randomly perturbed with amplitude @xmath31 , where @xmath32 is the sound speed . we have performed simulations at varying resolution to test for convergence ( the highest having 128x256x128 zones in @xmath33 , respectively ) . from the power spectrum of the velocity field @xmath34 , we have verified that our results are converged and unaffected by resolution changes at the level reported . turbulent motions induced by the mri develop in the simulation on a time scale @xmath35 , leading to turbulent transport of angular momentum . the fundamental output of the simulations is the luminosity as a function of time of the shearing box , which we denote as @xmath36 . , for shearing box simulations with cooling mimicking pair capture ( blue , @xmath37 ) and pair annihilation ( red , @xmath38 ) . the time is given in units of the orbital period at @xmath8 , the fiducial shearing box radial position . right : fourier transforms @xmath39 of @xmath36 . the reference power law has index -2 . ] luminosity variations for @xmath36 are shown in figure [ fig : luminosity ] for @xmath40 . rapid variability is present , reflecting the turbulent motions in the box . these changes can be characterized by their fourier transform , @xmath39 , where @xmath41 is the signal frequency ( figure [ fig : luminosity ] ) . roughly , the spectrum is a power law , @xmath42 , over a wide range of frequencies in the case of @xmath37 , with more complicated behavior superimposed in the case of @xmath38 . it is clear that the variations in energy release have a characteristic behavior , with the details varying according to the specific cooling mechanism . to find the global variability we require the superposition of a very large number of smaller regions represented by the shearing box calculations , and approximations are necessary to proceed further . we assume equatorial and azimuthal symmetry , and also take the disk to be in a stationary state . obtaining global quantities thus requires integrating over the polar angle @xmath43 $ ] and radius @xmath44 $ ] . we are still faced with the superposition of a potentially large number of zones , each uncorrelated with the other , and which could destroy any coherent signature of the processes occuring on small scales . a natural quantity in the disk , however , enforces causality for a given frequency : the local sound speed , @xmath32 . consider a disk annulus at radius @xmath45 , with sound speed @xmath46 . the correlation length over which signals with frequency @xmath41 propagate is @xmath47 . the azimuthal velocity is @xmath48 , and from vertical hydrostatic balance , the pressure scale height @xmath49 satisfies @xmath50 . thus @xmath51 , with @xmath52 being the orbital frequency at radius @xmath45 . in neutrino cooled accretion disks , @xmath53 cst @xcite , so that @xmath54 . the number of distinct azimuthal zones within the ring for frequency @xmath41 is @xmath55 . dynamical simulations of such disks show that @xmath56 . if @xmath57 , the signal at frequency @xmath41 is thus diminished with respect to the idealized case of entirely in - phase contributions by a factor @xmath58 , due to the random nature of the phase de - correlation . if , on the contrary , @xmath59 , the signal at frequency @xmath41 is added constructively . this acts as a broad low - pass filter by slightly suppressing the signal at high frequencies , when @xmath60 . carrying out this filtering procedure also requires scaling the fiducial model to different radii , which we now address . + the emissivity is not a function of azimuthal position , but it is one of radius . given the extreme sensitivity of the cooling rate to temperature , only the innermost regions of the disk contribute to the luminosity , and thus to its variability properties . the argument given above for the decoupling of azimuthal zones in the disk for a particular frequency is also applicable to the radial coordinate . advection may modify this slightly , but for our present purpose it is valid enough to say that as the flow is split up into @xmath61 zones in azimuth , it is divided into rings of characteristic size @xmath62 . the number of radial zones in the disk at frequency @xmath41 is @xmath63 , where @xmath64 is the outer radius of the disk . using the previous results , @xmath65 . a second low - pass filter needs to be considered in analogy with the azimuthal one , and the signal is now suppressed by a factor @xmath66 if @xmath67 . ) neutrino emissivities , @xmath68 , @xmath69 , for two dimensional , azimuthally symmetric global dynamical simulations of accretion disks around a black hole with @xmath70 . the blue ( red ) line corresponds to simulation @xmath71 ( @xmath72 ) , where only pair captures ( annihilation ) were used to compute the cooling rate . the solid black line is the result of a simulation including both processes . ] in order to carry out this procedure , we have used the results of global , 2d simulations of neutrino - cooled accretion disks @xcite orbiting a stellar mass black hole , evolved in the pseudo - relativistic potential of @xcite . for consistency purposes , we have re - computed their evolution for this work for two test cases , with @xmath73 and @xmath74 . in the first model , @xmath71 , we have eliminated all sources of cooling except from pair captures , where @xmath75 , while in the second , @xmath72 , we have only kept that due to pair annihilation , with @xmath76 . each model can then be more faithfully compared with the two fiducial runs for the shearing box with @xmath40 respectively , using a near - stationary state where the accretion rate is @xmath77 m@xmath78 s@xmath79 . we first compute the equatorial radial profile of emissivity in simulations @xmath71 and @xmath72 , @xmath80 , shown in figure [ fig : qdotr ] . at small radii the dominant contribution comes from pair capture , while annihilation dominates for @xmath81 . the combination is a strongly decreasing function of @xmath45 , so the radial convolution need not be extended to infinity , and pair capture dominates the total emission , as well as its general behavior . the inner radius of the disk is @xmath82 , where the effects of general relativity truncate the flow . we have used @xmath83 , and checked that the final result is insensitive to this choice , as long as @xmath84 . to perform the radial integral , the time series @xmath36 and @xmath39 need to be scaled to the disk at different radii @xmath45 . the characteristic break frequency in the fiducial spectrum for the volume - averaged total ( maxwell and reynolds ) stress , measured as an equivalent @xmath85 viscosity parameter , is roughly at the orbital frequency , @xmath52 . this is related to the nature of the mri , where , @xmath86 . the orbital angular frequency is known as a function of radius , @xmath87 , so we can convert from frequency to radius , @xmath39 to @xmath88 and integrate over different annuli ( the expression for @xmath89 is appropriate for the pseudo - relativistic potential used ) . , for models @xmath71 ( red ) and @xmath72 ( blue ) . the power is given in arbitrary units as a function of frequency ( in hz ) for a black hole with @xmath90 . the black line shows the result of a test calculation with local cooling in the shearing box simulation set to @xmath91 , i.e. , @xmath92 . the error bars for @xmath40 include the scatter from the fourier transform in the plotted frequency bins and sampling errors in the radial discretization of the convolution . ] the power associated to the variability in the neutrino luminosity can now be computed as : @xmath93 where @xmath80 corresponds to model @xmath71 or @xmath72 , and @xmath94 and @xmath95 account for the filtering at high frequencies . figure [ fig : pds ] shows the resulting spectra , where several features clearly stand out . at low frequencies , @xmath960.1 hz , the spectrum is the same in both cases , @xmath97 . for @xmath98 the results diverge : model @xmath71 approximately maintains the power law decay , while model @xmath72 shows an excess , and in general more complex behavior , with broad features at @xmath99 and 40 hz overlaid on a decay with lower index , @xmath100 . at @xmath101 100 hz , a break to a more rapidly decaying power law terminates both spectra . the errors are greater for model @xmath72 , due to greater scatter in the fourier transform . the thin black line in figure [ fig : pds ] is from a shearing box test calculation with @xmath91 , i.e. , with @xmath92 ( the smaller error bars are not plotted for clarity ) . as the sensitivity of the cooling term to the temperature rises , the trend is for an excess in power at higher frequencies , and a more moderate background power law decay . through shearing box mhd simulations we have characterized the time variability of the energy release at small scales in a neutrino - cooled accretion disk around a black hole , where energy losses primordially come from @xmath25 pair capture onto free nucleons and protons and pair annihilation . with the use of cooling profiles from large scale , two - dimensional simulations of full disks , we have convolved this local variability to obtain a global signature of time variations in the power output , through the power density spectrum of the neutrino luminosity . since accretion is enabled by the cooling through neutrinos , we take this as an indicator of central engine variability which will be reflected in the relativistic outflow eventually giving rise to a grb . the power spectrum exhibits characteristic features related to the general nature of the neutrino cooling in the optically thin regime , and particularly to its temperature dependence . a background power law decay , with index @xmath102 extends approximately from 0.1 - 100 hz . for emissivities @xmath103 , additional power appears at high frequencies as @xmath24 rises , with the particular values of the slopes and cuttoffs scaling with the cooling mechanism ( fig.[fig : pds ] ) . this result thus provides in principle a discriminant for grb central engines powered by neutrino - cooled accretion flows , and illustrates how any local mechanism can be used in the same way to test its viability . the reason for the difference when modifying the cooling mechanism is fundamentally related to the local energy balance . for @xmath104 ( model @xmath105 and a given internal energy supply , @xmath22 , the local cooling time @xmath106 is shorter under a temperature perturbation @xmath107 than when @xmath75 ( model @xmath71 ) . the power is thus higher at such frequencies , leading to the observed displacement in fig . [ fig : pds ] . the argument holds when comparing model @xmath71 with the test run at @xmath92 , and gives a way to characterize the accretion flow and discriminate between competing mechanisms , if the variations in the neutrino luminosity are directly reflected in the accretion power output which drives a relativistic flow . two possible scenarios in which the difference between the cooling regimes studied here may be of relevance are worthy of note . first , the mass of the black hole introduces a scaling into the problem when combined with neutrino cooling . if @xmath108 is too large , the density and temperature in the accretion flow can be too low for neutrino cooling to operate . as the bh mass is reduced , pair annihilation , with @xmath109 first becomes effective as an energy sink , followed by pair captures , with @xmath110 . a signature of the bh mass is thus in principle available in the variability of the flow . the second case is related to the time evolution of the flow , assuming a certain amount of mass @xmath111 is initially available for accretion , and no further feeding of the central engine takes place . global disk simulations show that the density and temperature drop as the disk drains into the bh on the viscous time scale . given enough time , the whole disk will lie on the branch cooled by pair annihilation where @xmath85 particles have formed , below @xmath112\simeq 6.5 $ ] and @xmath113\simeq 10 $ ] for the adopted black hole mass ( see also figure 2 in @xcite ) . thus the variability may initially behave as in case @xmath71 , and end as in case @xmath72 ( the luminosity will have decreased substantially by then , along with the accretion rate ) . the associated grb , if one occurs , need not necessarily be powered by neutrinos in order for these effects to be apparent . the fact that neutrinos are responsible for the cooling process allowing accretion makes them relevant in this context . a number of limitations apply to this study , and can be matters for further study . first , the scaling we have used to infer the variability at different radii is strictly valid if the flow is adiabatic . explicit cooling thus violates this assumption . however , given the choice of @xmath28 and @xmath114 the associated time scales are such that @xmath115 , making this a reasonable approximation . second , we have used the neutrino luminosity as a proxy for the manifestation of central engine activity , which could have a neutrino component , but need not be restricted to it . for example , magnetic fields may power relativistic outflows leading to the observed high - energy emission . in this sense , an alternative characterization based on the mass accretion rate @xmath116 through the disk may be of use as well , which can then be associated to the energy output of the relativistic outflow as @xmath117 . third , this is only the first filter any variability originating in a disk needs to go through before emerging as high - energy photons . a variable mass accretion rate and energy conversion efficiency , propagation through the stellar envelope ( for a massive star progenitor ) , external shocks , and general relativistic effects @xcite , among others , can each potentially leave their own fingerprint on the power spectrum . the background upon which they will do so , however , must ultimately be related to the flow in the disk itself if that is what is driving the energy release , and is what we focus on in this work . finally , other mechanisms can potentially be responsible for angular momentum transport and variability in the disk itself , such as thermal , viscous or gravitational instabilities , none of which we have considered here . balbus , s. a. , hawley , j. f. 1998 , rev . phys . , 70 , 1 beloborodov , a.m. , stern , b.e . , svensson , r. 1998 , , 508 , l25 beloborodov , a.m. , stern , b.e . , svensson , r. 2000 , , 535 , 158 beloborodov , a.m. , 2003 , , 588 , 931 birkl , r. aloy , m.a . , janka , h .- th . , mller , e. 2007 , 463 , 51 bosnjak , z. , daigne , f. , dubus , g. 2009 , , 498 , 677 chen , w.x . , beloborodov a.m. 2007 , , 657 , 383 di matteo , t. , perna , r. , & narayan , r.2002 , , 579 , 706 eichler , livio , m. , piran , t. , schramm , d.h . 1989 , nature , 340 , 126 gehrels , n. , ramirez - ruiz , e. & fox , d.b . 2009 , , 47 , 567 hawley , j.f . , gammie , c.f . , balbus , s.a . , 1995 , apj , 440 , 742 itoh , n. , hayashi , h. , nishikawa , a. , kohyama , y. 1996 , apjs , 102 , 411 kohri , k. , mineshige , s. 2002 , , 577 , 311 kumar , p. , nayaran , r. , johsnon , j.l . 2008 , , 388 , 1729 langanke , k. , martinez - pinedo , g. 2001 , at . data nucl . data tables , 79 , 1 lee , w. h. , ramirez - ruiz , & e. , page , d. 2005 , , 632 , 421 lee , w.h . , ramirez - ruiz , e. 2007 , new j. phys . , 9 , 17 lee , w.h . , ramirez - ruiz , e. , lopez - camara , d. 2009 , , 699 , l93 lopez - camara , d. , lee , w.h . , ramirez - ruiz , e. 2010 , , 716 , 1308 macfadyen , a. i. , & woosley , s. e. 1999 , , 524 , 262 masada , y. , kawanaka , n. , sano , t. , shibata , k. 2007 , , 663 , 437 mendoza , s. , hidalgo , j.c . , olvera , d. , cabrera , j.i . 2009 , , 395 , 1403 morsony , b.j . , lazzati , d. , begelman , m.c . 2010 , apj submitted , arxiv1002.0361 nakar , e. 2007 , phys . , 442 , 166 narayan , r. , paczyski , b. , piran , t. 1992 , , 395 , l83 narayan , r. , piran , t. , & kumar , p. 2001 , , 557 , 949 norris , j.p . , nemiroff , r.j . , bonnell , j.t . , scargle , j.d . , kouveliotou , c. , paciesas , w.s . , meegan , c.a . , fishman , g.j . 1996 , , 459 , 393 obergaulinger , m. , cerd durn , p. mller , e. , aloy , m.a . 2009 , , 498 , 241 paczyski , b. 1991 , acta astron . 41 , 257 pacznyski , b. , & wiita , p. j. 1980 , , 88 , 23 panaitescu , a. , spada , m. , mszros , p. 1999 , , 522 , l105 perna , r. , armitage , p.j . , zhang , b. 2006 , , 636 , l29 perna , r. , macfadyen , a.i . 2010 , , 710 , l103 popham , r. , woosley , s. e. , & fryer , c. 1999 , , 518 , 356 proga , d. , zhang , b. 2006 , , 370 , l61 ramirez - ruiz , e. , merloni , a. 2001 , , 320 , l25 ramirez - ruiz , e. , merloni , a. , rees , m.j . 2001 , , 324 , 1147 rossi , e.m . , armitage , p.j . , menou , k. 2008 , , 391 , 922 rosswog , s. 2007 , , 376 , l48 ryde , f. borgonovo , l. , larsson , s. , lund , n. , von kienlin , a. , lichti , g. 2003 , , 411 , l331 setiawan , s. , ruffert , m. , janka , h.th . , 2004 , , 352 , 753 stone , j. m. , & norman , m. l. 1992a , apjs,80,753 stone , j. m. , & norman , m. l. 1992b , apjs,80,791 taylor , p.a . , miller , j.c . , podsiadlowski , p. 2010 , mnras submitted , arxiv1006.4624 thompson , t.a . , quataert , e. , burrows , a. 2005 , , 620 , 861 usov , v.v . 1992 , nature , 357 , 472 woosley , s.e . 1993 , , 405 , 273 woosley , s.e . , bloom , j.s . 2006 , ara&a , 44 , 507 zhang , b. , mszros , p. 2004 , ijmp a , 19 , 2385
the central engine of gamma ray bursts is hidden from direct probing with photons mainly due to the high densities involved . inferences on their properties are thus made from their cosmological setting , energetics , low - energy counterparts and variability . if grbs are powered by hypercritical accretion onto compact objects , on small spatial scales the flow will exhibit fluctuations , which could in principle be reflected in the power output of the central engine and ultimately in the high energy prompt emission . here we address this issue by characterizing the variability in neutrino cooled accretion flows through local shearing box simulations with magnetic fields , and then convolving them on a global scale with large scale dynamical simulations of accretion disks . the resulting signature is characteristic , and sensitive to the details of the cooling mechanism , providing in principle a discriminant for grb central engine properties .
thermodynamic quantities such as the specific heat become non - analytic at a critical point associated with a second order phase transition . for a finite ( or confined ) system with a finite dimension such as a film characterized by a length @xmath0 , close enough to the critical point such that the correlation length becomes comparable or larger than @xmath0 , such thermodynamic quantities are significantly altered ; the reason is that the degrees of freedom of the system are correlated to each other over the entire system . examples of such confined systems are ( a ) a film of thickness @xmath0 where the system is confined in one spatial dimension , ( b ) a bar - like geometry with cross - section @xmath1 and infinite length ( such a pore or a wire ) where the system is confined in its two spatial dimensions or ( c ) a cubic geometry of size @xmath2 where the system is finite in all three dimensions . for any thermodymamic observable we can define a system specific dimensionless quantity , a scaling function@xcite ; for example , in the case of the specific heat near the critical point and for sufficiently large @xmath0 we may define the following scaling function : @xmath3 where @xmath4 is the reduced temperature and @xmath5 is the specific heat for the case of the system confined within a finite length @xmath0 . for a given case of confining geometry and given the condition which the order parameter satisfies at the boundaries of the system , as @xmath0 approaches infinity and @xmath6 , the scaling function depends only on the value of the combination @xmath7 , namely on the length @xmath0 in units of the correlation length @xmath8 . a dimensionless function such as @xmath9 defined by eq . [ scal1 ] can be thought of as a universal scaling function for the specific heat for a well - defined confining geometry . in other words , the scaling function does not depend on the microscopic details , but only depends on the nature of the universality class of the system , the confining geometry and the boundary conditions which are felt by the order parameter . in this limit ( @xmath10 and @xmath11 ) the scaling function is different for the three different cases mentioned previously for the following reason : for a fixed value of @xmath12 and for any large value of @xmath0 there is always a sufficiently small value of @xmath13 satisfying the condition where the correlation length is much larger than @xmath0 . however , in this limit the case ( a ) is the case of a 2d plane , the case ( b ) is the case of a 1d line and the case ( c ) is the zero - dimensional case . thus the value @xmath9 for sufficiently small values of @xmath14 should be very different for these three geometries . though earlier experiments on superfluid helium films of finite thickness @xcite seemed to confirm the validity of the finite - size - scaling(fss ) , there were later experiments@xcite where it was shown that the superfluid density of thick helium films does not satisfy fss when the expected values of critical exponents were used . similarly , in measurements of the specific heat of helium in finite geometries , other than the expected values for the critical exponents were found @xcite . more recent experiments in microgravity environment@xcite as well as earth bound experiments@xcite are consistent with scaling and they have determined the specific heat scaling function for the parallel plate ( film ) geometry ( case ( a ) ) and they are in reasonable agreement with the scaling function as was predicted by monte carlo simulations@xcite and renormalization group techniques@xcite . while the specific heat scaling function for case ( b ) confinement has been theoretically determined@xcite and it was found to be significantly suppressed relative to the case ( a ) there are so - far no experimental data to compare . more recently , the specific heat scaling function for the case ( c ) has been experimentally determined@xcite . the main goal of this paper is to present the results of our monte carlo simulations to determine the specific heat scaling function for cubes with open boundary conditions ( bc ) in all three directions ( confining case ( c ) ) . in this case the scaling function characterizes the zero - dimensional to three - dimensional transition . our results for the scaling function are compared to the very recently obtained experimental results for specific heat scaling function in the case of cubic confinement@xcite . we find satisfactory agreement with no free parameters . in addition , we present results for the specific heat scaling function for the parallel plate geometry on lattices of size @xmath15 with @xmath16 where we have applied periodic bc along the @xmath17-directions and open bc along the film - thickness direction of size @xmath0 . the latter case was carried out in order to compare the results for _ dirichlet bc _ ( on the top and bottom of the film ) obtained earlier@xcite . in refs @xcite it was found that while the results with periodic bc along the film - thickness direction were very different from those obtained with dirichlet bc , the results obtained with dirichlet bc fit the experimental results with no free parameter . in this paper we find that the scaling function obtained with open bc along the finite dimension is close to that obtained with dirichlet and also fits reasonably well the experimental results obtained by the so - called confined helium experiment@xcite ( chex ) . we have performed a numerical study of the scaling behavior of the specific heat of @xmath18 in a cubic and in a film geometry at temperatures close to the critical temperature @xmath19 . the superfluid transition of liquid @xmath18 belongs to the universality class of the three - dimensional @xmath20 model , thus , we are going to use this model to compute the specific heat at temperatures near @xmath19 using the cluster monte - carlo method @xcite . the @xmath20 model on a lattice is defined as @xmath21 where the summation is over all nearest neighbors , @xmath22 is a two - component vector which is constrained to be on the unit circle and @xmath23 sets the energy scale . we define the energy density of our model as follows : @xmath24 where @xmath25 for the cubes and @xmath26 for the film geometry . we have calculated the specific heat using the expression @xmath27 the above thermal averages denoted by the angular brackets are computed according to @xmath28 \exp ( - \beta { \cal h } ) , \label{ev}\end{aligned}\ ] ] where @xmath29 the energy in units of @xmath23 and @xmath30 . @xmath31 $ ] denotes the dependence of the physical observable @xmath32 on the configuration @xmath33 , and the partition function @xmath34 is given by @xmath35 we computed the specific heat @xmath36 of the @xmath20 model as a function of temperature on several cubic lattices @xmath2 ( with @xmath37 ) . open ( free ) boundary conditions were applied in all directions , namely the spins on the surface of the cube are free to take any value . these spins interact only with the 5 nearest neighbors , one in the interior and 4 on the surface of the cube and there is one missing neighbor . we have also calculated the specific heat scaling function @xmath38 ( to be defined in the following section ) for the case of the parallel plate geometry @xmath39 ( @xmath40 ) using periodic boundary conditions along the long directions of the film and open bc along the thickness direction @xmath0 . for this case we need to take the limit @xmath41 first ; in ref . @xcite it was found that using @xmath42 was large enough , in the sense that systematic errors due to the finite - size effects from the fact that @xmath17 are not infinite are smaller than the statistical errors for realistic computational time scales . the present simulations for films were done on lattices @xmath43 , @xmath44 , and @xmath45 . we used the monte carlo method and in particular wolff s cluster algorithm@xcite . we carried out of the order of 30,000 mc steps for thermalization to obtain equilibrium configurations . we made of the order of 10,000 - 50,000 measurements allowing 500 mc steps between successive measurements to obtain statistically uncorrelated configurations . defined by eq . [ scal1 ] . ] the main goal of this paper is to present a calculation of the specific heat scaling function for the case of cubic confinement and open bc . in this calculation we have used open bc in all three directions of the cube . we found that using open bc the results for the specific heat scaling were very close to those obtained with dirichlet bc . this will be demonstrated in the next section where we compare the previously published results@xcite for films with dirichlet bc and results reported in this paper for films with open bc . one can imagine a number of different scaling functions for the specific heat . any dimensionless combination such as the ratio given by eq . [ scal1 ] can be used as a scaling function . however , the various experimental groups have extracted two scaling functions , the so - called @xmath38 and @xmath46 with @xmath47 . these scaling functions are defined as follows : @xmath48 @xmath49t^{\alpha}= f_{2}(tl^{1/\nu } ) \label{scal - f2}\ ] ] we limit our goal to calculate the specific heat scaling function for the confined geometry and not the bulk critical exponents nor critical amplitude ratios . we take the values for the bulk critical exponents and the universal amplitude ratios as determined experimentally@xcite . previous mc work such as the work of ref . @xcite shows that the critical exponents are within error bars from the experimental values . our approach to use the experimentally determined values of the critical exponents and amplitude ratios and to determine the scaling function by applying fss on the calculated @xmath5 , has no fitting parameters and this allows no ambiguity . therefore we use @xmath50 as obtained from accurate experiments@xcite such as the so - called lambda point experiment ( lpe ) , an experiment in microgravity environment . the hyperscaling relation @xmath51 yields @xmath52 , and the correlation length @xmath53 = @xmath54 becomes equal to the system size @xmath0 at the reduced temperature @xmath55 , i.e. , @xmath56 with @xmath57 . in order to find the universal function @xmath9 defined by eq . [ scal1 ] , we need to know @xmath58 . we use the bulk values for @xmath58 obtained by studying the finite - size scaling of the specific heat of cubes with periodic bc@xcite . in fig . [ fi-1 ] the scaling function @xmath59 obtained for cubes with open bc in all three directions is compared to that obtained with periodic bc@xcite . the scaling function @xmath60 ( eq . [ scal - f1 ] ) can be calculated using our calculated @xmath5 and @xmath61 where the values of @xmath58 and @xmath62 for the @xmath20 model are obtained from reference@xcite . in order to calculate the universal function @xmath63 ( eq . [ scal - f2 ] ) , we need to know the bulk specific heat @xmath64 also . since we are restricting ourselves to the critical region we may write the following @xmath65 where @xmath66 is the universal amplitude ratio and it is most accurately determined experimentally@xcite from the critical properties of bulk helium to be @xmath67@xcite . inserting eqs . [ eq4.a ] and [ eq4.b ] into eq . [ scal - f2 ] , we obtain @xmath68t^{\alpha}+c_{1}^{+ } , \ \ t>0,\label{eq5a}\\ f_{2}(tl^{1/\nu } ) = [ c(0,\infty)-c(t , l)]|t|^{\alpha}+c_{1}^{+}/r \ \ t<0 . \label{eq5b}\end{aligned}\ ] ] which can be calculated by using our computed @xmath5 and the values of @xmath58 and @xmath62 from ref.@xcite . @xcite the specific heat scaling function for a parallel plate geometry on lattices of size @xmath69 with @xmath70 was calculated . in ref . @xcite periodic bc along the @xmath17-directions and staggered ( dirichlet ) bc or periodic along the film - thickness direction of size @xmath0 were applied . it was found that while the calculated scaling function for the parallel - plate geometry using periodic bc along all three directions was very different from that obtained with dirichlet bc along the top and bottom of the plate and periodic bc along the other two long directions , the latter scaling function fits the experimental results with no free parameter . this was explained on the basis that physically the order parameter along the film thickness vanishes at the boundaries of the film and therefore dirichlet bc are more appropriate . in this paper we have used open bc along the top and the bottom of the plate , instead of dirichlet , and periodic bc along the two long directions of the plate . since the film terminates on the top and on the bottom surface , for the pseudospins which belong to these two surfaces ( in language of the @xmath20 model ) there is no neighboring spins beyond the top and the bottom surface plane of the plate . therefore , even if we use open ( free ) boundary conditions this termination acts as `` a zero order parameter constraint '' beyond the top and the bottom of the plate . this implies that these two bc , namely staggered bc and open bc , are very similar for thick enough films . in order to make a direct comparison of our calculated @xmath60 to the experimental @xmath38 , we express all lattice units in physical units using the following equation@xcite @xmath71 where @xmath72 is the molar volume of liquid helium at the lambda point and saturated vapor pressure , @xmath73 is boltzmann s constant and @xmath74 the lattice spacing in the @xmath20 model required to make contact with the critical behavior of the correlation length in helium . this prefactor @xmath75 and it was determined in ref.@xcite . in fig . [ fi-7 ] we compare the results for @xmath76 for the case of films obtained with open bc along the direction of the film thickness to those obtained earlier@xcite and to the experimental results@xcite . it is clear that within error bars our results for the specific heat scaling function are the same for both cases of bc . with open bc ( solid circles ) is compared to the previously calculated scaling function using dirichlet bc@xcite(data shown as stars ) and periodic bc ( shown as plus signs ) and the experimental results of lipa _ et al._@xcite ( open circles ) and those of mehta _ et al_.@xcite ( open triangles ) . ] while the scaling function is sensitive to boundary conditions , this indicates that it is hard to distinguish dirichlet from open bc for the specific heat scaling function . we feel that when we use physical bc the agreement between the theoretical results for the specific heat scaling function and the experimental results is quite good taking into consideration the fact that there is no free parameter . in this section we present the results for the scaling functions @xmath38 and @xmath46 obtained for cubes of size @xmath2 with @xmath77 using open and periodic bc in all three directions . as was shown in the previous section open ( free ) bc are similar to using dirichlet bc and they both express the physical condition imposed by the confinement or the termination of the system . in fig . [ fi-2 ] we compare the scaling function @xmath60 obtained for open bc with that obtained for periodic bc@xcite . notice the suppression of @xmath38 when calculated with open bc relative to the case of periodic bc . this is similar to the case of the parallel - plate geometry ( fig . [ fi-7 ] ) . the scaling functions @xmath38 , however , are very different for cubic and parallel - plate geometry . notice , for instance , that for the case of cubic confinement with open bc @xmath38 is negative for all values of @xmath78 something very different of what happens for any of the calculated of the experimental scaling functions for parallel - plate confinement . obtained for cubes of size @xmath2 with open ( solid circles ) and that obtained for cubes with periodic bc ( open circles ) are compared . ] in fig . [ fi-3 ] we give the results of our present monte carlo calculation of the function @xmath46 for cubes with open bc using eqs [ eq5a],[eq5b ] . [ fi-4 ] shows the results of our calculation with periodic boundary conditions . [ fi-6 ] compares the scaling function @xmath63 obtained for open bc and for periodic bc . notice the qualitatively different behavior for the same scaling function for the same geometry but different boundary conditions . for cubes of size @xmath2 with open boundary conditions . ] for cubes of size @xmath2 with periodic boundary conditions ] experimentally the universal scaling function @xmath46 for cubic confinement has just become available@xcite . in order to make a direct comparison of our calculated scaling function @xmath63 to the experimentally determined , we express all lattice units in physical units . the prefactor is the same as in the case of the function @xmath38 : @xmath79 where @xmath80 is the constant given in the previous section by eq . [ lambda ] and its numerical value is @xmath75 . for open and periodic bc and for cubes is compared to the experimental results@xcite . ] in fig . [ fi-6 ] @xmath63 obtained from our mc calculation is compared with the experimental data@xcite . the agreement between the scaling function calculated with open bc and experiment is quite satisfactory considering the fact that there is no free parameters . in this paper we have used the @xmath20 model which describes the fluctuations of the superfluid order parameter near the critical point to calculate the scaling functions associated with the specific heat for the case where the superfluid is confined in a cubic geometry and in parallel - plate geometry . both in the theoretical calculations and in the experiments , the region very near the superfluid transition is probed such that the correlation length associated with the superfluid order parameter is of the size of the confining length . first , we calculated the specific heat scaling function for the case of parallel plate confining geometry using open boundary conditions along the top and the bottom surfaces of the film . our results are very close to those obtained@xcite with dirichlet ( staggered bc ) along the top and the bottom surfaces of the film . both calculations are in satisfactory agreement with the experimental results@xcite while the results of earlier calculations using periodic boundary conditions@xcite were found to disagree with the experimental scaling function near the superfluid transition . just recently , experimental measurements@xcite on superfluid helium confined in cubes became available . this prompted us to calculate the heat capacity scaling function of superfluids for cubic confinement . when we used open boundary conditions in all three directions of the cube we find very good agreement between the calculated and the measured@xcite scaling functions with no adjustable parameter . on the contrary , if periodic boundary conditions are used at the boundaries of the cube , which are unphysical boundary conditions for a confined system , there is great disagreement between the calculated and the measured specific heat scaling functions . this work was supported in part by nasa grants nag-1773 and nag-2867 and by the university of athens office of scientific research . the authors are grateful to m.o . kimball and f. m. gasparini who kindly provided access to their experimental results before publication and to j. a. lipa for providing a copy of the most recent analysis of the lpe experiment@xcite before acceptance for publication . m. e. fisher and m. n. barber , phys . * 28 * 1516 ( 1972 ) ; m. e. fisher , rev . phys . * 46 * 597 ( 1974 ) ; v. privman , finite size scaling and numerical simulation of statistical systems , singapore : world scientific 1990 ; e. brezin , j. physique * 43 * 15 ( 1982 ) ; v. privman , j. phys . * a23 * l711 ( 1990 ) . j. maps and r. b. hallock , phys . lett * 47 * 1533 ( 1981 ) ; d. j. bishop and j. d. reppy , phys . lett . * 40 * , 1727 ( 1978 ) . i. rhee , f. m. gasparini , and d. j. bishop , phys . * 63 * 410 ( 1989 ) . i. rhee , d. j. bishop , and f. m. gasparini , physica * b165&166 * 535 ( 1990 ) . t. chen and f. m. gasparini , phys . * 40 * 331 ( 1978 ) ; f. m. gasparini , t. chen , and b. bhattacharyya , phys . rev . * 23 * 5797 ( 1981 ) lipa , d.r . swanson , j.a . nissen , z.k . geng , p.r . williamson , d.a . stricker , t.c.p . chui , u.e . israelsson and m. larsen , phys . * 84 * , 4894 ( 2000 ) ; and j. low temp . phys . * 113 * , 849 ( 1998 ) . s. mehta and f.m . gasparini , phys . * 78 * , 2596 ( 1997 ) . s. mehta , m.o . kimball and f.m . gasparini , j. low temp . phys . * 114 * , 467 ( 1999 ) . n. schultka and e. manousakis , phys * 75 * , 2710 ( 1995 ) . n. schultka and e. manousakis , j. low temp . phys . * 109 * , 733 ( 1997 ) . r. schmolke , a. wacker , v. dohm , and d. frank , physica * b165 & 166 * 575 ( 1990 ) ; v. dohm , physica scripta * t49 * 46 ( 1993 ) ; p. sutter and v. dohm , physica * b194 - 196 * 613 ( 1994 ) ; w. huhn and v. dohm , phys . rev . lett . * 61 * 1368 ( 1988 ) ; m. krech and s. dietrich , phys . * a46 * 1886 ( 1992 ) . n. schultka and e. manousakis , j. low temp . phys . * 111 * , 783 ( 1998 ) . m. o. kimball , m. diaz - avila and f. m. gasparini , submitted to the editors of the lt23 conference in hiroshima , japan . m. o. kimball and f. m. gasparini , private communication . u. wolff , phys . lett . * 62 * , 361 ( 1989 ) . l.s . goldner and g. ahlers , b * 45 * , 13129 ( 1992 ) . n. schultka and e. manousakis , phys . rev . * b52 * , 7528 ( 1995 ) . j. a. lipa , j. a. nissen , d. a. stricker , d. r. swanson , t. c. p. chui , submitted for publication in phys * b * ; j. a. lipa , d. r. swanson , j. a. nissen , t. c. p. chui , u. e. israelsson , phys . lett , * 76 * , 944 ( 1996 ) . j. a. lipa and t. c. p. chui , phys . . lett . * 51 * , 2291 ( 1983 ) .
we study the specific heat scaling function of superfluids confined in cubic geometry and in parallel - plate ( film ) geometry with open boundary conditions ( bc ) along the finite dimensions using monte carlo simulation . for the case of cubic geometry for the superfluid order parameter we apply open bc in all three directions . we also calculated the specific heat scaling function for the parallel - plate confinement using open bc along the finite dimension and periodic bc along the other two dimensions and we find it to be very close to the earlier calculated using dirichlet instead of open bc . we find that the specific heat scaling function is significantly different for the two different geometries studied . in addition , we generally find that the scaling function for a fixed given geometry when calculated with open bc is quite close to that calculated with dirichlet bc , while when calculated with periodic bc is quite different . our results for both the scaling functions obtained for the parallel - plate geometry and for cubic geometry with open bc along the finite dimensions are in very good agreement with the recent very high quality experimental measurements with no free parameters .
the dynamics of many simple non - equilibrium systems are often studied through corresponding quantum hamiltonians . examples are the asymmetric @xmath9 chain hamiltonian and the attractive bose - hubbard hamiltonian for the single - step growth model @xcite and the directed polymers in random media ( dprm ) @xcite , respectively . the single - step growth model is a kardar - parisi - zhang ( kpz ) universality class growth model where the interface height @xmath10 grows in a stochastic manner under the condition that @xmath11 . the process is also called the asymmetric exclusion process ( asep ) in a different context . the evolution of the probability distribution for @xmath10 is generated by the asymmetric @xmath9 chain hamiltonian @xcite . the entire information about the dynamics is coded in the generating function @xmath12 . its time evolution , in turn , is given by the modified asymmetric @xmath9 chain hamiltonian @xcite , h_xxz()= -_i=1^l \ { e^2/l_i^-_i+1^++ 14(_i^z _ i+1^z-1 ) } , [ xxz ] with the @xmath13 s being the pauli matrices and @xmath3 the size of the system . ( we consider only the periodic boundary condition in this work . ) on the other hand , in the dprm problem , the partition function @xmath14 , the partition sum over directed polymer configurations with fixed ends at @xmath15 and @xmath16 , is the quantity of main interest . its generating function @xmath17 then evolves by the one - dimensional attractive bose - hubbard hamiltonian @xcite , h_bh(n)&=&-12_i=1^l(b_i b_i+1^+ b_i^b_i+1 - 2 ) + & & - u_i=1^lb_i^b_i(b_i^b_i-1)2 . [ bh ] here , the @xmath18s(@xmath19 s ) are the boson creation ( annihilation ) operators , @xmath20 is the conserved particle number , and @xmath21 is the attractive interaction . the two systems are closely related , at the level of continuum stochastic differential equations , through the cole - hopf transformation , @xmath22 @xcite . in particular , @xmath5 in eq . ( [ bh ] ) is related to the particle density , @xmath23 of eq . ( [ xxz ] ) , by @xmath24 @xcite . recently , the universal probability distribution function ( pdf ) at the stationary state was found for both models mentioned above @xcite . their common structure of the ground - state energies as functions of the scaling variables @xmath25 and @xmath26 , for the single - step growth model and the dprm problem , respectively , gives the universal pdf at the stationary state . while the ground - state energy of each hamiltonian gives information about the stationary state of the corresponding process , the first excited - state energy , combined with the ground - state energy , is related to the characteristic behavior of the process as it approaches the stationary state . for example , in the single - step growth model , as @xmath27 , the generating function takes the form e^h(x , t ) ~ \{-e_0()t + e^- ( ) t } , and similarly for @xmath28 . here , @xmath29 is the ground - state energy of eq . ( [ xxz ] ) , @xmath30 is the inverse of the relaxation time , as well as the gap between the ground - state energy and the first excited - state energy , @xmath31 , such that @xmath32 . the size dependence of @xmath30 , @xmath33 , defines the dynamic exponent @xmath1 . because the asymmetric @xmath9 chain hamiltonian , @xmath34 , is integrable by the bethe ansatz , the low - lying state energies , as well as the size dependence of the excitation gap , are well understood . when @xmath35 and the density of particles is finite in the limit @xmath36 , @xmath30 behaves as @xmath37 . however , when @xmath38 , @xmath30 behaves as @xmath39 @xcite . the dynamic exponent @xmath40 is a characteristic of the dynamic universality class of the kpz - type surface growth . when the number of particles is finite and the density of particles is very low , it is known that @xmath41 @xcite . however , when @xmath42 , which corresponds to the ferromagnetic phase , most bethe ansatz solutions are not available although the bethe ansatz equations continue to hold . as @xmath43 becomes negative , the quasi - particle momenta appearing in the bethe ansatz equations become complex , so solutions are difficult to obtain analytically . the attractive bose - hubbard hamiltonian is expected to have some resemblance to the ferromagnetic phase of the asymmetric @xmath9 chain hamiltonian considering the equivalence of @xmath43 and @xmath44 . the equivalence is identified indirectly by comparing the two scaling variables @xmath25 and @xmath26 under the relation @xmath24 or the two generating functions @xmath45 and @xmath17 under the relation @xmath22 . in contrast to the asymmetric @xmath9 chain hamiltonian , the bose - hubbard hamiltonian does not satisfy the bethe ansatz except in the two - particle sector @xcite . instead , the attractive delta - function bose - gas hamiltonian , h_d(n)= -12_i=1^n ^2 x_i^2 - u _ i < j ( x_i - x_j ) , [ delta ] which is the continuum version of the attractive bose - hubbard hamiltonian , is known to be integrable by the bethe ansatz . the attractive delta - function bose gas has been studied in refs . @xcite and @xcite . the ground - state energy is obtained from the bethe ansatz solution by using the symmetric distribution of the purely imaginary quasi - particle momenta . however , the structure of the energy spectra is not well known for the same reason as in the asymmetric @xmath9 chain hamiltonian with @xmath42 . the unknown energy spectra itself prevents one from understanding the dynamics of dprm near the stationary state . in this paper , we discuss in section ii the distribution of the quantum numbers appearing in the bethe ansatz equation for the first excited state of the attractive delta - function bose - gas hamiltonian , the knowledge of which is essential for solving the bethe ansatz equation . in section iii , the excitation gap of the attractive bose - hubbard hamiltonian with a small number of particles is investigated through the exact diagonalization method . we show that the gap decays as @xmath46 , i.e. , @xmath47 , but that the exponent becomes anomalous when @xmath48 . the emergence of the anomalous exponent is explained in connection with the transition of the first excited state from a positive energy state to a negative energy state . the bethe ansatz solutions in the two - particle sector show how the behavior of the gap varies with the interaction . we give a summary and discussion in section iv . in this section , we study the bethe ansatz solutions for the ground state and the first excited state of the attractive delta - function bose - gas hamiltonian . the eigenstate of @xmath49 , eq . ( [ delta ] ) , is of the form ( & & x_1,x_2, ,x_n ) + & & = _ p a(p)(ik_p1x_1+ik_p2x_2++ik_pnx_n ) , where @xmath50 is a permutation of @xmath51 and @xmath52 with no three @xmath53 s being equal . the quasi - particle momenta @xmath54 s are determined by solving the bethe ansatz equations , & & k_jl=2i_j + _ lj(k_j - k_l -u ) ( j=1,2, ,n ) , + & & ( x)=-2 ^ -1(x ) . [ bae ] if the distribution of quantum numbers @xmath55 is given , the set of quasi - particle momenta @xmath56 is uniquely determined . with such @xmath54 s , the energy eigenvalue is simply given by @xmath57 . for the ground state , the set of quantum numbers @xmath55 is i_j =- n+12+j , ( j=1,2, ,n ) , and the quasi - particle momenta are distributed symmetrically on the imaginary axis in the complex-@xmath58 plane . care should be taken when dealing with the first excited state . for the repulsive delta - function bose - gas hamiltonian , where @xmath5 is replaced by @xmath59 in eq . ( [ delta ] ) , the quantum numbers for one of the first excited states are i_j&=&-n+12+j ( j=1,2, ,n-1 ) , + i_n&=&n+12 . however , for the attractive case , by following the movement of the momenta as @xmath5 changes sign , we find that the quantum numbers for the first excited state should be given by i_j&=&-n-12+j ( j=1,2, ,n-1 ) , + i_n&=&-n-32 ( = i_1 ) . that is , the two quantum numbers @xmath60 and @xmath61 become the same . such a peculiar distribution of @xmath62 s does not appear in other bethe ansatz solutions such as those for the @xmath9 chain hamiltonian or the repulsive delta - function bose - gas hamiltonian . we remark that even though the two @xmath62 s are the same , all @xmath54 s are distinct ; otherwise , the wavefunction vanishes . such a distribution of quantum numbers is confirmed by the consistency between the energies obtained by diagonalizing the bose - hubbard hamiltonian exactly and those obtained by solving the bethe ansatz equations with the above quantum numbers for very weak interactions , for which the two hamiltonians possess almost the same energy spectra . when there is no interaction ( @xmath63 ) , all quasi - particle momenta , @xmath54 s , are zero for the ground state while for the first excited state , all the @xmath54 s are zero except the last one , @xmath64 . in the complex-@xmath58 plane , as the very weak repulsive interaction is turned on , the @xmath65 momenta are shifted infinitesimally from @xmath66 with @xmath67 , and the @xmath68th momentum is shifted infinitesimally to the left from @xmath69 . all the momenta remain on the real axis . when the interaction is weakly attractive , the @xmath65 momenta become complex with @xmath70 and @xmath71 for @xmath72 , and the @xmath68th momentum remains on the real axis , but is shifted to the left . figure [ momenta ] shows the distribution of the quantum numbers and the quasi - particle momenta in the presence of a very weak attractive interaction . the quasi - particle momenta are obtained by solving eq . ( [ bae ] ) . knowledge of the distribution of the quantum numbers is essential for solving the bethe ansatz equations of the attractive delta - function bose - gas hamiltonian . for the original attractive bose - hubbard hamiltonian , the bethe ansatz solutions are the exact solutions for the two - particle sector only , but are good approximate solutions in other sectors provided the density is very low and the interaction is very weak . this is because the bethe ansatz for the bose - hubbard hamiltonian fails once states with sites occupied by more than three particles are included . thus , for the sectors with three or more particles , the bethe ansatz solutions may be regarded as approximate eigenstates provided states with more than three particles at a site do not play an important role in the eigenfunctions . in ref . @xcite , it is shown that the error in the bethe ansatz due to multiply - occupied sites ( occupied by more than three particles ) is proportional to @xmath73 , where @xmath21 in ref . @xcite corresponds to @xmath59 in eq . ( [ bh ] ) . this applies to the attractive interaction case also . for the repulsive bose - hubbard hamiltonian , the bethe ansatz is a good approximation when the density is low and the interaction is strong because the strong repulsion prevents many particles from occupying the same site @xcite . for the attractive bose - hubbard hamiltonian , the bethe ansatz is good when the density is low and the interaction is weak because a weak attraction is better for preventing many particles from occupying the same site and because the error is proportional to @xmath73 . we are interested in the scaling limit @xmath36 with the scaling variable @xmath74 fixed because the common structure of the ground - state energies of the asymmetric @xmath9 chain hamiltonian and of the attractive bose - hubbard hamiltonian is found in this scaling limit . in this section , we investigate the size dependence of the excitation gap by using exact diagonalization of the attractive bose - hubbard hamiltonian in the two- and the three - particle sectors for @xmath3 up to @xmath75 . also , for the two - particle sector , we solve the bethe ansatz equation using the result of the previous section for larger @xmath3 . figure [ energy ] shows the ground - state energies and the first excited - state energies versus the size of the system in the two- and the three - particle sectors for three values of @xmath5 . note that for some value of @xmath5 , the sign of the first excited - state energy changes as the size of the system increases while for other values of @xmath5 , no such crossover is seen in the range of @xmath3 investigated here . the excitation gaps versus the size of the system are shown in fig . [ n=2gap ] and fig . [ n=3gap ] on a logarithmic scale . for @xmath76 and @xmath77 , the nearly straight lines indicate the power - law behavior of the gap , @xmath78 , and the slopes of the fitted lines indicate @xmath79 . however , for @xmath80 , the asymptotic behavior shows up only after a large crossover region where the effective @xmath1 , @xmath81 , is anomalously large . for the two - particle sector with @xmath80 , @xmath81 is about @xmath6 in the range @xmath82 . for the three - particle sector with @xmath80 , @xmath81 is about @xmath7 in the corresponding range @xmath83 . looking into the first excited - state energy in fig . [ energy ] , one can see that the anomalous value of @xmath81 appears in the range of @xmath3 where the first excited - state energies change their signs . we conjecture that the wavefunction has a transition as the sign of the energy changes . in order to confirm the connection between the anomalous exponent and the transition of the first excited state , we solved the bethe ansatz equation to evaluate the effective exponent @xmath81 near the transition point in the two - particle sector . in the two - particle sector , the quantum numbers for the ground state and the first excited state are @xmath84 and @xmath85 , respectively . the quasi - particle momenta are purely imaginary , i.e. , @xmath86 and @xmath87 for the ground state . for the first excited state , as noted already in ref . @xcite , when @xmath88 , the two quasi - particle momenta are real , i.e. , @xmath89 and @xmath90 , while when @xmath91 , @xmath58 is replaced by @xmath92 such that @xmath93 and @xmath94 . the examples of these distributions are shown in figs . [ root ] ( a ) and ( b ) with @xmath95 for the two values of @xmath96 and @xmath97 , respectively . the transition of the first excited state occurs when the quasi - particle momenta are imaginary . when @xmath91 , the ground - state energy(@xmath98 ) and the first excited - state energy(@xmath99 ) are , respectively , given by e_0=-4 ^2 ( 2 ) , [ energy_n=2_g ] and e_1=4 ^2 ( 2l ) ^2 ( q2 ) - 4 ^2 ( 2l ) ^2 ( q2 ) , [ energy_n=2_e ] where @xmath100 and @xmath101 are real and satisfy l=(2+u 2-u ) , [ bae_n=2_g ] and ql=(u+2(l)q u-2(l)q ) , [ bae_n=2_e ] respectively . we now consider the scaling limit @xmath36 with @xmath102 finite . let @xmath103 with @xmath104 . at this value of @xmath5 , the first excited - state energy , @xmath99 , is @xmath105 , @xmath106 , and @xmath107 . here , @xmath108 satisfies s_=(2 s_+s_u 2 s_-s_u ) , [ kappa ] which gives @xmath109 . when the size of the system @xmath3 is increased by @xmath110 with @xmath111 , the changes of @xmath100 and @xmath101 , @xmath112 and @xmath113 , are , from eqs . ( [ bae_n=2_g ] ) and ( [ bae_n=2_e ] ) , & = & -s_(4s_^2-s_u^2 ) 4s_^2-s_u^2 + ( 4/)s_u ll^2 - ll^2 , + q & = & ( s_u^2 - 4 ) ( 4/)s_u - s_u^2 + 4 ll^2 ll^2 . the perturbative expansion @xmath114 , under the assumption that @xmath115 , gives the value of @xmath81 at @xmath116 : z_eff=2 1+s_+s_^2 . numerical solutions of eq . ( [ kappa ] ) give @xmath117 . on the other hand , when @xmath118 , the ground - state energy , @xmath98 , is @xmath119 , and the first excited - state energy , @xmath99 , is @xmath120 , which are easily obtained from eqs . ( [ energy_n=2_g ] ) and ( [ energy_n=2_e ] ) . therefore , the excitation gap behaves as @xmath46 . also , when @xmath121 , the quasi - particle momenta of the ground state are @xmath122 , and those of the first excited state are @xmath123 , which lead to @xmath124 . for arbitrary @xmath5 , the effective value of @xmath81 is evaluated from the relation z_eff=((l+1)/(l-1))((l-1)/(l+1 ) ) by using the solutions of eqs . ( [ bae_n=2_g ] ) and ( [ bae_n=2_e ] ) for sufficiently large @xmath3 . as discussed above , the exponent @xmath81 shows an anomalous peak near @xmath111 or @xmath125 and approaches @xmath126 as @xmath127 or @xmath128 . figure [ z ] shows a plot of @xmath81 versus the scaling variable @xmath8 at @xmath129 . as the asymmetric @xmath9 chain generates the dynamics of the single - step growth model , the attractive bose - hubbard hamiltonian governs the dynamics of the dprm . we studied the attractive bose - hubbard hamiltonian and its continuum version , the attractive delta - function bose - gas hamiltonian concentrating on the behavior of the excitation gap , which is related to the characteristics of dprm relaxing into the stationary state . for the attractive delta - function bose gas hamiltonian , the quantum numbers for the first excited state in the bethe ansatz equation are found for the attractive delta - function bose gas hamiltonian , and the distribution of the quasi - particle momenta is discussed in the presence of a very weak attractive interaction . our result is the starting point for a further elucidation of the bethe ansatz solutions . we show that the excitation gap depends on the size of the system as a power law , @xmath130 , and that the exponent @xmath1 can be calculated by using an exact diagonalization of the attractive bose - hubbard hamiltonian in the two- and the three - particle sectors and by using the bethe ansatz solution in the two - particle sector . the exponent @xmath1 is @xmath126 . however , for the intermediate region where @xmath4 , the effective exponent @xmath81 shows a peak . the equivalence of the differential equations governing the single - step growth model and dprm implies some inherent equivalence in the corresponding hamiltonians . the power - law behavior of the excitation gap , @xmath46 , for the attractive bose - hubbard hamiltonian with a very weak interaction is the same as that for the asymmetric @xmath9 chain hamiltonian with a small number of particles , which is expected considering the relation @xmath24 . the fact that the excitation gap behaves anomalously for @xmath48 implies the possibility of an anomalous dynamic exponent @xmath1 for a finite scaling variable @xmath74 . if that is the case , one may expect the existence of some singularity at @xmath131 with finite @xmath5 , where the dynamic exponent @xmath1 takes an anomalous value , which is similar to the singularity at @xmath132 for the asymmetric @xmath9 chain hamiltonian . m. plischke , z. racz , and d. liu , phys . b * 35 * , 3485 ( 1987 ) . m. kardar , nucl . b * 290 * [ fs20 ] , 582 ( 1987 ) . l. h. gwa and h. spohn , phys . a * 46 * , 844 ( 1992 ) . b. derrida and j. l. lebowitz , phys . rev . lett . * 80 * , 209 ( 1998 ) . d .- s . lee and d. kim , phys . e * 59 * , 6476 ( 1999 ) . b. derrida and c. appert , j. stat . phys . * 94 * , 1 ( 1999 ) . j. krug and h. spohn , in _ solids far from equilibrium _ , edited by c. godrche ( cambridge university press , cambridge , 1991 ) , p. 412 . b. derrida and k. mallick , j. phys . a * 30 * , 1031 ( 1997 ) . park , j .- park , and d. kim , unpublished . e. brunet and b. derrida , phys . e * 61 * , 6789 ( 2000 ) . d. kim , phys . e * 52 * , 3512 ( 1995 ) . m. henkel and g. schtz , physica a * 206 * , 187 ( 1994 ) . t. c. choy and f. d. m. haldane , phys 90a * , 83 ( 1982 ) . e. h. lieb and w. liniger , phys . rev . * 130 * , 1605 ( 1963 ) . j. g. muga and r. f. snider , phys . a * 57 * , 3317 ( 1998 ) . w. krauth , phys . b * 44 * , 9772 ( 1991 ) .
the energy gap between the ground state and the first excited state of the one - dimensional attractive bose - hubbard hamiltonian is investigated in connection with directed polymers in random media . the excitation gap @xmath0 is obtained by exact diagonalization of the hamiltonian in the two- and three - particle sectors and also by an exact bethe ansatz solution in the two - particle sector . the dynamic exponent @xmath1 is found to be @xmath2 . however , in the intermediate range of the size @xmath3 where @xmath4 , @xmath5 being the attractive interaction , the effective dynamic exponent shows an anomalous peak reaching high values of @xmath6 and @xmath7 for the two- and the three - particle sectors , respectively . the anomalous behavior is related to a change in the sign of the first excited - state energy . in the two - particle sector , we use the bethe ansatz solution to obtain the effective dynamic exponent as a function of the scaling variable @xmath8 . the continuum version , the attractive delta - function bose - gas hamiltonian , is integrable by the bethe ansatz with suitable quantum numbers , the distributions of which are not known in general . quantum numbers are proposed for the first excited state and are confirmed numerically for an arbitrary number of particles . 2
the ubiquitous lyman @xmath16 lines detected in the spectra of qsos trace an important gaseous component of the universe . typically _ hundreds _ of forest lines are detected between the and ly@xmath17 qso emission lines in a high resolution spectrum of a _ single _ high @xmath0 qso ( e.g. , hu et al . 1995 ; lu et al . 1996 ; kirkman & tytler 1997 ) , and recent computations suggest that these absorbers contain a substantial fraction of the baryons in the universe at high and low redshifts ( e.g. , shull et al . 1996 ; miralda - escud et al . 1996 ; rauch et al . 1997 ; weinberg et al . 1997 ; bi & davidsen 1997 ) . cosmological simulations of structure formation suggest that at high redshift , the absorption lines trace the gas in large scale structures in between collapsed objects ( cen et al . 1994 ; petitjean , mcket , & kates 1995 ; miralda - escud et al . 1996 ; hernquist et al . 1996 ; zhang et al . 1997,1998 ) . the forest lines therefore provide a critical constraint on cosmological models as well as important clues about the formation and evolution of galaxies and large scale structures ( i.e. , voids , galaxy clusters , and galaxy superclusters ) . however , to make progress on this important topic and take advantage of the lines as a sensitive probe of cosmic phenomena from @xmath0 = 0 to @xmath0 = 5 , a better understanding of the _ nature _ of the absorbers is needed . the advent of the _ hubble space telescope ( hst ) _ has provided important new insights about the absorbers and an opportunity to directly study the relationship between the absorption lines and galaxies . spectroscopy of 3c 273 with _ hst _ in the first cycle of observations revealed that there are considerably more ly@xmath16 lines at low redshift than expected based on a simple extrapolation of the observed evolution of the number of lines per unit redshift ( @xmath18 ) at high @xmath0 ( morris et al . 1991 ; bahcall et al . subsequently , morris et al . ( 1993 ) carried out a redshift survey of galaxies brighter than @xmath19 19 within @xmath31@xmath20 of 3c 273 to study the relationship of the absorbers with galaxies , and they found no galaxies within a projected distance of 230 @xmath21 kpc of any of the thirteen lines toward 3c 273 with rest frame equivalent width @xmath22 exceeding 50 m . in fact , one line in the morris et al . ( 1993 ) sample turned out to be 4.8 @xmath21 mpc ( projected ) from the nearest bright galaxy . however , based on various statistical tests , morris et al . ( 1993 ) conclude that the absorbers are _ not _ randomly distributed with respect to the galaxies though the absorber - galaxy correlation is not as strong as the galaxy - galaxy correlation . in a similar program , stocke et al . ( 1995 ) and shull , stocke , & penton ( 1996 ) used the _ hst _ to search for lines in the cfa and arecibo redshift survey regions , and they find no galaxies within 450 @xmath9 kpc of the 10 - 11 absorbers they discovered , and 3 or 4 of their lines are found in galaxy voids ( the fourth is a tentative detection ) . overall , however , the majority of the stocke et al . and shull et al . absorbers are located within large scale galaxy structures supporting the statistical conclusions of morris et al . ( 1993 ) . in striking contrast to the results of morris et al . and stocke et al . , lanzetta et al . ( 1995 ) have measured the redshifts of galaxies within 1@xmath233 of six qsos observed for the _ hst _ qso absorption line key project ( bahcall et al . 1993 ) , and they find that between 32(@xmath610)% and 60(@xmath619)% of the lines in their sample are associated with luminous galaxies within @xmath3160 @xmath24 kpc of the qso sight lines , which suggests that a significant fraction of the clouds arise in large gaseous halos of intervening galaxies . furthermore , lanzetta et al . report that the line equivalent widths are anticorrelated with the intervening galaxy s projected distance from the qso sight line . these conclusions appear to be in conflict with the findings of morris et al . ( 1993 ) and stocke et al . ( 1995 ) . however , it may be possible to reconcile these discordant results . the lanzetta et al . sample is sensitive only to _ strong _ lines with @xmath25 300 m while the samples of morris et al . ( 1993 ) and stocke et al . ( 1995 ) are dominated by much weaker lines with @xmath26 200 m . based on this line strength difference it has been suggested that there are two populations of absorbers at low redshift : ( 1 ) strong absorbers with @xmath27 ( ) @xmath28 @xmath29 ( corresponding to @xmath5 300 m for doppler parameter @xmath30 35 ) that mostly occur in large gaseous halos of luminous galaxies , and ( 2 ) weaker absorbers with @xmath27 ( ) @xmath31 @xmath29 which are less closely tied to galaxies and may in some cases be truly intergalactic ( primordial ? ) gas clouds . currently this suggestion can not be rigorously tested , however , because the sample of weaker lines is rather small ; there are no more than 28 weak lines in the combined samples of morris et al . ( 1993 ) , stocke et al . ( 1995 ) , and shull et al . recently , grogin & geller ( 1998 ) have reanalyzed the morris et al . and stocke / shull et al . data with additional redshifts for 3c 273 , and they conclude that the absorbers on these sight lines are randomly distributed with respect to galaxies . evidently the nature of the low z absorbers is still an open question . to elucidate the nature of low redshift _ weak _ ly@xmath16 absorbers , we initiated a program in 1995 to study the relationship between low @xmath0 ly@xmath16 absorption line systems and galaxies in the directions of three qsos using the _ hst _ goddard high resolution spectrograph ( ghrs ) , the space telescope imaging spectrograph ( stis ) , and the wiyn multiobject fiber - fed spectrograph ( hydra ) . to significantly improve the size of the sample of weak lines , we designed the observations to obtain spectra adequate for detection of ly@xmath16 lines with @xmath32 = 50 m at the 3@xmath4 level . the wiyn hydra is used to measure the redshifts of bright galaxies in @xmath31@xmath20 fields centered on the qsos . in this paper we present a study of the relationship between clouds and galaxies in the directions of h 1821 + 643 ( @xmath2 = 0.297 ) and pg 1116 + 215 ( @xmath2 = 0.177 ) . these qsos have been observed previously with the _ hst _ faint object spectrograph ( fos , bahcall et al . 1992,1993 ; jannuzi et al . 1998 ) , but the fos spectra are not adequate for detecting _ weak _ lines ; their 4.5@xmath4 limiting equivalent widths are generally greater than 250 m throughout most of the cloud region . the third sight line for our program will be observed with stis in cycle 7 . the paper is organized as follows . in 2 we review the ghrs observations and data reduction and present the uv spectra . we also summarize the wiyn observations and data reduction in this section . we discuss the absorption line selection and identification in 3 . the properties of the clouds are examined in 4 , and the relationship between the absorbers and galaxies is analyzed in 5 . a discussion of the results and the nature of the absorbers is given in 6 . an overall summary of our investigation is provided in 7 . further details on this project can be found in tripp ( 1997 ) . throughout the rest of this paper we assume @xmath33 = 75 km s@xmath34 mpc@xmath34 and @xmath35 = 0 , and we rescale measurements from the literature to these values for the cosmological parameters . h 1821 + 643 was observed for 5.1 hours with the ghrs on march 29 , 1996 . standard target acquisition procedures were used to center the target in the large science aperture ( [email protected] ) , and the qso was observed with the g140l grating . these observations were obtained after the installation of costar , and the spectra have resolution @xmath37 2000 with a spectral spread function approximately described by a narrow gaussian profile containing @xmath3870% of the line profile area and a broader profile containing @xmath330% of the area ( see figure 4 in robinson et al . 1998 ) . in velocity units the resolution ( fwhm ) ranges from 160 at 1250 to 130 at 1540 . half - diode substepping was employed , which samples the ghrs spectral line spread function with @xmath31.2 diodes / fwhm . with this substepping pattern ( step pattern 4 ) , @xmath311% of the observing time is used for measuring the background with the science diodes . for accurate wavelength calibration , a pt - ne lamp was observed briefly at 1.7 hour intervals . the grating was positioned to provide complete wavelength coverage from 12521528 . combined with the earlier intermediate resolution ( 15 - 20 ) ghrs observations obtained by savage et al . ( 1995 ) , which included the 1231.71268.9 and 1521.21557.5 wavelength ranges , this allowed us to search for all clouds between = 0.013 and = 0.281 . since the qso systemic redshift is @xmath2 = 0.297 , these spectra cover all but a small fraction of the forest on this sight line . the fos data obtained by bahcall et al . ( 1992 ) can be used to fill in the missing redshift range , but their spectrum is not adequate for detection of the lines with @xmath39 250 m . bahcall et al . ( 1992 ) do not report detection of any lines at @xmath7 0.013 , but they do detect a strong line at = 0.297 , and a strong ly@xmath17 line is present at this redshift in our ghrs spectrum . the signal - to - noise ratio of high quality ghrs observations is ultimately limited by fixed pattern noise ( fpn ) due to scratches and manufacturing marks on the detector window and response irregularities of the photocathode . the detector diodes also introduce fixed pattern noise due to diode - to - diode sensitivity differences , but this noise is effectively reduced by comb addition and is generally weaker than the fpn due to the photocathode and faceplate . to reduce the fixed pattern noise , comb addition and the `` fp- split '' procedure were employed for the h 1821 + 643 observations . the fp- split procedure breaks an observation into subexposures and tilts the grating slightly between subexposures so that the position of the spectrum on the photocathode is shifted . when the subexposures are coadded _ in wavelength space _ , the fpn is smeared out and hence reduced . typically subexposures at four different grating tilts are obtained when the fp - split procedure is used , and s / n @xmath40 150 can be achieved when the subexposures are coadded ( cardelli & ebbets 1994 ) . to obtain higher s / n , the fixed pattern noise spectrum ( i.e. , the flat field ) must be derived from the fp - split subexposures and divided into the data . to prevent loss or degradation of data in case of _ hst _ glitches or orbital interrupts ( c.f . cardelli & ebbets ) , sets of four fp - split subexposures were read out roughly every four minutes , and a large number of subexposures at each of the four fp - split positions were obtained over the course of five hours . the preliminary data reduction was carried out at the ghrs computing facility at the goddard space flight center with the standard calhrs software . this included spectrum extraction and conversion to count rates , corrections for paired pulse events , background subtraction , and wavelength calibration . software developed at the university of wisconsin was then used to merge the individual subexposures and explore the usefulness of the fixed pattern noise flatfielding . due to non - repeatability of the grating carousel rotation , slight positioning errors occur every time the grating is rotated , and degradation of resolution occurs if it is assumed that the grating was rotated by exactly the instructed amount for merging of subexposures . to achieve the highest possible resolution , the centroids of well - detected and narrow absorption lines in the individual fp - split subexposures are calculated and used to determine the shifts for merging . finally , the zero point in the wavelength scale was determined by comparing the wavelengths of galactic interstellar absorption lines in the g140l spectrum to their known wavelengths derived from the high resolution interstellar absorption line study of savage et al . ( 1995 ) scale zero point and were not used for this purpose . ] . savage et al . also measured accurate wavelengths of the ly@xmath17 line at = 0.22489 and the 1031.93 line at = 0.22503 ( see also savage , tripp , & lu 1998 ) , so these lines were also used to determine the mean wavelength zero point . we estimate that the h 1821 + 643 g140l observations have an absolute velocity uncertainty of @xmath330 , which corresponds to a redshift uncertainty of @xmath30.0001 . the h 1821 + 643 g140l observation is close to the s / n regime where application of a fpn correction may be beneficial ( see 3 ) . after merging the data into four `` master '' fp - split subexposures ( i.e. , all fp - split subexposures at the same position on the photocathode combined ) we used the method of cardelli & ebbets ( 1994 ) to derive and apply the fpn correction . the final fpn corrected spectrum of h 1821 + 643 is shown in figure [ h1821grand ] where the observed count rate is plotted against heliocentric wavelength ( in ) . note that at @xmath41 1251.6 and at @xmath42 1537.9 , the spectrum falls off the end of the diode array in some of the fp - split subexposures and consequently the s / n is lower in these regions . since these spectral regions have already been observed by savage et al . ( 1995 ) at higher resolution , their only purpose in the g140l spectrum is to provide better constraints on the continuum placement . this is important at the short @xmath43 end of the g140l spectrum where the continuum rises ( see figure [ h1821grand ] ) ; without the spectrum extention a significantly different continuum fit could be selected . the final s / n ranges from 100 to 160 per resolution element . the combined s / n and resolution are adequate for detection of 50 m lines at the 3@xmath4 level or better throughout the g140l spectrum . pg 1116 + 215 was observed for 3.5 hours with the ghrs on 4 february 1997 . the observational setup and data reduction procedure were nearly identical to the h 1821 + 643 observation and reductions , but quarter - diode substepping ( step pattern 5 ) was used to provide more on - target exposure time and less ( but still ample ) time spent measuring the background . after the pg 1116 + 215 observations were fully reduced ( including the fpn correction ) , the data were binned to half - diode samples to improve the s / n and match the h 1821 + 643 data . the final fpn corrected and half - diode binned spectrum of pg 1116 + 215 is plotted in figure [ p1116grand ] . while h 1821 + 643 was actually brighter than expected when we observed it with the ghrs g140l grating , pg 1116 + 215 was apparently fainter than expected based on previous _ iue _ and _ hst _ measurements . consequently , the count rate was lower than expected . the s / n ( @xmath3 75 - 160 per resolution element ) of the pg 1116 + 215 ghrs g140l spectrum is adequate for detection of 75 m lines at the 3@xmath4 level or better throughout the entire spectrum and at the 50 m level for @xmath44 1400 . a special zero point velocity correction was not applied to the pg 1116 + 215 observations . the weak interstellar lines of and ( see 3.1.2 ) have average heliocentric velocities of @xmath317 while 21 cm emission in the direction of pg 1116 + 215 has an average heliocentric velocity of -20 ( lockman & savage 1995 ) . some of this difference could be due to the pronounced positive velocity wing of emission seen in the 21 cm profile . we estimate that the pg 1116 + 215 g140l spectrum has an absolute velocity uncertainty of @xmath350 , which corresponds to a redshift uncertainty of @xmath30.0002 . to study the relationship between the absorbers and galaxies in the directions of h 1821 + 643 and pg 1116 + 215 , we have conducted a survey of galaxy redshifts in the @xmath45 fields centered on the qsos using the fiber - fed multiobject spectrograph ( hydra ) on the wiyn telescope . the galaxy targets were selected primarily from ( 1 ) eleven 67 @xmath36 67 images in harris r clustered around h 1821 + 643 acquired with the wiyn imager in july 1995 , and ( 2 ) 1@xmath46 images obtained with the kpno schmidt telescope by b. jannuzi & r. green ( 1995 , private communication ) . for h 1821 + 643 , a few additional targets from the poss ii database were observed to extend the angular coverage at the lowest redshifts . objects in the wiyn and schmidt images were selected and classified as stars or galaxies by b. jannuzi and a. tanner using focas ( tyson & jarvis 1979 ; valdes 1982 ) . our original objective was to measure the redshifts of all galaxies brighter than @xmath47 19 in the 1@xmath20 fields centered on the qsos ( estimates of completeness of the final sample are provided below ) . however , a galaxy redshift sample complete to @xmath48 19 will not provide many redshifts with @xmath49 0.2 . since the h 1821 + 643 sight line probes clouds out to = 0.297 , we obtained the wiyn images for a deeper redshift survey near the center of the qso field . a full description of the wiyn observations and data reduction will be provided in a separate paper ( tripp et al . 1998 ) ; here we provide a summary . the observations were carried out on several observing runs between august 1995 and april 1997 . for all runs we used the simmons camera with the 400 line mm@xmath34 `` blue '' grating and a tektronics 2048@xmath362048 ccd ( t2kc ) . for most of the runs we used the red fibers which subtend 2 on the sky . however , on one of the early runs we used the blue fibers which subtend 3 . this setup provided a resolution of 4.6 ( @xmath3300 km s@xmath34 ) with the red fibers or 7.1 ( @xmath3 450 km s@xmath34 ) with the blue fibers . to include the h & k lines at @xmath0 = 0 , the grating was positioned to provide spectra extending from 3900 to 7100 . we typically observed 60 - 75 candidate galaxies per setup with 10 - 30 fibers positioned on the sky . three 50 minute exposures were obtained for each fiber configuration . targets were given priority based on brightness and proximity to the qso . due to fiber positioning constraints , in some setups not all of the fibers could be placed on new targets , so previously observed galaxies were reobserved to improve the s / n . spectra of objects observed in multiple setups were coadded to obtain the final redshift , but the individual spectra were also measured to test the reliability and uncertainties of the redshift measurements ( see tripp et al.1998 ) . the data were reduced with iraf using the dohydra package . we then used the standard method of tonry & davis ( 1979 ) to determine the galaxy redshifts by cross - correlating the target spectra with spectra of `` templates '' , i.e. , high s / n galaxy spectra with known redshifts , after masking out any emission lines or night sky residuals and passing both the target and the templates through a fourier bandpass filter . we used six templates for the cross - correlation measurements . the first template is a spectrum of m32 observed with hydra with one of the central fibers in the slit ( which have the highest throughput ) and reduced in an identical fashion to the target data reductions . the heliocentric velocity of m32 is 209 ( de vaucouleurs et al . the other templates are actual targets in the h 1821 + 643 and pg 1116 + 215 fields for which we obtained high s / n spectra , and the heliocentric velocities of these templates were established by cross - correlation with the m32 spectrum . the templates were intentionally selected to have a range of redshifts to check the reliability of the cross - correlation redshifts in marginal cases . for emission line galaxies , the redshifts were measured using templates with absorption lines and night sky residuals masked out . the galaxy redshifts measured in the fields of h 1821 + 643 and pg 1116 + 215 are summarized in tables [ h1821reds ] and [ p1116reds ] , respectively . we also list in these tables the poss ii @xmath50 magnitudes ( denoted @xmath51 since @xmath50 is roughly equivalent to @xmath52 ) and impact parameters ( projected distances from the sight line ) for each galaxy . the poss ii magnitudes have rms errors of @xmath30.21 magnitudes ( weir , djorgovski , & fayyad 1995 ) . since ccd calibration of the poss ii magnitudes was not available for these fields , we determined the magnitude zero - point by comparison to the measurements of kirhakos et al . absolute magnitudes calculated using the interstellar extinction corrections based on @xmath53 from lockman & savage ( 1995 ) and the k - correction from peebles ( 1993 ) are also provided in tables [ h1821reds ] and [ p1116reds ] . we have measured 98 redshifts in the field of h 1821 + 643 and 118 redshifts in the field of pg 1116 + 215 . for h 1821 + 643 , we have obtained 56 additional redshifts from schneider et al . ( 1992 ) , le brun , bergeron , & boiss ( 1996 ) , and bowen , pettini , & boyle ( 1998 ) , bringing the grand total to 154 . these additional redshifts and the other quantities when known are listed at the end of table [ h1821reds ] . the listed redshifts are on a heliocentric basis . in the direction of h 1821 + 643 , @xmath0(galactocentric ) - @xmath0(heliocentric ) = 0.0009 , while toward pg 1116 + 215 , @xmath0(galactocentric ) - @xmath0(heliocentric ) = -0.0003 . table [ complete1116 ] provides estimates of the completeness of the pg 1116 + 215 galaxy redshift survey for various limiting magnitudes out to various radii based on the number of galaxy candidates in the schmidt catalog and the number of redshifts actually obtained . it will be important to consider the impact of the angular coverage of the galaxy redshift survey ( at low redshifts ) and its limited depth ( at high redshifts ) , so we also list in table [ complete1116 ] the redshifts at which @xmath54 = 17 , 18 , 19 , and 20 correspond to a 1@xmath55 galaxy [ using @xmath56 = 19.5 from loveday et al . ( 1992 ) ] and the physical scale corresponding to a 10 , 20 , and 30 radius at each of these redshifts . for example , based on the schmidt catalog of targets within 20 of the qso , we estimate that the pg 1116 + 215 survey is 86.9% complete for @xmath57 19.0 . @xmath54 = 19 corresponds to a 1@xmath55 galaxy , and 20 corresponds to 2.4 mpc , at @xmath58 0.121 . the pg 1116 + 215 target list contained very little stellar contamination ; only six observed targets turned out to be stars . unfortunately , the h 1821 + 643 catalog suffered much more severe stellar contamination ; 64 observed candidates turned out to be stellar ( see tripp et al . this stellar contamination was partly due to the low latitude and direction of the qso ( @xmath594 ) which causes the sight line to pass over several spiral arms and through the warped part of the outer milky way ( see savage et al . 1995 ) , but it was also partly due to the fact that we observed the h 1821 + 643 field first experience gained from the h 1821 + 643 field was used to improve the star / galaxy separation procedure for pg 1116 + 215 . due to the large number of stars in the target list , estimates of the h 1821 + 643 survey completeness are more uncertain . if we remove objects which are observed to be stars and objects for which we did not get good spectra but are probably stars ( based on magnitude and poss ii classification ) , then we estimate that within a 20 radius from the qso , the h 1821 + 643 redshift survey is roughly 72.4% complete for @xmath60 18.0 and 50.8% complete for @xmath60 20.0 . our survey is substantially more complete than the previous surveys noted above because any redshifts from those surveys which we missed were added to our sample . table [ h1821reds ] shows that redshifts have been obtained for a number of galaxies as faint as @xmath61 21.0 in the h 1821 + 643 field . taking @xmath56 = 19.5 and the interstellar extinction and k - corrections described above , we find that @xmath62 21.0 corresponds to a roughly 0.5@xmath55 galaxy at @xmath0 = 0.20 . therefore we have not measured redshifts for galaxies fainter than 0.5@xmath55 beyond @xmath63 0.2 . furthermore , most of the galaxies at @xmath64 0.2 for which we have obtained redshifts are in the rich galaxy cluster which hosts the qso , so our redshift survey is severely incomplete for intervening galaxies at @xmath64 0.2 . consequently , in our analysis of the relationship between absorbers and galaxies in 5 , we shall consider a sample which only includes galaxies and absorbers with @xmath65 0.2 . absorption lines in the h 1821 + 643 g140l spectrum were selected for further analysis based on the statistical significance of the observed equivalent width @xmath32 . the automated objective procedure described in tripp et al . ( 1996 ) was used to measure the equivalent widths . the errors in the equivalent widths account for continuum placement uncertainty as well as statistical noise . for the lines detected at the @xmath66 level lines typically increases to 4 - 5@xmath4 if we only consider statistical noise . ] , the reality of the feature was checked by visually inspecting the four fp - split subexposures ; if the line is apparent in at least three of the subexposures , then the detection is probably reliable and the line was included in the line list . we have measured all of the equivalent widths with and without the fixed pattern noise correction , and for the most part , the fpn correction did not have a significant impact on the equivalent width measurements . however , for a few of the weakest lines the correction was important . figure [ fpndemo ] shows an example of a 3@xmath4 absorption line which is completely removed by the fpn correction . consequently , we elected to apply the correction to all lines in the final sample . table [ h1821list ] lists the observed equivalent widths of all reliably detected lines in the h 1821 + 643 g140l spectrum . table [ h1821list ] also lists the heliocentric vacuum wavelength of the centroid of each line , the line identification , the redshift , and , for ism lines , the heliocentric velocity of the line . for line identification and throughout this paper , we use wavelengths from morton ( 1991 ) . profile fitting was employed to extract the redshifts of the individual features using vpfit ( developed by j. webb and r. carswell , see carswell et al . 1991 ) and to infer the presence of additional components in some complex absorption blends ( see below ) . table [ p1116list ] provides the same information for lines detected in the spectrum of pg 1116 + 215 . the redshift uncertainties from vpfit are typically @xmath30.0001 . including the uncertainties in the velocity scale ( see 2.1,2.2 ) , we estimate that the total redshift uncertainties are @xmath30.0002 . as noted above , the previous _ hst _ observations of h 1821 + 643 obtained by bahcall et al . ( 1992,1993 ) and savage et al . ( 1995 ) can be used to expand the redshift range probed in this study . table [ g160mlist ] summarizes the equivalent widths , identifications , and redshifts of extragalactic absorption lines relevant to this study detected in these previous observations . note that while the savage et al . observations have a limiting equivalent width similar to that of our g140l spectrum , @xmath350 m at the 3@xmath4 level , their resolution is considerably better so we prefer the savage et al . data in cases where the lines are also present in the g140l spectrum . an important issue in the study of weak lines is possible contamination of the extragalactic line sample by absorption lines due to the ism of the milky way . while the strong ism lines are easily identified ( c.f . table 4 in morton , york , & jenkins 1988 ) , in spectra sensitive to 50 m equivalent widths , a significant number of weaker ism resonance lines which are less familiar may be detected as well , and these can easily be misidentified as clouds in low resolution spectra . galactic interstellar lines identified in the ghrs spectra of h 1821 + 643 and pg 1116 + 215 are listed in tables [ h1821list ] and [ p1116list ] , and absorption profiles of ism lines of interest are plotted in figures [ h1821ism ] and [ pg1116ism ] . in the following two sections we justify the galactic line identifications . in the wavelength range of the g140l spectrum shown in figure [ h1821grand ] , weak interstellar lines of and are of particular concern . neutral carbon is not the dominant ionization state of c in warm clouds ( the ionization potential of is 11.3 ev ) , but nevertheless it has been detected in absorption on interstellar sight lines with low ion column densities comparable to the those observed toward h 1821 + 643 . for example , lu , savage , & sembach ( 1994 ) measure @xmath32 = 104@xmath617 m for the galactic 1560.31 line at low velocity in the spectrum of the seyfert galaxy ngc 3783 . the equivalent widths in the low velocity interstellar gas toward ngc 3783 are comparable to the corresponding equivalent widths observed toward h 1821 + 643 ( savage et al . 1995 ) , so comparable lines may be expected toward h 1821 + 643 . our g140l h 1821 + 643 spectrum does not cover the 1560.31 multiplet , but it does cover the 1277.25 , 1280.13 , and 1328.83 resonance lines . since the value of @xmath67 , the product of the oscillator strength and wavelength , of the 1277.25 line is very similar to @xmath67 of the 1560.31 line ( morton 1991 ) , we might expect to see a roughly 100 m line due to the ism in the h 1821 + 643 spectrum , and indeed a 150 m line is observed at 1277.21 ( see figure [ h1821grand ] and table [ h1821list ] ) . figure [ h1821ism ] shows the normalized absorption profiles of the galactic 1277.25 , 1280.13 , and 1328.83 lines ( the 1280.13 profile is in the same panel as the 1277.25 line ) . for purposes of comparison , we also plot the milky way 1253.81 , 1526.71 , and 1334.53 lines from the g140l spectrum in figure [ h1821ism ] . the @xmath67-values of the 1280.13 and 1328.83 transitions are lower than the 1277.25 @xmath67-value by factors of 4.0 and 1.6 respectively . the 1280.13 line is not detected , but this is not surprising since it has a lower oscillator strength . an absorption feature of moderate strength is present at the expected wavelength of 1328.83 , but it is blended with the strong ly@xmath17 line at = 0.297 ( see figure [ h1821ism ] ) . it is possible that we have detected the galactic 1277.25 and 1328.83 lines . however , the 1277.25 line is rather broad compared to the strong galactic ism lines ( figure [ h1821ism ] ) . there are * and * * transitions near the 1277.25 line , but the strongest of these are only separated from 1277.2 by 70 ( see morton 1991 ) , so this seems an unlikely explanation of the breadth of the 1277.21 absorption line . this line may be blended with a cloud at @xmath40 0.05 . this is important because a cluster of galaxies is detected near this redshift in the wiyn survey ( see 5.2 ) . higher resolution spectroscopy with stis will be required to determine if absorption actually contributes to the feature at 1277.21 . singly ionized nickel also has several weak resonance lines which are likely to be detected in high s / n ism spectra . in the wavelength range of the g140l h 1821 + 643 spectrum , the strongest lines occur at 1317.22 , 1370.13 , and 1454.84 with f - values of 0.1458 , 0.1309 , and 0.05954 ( morton 1991 ) . figure [ h1821ism ] shows the normalized absorption profiles of these lines ; all three of the lines are well - detected in the g140l spectrum of h 1821 + 643 . from figure [ h1821ism ] we see that the profiles of the 1317.22 and 1370.13 lines appear to have complex component structure with a main component at the velocity of the other ism lines and weaker absorption extending from roughly 400 to + 400 . since the weak component structure appears in both profiles , one is tempted to attribute this high velocity absorption to interstellar . however , this weak high velocity absorption can not be entirely due to because it is not detected in the absorption profiles of strong ism lines . for example , savage et al . ( 1995 ) observed the interstellar 2796.35 , 2803.53 doublet at 15 resolution , and the absorption profiles of these lines , which are among the strongest of the ism lines , show no evidence of absorption with @xmath68 160 or @xmath69 + 60 ( see figure 2 in savage et al . ) . similarly , the strong 1334.53 and 1526.71 lines do not show high velocity interstellar gas of comparable strength ( see figure [ h1821ism ] ) . what then is the source of the high velocity absorption near the profiles ? for 1317.22 , continuum placement may play a role . with the uncertainty of the continuum placement , the extended absorption near 1317.33 is marginally detected . the extended absorption near 1370.13 , on the other hand , is real . the 1367.95 transition is close to the component at -400 in the 1370.13 profile , but this identification can not be correct because has a much stronger line at 1358.77 which is not apparent and because the wavelength does nt match the cu line well . given the close proximity of this line to the strong multicomponent line at 1363.15 , it is likely that the extra absorption near 1370.13 is extragalactic absorption . there are multiple absorption components near the 1454.84 line as well , but these occur at high positive velocities with different relative strengths than the components near the other lines ( see figure [ h1821ism ] ) and thus are not due to . the absorption line at 1414.41 may have a contribution from galactic 1414.40 . however , toward the well studied star @xmath70 oph with @xmath27(h ) = 1.38@xmath71 atoms @xmath29 , this ga line has @xmath32 = 4 m ( savage , cardelli , & sofia 1992 ) . since the line at 1414.41 in the h 1821 + 643 spectrum has @xmath32 = 63@xmath610 m , it is unlikely that this line is entirely due to , but ga is significantly depleted from the gas phase by dust in the @xmath70 oph interstellar cloud . if we assume that ga is not depleted at all in the interstellar gas toward h 1821 + 643 and scale the @xmath70 oph equivalent width by the h 1821 + 643/@xmath70 oph hydrogen column densities , then we predict @xmath72 12 m for the line in the h 1821 + 643 spectrum . this is a conservative upper limit since ga probably is at least somewhat dust depleted on the h 1821 + 643 sight line . therefore we conclude that the line at 1414.41 is primarily due to extragalactic with possibly a small contribution from . the interstellar lines in the ghrs g140l spectrum of pg 1116 + 215 are listed in table [ p1116list ] and many of them are displayed on a velocity basis in figure [ pg1116ism ] . close inspection of these interstellar profiles reveals evidence of a galactic high velocity cloud ( hvc ) at @xmath3200 ( see , e.g. , the and profiles in figure [ pg1116ism ] ) . we have taken this hvc into account in our identification of the lines . in contrast to h 1821 + 643 , the 1277.25 multiplet is not detected toward pg 1116 + 215 . the difference may be due to the fact that the galactic sight line to pg 1116 + 215 contains a considerably lower column density of than that toward h 1821 + 643 : @xmath27 ( ) = 1.40@xmath73 versus 3.84@xmath73 @xmath29 ( lockman & savage 1995 ) . we attribute the line found at 1328.46 to rather than 1328.83 because of the absence of the 1277.25 multiplet which should be @xmath31.6 times stronger . the line at 1239.40 with @xmath32 = 170@xmath624 lies close to the wavelengths of galactic 1238.80 and 1239.93 and 1240.39 . however , we identify this feature as dominated by absorption at = 0.1950 because there is no evidence for the weaker member of the doublet at 1242.80 ( see fig . [ pg1116ism ] ) and the velocity centroid implied for the absorption ( @xmath3145 km s@xmath34 ) is larger than we would expect to observe given that the ism 1393.76 and 1402.77 lines are at + 43 km s@xmath34 . also , the 1239.93 and 1240.39 doublet lines have small f - values and these lines typically have equivalent widths of 10 to 25 m in the spectra of distant stars in the milky way halo ( savage & sembach 1996 ) . again , it would be useful to obtain a high resolution stis spectrum to confirm the identification of this line and to properly remove any absorption due to interstellar and . we identify the line at 1317.42 as 1317.21 . with this identification we would expect to see a line of comparable strength at 1370.13 . while there is no statistically significant line detected at that wavelength , a feature of low significance is evident in figure [ pg1116ism ] . we believe the 1317.21 identification is the correct one and that noise and/or continuum placement uncertainties have probably combined to cause the detection of one line and the apparent weakness of the second . the interstellar triplet lines at 1259.52 , 1253.81 , and 1250.58 ( with f - values of 0.0162 , 0.0109 , and 0.00545 , respectively ) shown in figure [ pg1116ism ] are blended with various other absorption features . the high s / n of the observations permits a clear separation of the overlapping absorption features . 1259.52 blends with the very strong interstellar 1260.42 line and separate equivalent widths are given in table [ p1116list ] . 1253.81 blends with a weaker line at 1255.00 that we identify as a feature at = 0.03223 . 1250.58 blends with a broader and stronger feature we identify as a line at = 0.02845 . this is evident from the strength of the 1250.30 line ; if this line is entirely due to galactic 1250.58 , then it should be a factor of @xmath32.0 weaker than the 1253.81 line . instead we see from figures [ p1116grand ] and [ pg1116ism ] that the 1250.30 line is clearly stronger than the the line at 1253.86 . also , the centroid of the 1250.30 line is blueshifted with respect to the other ism lines ( see figure [ pg1116ism ] ) . with these identifications the relative strengths and velocities of the three lines from profile fitting are consistent with their origin in the galactic ism . the two lines that blend with the lines differ in strength by about a factor of two . we considered the possibility that these lines might be a metal line doublet but could not find any plausible identifications involving the stronger systems we see in the spectrum of pg 1116 + 215 . while the line at 1255.0 may contain a contribution from in the galactic hvc , this feature can not be entirely due to high velocity interstellar . this is clear from comparison of the 1250.58 and 1253.81 lines ; if the line at 1255.0 were mostly due to high velocity , then it would be much stronger than observed in the 1250.58 profile . the other ism lines found in the spectrum of pg 1116 + 215 are definite detections of lines known to be strong in the galactic ism . there are no known galactic ism lines close to the observed line at 1448.43 . there are also no reasonable identifications of this feature as a metal line associated with one of the stronger lines seen in the spectrum of pg 1116 + 215 . therefore this line is left as unidentified in table [ p1116list ] . heavy elements are detected in two extragalactic absorption line systems in the g140l spectrum of h 1821 + 643 . we detect the 1031.93 , 1037.62 doublet and 977.02 in the associated absorber at = 0.2967 , and the doublet is also detected in the intervening absorption system at = 0.2250 . these metal absorbers have been analyzed in a separate paper ( savage , tripp , & lu 1998 ) . note that the line at 1478.27 could be identified as 1206.50 associated with the absorber at = 0.2250 . however , this identification seems unlikely because neither nor lines are detected at this redshift with sensitive upper limits ( see savage et al . 1998 ) , and it is difficult to simultaneously satisfy the lower limits on the / and / ratios in photoionized or collisionally ionized gas . therefore we attribute the line at 1478.27 at least partially to extragalactic absorption . corresponding ly@xmath17 absorption is detected in the g160 m spectrum ( see table [ g160mlist ] ) , but the redshifts derived from the ly@xmath17 and lines differ by 0.0008 , so this line is probably blended , perhaps with associated with the absorber . once again , higher resolution observations are needed to resolve this line identification ambiguity . in the spectrum of pg 1116 + 215 , a statistically significant line is detected at the expected wavelength of 1206.50 at = 0.13852 , within @xmath330 km s@xmath34 of the strong line at = 0.13861 . 1260.42 is also marginally detected at this redshift . the rest of the lines not identified as galactic ism lines or extragalactic metals ( or ly@xmath17 ) are identified as lines provided @xmath74 . in a few cases , the @xmath75 statistic from profile fitting with vpfit is poor unless additional components are added to the fit . on this basis we infer the presence of two components in the line at 1384.16 in the spectrum of pg 1116 + 215 . there are some indications that there is an additional line ( not listed in table [ h1821list ] ) blended with interstellar 1393.76 toward h 1821 + 643 ( see tripp 1997 ) , but higher resolution observations are required to reliably identify this line . the high detection sensitivity of our ghrs spectra provides a rare opportunity to assess the line density , @xmath18 ( the number of lines per unit redshift ) , of weak ly@xmath16 lines and its evolution at low redshift . in the following analysis we exclude from consideration the region within 3000 km s@xmath34 of the emission redshift of each quasar for the reasons given in 5.2 . combining the g140l spectrum obtained here with the g160 m spectra of savage et al . ( 1995 ) , we detect a total of 26 ly@xmath16 lines with @xmath76 m ( 19 with @xmath77 m ) in the spectrum of h 1821 + 643 over the wavelength region 1233 - 1556 , corresponding to a redshift path @xmath78 = 0.247 after applying a correction of @xmath79 for the redshift path blocked out by strong ism lines . this implies @xmath80 for ly@xmath16 lines with @xmath76 m , where the error is determined from @xmath81 statistics . a similar analysis for the 3c 273 sightline ( morris et al . 1993 ) with 13 ly@xmath16 lines with @xmath76 m ( 8 with @xmath77 m ) over the wavelength region 1218 - 1382 with a redshift path of 0.135 ( including the two lines associated with the virgo cluster and the line blended with galactic absorption absorption at the same redshift . ] ) yields @xmath82 for ly@xmath16 lines with @xmath76 m . for pg 1116 + 215 , 13 ly@xmath16 lines with @xmath76 m ( 10 with @xmath77 m ) are found in the g140l spectrum between 1226 and 1417 , corresponding to @xmath83 after applying a correction of 0.013 for the redshift path blocked out by the ism lines . the resulting line density is then @xmath84 for ly@xmath16 lines with @xmath76 m , where the inequality results from the fact that not all spectral regions of the pg 1116 + 215 g140l spectrum have adequate s / n to detect ly@xmath16 lines with @xmath85 m . these estimates are in essential agreement with each other . combining the h 1821 + 643 and the 3c 273 sightlines , we find @xmath86 for ly@xmath16 lines with @xmath76 m at @xmath87 . similarly we find @xmath88 for ly@xmath16 lines with @xmath77 m combining the data for all three sightlines . the line density at @xmath87 , @xmath89 for ly@xmath16 lines with @xmath76 m , is consistent with the number of ly@xmath16 absorbers with @xmath27()@xmath90 @xmath29 at these redshifts predicted by riediger , petitjean , & mcket ( 1998 ) based on their study of ly@xmath16 cloud evolution in the context of a cdm model . note that an absorber with @xmath27()@xmath91 @xmath29 would produce a ly@xmath16 absorption line with @xmath22 = 50 m if @xmath92 km s@xmath34 . recently , shull ( 1997 ) has reported preliminary results from analysis of 7 sight lines observed in _ hst _ cycle 6 . based on these new spectra , which are reported to be adequate for detection of 20 m lines , shull ( 1997 ) derives @xmath18 = 250@xmath640 . evidently the number of lines detected continues to rise as the sensitivity improves . this is also predicted by riediger et al . and may have important implications for the nature of the absorbers and the baryon content of the low @xmath0 forest ( see shull 1997 and discussion below ) . previous studies of the redshift evolution of ly@xmath16 clouds at low redshift have been limited to relatively strong lines ( @xmath93 m , e.g. , weymann et al . 1998 ) . the redshift evolution of weaker ly@xmath16 lines is of considerable interest since , according to cosmological simulations of structure formation and evolution , they may arise from the tenuous gas in void regions and therefore may evolve differently from the stronger lines which mostly trace the gas distribution near galaxy structures ( compare the @xmath94 and @xmath95 populations in figure 6 of riediger et al . the distribution of ly@xmath16 lines in redshift is generally described by the expression @xmath96 where @xmath97 is the line density at @xmath98 . dividing the data for the above three sightlines into two redshift bins at @xmath99 and @xmath100 and considering only ly@xmath16 lines with @xmath22 between 75 and 300 m , we find @xmath101 and @xmath102 . the data seem to indicate some evolution in the number density of the weak lines but the relatively small number of lines and the small redshift span of the current sample ( @xmath103 ) make any determination of the evolution index @xmath104 very uncertain . a formal fit to the equation above using the technique of lu , wolfe , & turnshek ( 1991 ) yields @xmath105 for ly@xmath16 lines with @xmath106 m at @xmath103 . a similar value is obtained if lines as weak as 50 m are included . a striking aspect of figure [ h1821grand ] is the complex component structure of the strong ly@xmath16 lines . to allow the reader to inspect this more closely , the normalized absorption profiles of the four strongest ly@xmath16 lines in the spectrum of h 1821 + 643 are shown on expanded scales in figure [ lyapros ] , and the rest equivalent widths of the ly@xmath16 lines are plotted versus redshift in figure [ lyastick ] . though there is some confusion from adjacent unrelated lines , we see from figures [ lyapros ] and [ lyastick ] that the absorption profiles of three out of the four strongest ly@xmath16 lines contain complex component structure with a main strong component and several partially resolved weaker outlying components spanning 1000 - 1500 km s@xmath34 . even the ly@xmath16 line at @xmath107 which does not show resolved weak components has an asymmetric profile which is evidence for unresolved components . similarly , two out of the three strongest ly@xmath16 lines seen toward pg 1116 + 215 show evidence of multiple components ( see table [ p1116list ] ) . this seems to suggest that there is some clustering of weak ly@xmath16 lines at low redshift . clustering of strong ly@xmath16 lines ( @xmath108 m ) at low @xmath0 has been reported by lanzetta et al . ( 1996 ) and ulmer ( 1996 ) , but clustering of weak ly@xmath16 lines has not been detected previously . to quantitatively measure the degree to which the ly@xmath16 lines are clustered , we use the standard technique involving the two - point velocity correlation function , @xmath109 where @xmath110 is the number of observed line pairs with velocity separation @xmath111 and @xmath112 is the number of pairs expected in a random distribution . if the clouds are not clustered on velocity scale @xmath111 , then @xmath113 . to expand our ly@xmath16 line sample , we have combined our data with those of morris et al ( 1993 ) for 3c 273 , stocke et al ( 1995 ) and shull et al ( 1996 ) for i zw 1 , mrk 335 , mrk 421 , and mrk 501 , and bruhweiler et al . ( 1993 ) and savage et al . ( 1997 ) for pks 2155@xmath114304 . all these spectra were obtained with the ghrs and have enough sensitivity to detect ly@xmath16 lines as weak as 75 m throughout the spectra , except the g140l spectrum of pks 2155@xmath114304 obtained by bruhweiler et al ( 1993 ) which has somewhat lower sensitivity and is adequate for detection of lines with @xmath115 m . figure [ twop]a shows the two - point correlation function for lines with @xmath116 no evidence for clustering on any velocity scale is evident . figure [ twop]b shows the two - point correlation function for lines with @xmath115 m , which shows a marginal signal ( @xmath66 ) at @xmath117 km s@xmath34 , and the signal may extend up to @xmath118 km s@xmath34 . the number of lines in the sample becomes too small for this analysis if a higher equivalent width limit is imposed . clearly a much larger sample of lines is required to further explore the clustering properties of weak ly@xmath16 lines . in this section we examine the relationship between absorbers and galaxies . we begin with the sample definitions and nearest neighbor distances ( 5.1 ) . we then present some individual cases of interest and examine the correlations between the line equivalent widths and the nearest galaxy impact parameters and distances in 5.2 . finally , in 5.3 we carry out a statistical analysis of the nature of the absorber - galaxy relationship . in figures [ pie1821 ] and [ pie1116 ] we overplot the line redshifts ( indicated with vertical lines ) on ra and dec slices showing the galaxy redshifts ( circles ) ; the length of the line indicates the absorption line equivalent width as shown in the figure legend . figure [ histreds ] shows the number of galaxies in the survey as a function of redshift in 1000 bins with the absorber redshifts plotted at the top of each panel . for the absorption lines we consider the following samples : ( 1 ) a `` total '' sample which includes all lines listed in tables [ h1821list ] - [ g160mlist ] , ( 2 ) a `` complete '' sample which only includes lines with @xmath119 0.20 but is otherwise identical to the total sample , and ( 3 ) a `` 1 mpc complete '' sample which only includes lines with 0.048 @xmath119 @xmath119 0.20 . the second sample is defined because the h 1821 + 643 galaxy redshift survey is severely incomplete at @xmath49 0.20 ( see 2.3 ) . the `` 1 mpc complete '' sample is defined to ensure good spatial coverage with a high degree of galaxy redshift survey completeness out to at least 1 mpc from the qso ( c.f . , table [ complete1116 ] ) . at @xmath120 0.048 , a 20 radius corresponds to less than 1 mpc , so nearest galaxies could be missed because they are outside of the field covered by the redshift survey . we will compare results derived from the `` complete '' and `` 1 mpc complete '' samples to evaluate the impact of this bias . for reasons given in 5.2 , in all samples we exclude lines within 3000 of the qso redshift . following morris et al . ( 1993 ) and stocke et al . ( 1995 ) , we further delineate the samples based on the assumption used to calculate the radial distances between absorbers and galaxies . we either ( a ) assume the radial distance between a given absorber and galaxy is given by the hubble law ( referred to as the `` pure hubble flow '' sample ) , or ( b ) assume the absorber and galaxy are at the same radial distance if @xmath121 300 and calculate the radial distance between the absorber and galaxy from ( @xmath122 for @xmath123 300 ( the `` perturbed hubble flow '' sample ) . the perturbed sample is intended to account for motions from , e.g. , galaxy rotation , which cause departures from the hubble flow . the projected ( i.e. , transverse ) distance between clouds and galaxies , also referred to as the impact parameter , is calculated assuming @xmath124 = 75 mpc@xmath34 and @xmath125 = 0 . tables [ h1821near ] and [ p1116near ] list the projected and three - dimensional distances from the absorbers to their nearest galaxies , for the pure hubble flow sample , along with the absorber - galaxy velocity separation . we now turn to the relationship between galaxies and absorbers implied by our study , and we begin with inferences drawn from direct inspection of the data . _ individual galaxies . _ one of the major conclusions of the cloud study of lanzetta et al . ( 1995 ) is that a large fraction ( @xmath332 - 60% ) of the _ strong _ lines originate in individual luminous galaxies within projected distances of @xmath3200 kpc from the line of sight . more recently , lanzetta et al . ( 1997 ) have suggested that the fraction is even larger and that possibly all of the absorbers arise in such galaxies . furthermore , lanzetta et al . claim that the equivalent width is anticorrelated with impact parameter , which indicates that there is a physical association between the galaxies and the absorbers and that the density of the gaseous halo decreases with increasing distance from the galaxy . this anticorrelation has been contested by le brun et al . ( 1996 ) and bowen , blades , & pettini ( 1996 ) , but chen et al . ( 1998 ) argue that le brun et al . and bowen et al . have effectively diluted the anticorrelation by including galaxy - absorber pairs at large impact parameters ( up to @xmath3 5 mpc ) which are unlikely to be physically associated . we can test the conclusions of lanzetta et al . because our galaxy samples include 42 galaxies with impact parameters @xmath126 600 kpc ( see tables [ h1821reds ] and [ p1116reds ] ) . however , we should exclude any galaxies and absorption lines which have @xmath127 because ( 1 ) the absorbers with @xmath128 are now recognized to be a special class of absorption systems ; in some cases there is clear evidence that these absorbers are close to the qso nucleus ( e.g. , hamann et al . 1997a , b ; barlow & sargent 1997 ) , ( 2 ) these absorbers can be significantly photoionized by the qso itself ( the `` proximity effect '' ) , and ( 3 ) in the case of h 1821 + 643 , the qso is located in a rich galaxy cluster ( see figure [ pie1821 ] and hall , ellingson , & green 1997 ) which is a special environment . for consistency with lanzetta et al . and chen et al . , we exclude objects within 3000 of the qso redshift . this leaves us with 20 galaxies with @xmath126 600 kpc in the fields of h 1821 + 643 and pg 1116 + 215 . first consider the closer galaxies : if we find a galaxy with an impact parameter of 200 kpc or less , do we find a strong line at that redshift ? our sample is small we only have three galaxies with impact parameters of 200 kpc or less but in all three cases we find a strong line with @xmath25 300 m within 150 of the galaxy redshift . furthermore , as we consider larger impact parameters , we continue to find nearby absorbers . in fact , all galaxies within projected distances of 600 kpc from the sight lines have associated absorbers with @xmath129 1000 ( @xmath130 is the mean of and @xmath131 ) . in some cases there are several galaxies near the redshift of the cloud ; following chen et al . ( 1998 ) and lanzetta et al . ( 1995 ) , we first assume that the galaxy with the smallest impact parameter is the most likely to be associated with the absorber ( we consider an alternate selection criterion below ) . this reduces our list of 20 close galaxies to 12 close galaxies . we also treat lines within 350 of each other as a single absorber . given these assumptions , we find 12 galaxy - absorber pairs with @xmath11 600 kpc and @xmath132 1000 ( 7 pairs toward h 1821 + 643 and 5 pairs toward pg 1116 + 215 ) . we have used @xmath133 1000 because this is the selection criterion originally employed by lanzetta et al . while there are situations which could lead to such high velocity gas flows in galaxy clusters ( e.g. , roettiger , stone , & mushotzky 1998 ; burns 1998 ) , in general @xmath133 1000 is a rather large velocity interval which corresponds to many mpc in an unperturbed hubble flow . therefore one might reasonably wonder if these 12 galaxy - absorber pairs are coincidental alignments of randomly distributed lines which occasionally happen to be close to a galaxy . to address this question , we have used monte carlo simulations ( see 5.3 ) to randomly distribute lines along the h 1821 + 643 and pg 1116 + 215 sightlines . based on 10000 monte carlo trials , we find that the probability of 7 chance galaxy - absorber pairs toward h 1821 + 643 and 5 chance pairs toward pg 1116 + 215 is 4.7@xmath134 . furthermore , nine of these 12 observed galaxy - absorber pairs have @xmath11 600 kpc and @xmath133 340 , and the probability of this occuring if the clouds are randomly distributed is 1.3@xmath135 , again based on 10000 monte carlo trials . taking into account the uncertainties in the galaxy and cloud redshifts , it seems reasonable to attribute galaxy - absorber velocity differences of @xmath3350 to normal galaxy motions such as rotation . a similar corroborating result is obtained from the 3c 273 data obtained by morris et al . there are five galaxies 600 kpc in the 3c 273 sample , but four of these are in virgo at roughly the same redshift with @xmath136 = 0.0065 , and again we use the galaxy with the smallest impact parameter . ] in the sample of morris et al . with impact parameters of 600 kpc or less , and in all five cases there are clouds detected within 500 . in four out of the five 3c 273 close galaxy - absorber pairs , the absorber is within 250 of the galaxy redshift . in all , there are 17 galaxies with @xmath11 600 kpc along the three sight lines ( 3c 273 , h 1821 + 643 , and pg 1116 + 215 ) ; every one of them has a corresponding absorption line with @xmath137 1000 , and 14 out of the 17 galaxies are within 500 of a line . this is a strong indication that absorbers are somehow associated with luminous galaxies but does not necessarily indicate that the absorption lines originate in the large gaseous halos of the identified individual luminous galaxies . the lines could be due to undetected faint dwarf galaxies which are clustered with the observed luminous galaxies , or they could be due to broadly distributed gas within the large scale structures where luminous galaxies are found , for example . since in some cases there are several galaxies close to a line of sight at a given absorber redshift ( see figures [ pie1821 ] and [ pie1116 ] ) , it is nt clear that the galaxy with the smallest impact parameter is the most likely one ( or the only one ) to be associated with the absorber ( as assumed above ) . in some cases the galaxy with the smallest impact parameter has a larger velocity separation from the absorber than another galaxy at a somewhat larger impact parameter . an alternative criterion for identifying galaxy - absorber pairs is to match absorbers with the galaxy with the smallest _ three - dimensional _ distance . assuming the perturbed hubble flow described above , this selection method leads to the identification of 17 galaxy - absorber pairs within projected distances of 1 mpc with @xmath138 350 in the h 1821 + 643 and pg 1116 + 215 sight lines . these galaxy - absorber pairs are summarized in table [ galx600 ] . monte carlo simulations indicate that the probability of drawing this many pairs from a randomly distributed population is 3.6@xmath10 . we now turn to the equivalent width - impact parameter anticorrelation which lanzetta et al . use to argue that the strong lines are truly due to luminous _ individual _ galaxies . does this anticorrelation extend to larger impact parameters ? lanzetta et al . ( 1995 ) and chen et al . ( 1998 ) could not detect such an extension because their fos line measurements generally do not have sufficient sensitivity to detect the weak absorption lines , but our data are adequate for this purpose . to explore this question , we have combined the h 1821 + 643 and pg 1116 + 215 data with the 3c 273 data from morris et al . ( 1993 ) and , for the stronger lines at smaller impact parameters , the data from chen et al . ( 1998 ) . from these papers we have selected all galaxy - absorber pairs with impact parameters of 600 kpc or less and @xmath139 1000 . the results are plotted in figure [ rhocor ] . this figure shows that the lanzetta et al . ( 1995 ) anticorrelation does appear to extend to larger impact parameters . the spearman rank - order correlation coefficient for the data in figure [ rhocor ] is @xmath140 = -0.762 , and with 44 data points , this indicates that the data are anticorrelated at the @xmath141 significance level .. ] fitting a power - law of the form @xmath142 where @xmath143 is the galaxy impact parameter and @xmath144 is a constant , to the data in figure [ rhocor ] yields @xmath16 = [email protected] and @xmath144 = [email protected] where @xmath22 is in m and @xmath143 is in kpc ( fit shown as a solid line in figure [ rhocor ] ) . based on the strong line data alone , chen et al . derive @xmath16 = [email protected] , in agreement within the 1@xmath4 uncertainties . in figure [ rhocor ] , we have only included galaxies with impact parameters of 600 kpc or less . if we increase the impact parameter cutoff , we continue to find galaxies apparently associated with lines , be we also find some galaxies which do not have associated absorption lines , within @xmath137 1000 , of the strength expected from equation ( 3 ) . for example , a galaxy at @xmath131 = 0.18967 in the h 1821 + 643 field is at a projected distance of 635 kpc , and therefore from eqn . ( 3 ) we expect to find a line with @xmath145 95 m near this redshift , which should be easily detected . from figure [ h1821grand ] , it is clear that no line of this strength is present at this redshift ( a weak feature is apparent at the expected wavelength , but it is lower than 3@xmath4 significance and much weaker than @xmath145 95 m ) . from the h 1821 + 643 , pg 1116 + 215 , and 3c 273 sight lines , we find 15 cases of galaxies with 600 @xmath146 2200 kpc that do not have unambiguously associated lines within 1000 . however , five of these cases are at redshifts where the associated line could be hidden within strong ism lines or extragalactic lines . the remaining 10 cases which do not have associated lines are plotted as upper limits in figure [ bigrhocor ] , which also shows all galaxies with @xmath147 2200 kpc that do have lines within 1000 . remarkably , using the sample shown in figure [ bigrhocor ] , we find that the anticorrelation persists at high significance levels out to @xmath12 2200 kpc . there is a lot of scatter in the data shown in figure [ bigrhocor ] , so the cases where we do not find associated lines may simply be cases where the line is weak enough to fall below our detection threshold ; the upper limits for these cases are consistent with the anticorrelation . if we select only galaxy - absorber pairs with @xmath126 2200 kpc and @xmath148 500 , the anticorrelation is still strong because there are only five galaxy - absorber pairs in figure [ bigrhocor ] with 500 @xmath149 1000 ( the triangles encased in squares in figure [ bigrhocor ] ) . this strong anticorrelation is another indication that the absorbers and galaxies shown in figure [ bigrhocor ] are in some way physically related and may have important implications regarding the nature of the absorbers . we discuss possible interpretations of the anticorrelation in 6 . however , it is important to emphasize that there are some potential selection biases which could artificially tighten the anticorrelations shown in figures [ rhocor ] and [ bigrhocor ] . for example , the current galaxy redshift surveys only probe relatively bright galaxies ( see table [ complete1116 ] ) , so there could be fainter galaxies at smaller impact parameters than those plotted in figure [ bigrhocor ] that might be revealed by a deeper redshift survey . discovery of such galaxies would modify the appearance of figure [ rhocor ] and weaken the anticorrelation by moving points from the lower right corner of figure [ bigrhocor ] to the lower left corner of the plot . going deeper might also reveal close galaxies which do not have corresponding absorption . along these lines , linder ( 1998 ) has used simulations to argue that most absorbers could be due to faint galaxies rather than the observable luminous galaxies at the same redshift . some deep searches near the lowest redshift lines have not found such faint galaxies ( e.g. , morris et al . 1993 ; van gorkom et al . 1993 ; rauch et al . 1996 ) , but van gorkom et al . ( 1996 ) and hoffman et al . ( 1998 ) have located faint dwarf galaxies at the redshifts of a few low @xmath0 lines . it will be important to do more deep searches for faint galaxies to establish whether or not the anticorrelation in figure [ rhocor ] is real . another potential bias in figure [ bigrhocor ] ( and in the analyses of lanzetta et al . and chen et al . ) may be introduced by the selection of _ only _ lines which are known to be within a certain projected distance and velocity of a galaxy . due to these selection criteria , there are many lines in our sight lines and in the chen et al . ( 1998 ) sight lines which are not plotted in figure [ bigrhocor ] . in some cases , these unplotted lines may not have known galaxies nearby simply because the appropriate galaxy redshift survey is incomplete . alternatively , these lines may not have known nearby galaxies because the nearest galaxy is outside of the angular extent of the redshift survey . if this is the case , then these missing galaxy - absorber pairs would have relatively large impact parameters and would fill in the upper right corner of figure [ bigrhocor ] , and this would weaken the anticorrelation . to illustrate the possible impact of this selection effect , we plot on the far right side of figure [ bigrhocor ] some of the missing lines from the chen et al . ( 1998 ) sample ( using equivalent widths reported by jannuzi et al . 1998 and bahcall et al . 1993 ) and the missing lines from our sight lines . from this we see that the missing lines could indeed substantially dilute the anticorrelation . to check to see if this selection bias is occuring , more galaxy redshift measurements are needed . _ specific galaxy structures . _ a visual inspection of figures [ pie1821 ] , [ pie1116 ] , and [ histreds ] and tables [ h1821near ] and [ p1116near ] yields the following information on the relationship between the clouds and galaxy structures . first , there are no unambiguously detected lines in the vicinity of the prominent galaxy clusters at @xmath58 0.050 and 0.190 in the direction of h 1821 + 643 . the galaxy survey of morris et al . ( 1993 ) revealed a similar cluster of galaxies at @xmath150 0.08 with no associated absorbers in the direction of 3c 273 . this may indicate that absorbers are destroyed ( or are not created in the first place ) in this type of environment . however , do we expect to see absorbers in these clusters based on the impact parameters of the nearest galaxies ? as already noted above , in the cluster at @xmath58 0.190 , there is a galaxy at a projected distance of 635 kpc and we should see an absorber there based on equation ( 3 ) . in the cluster at @xmath58 0.050 , the nearest galaxy has an impact parameter of 590 kpc , so we might expect a line at that redshift with @xmath151 100 m . in this case a line with the right strength is well - detected at the expected wavelength , but its identification is ambiguous since this absorption may be due to in the milky way ism ( see 3.1.1 ) . morris & van den bergh ( 1994 ) have shown that a substantial fraction of lines at low @xmath0 could be due to material stripped out of galaxies in tidal interactions , and they point out that in this model absorbers would be less likely to be found in rich clusters because ( 1 ) there are fewer gas - rich spirals , and ( 2 ) tidal debris should be less common due to the higher velocity dispersion of rich clusters . furthermore , rich galaxy clusters contain hot intracluster media , so clouds may be fully ionized in this environment . therefore it is not necessarily surprising to find that absorbers may avoid rich clusters . however , more sight lines crossing rich clusters must be studied to give this suggestive result statistical significance . we note that there is a strong cloud in the rich cluster which hosts h 1821 + 643 , but since this is a @xmath152 absorber , this is a special case . this cloud may be quite close to the qso nucleus or gas ejected by the qso , for example ( see savage et al . 1998 ) . given the very large number of galaxies in this rich cluster ( see figure [ pie1821 ] ) , it is noteworthy that the absorption only shows one strong component , albeit at the fos g130h resolution ( fwhm @xmath40 250 ) . examination of figures [ pie1821 ] - [ histreds ] reveals other smaller galaxy groups which do not have associated lines within @xmath153 = 1000 such as the group at @xmath0 = 0.072 toward h 1821 + 643 . however , such inspection also shows several galaxy groups which _ do _ have associated absorbers , e.g. , the group at @xmath0 = 0.121 toward h 1821 + 643 or the groups at @xmath0 = 0.060 , 0.138 , and 0.166 toward pg 1116 + 215 . evidently the absorbers do not always avoid the regions of higher galaxy concentration . this was known previously from the results of stocke et al . ( 1995 ) and shull et al . ( 1996 ) who showed that the majority of their lines are found within large scale structures even though their sight lines probe substantial paths through galaxy void regions , but it is still possible that the lines avoid the regions of _ highest _ galaxy density , i.e. rich clusters . second , as found for other sight lines by morris et al . ( 1993 ) , stocke et al . ( 1995 ) , and shull et al . ( 1996 ) , our study reveals some absorbers apparently in galaxy voids at very large distances from the nearest galaxy in the survey ( see tables [ h1821near ] and [ p1116near ] and figures [ pie1821 ] - [ pie1116 ] ) . at @xmath154 0.20 , absorbers apparently in voids should not be taken too seriously because at these redshifts our galaxy survey is very incomplete . at lower redshifts , our survey is deep enough to reveal faint galaxies . consider the examples of lines at large distances from galaxies at = 0.04105 toward h 1821 + 643 and at = 0.01639 toward pg 1116 + 215 . at @xmath0 = 0.04105 , @xmath54 = 19.0 corresponds to 0.14@xmath55 , while at @xmath0 = 0.01639 , @xmath54 = 19.0 corresponds to 0.02@xmath55 . therefore based on the completeness estimates from 2.3 , there is a good chance that we would have identified a galaxy as bright as the lmc near these void absorbers if such galaxies were present . however , at these redshifts the limited angular extent of the galaxy redshift survey ( see table [ complete1116 ] ) is a source of concern ; in these cases a galaxy at a projected distance of one to a few mpc might be missed because it is outside of the region surveyed . therefore more redshift measurements with broader angular coverage are needed to establish that these are truly void absorbers . figures 9 - 11 give a visual impression that the absorption lines in voids tend to be weaker . to explore this possibility , we have divided ( somewhat arbitrarily ) our complete perturbed hubble flow sample into `` void '' and `` non - void '' samples , the non - void absorbers defined by having three - dimensional nearest neighbor distances of 3 mpc or less and the void absorbers defined by nearest neighbor distances greater than 3 mpc ( we have also included the morris et al . and stocke / shull et al . data in this exercise ) . we then compared the equivalent width distributions . this division at 3 mpc is not entirely arbitrary though because it correctly separates the lines which stocke et al . ( 1995 ) and shull et al . ( 1996 ) define as void and non - void absorbers on different grounds . figure [ voidnot ] compares the resulting equivalent width distributions of the void and non - void absorbers . while it does appear that the strongest lines tend to be non - void absorbers ( which is not surprising given the discussion above ) , the two distributions are similar , and a ks test indicates that there is a 15% probability that they were drawn from the same parent distribution . therefore there is no statistically significant difference between the equivalent width distribution of the void and non - void absorbers in the present sample . applying the same test to the `` 1 mpc complete '' sample , which is less prone to the problems discussed in the previous paragraph , yields a very similar result . in this case we find a 14% probability that the void and non - void absorbers were drawn from the same equivalent width distribution . in 5.2 we addressed the question : if there is a galaxy within some impact parameter , do we find a absorber near that redshift ? the other obvious question to ask about our data is : given that we have detected a absorber at redshift @xmath0 , is there a galaxy or galaxy structure nearby ? toward pg 1116 + 215 , 69% of the absorbers in the total sample have nearest neighbor galaxies within three - dimensional distances of 2 mpc , while toward h 1821 + 643 , 47% of the absorbers have galaxies within 2 mpc , assuming the perturbed hubble flow . in the `` 1 mpc complete '' sample , 75% of the pg 1116 + 215 absorbers and 53 % of the h 1821 + 643 absorbers are within 2 mpc of the nearest galaxy . specifically , we consider the following issues . ( 1 ) previous studies have indicated that overall , the absorbers are not randomly distributed with respect to galaxies . is this confirmed by our data ? recently this conclusion has been challenged by grogin & geller ( 1998 ) . ( 2 ) as noted in 1 , there is considerable interest in comparing the relationship between the weak absorbers and galaxies to that between the strong absorbers and galaxies . is there a statistically significant difference ? stocke et al . ( 1995 ) split their sample into weak and strong absorbers and found that the weak lines are statistically consistent with a random distribution . however , stocke et al . emphasize that their sample of weak clouds is rather small . we can check this result with a larger sample . following the previous studies of morris et al . ( 1993 ) and stocke et al . ( 1995 ) , we use monte carlo nearest neighbor tests to examine these issues . to do this , we randomly distributed absorbers along the h 1821 + 643 , pg 1116 + 215 , and 3c 273 sight lines with a line density in accordance with eqn . ( 1 ) ; we tried various values of @xmath104 ranging from 0.0 to 0.5 and found no significant changes in the results . the number of absorbers randomly distributed on each sight line was set equal to the number in the observed sample , and we ran 1000 simulations for each qso . we then determined the three - dimensional nearest neighbor distance for each real absorber - galaxy , random absorber - galaxy , and galaxy- galaxy pair as discussed above and plotted their cumulative nearest neighbor distributions . the results for the total and complete samples are shown in figure [ cumdist ] ( see 5.1 for the sample definitions ) . the cumulative percentage on the y - axis in this figure indicates the fraction of objects ( absorbers , galaxies , and random absorbers ) with nearest neighbor distances equal to or less than the corresponding value on the x - axis . the results of kolmogorov - smirnov ( ks ) tests which compare the nearest neighbor distributions for all of the samples defined in 5.1 are summarized in table [ kstests ] . column 6 of table [ kstests ] shows that _ the probability that the clouds are drawn from a random distribution is extremely small for all samples , _ in agreement with the previous conclusions of morris et al . ( 1993 ) and stocke et al . the `` complete '' and `` 1 mpc complete '' samples yield similar results , which indicates that the limited spatial coverage of the redshift survey at low @xmath0 does not have a severe impact . table [ kstests ] also shows that for the complete sample , the galaxy - galaxy and galaxy - distributions are statistically difficult to distinguish . however , this does not necessarily indicate that the absorbers are clustered as strongly as galaxies in general because the galaxy clustering implied by the galaxy - galaxy nearest neighbor distribution in figure [ cumdist ] is underestimated due to incompleteness of the redshift surveys . we have also divided our complete sample into `` weaker '' and `` stronger '' absorbers based on the line equivalent widths . figure [ cumweak ] compares the resulting nearest neighbor distribution of the lines with @xmath5 100 m to the nearest neighbor distribution of the lines with @xmath15 100 m . we selected 100 m as the cutoff between the weak and strong absorbers because this divides the sample roughly in half ( 28 weak vs. 21 strong absorbers ) and because this is the cutoff used by stocke et al . table [ ksweak ] lists the ks probabilities that the weak and strong absorbers are drawn from a random distribution . figure [ cumweak ] and table [ ksweak ] clearly show that _ neither the weak absorbers nor the strong absorbers are randomly distributed with respect to the galaxies in the survey . _ figure [ cumweak ] also shows that the weaker absorbers are difficult to distinguish from the stronger absorbers , at least based on the nearest neighbor distribution with 100 m delineating weak lines from strong . however , previous studies indicate that the cutoff between the hypothesized two populations of absorbers ( see 1 ) may be closer to 300 m than 100 m . unfortunately , we have very few lines in our sample with @xmath25 300 m , so we can not apply the same test using 300 m as the cutoff . however , the sparse data that we do have suggest that there is a difference between absorbers with @xmath25 300 m and weaker absorbers , as expected from the anticorrelation discussed in 5.2 . for example , four out of six lines in the h 1821 + 643 and pg 1116 + 215 spectra with @xmath25 300 m ( not including lines within 3000 of the qso redshift ) have a nearby galaxy with @xmath155 200 kpc and @xmath137 300 , and the fifth has a galaxy at @xmath143 = 373 kpc and @xmath156 = 246 . the only line with @xmath25 300 m that does not have a nearby galaxy is at @xmath64 0.2 toward h 1821 + 643 where the galaxy redshift survey is very incomplete . in contrast , out of the 41 remaining lines with @xmath26 300 m , only two have @xmath155 200 kpc and @xmath137 300 , and these are both within 350 of a strong line with @xmath25 300 m ( see tables [ h1821near ] and [ p1116near ] ) . if we divide the sample into lines with @xmath25 200 m and @xmath26 200 m , we obtain the nearest neighbor distributions shown in figure [ weak200 ] and the statistical results in the lower half of table [ ksweak ] . while figure [ weak200 ] does suggest that there is a difference between the weaker and stronger lines , the difference has a marginal statistical significance ; a ks test indicates that there is a 6.5% probability that the weaker and stronger absorbers in figure [ weak200 ] are drawn from the same distribution . however , there are only 9 lines in the sample with with @xmath25 200 m . clearly more observations are needed . in this section we discuss the implications of our results for the relationship between absorbers and galaxies . to facilitate the discussion , we first summarize the main results of our study in the context of earlier studies : \(1 ) the distribution of absorbers in space is not random but rather is correlated with the distribution of galaxies as found previously ( morris et al . 1993 ; lanzetta et al . 1995 ; stocke et al . 1995 ; lanzetta et al . a new result of this work is that , because of our larger sample of weak absorption lines , we find that even the very weak absorbers ( @xmath15 100 m ) are correlated with the galaxy distribution . on the other hand , the correlation between galaxies and absorbers is not as strong as the correlation between galaxies themselves ( morris et al . 1993 ; mo & morris 1994 ; stocke et al . 1995 ; lanzetta et al 1997 ) . there appears to be marginal evidence that the weak absorption lines are less correlated with the galaxy distribution than the strong absorption lines , but the difference is not yet significant . \(2 ) earlier work by lanzetta and co - workers ( lanzetta et al . 1995 , 1997 ) showed that luminous galaxies at impact parameters @xmath7 200 kpc frequently have corresponding absorption lines . they also suggested that galaxies at impact parameters greater than 200 kpc are almost never associated with corresponding absorption lines . however , we find that all galaxies within 600 kpc of the quasar sightlines do have corresponding absorption with @xmath137 1000 , and a significant number of galaxies at 600 @xmath157 2200 kpc have corresponding absorption lines with @xmath137 1000 . furthermore , most of these galaxy - absorber pairs are separated by less than @xmath3350 . these different conclusions result mostly from the fact that the absorption lines tend to be weak for galaxies with large impact parameters , and the fos spectra used by lanzetta et al . ( 1997 ) are not sufficiently sensitive to detect these weak lines . \(3 ) it was shown by lanzetta and co - workers ( lanzetta et al . 1995,1997 ; chen et al . 1998 ) that the equivalent widths of the absorption lines are anticorrelated with the impact parameters to galaxies for impact parameter up to 200 kpc and @xmath111 up to 500 . we find that the same relation extends to impact parameters as large as @xmath32 mpc for @xmath137 1000 . in addition , there is no noticeable increase in the scatter around the mean anticorrelation relation going from small @xmath143 to large @xmath143 ( see figure [ bigrhocor ] ) , while increasingly large scatter with increasing @xmath143 might be expected if the absorbers and the galaxies are physically related at small impact parameters but are less physically related at larger projected distances . however , there are potential selection effects which can artificially tighten this anticorrelation , and follow - up studies are needed to determine whether or not it is real . \(4 ) some prominent groups of galaxies do not have clearly associated lines . however other galaxy groups do have associated absorbers . evidently absorbers do not always avoid regions of higher galaxy density , but they may avoid the densest regions . \(5 ) dividing the absorbers into `` void '' and `` non - void '' absorbers based on whether or not there is a galaxy within 3 mpc in 3-dimensional space , assuming a perturbed hubble flow for the determination of radial distance , we find that the equivalent width distributions of void and non - void absorbers are statistically indistinguishable . there are two main competing interpretations of the relationship between galaxies and absorbers at low redshifts . lanzetta and co - workers ( e.g. , lanzetta et al . 1995,1997 ; chen et al . 1998 ) have suggested that low redshift absorbers ( mostly strong absorption lines with @xmath5 300 m ) are physically associated with individual galaxies which are surrounded by extended gaseous envelopes of @xmath3200 kpc radius of unknown geometry which give rise to the observed absorption . two main arguments are cited in support of the physical association interpretation : ( 1 ) the galaxy - absorber cross - correlation function appears to be strong for @xmath147 200 kpc and @xmath137 500 and is much weaker at larger impact parameters and velocity separations ( lanzetta et al . 1997 ) ; ( 2 ) there is an anticorrelation between absorber equivalent width and impact parameter for galaxy - absorber pairs that are likely to be physically associated ( chen et al . 1998 ) , where `` physically associated '' pairs refer to those with galaxy - absorber cross - correlation significantly above zero ( essentially galaxy - absorber pairs with @xmath147 200 kpc and @xmath137 500 ) . the latter point suggests that the density of the gas distribution near galaxies drops off with increasing galactocentric radius . some independent support for this picture is provided by zaritsky et al . ( 1997 ) who conclude , based on the velocity dispersion of satellites around isolated spiral galaxies , that the dark matter halos of spirals have radii @xmath158 200 kpc . however , there are difficulties with the gaseous halo interpretation ( see , e.g. , 5 in stocke et al . 1995 ) . it is difficult to understand how galaxies would be able to maintain such large gaseous halos such halos would probably be weakly bound and vulnerable to disruption by interactions in groups and clusters . our finding that the equivalent width versus impact parameter anticorrelation extends to impact parameters as large as @xmath3600 kpc and may extend to several mpc further aggrevates the problem and challenges the lanzetta et al . interpretation . in addition , there are several instances in our sample and elsewhere ( e.g. , rauch et al . 1996 ) in which a group of galaxies within a few 100 of each other with separations comparable to the impact parameter to the nearest galaxy appear to be responsible for the same absorption . as noted by rauch et al . ( 1996 ) , the `` cloud size '' or `` coherence length '' of absorbers inferred by dinshaw et al . ( 1995,1997 ) from common lines detected toward qso pairs , @xmath3700 kpc , is much larger than the transverse separation between galaxies so it is doubtful that the large cloud size could refer to absorption produced by a single galactic halo or disk . rather , the large transverse sizes and small velocity differences of common absorbers measured by dinshaw et al . appear to favor spatially coherent and rather quiescent gaseous structures on scales significantly larger than the virial radii of individual galaxies . an alternative to the lanzetta et al . interpretation is that , rather than being associated with individual galaxies , the absorbers trace the large scale structures in which both galaxies and gas reside . recent cosmological simulations of structure formation involving gas hydrodynamics suggest a picture in which galaxies are surrounded and connected by filamentary or sheet - like gaseous structures . such structures have properties which are consistent with the absorption lines at @xmath0 = 2- 4 , as has been shown by a number of recent studies ( e.g. , cen et al . 1994 ; petitjean , mcket , & kates 1995 ; miralda - escud et al . 1996 ; hernquist et al . 1996 ; zhang et al . 1997,1998 ) . in these simulations , the density of the gas distribution drops with increasing distance from galaxies , and often even void regions contain enough gas to produce absorption lines , albeit with much lower column densities . it is reasonable to expect that the overall structure of the gas distribution at @xmath0 = 0 will remain similar to that at @xmath159 2 , although the amount of gas in void regions and in the filamentary structures should decrease as the gas `` drains '' through the filaments into galaxies . the association of absorbers with large scale structures appears to be consistent with all of the observational evidence obtained so far for low @xmath0 absorbers . the distribution of absorbers is obviously correlated with that of galaxies because they trace the same large scale structures , but clearly the absorber - galaxy correlation function will be weaker than the galaxy - galaxy correlation function . since most ( if not all ) galaxies are likely to be engulfed by diffuse gas in the filamentary structures , an absorption line is likely to be found when a galaxy is near the qso sightline ( result ( 2 ) above ) . the anticorrelation between absorber equivalent width and impact parameter and the persistence of such an anticorrelation to very large impact parameters can also be understood in analogy with the simulations at large redshift ; the density of the gas distribution is expected to decrease on average as one moves away from galaxies . however , since the gas distribution is not regular and smooth , a large scatter in this anticorrelation relation is expected , as observed ( see figures [ rhocor ] and [ bigrhocor ] ) . chen et al . ( 1998 ) attemped to reduce the scatter in the equivalent width vs impact parameter relation by incorporating various galaxy properties as a secondary parameter , such as galaxy luminosity , mean surface brightness , effective radius , disk - to - bulge ratio , redshift , and shape of the gas distribution ; only marginal improvement in the relation was found even in the best case when galaxy b - band luminosity was incorporated into the relation to 4.2 - 5.9@xmath4 , depending on the statistical test applied . ] . the apparent insensitivity of this relation to galaxy properties is naturally explained if the absorption arises from the overall large scale structure rather than in individual galaxies . some absorption lines should also occur when the quasar sightlines pass through regions in the large scale structure that are relatively far ( many mpc ) away from galaxies , where there may still be gas either in filaments or in the voids . this is consistent with result ( 5 ) , namely , the equivalent width distributions of void and non - void absorbers are statistically indistinguishable . however , the absorption lines arising from regions in space far away from galaxies should , to some degree , be weaker on average . a much larger line sample is required to search for such a trend . a third interpretation of the absorbers is that the lines are associated with faint dwarf galaxies which may be able to generate large ( @xmath3100 kpc ) gaseous halos via supernova - driven winds ( wang 1995 ; nath & trentham 1997 ) . at low @xmath0 , this interpretation has gained support from the 21 cm imaging study of van gorkom et al . ( 1996 ) which has located gas - rich dwarf galaxies at the redshift of some low @xmath0 lines . there are some potential problems with this interpretation . first , the dwarf galaxies are usually more than @xmath3100 kpc from the sight lines . several sensitive imaging searches for very faint galaxies in the vicinity of the two lowest - redshift absorbers toward 3c 273 have failed to find faint galaxies within 100 kpc of the sight line ( e.g. , morris et al . 1993 ; van gorkom et al . 1993 ; rauch et al . 1996 ) . by searching a somewhat larger field , salzer ( 1992 ) and hoffman et al . ( 1998 ) have found two dwarf galaxies at projected distances of @xmath3200 kpc near the lowest redshift 3c 273 absorber . hoffman et al . ( 1998 ) have measured the rotation curve of one of these dwarfs , and they show that extension of the observed rotation curve can not explain the velocity of the associated absorber ( see their figure 6 ) , but this may not be surprising if the line is due to a galactic wind . a second criticism of the dwarf galaxy model of absorbers stems from the recent study of martin ( 1998 ) . martin ( 1998 ) has explored the likelihood that dwarf galaxies will drive galactic winds by surveying the expansion rates and sizes of their h@xmath16 shells . she concludes that `` many of the shells will breakthrough the surrounding layer supersonically , but the projected expansion speeds are typically less than the lower limits on the escape velocity . '' martin notes that in a few objects in her sample ( e.g. , ngc 1569 , see also heckman et al . 1995 ) , an unbound galactic wind is likely , but since she has exclusively selected dwarfs with prominent shells and only @xmath325% of dwarf galaxies show prominent h@xmath16 filaments ( hunter , hawley , & gallagher 1993 ) , her study seems to suggest that the majority of nearby dwarf galaxies will not drive winds . since many recent studies have shown that galaxy clusters and large scale structures contain large numbers of dwarf galaxies ( e.g. , phillipps et al . 1998 , and references therein ) and since gallagher , littleton , & matthews ( 1995 ) and phillipps et al . have shown that in these structures the distribution of the dwarfs tends to be more spatially extended than that of the giant galaxies , it seems more likely that lines with dwarfs 100 - 200 kpc away are due to gas in the large scale structures where the dwarfs are found rather than the dwarfs themselves . the hypothesis that many low redshift absorption lines with @xmath160 in the range from 50 to @xmath3500 m trace the overall gas distribution in the large scale structures rather than gaseous halos of individual galaxies is physically plausible and is consistent with existing observations . however , it is probably true that a variety of phenomena cause absorption lines in qso spectra . the gaseous disks and halos of individual galaxies will certainly produce absorption if the qso sight line passes close enough , and these will probably be stronger lines . additional studies at low redshift should ultimately allow the determination of all sites of origin for this important tracer of cosmic phenomena . we have obtained high signal - to - noise uv spectra of the qsos h 1821 + 643 ( @xmath161 = 0.297 ) and pg 1116 + 215 ( @xmath161 = 0.177 ) with the ghrs in order to study the nature of low @xmath0 absorbers and their relationship with galaxies . the spectra have s / n ranging from 75 to 160 per 150 resolution element and permit the detection of weak lines . toward h 1821 + 643 and pg 1116 + 215 we find 26 and 13 lines , respectively , with 3@xmath4 rest equivalent widths @xmath5 50 m ( not including lines within 3000 of @xmath2 ) . the previous best studied sight line for lines as weak as 50 m , 3c 273 , showed 13 lines with @xmath5 50 m at 3@xmath4 significance or better ( morris et al . we have also measured 98 and 118 galaxy redshifts in the @xmath8 fields centered on h 1821 + 643 and pg 1116 + 215 with the multiobject spectrograph on the wiyn telescope . with 56 additional redshifts from the literature , we have a total of 154 redshifts in the h 1821 + 643 field . within a 20 radius from the qso , we estimate that our h 1821 + 643 and pg 1116 + 215 redshift surveys are respectively 72.4% and 85.0% complete for @xmath60 18.0 and 50.8% and 68.5% complete for @xmath60 20.0 . \(2 ) several of the stronger profiles show complex multiple component structure , but there is no compelling evidence of clustering in the two - point velocity correlation function of the whole sample . however , if we exclude the weakest lines ( @xmath15 100 m ) , the two - point correlation function shows a marginal signal ( 3@xmath4 ) on velocity scales of @xmath3500 . \(3 ) all galaxies within projected distances of 600 @xmath9 kpc have associated absorbers with @xmath137 1000 , and the equivalent widths are anticorrelated at high significance with the projected distance of the nearest galaxy out to @xmath12 600 @xmath9 kpc . further studies are needed to establish whether this anticorrelation is real or due to selection effects . for @xmath163 kpc , we find galaxies which do not have associated lines . however , the anticorrelation continues to @xmath164 mpc for galaxies within 500 or 1000 of a absorber . most of these galaxy - absorber pairs have @xmath165 350 . \(5 ) as in previous studies , we find lines in voids at large distances from the nearest galaxies , but this may be due to the limited spatial coverage ( at low @xmath0 ) or limited depth ( at high @xmath0 ) of the galaxy redshift survey . statistically , the equivalent width distributions of the void and non - void absorbers are indistinguishable , but the sample is small . \(6 ) statistical tests show that the absorbers are not randomly distributed with respect to the galaxies . splitting the sample into roughly equal groups of weak ( @xmath15 100 m ) and strong absorbers shows that the weak absorbers are also not randomly distributed . comparison of the nearest neighbor distances of the weak and strong absorbers suggests that the weak absorbers may be less closely associated with galaxies than the strong absorbers , but with the small sample this difference is not yet statistically significant . \(7 ) the observations are consistent with the hypothesis that many of the low redshift absorbers with rest equivalent widths in the range from 50 to @xmath3500 m trace the overall gas distributions in the large scale structures in which galaxies reside rather than the gaseous halos of individual galaxies . however , it is likely that not all lines are produced in this way . some of the stronger lines may arise in the halos of individual galaxies . the galaxy redshift survey employed in this paper is a collaboration with buell jannuzi and coworkers , and we thank this group , especially buell and angelle tanner , for allowing us to use these redshifts in advance of publication . we acknowledge michael rauch for helpful discussions , and we thank r. carswell for sharing his profile fitting software vpfit . we also thank jay gallagher , ed jenkins , and especially john stocke for valuable comments on the manuscript . our observational program has utilized data from the _ hubble space telescope _ and the wiyn 3.5 m optical observatory on kitt peak . we very much appreciate the work of the many people who helped create and operate these facilities . and t.m.t . appreciate support from nasa through grant go-06499.02 - 95a . l.l . appreciates support through nasa funded hubble fellowship hf 1062.01- 94a . jason cardelli made many contributions to the success of the goddard high resolution spectrograph on _ hst _ , and some of jason s work played an important role in this paper . sadly , jason died suddenly on may 14 , 1996 at the age of 40 . we dedicate this paper to jason . bahcall , j. n. , et al . 1993 , apjs , 87 , 1 bahcall , j. n. , jannuzi , b. t. , schneider , d. p. , hartig , g. f. , bohlin , r. , & junkkarinen , v. 1991 , apj , 377 , l5 bahcall , j. n. , jannuzi , b. t. , schneider , d. p. , hartig , g. f. , & green , r. f. 1992 , apj , 397 , 68 barlow , t. a. , & sargent , w. l. w. 1997 , aj , 113 , 136 bi , h. , & davidsen , a. f. 1997 , apj , 479 , 523 bowen , d. v. , blades , j. c. , & pettini , m. 1996 , apj , 464 , 141 bowen , d. v. , pettini , m. , & boyle , b.j . 1998 , mnras , submitted brandt , j. c. et al . 1997 , aj , 114 , 554 bruhweiler , f. c. , boggess , a. , norman , d. j. , grady , c. a. , urry , c. m. , & kondo , y. 1993 , apj , 409 , 199 burns , j. o. 1998 , science , 280 , 400 cardelli , j. a. , & ebbets , d. c. 1994 , in calibrating hubble space telescope : proceedings of a workshop held at stsci , ed . j. c. blades & s. j. osmer ( baltimore : stsci ) , 322 carswell , r. f. , lanzetta , k. m. , parnell , h. c. , & webb , j. k. 1991 , apj , 371 , 36 cen , r. , miralda - escud , j. , ostriker , j. p. , & rauch , m. 1994 , apj , 437 , l9 chen , h .- w . , lanzetta , k. m. , webb , j. k. , & barcons , x. 1998 , apj , 498 , 77 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , buta , r. j. , paturel , g. , & fouque , p. 1991 , third reference catalog of bright galaxies ( springer , new york ) dinshaw , n. , foltz , c. b. , impey , c. d. , weymann , r. j. , & morris , s. l. 1995 , nature , 373 , 223 dinshaw , n. , weymann , r. j. , impey , c. d. , foltz , c. b. , morris , s. l. , & ake , t. 1997 , apj , 491 , 45 gallagher , j. s. iii , littleton , j. e. , & matthews , l. d. 1995 , aj , 109 , 2003 grogin , n. a. , & geller , m. j. 1998 , apj , in press hall , p. b. , ellingson , e. , & green , r. f. 1997 , aj , 113 , 1179 hamann , f. , barlow , t. a. , & junkkarinen , v. 1997a , apj , 478 , 87 hamann , f. , barlow , t. a. , junkkarinen , v. , & burbidge , e. m. 1997b , apj , 478 , 80 heckman , t. m. , dahlem , m. , lehnert , m. d. , fabbiano , g. , gilmore , d. , & waller , w. h. 1995 , apj , 448 , 98 hernquist , l. , katz , n. , weinberg , d. h. , & miralda- escud , j. 1996 , apj , 457 , l51 hoffman , g. l. , lu , n. y. , salpeter , e. e. , connell , b. m. , & fromhold - treu , r. 1998 , apj , in press hu , e. m. , kim , t .- s . , cowie , l. l. , songaila , a. , & rauch , m. 1995 , aj , 110 , 1526 referencehunt93 hunter , d. a. , hawley , w. n. , & gallagher , j. s. 1993 , aj , 106 , 1797 jannuzi , b. t. et al . 1998 , apjs , 118 , in press kirhakos , s. , sargent , w. l. w. , schneider , d. p. , bahcall , j. n. , jannuzi , b. t. , maoz , d. , & small , t. a. 1994 , pasp , 106 , 646 kirkman , d. , & tytler , d. 1997 , apj , 484 , 672 lanzetta , k. m. , bowen , d. v. , tytler , d. , & webb , j. k. 1995 , apj , 442 , 538 lanzetta , k. m. , webb , j. k. , & barcons , x. 1996 , apj , 456 , l17 lanzetta , k. m. , webb , j. k. , & barcons , x. 1997 , in proceedings of the 13th iap colloquium , structure and evolution of the intergalactic medium from qso absorption lines , ed . p. petitjean & s. charlot , ( paris : editions frontieres ) , 213 le brun , v. , bergeron , j. , & boiss , p. 1996 , a&a , 306 , 691 linder , s. , 1998 , apj , 495 , 637 lockman , f. j. , & savage , b. d. 1995 , apjs , 97 , 1 loveday , j. , peterson , b. a. , efstathiou , g. , & maddox , s. j. 1992 , apj , 390 , 338 lu , l. , sargent , w. l. w. , womble , d. s. , & takada - hidai , m. 1996 , apj , 472 , 509 lu , l. , savage , b. d. , & sembach , k. r. 1994 , apj , 426 , 563 lu , l. , wolfe , a. m. , & turnshek , d. a. 1991 , apj , 367 , 19 martin , c. 1998 , apj , in press miralda - escud , j. , cen , r. , ostriker , j. p. , & rauch , m. 1996 , apj , 471 , 582 mo , h. j. , & morris , s. l. 1994 , mnras , 269 , 52 morris , s. l. , & van den bergh , s. 1994 , apj , 427 , 696 morris , s. l. , weymann , r. j. , dressler , a. , mccarthy , p. j. , smith , b. a. , terrile , r. j. , giovanelli , r. , & irwin , m. 1993 , apj , 419 , 524 morris , s. l. , weymann , r. j. , savage , b. d. , & gilliland , r. l. 1991 , apj , 377 , l21 morton , d. c. 1991 , apjs , 77 , 119 morton , d. c. , york , d. g. , & jenkins , e. b. 1988 , apjs , 68 , 449 nath , b. b. , & trentham , n. 1997 , mnras , 291 , 505 peebles , p. j. e. 1993 , principles of physical cosmology ( princeton : princeton univ . press ) , 331 petitjean , p. , mcket , j. p. , & kates , r. e. 1995 , a&a , 295 , l9 phillipps , s. , driver , s. p. , couch , w. j. , & smith , r. m. 1998 , 498 , l119 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992 , ( cambridge : cambridge university press ) , 634 rauch , m. , et al . 1997 , apj , 489 , 7 rauch , m. , weymann , r. j. , & morris , s. l. 1996 , apj , 458 , 518 riediger , r. , petitjean , p. , & mcket , j. p. 1998 , a&a , 329 , 30 robinson , r. d. , et al . 1998 , pasp , 110 , 68 roettiger , k. , stone , j. m. , & mushotzky , r. f. 1998 , apj , 493 , 62 salzer , j. j. 1992 , aj , 103 , 385 savage , b. d. , cardelli , j. a. , & sofia , u. j. 1992 , apj , 401 , 706 savage , b. d. , & sembach , k. r. 1996 , apj , 470 , 893 savage , b. d. , sembach , k. r. , & lu , l. 1995 , apj , 449 , 145 savage , b. d. , sembach , k. r. , & lu , l. 1997 , aj , 113 , 2158 savage , b. d. , tripp , t. m. , & lu , l. 1998 , aj , 115 , 436 schneider , d. p. , bahcall , j. n. , gunn , j. e. , & dressler , a. 1992 , aj , 103 , 1047 shull , j. m. 1997 , in proceedings of the 13th iap colloquium , structure and evolution of the intergalactic medium from qso absorption lines , ed . p. petitjean & s. charlot , ( paris : editions frontieres ) , 101 shull , j. m. , stocke , j. t. , & penton , s. 1996 , aj , 111 , 72 stocke , j. t. , shull , j. m. , penton , s. , donahue , m. , & carilli , c. 1995 , apj , 451 , 24 tonry , j. , & davis , m. 1979 , aj , 84 , 1511 tripp , t. m. 1997 , phd dissertation , university of wisconsin - madison tripp , t. m. , jannuzi , b. t. et al . 1998 , in preparation tripp , t. m. , lu , l. , & savage , b. d. 1996 , apjs , 102 , 239 tyson , j. a. , & jarvis , j. f. 1979 , apj , 230 , l153 ulmer , a. 1996 , apj , 473 , 110 valdes , f. 1982 , in spie , instrumentation in astronomy iv , 331 , 465 van gorkom , j. h. , bahcall , j. n. , jannuzi , b. , & schneider , d. 1993 , aj , 106 , 2213 van gorkom , j. h , carilli , c. l. , stocke , j. t. , perlman , e. s. , & shull , j. m. 1996 , aj , 112 , 1397 wang , b. 1995 , apj , 444 , l17 weinberg , d. h. , miralda - escud , j. , hernquist , l. , & katz , n. 1997 , apj , 490 , 564 weir , n. , djorgovski , s. , & fayyad , u. m. 1995 , aj , 110 , 1 weymann , r. j. et al . 1998 , apj , submitted zaritsky , d. , smith , r. , frenk , c. , & white , s. d. m. 1997 , apj , 478 , 39 zhang , y. , anninos , p. , norman , m. l. , & meiksin , a. 1997 , 485 , 496 zhang , y. , meiksin , a. , anninos , p. , & norman , m. l. 1998 , apj , in press llccccc 18:17:12.35 & 64:33:56.0 & 0.02089 & 12 & & & 0.78818:21:41.17 & 63:51:38.0 & 0.02404 & 17 & 17.4 & -17.8 & 0.78318:21:27.36 & 64:12:11.3 & 0.02752 & 12 & 20.6 & -14.9 & 0.27718:23:25.35 & 64 : 8:34.2 & 0.05003 & 48 & 15.6 & -21.2 & 0.83218:21:49.15 & 64:37:49.0 & 0.05024 & 36 & 15.8 & -21.0 & 0.93618:22:31.72 & 64:35:29.9 & 0.05051 & 49 & 16.7 & -20.2 & 0.83818:18:14.76 & 64:18:23.5 & 0.05059 & 25 & 18.2 & -18.6 & 1.32218:21:52.71 & 64:39 : 0.1 & 0.05062 & 16 & 17.5 & -19.4 & 1.00618:16:24.99 & 64 : 5:11.9 & 0.05062 & 18 & 18.9 & -18.0 & 2.14718:22:41.28 & 64:10:27.8 & 0.05067 & 12 & 18.0 & -18.8 & 0.61318:22:50.89 & 64 : 9:20.2 & 0.05079 & 55 & & & 0.69618:24:21.41 & 64 : 6:26.3 & 0.05090 & 48 & 15.5 & -21.3 & 1.16118:23:42.37 & 64:32:54.2 & 0.05231 & 35 & 17.9 & -19.0 & 0.94318:24:33.29 & 64:16:14.5 & 0.05245 & 31 & 17.9 & -19.0 & 0.98718:16:31.37 & 64:43:54.6 & 0.06553 & 36 & 16.9 & -20.5 & 2.91618:22:30.84 & 64 : 4:13.7 & 0.07103 & 30 & 17.6 & -20.0 & 1.25118:24:33.51 & 64 : 6:21.0 & 0.07149 & 34 & 18.9 & -18.7 & 1.66318:20:56.89 & 64:27:54.5 & 0.07170 & 27 & 18.3 & -19.3 & 0.73618:27:46.25 & 64:14:28.2 & 0.07177 & 73 & 17.0 & -20.6 & 2.88618:25:46.50 & 64 : 7:56.7 & 0.07192 & 44 & 18.4 & -19.2 & 2.10718:25:43.04 & 64:53:37.3 & 0.07608 & 29 & 18.4 & -19.3 & 3.24618:18:17.51 & 64:34:37.3 & 0.07843 & 37 & 18.7 & -19.1 & 2.24218:26:16.03 & 63:54 : 5.7 & 0.08343 & 28 & 18.2 & -19.7 & 3.33418:24:29.24 & 64 : 1 : 5.2 & 0.08432 & 37 & 18.8 & -19.1 & 2.22418:23:38.83 & 64:41:47.7 & 0.08796 & 41 & & & 2.15118:22:29.42 & 64:41 : 6.7 & 0.08822 & 40 & 17.7 & -20.3 & 1.88118:23:44.86 & 64:42:47.0 & 0.08834 & 45 & 18.5 & -19.5 & 2.26618:19:44.77 & 64:23 : 7.3 & 0.08846 & 100 & 17.7 & -20.3 & 1.31918:20:42.96 & 64:19:45.8 & 0.08930 & 67 & 19.4 & -18.7 & 0.73918:18:35.52 & 64:14:22.3 & 0.09532 & 57 & 18.1 & -20.1 & 2.20118:25:52.36 & 64:40 : 7.5 & 0.09574 & 57 & 17.8 & -20.4 & 3.10518:24:28.05 & 64:16:42.7 & 0.09701 & 66 & 18.4 & -19.8 & 1.65118:23:29.55 & 64:40:22.6 & 0.10614 & 96 & 18.8 & -19.7 & 2.35118:24:40.32 & 64:34:54.8 & 0.10642 & 35 & 17.8 & -20.6 & 2.41318:21:36.59 & 64:45:23.6 & 0.10682 & 32 & 17.8 & -20.7 & 2.65818:20 : 7.54 & 65 : 4 : 0.3 & 0.10689 & 38 & 17.4 & -21.1 & 4.80418:25:53.45 & 64:19:19.7 & 0.11156 & 79 & 18.8 & -19.8 & 2.83918:20 : 8.65 & 64:21 : 4.2 & 0.11208 & 102 & 20.2 & -18.3 & 1.30818:25 : 4.45 & 64:25:22.9 & 0.11957 & 99 & 18.9 & -19.8 & 2.44518:21 : 0.88 & 64 : 3:55.7 & 0.12120 & 105 & & & 2.11218:22 : 2.65 & 64:21:39.3 & 0.12154 & 70 & 18.3 & -20.5 & 0.14418:21:22.38 & 64:28:29.0 & 0.12224 & 34 & 18.5 & -20.2 & 1.04618:23:58.19 & 64:26:52.3 & 0.12258 & 76 & 17.3 & -21.5 & 1.74018:21:20.26 & 64 : 3:44.8 & 0.13774 & 59 & 18.5 & -20.5 & 2.29118:21:57.42 & 63:52:44.3 & 0.14220 & 206 & & & 3.78218:28:15.51 & 64:33:39.5 & 0.14228 & 70 & 18.3 & -20.8 & 5.81618:22 : 4.38 & 64 : 8:38.5 & 0.15465 & 59 & 18.9 & -20.4 & 1.74118:22:49.76 & 64:19:28.6 & 0.16376 & 100 & & & 0.88218:16:24.55 & 64 : 2 : 6.2 & 0.16556 & 94 & 18.2 & -21.2 & 6.23818:16:43.64 & 64:34:56.2 & 0.16564 & 39 & 17.9 & -21.6 & 5.63518:21:36.61 & 64:21:25.0 & 0.17086 & 74 & 19.1 & -20.4 & 0.37318:19:24.36 & 64 : 7:53.6 & 0.17919 & 100 & 18.4 & -21.2 & 3.41418:19:55.22 & 64:35 : 8.4 & 0.18012 & 56 & 18.9 & -20.7 & 3.21318:21:14.82 & 64:12:16.4 & 0.18504 & 35 & 20.9 & -18.7 & 1.59218:20:24.54 & 63:56:31.0 & 0.18532 & 97 & 19.4 & -20.2 & 4.37718:20:21.11 & 63:59 : 1.2 & 0.18547 & 39 & 19.3 & -20.4 & 4.02218:19:45.65 & 64:30:52.5 & 0.18596 & 95 & 19.1 & -20.5 & 2.94518:20:25.56 & 64:23:59.0 & 0.18599 & 52 & 21.0 & -18.7 & 1.76018:20:25.99 & 64:14:57.5 & 0.18742 & 63 & & & 1.92518:20:43.29 & 65 : 0:15.1 & 0.18921 & 40 & 18.6 & -21.1 & 6.88518:22:25.14 & 64:22:46.2 & 0.18967 & 40 & 19.9 & -19.8 & 0.63518:22:53.12 & 64:55:24.0 & 0.19117 & 56 & 19.1 & -20.7 & 6.06018:21:18.06 & 63:53:47.8 & 0.19161 & 39 & 18.7 & -21.0 & 4.66718:22:29.69 & 64:23 : 8.2 & 0.19162 & 53 & 19.1 & -20.6 & 0.74518:22:20.96 & 64:26:52.2 & 0.19212 & 35 & 18.1 & -21.6 & 1.16718:23:59.02 & 64:25:58.7 & 0.19213 & 59 & 18.8 & -21.0 & 2.45018:22:14.04 & 64 : 3 : 5.4 & 0.20057 & 54 & 18.9 & -21.0 & 3.13418:21:38.82 & 64:20:31.7 & 0.22650 & 32 & 19.8 & -20.3 & 0.38818:23:47.60 & 64 : 3:16.8 & 0.24433 & 49 & 19.6 & -20.7 & 4.34018:21:56.28 & 64:22:50.4 & 0.24435 & 64 & 20.2 & -20.1 & 0.46218:22:19.11 & 64:18:43.5 & 0.24568 & 86 & 20.8 & -19.5 & 0.62518:20:31.50 & 64:20:24.0 & 0.25147 & 72 & 19.8 & -20.5 & 1.95018:22:10.26 & 64:17:15.6 & 0.26669 & 80 & & & 0.79518:24 : 4.97 & 63:52:38.0 & 0.27548 & 116 & 19.1 & -21.5 & 7.00118:20:31.98 & 64:22:19.7 & 0.27889 & 114 & 20.1 & -20.4 & 2.11918:22:30.01 & 64:13:16.4 & 0.27987 & 55 & 20.6 & -19.9 & 1.84518:23:50.02 & 63:52 : 3.2 & 0.28420 & 100 & 19.5 & -21.1 & 7.11518:22 : 0.23 & 64:18:48.9 & 0.28787 & 62 & 21.1 & -19.5 & 0.41918:21:48.70 & 64:24:20.4 & 0.28858 & 78 & 20.9 & -19.7 & 0.89118:21:39.29 & 64:22 : 4.8 & 0.28880 & 92 & 20.5 & -20.2 & 0.56418:21:34.40 & 64:20:31.0 & 0.29044 & 82 & 21.4 & -19.2 & 0.57418:21:53.32 & 64:20:23.7 & 0.29056 & 83 & & & 0.10918:21:58.42 & 64:18:24.1 & 0.29116 & 89 & 20.9 & -19.7 & 0.51218:21:19.97 & 64:22:57.1 & 0.29216 & 53 & 20.2 & -20.5 & 1.08818:22:19.84 & 64:23:36.1 & 0.29305 & 79 & 20.7 & -20.0 & 0.90518:21:44.81 & 64:24:11.2 & 0.29359 & 97 & 20.3 & -20.4 & 0.89618:22:33.10 & 64:18:50.8 & 0.29481 & 54 & 21.1 & -19.6 & 1.00018:21:58.69 & 64:26:45.7 & 0.29552 & 84 & 21.3 & -19.4 & 1.44818:21:37.00 & 64:23:28.6 & 0.29606 & 101 & 20.7 & -20.0 & 0.85018:22:27.25 & 64:18:20.3 & 0.29639 & 74 & & & 0.93318:22 : 3.51 & 64:23 : 0.9 & 0.29766 & 80 & 19.8 & -20.9 & 0.59218:22:12.61 & 64:26:32.9 & 0.29828 & 66 & 20.2 & -20.5 & 1.46018:21:29.76 & 64:27:42.3 & 0.30025 & 50 & 21.0 & -19.7 & 1.82818:21:55.72 & 64:20 : 3.1 & 0.30294 & 55 & & & 0.13718:24:45.87 & 64:33 : 8.2 & 0.30512 & 85 & 19.9 & -20.8 & 5.30218:21:39.49 & 64:15:20.8 & 0.30529 & 50 & 20.7 & -20.0 & 1.34318:22:11.36 & 64:28:50.5 & 0.33069 & 144 & 20.8 & -20.1 & 2.12118:20:33.20 & 64:22:13.1 & 0.33233 & 40 & 20.3 & -20.7 & 2.344 18:21:54.40 & 64:20 : 9.3 & 0.22560 & 150 & 19.5 & -20.6 & 0.10518:21:55.98 & 64:21 : 0.8 & 0.29160 & 150 & 19.8 & -20.9 & 0.10118:21:59.36 & 64:19:49.2 & 0.29360 & 300 & 21.4 & -19.2 & 0.19118:21:54.23 & 64:20:13.4 & 0.29860 & 300 & & & 0.11718:21:52.80 & 64:20:43.9 & 0.30060 & 300 & 21.3 & -19.4 & 0.11818:21:54.97 & 64:21:17.1 & 0.30310 & 300 & 19.0 & -21.7 & 0.174 18:22:20.16 & 64:21:46.7 & 0.17850 & & 20.8 & -18.8 & 0.44718:22 : 4.57 & 64:20:52.5 & 0.28440 & & & & 0.19318:21:30.35 & 64:20:44.8 & 0.29090 & & 20.8 & -19.8 & 0.67718:22:10.61 & 64:20:29.0 & 0.29170 & & & & 0.33918:21:47.72 & 64:20 : 8.9 & 0.29200 & & & & 0.26218:21:55.01 & 64:20 : 4.7 & 0.29300 & & & & 0.13418:21:51.21 & 64:20:52.1 & 0.29380 & & & & 0.16418:22 : 1.07 & 64:20:29.0 & 0.29500 & & & & 0.10218:22:28.30 & 64:20:39.5 & 0.29530 & & & & 0.79118:21:34.30 & 64:21 : 0.9 & 0.29660 & & & & 0.59218:22:13.61 & 64:20:14.5 & 0.29690 & & & & 0.42718:21:44.59 & 64:21:18.5 & 0.29700 & & 19.9 & -20.8 & 0.36318:21:17.91 & 64:21:23.5 & 0.29770 & & & & 1.02218:21:27.64 & 64:21:59.6 & 0.29980 & & & & 0.82818:22:20.16 & 64:20:39.8 & 0.30000 & & & & 0.59018:22:31.94 & 64:19:34.7 & 0.30100 & & & & 0.92718:22:34.80 & 64:20:51.6 & 0.30100 & & & & 0.97018:21:57.96 & 64:20:44.1 & 0.30130 & & 20.9 & -19.8 & 0.03718:22 : 6.59 & 64:19:50.2 & 0.30180 & & & & 0.303 18:23:16.01 & 64 : 4:52.9 & 0.02754 & & 16.6 & -18.9 & 0.55018:20 : 2.47 & 64:18:53.0 & 0.02788 & & 18.5 & -17.0 & 0.39018:22:36.11 & 64:45:44.9 & 0.04970 & & & & 1.37018:22 : 7.61 & 64:39:32.4 & 0.05039 & & 18.1 & -18.7 & 1.03318:24:34.11 & 64:19 : 4.1 & 0.05050 & & 17.2 & -19.7 & 0.93118:23:34.34 & 64:18:35.1 & 0.05099 & & 17.8 & -19.0 & 0.59018:21:21.38 & 64:49:40.4 & 0.05683 & & 18.5 & -18.5 & 1.78418:22:47.01 & 63:59:28.6 & 0.07201 & & 17.3 & -20.3 & 1.64618:23:41.16 & 64 : 1 : 8.3 & 0.07249 & & 17.8 & -19.8 & 1.71018:26 : 1.05 & 64:23:16.4 & 0.07623 & & 17.7 & -20.0 & 2.10618:17:26.18 & 64:19:43.7 & 0.08210 & & 18.6 & -19.3 & 2.49218:22 : 9.95 & 64 : 9:36.9 & 0.08410 & & 18.3 & -19.6 & 0.96018:19:10.44 & 64 : 7:27.4 & 0.09458 & & 18.6 & -19.5 & 2.15218:25:26.87 & 64:36:39.1 & 0.09517 & & 18.4 & -19.8 & 2.67818:25 : 4.56 & 64:40:35.8 & 0.09583 & & 18.4 & -19.8 & 2.76118:24:33.84 & 64:38:51.2 & 0.09605 & & 17.4 & -20.8 & 2.42118:24:40.55 & 64 : 2:45.1 & 0.09681 & & 17.5 & -20.8 & 2.47118:25:20.57 & 64:16:35.7 & 0.09697 & & 18.8 & -19.5 & 2.20118:20:53.45 & 64:19:37.1 & 0.11155 & & 18.6 & -20.0 & 0.77318:22:40.55 & 64 : 9:43.7 & 0.12050 & & 18.5 & -20.2 & 1.40218:18 : 2.25 & 64:14 : 4.2 & 0.12103 & & 17.9 & -20.8 & 3.12418:25 : 8.16 & 64 : 9:17.3 & 0.13853 & & 18.5 & -20.5 & 3.13918:21 : 8.73 & 63:52 : 9.0 & 0.15039 & & 18.4 & -20.8 & 4.11118:19:34.62 & 64:42:21.0 & 0.16410 & & 19.3 & -20.1 & 4.05418:18:28.71 & 64:38 : 3.6 & 0.16550 & & 18.9 & -20.5 & 4.36218:19:40.64 & 64:30:38.5 & 0.17955 & & 18.7 & -20.9 & 2.916 18:18:17.71 & 64:16:15.9 & 0.18002 & & 18.2 & -21.4 & 3.96218:21:35.27 & 64:25:24.0 & 0.18900 & & & & 0.91118:23:15.55 & 64:33 : 5.6 & 0.19188 & & 19.4 & -20.3 & 2.59618:21:25.61 & 64 : 2:11.0 & 0.19406 & & 18.2 & -21.5 & 3.25318:21 : 0.73 & 64:37:54.3 & 0.26650 & & 19.0 & -21.5 & 4.016 llccccc 11:20:46.06 & 21:11:15.2 & 0.02053 & 42 & 17.6 & -17.1 & 0.55711:19:16.67 & 20:48:49.3 & 0.02077 & 34 & 13.8 & -20.9 & 0.71511:21 : 0.27 & 21:20:13.8 & 0.02122 & 26 & 14.4 & -20.3 & 0.62211:18:44.44 & 21:33:51.6 & 0.02128 & 34 & 15.1 & -19.6 & 0.37411:18:45.81 & 21:28:27.9 & 0.02131 & 54 & 16.3 & -18.4 & 0.25411:19 : 1.68 & 21:46 : 6.3 & 0.02140 & 37 & 17.1 & -17.6 & 0.64711:18:22.07 & 21:30 : 4.3 & 0.02153 & 48 & 18.2 & -16.6 & 0.37111:20:38.52 & 21:11:36.8 & 0.02156 & 45 & 18.0 & -16.8 & 0.54111:18:46.12 & 21 : 1:55.3 & 0.02612 & 37 & 18.9 & -16.2 & 0.53111:18:43.36 & 21:27:23.6 & 0.03235 & 44 & 17.1 & -18.5 & 0.35911:20:38.81 & 21 : 0:49.8 & 0.03744 & 35 & 17.1 & -18.8 & 1.15311:21:29.42 & 21:35:32.2 & 0.04107 & 51 & 16.6 & -19.5 & 1.64311:19 : 9.67 & 21 : 2:43.2 & 0.04108 & 47 & 17.9 & -18.2 & 0.74611:19:24.29 & 21:10:30.3 & 0.05916 & 40 & 16.6 & -20.4 & 0.60111:18:33.69 & 21:13 : 0.8 & 0.05963 & 68 & 17.2 & -19.7 & 0.65511:19:59.04 & 21:31:53.8 & 0.05976 & 20 & 18.1 & -18.9 & 1.09611:21:25.04 & 20:58:37.3 & 0.05986 & 43 & 17.9 & -19.1 & 2.42011:16:55.43 & 21 : 8:59.8 & 0.06003 & 51 & 15.6 & -21.4 & 2.09411:21:24.08 & 21:14 : 2.0 & 0.06016 & 38 & 16.8 & -20.1 & 2.05011:19:43.67 & 21:26:52.0 & 0.06055 & 45 & 16.6 & -20.4 & 0.71711:21 : 9.19 & 21:40:43.0 & 0.06056 & 38 & 18.6 & -18.4 & 2.27511:21:29.82 & 21:37 : 7.4 & 0.06076 & 71 & 18.0 & -19.0 & 2.41711:19:42.06 & 21:26:10.7 & 0.06134 & 36 & 19.2 & -17.8 & 0.67711:21:32.33 & 20:42 : 2.3 & 0.06750 & 87 & 17.2 & -20.0 & 3.56811:21:35.88 & 21:26:52.5 & 0.07017 & 56 & 17.9 & -19.4 & 2.58711:21:39.67 & 20:58:11.5 & 0.07048 & 75 & 16.9 & -20.5 & 3.03911:18 : 5.82 & 20:59:42.8 & 0.08076 & 24 & & & 2.04611:19:52.05 & 21:30 : 2.7 & 0.08247 & 53 & 17.4 & -20.3 & 1.25611:20:39.86 & 21 : 0 : 4.4 & 0.08353 & 42 & 18.0 & -19.7 & 2.47011:18:54.00 & 21 : 5:38.1 & 0.08369 & 27 & 17.6 & -20.1 & 1.21611:18:23.11 & 21 : 7:17.4 & 0.08373 & 50 & 16.8 & -20.9 & 1.38511:20:30.00 & 21:31:31.7 & 0.08442 & 49 & 16.7 & -21.0 & 1.96011:18:51.61 & 21 : 4:42.6 & 0.09181 & 91 & 17.5 & -20.4 & 1.41711:19 : 0.09 & 21 : 6:39.8 & 0.09195 & 35 & 16.7 & -21.2 & 1.20011:19:10.35 & 21 : 4:31.2 & 0.09282 & 33 & 17.2 & -20.7 & 1.39811:21:40.80 & 21:29 : 0.7 & 0.10225 & 77 & 18.2 & -19.9 & 3.77611:21:53.61 & 20:50:37.8 & 0.10252 & 38 & 17.4 & -20.7 & 4.94611:17:12.60 & 21:35:39.9 & 0.11236 & 52 & 17.9 & -20.5 & 3.52211:19 : 5.59 & 21:34 : 9.2 & 0.11492 & 73 & 17.6 & -20.8 & 1.69011:19 : 2.35 & 21:34:58.2 & 0.11542 & 48 & 17.4 & -20.9 & 1.79611:17:21.62 & 21:28:38.6 & 0.11767 & 44 & 17.6 & -20.9 & 3.08711:20 : 3.85 & 21:43:48.5 & 0.11786 & 61 & 18.1 & -20.3 & 3.21211:17:31.41 & 20:38:53.6 & 0.13213 & 143 & 17.3 & -21.4 & 5.92311:20 : 9.44 & 21 : 8:28.8 & 0.13398 & 36 & 18.7 & -20.0 & 2.30311:19:13.61 & 21:34:44.7 & 0.13450 & 48 & 17.9 & -20.8 & 2.00811:19 : 5.08 & 21:15 : 3.0 & 0.13736 & 70 & 18.6 & -20.2 & 0.57211:19 : 6.67 & 21:18:28.3 & 0.13814 & 50 & 18.4 & -20.4 & 0.12711:18:50.00 & 21:16:11.0 & 0.13845 & 28 & 19.8 & -19.0 & 0.71111:18:46.04 & 21 : 6:11.2 & 0.13861 & 30 & 19.2 & -19.6 & 1.87911:19:47.84 & 21:15:44.7 & 0.13863 & 40 & 18.1 & -20.7 & 1.30111:20:58.83 & 20:39:38.2 & 0.14004 & 82 & 19.2 & -19.6 & 6.33611:21:26.73 & 20:41:58.9 & 0.14021 & 65 & 18.5 & -20.4 & 6.61511:19:44.75 & 21:32:11.3 & 0.14058 & 38 & & & 2.06811:21:10.72 & 20:39:17.9 & 0.14072 & 60 & 19.5 & -19.3 & 6.60911:18:57.41 & 21:44:42.3 & 0.14096 & 45 & 18.5 & -20.3 & 3.44211:17:35.31 & 21:26:34.1 & 0.14481 & 46 & 18.2 & -20.7 & 3.15911:18 : 5.00 & 21 : 9:47.8 & 0.14549 & 53 & 18.4 & -20.5 & 2.43811:21:16.99 & 20:51:34.1 & 0.14616 & 59 & 18.2 & -20.7 & 5.66411:20:20.93 & 20:48:28.0 & 0.14652 & 66 & 19.0 & -19.9 & 4.88811:20 : 8.52 & 20:51:14.5 & 0.14656 & 63 & 18.2 & -20.7 & 4.36011:19:59.62 & 21:12 : 9.8 & 0.14695 & 79 & 18.6 & -20.4 & 1.93011:18 : 0.70 & 21:23:50.2 & 0.14828 & 40 & 19.4 & -19.5 & 2.31311:19:16.56 & 21 : 1:50.6 & 0.15010 & 63 & 18.1 & -20.8 & 2.49011:18 : 1.86 & 20:56:26.6 & 0.15263 & 41 & 18.0 & -21.0 & 3.97811:19:26.00 & 21:12:30.7 & 0.16362 & 35 & 18.0 & -21.2 & 1.20011:20:27.74 & 21:10:57.0 & 0.16431 & 34 & 19.7 & -19.5 & 3.08411:20:20.20 & 21 : 9:53.2 & 0.16444 & 67 & 20.3 & -18.9 & 2.92011:20:21.95 & 21 : 6:43.4 & 0.16478 & 54 & 18.6 & -20.6 & 3.24111:20 : 9.85 & 21:21:30.6 & 0.16487 & 20 & 18.9 & -20.3 & 2.20311:20:13.05 & 21 : 9:44.7 & 0.16496 & 64 & 18.6 & -20.6 & 2.72011:19 : 3.12 & 21:11 : 2.6 & 0.16535 & 69 & 17.6 & -21.6 & 1.28111:20:52.18 & 21:21:56.1 & 0.16540 & 110 & 19.3 & -19.9 & 3.71611:20:24.16 & 20:58:54.0 & 0.16552 & 60 & 17.8 & -21.4 & 4.13211:19:18.07 & 21:15 : 3.5 & 0.16581 & 51 & 18.3 & -20.8 & 0.73311:18:34.78 & 21:25:37.0 & 0.16604 & 64 & 18.9 & -20.3 & 1.55511:19:19.33 & 21:12:42.4 & 0.16627 & 63 & 18.9 & -20.3 & 1.08411:20:14.66 & 20:44:30.1 & 0.16645 & 38 & 18.2 & -21.0 & 5.86311:18:43.84 & 20:44 : 6.7 & 0.16660 & 72 & 18.7 & -20.5 & 5.49911:19:23.36 & 21 : 2:27.7 & 0.16661 & 71 & 18.6 & -20.6 & 2.64911:18:46.74 & 20:43:36.2 & 0.16696 & 134 & 18.7 & -20.5 & 5.57011:18:59.62 & 20:39 : 3.1 & 0.16712 & 36 & 18.7 & -20.5 & 6.23011:20 : 1.20 & 21:35:54.4 & 0.16934 & 71 & 18.6 & -20.7 & 3.22011:18 : 4.50 & 21:35:55.0 & 0.16995 & 47 & 18.4 & -20.8 & 3.50011:19:19.27 & 21 : 5 : 7.0 & 0.17305 & 30 & 19.3 & -20.0 & 2.28711:19:25.97 & 21 : 6:47.9 & 0.17311 & 44 & 18.5 & -20.8 & 2.08711:19:53.62 & 21 : 7:29.0 & 0.17321 & 89 & 18.7 & -20.6 & 2.51011:18:55.59 & 21 : 3:56.6 & 0.17365 & 36 & 18.6 & -20.7 & 2.49411:18:53.56 & 21 : 2:42.6 & 0.17503 & 53 & 18.4 & -20.9 & 2.71911:18:17.00 & 21:33:47.9 & 0.17522 & 52 & 18.4 & -20.9 & 3.02311:19:27.74 & 21 : 6:36.9 & 0.17523 & 58 & 18.3 & -21.0 & 2.15611:21:23.89 & 21 : 7:59.2 & 0.17532 & 41 & 19.0 & -20.3 & 5.37311:18:17.59 & 20:57:40.0 & 0.17532 & 57 & 18.9 & -20.4 & 3.96411:19:31.52 & 21:18 : 6.5 & 0.17585 & 34 & 18.9 & -20.4 & 0.87611:18:29.05 & 21:39:46.0 & 0.17605 & 91 & 18.7 & -20.6 & 3.61511:19:23.63 & 21 : 7 : 3.6 & 0.17625 & 49 & 19.1 & -20.2 & 2.05111:18:51.14 & 21:35:16.3 & 0.17640 & 100 & 18.8 & -20.5 & 2.65911:18:41.96 & 21:28:28.4 & 0.17784 & 71 & 18.7 & -20.6 & 1.79911:18 : 0.53 & 21:36:44.8 & 0.17895 & 53 & 18.7 & -20.6 & 3.84611:18:34.04 & 21:31:35.3 & 0.17915 & 71 & 18.2 & -21.2 & 2.40011:17:59.33 & 21:36:29.6 & 0.17928 & 99 & 18.6 & -20.7 & 3.85211:19:34.11 & 21:37:20.7 & 0.17997 & 74 & 18.7 & -20.7 & 3.11111:19:16.68 & 21:18:58.3 & 0.18018 & 36 & 19.3 & -20.1 & 0.30911:18:35.21 & 21:48:31.9 & 0.18083 & 85 & 18.1 & -21.3 & 4.97411:20:24.89 & 20:50:32.3 & 0.18821 & 57 & & & 5.73411:20:35.98 & 21 : 5:38.0 & 0.18884 & 64 & 18.9 & -20.6 & 4.16611:20:54.43 & 20:42:16.8 & 0.20676 & 77 & 19.7 & -20.0 & 8.10411:20:12.33 & 21:50:47.0 & 0.21159 & 56 & 18.4 & -21.3 & 6.44711:18 : 0.69 & 21 : 7 : 2.9 & 0.21189 & 61 & 19.8 & -20.0 & 3.71711:20:25.71 & 21 : 2:36.0 & 0.21236 & 114 & 18.7 & -21.0 & 4.55611:19:47.13 & 21:28:56.1 & 0.21409 & 65 & 19.2 & -20.5 & 2.45811:21:44.40 & 20:56:12.0 & 0.21418 & 88 & 18.6 & -21.2 & 8.04811:20:42.03 & 20:39:46.6 & 0.21465 & 59 & 19.1 & -20.7 & 8.45311:19:45.21 & 21:31 : 9.9 & 0.22931 & 63 & 18.9 & -21.0 & 2.87011:19:18.91 & 21:39:52.4 & 0.25950 & 68 & 18.8 & -21.3 & 4.45011:19:10.61 & 21:43 : 2.3 & 0.25982 & 67 & 19.3 & -20.8 & 5.10611:21:23.82 & 21:33:49.3 & 0.27284 & 62 & 19.2 & -21.1 & 7.70611:20:32.81 & 21:25:42.8 & 0.27577 & 39 & 19.5 & -20.8 & 4.61911:20:20.17 & 21:36 : 1.3 & 0.27807 & 37 & 18.9 & -21.4 & 5.317 cccccccc @xmath166 17.0 & 0.048 & 100% & 0.5 & 100% & 1.0 & 83.3% & 1.6 @xmath166 18.0 & 0.076 & 85.7% & 0.8 & 85.0% & 1.6 & 81.1% & 2.4 @xmath166 19.0 & 0.121 & 88.2% & 1.2 & 86.9% & 2.4 & 78.1% & 3.6 @xmath166 20.0 & 0.191 & 70.4% & 1.7 & 68.5% & 3.4 & 63.8% & 5.1 cllllc 1253.66 & 153@xmath611 & & 1253.81 & 0.0 & -51 1256.43 & 477@xmath612 & & 1025.72 & 0.22484 & 1260.15 & 1379@xmath618 & ( 214 ) & 1259.52 & 0.0 & -51 & & ( 1105 ) & 1260.42 & 0.0 & -68 & & ( 243 ) & 972.54 & 0.29678 & 1264.11 & 157@xmath614 & & 1031.93 & 0.22501 & 1266.92 & 246@xmath612 & & 977.02 & 0.29676 & 1271.00 & 107@xmath610 & & 1037.62 & 0.22501 & 1277.21 & 100@xmath615 & & 1277.46 & 0.0 & + 21 1284.82 & 92@xmath615 & & 1215.67 & 0.05704 & 1293.55 & 66@xmath616 & & 1215.67 & 0.06432 & 1297.39 & 170@xmath614 & & 1215.67 & 0.06722 & 1302.04 & 921@xmath616 & & 1302.17 & 0.0 & -29 1304.08 & 542@xmath611 & & 1304.37 & 0.0 & -68 1317.20 & 161@xmath619 & & 1317.22 & 0.0 & -46 1323.93 & 51@xmath612 & & 1215.67 & 0.08910 & 1328.96 & 72@xmath610 & & 1329.34 & 0.0 & + 21 1330.09 & 422@xmath613 & & 1025.72 & 0.29677 & 1334.25 & 873@xmath610 & & 1334.53 & 0.0 & -63 1335.43 & 174@xmath69 & & * 1335.71 & 0.0 & -31 1338.60 & 214@xmath616 & & 1031.93 & 0.29676 & 1346.01 & 123@xmath614 & & 1037.62 & 0.29676 & 1351.19 & 73@xmath611 & & 1215.67 & 0.11152 & 1361.16 & 114@xmath610 & & 1215.67 & 0.11974 & 1363.15 & 719@xmath619 & ( 438 ) & 1215.67 & 0.12123 & & & ( 396 ) & 1215.67 & 0.12157 & 1366.23 & 39@xmath613 & & 1215.67 & 0.12385 & 1368.44 & 49@xmath615 & & 1215.67 & 0.12566 & 1369.86 & 89@xmath615 & & 1370.13 & 0.0 & -46 1370.91 & 46@xmath615 & & 1215.67 & 0.12781 & 1393.40 & 452@xmath620 & & 1393.76 & 0.0 & -75 1395.08 & 263@xmath617 & & 1215.67 & 0.14760 & 1402.26 & 200@xmath621 & & 1402.77 & 0.0 & -75 1407.17 & 79@xmath617 & & 1215.67 & 0.15727 & 1414.41 & 63@xmath610 & & 1215.67 & 0.16350 & 1422.21 & 612@xmath620 & & 1215.67 & 0.16990 & 1433.46 & 84@xmath613 & & 1215.67 & 0.17915 & 1435.09 & 89@xmath615 & & 1215.67 & 0.18049 & 1454.67 & 86@xmath620 & & 1454.84 & 0.0 & -46 1456.33 & 69@xmath618 & & 1215.67 & 0.19794 & 1457.67 & 35@xmath614 & & 1215.67 : & 0.19905 & 1470.44 & 39@xmath615 & & 1215.67 : & 0.20961 & 1473.03 & 126@xmath617 & & 1215.67 & 0.21176 & 1474.87 & 586@xmath622 & & 1215.67 & 0.21325 & 1478.27 & 177@xmath618 & & 1215.67 & 0.21577 & 1488.90 & 905@xmath627 & & 1215.67 & 0.22484 & 1490.56 & 343@xmath625 & & 1215.67 & 0.22621 & 1492.61 & 87@xmath614 & & 1215.67 & 0.22782 & 1506.05 & 57@xmath619 & & 1215.67 & 0.23864 & 1509.05 & 98@xmath620 & & 1215.67 & 0.24132 & 1513.28 & 98@xmath620 & & 1215.67 & 0.24514 & 1526.39 & 902@xmath632 & & 1526.71 & 0.0 & -51 1529.75 & 144@xmath642 & & 1215.67 & 0.25822 & 1533.36 & 229@xmath643 & & 1215.67 & 0.26163 & cllllc 1235.41 & 112@xmath629 & & 1215.67 & 0.01639 & 1239.40 & 170@xmath624 & & 1215.67 & 0.01950 & 1250.30 & 322@xmath628 & ( 173 ) & 1215.67 & 0.02845 & & & ( 72 ) & 1250.58 & 0.0 & + 141253.86 & 189@xmath623 & & 1253.81 & 0.0 & + 141255.00 & 95@xmath617 & & 1215.67 & 0.03223 & 1260.51 & 1077@xmath628 & ( 196 ) & 1259.52 & 0.0 & + 14 & & ( 851 ) & 1260.42 & 0.0 & + 601265.66 & 171@xmath637 & & 1215.67 & 0.04125 & 1287.46 & 166@xmath624 & & 1215.67 & 0.05905 & 1289.55 & 66@xmath622 & & 1215.67 & 0.06079 & 1302.30 & 467@xmath633 & & 1302.17 & 0.0 & + 351304.51 & 466@xmath630 & & 1304.37 & 0.0 & + 601314.42 & 119@xmath619 & & 1215.67 & 0.08118 & 1317.42 & 147@xmath628 & & 1317.21 & 0.0 & + 26 1328.46 & 70@xmath623 & & 1215.67 & 0.09279 & 1335.00 & 1048@xmath635 & ( 845 ) & 1334.53 & 0.0 & + 61 & & ( 176 ) & * 1335.71 & 0.0 & + 541360.48 & 132@xmath625 & & 1215.67 & 0.11910 & 1373.69 & 91@xmath616 & & 1206.50 & 0.13852 & 1375.70 & 61@xmath615 & & 1215.67 & 0.13159 & 1384.16 & 543@xmath632 & ( 509 ) & 1215.67 & 0.13861 & & & ( 96 ) & 1215.67 & 0.13862 & 1393.83 & 484@xmath636 & & 1393.76 & 0.0 & + 431402.88 & 228@xmath629 & & 1402.77 & 0.0 & + 431417.78 & 865@xmath624 & ( 177 ) & 1215.67 & 0.16547 & & & ( 121 ) & 1215.67 & 0.16616 & & & ( 314 ) & 1215.67 & 0.16697 & 1426.81 & 308@xmath615 & & 1215.67 & 0.17366 & 1435.10 & 46@xmath615 & & 1260.42 : & 0.13858 & 1448.43 & 67@xmath614 & & uid & & ccllc 1244.466 & 183@xmath623 & 1025.72 & [email protected] & 1 1245.500 & 297@xmath623 & 1215.67 & [email protected] & 1 1247.858 & 052@xmath616 & 1025.72 & [email protected] & 1 1256.400 & 459@xmath627 & 1025.72 & [email protected] & 1 1261.124 & 227@xmath616 & 972.54 & [email protected] & 1 1264.138 & 206@xmath618 & 1031.93 & [email protected] & 1 1265.579 & 047@xmath614 & 1215.67 & [email protected] & 1 1266.917 & 247@xmath619 & 977.02 & [email protected] & 1 1529.486 & 168@xmath621 & 1215.67 & [email protected] & 1 1533.635 & 206@xmath617 & 1215.67 & [email protected] & 1 1539.773 & 206@xmath622 & 1215.67 & [email protected] & 1 1576.93 & 680@xmath610 & 1215.67 & 0.2972 & 2 cccccccc 0.02454 & 290 & 2.103 & 0.783 & 146 & 18:21:41.17 & 63:51:38.0 & 0.024040.04105 & 45 & 33.136 & 1.370 & -2481 & 18:22:36.11 & 64:45:44.9 & 0.049700.05704 & 87 & 1.954 & 1.784 & 60 & 18:21:21.38 & 64:49:40.4 & 0.056830.06432 & 62 & 5.405 & 2.916 & -341 & 18:16:31.37 & 64:43:54.6 & 0.065530.06722 & 159 & 6.985 & 2.916 & 475 & 18:16:31.37 & 64:43:54.6 & 0.065530.08910 & 47 & 1.043 & 0.739 & -55 & 18:20:42.96 & 64:19:45.8 & 0.089300.11152 & 66 & 0.780 & 0.773 & -8 & 18:20:53.45 & 64:19:37.1 & 0.111550.11974 & 102 & 2.520 & 2.445 & 46 & 18:25 : 4.45 & 64:25:22.9 & 0.119570.12123 & 391 & 1.122 & 0.144 & -83 & 18:22 : 2.65 & 64:21:39.3 & 0.121540.12157 & 353 & 0.180 & 0.144 & 8 & 18:22 : 2.65 & 64:21:39.3 & 0.121540.12385 & 35 & 4.871 & 1.740 & 339 & 18:23:58.19 & 64:26:52.3 & 0.122580.12566 & 44 & 11.164 & 1.740 & 821 & 18:23:58.19 & 64:26:52.3 & 0.122580.12781 & 41 & 18.790 & 1.740 & 1393 & 18:23:58.19 & 64:26:52.3 & 0.122580.14760 & 229 & 10.627 & 4.111 & -728 & 18:21 : 8.73 & 63:52 : 9.0 & 0.150390.15727 & 68 & 9.319 & 1.741 & 679 & 18:22 : 4.38 & 64 : 8:38.5 & 0.154650.16350 & 54 & 1.263 & 0.882 & -67 & 18:22:49.76 & 64:19:28.6 & 0.163760.16990 & 523 & 3.340 & 0.373 & -246 & 18:21:36.61 & 64:21:25.0 & 0.170860.17915 & 71 & 2.278 & 0.447 & 165 & 18:22:20.16 & 64:21:46.7 & 0.178500.18049 & 75 & 3.455 & 3.213 & 94 & 18:19:55.22 & 64:35 : 8.4 & 0.180120.19794 & 58 & 9.446 & 3.134 & -657 & 18:22:14.04 & 64 : 3 : 5.4 & 0.200570.19905 & 29 & 6.027 & 3.134 & -380 & 18:22:14.04 & 64 : 3 : 5.4 & 0.200570.20961 & 32 & 30.669 & 3.134 & 2249 & 18:22:14.04 & 64 : 3 : 5.4 & 0.200570.21176 & 104 & 37.867 & 3.134 & 2781 & 18:22:14.04 & 64 : 3 : 5.4 & 0.200570.21326 & 483 & 41.249 & 0.105 & -3034 & 18:21:54.40 & 64:20 : 9.3 & 0.225600.21656 & 145 & 30.185 & 0.105 & -2219 & 18:21:54.40 & 64:20 : 9.3 & 0.225600.22489 & 739 & 2.367 & 0.105 & -174 & 18:21:54.40 & 64:20 : 9.3 & 0.225600.22621 & 280 & 1.040 & 0.388 & -71 & 18:21:38.82 & 64:20:31.7 & 0.226500.22782 & 71 & 4.407 & 0.388 & 322 & 18:21:38.82 & 64:20:31.7 & 0.226500.23864 & 46 & 18.822 & 0.462 & -1379 & 18:21:56.28 & 64:22:50.4 & 0.244350.24132 & 79 & 9.987 & 0.462 & -731 & 18:21:56.28 & 64:22:50.4 & 0.244350.24514 & 79 & 1.882 & 0.625 & -130 & 18:22:19.11 & 64:18:43.5 & 0.245680.25814 & 134 & 21.884 & 1.950 & 1594 & 18:20:31.50 & 64:20:24.0 & 0.251470.26156 & 163 & 16.547 & 4.016 & -1172 & 18:21 : 0.73 & 64:37:54.3 & 0.266500.26660 & 163 & 0.847 & 0.795 & -21 & 18:22:10.26 & 64:17:15.6 & 0.266690.29674 & 524 & 0.665 & 0.427 & -37 & 18:22:13.61 & 64:20:14.5 & 0.29690 cccccccc 0.01639 & 110 & 16.261 & 0.557 & -1219 & 11:20:46.06 & 21:11:15.2 & 0.020530.01950 & 167 & 4.075 & 0.557 & -303 & 11:20:46.06 & 21:11:15.2 & 0.020530.02845 & 168 & 9.085 & 0.531 & 680 & 11:18:46.12 & 21 : 1:55.3 & 0.026120.03223 & 92 & 0.588 & 0.359 & -35 & 11:18:43.36 & 21:27:23.6 & 0.032350.04125 & 164 & 0.991 & 0.746 & 49 & 11:19 : 9.67 & 21 : 2:43.2 & 0.041080.05905 & 157 & 0.731 & 0.601 & -31 & 11:19:24.29 & 21:10:30.3 & 0.059160.06079 & 62 & 1.155 & 0.717 & 68 & 11:19:43.67 & 21:26:52.0 & 0.060550.08118 & 110 & 2.572 & 2.046 & 116 & 11:18 : 5.82 & 20:59:42.8 & 0.080760.09279 & 64 & 1.402 & 1.398 & -8 & 11:19:10.35 & 21 : 4:31.2 & 0.092820.11910 & 118 & 5.496 & 3.212 & 332 & 11:20 : 3.85 & 21:43:48.5 & 0.117860.13159 & 54 & 6.227 & 5.923 & -143 & 11:17:31.41 & 20:38:53.6 & 0.132130.13861 & 447 & 0.909 & 0.711 & 42 & 11:18:50.00 & 21:16:11.0 & 0.138450.13862 & 84 & 0.932 & 0.711 & 45 & 11:18:50.00 & 21:16:11.0 & 0.138450.16547 & 152 & 1.347 & 1.281 & 31 & 11:19 : 3.12 & 21:11 : 2.6 & 0.165350.16616 & 104 & 1.149 & 1.084 & -28 & 11:19:19.33 & 21:12:42.4 & 0.166270.16697 & 269 & 2.658 & 1.084 & 180 & 11:19:19.33 & 21:12:42.4 & 0.166270.17366 & 262 & 2.495 & 2.494 & 3 & 11:18:55.59 & 21 : 3:56.6 & 0.17365 ccccclccl 11:20:46.06 & 21:11:15.2 & 0.02053 & 557 & & 0.01950 & 167 & -303 & 11:18:43.36 & 21:27:23.6 & 0.03235 & 359 & & 0.03223 & 92 & -35 & 11:19:09.67 & 21:02:43.2 & 0.04108 & 746 & & 0.04125 & 164 & 49 & 11:19:24.29 & 21:10:30.3 & 0.05916 & 601 & & 0.05905 & 157 & -31 & 11:19:42.06 & 21:26:10.7 & 0.06134 & 677 & & 0.06079 & 62 & -155 & 11:19:06.67 & 21:18:28.3 & 0.13814 & 127 & & 0.13861,0.13862 & 531 & 125 & , 18:21:41.17 & 63:51:38.0 & 0.02404 & 783 & & 0.02454 & 290 & 146 & 18:20:42.96 & 64:19:45.8 & 0.08930 & 739 & & 0.08910 & 47 & -55 & 18:20:53.45 & 64:19:37.1 & 0.11155 & 773 & & 0.11152 & 66 & -8 & 18:22:02.65 & 64:21:39.3 & 0.12154 & 144 & & 0.12123,0.12157 & 846 & -40 & 18:22:49.76 & 64:19:28.6 & 0.16376 & 882 & & 0.16350 & 54 & -67 & 18:21:36.61 & 64:21:25.1 & 0.17086 & 373 & & 0.16990 & 523 & -246 & 18:22:20.16 & 64:21:46.7 & 0.17850 & 447 & & 0.17915,0.18049 & 146 & 340 & 18:21:54.40 & 64:20:09.3 & 0.22560 & 105 & & 0.22489,0.22621 & 1019 & -85 & 18:21:38.82 & 64:20:31.7 & 0.22650 & 388 & & 0.22782 & 71 & 322 & 18:21:56.28 & 64:22:50.4 & 0.24435 & 462 & & 0.24514 & 79 & 190 & 18:22:10.26 & 64:17:15.6 & 0.26669 & 795 & & 0.26660 & 163 & -21 & llllcc total & perturbed & galaxy - ly@xmath16 & galaxy - random & 0.305 & 1.4@xmath10 total & pure & galaxy - ly@xmath16 & galaxy - random & 0.250 & 6.5@xmath167 complete & perturbed & galaxy - ly@xmath16 & galaxy - random & 0.340 & 1.5@xmath10 complete & pure & galaxy - ly@xmath16 & galaxy - random & 0.285 & 5.2@xmath167 `` 1 mpc '' complete & perturbed & galaxy - ly@xmath16 & galaxy - random & 0.368 & 2.4@xmath10 `` 1 mpc '' complete & pure & galaxy - ly@xmath16 & galaxy - random & 0.299 & 1.2@xmath135 total & perturbed & galaxy - ly@xmath16 & galaxy - galaxy & 0.209 & 0.016 total & pure & galaxy - ly@xmath16 & galaxy - galaxy & 0.239 & 3.8@xmath135 complete & perturbed & galaxy - ly@xmath16 & galaxy - galaxy & 0.135 & 0.38 complete & pure & galaxy - ly@xmath16 & galaxy - galaxy & 0.200 & 0.05 `` 1 mpc '' complete & perturbed & galaxy - ly@xmath16 & galaxy - galaxy & 0.118 & 0.67 `` 1 mpc '' complete & pure & galaxy - ly@xmath16 & galaxy - galaxy & 0.188 & 0.14 llllcc complete & perturbed & galaxy - ly@xmath168 100 m ) & galaxy - random & 0.380 & 3.2@xmath135 complete & perturbed & galaxy - ly@xmath169 100 m ) & galaxy - random & 0.308 & 7.6@xmath135 complete & pure & galaxy - ly@xmath168 100 m ) & galaxy - random & 0.345 & 9.9@xmath135 complete & pure & galaxy - ly@xmath169 100 m ) & galaxy - random & 0.288 & 0.015 complete & perturbed & galaxy - ly@xmath168 100 m ) & galaxy - galaxy & 0.133 & 0.85 complete & perturbed & galaxy - ly@xmath169 100 m ) & galaxy - galaxy & 0.160 & 0.48 complete & pure & galaxy - ly@xmath168 100 m ) & galaxy - galaxy & 0.255 & 0.13 complete & pure & galaxy - ly@xmath169 100 m ) & galaxy - galaxy & 0.229 & 0.11 complete & perturbed & galaxy - ly@xmath168 200 m ) & galaxy - random & 0.544 & 5.5@xmath135 complete & perturbed & galaxy - ly@xmath169 200 m ) & galaxy - random & 0.333 & 1.9@xmath167 complete & pure & galaxy - ly@xmath168 200 m ) & galaxy - random & 0.283 & 6.0@xmath167 complete & pure & galaxy - ly@xmath169 200 m ) & galaxy - random & 0.277 & 3.3@xmath135 complete & perturbed & galaxy - ly@xmath168 200 m ) & galaxy - galaxy & 0.366 & 0.15 complete & perturbed & galaxy - ly@xmath169 200 m ) & galaxy - galaxy & 0.143 & 0.43 complete & pure & galaxy - ly@xmath168 200 m ) & galaxy - galaxy & 0.200 & 0.054 complete & pure & galaxy - ly@xmath169 200 m ) & galaxy - galaxy & 0.213 & 0.064
to study the nature of low @xmath0 absorbers in the spectra of qsos , we have obtained high signal - to - noise uv spectra of h 1821 + 643 ( @xmath1 = 0.297 ) and pg 1116 + 215 ( @xmath2 = 0.177 ) with the goddard high resolution spectrograph on the _ hubble space telescope_. the spectra have minimum s / n @xmath3 70 - 100 and 3@xmath4 limiting equivalent widths of 50 - 75 m at a resolution of @xmath3150 . excluding lines within 3000 of @xmath2 , we detect 26 lines with @xmath5 50 m toward h 1821 + 643 and 13 toward pg 1116 + 215 ( comparable to the 13 lines observed toward 3c 273 by morris et al . ) , which implies a density of 102@xmath616 lines per unit redshift for @xmath5 50 m and @xmath7 0.28 . the two - point velocity correlation function shows marginal evidence of clustering of lines on @xmath3500 scales , but only if the weakest lines are excluded . we have also used the wiyn observatory to measure galaxy redshifts in the @xmath8 fields centered on each qso in order to study the relationship between the absorbers and galaxies . we find 17 galaxy - absorber pairs within projected distances of 1 @xmath9 mpc with velocity separations of 350 or less . monte carlo simulations show that if the lines are randomly distributed , the probability of observing this many close pairs on the two sight lines is 3.6@xmath10 . we find that _ all _ galaxies with projected distances @xmath11 600 @xmath9 kpc from the qso sight lines have associated absorbers within 1000 , and the majority of these galaxies have absorbers within 350 . we also find that the equivalent width is anticorrelated with the projected distance of the nearest galaxy out to at least @xmath12 600 @xmath9 kpc . for @xmath13 600@xmath9 kpc , we find galaxies which do not have associated lines , but nevertheless the anticorrelation persists if we select galaxies with @xmath14 2 @xmath9 mpc which are within 500 or 1000 of a absorber . this anticorrelation has a high significance but should be interpreted cautiously because there are potential selection biases which could lead to an artificial correlation . statistical tests also show that the absorbers are not randomly distributed with respect to the galaxies . splitting the sample into roughly equal sets with @xmath5 100 m and @xmath15 100 m shows that _ the weakest absorbers are not randomly distributed _ either . comparison of the nearest neighbor distances of the weaker and stronger absorbers suggests that the weakest absorbers are less closely associated with galaxies , but the difference is not yet statistically significant . we find several galaxy groups which do not have clearly associated absorbers . however , given the projected distance of the nearest galaxy , we do not necessarily expect to find detectable lines in these groups based on the equivalent width - projected distance anticorrelation . furthermore , we find several counterexamples of comparable galaxy groups which _ do _ have associated lines . as in previous studies , we find some absorbers in regions apparently devoid of galaxies , although this may be due to the limited spatial extent and/or limited depth of the redshift survey . the equivalent width distributions of the absorbers apparently in voids and non - void absorbers are statistically indistinguishable , but the sample is small . we discuss the nature of the absorbers in light of the new data . the observations are consistent with the hypothesis that many of the low redshift absorption lines with rest equivalent widths in the range from 50 to @xmath3500 m trace the overall gas distributions in the large scale structures of galaxies rather than the gaseous halos of individual galaxies . other phenomena may also cause absorption lines . .
generalized langevin equations ( gle ) @xcite and non - markovian master equations@xcite , which arise in the treatment of systems interacting with environmental degrees of freedom , often have an integro - differential form . unlike ordinary differential equations which can be readily solved using runge - kutta , predictor - corrector and other well known numerical schemes@xcite there are no general methods for solving equations of integro - differential type . here we show that these integro - differential equations can be converted to ordinary - differential equations at the expense of introducing a new time variable which is treated as if it is of spatial type . [ similar schemes are employed to numerically solve the schrdinger equation for time - dependent hamiltonians@xcite and as analytical tools@xcite . there is also some resemblence to schemes for solving intego - differential equations of viscoelasticity@xcite . ] we then develop a numerical method based on this exact transformation and show that it can be used to accurately solve a variety of physically motivated examples . neglecting inhomogeneous terms resulting from noise , for simplicity , the generalized langevin equations@xcite for position @xmath0 and momentum @xmath1 of a damped oscillator in one dimension can be expressed in the form @xmath2 where @xmath3 and @xmath4 are the mass and frequency of the oscillator and @xmath5 is the memory function . defining a space - like time variable @xmath6 and a function @xmath7 it can be verified by direct substitution that @xmath1 and @xmath8 satisfy the following ordinary differential equations @xmath9 here we have introduced a differentiable damping function @xmath10 ( with @xmath11 ) which plays a useful role in the numerical scheme we will introduce to solve the ordinary differential equations ( [ gle1 ] ) , ( [ gle3 ] ) and ( [ gle4 ] ) . [ note that @xmath12 . ] neglecting inhomogeneous terms , non - markovian master equations@xcite can be written in the form @xmath13-\int_{-\infty}^tk(t , t')\rho(t')~dt ' \label{mastera}\ ] ] where @xmath14 is the time - evolving reduced density matrix of the subsystem , @xmath15 is an effective hamiltonian , and @xmath16 is a memory operator . [ we employ units such that @xmath17 . ] defining an operator @xmath18 it can be verified by direct substitution that @xmath14 and @xmath8 satisfy ordinary differential equations @xmath19-\chi(t,0)\label{master1}\\ d\chi(t , u)/dt&=&f(u)k(t+u , t)\rho(t)+\frac{\partial \chi(t , u)}{\partial u}\nonumber \\ & -&\frac{f'(u)}{f(u)}~\chi(t , u).\label{master2}\end{aligned}\ ] ] here @xmath10 is again a differentiable damping function such that @xmath11 . thus , the integro - differential langevin equations ( [ gle1])-([gle2 ] ) can be expressed in the ordinary differential forms ( [ gle1 ] ) and ( [ gle3])-([gle4 ] ) and the integro - differential master equation ( [ mastera ] ) can be expressed as the ordinary differential equations ( [ master1])-([master2 ] ) . to exploit these transformed equations as a practical numerical scheme we must discretize the @xmath6 variable on a grid of points so that the number of ordinary differential equations is finite . once this is achieved the ordinary differential equations can be solved using standard techniques@xcite . we use an eighth order runge - kutta routine@xcite in our calculations . to minimize the number of grid points we choose a damping function @xmath10 which decreases rapidly with @xmath6 . in the calculations reported here we used @xmath20 . in practice fewer grid points are needed for positive @xmath6 than for negative @xmath6 , and we found that the points @xmath21 for @xmath22 worked well when we chose @xmath23 . here @xmath24 is the largest positive @xmath6 value . while accurate solutions can be obtained for almost any non - zero value of @xmath25 we found the most rapid convergence when values were optimized for the type of equation . hence , @xmath25 is specified differently below for each type of equation . to complete the numerical method we need a representation of the partial derivative with respect to @xmath6 on the grid . this could be performed via fast fourier transform techniques@xcite . we chose instead to employ a matrix representation @xmath26 which is known as the sinc - dvr ( discrete variable representation)@xcite . a discrete variable representation ( dvr ) is a complete set of basis functions , associated with a specific grid of points , in which functions of the variable are diagonal and derivatives have simple matrix representations@xcite . dvrs are often used in multi - dimensional quantum mechanical scattering theory calculations@xcite . in the sinc - dvr@xcite , which is associated with an equidistantly spaced grid on @xmath27 , partial derivatives can thus be evaluated with a sum @xmath28 for any function or operator @xmath29 . in our calculations we chose @xmath30 to equal the time interval @xmath31 between output from the runge - kutta routine . for the generalized langevin equation we chose an initial value problem ( i.e. @xmath32 for @xmath33 and @xmath34 for @xmath35 ) where @xmath36 has one of the following forms @xmath37 which are displayed graphically in figure 1 . the solid curve is ( [ mem1 ] ) , the dashed is ( [ mem2 ] ) , the short - dashed is ( [ mem3 ] ) and the dotted is ( [ mem4 ] ) . these memory functions were chosen to roughly represent the various functional forms which can occur physically@xcite and for ease in obtaining exact solutions . the constants appearing in equations ( [ gle1 ] ) , ( [ gle3 ] ) and ( [ gle4 ] ) are chosen as @xmath38 and @xmath39 . figure 2 shows the functional form of the exact solutions @xmath0 ( solid curve ) and @xmath1 ( dashed ) , which evolve from initial conditions @xmath40 and @xmath41 , for memory function ( [ mem1 ] ) over a timescale of 20 units with @xmath42 . solutions for the other memory functions ( and the same initial conditions ) are similar in appearance . these exact solutions were obtained by expoiting the fact that the above memory functions are sums of exponentials ( i.e. @xmath43 ) from which it follows that one may write @xmath44 for @xmath45 , and solve these ordinary differential equations using standard methods . this approach only works for memory functions of this type . the negative logarithm of the absolute error in @xmath0 , @xmath48 is shown in figure 3 plotted against time for the values of @xmath49 indicated in the inset . [ the error in @xmath1 is similar . ] as @xmath49 increases @xmath50 increases ( on average ) and hence the error decreases . the oscillations in @xmath50 are caused by periodic intersections of the two solutions . in practice it is impossible to visually distinguish the two solutions when @xmath51 . note that after a short transient the error ( on average ) does not increase . this is probably a consequence of the linearity of these equations . some decline in accuracy with time should be expected when the langevin equations are non - linear ( e.g. a particle in a double - well ) . figure 4 compares the exact solutions for @xmath0 ( solid curve ) and @xmath1 ( short - dashed ) with those obtained using our method for @xmath52 ( dashed and dotted , respectively ) over a time of 40 units . no disagreement is visible . convergence for memory function ( [ mem2 ] ) is similar . memory functions ( [ mem3 ] ) and ( [ mem4 ] ) which take negative values and have long time tails require many grid points for convergence . figure 5 shows the negative logarithm ( base ten ) of the absolute error in @xmath0 for this case . while many grid points are required , high accuracy solutions can clearly be obtained using our method . for the master equation we chose an initial value problem consisting of a dissipative two - level system representing a spin interacting with environmental degrees of freedom . if the spin hamiltonian is @xmath53 and the coupling to the environment is proportional to @xmath54 then the equation for the density matrix @xmath14 is of the form@xcite @xmath55\nonumber \\ & & -c\int_0^t w(t - t')\{\sigma_x^2\rho(t')+\rho(t')\sigma_x^2 - 2\sigma_x\rho(t')\sigma_x\}~dt'\nonumber\\ & & ~~~~\end{aligned}\ ] ] where the sigmas denote pauli matrices . parameters were set as @xmath56 and @xmath57 . we chose to define @xmath58 which differs somewhat from the general definition employed in ( [ smo ] ) . the transformed equations are then @xmath59 - 2c\{\chi(t,0)\nonumber \\ & -&\sigma_x\chi(t,0)\sigma_x\ } \label{stu1}\\ \frac{d\chi(t , u)}{dt}&=&e^{-gu^2}w(u)\rho(t)+\frac{\partial \chi(t , u)}{\partial u}\nonumber \\ & + & 2 g u~\chi(t , u)\label{stu2}.\end{aligned}\ ] ] theory predicts that the memory function @xmath36 for this problem is approximately gaussian in form@xcite . however , we were unable to obtain an exact solution of the master equation for this case@xcite . instead we approximate the gaussian via the similar function @xmath60 . exact solutions for @xmath61 and initial conditions @xmath62 and @xmath63 were obtained in the same way as for the generalized langevin equations and are plotted vs time in figure 6 . for the approximate method we used @xmath64 ^ 2 $ ] and for negative @xmath6 we set @xmath47 . from figure 7 where we plot @xmath65 against time we see that convergence of the numerical method is very rapid for these equations . [ similar accuracies are achieved for @xmath66 and @xmath67 . ] thus , we have shown that accurate solutions of integro - differential equations can be obtained via transformation to a larger set of ordinary differential equations . because this transformation is exact we expect that the method will also work for equations not considered in this manuscript . it should be possible to obtain accurate solutions for such equations via the following steps . first find an approximation of the memory function or operator which will allow exact solutions to be obtained . optimize the numerical method by finding the best @xmath25 for the model equations . finally , apply the numerical method to the original equations and look for convergence of the solutions with increasing @xmath49 .
we show that integro - differential generalized langevin and non - markovian master equations can be transformed into larger sets of ordinary differential equations . on the basis of this transformation we develop a numerical method for solving such integro - differential equations . physically motivated example calculations are performed to demonstrate the accuracy and convergence of the method .
this work was supported by nwo under grant no . 643.200.503 , the netherlands research school for astronomomy ( nova ) and the national science foundation under grant no . phy 05 - 51164 to the kavli institute for theoretical physics in santa barbara .
* supernova 2006gy in the galaxy ngc 1260 is the most luminous one recorded @xcite . its progenitor might have been a very massive ( @xmath0 ) star @xcite , but that is incompatible with hydrogen in the spectrum of the supernova , because stars @xmath1 are believed to have shed their hydrogen envelopes several hundred thousand years before the explosion . alternatively , the progenitor might have arisen from the merger of two massive stars @xcite . here we show that the collision frequency of massive stars in a dense and young cluster ( of the kind to be expected near the center of a galaxy ) is sufficient to provide a reasonable chance that sn 2006gy resulted from such a bombardment . if this is the correct explanation , then we predict that when the supernova fades ( in a year or so ) a dense cluster of massive stars becomes visible at the site of the explosion . * # 1*[#1 ed ] * # 1*[#1 ed ] * # 1*[#1 simon ] * # 1*[#1 simon ] * the presence of hydrogen in supernova sn2006gy is hard to reconcile with the explosion of a @xmath2 star , as such a star loses its hydrogen - rich envelope several hundreds of thousands of years before the star explodes . also the location of the supernova , at a projected distance of about 1 ( @xmath3pc ) from the nucleus of the host galaxy ngc1260 is remarkable . a merger between a very massive ( @xmath0 ) hydrogen - depleted star that already had a core in an advanced phase of helium burning , with a hydrogen rich main - sequence star of 10 to 40 , @xmath4 to @xmath5 years prior to the supernova explosion may explain the unusual brightness of the supernova , the presence of the hydrogen in the interstellar medium surrounding the supernova and the presence of hydrogen in the supernova itself @xcite . the existence of young star clusters which are in a state of dynamical core collapse is crucial for the proposed scenario . during core collapse and the subsequent post - core collapse evolution of the star cluster a runaway collision product can grow , and even though the star is likely to be much more extended than usual , subsequent bombardment will result in a net increase in mass @xcite . eventually the massive star is expected to prostrate to a black hole of intermediate mass @xcite . the supernova in which the black hole forms is likely to be unusually bright with some hydrogen in its envelope left - over from the last collision . the inner few hundred parsec around the center of the milky way is populated with several bright and dense star clusters , of which the arches cluster @xcite and quintuplet @xcite are the most well known , but many others exist . the proximity of the galactic center and the depth of the potential well of the bulge causes these clusters to be denser than elsewhere in the galaxy @xcite . the sb / sb0 host galaxy ngc 1260 appears rather ordinary , though the presence of a dust lane and of hii emission near its center suggests that a recent burst of star formation occurred near its center @xcite . we estimate , taking an intergalactic extinction of @xmath6mag @xcite into account , that within the observed isophotal magnitude @xmath7 ( @xmath8kpc ) @xcite and adopting @xmath9 @xcite , ngc1260 has a mass of @xmath10 . assuming that the mass enclosed within a radius @xmath11 from the center of ngc1260 is , like in the milky way @xcite described with @xmath12 , but for ngc1260 @xmath13 . we can then calculate the lower limit to the tidal radius @xcite for a cluster of mass @xmath14 in a circular orbit at distance @xmath11 from the center of ngc1260 . star clusters that experience core collapse before the most massive stars have left the main sequence can grow a supermassive star via collision runaway . the mass which can then grow within @xmath15myr can be estimated using eq.2 of @xcite . here we have to make some assumption about the stellar mass function in the cluster , but for clarity adopting a mean mass of @xmath16 is sufficient without detailed knowledge of the exact shape of the initial mass function . for a reasonable range of cluster densities and distances from the center of ngc 1260 we can now calculate the mass that can be grown in the cluster in @xmath15myr . in fig.[fig : mrunaway ] we present the results of our calculations using a king @xcite model with a depth of the central potential expressed in the dimension - less parameter @xmath17 , which can produce at most a @xmath18 star in a collision runaway . for shallower as well as for more concentrated king models the maximum mass for the supermassive star decreases , as well as the mass of the cluster that produces such stars . the last collision before the supernova , must have occurred with a relatively unevolved main - sequence star , and deposited large quantities of hydrogen on the surface of the collision product . by the time of the supernova not all surface hydrogen of the last collision was blown away , as about one of hydrogen was observed in the supernova @xcite . the remainder of the hydrogen deposited on the stellar surface during the last collision was found in the interstellar medium surrounding the supernova , and exceeds some 0.55 @xcite but could be as high as 2030 @xcite . this mass may have come from the strong stellar wind in the last few @xmath19 years before the supernova , blowing away the hydrogen which was deposited to the stellar surface during the collision . a tentative upper limit for the rate of mass loss of the progenitor star is @xmath20/yr to @xmath21/yr @xcite . these observed mass loss rates are consistent with those of detailed evolutionary calculations of stars of 500@xmath22 @xcite . at this mass loss rate it takes roughly @xmath23 to @xmath24 years to blow 20 in the form of a stellar wind from the surface of the supermassive star . this time scale is of the same order as our estimated average time between collisions of @xmath25 years ( see fig.[fig : mrunaway ] ) . the luminosity of the supernova explosion in collapsar models is driven by the angular momentum transfer from the critically rotating black hole to its surrounding torus . the available energy reservoir , and thus the supernova brightness , would then be proportional to the mass of the black hole @xcite . the observed brightness of sn2006gy would then be consistent with the collapse of an unusually massive star , and the consequent formation of a rather massive ( @xmath26 ) black hole . we are unaware of detailed simulations of such an unusual supernova to bolster our arguments , but the consequences for the supernova seem to be profound and we encourage further research in this direction . the amount of hydrogen in the pre - supernova stellar envelope , the amount of hydrogen in the interstellar medium , the mass - loss rate of the supernova progenitor derived from the observations and the enormous brightness of the supernova , bracket the values we derive based on the collision runaway scenario . we therefore conclude that a collision of a @xmath27 main - sequence star with a supermassive star @xmath28 years before the supernova could conveniently explain the range of oddities surrounding sn2006gy . we predict that a young ( @xmath29myr ) , dense and massive ( @xmath30 ) star cluster is present at the location of the supernova . at this moment the star cluster can not be seen , but adopting a mass - to - ligh ratio of @xmath31 , which is consistent with the starburst99 @xcite models for a @xmath29myr old stellar population , the cluster should become noticeable as soon as the supernova fades below an absolute magnitude of about -8.2 mag for @xmath32 and -5.7mag , for a @xmath19 star cluster . the environment in which the collision runaway can be initiated is rather exotic , as the cluster has to be sufficiently massive and dense to warrant dynamical core collapse within a few myr . in star clusters sufficiently massive to grow a massive collision product there are typically between 60 and 600 stars @xmath33 , and consequently only one out of 60600 type ib / c or type ii supernovae in these clusters will be of this peculiar bright type , like sn2006gy . if in a nuclear or a normal starburst ten per cent of all stars are formed in sufficiently dense clusters , one would expect that about one out of 600 - 6,000 supernovae to be of this type . ./journals_nature.def 10 , r. , 11 ( 2006 ) . , a. _ et al . _ . , 11 ( 2006 ) . , j. l. , garnavich , p. , chronister , a. & connick , p. . , 11 ( 2006 ) . , r. j. _ et al . _ . , 11 ( 2006 ) . , n. _ et al . _ . ( 2006 ) . , g. & maeder , a. . , 581598 ( 2005 ) . , e. o. _ et al . _ . , l13l16 ( 2007 ) . , s. f. , makino , j. , mcmillan , s. l. w. & hut , p. . , 117126 ( 1999 ) . , t. k. _ et al . _ . ( 2007 ) . , c. l. & kalogera , v. theoretical black hole mass distributions . , 548560 ( 2001 ) . , s. f. , baumgardt , h. , hut , p. , makino , j. & mcmillan , s. l. w. . , 724726 ( 2004 ) . , a. s. _ et al . _ . , 750761 ( 1996 ) . , a. s. _ et al . _ . in _ bulletin of the american astronomical society _ , 12621263 , ( 1992 ) . , c. m. & bica , e. `` new star clusters projected close to the galactic centre '' . , l9l12 ( 2000 ) . , c. m. _ et al . _ . , 127134 ( 2003 ) . , j. p. , paumard , t. , stolovy , s. r. & rigaut , f. . , 155167 ( 2004 ) . , s. f. _ et al . _ . , 319326 ( 2006 ) . , j. & meusinger , h. . , 141161 ( 1999 ) . , g. _ et al . _ . , 22682280 ( 2003 ) . , p. , ashman , k. m. & persic , m. . , 8993 ( 1991 ) . , r. h. & lowinger , t. . , 292297 ( 1972 ) . , j. & tremaine , s. . princeton , nj , princeton university press , pp.747 , ( 1987 ) . , s. f. & mcmillan , s. l. w. . , 899907 ( 2002 ) . , m. a. , freitag , m. & rasio , f. a. . , 632652 ( 2004 ) . , m. , grkan , m. a. & rasio , f. a. . , 141161 ( 2006 ) . , i. r. the structure of star clusters . iii . some simple dvriamical models . , 6475 ( 1966 ) . , l. r. . ( 2006 ) . , l. r. , van den heuvel , e. , vinck , j. , portegies zwart , s. & de kooter , a. . ( 2007 ) . , h. k. , wijers , r. a. m. j. & brown , g. e. , 83114 ( 2000 ) . , c. _ et al . _ . , 340 ( 1999 )
coherent states have become a tool of the utmost importance in quantum optics and many efforts have been devoted to investigate them , also with respect to their prospective applications in the growing field of quantum technologies ( qt ) such as quantum information ( coding , transmission and elaboration of information by exploiting properties of quantum states ) , quantum computing , quantum imaging and quantum metrology . in particular , their original definition has been generalized , so that nowadays several quantum optical extensions of the glauber states have been constructed and analyzed from both the theoretical @xcite-@xcite and the experimental @xcite-@xcite point of view . such states can find application also in quantum calculus , as shown , e.g. , in @xcite . in the present paper the focus is on the @xmath0-boson and single - boson nonlinear coherent states ( nlcs ) which can be realized experimentally by means of trapped ions interacting with driving electromagnetic fields , as reported in references @xcite and @xcite , respectively . due to their mathematical structure these states are affected , for specific values of the lamb - dicke parameter , by singularity problems which do not appear to have been tackled systematically so far . the paper is organized as follows : in section 2 the hamiltonian of a ion trap driven by two lasers is illustrated , while the construction of the corresponding @xmath0-boson nlcs is reported in section 3 resorting to standard tools of quantum optics , i.e. , the interaction representation and the rotating wave approximation ( rwa ) . sections 4 and 5 are devoted to the special case of the single - boson nlcs . the relaxing of the rwa in order to remove the singularity problems is discussed in section 6 . the properties of the ion - trap systems and of their hamiltonians in various physical regimes have been widely investigated @xcite-@xcite . as for the construction of coherent states for ion - trap systems , in @xcite it has been proved that single - boson nlcs can be realized by exploiting the hamiltonian @xmath1 @xmath2 @xmath3 @xmath4 @xmath5 of a two - level ion trapped in an external harmonic oscillator potential and interacting with two laser fields . the free hamiltonian @xmath3 describes the motion of both the electronic ( or internal ) and vibrational ( or external ) degrees of freedom of the ion , while the interaction hamiltonian @xmath6 corresponds to the interaction of the ion with the electromagnetic fields of the driving lasers . the external states of the ion are the ground state @xmath7 and the excited state @xmath8 , their transition frequency is @xmath9 @xmath2 @xmath10 , @xmath11 is the frequency of the trap harmonic potential . with @xmath12 the annihilation operator for the vibrational motion of the trapped ion , @xmath13 @xmath2 @xmath14 , @xmath15 the pauli matrix , which acts as a third component pseudospin operator , @xmath16 in eq . the zero energy level of the ion has been set at halfway between @xmath7 and @xmath8 , so that in outer product notation @xmath15 @xmath2 @xmath17 - @xmath18 . the interaction hamiltonian reads @xmath19 in eq . @xmath20 is the dipole coupling matrix element , @xmath21 , @xmath22 , and @xmath23 , @xmath24 are the amplitudes and the frequencies of the lasers fields , @xmath25 , @xmath26 their wavectors ; @xmath27 @xmath2 @xmath28 is the position operator of the ion center of mass , where @xmath29 @xmath30 @xmath31 is the value of both lasers wavevectors , @xmath32 is the lamb - dicke parameter , with @xmath33 the ion mass . with @xmath34 and @xmath35 pauli matrices , @xmath36 @xmath2 @xmath37 @xmath2 @xmath38 and @xmath39 @xmath2 @xmath40 are the two - level electronic flip operators . for the lasers tunings one sets @xmath41 , which corresponds to the resonant condition of the first laser , and @xmath42 , which means that the second laser is tuned on @xmath0-th lower ( or red - sideband ) vibrational level , @xmath0 @xmath43 1 . from eq . one then obtains @xmath44 s_+ + { \textrm{h.c . } } \ ; . \label{hamintapprox}\ ] ] + the underlying idea for the construction of nonlinear bosonic coherent states from the hamiltonian can be summarized as follows . + 1 . through the unitary transformation @xmath45 one writes the hamiltonian in the interaction representation @xmath46 @xmath2 @xmath47 ; utilizing the rwa @xmath46 is put in the form @xcite @xmath48 where @xmath49 is a non - hermitian operator acting in the fock space ; 2 . the condition @xmath50 @xmath2 0 @xcite,@xcite determines the states @xmath51 upon expanding them in terms of number states ; besides , @xmath52 @xmath2 0 by construction ; 3 . the states @xmath51 are recognized to be the eigenstates of the annihilation operator @xmath53 with the appropriate identification of the nonlinear function @xmath54 , i.e. , they are coherent states in the barut - girardello sense @xcite . let us detail the above steps . in the interaction representation one obtains @xmath55 s_+ + { \textrm{h.c . } } \ ; . \label{hamint2}\end{aligned}\ ] ] the link between the physical system of the laser - driven ion trap and its mathematical structure is more clearly grasped rewriting eq . in the form @xmath56 s_+ + { \textrm{h.c . } } \ ; , \label{hamintoper}\ ] ] where @xmath57 and @xmath58 are the operators @xmath59 @xmath60 in eqs . , , @xmath61 , with @xmath62 integer @xmath43 0 , is the ` laguerre polynomial operator ' . by this we means an appropriate integral representation @xcite . its action in the fock space results into the generalized laguerre polynomial of degree @xmath63 and parameter @xmath62 @xcite @xmath64 applying the rwa to the interaction hamiltonian implies that all terms oscillating with frequency @xmath11 or larger than @xmath11 are disregarded in eqs . and so that in this approximation @xmath65 and @xmath66 reduce to @xmath67 this leads to the form @xmath68 where @xmath69 the states @xmath51 , which are the stationary states of the master equation of the system , are found from the condition @xmath50 @xmath2 0 . their physical meaning is twofold : they are both noiseless codes in the sense specified in @xcite , as @xmath52 @xmath2 0 , and the coherent states of the system , defined as the eigenstates of the operator @xmath70 where @xmath71 while the nonlinear operator @xmath54 is @xmath72 one finds that the following self - splitting into @xmath0 states results from the condition @xmath50 @xmath2 0 with @xmath73 @xmath2 @xmath74 @xmath75 the states @xmath76 , @xmath77 are fixed by the corresponding sets of coefficients @xmath78^{-1 } c_\ell^{(k ) } \ ; , \label{cnkl}\ ] ] @xmath79 being the normalization constants , chosen to be real to retrieve the glauber states when @xmath80 and @xmath81 @xmath82^{-2 } \right\}^{- \frac{1}{2 } } \ ; . \label{normkl}\ ] ] in eqs . , the scalar functions @xmath83 are obtained from the action of the operator in the fock space , while @xmath84 for @xmath85 . it is readily verified that the states are eigenstates of the operator , @xmath86 @xmath2 @xmath87 , with the @xmath0-degenerate eigenvalue ; since @xmath88 @xmath2 @xmath89 , they form an orthonormal basis of a @xmath0-dimensional hilbert space . on the other hand , due to the splitting , @xmath51 is eigenstate of @xmath90 with the nondegenerate eigenvalue @xmath91 . in the nondegenerate case @xmath80 one retrieves the construction reported in @xcite . the notation adopted so far can be simplified dropping the index @xmath0 throughout . + eqs . and show that @xmath92 is by construction eigenstate of the operator with @xmath0 @xmath2 1 , @xmath93 @xmath2 @xmath94 , the eigenvalue being @xmath95 @xmath2 @xmath96 . the relevant commutation relations are @xmath97 $ ] @xmath2 @xmath98 , @xmath99 $ ] @xmath2 @xmath100 and @xmath101 \,=\ , ( \hat{n}+1 ) f^2(\hat{n } ) - \hat{n } f^2(\hat{n}-1 ) \doteq c(\hat{n } ) \ ; . \label{com_aa}\ ] ] in terms of number states @xmath102 where the normalization constant @xmath103 , which is the new label of @xmath104 in eq . , has been factorized so that from eq . @xmath105^{-1 } \,=\ , \xi^n \sqrt{n ! } \prod^{n-1}_{j=0 } \frac{l_{j}^{(0 ) } ( \eta^2)}{l_{j}^{(1 ) } ( \eta^2 ) } \ ; , \label{gamma}\ ] ] with @xmath106 @xmath107 1 , @xmath108 @xmath2 @xmath109 $ ] and @xmath110^{-\frac{1}{2 } } \ ; . \label{coh - norm}\ ] ] with @xmath111 , the following recurrence formula for the coefficients holds @xmath112^{-1 } \xi^m c_n \ ; . \label{gamrecurr}\ ] ] + for the dimensionless operators @xmath113 @xmath2 @xmath114 , @xmath115 @xmath2 @xmath116 , in view of eq . , one has @xmath117 $ ] @xmath2 @xmath118 , while the variances @xmath119 \ ; , \label{varxgen}\ ] ] @xmath120 \ ; , \label{varpgen}\ ] ] calculated in the coherent states @xmath92 , minimize the heisenberg uncertainty relation @xmath121 @xmath43 @xmath122 \rangle|^2 /4 $ ] @xmath123 in eq . @xmath124 @xmath2 @xmath125 @xmath2 @xmath126 , where @xmath127 @xmath2 @xmath128 and @xmath129 @xmath2 @xmath130 with @xmath131 @xmath132 . confirms that the states are intelligent states along with the glauber states . on the other hand they are not coherent states according to the perelomov definition @xcite since , in view of , @xmath93 and @xmath133 generate an infinite - dimensional algebra . eq . shows that for @xmath134 @xmath135 has @xmath136 zeros and as many poles due to the roots of the @xmath63 laguerre polynomials @xmath137 and @xmath138 , respectively , whose effects on the coherent states have to be investigated . it is worth acknowledging that the above problems are referred to briefly in @xcite where it is stated that the values of @xmath139 must be such as to prevent them . furthermore , in @xcite the effects of an ill - defined function @xmath108 are investigated in relation to the representation of the fock states in terms of single - boson nlcs . + let us investigate the effect produced by a single zero , @xmath139 @xmath2 @xmath140 , of a given @xmath135 , say @xmath141 . this means that @xmath142 @xmath2 0 and , equivalently , @xmath143 @xmath144 @xmath145 . since @xmath146 @xmath2 0 , from eq . it follows that @xmath147 @xmath2 0 , @xmath148 @xmath111 . due to the above conditions the sums in eqs . and can be truncated so that the states and their normalization constant read @xmath149 @xmath150^{-\frac{1}{2 } } \ ; . \label{normzero}\ ] ] the vanishing of the coefficient @xmath146 violates the coherence of the states as the latter do not satisfy the eigenvalue equation of the annihilation operator @xmath151 notice that the truncactions and do not remove the singularity of the states which influences the variances of the observables @xmath152 , @xmath153 . in fact , addressing now the minimum - uncertainty property in the states @xmath154 , from @xmath155 \ ; , \ ] ] and from eqs . , , one gets @xmath156 @xmath2 @xmath157 , @xmath158 @xmath2 @xmath159 , where @xmath160 \ ; , \nonumber\\ \xi & = & |\xi|^2 p_{\beta-1}(\beta ) \left[1 - p_{\beta-1}(\beta ) \right ] + \frac{1}{2 } \sum^{\beta-1}_{n=0 } p_n(\beta ) c(n ) \ ; , \nonumber\end{aligned}\ ] ] while @xmath161 @xmath2 @xmath162 , @xmath163 @xmath164 and @xmath165 denotes complex conjugation . @xmath166 for @xmath167 , @xmath168,scaledwidth=90.0% ] as an example , in fig . [ fig : heisenzero ] the product @xmath169 @xmath166 @xmath2 @xmath170 is plotted in the range @xmath171 assuming that the singularity problem corresponds to @xmath167 , i.e. , @xmath140 is one of the roots of @xmath172 . of course , in fig . [ fig : heisenzero ] , @xmath139 varies with continuity so that the product of the variances contain the roots in the considered range of all the laguerre polynomials involved in the @xmath167 case , i.e. , @xmath173 and @xmath174 , @xmath175 . one emphasizes that a root of the polynomials @xmath173 ( the single zero case ) corresponds to a singularity of the product @xmath169 @xmath166 , while when a root of the polynomials @xmath174 occurs ( the single pole case of sect . 5.2 ) the product vanishes . notice also that in the example considered the minimum - uncertainty product @xmath176 is still maintained over a wide range of values of @xmath139 . this feature strongly depends on the parameter @xmath177 , in the sense that the lower the value of @xmath177 , the wider the ranges where , due to the minimum - uncertainty property , the truncated states can still be considered , approximately , intelligent states and conveniently utilized for qt purposes . + here @xmath140 is a single pole of a given @xmath178 , i.e. , @xmath178 @xmath144 @xmath145 . this implies that @xmath179 @xmath2 0 and correspondingly @xmath180 @xmath2 0 . from the recurrence relation it follows that @xmath181 @xmath144 @xmath145 , @xmath148 @xmath111 , as rapidly as @xmath178 . the above conditions imply that the contribution of the first @xmath182 terms to the sums and can be disregarded with respect to the singular terms , leading to the following expressions of the states and their normalization constant @xmath183 @xmath184^{-\frac{1}{2 } } \ ; . \label{normpolo}\ ] ] extracting the singular coefficient @xmath185 from both eqs . and and utilizing the recurrence property , one finds @xmath186^{-\frac{1}{2 } } \ , \sum^{\infty}_{j=0 } \frac{\xi^{\mu+j}}{\sqrt{(\mu+j ) ! } } \prod^{\mu+j-1}_{\ell=\mu } f^{-1}(\ell ) |n\rangle \ ; , \nonumber\end{aligned}\ ] ] where @xmath187 is the series @xmath188 ^ 2_{\eta^2 = { \bar\eta}^2 } \ ; , \label{series}\end{aligned}\ ] ] the question whether or not the normalization constant @xmath187 converges , and hence the states exist ( in the sense that they are normalizable ) , can be tackled , for example , through the ratio test . according to this criterion the convergence of the series , written as @xmath189 @xmath2 @xmath190 , implies the evaluation of the ratio @xmath191 ^ 2_{\eta^2 = \bar{\eta}^2 } \ ; , \ ] ] when @xmath192 . resorting to the asymptotic formulas for the laguerre polynomials of large degree reported , e.g. , in @xcite , one finds that @xmath193 ^ 2 \ ; . \label{ratiolimodd}\end{aligned}\ ] ] the result proves to be wildly oscillating as @xmath194 increases , which means that the very definition of the states @xmath195 is questionable when a pole of a coefficient @xmath185 is encountered . this pathology becomes manifest in the example of fig . [ fig : heisenzero ] where the poles correspond to the vanishing of @xmath169 @xmath166 . the above analysis shows that there are values of @xmath139 which alter significantly the structure of the single - boson nlcs , resulting either in the truncated states or in the states whose definition is not certain . these singularities are inherent in the very construction of such states which relies on the application of the rwa to the two - level ion model . one is therefore led to investigate whether relaxing the rwa can remove the above difficulties . a detailed analysis of the possible corrections of the rwa is out of the aims of the present paper . here we discuss two ways through which a ` first - order ' correction of the rwa can be implemented , where first - order means that only the terms of eqs . , associated to the exponentials @xmath196 are retained . specifically , one adds the following time - dependent operator @xmath197 to the interaction hamiltonian @xmath198 s_+ + { \textrm{h.c . } } \ ; , \ ] ] where @xmath199 \ ; , \label{ofirst}\ ] ] @xmath200 the first approach develops along the general scheme of section 3 , i.e. , one assumes that , in this approximation , the coherent states @xmath201 , @xmath202 are determined from the condition @xmath203 @xmath2 0 , @xmath204 @xmath2 @xmath205 + @xmath206 with @xmath205 from eq . and @xmath206 @xmath2 @xmath207 , where @xmath208 and @xmath209 are given in eqs . and . the coefficients @xmath210 can be calculated from the relations @xmath211 @xmath212 where @xmath213 , if @xmath214 and @xmath215 are known . to fix the latter coefficients one could resort , e.g. , to suitable orthonormality conditions . the second approach entails the utilization of the rwa according to the procedure proposed in @xcite : one defines @xmath216 @xmath2 @xmath217 + @xmath218 @xmath2 @xmath205 - @xmath219 , so that @xmath220 @xmath2 @xmath221 . the condition @xmath203 @xmath2 @xmath222 , with @xmath204 @xmath2 @xmath216 + @xmath206 , is then imposed upon expanding @xmath201 in terms of the coherent states @xmath92 @xmath223 on account of the completeness relation @xmath224 @xmath225 @xmath226 , @xmath227 @xmath2 @xmath228 , with @xmath229 @xmath2 @xmath230@xmath231 , being the measure in the complex @xmath95-plane . since @xmath232 @xmath2 @xmath233 , where the coefficients @xmath234 are readily calculated in view of eqs . , , one has @xmath235 |\xi^{\prime}\rangle \ ; . \label{fz1}\ ] ] in eq . , the expansion of @xmath236 in terms of the coherent states @xmath237 , @xmath236 @xmath2 @xmath238 , with @xmath239 @xmath2 @xmath240 , has been utilized . left - multiplying @xmath203 @xmath2 @xmath222 by @xmath241 and resorting to eq . , one finds @xmath242 where @xmath243 denotes the kernel @xmath244 the time - dependent shift of the eigenvalue @xmath245 @xmath2 @xmath246 of the operator @xmath216 implies that also the values of the lamb - dicke parameter , possibly giving rise to a singularity , are shifted from the pathological value to a close one , most likely not singular . to summarize , we have analyzed the properties of the nlcs , which correspond to the experimental scheme of laser - driven trapped ions . upon verifying that they can be defined as both barut - girardello states ( eigenstates of annihilation operator ) and minimum - uncertainty states , we have carried out a mathematical investigation whose main conclusions , in view of physical applications , are : + * for single - boson nlcs , if the squared lamb - dicke parameter happens to be a root of a given laguerre polynomial @xmath247 both the coherence and the minimum - uncertainty properties do not remain valid ; however , the latter feature is still maintained over a wide range of typical values of @xmath248 if @xmath177 is not exceedingly high ( i.e. , one does not consider very excited vibrational levels ) ; if the squared lamb - dicke parameter happens to be a root of a given laguerre polynomial @xmath249 , the very existence of the single - boson nlcs is compromised , due to the oscillating behaviour of the defining series ; * the general construction which leads to the @xmath0-boson nlcs is naturally affected by the same kind of pathologies ; * we have proposed to relax the rwa conditions to get around the above difficulties and have discussed , for the single - boson nlcs , first - order corrections of the known theory ; * from an experimental point of view our results suggest that , due to the finite accuracy in fixing the parameter @xmath248 when preparing single - boson and @xmath0-boson nlcs for actual experiments , one should be cautiously far from the singularities analyzed in this paper in order to maintain the rwa approach . 99 w. vogel and d. g. welsch , _ quantum optics _ ( wiley - vch , weinheim , 2006 ) . v. v. dodonov , m. a. marchiolli , ya . a. korennoy , v. i. manko , and y. a. moukhin , phys . a * 58 * , 4087 ( 1998 ) . w. wu and l. wu , j. opt . b * 5 * , 364 ( 2003 ) . i. jex and v. buek , j. mod . opt * 40 * , 771 ( 1993 ) . k. fujii , e - print arxiv : quant - ph/0112090 . s. sivakumar , j. opt . b : quantum semiclass . opt . * 2 * , r61 ( 2000 ) . r. l. de matos filho and w. vogel , phys . a * 54 * , 4560 ( 1996 ) . f. a. raffa and m. rasetti , laser phys . * 16 * , 1486 ( 2006 ) . f. a. raffa and m. rasetti , int . j. quant . inf . * 5 * , 229 ( 2007 ) . a. napoli and a. messina , eur . j. d * 5 * , 441 ( 1999 ) . j. katriel and a. solomon , phys . a * 49 * , 5149 ( 1994 ) . v. i. manko , g. marmo , e. c. g. sudarshan , and f. zaccaria , phys . scripta * 55 * , 528 ( 1997 ) . m. k. tavassoly and a. parsaiean , j. phys . a : math . theor . * 40 * , 9905 ( 2007 ) . b. r. lavoie and h. de guise , j. phys . theor . * 40 * , 2825 ( 2007 ) . c. c. gerry , opt . comm . * 91 * , 247 ( 1992 ) . m. genovese and c. novero , phys . a * 61 * , 032102 ( 2000 ) . j. s. neergard - nielsen , b. melholt nielsen , c. hettich , k. mlmer , and e. s. polzik , phys . . lett . * 97 * , 083604 ( 2006 ) . a. ourjumtsev , r. tualle - brouri , j. laurat , and p. grangier , science * 312 * , 83 ( 2006 ) . n. ba an , chinese j. phys . * 39 * , 594 ( 2001 ) , e - print arxiv : quant - ph/0103077 . h. moya - cessa , a. vidiella - barranco , j. a. roversi , d. s. freitas , and s. m. dutra , phys . a * 59 * , 2518 ( 1999 ) . k. mlmer and a. srensen , phys . * 82 * , 1835 ( 1999 ) . d. wang , t. hansson , . larson , h. o. karlsson , and j. larson , phys . a * 77 * , 053808 ( 2008 ) . v. manko , g. marmo , a. porzio , s. solimeno , and f. zaccaria , phys . a * 62 * , 053407 ( 2000 ) . z. kis , w. vogel , and l. davidovich , phys . a * 64 * , 033401 ( 2001 ) . r. l. de matos filho and w. vogel , phys . lett . * 76 * , 608 ( 1996 ) . c. a. blockey and d. f. walls , phys . a * 47 * , 2115 ( 1993 ) . j. rnugelj , m. martinis , and v. mikuta - martinis , phys . a * 50 * , 1785 ( 1994 ) . l. f. wei , s. y. liu , and x. l. lei , phys . rev . a * 65 * , 062316 ( 2002 ) . c. c. gerry , s .- c . gou , and j. steinbach , phys . a * 55 * , 630 ( 1997 ) . a. o. barut and l. girardello , commun * 21 * , 41 ( 1971 ) . m. abramowitz and i. a. stegun ( eds . ) , _ handbook of mathematical functions _ ( dover , new york , 1970 ) . p. zanardi and m. rasetti , phys . lett . * 79 * , 3306 ( 1997 ) . a. perelomov , _ generalized coherent states and their applications _ ( springer , berlin , 1986 ) . g. szeg , _ orthogonal polynomials _ ( 4th ed . , american mathematical society , providence , rhode island , 1975 ) .
in order to provide experimentalists with a useful tool for experiments on the @xmath0-boson nonlinear coherent states , we investigate some critical features of these states , as regards their realization as stationary states of a trapped laser - driven ion . after illustrating the @xmath0-boson case , we analyze the single - boson nonlinear coherent states and prove that their coherence and minimum - uncertainty properties are violated whenever the lamb - dicke parameter is one of the roots of the laguerre polynomials which enter such states . we discuss the possibility of circumventing this problem through the relaxing of the rotating wave approximation .
let @xmath0 be a closed orientable euclidean cone 3-manifold with cone angles @xmath4 . its singular locus @xmath5 is a trivalent graph consisting of @xmath6 circles and edges , that we enumerate from @xmath7 to @xmath6 . to those edges one can associate a multiangle @xmath8 , where @xmath9 is the angle of the @xmath10-th circle or edge of @xmath5 . let further @xmath11 denote the vector of lengths of singular circles and edges . we are interested in the space of cone structures on @xmath0 obtained by changing the multiangle . a _ hyperbolic cone structure _ on @xmath0 is a hyperbolic cone manifold @xmath12 with an embedding @xmath13 such that @xmath14 is a retract of @xmath15 . a euclidean cone 3-manifold @xmath0 is _ almost product _ if @xmath16 where @xmath17 is a two - dimensional cone manifold and @xmath18 is a finite subgroup . [ thm : main ] let @xmath0 be a closed orientable euclidean cone 3-manifold with cone angles @xmath19 . if @xmath0 is not almost product , then for every multiangle @xmath20 there exists a unique cone manifold structure of constant curvature in @xmath21 on @xmath0 with those cone angles . if all cone angles of @xmath0 are @xmath22 , then every point in @xmath23 is the multiangle of a hyperbolic cone structure on @xmath0 . if some of the cone angles is @xmath3 , then the subset @xmath24 of multiangles of euclidean cone structures is a smooth , properly embedded hypersurface that splits @xmath25 into 2 connected components @xmath26 and @xmath27 , corresponding to multiangles of spherical and hyperbolic cone structures respectively . furthermore , for each @xmath28 the tangent space of @xmath29 at @xmath30 is orthogonal to the vector of singular lengths @xmath31 . this theorem describes the structures corresponding to multiangles in the open cube . we can also describe the structures corresponding to multiangles contained in some parts of the boundary . for instance , we show that the multiangles contained in @xmath32 ( i.e. none of the angles is @xmath22 and at least one of them is zero ) are angles of hyperbolic cone structures . however , we do not give a description of structures in all points of @xmath33^q$ ] , as this would involve studying collapses at cone angle @xmath22 and deformations and regenerations of seifert fibered geometries . [ cor : sphericity ] let @xmath34 be an irreducible closed orientable @xmath35-orbifold . if there exists a euclidean cone structure @xmath0 on @xmath34 with cone angles strictly less than the orbifold angles of @xmath34 , then @xmath34 is spherical . to prove this corollary , we show in lemma [ lem : cnotap ] that the euclidean cone structure is not almost product , using the irreducibility of @xmath34 . by theorem [ thm : main ] we can obtain a spherical cone structure on @xmath0 by increasing any of the cone angles . we are using here that @xmath29 is orthogonal to the vector of singular lengths @xmath36 . by proposition [ prop : largerspherical ] the orbifold angles can be realized by a spherical cone structure . this corollary gives an alternative argument to the last step in the proof of the orbifold theorem in @xcite , which is more natural from the point of view of cone manifolds . d. cooper and s.p . kerckhoff have announced a different approach to the spherical uniformisation . a special case of this theorem was proved in @xcite , assuming that the singular locus was a knot and a technical hypothesis involving cohomology . even if some of the techniques of @xcite are used , this paper does not rely on it . now we describe the organization of the paper : we are interested in the rotational part of the holonomy of the euclidean cone manifold in @xmath37 ( in fact its lift to @xmath38 ) , that we denote by @xmath39 . some basic properties of this representation are studied in section [ sec : almostproduct ] . in order to deform the structure , we view @xmath37 as the stabilizer of a point in @xmath40 or @xmath41 , thus we study the local properties of the varieties of representations in the isometries of @xmath40 and @xmath41 around @xmath42 . this is done in section [ sec : varofrep ] , using the cohomological results of section [ sec : cohomologytangentbundle ] . in section [ section : regenerationstructures ] we give the conditions for a deformation of the representation @xmath39 to correspond to a path of hyperbolic or spherical cone manifolds . this is applied in section [ sec : pathsofstructures ] to construct paths of hyperbolic and spherical structures by deforming one of the cone angles . the analysis of the local parametrization of the variety of representations is completed in section [ sec : foldlocus ] , where all deformations of the structure are constructed . the global results are established in section [ sec : global ] . finally , in section [ sec : example ] we illustrate the main theorem for cone manifold structures on the 3-sphere with singular locus given by the whitehead link . [ [ acknowledgments ] ] acknowledgments + + + + + + + + + + + + + + + we thank bernhard leeb for helpful conversations on the contents of this paper and for sharing his ideas with us . given a euclidean cone manifold @xmath0 , its smooth part is denoted by @xmath43 . then @xmath44 has a non - complete and non - singular flat riemannian metric . in particular it has a developing map @xmath45 and a holonomy representation @xmath46 we consider the composition of @xmath47 with the projection @xmath48 . [ lem : almostproduct ] @xmath0 is almost product if and only if the image of @xmath49 is contained in @xmath50 . if @xmath0 is almost product then the lemma is straightforward . to prove the converse , we assume that the image of @xmath51 is contained in @xmath52 . since @xmath53 has two components , composition with @xmath54 defines a morphism @xmath55 and , if it is non - trivial , we take the corresponding covering . this induces a branched covering of @xmath0 , with branching locus contained in the edges and circles of @xmath5 that have cone angle @xmath22 . hence we may assume that the image of @xmath51 is contained in @xmath56 . the vertical foliation of @xmath57 invariant under @xmath56 induces an isometric flow of @xmath44 . it extends to a flow on the whole of @xmath0 , since this restriction of the holonomy implies that the components of @xmath5 are parallel to the leaves of the flow . if @xmath58 , then @xmath0 is a smooth flat manifold and the lemma is well known . hence we assume that @xmath59 . since the components of @xmath5 are leaves of the flow , either all leaves are closed or they accumulate on tori at constant distance from the components of @xmath5 . the latter is not possible , since the concentric tori of accumulation of fibers have principal curvatures of constant sign . hence the leaves of the flow are closed , and they are leaves of a seifert fibration , whose basis has a natural structure of a euclidean cone 2-manifold . one can construct a horizontal surface @xmath27 perpendicular to the flow , and we can enlarge it as much as possible to be complete . since @xmath27 has cone angles @xmath60 , the singular locus of @xmath27 is finite . thus @xmath27 is compact and therefore the seifert fibration is virtually a product . the euclidean cone manifold @xmath0 is _ seifert fibered _ if it admits a seifert fibration such that @xmath5 is a union of fibers . it is _ almost seifert fibered _ if it is the quotient of a seifert fibered cone manifold by a finite subgroup of isometries that preserve the fibration . a cone manifold which is almost seifert fibered but not seifert fibered admits a partition by circles and intervals , so that the end - points of the intervals lie in singular edges with cone angle @xmath22 . this corresponds to an orbifold seifert fibration . in particular , if cone angles are @xmath3 , then almost seifert fibered implies seifert fibered . [ cor : almostseifert ] the euclidean cone manifold @xmath0 is almost product iff it is almost seifert fibered . in particular , if all cone angles of @xmath0 are @xmath3 and its smooth part is not seifert fibered , then @xmath0 is not almost product . almost product implies trivially almost seifert fibered . assume now that @xmath0 is seifert fibered . then the smooth part of @xmath0 is a product . let @xmath61 represent the fiber of this fibration . since the components of @xmath5 are also fibers , the holonomy of @xmath62 is either a nontrivial translation or a screw motion with non - trivial translation length . since @xmath62 is central , the direction of this vector must be preserved . thus the image of @xmath51 is contained in @xmath56 . it follows that when @xmath0 is almost seifert fibered , then the image of @xmath51 is contained in @xmath53 , and by lemma [ lem : almostproduct ] @xmath0 is almost product . we consider the action by conjugation of @xmath37 on @xmath63 , the vector space of @xmath64 matrices with real entries . [ lem : invariantsubspace ] @xmath0 is not almost product if and only if the only subspace of @xmath65 invariant under @xmath66 is the space of diagonal matrices . as @xmath37-module acting by conjugation , we have the splitting : @xmath67 where @xmath68 is the set of diagonal matrices , @xmath69 is the subspace of antisymmetric ones and @xmath70 is the subspace of traceless symmetric matrices . the action on @xmath68 is trivial and this space is always invariant . the action on @xmath71 is equivalent to the usual action of @xmath37 on @xmath57 , and having a nonzero invariant vector in @xmath57 is equivalent for @xmath51 to be contained in @xmath56 . since symmetric matrices diagonalize orthogonally , matrices in @xmath72 have an orthogonal basis of eigenvectors . furthermore the sum of eigenvalues is zero , and therefore every non - trivial element has at least a one dimensional eigenspace . thus having a non - trivial invariant element of @xmath72 is equivalent for @xmath51 to have an invariant line , i.e. to be contained in @xmath53 . [ lem : cnotap ] let @xmath0 and @xmath34 be as in the situation of corollary [ cor : sphericity ] . then @xmath0 is not almost product . by hypothesis the cone angles of @xmath0 are @xmath3 . therefore if @xmath0 is almost product , then @xmath0 is a quotient of @xmath73 where @xmath74 is a euclidean turnover ( a cone manifold structure on @xmath75 with three cone points of cone angles satisfying @xmath76 ) . this cone manifold @xmath17 embeds in @xmath0 and induces an essential spherical 2-orbifold in @xmath34 , contradicting the irreducibility of @xmath34 . we assume from now on that @xmath44 is the smooth part of a closed euclidean cone manifold @xmath0 which is not almost product . the latter assumption may be viewed as a nondegeneracy condition and is in that respect similar to the assumption of not being seifert fibered in the deformation theory of spherical cone - manifolds , cf . @xcite , @xcite . in the euclidean case the flat bundle of infinitesimal isometries @xmath77 contains the bundle of infinitesimal translations @xmath78 as a parallel subbundle . the inclusion @xmath79 fits into a short exact sequence of flat vectorbundles and connection - preserving maps @xmath80 where @xmath81 carries the _ quotient _ connection ( @xmath82 is _ not _ parallel ! ) . in fact @xmath83 is again isomorphic to @xmath84 as a flat bundle . this can for instance be seen if we look at the corresponding short exact sequence of @xmath85-modules @xmath86 now we have isomorphisms of @xmath85-modules @xmath87 , which in turn yield isomorphisms of flat bundles @xmath88 . note that the first isomorphism is particular to our 3-dimensional situation . we will freely identify @xmath84 with either @xmath89 or @xmath83 from now on . + the translational part of a euclidean isometry @xmath90 w.r.t . a base point @xmath91 @xmath92 gives rise to a cocycle @xmath93 twisted by @xmath94 , i.e. @xmath95 for @xmath96 . the _ holonomy cocycle _ is the composition @xmath97 for some fixed @xmath98 . this is a cocycle twisted by @xmath51 , i.e. @xmath99 for @xmath100 . the cohomology class of this cocycle is independent of @xmath101 and is denoted by @xmath102\in h^1(\pi_1 m ; \r^3_{rot \circ hol})\,.\ ] ] let @xmath103 be closed and @xmath104 a loop based at @xmath105 . we define the integral of @xmath106 along @xmath104 as @xmath107 where @xmath108 denotes parallel transport along @xmath104 from @xmath109 to @xmath110 and we identify @xmath111 with @xmath112 by developing @xmath44 on @xmath112 . for @xmath113 closed the assignment @xmath114 defines a group cocycle @xmath115 , the so - called period cocycle of @xmath106 . the period map @xmath116 descends to an isomorphism in cohomology @xmath117 & \lmt [ z_{\omega}]\,.\end{aligned}\ ] ] observe that on each side there is a canonically defined cohomology class . on the left hand side we have @xmath118 . @xmath119 is parallel and gives rise to a class @xmath120 \in h^1(m;tm)$ ] . on the right hand side we have the class of the holonomy cocycle @xmath121 . in fact , these classes coincide modulo the period map and correspond geometrically to a global rescaling of @xmath44 : [ lem : pid = v ] @xmath122)=v$ ] and the image of @xmath121 in @xmath123 is tangent to a deformation , which globally rescales the euclidean metric on @xmath44 . let @xmath104 be a loop based at @xmath105 and @xmath124 a lift of @xmath104 to @xmath125 . we compute @xmath126 which shows that the periods of @xmath119 reproduce the holonomy cocycle defined with respect to @xmath127 . furthermore for @xmath128 let @xmath129 then @xmath130 and @xmath131 is the holonomy representation of @xmath44 with rescaled euclidean metric @xmath132 . we obtain @xmath133 and consequently @xmath134 . this means the corresponding path of characters is precisely tangent to @xmath121 , cf . the later discussion in section [ sec : varofrep ] . let @xmath135 denote the @xmath136-tube around @xmath5 in @xmath0 minus the singular locus itself . let @xmath137 denote a compact core of @xmath44 . @xmath138 is a disjoint union of tori and surfaces of higher genus . note that the pair @xmath139 is homotopy equivalent to the pair @xmath140 , by homotopy invariance we may identify the corresponding cohomology groups . + if @xmath141 are cylindrical coordinates around the @xmath10-th edge and @xmath142 is a function with @xmath143 near @xmath144 and @xmath145 near @xmath146 , we define as in @xcite @xmath147 as in @xcite we obtain : [ l2 ] the forms @xmath148 are bounded on @xmath135 , in particular @xmath149 . geometrically , the classes @xmath150 $ ] correspond to deformations of @xmath135 changing the length of the @xmath10-th edge . these deformations are independent . therefore the following does not come as a surprise : [ lem : nontriv ] the classes @xmath150 $ ] are linearly independent in @xmath151 , furthermore the relation @xmath120|_{\partial \overline{m}}=\sum_i [ \omega^i_{len}]$ ] holds . the first assertion follows as in @xcite . the periods of both forms @xmath119 and @xmath152 reproduce the holonomy cocycle restricted to @xmath153 . since the period map descends to an isomorphism in cohomology , the result follows . @xmath154 . this group is isomorphic to the subspace of @xmath155 invariant under @xmath51 , which is trivial since we assume that @xmath0 is not almost product ( see the proof of lemma [ lem : invariantsubspace ] ) . from the long exact cohomology sequence we obtain : @xmath156 and the map @xmath157 is an injection . to compute @xmath158 we use the @xmath149-cohomology : @xmath159 . our starting point is the following theorem of @xcite , where the hypothesis about cone angles is used : [ thm : l2 ] if @xmath44 is the smooth part of a closed euclidean cone manifold with cone angles @xmath60 , then @xmath160 [ cor : l2 ] @xmath161 . since every element of @xmath159 is represented by a parallel form , this cohomology group is isomorphic to the space of elements in @xmath162 which are invariant under the action of @xmath51 by conjugation . by lemma [ lem : invariantsubspace ] , this subspace is one dimensional . let @xmath163 denote a compact core of @xmath44 . by homotopy invariance @xmath164 . [ lem : inj ] the map @xmath165 is an injection . looking at the exact sequence of the pair @xmath166 , it suffices to show that the image of the map @xmath167 is zero . to prove it , we observe that @xmath168 factors through @xmath159 . by corollary [ cor : l2 ] , @xmath159 is one - dimensional and its generator is represented by the identity of @xmath84 . therefore the image of @xmath168 is contained in the span of @xmath120 $ ] . by lemma [ lem : nontriv ] , @xmath120 $ ] restricts to a nontrivial class in @xmath151 . on the other hand , by exactness the image of @xmath168 is contained in the kernel of @xmath169 , hence it is zero . as a corollary of the proof above we notice : [ l2:inj ] the map @xmath170 is an injection . we may thus view @xmath149-cohomology as a subspace of ordinary cohomology in degree 1 . [ lem : h1q ] @xmath171 . by lemma [ lem : inj ] , the long exact sequence of the pair and poincar duality , we have a short exact sequence : @xmath172 since @xmath173 and @xmath174 are poincar dual , @xmath175 in addition , @xmath176 ( see @xcite or @xcite ) . continuing the long exact sequence we get : [ lem : h2isomorphic ] the inclusion induces an isomorphism @xmath177 . let @xmath178 denote the cohomology of the @xmath179-module @xmath57 with the action induced by @xmath51 . notice that the induced flat bundle over @xmath44 is isomorphic to @xmath84 . [ lem : cohomologies ] there is a natural map @xmath180 which is an isomorphism for @xmath181 and an injection for @xmath182 . when @xmath7 , this map is the inverse of the isomorphism @xmath183 in lemma [ lem : pid = v ] . there is a natural isomorphism @xmath184 , where @xmath185 denotes any aspherical cw - complex with the same fundamental group as @xmath44 , and @xmath186 also denotes the flat bundle over @xmath185 with fiber @xmath187 and holonomy @xmath51 . the space @xmath185 can be constructed from @xmath44 itself by adding cells of dimension @xmath188 . thus the cohomology of the pair @xmath189 is trivial in dimension @xmath190 , and the lemma follows easily from the long exact sequence in cohomology for the pair @xmath189 . [ cor : injectiontoboundary ] let @xmath191 denote the components of @xmath192 . the restriction induces an injection : @xmath193 the result follows from corollary [ lem : h2isomorphic ] , lemma [ lem : cohomologies ] and the asphericity of the surfaces @xmath194 . since @xmath44 is three dimensional by a theorem of culler , cf . @xcite , @xmath47 lifts to a representation @xmath195 and similarly @xmath51 lifts to @xmath196 where @xmath197 is a lift of @xmath94 . we will drop the notational distinction between @xmath198 and @xmath47 , resp . between @xmath199 and @xmath94 from now on . let @xmath200 and @xmath201 denote the varieties of representations in @xmath202 and @xmath203 respectively . let @xmath204 and @xmath205 be the varieties of characters . let @xmath206 denote the character of @xmath39 . we recall some general facts about spaces of representations : let @xmath207 be a finitely generated discrete group and @xmath208 a lie group . we equip @xmath209 with the compact - open topology . once and for all we fix a presentation @xmath210 . without loss of generality we may assume that the set of generators contains with any of its elements its inverse . the relations @xmath211 determine functions @xmath212 which identify @xmath213 with the subset @xmath214 . if @xmath208 is a real algebraic group ( as for example @xmath215 or @xmath216 ) , the @xmath217 are polynomial functions and @xmath213 acquires the structure of a real algebraic set . similarly , if @xmath208 is a complex algebraic group ( for example @xmath218 ) , then @xmath213 is a complex algebraic set . let @xmath219 be the semidirect product of @xmath208 and @xmath220 defined via @xmath221 . let @xmath222 be given . a simple but important observation is that @xmath223 if and only if @xmath224 . the relations therefore determine functions @xmath225 such that @xmath226 and @xmath227 . note that for @xmath228 fixed the functions @xmath229 are linear . if we identify @xmath230 with @xmath231 via right translation , then we have : @xmath232 . the differential of the group multiplication @xmath233 at @xmath234 considered as a map @xmath235 via right translation is given by @xmath236 on the other hand the group multiplication law in @xmath237 reads @xmath238 which proves the claim in this case . the general case is easily reduced to this one . in particular if the @xmath217 cut out @xmath213 transversely , then @xmath239 is identified with @xmath240 ) . in general , if @xmath241 is a smooth point , there is a linear injection @xmath242 which identifies @xmath239 with a subspace of @xmath243 . this injection is known as _ weil s construction _ , and it extends naturally to the non - smooth case by using the zariski tangent space . a cocycle @xmath244 is a coboundary , i.e. @xmath245 , if and only if @xmath244 is tangent to the orbit of the action of @xmath208 on @xmath213 via conjugation . [ rep_var_smooth ] @xmath246 is a smooth point with tangent space @xmath247 . similarly , @xmath248 is a smooth point with tangent space @xmath249 . there is an infinite sequence of obstructions for a cocycle @xmath250 to be integrated into an actual path of representations tangent to @xmath244 , cf . these obstructions live in @xmath251 . using the injection of corollary [ cor : injectiontoboundary ] and the fact that the obstructions are natural , we conclude that they vanish , because they vanish in @xmath252 . then using artin s theorem , cf . @xcite , they are in fact integrable . thus every element in @xmath247 is integrable , and therefore @xmath39 is a smooth point of @xmath253 with tangent space @xmath247 . for @xmath254 the same argument applies since @xmath255 is the complexification of the module @xmath256 . @xmath257 is a smooth point of local real dimension @xmath6 and tangent space isomorphic to @xmath258 . similarly , @xmath259 is a smooth point of local complex dimension @xmath6 and tangent space isomorphic to @xmath260 . since @xmath39 is irreducible , @xmath261 in each case . since @xmath202 is compact ( resp . the action of @xmath262 is proper on the irreducible part of @xmath263 , cf . * lemma 6.24 ) ) and the tangent space to the orbit is given by @xmath247 ( resp . by @xmath249 ) , the result follows with lemma [ rep_var_smooth ] . the fiber of the map @xmath264 at @xmath265 is the the space of cocycles @xmath266 , which is given by a system of linear equations whose coefficients depend continuously on @xmath228 . therefore its dimension is an upper semi - continuous function of @xmath228 in general . if @xmath246 is a smooth point with @xmath267 ( which is the case if the @xmath217 cut out @xmath253 transversely or for @xmath268 according to lemma [ rep_var_smooth ] ) , then @xmath269 injects into @xmath266 for @xmath228 in a neighbourhood of @xmath39 via weil s construction ( [ eqn : weil ] ) : @xmath270 and hence @xmath271 is locally the projection of a vector bundle . more precisely we have : @xmath272 is isomorphic to the tangent bundle @xmath273 near @xmath39 . the map @xmath274 is a vector bundle isomorphism near @xmath39 . @xmath275 is a smooth point . let @xmath276 denote the character of @xmath275 . [ vectorbundle ] @xmath276 is a smooth point of local real dimension @xmath277 . the induced map @xmath278 is locally the projection of a vector bundle isomorphic to the tangent bundle @xmath279 near @xmath39 . let us digress into a more general situation first : let @xmath208 be a compact lie group , @xmath44 a smooth manifold ( not necessarily compact ) and @xmath280 a smooth free action . the associated infinitesimal action is the lie algebra homomorphism @xmath281 defined by @xmath282 . since the original action was free , this homomorphism is injective and we will identify @xmath231 with its image in @xmath283 . let @xmath219 be the semidirect product of @xmath208 and @xmath284 defined via @xmath285 . we extend the natural action of @xmath208 on @xmath84 via the differential to an action of the group @xmath286 in the following way : @xmath287 the relation @xmath288 ensures that this defines a group action . this action is free since the original action was free , it is clearly proper . [ extendedaction ] @xmath289 . the fibers of the map @xmath290 are precisely the orbits of the action of the group @xmath237 on @xmath84 . we return to the situation of proposition [ vectorbundle ] : clearly @xmath291 via @xmath292 as @xmath202-modules . for a representation @xmath293 and @xmath294 we have @xmath295 therefore the map @xmath296 is @xmath216-equivariant and lemma [ extendedaction ] yields the proposition . from the long exact cohomology sequence and proposition [ vectorbundle ] we obtain : [ shortexact ] there is a short exact sequence @xmath297 modulo the choice of a splitting of the exact sequence in corollary [ shortexact ] , an infinitesimal deformation of the holonomy of the euclidean structure on @xmath44 is therefore determined by the infinitesimal deformation of its rotational part and of its translational part , and both can be independently prescribed . let @xmath298 a system of meridians for @xmath179 ( i.e. one for each component of @xmath5 ) . we define the angle function @xmath299 in a neighborhood @xmath300 of @xmath301 , so that @xmath302 . it is related to the trace by the equality : @xmath303 in particular @xmath304 is analytic . we extend @xmath304 to a neighborhood in @xmath305 as a complex analytic function : @xmath306 so that the complex length of @xmath307 is @xmath308 ( i.e. a translation of length @xmath309 plus a rotation of angle @xmath310 ) . + the differentials @xmath311 live in the cotangent space to the varieties of characters , thus @xmath312 [ prop : kerdmu ] @xmath313 , where @xmath121 is the class of the holonomy cocycle . to check that @xmath314 , we observe that @xmath315 maps @xmath121 to zero , because @xmath316 is a rotation with a fixed axis . in particular @xmath311 evaluated at @xmath121 is zero . to see the other inclusion , we use that @xmath317 , \ldots , [ \omega^q_{len } ] \right\rangle \,,\ ] ] where we view @xmath158 as a subspace of @xmath318 via lemma [ lem : inj ] . since the forms @xmath319 are @xmath149 near the singular locus according to lemma [ l2 ] , we conclude that @xmath320 , which we view as a subspace of @xmath158 via corollary [ l2:inj ] . since @xmath159 is spanned by @xmath120 $ ] the result follows . [ cor : dimdmu ] @xmath321 along this section @xmath322 denotes either @xmath41 or @xmath40 . the stabilizer of a point @xmath323 is isomorphic to @xmath324 . thus we view @xmath39 as a representation of @xmath179 in this stabilizer . we consider a path of characters @xmath325 with @xmath326 . assume that this path is differentiable to the right at @xmath327 . the derivative @xmath328 is an element of @xmath329 , where @xmath284 is the lie algebra of @xmath330 ( using weil s construction ) . the lie algebra @xmath284 decomposes into rotational and translational part with respect to @xmath101 : @xmath331 where @xmath332 is the lie algebra of the stabilizer of @xmath323 . thus @xmath333 [ thm : regeneration ] let @xmath334 be a path in @xmath335 with @xmath326 and differentiable to the right at @xmath327 . if @xmath336 , then @xmath337 is the holonomy character of a @xmath322-structure on @xmath44 , for @xmath338 and some @xmath339 . if in addition @xmath340 is a rotation for each meridian @xmath341 , then the structure on @xmath44 completes to a cone manifold structure . consider @xmath342 , the space of constant sectional curvature @xmath343 , where @xmath344 is the curvature of @xmath322 , for @xmath345 . equivalently , @xmath342 is @xmath322 with the metric tensor scaled by @xmath346 . we fix a base point @xmath101 in @xmath342 independently of @xmath347 , so that the pointed euclidean space @xmath348 is the limit of @xmath349 when @xmath350 , and consider the union @xmath351}\mathbf x_t.\ ] ] we equip @xmath352 with a manifold structure , using the local charts given by the exponential maps at @xmath101 and the parameter @xmath345 , after identifying isometrically @xmath353 for every @xmath354 . notice that the choice of @xmath101 is relevant for the topology of @xmath352 . choose a smooth path of representations @xmath131 with character @xmath337 so that @xmath355 and @xmath356 . for every @xmath357 , the action of @xmath358 on @xmath342 for @xmath359 extends continuously to the action of @xmath360 on @xmath361 for the @xmath362-topology . notice that @xmath131 acts on @xmath342 isometrically , since rescaling the metric does not change the isometry group . the previous lemma provides an action on @xmath352 , which is isometric on each @xmath342 . we start describing the local coordinates . let @xmath363 denote the riemannian exponential . using the isometric identification @xmath364 , we have @xmath365 where @xmath366 . thus , the ( inverse of ) the local charts is given as follows . given an open set @xmath367 , @xmath368 parametrizes a subset of @xmath352 via the map : @xmath369 to prove the lemma it suffices to show that @xmath370 uniformly for @xmath371 in a compact set for the @xmath362-topology . the lie algebra of @xmath372 decomposes as a sum @xmath373 , where @xmath374 is the subalgebra of infinitesimal rotations around @xmath101 and @xmath375 the subspace of infinitesimal translations with respect to @xmath101 . there is an isometric identification @xmath376 , so that for any @xmath377 : @xmath378 here we view @xmath379 when we write @xmath380 and @xmath381 for @xmath382 . according to global cartan s decomposition , we write @xmath383 , where @xmath384 and @xmath385 belongs to the stabilizer of @xmath101 in @xmath386 . notice that @xmath387 as @xmath350 by hypothesis . using this notation : @xmath388 where @xmath389 is an analytic function . here we use that @xmath390 together with the campbell - hausdorff formula . using again this formula and the fact that @xmath391 , we get : @xmath392 where @xmath393 is also analytic . hence @xmath394 when @xmath350 this converges to @xmath395 uniformly for @xmath121 in a compact subset of @xmath187 , because @xmath396 and @xmath397 is a lift of @xmath398 , by hypothesis . to prove convergence in the @xmath362-topology , we write @xmath399 with @xmath400 , and @xmath401 small . the previous calculation yields easily : @xmath402 we can compute the derivative with respect to @xmath136 when @xmath403 and we get @xmath362-convergence . we modify slightly goldman s construction to deform the structure . let @xmath163 be a compact core of @xmath404 . consider @xmath405 , which is a bundle over @xmath163 with fiber @xmath406 . it is in fact a union of bundles @xmath407 with fiber @xmath408 . each @xmath409 has a natural flat connection that varies continuously with @xmath347 . the developing map of the euclidean structure induces a section @xmath410 with values in @xmath361 transverse to the flat connection . since the image of @xmath411 is compact , we compose it with the flow @xmath412 of the vector field tangent to the direction of @xmath347 ( this is defined globally when @xmath413 but not when @xmath414 , because @xmath361 is not homeomorphic to @xmath342 ) . for small values of @xmath359 , @xmath415 is a section , still transverse to the flat connection by @xmath362-continuity , hence inducing a developing map . this provides a structure on @xmath163 , that can be completed by controlling its behavior on @xmath416 , using the hypothesis about the meridians . the volume of these cone manifolds is increasing with @xmath347 , because the starting euclidean structure is viewed as totally degenerate with volume zero . in fact we get more precisely : [ prop : schlafli ] let @xmath417 be the family of cone manifolds constructed in theorem [ thm : regeneration ] of constant curvature @xmath418 . assume moreover that the path of characters is analytic . then @xmath419 where the sum runs over the edges and circles @xmath420 of @xmath5 . _ sketch of proof . _ the construction of developing maps in the proof of theorem [ thm : regeneration ] can be made so that we have a set of points @xmath421 such that the balls @xmath422 cover @xmath0 , where @xmath211 is much smaller that the injectivity radius , and @xmath423 varies analytically ( we assume that the path of characters is analytic ) . let @xmath424 the cone structure with holonomy @xmath337 and define @xmath425 by construction , this @xmath426 is an analytic family of polyhedra in the space @xmath322 . analyticity implies that the topological type of @xmath426 changes in a discrete subset of times @xmath347 . thus we may apply schlfli s formula to them . adding all the terms , we get the formula of the proposition , because the contribution of nonsingular edges is trivial . see ( * ? ? ? 4.2 ) for further details . [ rem : trans ] in the hyperbolic case , @xmath427 and @xmath428 . therefore in this case @xmath429 can be viewed as the imaginary part . in the spherical case , @xmath430 and the lie algebra of the stabilizer of a point @xmath431 is conjugate to the diagonal subalgebra of @xmath432 . therefore in this case @xmath433 can be viewed as the difference between factors . [ rem : rotation ] for every @xmath228 with character @xmath434 , @xmath435 for every @xmath436 with character @xmath437 , @xmath438 up to changing the indices , by corollary [ cor : dimdmu ] we may assume that @xmath439 are linearly independent . thus the set @xmath440 is a smooth complex curve in a neighborhood @xmath300 of @xmath301 . notice that the class of the holonomy cocycle @xmath121 is the tangent vector to @xmath441 at @xmath301 , by proposition [ prop : kerdmu ] . thus , using remark [ rem : trans ] we have : to every path @xmath442 with @xmath443 and @xmath444 , theorem [ thm : regeneration ] applies . [ lem : deg2 ] the restricted map @xmath445 is a branched covering of degree two . by proposition [ prop : kerdmu ] , @xmath446 , thus @xmath447 is either constant or a branched covering of degree @xmath448 . seeking a contradiction , assume that @xmath447 is constant , then we have a path of characters in the curve @xmath449 to which we apply theorem [ thm : regeneration ] . since each @xmath450 stays constant on this curve , this path corresponds to hyperbolic cone manifolds with constant cone angle . thus by schlfli s formula ( proposition [ prop : schlafli ] ) they have constant volume , contradicting the fact that the volume increases from zero . thus @xmath447 is a branched covering of degree @xmath451 . to prove that the degree @xmath452 is precisely @xmath453 , we assume that @xmath454 and seek again a contradiction . we look at the inverse image or real points @xmath455 , because for characters here , the image of @xmath456 is a rotation . this inverse image is a graph with @xmath457 branches starting at @xmath301 , and the angle between the branches is @xmath458 . thus two of the branches are real and the remaining @xmath459 have nontrivial imaginary part . hence theorem [ thm : regeneration ] applies to those @xmath459 branches . in addition , if @xmath460 , then there are branches for which @xmath456 is strictly larger than @xmath461 and branches for which @xmath456 is strictly less than @xmath461 . since @xmath456 is the cone angle , we have constructed regenerating families of hyperbolic cone manifolds with both increasing and decreasing cone angles , contradicting schlfli s formula . [ cor : hyperbolicregeneration ] there is a family of hyperbolic cone structures obtained by decreasing @xmath462 . since @xmath447 is a branched covering of degree 2 , @xmath455 has 4 branches starting at @xmath301 . two of them are real , and to the other two one applies theorem [ thm : regeneration ] . the cone angle @xmath462 of those branches must decrease by schlfli s formula . these two branches correspond to complex conjugate representations , i.e. with opposite orientations , because changing the sign of @xmath121 corresponds to changing the orientation . there is a family of spherical cone structures obtained by increasing @xmath462 . to construct the spherical structure we consider @xmath463 this is a real analytic curve in a neighborhood @xmath464 of @xmath301 . it is the set of real points of @xmath441 . by lemma [ lem : deg2 ] , the map @xmath465 is locally equivalent to the map @xmath466 in a neighborhood of @xmath467 . thus the inverse fiber of the map @xmath468 consists of pairs of points , except for @xmath461 , which consists of a single point . let @xmath469 be two subintervals of @xmath470 such that @xmath471 and @xmath472 . we consider the set @xmath473 because @xmath474 is the condition that guarantees that @xmath475 is the character of a representation @xmath228 in @xmath476 so that @xmath477 is a rotation , by remark [ rem : rotation ] . theorem [ thm : regeneration ] applies to the latter two branches , because the difference of tangent vectors gives the translational part , that is necessarily a non - zero multiple of @xmath121 ( see remark [ rem : trans ] ) . if we replace @xmath478 by @xmath479 ( i.e. we change the order of the factors ) then we get the same structures with different orientation . [ rem : furtherregenerations ] we have constructed regenerations by deforming the cone angle @xmath462 because we assumed that the forms @xmath480 are linearly independent . below in corollary [ cor : linearrelation ] we prove that the linear relation satisfied by these forms is : @xmath481 thus we can regenerate by deforming any of the cone angles . now we analyze the behavior of the real analytic map @xmath482 in a neighborhood of @xmath301 . [ lem : equivalencex2 ] in a neighborhood of @xmath301 , @xmath483 is equivalent to the map @xmath484 in a neighborhood of the origin via a real analytic change of coordinates at both the source and at the target . by corollary [ cor : dimdmu ] and up to permuting coordinates , we may assume that the forms @xmath485 are linearly independent . thus the set @xmath486 is a smooth ( real analytic ) curve in a neighborhood @xmath300 of @xmath301 . by lemma [ lem : deg2 ] we can choose real analytic coordinates @xmath487 in a neighborhood of @xmath301 with @xmath488 such that @xmath489 and @xmath490 for @xmath491 . note that the curve @xmath492 corresponds to the @xmath493-axis in these coordinates . since @xmath494 and @xmath495 , the implicit function theorem yields that the set @xmath496 is locally around @xmath327 the graph of a real analytic function @xmath497 . let @xmath498 be the real analytic function defined by @xmath499 let further @xmath500 be the unique real analytic function satisfying @xmath501 and @xmath502 . then the map @xmath90 defined by @xmath503 is a local diffeomorphism . we obtain @xmath504 and @xmath505 for @xmath491 . with the local diffeomorphism @xmath506 defined by @xmath507 we obtain @xmath508 . the same result holds true for the complex analytic extension @xmath509 just by composing with the complex analytic extensions of the coordinate changes . the _ fold locus _ @xmath510 is the set of points where @xmath483 is not a local diffeomorphism . in the coordinate system of lemma [ lem : equivalencex2 ] the fold locus is the set @xmath511 , hence : [ cor : jmanifold ] there exists a neighborhood @xmath300 of @xmath301 such that both @xmath512 and @xmath513 are codimension one submanifolds . in addition , for each @xmath514 , @xmath515 . each @xmath516 is the rotational part of the holonomy of a euclidean cone manifold structure . the translational part lies in @xmath517 . when we deform a character @xmath518 in @xmath519 we can also deform continuously the vector @xmath121 in @xmath520 , which is a one dimensional subspace that varies continuously , by lemma [ lem : equivalencex2 ] . this corresponds to deforming continuously the holonomy @xmath47 by a family of representations of @xmath179 in @xmath216 . the condition @xmath521 ensures that the image of meridians are rotations ( cf . * prop . 9.6 ) ) . since those representations map the meridians to rotations , it follows that they correspond to holonomy representations of euclidean cone manifolds . [ cor : eeucl ] points of @xmath522 are angles of euclidean cone manifold structures on @xmath0 . let @xmath523 denote the lengths of the singular circles and components of @xmath0 . those are unique up to homothety . [ prop : lnormal ] the vector @xmath36 is normal to @xmath522 at @xmath524 . this is a consequence of schlfli s formula ( prop . [ prop : schlafli ] ) . [ cor : linearrelation ] on @xmath525 , @xmath526 . it holds on @xmath527 by proposition [ prop : lnormal ] . in addition , it also holds on @xmath528 $ ] . by lemma [ lem : equivalencex2 ] and proposition [ prop : kerdmu ] , those spaces generate the whole @xmath529 . it follows from this corollary that we can regenerate into hyperbolic or spherical cone manifold structures by decreasing or increasing any of the cone angles , the geometry depends on the sign of @xmath530 . ( see remark [ rem : furtherregenerations ] . ) [ cor : lrecm ] let @xmath0 be a closed euclidean cone manifold with cone angles @xmath60 which is not almost product . then deformations up to dilations of @xmath0 into euclidean cone manifold structures are parametrized by the @xmath6 cone angles in a manifold @xmath29 of dimension @xmath531 and transverse to the vector of singular lengths @xmath532 . the previous analysis determines all representations of @xmath179 into @xmath216 up to conjugation that map meridians to rotations , i.e. pairs @xmath533 such that @xmath534 and further @xmath535 , see figure [ fig : foldlocus ] . [ cor : lrr ] let @xmath0 be as above . then deformations of @xmath0 into constant curvature structures up to dilations are parametrized by the @xmath6 cone angles , and the type of geometry depends on the side of @xmath29 . the analysis of this section and previous ones determine the structure of the spaces of characters of @xmath179 in @xmath330 . namely , spherical regenerations correspond to pairs of @xmath202-characters @xmath475 such that @xmath536 . according to lemma [ lem : equivalencex2 ] , these pairs are determined by the value of @xmath483 , which is the multiangle of the structure , cf . figure [ fig : foldlocus ] . hyperbolic regenerations correspond to @xmath262-characters @xmath518 with @xmath537 , which are not @xmath202-characters , cf . figure [ fig : hypreg ] . again such a character is determined up to complex conjugation by the value of @xmath483 . next , we shall show that the deformations of the structures are uniquely determined by the deformations of the characters , adapting an argument from @xcite . let @xmath163 be a compact core of @xmath404 , so that we will consider deformations of structures on @xmath163 , knowing that the structure on the surface @xmath416 is determined by the holonomy , and the structure on @xmath538 is given by taking the cone . using the notation of section [ section : regenerationstructures ] , @xmath539 is the union of flat bundles @xmath409 with fiber @xmath342 the space of constant curvature @xmath540 , and holonomy the deformation of the euclidean holonomy representation . since @xmath409 is flat , there is a horizontal foliation , which is also used to define locally a projection onto the fibers @xmath342 . the euclidean structure is induced by a section @xmath541 transverse to the horizontal foliation . in section [ section : regenerationstructures ] , @xmath542 is deformed to @xmath543 , following the flow in the time direction on @xmath29 . let @xmath544 be another deformation , i.e. a section of class @xmath362 such that @xmath545 and @xmath546 . we claim that , for @xmath547 sufficiently small , there exist a diffeomorphism @xmath548 of @xmath44 isotopic to the identity so that the developing maps corresponding to @xmath549 and @xmath550 are equivalent ( namely , the sections @xmath549 and @xmath550 composed with the local projections to @xmath342 differ by isometries of @xmath551 ) . following @xcite , we take a foliated neighborhood @xmath552 of @xmath553 in @xmath554 , so that the intersection of leaves with @xmath552 are balls , and consider @xmath555 in @xmath29 , via the flow as in section [ section : regenerationstructures ] . for @xmath347 sufficiently small , closeness and transversality imply that the image of both @xmath556 and @xmath549 meet each horizontal leave in @xmath557 precisely once . the intersection with the horizontal leaves can be used to define the map @xmath548 as in @xcite . in this section we finish the proof of theorem [ thm : main ] . we first establish some properties of the set of euclidean angles and recall some results about the set of hyperbolic and spherical ones . let @xmath29 denote the set of multiangles with a euclidean cone manifold structure . [ prop : completesubmfld ] @xmath29 is a properly embedded hypersurface of @xmath23 . first we show that every euclidean cone manifold structure on @xmath0 with cone angles in @xmath558 is not almost product . this will imply that @xmath29 is a @xmath531 submanifold , by corollary [ cor : eeucl ] . by contradiction , we assume that there exists such a euclidean cone structure on @xmath0 which is almost product . since all cone angles are in @xmath558 , this implies that @xmath0 is seifert fibered and @xmath5 is a union of fibers . by corollary [ cor : almostseifert ] , in this case @xmath0 is almost product for any cone angle , contradicting the hypothesis of theorem [ thm : main ] . to prove properness , we show that if @xmath559 is a sequence in @xmath29 converging to @xmath560 , then @xmath561 . let @xmath562 be the corresponding sequence of cone structures on @xmath0 with multiangles @xmath559 . we can rescale them so that @xmath563 . since the cone angles are uniformly @xmath3 , by theorem 5.3 and corollary 5.4 of @xcite the cone manifolds @xmath562 are uniformly thick : they have a base point @xmath564 with @xmath565 . thus they have a convergent subsequence to a euclidean cone manifold of diameter 1 , hence a cone structure on @xmath0 with cone angles @xmath566 . in ( * ? ? 3 ) the following proposition is proved , which applies to @xmath0 under the assumptions of the paper . [ prop : largerspherical ] if @xmath567^q$ ] is a multiangle of a spherical cone structure on @xmath0 which is not seifert fibered , then every @xmath568^q$ ] with @xmath569 is the multiangle of a spherical cone structure with cone angles @xmath570 . if @xmath567^q$ ] is a multiangle of a hyperbolic cone structure on @xmath0 , then every @xmath568^q$ ] with @xmath571 is the multiangle of a hyperbolic cone structure with cone angles @xmath570 . [ rem : smallness ] when all cone angles are @xmath22 , the orbifold is spherical and hence small ( without essential 2-suborbifolds ) . in particular all turnovers in @xmath0 are compressible or boundary parallel . before proving theorem [ thm : main ] , we need to establish incompatibility between structures of different sign . we normalize the constant curvature to be @xmath572 , @xmath327 or @xmath573 . [ lem : incompatibility ] given @xmath574^q$ ] , @xmath575 can not be the multiangle of two cone structures with different normalized constant curvatures ( @xmath572 , @xmath327 or @xmath573 ) . seeking a contradiction , assume for instance that @xmath576^q$ ] is the multiangle of a hyperbolic and euclidean cone structure simultaneously . using corollary [ cor : hyperbolicregeneration ] , the euclidean cone structure can be regenerated to hyperbolic ones , which , by construction , have arbitrarily small diameter , and also cone multiangle arbitrarily close to @xmath575 . on the other hand , when we perturb the hyperbolic cone structure with cone multiangle @xmath575 , we find hyperbolic cone manifolds with the same multiangles but diameter bounded below away from zero . this contradicts the global rigidity of hyperbolic cone manifolds proved in ( * ? ? ? the same argument works for spherical and euclidean ( global rigidity in the spherical case is the content of ( * ? ? ? * thm . 1.7 ) ) . finally we deal with the case that @xmath575 is simultaneously the multiangle of a hyperbolic and spherical structure on @xmath0 . consider @xmath577 sufficiently large so that @xmath578 for @xmath579 . by proposition [ prop : largerspherical ] , @xmath580 is the multiangle of hyperbolic cone manifold , hence a hyperbolic orbifold . notice that this orbifold can not be spherical , because its fundamental group is infinite . consider now a path of cone multiangles between @xmath575 and @xmath580 which is decreasing along each component . by proposition [ prop : largerspherical ] every multiangle in this path corresponds to a hyperbolic cone structure . however , sphericity must fail at some multiangle of this path , because @xmath580 is not the multiangle of a spherical structure . by the arguments in the proof of proposition [ prop : completesubmfld ] , when sphericity fails we find precisely a euclidean structure , and we have reduced to the first case of incompatibility between hyperbolic and euclidean structures . the following proposition is also used in the proof of theorem [ thm : main ] . [ prop : angle0 ] every multiangle in @xmath581 with at least one angle @xmath327 is the multiangle of a hyperbolic cone structure . by ( 1.5 ) , the smooth part of @xmath0 is hyperbolic . applying thurston s dehn filling theorem , we have the proposition for a neighborhood of the origin . to cover the rest of the multiangles , we just have to enlarge some of the angles , keeping the other ones equal to zero ( i.e. complete cusps ) . thus we have a lower bound on the cone angles , and in addition the diameter is infinity . applying the results of @xcite , we can enlarge each one of the cone angles up to @xmath3 . notice that some convergence results of @xcite apply to cone manifolds without essential turnovers . in our case there are no turnovers by remark [ rem : smallness ] except if we allow cusped turnovers ( i.e.turnovers with some cone angle 0 ) . however cusped turnovers are not a problem in those arguments , because all cone angles are @xmath3 , and therefore they can not converge to a euclidean cone 2-manifold . [ rem : opennessat0 ] the hyperbolic structures of the previous proposition can be deformed in a neighborhood in @xmath581 . this remark can be easily proved adapting the arguments of the proof of hyperbolic dehn filling for orbifolds in ( * ? ? ? b ) and using the infinitesimal rigidity results established in @xcite . _ proof of theorem [ thm : main ] : _ assume first that @xmath0 is a euclidean cone manifold as in the statement and all cone angles are @xmath22 . then using the regeneration results of section [ sec : pathsofstructures ] and proposition [ prop : largerspherical ] , every point in @xmath23 is the multiangle of a hyperbolic cone structure on @xmath0 . assuming that at least one of the angles of @xmath0 is @xmath3 , then there exists a spherical cone structure on @xmath0 with all cone angles @xmath22 , again by section [ sec : pathsofstructures ] and proposition [ prop : largerspherical ] . we consider all segments in @xmath581 starting at @xmath582 and ending at some point with at least one coordinate zero . the first point of the segment is the multiangle of a spherical cone structure and the last one is the multiangle of a hyperbolic one , by proposition [ prop : angle0 ] . by remark [ rem : opennessat0 ] , we can assume that the multiangle of the hyperbolic structure lies in the open cube @xmath23 . starting at @xmath583 , we move along the segment by decreasing the cone angles and obtaining a family of spherical cone manifolds . this family can not be spherical all the time because the endpoint of the segment corresponds to a hyperbolic structure and they are incompatible by lemma [ lem : incompatibility ] . since all cone angles are @xmath3 , then the end of the spherical subsegment is the multiangle of a euclidean structure , by the same argument as in proposition [ prop : completesubmfld ] . by the regeneration results of section [ sec : pathsofstructures ] and proposition [ prop : largerspherical ] , we connect the multiangle of this euclidean structure to the endpoint of the segment by multiangles of hyperbolic cone structures . the previous argument shows that every point in @xmath23 is the multiangle of a constant curvature cone structure on @xmath0 , and that @xmath29 is a hypersurface that divides @xmath23 in two components @xmath27 and @xmath26 corresponding respectively to hyperbolic and spherical multiangles . for the uniqueness , notice first that in lemma [ lem : incompatibility ] we establish incompatibility between hyperbolic , spherical and euclidean for a given multiangle . global rigidity for hyperbolic and spherical structures is proved in @xcite . using local rigidity of the regenerations ( corollary [ cor : lrr ] ) and global rigidity for the hyperbolic structures , we get global rigidity for the euclidean ones . in this section we illustrate the main theorem for cone manifold structures on the 3-sphere with singular locus given by the whitehead link , which is the 2-component link depicted in figure [ fig : whitehead ] . furthermore we discuss geometric structures corresponding to multiangles contained in the boundary of @xmath584 ^ 2 $ ] to some extent . shmatkov has computed the curve of multiangles of euclidean structures by constructing the corresponding cone manifolds explicitly , cf . these structures are not almost product . the hyperbolic region had earlier been computed by a. mednykh in a similar way , cf . @xcite and references therein . alternatively , one can proceed as follows : let @xmath44 denote the complement of the whitehead link in @xmath585 . if @xmath586 are meridians around the two components of @xmath587 , the fundamental group has the following presentation , cf . @xcite : @xmath588 with @xmath589 . the @xmath254-character variety of @xmath44 has been computed in @xcite . namely , after identifying @xmath590 with the image of the map @xmath591 in @xmath592 , it is given by @xmath593 with @xmath594 and @xmath595 holonomy representations of hyperbolic cone manifold structures lift to irreducible representations @xmath228 with @xmath598 and @xmath599 . we may write @xmath600 and @xmath601 with @xmath602 and @xmath603 the cone angles around the two components of @xmath587 . rotational parts of euclidean holonomies correspond to representations as above where in addition @xmath604 and @xmath605 fail to be local coordinates . this is precisely the locus where the discriminant of @xmath101 computed with respect to the variable @xmath244 vanishes . the discriminant is given by the following polynomial : @xmath606 for the resulting curve of multiangles of euclidean cone manifold structures see figure [ fig : euclideanregion ] . for multiangles contained in some parts of the boundary of @xmath607 ^ 2 $ ] we can construct cone manifold structures as well using our main theorem : for multiangles of the form @xmath608 , @xmath609 , resp . @xmath610 , @xmath611 , we obtain hyperbolic cone manifold structures , whereas for multiangles of the form @xmath612 , @xmath613 , resp . @xmath614 , @xmath615 , we obtain spherical ones . let us suppose there existed such a structure . then we consider the double branched cover of @xmath585 branched along the component of the whitehead link with cone angle @xmath22 . this branched cover is again @xmath585 since the components of the whitehead link are unknotted . the other component of the whitehead link lifts to the torus link @xmath622 , whose complement is known to be non - hyperbolic . on the other hand the lift of the intital hyperbolic cone metric will be a hyperbolic cone metric on @xmath585 with singular locus @xmath622 , which is a contradiction in view of the results in @xcite . , _ notes on notes of thurston _ , in _ analytical and geometric aspects of hyperbolic space _ ( ed . by d. b. a. epstein ) , london math . lecture notes , ser . * 111 * , 1987 , cambridge univ . press , cambridge , pp . 392 . , _ geometric structures on manifolds and varieties of representations _ , in _ geometry of group representations _ ( boulder , co , 1987 ) , pp . 169198 , contemp . math * 74 * , amer . soc . , providence , ri , 1988 .
given a closed orientable euclidean cone 3-manifold @xmath0 with cone angles @xmath1 and which is not almost product , we describe the space of constant curvature cone structures on @xmath2 with cone angles @xmath3 . we establish a regeneration result for such euclidean cone manifolds into spherical or hyperbolic ones and we also deduce global rigidity for euclidean cone structures .
understanding the intrinsic mechanism of collective behaviors of coupled units has become a focus for a variety of fields , such as biological neurons , circadian rhythm , chemically reacting cells , and even social systems @xcite . some properties of collective behaviors depend on the complexity of the system , while the other properties , such as phase transition , may be described by low - dimensional dynamics with macroscopic variables @xcite . discovering the method to simplify the system is just as important and as fascinating as the discovery of the complexity of it . like most cases in physics , simplification and low - dimensional reduction are associated with some symmetry of the system . as the identity of gas particles is the foundation of statistical mechanics and collective variables as temperature and pressure , the identity of the coupled units in a complex system is also related to some order parameters . in previous works , in the limit of large number of oscillators @xmath0 with special coupling function this work has been done by ott - antonsen(oa ) ansatz @xcite for oscillators with nonidentical parameters . as for the oscillators with identical parameters the same result was also got from group theory analysis in @xcite called watanabe - strogatz s approach . in this paper , we will show a different way in which the low - dimensional reduction is a natural consequence of the symmetry of the system by taking the order parameters as collective variables , namely the order parameter analysis . our approach is simple and concise for oscillators with both identical and nonidentical parameters . we can also get the scope of the oa ansatz . two more cases beyond the scope are further discussed with appropriate approximations . with our approach we show that the oa ansatz can be used beyond its scope with the approximations works . the model discussed in this paper is the all - connected coupled phase - oscillators . in the first section , the dynamical equations for order parameters are derived . the oa ansatz is got naturally from the symmetry of the order parameter equations , with its scope as the limit of infinitely many oscillators and the condition that only three fourier coefficients of the coupling function are nonzero . in the second and third sections , we will consider two approximate use of this ansatz beyond its scope , the case of a finite - size system and the case of coupled oscillator systems with more complicated coupling functions , with approximations respectively , i.e. , the ensemble approach and the dominating - term assumption . the famous kuramoto model for the process of synchronization attracts much attentions upon it is proposed and has been developed for decades . this model consists of a population of n coupled phase oscillators @xmath1 with natural frequencies @xmath2 , and the dynamics are governed by @xmath3 with the mean - field coupling @xmath4 and the definition of order parameter @xmath5 the equations eq . ( [ e1 ] ) can be rewritten as @xmath6 where @xmath7 is the imaginary unit , @xmath8 is the complex conjugate of @xmath9 . apart form parameters @xmath10 and @xmath11 , the dynamics of each phase variable @xmath12 depends only on itself and @xmath9 . it is the important character of this mean - field model , and the order parameter @xmath9 is always used to describe the state of the system , as the system is in synchronous states if and only if @xmath13 . a more general form of the mean - field model with phase oscillators can be written as @xmath14 where @xmath15 is any smooth , real , @xmath16-periodic function for @xmath12 . @xmath17 is the order parameter with the @xmath18th order parameter @xmath19 defined as @xmath20 @xmath21 is the identical parameter such as the coupling strength @xmath10 which is identical for all the oscillators . @xmath22 is the nonidentical parameter such as the natural - frequency @xmath2 which is nonidentical for the oscillators and usually has a distribution among the system . almost all the mean - field models based on the kuramoto model belong to this general category . in the following , we will build our approach for this general model to explore the conditions in which we can get the low - dimensional description of the system eq . ( [ ew1 ] ) . to begin with , let us consider the simple case of identical oscillators , e.g. , @xmath23 for @xmath24 in the kuramoto model , which reads @xmath25 in the limit of infinitely many oscillators @xmath0 , let @xmath26 denote the fraction of oscillators that lie between @xmath27 and @xmath28 . because each oscillator in eq.([e2 ] ) moves with the angular velocity @xmath29 , the single oscillator density @xmath30 obeys the continuity equation as @xmath31 if the phase density @xmath30 is known , all of macroscopic properties of the system can be got through some statistical average , such as order parameters @xmath19 as @xmath32 if we are only concerned with the collective or macroscopic state of the system , as in this paper , the macroscopic description of the phase density is equivalent to the microscopic description of phases of all the oscillators . moreover , for the coupling function @xmath33 is always @xmath16-periodic for @xmath27 , we have the fourier expansion of @xmath33 , @xmath34 with the the definition eq . ( [ e3 ] ) , the dynamics of order parameters can be got as @xmath35 where @xmath36 . substituting the expansion of @xmath33 into these equations , we have the closed equations for the dynamics of order parameters as @xmath37 on the other hand , from eq . ( [ e3 ] ) the order parameters are exactly fourier components of the phase density . if all the order parameters are known , we have @xmath38 hence if we know all the initial values of order parameters , with the dynamical equations eq . ( [ e4 ] ) and the fourier transformation eq . ( [ e5 ] ) , the system is identified , which is the same as the dynamical equations eq . ( [ e2 ] ) for phase variables @xmath1 and the continuity equation eq . ( [ ea1 ] ) for the phase density @xmath30 . therefore , we can choose either the order parameters , the phase variables or the phase density to perform the analysis of the collective behaviors of the coupled phase oscillators . these three descriptions of the system , i.e. phase variables , density of phase and order parameters , correspond to the dynamical , statistical and macroscopic descriptions of the system , respectively , and they are equivalent to each other in the limit of infinitely many oscillators @xmath0 . as a matter of fact , the ott - antonsen ansatz @xcite is based on the representation of the density of phase , and the watanabe - strogatz s approach @xcite is based on the scenario of the phase variables . in the following , we will take our approach on the base of order parameters . the complexity of dynamical equations for order parameters eq . ( [ e4 ] ) depends on the coupling function , or explicitly , on the fourier expansion of @xmath39 . first , let us consider the simplest case , with only the first three nonzero terms of the fourier expansion , @xmath40 because @xmath39 is a real function , we have @xmath41 where @xmath42 means the complex conjugate of @xmath43 . ( [ e6 ] ) , the dynamics of order parameters eq . ( [ e4 ] ) becomes @xmath44 with @xmath45 and @xmath46 . the recursion form of these equations shows that there is some structure for order parameters with which we can simplify the system and get some low - dimensional equations for the high - dimensional coupled oscillator dynamics of the system . in fact , by choosing an invariant manifold as @xmath47 , all the equations for @xmath19 is reduced to @xmath48 for @xmath36 . the infinitely many dynamical equations of order parameters is thus reduced to a single equation on this manifold and the corresponding phase density defined by eq . ( [ e5 ] ) is the so - called poisson kernel distribution as @xmath49 where @xmath50 . if we choose the initial state on this manifold , the state will never evolve out of it , which is called the invariant manifold of the dynamical system eq . ( [ e7 ] ) . this is exactly the low - dimensional behavior of the system we are looking for . with our approach , the manifold is derived naturally and concisely . we will call this manifold the poisson manifold in this paper as in @xcite where the dynamics of @xmath51 , as eq . ( [ ew6 ] ) , is got from the group theory analysis for josephson junction arrays . furthermore , the low - dimensional behavior is not confined to the system of identical oscillators . the order parameter analysis we used above can also be used in a more general case of nonidentical oscillators with nonidentical parameters . firstly , let us consider a system of nonidentical oscillators with a discrete distribution of the parameter @xmath52 . this distribution naturally separates the oscillators into @xmath53 groups , and each group has the same parameter @xmath54 . in this case , the phase variable for oscillators is denoted by @xmath55 which means the @xmath56th oscillator with the same parameter @xmath57 . for each group we can define a local order parameter as @xmath58 where @xmath59 is the number of oscillators in the group with @xmath57 . the order parameter of the system denoted by @xmath19 is defined by @xmath60 assume that the coupling function for each oscillator @xmath61 depends only on the order parameters whether for the whole system or the groups , then equations for phase oscillators are @xmath62 where @xmath63 . @xmath64 and @xmath17 are the local and global order parameters , respectively . as the coupling function @xmath65 is @xmath16-periodic for @xmath55 , by performing fourier expansion of @xmath39 , the equations for order parameters of each group read @xmath66 where @xmath63 , @xmath36 and @xmath67 . the equations eq . ( [ e8 ] ) are closed for local order parameters , and the local order parameters here can be regarded as the coordinates of the system which are equivalent to phase variables . each group is almost separated from others and the only relation for the groups is the dependence of the coupling function on order parameters of the system . in the simplest case when only the first three terms of fourier expansion are nonzero , we have @xmath68 with @xmath63 and @xmath36 . for each group , @xmath69 is obviously a solution for the equations which reduce all the equations for @xmath70 to the same one as @xmath71 where @xmath63 . the poisson manifold is an invariant manifold for each group , and the order parameters of the system read @xmath72 which is determined by @xmath73 . therefore , the fourier coefficients of the coupling function @xmath74 depends on @xmath73 only . the system described by the local order parameters is governed by a group of poisson manifolds with the low - dimensional dynamics eq . ( [ e9 ] ) . in the limit with @xmath75 and @xmath76 , the distribution of parameter @xmath77 as @xmath78 for @xmath79 has the continuous form denoted by the function @xmath80 and the local order parameters @xmath73 become the function of @xmath77 , i.e. , @xmath81 . replacing the summation by the integral over @xmath77 , we get the continuous form of the dynamical equations as @xmath82 where @xmath83 are functionals of @xmath81 as @xmath84 even though for each specific @xmath77 the approach has already reduced the dynamics of the oscillators to the dynamic of @xmath81 as eq . ( [ e10 ] ) , it is still hard to get any analytical results as eq . ( [ e10 ] ) depends on the integral eq . ( [ e11 ] ) which can not be expressed by functions of @xmath81 . on the other hand , for some specific choice of @xmath80 , the integral eq . ( [ e11 ] ) can be obtained analytically , in which case we can get the simpler form of eq . ( [ e10 ] ) . what we are looking for is a two - step reduction . the first step is finished by introducing the poisson manifold with which we have reduced the dynamic of phase oscillators to the dynamic of a group of poisson manifolds , i.e. , eq . ( [ e10 ] ) . the second step is rooted in the relation between the poisson manifolds in the group which will reduce the dynamics of the group further to the dynamic of a specific poisson manifold in this group . for instance , in the case that the function @xmath81 is analytical which can be extended to the complex plane , and the distribution of @xmath77 is the lorentzain distribution as @xmath85 the integral eq . ( [ e11 ] ) can be obtained by residue theorem as @xmath86 . setting @xmath87 , eq . ( [ e10 ] ) becomes @xmath88 which is closed for order parameter @xmath89 . then , the behavior of @xmath90 , together with the collective behaviors of the system , is determined by eq . ( [ e12 ] ) . another solvable example is the case when the distribution of natural frequencies of oscillators is the dirac s delta distribution function @xmath91 . the integral eq . ( [ e11 ] ) can be worked out as @xmath92 . setting @xmath93 in eq . ( [ e10 ] ) , the model is reduced to the network of identical oscillators which we discussed above . the approach shown above indicates there are two steps of the reduction scheme . the first step is the reduction from the equation eq . ( [ e8 ] ) to eq . ( [ e9 ] ) or the continuous form eq . ( [ e10 ] ) , which means the choice of the poisson manifold of the system . this is related to the the symmetry of the system or the recursion form of order parameters equations . the second step is the further operation from eq . ( [ e10 ] ) to eq . ( [ e12 ] ) , which depends on the special choice of distribution of nonidentical parameters . this reduction with the lorentzain distribution was first found in @xcite with an ansatz for low - dimensional manifold , namely the oa ansatz . for the case of the dirac s delta distribution function , this reduction was discussed comprehensively in @xcite with group theory analysis of the system . in our approach , the reduction comes from the same basis , i.e. , the symmetry of order parameter equations . along the approach , we can also consider more nonidentical parameters , such as the location of oscillators @xmath94 for example in @xcite , where the coupling function depends on the locations @xmath95 as @xmath96 with the distributions @xmath80 and @xmath97 . the order parameter @xmath98 of the system is defined as @xmath99 when the fourier expansion of the coupling function for @xmath27 contains only the first three terms , for each specific @xmath77 and @xmath94 , the relation @xmath100 is exactly the solution for dynamical equations , which reads @xmath101 if @xmath102 is analytic for @xmath77 and @xmath80 is the lorentzain distribution , the integral eq . ( [ e13 ] ) can be obtained for @xmath77 as @xmath103 where @xmath36 . setting @xmath104 and @xmath87 in the dynamical equation eq . ( [ e14 ] ) , we have @xmath105 where @xmath17 is the order parameter of the system and the functional of @xmath106 as @xmath107 the solution of this integral differential equation eq . ( [ e15 ] ) describes the structure of the local order parameter @xmath106 along @xmath94 . up to now , we have discussed the system of all - connected phase oscillators in the limit of infinitely many oscillators and the case that only three fourier coefficients of the coupling function are nonzero . the low - dimensional invariant manifold , namely the poisson manifold , is got for both cases . we will see in the next two sections that these two conditions are exactly the scope of the oa ansatz . two cases beyond this cope will be discussed respectively . we have considered the general model eq . ( [ e2 ] ) in the limit of infinitely many oscillators , @xmath0 , in which we could get the phase density of oscillators @xmath30 and the corresponding continuity equation eq . ( [ ea1 ] ) . order parameters @xmath19 could be considered as fourier coefficients of @xmath30 , from which we get the equivalent expression of the dynamics of the system as the order parameter equations eq . ( [ e4 ] ) and build the approach above . in the limit @xmath0 , the macroscopic variable , namely the order parameter @xmath108 , has the limit @xmath109 for steady states as the synchronous state and incoherent states , which gives us the basis to discuss the system analytically . in the case of a finite but large number of oscillators , i.e. , @xmath110 , the order parameter defined as @xmath111 will fluctuate around the value @xmath112 . this fluctuation depends on the number of oscillators as @xmath113 @xcite . when the fluctuation is small enough , as @xmath114 for @xmath115 , the collective behaviors of the system with finite number of oscillators could be described by the approximation of infinitely many oscillators , for which analytical methods can be applied . some analytical methods , e.g. , the self - consistent method and the oa ansatz , are applicable for the case of infinitely many oscillators or equivalently the phase density description . on the other hand , in the case of only a few number of oscillators , e.g. , @xmath116 , the difference @xmath117 would be so large , as @xmath118 , whose magnitude comparable with the value of @xmath9 . obviously , it is not reliable to treat the system with the approximation @xmath119 in this case . it is necessary to propose new approaches in analytically dealing with the collective behaviors of finite - oscillator systems . in our approach , order parameters @xmath19 could be considered as not only the fourier coefficients of @xmath30 but also the collective variables as @xmath120 which does not depend on the approximation of infinitely many oscillators . for the system as @xmath121 with the definition of order parameters ( [ ew7 ] ) , the dynamical equations for order parameters are @xmath122 for the coupling function @xmath33 is @xmath16-periodic for @xmath27 , we have the fourier expansion as @xmath34 substituting the expansion into eq . ( [ e16 ] ) , together with eq . ( [ ew7 ] ) , we have the closed form of equations for order parameters as @xmath123 which is the same as we get in the limit of infinitely many oscillators . in the case of a few number of oscillators , the definition of order parameters can also be considered as the coordinate transformation that transforms the microscopic variables @xmath1 to macroscopic variables @xmath124 . therefore it is inspiring that the dynamics of phase oscillators with corresponding initial conditions @xmath125 is equivalent to the dynamics of order parameters with corresponding initial conditions @xmath126 . the number of oscillators , no matter finite or infinite , has no influence on the dynamical equations of order parameters . this forms the important basis of our approach . following this approach , when only the first three fourier coefficients of the coupling function are nonzero , by setting @xmath127 and substituting the relation to the dynamical equations , the equations eq . ( [ e17 ] ) for all the @xmath19 will be reduced to a single one as @xmath128 it appears that we could get the poisson manifold again even the size of the system is finite . however , the finite - size effect should be taken into account with care . as a matter of fact , in the case of a finite number of oscillators , the poisson manifold with @xmath127 is not attainable for the system . to see this , take the system of @xmath129 as an example . for the first two order parameters @xmath51 and @xmath130 , by definition @xmath131 if the system described by @xmath132 is in the poisson manifold , the relation @xmath133 could be rewritten in terms of the definition eq . ( [ ew7 ] ) as @xmath134 however , the equality eq . ( [ e18 ] ) is not naturally valid . by using a simple calculation of the difference between the left - hand and the right - hand terms in eq . ( [ e18 ] ) one obtains @xmath135 this indicates that the system can evolve on the poisson manifold only when @xmath136 . in general , the relation @xmath127 gives exactly infinitely many independent constraints like eq . ( [ e18 ] ) with @xmath137 , and any solutions for finite oscillators will be determined as the trivial one as @xmath138 for all the index @xmath56 , which means that except the synchronous state , all the states of the system of finite oscillators lie out of this poisson manifold . the system can only evolve on the poisson manifold in the limit of infinitely many oscillators . whereas , as a matter of fact , the equations eqs . ( [ e17 ] ) hold for all @xmath139 , whether the size of system is finite or infinite , and the synchronous state always satisfies the relation of the manifold . the poisson manifold can be used at least approximately in the vicinity of synchronous state . however , we need some new methods and approximations . let us consider an ensemble with @xmath140 identical systems of @xmath139 oscillators which have the same dynamical equations and same parameters . the initial phases of oscillators are chosen from the same distribution , which makes the mean values of order parameters @xmath19 over the ensemble have the limit when @xmath141 . this leads to the definition of the ensemble order parameter @xmath142 as @xmath143 where @xmath140 is the number of sampling systems in the ensemble and @xmath144 is the @xmath18th order parameter for the @xmath56th system in the ensemble . taking the ensemble average for both sides of the dynamical equations eq . ( [ e17 ] ) , we get @xmath145 where @xmath36 , and @xmath146 means ensemble average . in general eq . ( [ e20 ] ) is not solvable because the terms in the right side as @xmath147 can not be simply described by @xmath142 in general . however , if @xmath148 is independent of all the systems in the ensemble as @xmath149 , then we can get exactly the dynamical equations of the ensemble order parameters , as @xmath150 where @xmath36 and the terms @xmath151 may depend on @xmath142 . similar to eq . ( [ e17 ] ) , there is a solution for the ensemble order parameter equations eqs . ( [ e21 ] ) as @xmath152 , which is exactly the poisson manifold . moreover , for more general cases @xmath153 , but when the terms @xmath154 can be approximated by @xmath155 , namely the statistical independence , we can also get the dynamical equations of the ensemble order parameters @xmath142 with similar approach . the validity of this approach can be measured by the error terms @xmath156 with @xmath157 and @xmath36 . in the following , let us take the star sakaguchi - kuramoto model as an example , which is a typical topology and model in grasping the essential properties of heterogeneous networks and synchronization process , as @xmath158 where @xmath159 , @xmath160 and @xmath161 are the natural frequency and the phase of hub and leaf nodes respectively , @xmath162 is the coupling strength and @xmath163 is the phase shift . by introducing the phase difference @xmath164 , the dynamical equation can be transformed into @xmath165 where @xmath166 is the natural - frequency difference between hub and leaf nodes . by introducing the order parameter @xmath167 , we could rewrite the dynamics as @xmath168 where @xmath169 , @xmath170 . hence the system is in the framework which we discussed above , with the first three nonzero fourier coefficients of the coupling function . with our approach , the dynamical equations for order parameters can be obtained as @xmath171 in this specific model , we can not simply make the approximation as @xmath0 because @xmath170 diverges with @xmath0 . there are two ways in further simplifying the system . first , we could set a hypothetical system as @xmath172 where @xmath173 , @xmath174 is the hypothetical phase oscillator and @xmath139 is considered as a parameter in this system . defining the order parameter for this system as @xmath175 , eq . ( [ e24 ] ) becomes @xmath176 where @xmath169 , @xmath170 . following our approach , the corresponding order parameter equations read @xmath177 which are exactly the same as eq . ( [ e23 ] ) . but for this hypothetical system , we could take the limit of infinitely many oscillators as @xmath141 , which could be discussed analytically with the traditional oa anstaz . obviously , these two systems are quite different for phase variables , one with finite oscillators the other with infinitely many oscillators , but they have the same order parameter equations . in the following we will see that the hypothetical system eq . ( [ e24 ] ) is exactly a representation of the ensemble of the original model eq . ( [ ea2 ] ) . on the other hand , following our ensemble approach , for the system eq . ( [ ea2 ] ) , let us choose an ensemble consisting of systems with the same parameters @xmath178 and @xmath139 , and different initial conditions chosen from the same distribution , as the poisson kernel . we have the ensemble average of the dynamical equation as @xmath179 where @xmath36 . setting @xmath180 for @xmath36 . if @xmath181 and @xmath182 , or typically @xmath183 and @xmath184 , they can be seen as the perturbation terms for the dynamics of ensemble order parameters in eq . ( [ e26 ] ) . ignoring the perturbations , we could get @xmath185 and this is the same as eq . ( [ e25 ] ) for hypothetical system . in this case , the ensemble order parameters are the same as the order parameters for the hypothetical system . for the case of a finite but large number of oscillators , the hypothetical system is exactly the continuous model which shares the same order parameter equations but with infinitely many oscillators . as in the case of a few oscillators , the hypothetical system is a presentation of the ensemble for the systems , where the difference terms @xmath186 and @xmath187 by this method will not only introduce some fluctuations but also some systematic errors , which can be described by @xmath188 with @xmath36 for this specific model . in terms of numerical computation , we can check the above approximation by examining @xmath189 and @xmath190 for the ensemble of @xmath140 systems . the approximation @xmath189 and @xmath190 can be divided into two parts as @xmath191 and @xmath192 where @xmath193 and @xmath194 are the fluctuation parts proportional to @xmath195 which depend on the size of the ensemble and @xmath196 and @xmath197 are the systematic parts which are introduced by the statistical independence assumption . take @xmath116 as an example . when @xmath198 the fluctuation parts are ignorable . we find that except for @xmath199 is around @xmath200 , all of @xmath189 and @xmath190 for @xmath201 are much smaller than @xmath202 in this case as plotted in fig . [ fig:2](a ) and ( b ) . comparing the typical magnitude of values of order parameters as @xmath203 , the approximation of the ensemble approach is obviously reasonable . for the initial conditions chosen from the poisson kernel , with the dynamical equations eq . ( [ e27 ] ) , the ensemble order parameters will evolve on an invariant manifold , i.e. , the poisson manifold with @xmath152 . denote @xmath204 by @xmath205 , we have @xmath206 the difference between @xmath205 and the ensemble average of numerical simulations is checked in fig . [ fig:2](d ) , which shows that it is reasonable to describe the behavior of the ensemble by @xmath205 . moreover , for every single system in the ensemble , with some fluctuation , it can also be described by the ensemble order parameter and hence by @xmath205 , as shown in fig . [ fig:2](c ) . and @xmath190 versus time for the system with @xmath116 . ( c ) and ( d ) order parameters getting from numerical simulation and approximated oa ansatz . results from approximated oa ansatz is the red line , and every single simulation results is the blue lines in ( c ) with ensemble average the light blue line in ( d).,height=264 ] with this ensemble assumption works , the two - dimensional equation eq . ( [ ea5 ] ) in the bounded space as @xmath207 describes the dynamics of the coupled oscillator system . every stationary state for phase oscillators system has its counterpart in this space for @xmath208 . for example , the in - phase state(ips ) defined as @xmath209 corresponds to a limit cycle as @xmath210 . the synchronous state ( ss ) defined as @xmath211 corresponds to a fixed point with @xmath210 and @xmath212 . the splay state(sps ) defined as @xmath213 with @xmath214 the period of @xmath215 corresponds to a fixed point with @xmath216 and @xmath212 . in the two - dimensional order - parameter space , it is easy to analyze the existence and stability conditions of these states , and the basion of attraction for each state in the coexistence region can also be conveniently determined @xcite . this is the reason why we introduce the invariant manifold governed by low - dimensional dynamics . as we see , in the case with only finite oscillators , the ensemble order parameter plays the dominant role in revealing the key collective structure of the system . the price of the ensemble order parameter description is the error induced by the statistical independence assumption , whose validity can be checked by numerical results . in the next section we will see that when more than three fourier coefficients of the coupling function are nonzero , where the traditional order parameter scheme fails , one can still discuss the collective dynamics in terms of the ensemble order parameter approach . in the above sections , for the coupling functions with only first three terms as @xmath217 we have the dynamical equations for order parameters @xmath19 as @xmath218 on the invariant manifold @xmath127 , all these equations are reduced to a single one @xmath219 for general situations , the coupling function can not be simply truncated to only the first terms . it is thus important to consider the order parameter approach for more complicated cases . firstly , let us consider a coupling function with higher order fourier coefficients like @xmath220 where @xmath221 is an integer and @xmath222 are the order parameters for the system . note that eq . ( [ e29 ] ) can be regarded as the transformation of eq . ( [ ea3 ] ) with @xmath223 and @xmath224 . the order parameters @xmath98 for @xmath27 is related to the order parameters @xmath222 for @xmath225 as @xmath226 and the phase density of @xmath225 with the transformation reads @xmath227 this indicates that all the terms @xmath228 with @xmath229 are zero . moreover , we can transform the poisson manifold @xmath230 for @xmath27 to the invariant manifold @xmath231 for @xmath225 , which is governed by @xmath232 corresponding to eq . ( [ ea4 ] ) for @xmath51 . this is exactly the low - dimensional manifold for the coupling function eq . ( [ e29 ] ) with higher order fourier coefficients . this manifold is based on the @xmath233th order parameter @xmath234 and the statement that all the terms @xmath228 with @xmath229 are zero . this is a bit queer in this respect . let us take the case @xmath235 as an example @xcite . in this case only @xmath236 and @xmath237 among the fourier components of the coupling function are nonzero , i.e. , @xmath238 according to the above analysis , the system has the invariant manifold as @xmath239 , or equivalently @xmath240 $ ] , with dynamics @xmath241 the phase density of oscillators in this case is @xmath242 by using our order - parameter - analysis approach , for the case of @xmath235 , it is not hard to get the the equations of order parameters @xmath19 as @xmath243 with @xmath36 . the manifold with @xmath244 is obviously the solution of these equations . this invariant manifold separates the relation @xmath127 into two parts , where the part @xmath245 conserves the relation , and the odd part @xmath246 can be regarded as a special choice of the relation . even if in the state defined as @xmath247 , we still have @xmath248 , which represents the state of cluster synchrony as discussed in @xcite . and the manifold with @xmath244 can be regarded as a sub - manifold of the poisson manifold . an interesting issue here is the reason why we set all the @xmath246 other than @xmath249 . to see this , we just substitute the relation @xmath250 into the odd part of eq . ( [ e31 ] ) , then the equations will be reduced to the following equations , @xmath251 where the first equation is got from @xmath252 in eq . ( [ e31 ] ) , and the second equation is got from all the others @xmath253 in eq . ( [ e31 ] ) . comparing the two equations in eq . ( [ ew4 ] ) , the consistent condition requires @xmath254 , and this implies that @xmath255 . hence , there is nt a general relation of @xmath256 , apart from the vicinity of synchronous state . thus we should consider another special solution for the odd part of eq . ( [ e31 ] ) as @xmath257 . on the other hand , for the even part of eq . ( [ e31 ] ) , by setting @xmath245 , the equations will be reduced to a single one as @xmath258 the conjugate part @xmath259 is separated from @xmath130 naturally in this equation , and we can get the low - dimensional manifold governed by the order parameter equation eq . ( [ ew5 ] ) , as a sub - manifold of the poisson manifold . as mentioned above , the poisson manifold no longer exists when the order parameter relation @xmath127 can not be used to reduce the order parameter equations to the single one . for the case of only higher terms this problem can be fixed by separating the order parameters into groups and setting some of them zero with which a special sub - manifold of the poisson manifold can be used to get low - dimensional behaviors . however , when more than three terms in fourier expansion are considered , even the sub - manifold on longer exists . consider a coupling function with the first five nonzero fourier coefficients , as @xmath260 the corresponding equations of order parameters read now @xmath261 with @xmath46 , @xmath36 . in this case , if we suppose @xmath127 , then two different dynamical equations will be obtained @xmath262 hence we will have @xmath263 as the requirement of coincidence of these two equations . the relation of @xmath264 is broken by the conjugate terms and there is no way in finding out a separated solution as we did for eq . ( [ e31 ] ) . the only solutions with the relation @xmath127 for this system seems to be either @xmath265 or @xmath266 . none of them are expected for our analysis , unless in the vicinity of synchronous state with @xmath255 or in the vicinity of the incoherent state with @xmath248 . on the other hand , for @xmath267 all the dynamical equations for the @xmath18th order parameter in eq . ( [ e32 ] ) are reduced to the same equation by @xmath127 , and the symmetry of eq . ( [ e32 ] ) can be represented by @xmath127 at least for all the order parameter equations for @xmath267 . moreover , due to the bound @xmath268 , the higher the order components @xmath19 become smaller with increasing @xmath18 . if we choose the initial conditions as @xmath127 , then along the evolution of eq . ( [ e32 ] ) , the differences of higher terms @xmath19 with @xmath269 will keep small enough . by ignoring the differences and making the approximation that @xmath270 , we can obtain the dynamical equations for @xmath51 as @xmath271 which can be regarded as the main term of the dynamics eq . ( [ e32 ] ) . from now on we call this approach _ dominating - term assumption_. we will further show that this approximation is pretty efficient in dealing with systems with complicated coupling functions . let us take the system described by eq . ( [ ea2 ] ) as an example . by considering higher terms , we have the dynamical equations @xmath272 where @xmath273 and @xmath274 , @xmath275 , @xmath276 with @xmath277 . by introducing the dominating - term assumption , the dynamical equation for @xmath51 reads @xmath278 if the dominating - term assumption works , eq . ( [ e33 ] ) can be used to study the collective behaviors of the coupled oscillators , which can be regarded as a sort of approximate low - dimensional behaviors . the validity of the dominating - term assumption can be checked via numerical simulations . for the initial state chosen from the poisson kernel distribution , by using numerical simulation , we have the order parameter from the continuity equation of the phase density and the approximate manifold . in fig . [ fig:5](a ) the order parameter @xmath51 is plotted . it can be found that the result in terms of the dominating - term approximation coincides very well with numerical results . in fig . [ fig:5](b ) , the density of oscillators at a given time is also plotted , and the result shows that though the distribution is indeed no longer the poisson kernel distribution , it is instead close to it . this gives us the mechanism of how this assumption works . ) ( blue line ) . ( b ) distribution of phase variables getting form numerical simulation ( red ) and reconstituting poisson kernel with order parameter @xmath51 ( blue ) . the system considered as @xmath279,height=132 ] according to numerical results , we can see that the dominating - term assumption works pretty well , and the system shows approximated low - dimensional behaviors . with the dynamical equations for the first order parameter eq . ( [ e33 ] ) , we can do the study further to find the property of collective behaviors of the system , which will be discussed in detail in another paper . in this paper , we focused on the theoretical description of low - dimensional order - parameter dynamics of the collective behaviors in coupled phase oscillators . we have derived the closed form of dynamical equations for order parameters , from which we find the poisson kernel as an invariant manifold and the well - known oa ansatz . different from the traditional ott - antonsen ansatz and watanabe - strogatz s approach , our approach is suitable for systems of both finite and infinite , identical and nonidentical oscillators . in our approach , the scope of the oa ansatz is determined as two parts , i.e. , the limit of infinitely many oscillators and the condition that only the first three fourier coefficients of the coupling strength are nonzero . by using our order parameter analysis , we also discussed two cases that go beyond the scope of the oa ansatz , i.e. , the case of a finite - size system and the case of coupled oscillator systems with more complicated coupling functions . we have discussed the reasons why the oa ansatz can not be used directly to these two cases and we further developed the approximation methods to deal with these difficulties with the oa ansatz . we developed two methods , namely the ensemble method and dominating - term assumption . it is shown that these schemes work pretty well , and their validity has been checked by the numerical simulations . for the case of a finite number of coupled oscillators , it is shown that the system is out of the domain of low - dimensional manifold of the coupled order parameter equations . hence , from the view of the geometry theory of manifolds , the finite size of system brings some fluctuations along the manifold which is introduced from initial states . if the dynamics of order parameter equations is not too complicated and the manifold satisfying @xmath127 can be used to describe the dynamics approximately , all the trajectories from different initial conditions around the poisson manifold will be described by the mechanism of this manifold approximately . this is the meaning of the ensemble method and the reason for why it works in some models . in other case , with a more complicated coupling function , the poisson manifold is not a solution of the dynamical system . however , for the infinitely many coupled equations for @xmath19 , except for few of them , the relation @xmath127 indeed reflects the symmetry of these order parameter equations . when we choose the initial states in this manifold , the evolution of the dynamics will certainly be influenced by the symmetry of it , which is described by @xmath127 approximately . the method with dominating - term assumption is indeed designed to show this effect , and consequently the system is simplified by it for some models . in this paper , we have developed the order parameter analysis , with which we get the low - dimensional behaviors of coupled phase oscillators , such as the oa ansatz , the ensemble order parameter approach and the dominating - term approximation . however , to fully understand the low - dimensional collective behaviors we need a more comprehensive understanding of the poisson manifold and its relation with the dynamics of the system , which is still an open question . we believe that the order parameter analysis should be a powerful tool in helping to reveal the mechanism of low - dimensional collective behaviors . this work is partially supported by the national natural science foundation of china ( grant no . 11075016 and 11475022 ) . kuramoto , y. _ chemical oscillations , waves and turbulence . _ . 7576 ( springer , berlin , 1984 ) . acebron , j. a. , bonilla , l. l. , vicente , c. j. p. , ritort , f. & spigler , r. the kuramoto model : a simple paradigm for synchronization phenomena . phys . _ * 77 * , 137185 ( 2005 ) . strogatz , s. h. from kuramoto to crawford : exploring the onset of synchronization in populations of coupled oscillators . _ physica d _ * 143 * , 120 ( 2000 ) . pikovsky , a. , rosenblum , m. & kurths , j. _ synchronization : a universal concept in nonlinear sciences . _ pp . 279296 ( cambridge university press , cambridge , england , 2001 ) . dorogovtsev , s. n. , goltsev , a. v. & mendes , j. f. f. critical phenomena in complex networks . phys . _ * 80 * , 1275 ( 2008 ) . arenas , a. , diaz - guilera , a. , kurths , j. , moreno , y. & zhou . c. synchronization in complex networks . rep . _ * 469 * , 93153 ( 2008 ) . zheng , z. , hu , g. & hu , b. phase slips and phase synchronization of coupled oscillators . _ phys . lett . _ * 81 * , 53185321 ( 1998 ) . watanabe , s. & strogatz , s. h. constants of motion for superconducting josephson arrays . _ physica d _ * 74 * , 197253 ( 1994 ) . ott , e. & antonsen , t. m. low dimensional behavior of large systems of globally coupled oscillators . _ chaos _ * 18 * , 037113 ( 2008 ) . marvel , s. a. , mirollo , r. e. & strogatz , s. h. identical phase oscillators with global sinusoidal coupling evolve by mbius group action . _ chaos _ * 19 * , 043104 ( 2009 ) . marvel , s. a. & strogatz , s. h. invariant submanifold for series arrays of josephson junctions . _ chaos _ * 19 * , 013132 ( 2009 ) . omelchenko , o. e. & wolfrum , m. nonuniversal transitions to synchrony in the sakaguchi - kuramoto model . lett . _ * 109 * , 164101 ( 2012 ) . watanabe , s. & strogatz , s. h. integrability of a globally coupled oscillator array . * 70 * , 2391 ( 1993 ) . laing , c. r. the dynamics of chimera states in heterogeneous kuramoto networks . _ physica d _ * 238 * , 15691588 ( 2009 ) . daido , h. scaling behaviour at the onset of mutual entrainment in a population of interacting oscillators . _ j. phys . a : math . gen . _ * 20 * l629l636 ( 1986 ) . daido , h. intrinsic fluctuation and its critical scaling in a class of populations of oscillators with distributed frequencies . * 81 * 727731 ( 1989 ) . daido , h. intrinsic fluctuations and a phase transition in a class of large populations of interacting oscillators . _ journal of statistical physics _ * 60 * 753800 ( 1990 ) . skardal , p. s. , ott , e. , restrepo , j. g. cluster synchrony in systems of coupled phase oscillators with higher - order coupling . * 84 * , 036208 ( 2011 ) . xu , c. , gao , j. , sun , y. , huang , x. , & zheng , z. , explosive or continuous : incoherent state determines the route to synchronization . _ scientific report _ * 5 * , 12039 ( 2015 ) .
coupled phase - oscillators are important models related to synchronization . recently , ott - antonsen(oa ) ansatz is developed and used to get low - dimensional collective behaviors in coupled oscillator systems . in this paper , we develop a simple and concise approach based on the equations of order parameters , namely , order parameter analysis , with which we point out that the oa ansatz is rooted in the dynamical symmetry of the order parameters . with our approach the scope of the oa ansatz is identified as two conditions , i.e. , infinite size of the system and only three nonzero fourier coefficients of the coupling function . coinciding with each of the conditions , a distinctive system out of the scope is taken into account and discussed with the order parameter analysis . two approximation methods are introduced respectively , namely the ensemble approach and the dominating - term assumption .
is a high - energy transient in a low - mass x - ray binary ( lmxb ) system , and only the second source , besides the rapid burster @xcite , observed to emit multiple type ii x - ray bursts , i.e. , due to spasmodic accretion rather than thermonuclear burning . the source was discovered in 1996 with the burst and transient source experiment ( batse ) on board the _ compton gamma ray observatory ( cgro ) _ , when it emitted a series of hard x - ray bursts during a period lasting @xmath6 days @xcite . soon after its discovery , @xcite reported that the timing properties of the persistent x - ray emission pointed towards a magnetized neutron star pulsating at 2.14 hz , accreting material from a low - mass companion in a nearly circular orbit with an orbital period of 11.8 days . at that time was the first source to show bursts and pulsations , hence the source was nicknamed `` the bursting pulsar '' ( hereafter bp ) . the bp emerged from quiescence again almost exactly one year after its first outburst , in december 1996 @xcite . this second outburst was very similar to the first including both burst and persistent x - ray emission characteristics @xcite . the bp and its two outbursts were studied extensively during the first few years after its discovery . the x - ray bursts from the source were classified as type ii bursts , based on their spectra , energetics @xcite , and their resemblance to the bursts observed from the rapid burster @xcite . type ii bursts are most likely the result of some sort of instability ( whose origin is still unknown ) in the accretion disk resulting in the onset of mass inflow onto the neutron star which is responsible for the bursting activity . the bp average burst duration was 10 s and each burst was followed by a dip in flux below that of the pre - burst persistent emission . the flux recovered exponentially back to the pre - burst persistent emission level on time - scales of a few hundred seconds @xcite . pulsations at the spin frequency of the source were also detected during bursts , albeit with an average time lag of about 50 ms compared to the pre - burst pulses @xcite . the pre - burst pulse profile was subsequently recovered on timescales of a few hundred seconds @xcite . @xcite attributed these lags to the accretion column geometry at the pole . there is no direct estimate as yet of the magnetic field of the bp . * see also @xcite ) placed an upper limit on the dipole magnetic field of @xmath7 g based on the spin - up rate of the source and the persistent pulsed luminosity . @xcite deduced from binary evolution calculations that the dipole magnetic field of lies in the range of @xmath8 g , with a most probable value of @xmath9 g. finally , @xcite derived a surface magnetic field of @xmath10 g , assuming that the propeller effect is the reason for the non - detection of x - ray pulsations when the source persistent flux dropped below a certain level . it is , therefore , likely that the surface magnetic field of the bp lies between classical x - ray accreting pulsars ( @xmath11 g ) and lmxbs ( @xmath12 g ) . this intermediate strength surface field could be an important parameter defining the unusual properties of this source ; hence , the determination of its exact value is of crucial importance . @xcite studied the @xmath13kev spectrum of during outbursts using _ asca_. they found a spectrum well described by an absorbed power - law ( pl ) and line - like emission between 6 and 7 kev , most likely from fe reprocessed in the accretion disk . the heavy absorption towards the source ( @xmath14 @xmath15 ) places the bp at the galactic center , likely at 8 kpc . the quiescent x - ray counterpart of the bp was discovered with _ chandra _ @xcite and was confirmed one month later with _ xmm - newton _ the spectrum in quiescence is soft and could be fit with either a pl model with @xmath16 or a blackbody with @xmath17 kev , implying a quiescent @xmath13 kev x - ray luminosity of @xmath18 erg s@xmath19 at 8 kpc . using the _ chandra _ position , ( * ? ? ? * see also @xcite ) found two potential infrared counterparts within the bp error circle , with the most likely candidate being a giant star of type g4 iii . on 2014 january 18 , the _ monitor of all - sky x - ray image ( maxi _ ) gas slit camera detected enhanced hard x - ray emission from the galactic center region @xcite . following the detection , they examined archival data from the _ swift _ burst alert telescope ( bat ) , and found that the x - ray emission from the bp had increased compared to its quiescent level . soon after , the source triggered bat on 2014 january 18 @xcite . @xcite detected pulsations from the direction of at the 2.14 hz spin period of the source during january @xmath20 using the _ fermi _ gamma - ray burst monitor . finally , the _ swift _ x - ray telescope ( xrt ) observed the bp on 2014 february 2 @xcite , detecting a bright source at the _ chandra _ location , confirming that the source entered a new outburst after about 18 years of quiescence ( see also * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? @xcite discovered infrared brightening of the g4 iii candidate counterpart contemporary with the x - ray outburst , confirming its identification as the bp companion . here we report our results of the analysis of a 10 ks simultaneous _ chandra _ and _ nustar _ observation of the bp taken on 2014 march 3 . section [ obs ] describes the observations and data reduction techniques . our results are presented in section [ res ] , and discussed in section [ discuss ] . we observed the bp with _ chandra _ using the high energy transmission grating ( hetg ) in continuous clocking mode ( cc - mode ) with all six ccds of the acis - s array . the hetg comprises two sets of gratings , the medium energy grating ( meg ) , operating in the energy range of @xmath21 kev , and the high energy grating ( heg ) with energy coverage in the range of @xmath22 kev , and a spectral resolution ( fwhm ) @xmath23 ev . each grating spectrum is dispersed along the acis - s ccds into positive and negative spectral orders . in addition , each grating observation results in an on - axis undispersed image with the ccd spectral resolution . we used the cc - mode to obtain the highest possible temporal resolution of 2.85 ms , at the expense of obtaining a one dimensional image of the source . the observation took place on 2014 march 3 , 08:59:06 utc , with 10 ks of good time intervals ( gti ) . a comparison between the zeroth order and the dispersed heg @xmath24 first order light curve reveals that the bursts are completely missing from the zeroth order light curve due to heavy pile - up . the source persistent emission also suffered a @xmath25 pile - up effect in the zeroth order . on the other hand , the dispersed grating spectra have a much lower total count rate compared to that of the zeroth order spectra . this results in spectra free of pile - up except during the peak of the bursts when pile - up still occurred at the @xmath25 level . to use the grating arms in our timing analysis , the photons assigned times needed to be corrected for their diffraction angle , which is directly proportional to the grating time offset with respect to the zeroth order . this time offset , @xmath26 , relative to the zeroth order location is @xmath27\ ] ] where @xmath28 is the diffraction angle of each photon @xmath29 , @xmath30 is the rowland spacing , @xmath31 is the grating clocking angle , @xmath32 is the pixel size , and @xmath33 is the read time per row ( 2.85 ms ) . the use of cc - mode with any _ chandra _ grating observation introduces some complications to the data reduction and analysis . for instance , due to the fact that there is no spatial information ( the clocking rows are all collapsed into one pixel ) , the soft x - ray background , usually low in te mode , is enhanced by 3 orders of magnitude . and @xmath0 arms of both the heg and meg 1st order spectra . the wiggle seen in the heg p1 spectrum between 2 and 6 kev is not detected in any of the other three spectra . we conclude that this shape is artificial and refrain from using the heg p1 in the non - burst spectral analyses.,title="fig:",scaledwidth=45.0% ] + a more pressing issue , that can alter grating dispersed spectra in cc - mode , is the dust scattering halo , usually present around bright absorbed sources . the bp resides in the galactic center region , hence , it is heavily absorbed , with a hydrogen column density @xmath14 @xmath15 . the brightness of the source produces a diffuse scattering halo , the emission from which disperses and blends with the source dispersed spectrum . the significance of this effect depends on the incident source spectrum , with hard sources affected less than soft ones . luckily , the bp has a hard x - ray spectrum ( section [ specana ] ) which reduces the impact of this background on the source spectrum . since no spatial information exists when using cc - mode , we extract the meg and heg backgrounds using the order sorting plots , which display the energies of the dispersed events versus the ratios of these energies over the event positions on the grating arm@xmath34 . on - axis point - source photons should distribute tightly and symmetrically around the extraction order , while diffuse photons have a larger scatter . finally , the extracted background is normalized to the excluded source region . these backgrounds are used for both timing and spectral analyses . + and @xmath35 kev energy ranges , respectively , plotted with 0.5 s time resolution.,scaledwidth=45.0% ] in addition to the background complications when dealing with cc - mode observations , there are calibration uncertainties@xmath34 between the different orders , e.g. , complicated charge transfer inefficiency ( cti ) corrections on the events . to check for potential differences , all _ chandra _ analyses were initially performed on the separate heg and meg arms . temporal analysis returned consistent results between all the different arms . spectral analyses , on the other hand , showed that the heg p1 ( hereafter , p1 refers to the positive and m1 to the negative first order grating arms ) spectrum is markedly different from the rest , i.e. , meg p1 , meg m1 , and heg m1 . figure [ calibration ] shows a pl fit to the different spectral arms , where a wiggle between 2 and 6 kev is present only in the heg p1 spectrum . we conclude that this feature is not real , and most likely due to either miscalibration , an improper modeling of the background , or both . hence , the heg p1 arm is excluded from the spectral analyses , except for bursts , where the above feature is not present ( likely due to the small integration times during bursts and/or the fact that the emission during burst intervals includes minimal background ) . all narrow features in the _ chandra _ spectra are seen above 6 kev . due to the lower spectral resolution and collecting area of the megs at energies @xmath36 kev , we also exclude these spectra in the analysis . for our timing analyses , we use the heg first order gratings ( positive and negative arms combined , i.e. , heg 1 ) . the _ nuclear spectroscopic telescope array _ ( _ nustar _ ) is a nasa small explorer ( smex ) satellite launched on 2012 june 13 @xcite . it is the first orbiting focusing hard x - ray telescope , observing the sky in an energy range from 3 to 79 kev with two co - aligned x - ray optics which focus x - rays onto two independent detector planes ( fpma and fpmb ) , each composed of four cdznte detectors . the field of view of _ nustar _ is roughly 12x 12 with a point - spread function with a fwhm of 18and a half - power diameter of 58 . _ _ obtained simultaneous observations of the bp during the _ chandra _ observation . the broader _ nustar _ bp data set is reserved for future work ; here we concentrate on the data obtained simultaneously with _ chandra_. we reduced the _ nustar _ data using nustardas v 1.3.1 and the _ nustar _ caldb 20131210 , with the standard pipeline filtering . we extracted the source photons from a circular region with a radius of 60 ; these regions were centroided separately for fpma and fpmb to account for the small misalignments in the absolute aspect reconstruction for the two telescopes . _ produces event files ( e.g. , each row in the event file represents a single time - tagged photon ) , which we can then filter based on the source region described above to produce source " event files . we produced response files ( arfs and rmfs ) for the _ nustar _ spectral analysis using the custom time - intervals defined in section [ timeana ] . these response files capture the response of the instrument over the specified time ranges . we show in figure [ channuslc ] the @xmath37 kev _ chandra _ heg 1st order light curve ( top panel ) of the entire 10 ks observation , with 10 s time bins . seven bursts are detected from the source during this observation . following each burst , a dip in the count rate is observed , which recovers exponentially back to the persistent level . the fpma light curve of the simultaneous _ nustar _ observation is shown in the second panel in figure [ channuslc ] , in the energy range of @xmath35 kev and also at 10 s resolution . only three bursts are detected simultaneously by both _ chandra _ and _ nustar_. figure [ channuslcbursts ] is a zoom in on these three bursts plotted with a 0.5 s resolution . kev light curves of bp bursts with 0.5 s resolution ( left panels ) , along with their bayesian blocks representation ( right panels ) . times are from the start of the _ chandra _ observation.,title="fig:",scaledwidth=43.0% ] + kev light curves of bp bursts with 0.5 s resolution ( left panels ) , along with their bayesian blocks representation ( right panels ) . times are from the start of the _ chandra _ observation.,title="fig:",scaledwidth=43.0% ] we use a bayesian blocks algorithm @xcite to identify the beginning and end times of the 7 bursts detected with _ chandra _ , and to search for weaker bursts in the _ chandra _ light curve binned with 0.5 s resolution . we chose this temporal resolution as a trade - off between speed and accuracy , considering that the source has a comparable pulse period . bayesian blocks have been frequently used for the temporal analysis of gamma - ray bursts , magnetars , and even flares from sgr a@xmath38 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? we find that only the 7 bursts clearly visible in figure [ channuslc ] show significant deviations ( @xmath39 ) from the persistent level of the source . we do not find any weaker ( mini@xmath0 ) bursts similar to the ones seen during the first two outbursts , e.g. , @xcite . the start and end times of the 7 _ chandra _ bursts are recorded from the bayesian blocks analysis ( figure [ burstsbayes ] ) . table [ burstprop ] shows these start times in mjd , and durations in seconds . the uncertainty on these durations is dominated by the temporal resolution we used , and it is @xmath40 s ( @xmath41 s error at the beginning and end of each burst ) . all durations , except the last , are narrowly distributed with a mean and 1@xmath42 standard deviation of @xmath43 s. the last burst consists of two pulses with a total duration of @xmath44 s ( figure [ burstsbayes ] ) . the rise and decay times of all bursts have a mean and 1@xmath42 standard deviation of @xmath45 s , and @xmath46 s , respectively ( excluding the decay time of the last burst ) . kev light curve with 10 s resolution . the black solid line is the mean rate of a pre - defined persistent emission interval far away from bursts and dips . the grey areas are the persistent emission intervals used in our spectral analysis . the red solid curves are exponential fits to the dip intervals . _ bottom panel . _ hardness ratio , @xmath47 , evolution during the entire _ chandra _ observation ( @xmath48 is computed in the @xmath49 and @xmath50 kev range , and @xmath51 in the @xmath52 kev range).,title="fig:",scaledwidth=45.0% ] + we identify the dip durations using the following method on the 2@xmath010 kev _ chandra _ light curve binned at 10 s. we search for the time - bin with the minimum count rate immediately after the end of a burst and up to the start of the following burst ( the end of the observation after the last burst ) . we then fit the light curve of each of these time intervals with an exponential function of the form , @xmath53 absorbed pl with @xmath54 @xmath15 , and @xmath55 . @xmath56 assuming a distance of 8 kpc . @xmath57 combined _ chandra _ and _ nustar _ data fit with an absorbed _ cutoffpl _ with @xmath58 @xmath15 , @xmath59 , and @xmath60 kev . @xmath61 derived in the energy range @xmath13 kev . @xmath53 dip times at minimum count rate after burst . @xmath56 dip characteristic time - scale for recovery . @xmath57 duration of interval from end of burst to @xmath62 . @xmath61 calculated in an 80 s interval centered on @xmath62 . @xmath63 calculated at @xmath62 over 10 s by converting count rates to fluxes with pimms . errors reflect the count rate errors only . @xmath64 fluence deficiency in the dip . @xmath65 combined _ chandra _ and _ nustar _ data . @xmath66 where @xmath67 is the minimum count rate at time @xmath62 , @xmath68 is the persistent count rate level , and @xmath69 is the characteristic time - scale representing 63.2% recovery of the count rate to @xmath68 . we first performed fits keeping both @xmath68 and @xmath69 as free parameters . we find that @xmath68 is similar in all cases with good enough coverage after the recovery , hence we keep @xmath68 constant at the mean count rate value calculated from time intervals far away from bursts and dips ( black solid line in figure [ fitdips ] ) . figure [ fitdips ] shows in red our exponential fits to the dips , and table [ dipprop ] lists the dip temporal properties . we find that @xmath69 ranges between @xmath70 and @xmath71 s , with a mean and 1@xmath42 standard deviation of @xmath72 s , whereas the average time from the end of a burst to the minimum count rate of the dip , @xmath73 , is @xmath74 s. we note that no bursts are seen during the dipping intervals , i.e. , all bursts are emitted after the dip recovered to at least the @xmath75 level of the persistent emission . the grey areas in figure [ fitdips ] represent the persistent emission time intervals excluding bursts and dip intervals . finally , we searched for any strong spectral variations in the _ chandra _ observation , especially during bursts , by looking at the evolution of the source flux hardness ratio , @xmath47 , where @xmath48 includes the energy ranges @xmath49 and @xmath50 kev ( to avoid contamination from the fe line complex , see section [ specana ] ) , and @xmath51 includes the @xmath52 kev range . the bottom panel of figure [ fitdips ] shows the @xmath47 derived from light curves with a 0.5 s resolution during bursts and with 10 s bins elsewhere . we do not find any spectral variations during bursts in the _ chandra _ observation compared to the non - burst emission , at the above temporal resolution . for each of the intervals defined in the previous section , we first apply a barycenter correction for the _ chandra _ and _ nustar_time - tagged events ( see also section [ chanobs ] for a description of the _ chandra _ grating time correction ) . we then correct these times for the binary motion of the system , using the orbital parameters provided by the gbm pulsar team . we estimate the spin frequency of the pulsar from the persistent data by locating the peak rayleigh power , @xmath76 , in a frequency range expected to contain the spin frequency . the rayleigh power is given by @xmath77 @xcite , where @xmath78 is the number of events , @xmath79 a trial frequency , and @xmath80 a barycenter and binary corrected event time . the peak power of 891 occurs at frequency 2.1411203(16 ) hz . the one sigma error is determined by the change in frequency required for the rayleigh power to drop by 0.5 . finally , we epoch - fold the data at the spin frequency derived above to compute a pulse profile ( pp ) . for the persistent and dip intervals , we extract _ chandra _ and _ nustar _ pulse profiles in different energy bands , chosen to have comparable number of events ( figure [ pp_persisnu ] ) . we fit the different pps with a sine plus cosine function of the form @xcite , @xmath81\ ] ] where @xmath82 is the count rate at phase bin @xmath83 , @xmath84 is the average count rate throughout the pp , and @xmath85 , @xmath86 are the coefficients of the different harmonics @xmath87 of the sine and cosine functions . the pp is nearly sinusoidal . the second harmonic contribution is highest in between 8 and 10 kev ( @xmath88 ) for _ chandra _ , and between 6 and 8 kev ( @xmath89 ) for _ nustar_. the rms pulsed fraction ( pf ) is defined as @xcite , @xmath90^{0.5}/c_{\rm mean}\ ] ] where @xmath91 and @xmath92 are the 1 sigma standard deviations on the model coefficients . we calculate the pf of the persistent emission pps in different energy bands and find a slight energy dependence in both _ chandra _ and _ data . at the lowest energies , @xmath93 kev , the pf is about @xmath25 . it increases to @xmath94 in the @xmath95 kev energy range , and remains constant at higher energies ( figure [ rmsvsene ] ) . we find the same dependence and pf values during the dip intervals in both _ chandra _ and _ nustar_. we do not perform timing analysis on the burst intervals . both _ chandra _ and _ nustar _ suffer high instrumental dead - time during bursts , and _ chandra _ data suffer a small pile - up percentage at the peak of the bursts ( @xmath25 ) , all of which distorts the burst pp . we perform our spectral analysis using xspec @xcite version 12.8.1 . the photo - electric cross - sections of @xcite and the abundances of @xcite are used throughout to account for absorption by neutral gas . all quoted uncertainties are at the @xmath96 level , unless otherwise noted . we extract the heg m1 spectrum of the persistent emission intervals as defined in section [ timeana ] and fit it ( binned to a s / n ratio of 10 ) with an absorbed pl . residuals in the form of narrow emission lines are present in the spectra in the energy range @xmath97 kev ( figure [ specpersisaio ] ) . to properly model the continuum we ignore data in the energy range @xmath98 kev . the absorbed pl provides a good fit to the data with a reduced @xmath99 of 0.7 for 247 degrees of freedom ( d.o.f . ) . we also fit the continuum with a black - body function ( bbody in xspec , bb hereafter ) , and although we find a statistically acceptable fit with a reduced @xmath99 of 0.9 , significant fit residuals at low and high energies ( @xmath100 and @xmath101 kev ) are present . a diskbb model gives as good a fit as the pl with a reduced @xmath99 of 0.76 ; however , the temperature @xmath102 kev of the inner disk , is too high for an accreting pulsar . we conclude that a simple pl is sufficient to explain the _ chandra _ data alone . we add the 6.0 to 8.0 kev data , including three gaussian lines at 6.4 , 6.65 , and 7.0 kev to account for the residuals , and re - fit . we find a reduced @xmath99 of 0.7 for 291 d.o.f ( figure [ specpersisaio ] ) . table [ persisprop ] lists the best fit parameters to the persistent emission continuum and features , along with their @xmath96 uncertainties . the addition of the 6.4 , 6.65 , and 7.0 kev gaussian components ( one at a time ) improve the fit by a @xmath103 of 24 , 31 , and 10 , respectively . we find that all three lines are narrow with comparable fluxes and equivalent widths ( ew ) within uncertainties . most likely the lines are due to neutral or near - neutral fe , and highly ionized fe xxv ( he - like ) and fe xxvi ( h - like ) . we note here that there is a ccd gap in the heg m1 at 6.3 kev , right below the energy of the 6.4 kev fe feature , resulting in loss of counts at that energy , which renders the fit parameters of this line uncertain . @xmath53 in units of photons kev@xmath19 @xmath15 s@xmath19 . @xmath56 in units of erg @xmath15 s@xmath19 . @xmath57 total unabsorbed flux . @xmath104 and @xmath105 are caclualted in the 0.5 - 70 kev energy range . fluxes of the gaussian components are in the 6 - 8 kev range . + next , we extract the _ nustar _ fpma and fpmb spectra , in the 3@xmath070 kev range for the persistent intervals simultaneous with _ chandra _ , and group them to have a s / n ratio of 25 . we fit the _ nustar _ spectra and the heg m1 spectrum simultaneously , including a normalization factor to all model fits to take into account any cross - calibration uncertainties between the three instruments . we link all fit parameters between the three spectra except for the normalization factor . both _ spectra show a broad emission line centered at around 6.65 kev , most likely corresponding to the fe line complex detected with _ chandra_. hence , for our initial fits , we exclude from both _ chandra _ and _ nustar _ the @xmath98 kev energy range . a single component model , i.e. , pl or bb , does not give a satisfactory fit with a reduced @xmath106 . an absorbed _ cutoffpl _ ( a pl with an exponential rolloff ) model gives a better fit , but it is also statistically unacceptable ( reduced @xmath107 ) . the addition of a bb model improves the fit dramatically with a reduced @xmath108 ( figure [ nuschanfit ] , panel e ) . however , this model results in an absorption hydrogen column density three times lower than what we get from _ chandra _ alone , resulting in large residuals at the lower end of the spectrum , along with residuals at 10 kev . hence , we fix @xmath109 to the _ chandra _ value of @xmath110 @xmath15 . this gives a similar fit quality compared to the above , but emphasizes the residuals around 10 kev in the form of a broad trough ( figure [ nuschanfit ] , panel d ) . adding a negative broad feature ( _ cyclabs _ for the same number of d.o.f . hence , we use _ cyclabs _ in the rest of the analysis . ] in xspec ) to the model and thawing @xmath109 results in the best fit to the continuum with a reduced @xmath99 of 0.95 ( for 767 d.o.f . ) . we also investigate the effect of other model continua on the presence and shape of the 10 kev feature . first , instead of a _ cutoffpl _ model , we fit the spectrum using ( 1 ) a fermi - dirac form of cutoff ( _ fdcut _ , * ? ? ? * ) , and ( 2 ) a negative@xmath0positive pl exponential ( _ npex _ , * ? ? ? * ) , both used for fitting accreting x - ray pulsar spectra . neither model , modified by absorption , gives a good fit to the data ( reduced @xmath111 ) . adding a bb results in a reduced @xmath99 of 1.3 . we then remove the bb component and add a negative broad feature to the models . both _ fdcut _ and _ npex _ give a reduced @xmath99 of 1.1 but fail to reproduce the soft part of the spectrum . adding both a bb component and a negative feature gives a good fit to the data in both cases ( reduced @xmath99 of 0.94 and 0.95 for 765 and 766 d.o.f for _ npex _ and _ fdcut _ , respectively ) . we conclude that the 10 kev feature and the bb component are present in the data regardless of the shape of the continuum used . we , therefore , adopt the simplest empirical model , i.e. , _ cutoffpl _ , for the rest of our analysis since it has less free parameters than the above two for comparable fit results . moreover , the parameters of the bb component and the 10 kev feature are consistent within 1@xmath42 in all three models . + finally , using the _ cutoffpl _ continuum model , we add the [email protected] kev data . we include three gaussian lines with centroid energies and widths fixed to the values derived from the _ chandra_data ( reduced @xmath99 of 0.91 for 919 d.o.f . ) . we also fit one gaussian line to the data , with all parameters left free to vary . we find a line centroid energy @xmath112 kev and a width @xmath113 kev ( reduced @xmath99 of 0.92 for 919 d.o.f . ) . we conclude that our best fit model for the _ nustar _ and heg m1 persistent emission spectra consists of a bb , a _ cutoffpl _ , a @xmath1 kev feature , three gaussian lines with centroid energies and widths fixed to the _ chandra_-alone values ( or 1 gaussian component with all parameters left free to vary ) , all affected by neutral absorption ( _ tbabs _ in xspec ) , and a constant normalization . table [ nuschanpersisprop ] gives the best fit parameters . @xmath53 in units of photons kev@xmath19 @xmath15 s@xmath19 . @xmath56 in units of erg @xmath15 s@xmath19 . @xmath114 fitting one gaussian line to the fe line complex . @xmath115 fixed to the _ chandra _ best fit results . @xmath63 total unabsorbed flux . @xmath116 , @xmath104 , and @xmath105 are caclualted in the 0.5 - 70 kev energy range . fluxes of the gaussian components are in the 6 - 8 kev range . we extract the _ chandra _ heg m1 spectra for each of the dip intervals separately ( here we define the duration of a dip interval as starting from the end of a burst until the time it recovers to 95% of the persistent level ) . we fit all 7 intervals simultaneously with an absorbed pl model . the hydrogen column density is linked for all spectra to ensure that they are all equally absorbed . first , we allow the pl indices and normalizations to vary and we find that the pl index is consistent across all spectra . therefore , we also keep the indices linked . we do not find any flux variability between the different dips , and estimate an average flux and 1@xmath42 standard deviation of @xmath117 erg s@xmath19 @xmath15 . we also find no variability in the fluence deficiency during the different dips ( the actual deficit of fluence from the persistent emission during a dip ) with an average of @xmath118 erg @xmath15 . table [ dipprop ] gives the spectral results for the individual dip intervals . motivated by the differences in the count rates at the dip minima ( figure [ fitdips ] ) , we extract the heg m1 spectra of all seven dips in an 80 s interval centered at @xmath62 and fit them simultaneously with an absorbed pl model , keeping only the normalizations free . we do not find any flux variability ( at the @xmath119 level ) in the minimum level the dips reach after each burst ( table [ dipprop ] ) . we repeat the analysis for a 40 s interval centered at @xmath62 , and reach the same conclusions . next , we extract the heg m1 spectrum for dip intervals _ collectively _ , and group them so that each bin has a s / n ratio of 10 . we fit the spectrum with an absorbed pl and find an absorption hydrogen column density and a pl index consistent within errors with the results we find for the persistent emission intervals ( table [ persisprop ] ) . we find an unabsorbed average flux @xmath120 erg s@xmath19 @xmath15 . the ratio of the average flux ( over the entire dip intervals ) to the persistent emission flux is @xmath121 , or a @xmath122% drop in the persistent emission flux . finally , we find that some residuals are present in the heg m1 at high energies ( @xmath123 kev ) , in the form of excess emission , similar to what is seen in the persistent emission spectrum . hence , we add three gaussian lines at 6.4 , 6.65 , and 7.0 kev . table [ persisprop ] lists the best fit parameters to the dip intervals . unfortunately , due to the low number of counts in the dip spectra , we are not able to adequately constrain the parameters of the gaussian components , except for their energies . nonetheless , comparing the dip and persistent emission fit parameters , we do not find dramatic changes in the line properties . these results are discussed in section [ discuss ] . we then apply the same model we used to fit the persistent interval _ chandra_+_nustar _ broadband spectrum to the dip broadband spectrum . we find very similar values between persistent and dip intervals , except for the flux of the _ cutoffpl _ component , which decreased by 10% during dips . the fluxes of the bb and the fe line component ( considering one gaussian line fit to the 6.7 kev excess ) did not change within their @xmath96 error . these results are shown in table [ nuschanpersisprop ] . to achieve better s / n ratio for the spectral fitting of the lines , we extract the heg m1 spectrum of the persistent and dip intervals together . we fit the spectrum with an absorbed pl model and three gaussian emission lines ( figure [ specpersisaio ] ) . table [ persisprop ] lists the best fit parameters along with their @xmath96 uncertainties . according to the persistent+dip spectrum , which has better statistics than the persistent or dip spectra alone , the highly ionized lines are narrow with widths of about 30 ev . they also contribute similarly to the total flux . the neutral fe , on the other hand , has a larger width and flux compared to the other two . the broadband model used to fit the persistent and dip spectra alone is also successfully fit to the dips+persistent _ nustar _ and _ chandra _ spectra ( table [ nuschanpersisprop ] ) . all the parameters of the best - fit model were compatible with the persistent and dip fits . a one gaussian emission line fit to the 6.7 kev excess results in a width @xmath124 kev and an ew=129 ev . finally , we also fit the 6.7 kev excess using a diskline model and find @xmath125 kev , @xmath126 gm / c@xmath127 ( @xmath128 cm ) , and @xmath129@xmath130 ( all quoted uncertainties are at the @xmath131 level ) . we extract the heg 1 spectrum for each of the seven bursts seen with _ chandra _ , and fit them simultaneously with an absorbed pl . we link the hydrogen column density in the fit . we find a consistent pl index for all spectra and , therefore , we also link the index thereafter . the pl normalizations are left free to vary , to account for any flux variability between the bursts . we find @xmath132 @xmath15 , and a pl index of @xmath133 . we report in table [ burstprop ] all burst spectral parameters . we find very similar energetics between the different bursts ; flux and luminosity variability ( at the @xmath131 level ) is observed only between bursts # 4 and # 7 ( these are the shortest and the longest burst , respectively , table [ burstprop ] ) . in terms of fluence , all bursts emitted comparable ( at the @xmath134 level ) amounts of energy . the mean and 1@xmath42 standard deviation of the fluxes and fluences are @xmath135 erg @xmath15 s@xmath19 , and @xmath136 erg @xmath15 , respectively . we also convert the 1 s peak count rates of the seven bursts to fluxes using pimms ( due to the low count statistics ) . we find an average peak flux of @xmath137 erg s@xmath19 @xmath15 , equivalent to a luminosity of @xmath138 erg s@xmath19 at 8 kpc . this value should be regarded as a lower limit due to a @xmath25 pile - up , and the narrow energy band for which it was derived ( @xmath13 kev ; _ nustar _ data were excluded from this analysis due to severe dead - time effects ) . we extract the heg 1 spectrum of all seven bursts collectively , grouped to a s / n of 7 , to search for any features present in their added spectrum . we fit the spectrum with an absorbed pl , and find a hydrogen column density @xmath139 @xmath15 , and a pl index @xmath140 . no prominent absorption and/or emission features are seen in the spectrum . we then extract the @xmath141 kev _ nustar _ spectrum of the three bursts seen simultaneously with _ chandra _ , binned to a s / n ratio of 15 . we fit the spectrum with an absorbed _ cutoffpl _ resulting in a reduced @xmath99 of 0.89 for 637 d.o.f .. some residuals around 10 kev can be seen , similar to what we find in the persistent and dip spectra . including a _ cyclabs _ feature to the model improved the fit slightly , resulting in a reduced @xmath99 of 0.86 for 634 d.o.f . , which is a @xmath103 of 17 for 3 additional parameters . figure [ specburstaio ] shows the data and best fit model , while the parameter values are listed in table [ persisprop ] . we see no excess emission between 6 and 8 kev . we derive a 3@xmath42 upper limit of 119 ev on the ew of a line with centroid energy at 6.7 kev . we also derive a 3@xmath42 upper limit on the flux of a line at the same energy and a width of 0.4 kev of @xmath142 erg s@xmath19 @xmath15 . using the co - added _ chandra _ burst spectra , we derive a 30 ev @xmath131 upper limit on the ew of the neutral fe , and @xmath143 ev on the highly ionized species . assuming a width of 0.05 kev for the fe k and the highly ionized species , we find a @xmath131 flux upper limits of the order of @xmath144 erg s@xmath19 @xmath15 for all three lines . these upper limits indicate that if any of the lines we detect in the dip and persistent spectra brightened proportionally to the burst flux ( on the average by a factor of 5 ) , we should have been able to detect them . however , if the line fluxes remained constant during the bursts , their presence could be masked by the much brighter burst continuum . finally , we note that the bb component is not required by the fit to the bursts spectrum at a high significance . however , the hydrogen column density is lower at the @xmath131 level than the value derived for the dips and persistent intervals ( table [ nuschanpersisprop ] ) . hence , fixing the column density at @xmath145 @xmath15 , we find residuals at the lower end of the spectrum , which are well fit with a bb component with @xmath146 kev and a @xmath3 kev flux of @xmath147 erg s@xmath19 @xmath15 , consistent with the bb temperatures and fluxes during the persistent and dip intervals . these results are discussed in section [ discussbbs ] . + we divide the broadband persistent emission spectrum into five pulse phase bins , which we fit simultaneously with our best fit model described above . we fit the @xmath148 kev excess energy with one gaussian line . we first leave all model parameters free to vary . we link one model parameter after another ( starting from the least variable according to a @xmath99 test ) , and record the f - test significance at each step to assess the significance of leaving the parameter free in the fit . we find that the fit parameters of the 6.7 and 10 kev features do not show significant changes with pulse phase . on the other hand , the bb and _ cutoffpl _ fit parameters tightly follow the pp shape ( figure [ phaseresspec ] ) . we find a decrease in the photon index @xmath149 and an increase in the bb temperature indicating that the x - ray spectrum hardens at pulse maximum . the rolloff energy is also anti - correlated with the pulse shape . our temporal analysis shows that six out of the seven detected bursts have comparable durations with an average of about @xmath150 s. these consist of a single pulse with a faster rise than decay time . the seventh burst detected by _ chandra _ is the only outlier with a duration of about 25 s , consisting of two pulses . these temporal properties are similar to the properties of hundreds of bursts recorded during the first two outbursts from the bp ( e.g. , * ? ? ? * ) . from the _ chandra _ data , we can derive the @xmath151 parameter , the ratio of the fluence in the persistent emission to the fluence in the bursts . we find @xmath152 for all bursts ( except for the sixth one where @xmath153 ) , with an average value of 10 . the same value was also derived during the first two outbursts from the source @xcite . as pointed out by @xcite , this small value of @xmath151 is inconsistent with thermonuclear burning as the origin for the bp bursts . this value is consistent with the observed bursts being type ii bursts , similar to what is seen in the rapid burster , caused by some sort of instability associated with the accretion disk @xcite . unlike the rapid burster , however , the bp does not display any correlation between the fluence in a burst and the time to the following burst ( e.g. , * ? ? ? this is also evidenced by our data , where the fluence emitted during all seven bursts is constant while the intervals between bursts changed by up to a factor of 4 . similar to the previous two outbursts , the bursts we detect in _ chandra _ and _ nustar _ are followed by a dip , where the x - ray emission decreases by 10% on average , and 40% at dip minimum . the emission exponentially recovers back to the pre - burst persistent level , on time - scales of a few hundred seconds . we find that the fluence in a burst and the integrated flux deficiency of the following dip are consistent within @xmath96 ( tables [ burstprop ] and [ dipprop ] ) . we estimate an average burst fluence ( @xmath37 kev ) of @xmath154 erg @xmath15 , and an average _ missing _ dip fluence of @xmath155 erg @xmath15 . such a correlation was also seen during the previous two outbursts ( e.g. , * ? ? ? these authors suggested that the energy emitted during a burst could be compensated by the deficit in energy during the following dip . a very simple picture would be that accretion - disk instabilities would allow for a sudden and rapid increase of the mass - inflow rate onto the polar cap of the neutron star from a reservoir ( e.g. , the accretion disk ) . the dips , then , would be the result of a small fraction of the continuously accreted matter disappearing to replenish this reservoir . the broadband spectrum of the bp is the typical spectrum of an accreting x - ray pulsar at high accretion rates ( see e.g. , * ? ? ? * for a review ) . it is well fit with a hard component , modeled as a pl with an exponential rolloff , an fe line complex , a soft component modeled with a bb , and a 10 kev feature , all modified by absorption . in the following , we will discuss these different components and their interplay between persistent , dip , and burst emission , except for the fe line complex which is discussed in section [ disfe ] . the high persistent x - ray luminosity of during the present observation ( @xmath156 erg s@xmath19 ) implies that the emission is coming from an accretion column , where the kinetic energy of the infalling gas onto the polar cap is converted to radiation via a radiative shock above the thermal mound @xcite . thermal photons from the mound , as well as cyclotron and bremsstrahlung radiation , are converted to high - energy photons via inverse compton scattering . hence , the resulting x - ray spectrum will depend on several parameters such as the geometry of the system and the properties of the compact source @xmath0 mainly its dipole magnetic field @xmath0 among others @xcite . even with the small dipole magnetic field of , and its complicated accretion geometry @xcite , the parameters we derive from our phenomenological fit compare reasonably well to other accreting x - ray pulsars ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the photon index of the _ cutoffpl _ is slightly lower than in most cases implying a harder spectrum , which could be the result of the higher luminosity of the source . simply put , a higher accretion rate onto the poles , would lead to a higher electron density in the accretion column and to higher compton @xmath157parameter causing a harder spectrum . the spectrum also shows a lower energy rolloff compared to other sources , which could be due to the relatively low magnetic field of the source @xcite . the 10 kev feature is not unique to , and has previously been reported in other accreting x - ray pulsars , e.g. , vela x-1 ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * see also @xcite for a review ) . this feature is not always necessarily observed as an absorption trough and sometimes manifests itself as a broad emission feature or a wiggle . it is believed to be the result of modeling accreting x - ray pulsar spectra with simple empirical functions , when the true physics giving rise to their x - ray spectra is far more complicated , especially when they are emitting near the eddington limit ( see * ? ? ? * for a discussion ) . in a few cases , however , such as in the case of the be / x - ray binary swift j1626.6@xmath05156 @xcite , an absorption line at 10 kev was interpreted as a cyclotron resonance scattering feature ( crsf ) , evidenced by the presence of a weak second harmonic and the fact that the b field strength , derived from the line energy , was consistent to the value derived from the spin - up rate of the source @xcite . in the 10 ks of _ nustar _ data that we consider here , we find no evidence of a second harmonic at about twice the energy of the 10 kev feature ( i.e. , @xmath158 kev ) , and the @xmath159field strength corresponding to the line energy ( @xmath160 g for a 10 kev line energy ) is significantly larger than the estimates we derive in section 4.4 . moreover , crsfs usually show strong dependence with pulse phase ( e.g. , * ? ? ? * ) , which we do not observe in our phase - resolved spectroscopy . hence , we consider the 10 kev feature in spectrum to be a defect of our continuum modeling . we note , however , that unlike the bp , other x - ray pulsars showing 10 kev features invariably show crsfs . soft excess emission is often modeled with a bb component in accreting x - ray pulsars . * see also @xcite ) showed that in luminous sources such as , the most likely source for this bb - like emission is the inner region of the accretion disk , from where the reprocessed hard x - ray emission of the accretion column is emitted . such reprocessed emission also pulsates at the pulse period of the hard x - ray component , most likely with a lower pulsed fraction due to the large area where the reprocessing is taking place . this is in agreement with both the change of the bb temperature and flux with pulse phase ( figure [ phaseresspec ] ) , and the slight decrease of the pf at low energies compared to the high energies ( figure [ rmsvsene ] ) . under this assumption and for isotropic emission , the inner radius of the disk is , @xmath161 @xcite , where @xmath42 is the stefan - boltzmann constant , @xmath162 is the non - thermal x - ray luminosity , and @xmath163 is the bb temperature in k. the temperature we calculate , however , is the apparent temperature of the plasma and is related to the effective temperature through a color - correction ( or hardness ) factor , @xmath164 @xcite , which is usually taken to be between 1.5 and 2 . hence , the true inner radius of the bb emission area is @xmath165 @xcite . we find @xmath166 cm ( 3@xmath42 confidence ) , for @xmath167 . this is consistent with the expected small accretion disk radius considering the low b field of the source and its high luminosity . the burst broadband spectrum requires only emission from the non - thermal component , with fit parameters similar to the ones we derive for the persistent emission . this reinforces our above picture where we envisioned the burst emission to be the result of a sudden increase of the mass accretion rate onto the neutron - star pole . the non - detection of the bb component implies that the reprocessing of the non - thermal emission may not have taken place during bursts . this is possible , for instance , if the burst emission is anisotropic away from the reprocessing material , i.e. , the inner accretion disk . such anisotropy for the bp has already been discussed by @xcite and @xcite to explain the extremely high luminosities of the bursts during the previous two outbursts , which reached luminosities two orders of magnitude above eddington . this conclusion is also supported by the timing properties of the source , for which the hard x - ray pf has been seen to increase prominently during the bursts @xcite . finally , the broadband spectrum of the dip intervals is similar to the persistent emission spectrum . the flux deficiency during dips is primarily seen in the non - thermal component , where the _ cutoffpl _ flux decreased by @xmath25 compared to the persistent emission flux . this is again in agreement with the accretion picture where the dips are essentially the result of a fraction of the long - term accreted matter not reaching the neutron star pole , instead replacing the matter that produced the preceding burst . the _ asca _ observations during the first outburst of revealed a feature between 6 and 8 kev in its persistent emission spectrum . @xcite modeled the spectrum with a gaussian line with a centroid energy of @xmath148 kev and an ew of about 300 ev . the line was not resolved by the spectral resolution of _ asca _ , but its energy is indicative of a blend of emission lines from different species . the _ nustar _ persistent and dip spectra show a similar emission excess at the same centroid energy and a somewhat smaller ew ( although consistent at the 3@xmath42 level ) . using the _ chandra _ hetgs we are able to resolve the broad feature into three emission lines , which we identify as fe k from neutral and/or lowly ionized species at @xmath168 kev , and highly ionized fe xxv and fe xxvi at @xmath169 and @xmath170 kev ( these are the best estimates of the line energies from the dip@xmath171persistent emission spectrum , see also @xcite ) . we discuss first the fe emission lines from highly ionized species in the x - ray spectrum of . the gas producing the lines is most likely photo - ionized by the x - ray emission of the neutron star . in photo - ionized gas , he - like fe emission lines are produced by recombination and resonant scattering @xcite , and include four different transitions at slightly different energies , the resonant line @xmath172 , the two inter - combination lines @xmath173 and @xmath174 , and the forbidden line @xmath175 ( see e.g. , * ? ? ? * ; * ? ? ? , we could not resolve the different resonances , however , from the centroid energy of the fe xxv line , @xmath169 kev , we can safely conclude that the resonant line @xmath172 ( with mean energy at 6.700 kev ) contributed minimally to the line strength , hence the emission is dominated by recombination @xcite . to investigate the origin of the highly ionized species , we simulate xstar grids @xcite based on the broadband x - ray spectrum of the source . we choose a covering fraction of 0.2 ( assuming an accretion disk ) , and solar abundances as in @xcite . due to the high accretion rate of the source , one would expect the photo - ionized gas to have a very large density ; hence , we examined different values of the gas density @xmath78 from @xmath176 to @xmath177 @xmath178 , each time multiplying by 10 , to test the effects of density on reproducing the line shapes . we find that the best densities to reproduce the lines , resulting in reasonable values of the ionization parameter ( @xmath179 ) , are 10@xmath180 and 10@xmath181 @xmath178 . here we consider a density of @xmath182 @xmath178 . finally , we assumed no turbulence in the gas . fitting this xstar simulated grid to the persistent+dip spectrum , we find a best fit value for the column density in the gas of @xmath183 @xmath15 ( @xmath184 @xmath15 ) , and for the ionization parameter @xmath185 . the @xmath186 parameter is related to the total x - ray luminosity of the source , @xmath187 , the density of the ionized gas , @xmath78 , and its distance from the ionizing source , @xmath188 , by @xmath189 . solving for @xmath188 , we find @xmath190 cm ( 3@xmath42 confidence ) . this distance is similar to the estimates of the ionized gas location in other sources ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , and points towards reprocessing in an accretion disk corona . the other interesting feature in our spectra is the ( quasi@xmath0 ) neutral fe at 6.4 kev . there are three possibilities for the formation site of the fluorescence fe k line in x - ray binaries : ( i ) a wind from the companion seems to be unlikely in the case of , since the companion is a low - mass star and accretion is most likely occurring through roche - lobe overflow @xcite ; ( ii ) the companion surface via reflection , which is also hard to achieve , because of the very low inclination of the system @xcite would result in a very low ew for any fe features @xcite ; or ( iii ) the outer regions of the accretion disk , by means of irradiation from the central source . to test this third possibility , we fit a second xstar grid , similar to the one above , to the dips+persistent spectrum . we find that the ( quasi@xmath0 ) neutral fe line is well reproduced with an ionization parameter @xmath191 ( @xmath192 at 3@xmath42 confidence ) , much lower than the value required to model the highly ionized lines . this ionization parameter corresponds to a distance from the neutron star of @xmath193 cm ( @xmath194 cm at 3@xmath42 confidence ) . this seems to point to the outer regions of the accretion disk as the likely origin of the fe k. other x - ray binary sources showed , similar to , fe k@xmath151 lines most likely from the outer region of an irradiated disk ( e.g. , * ? ? ? the 6.7 kev excess in the simultaneous _ nustar _ and _ chandra _ data is also consistent with a broad line which we fit using a diskline model ( see also @xcite ) . the inner - disk radius that we find ( @xmath195 cm ) is consistent with the results of @xcite and in agreement with a magnetically truncated accretion disk . this radius , however , is more than an order of magnitude smaller than the result we get from the xstar fits to the highly ionized lines ( assuming that the broad line is consistent with fe xxv , @xcite ) . this could either be due to the uncertainties in the density of the ionizing gas and/or in the distance to the source , or the fact that other broadening mechanisms , e.g. , compton scattering , are contributing to the line profile . due to the low statistics of the present observation , we could not constrain any variations in the separate _ chandra _ lines during dips ( table [ persisprop ] ) . the _ nustar _ gaussian line fit to the @xmath148 kev excess has energy , width , ew , and flux consistent within @xmath96 between persistent and dip emission ( table [ persisprop ] ) . the excess emission is not detected during bursts , which means that either the super - eddington burst x - ray luminosity fully ionized the line - emitting region , including the fe k region , or that the line strength remained more or less constant during bursts , but was masked by the very bright continuum . to explore the first possibility , we simulated the same xstar grid as above , but instead of the persistent x - ray luminosity , we used the x - ray luminosity as derived from the bursts . we fit this xstar table to the burst spectrum and derived a 3@xmath42 lower limit on the ionization parameter @xmath196 . this limit represents the lowest ionization state that would result in the featureless spectrum that we see during bursts . this lower limit translates into an upper - limit on the radius of the ionized material of @xmath197 cm . hence , the burst luminosity is capable of fully ionizing the region of the disk where the highly ionized lines are thought to originate ( @xmath198 cm ) . at the @xmath131 upper limit , it is capable of fully ionizing the region of the neutral fe ( @xmath199 cm ) . this result does not exclude the second possibility . similar to the previous two outbursts from , the increase in the x - ray luminosity of the source is accompanied by an increase in the spin period of the neutron star ( gbm pulsar team , see footnote 2 ) . this spin - up factor and persistent pulsed emission indicate that the accretion onto the neutron star is not quenched at the disk - magnetosphere boundary , i.e. , the propeller effect is not acting . for spin - up to occur during accretion , the inner disk rotational frequency at the magnetospheric radius has to be greater than the neutron star spin frequency , which results in an upper limit on the magnetic dipole field of , @xmath200 where @xmath187 is the total x - ray luminosity assuming a distance of 8 kpc , @xmath201 is the gravitational constant , @xmath202 and @xmath188 are the mass and radius of the neutron star , taken to be 1.4 @xmath203 and 10 km , @xmath79 is the neutron star spin frequency , and @xmath204 is a dimensionless parameter between 0.5 and 1 @xcite . we find @xmath205 g. this value is consistent with the estimates of the previous two outbursts ( e.g. , * ? ? ? * ; * ? ? ? we note that the true upper - limit is lower than the above derived value since the source started spinning - up at earlier stages in the outburst when the source luminosity was lower . assuming that the bb component is the result of reprocessing in the inner regions of the accretion disk , we could also use the bb radius estimate ( section [ discussbbs ] ) to derive the strength of the dipole field of the source . the inner accretion disk radius can be written as @xmath206 with @xmath207 . we find @xmath208 which is consistent with the above upper - limit and the expected low dipole field of the source . we studied the broad - band x - ray emission ( @xmath3 kev ) of the bp from a @xmath209 h simultaneous _ nustar_-_chandra _ observation during its third detected outburst since discovery and after nearly 18 years of quiescence . these data were taken a few days before the outburst reached its peak . a total of seven bursts are detected during our observation . temporal analysis revealed that the first six bursts have comparable shapes , consisting of a single pulse with duration of @xmath150 s , and a faster rise than decay time . the last burst has a double - peaked morphology with a duration of about 25 s. all seven bursts , however , have equal fluences with an average of about @xmath210 erg @xmath15 . similar to previous results , we find an average ratio of the burst to the persistent emission fluence @xmath211 ( with the exception of one burst where @xmath153 ) , pointing to the type ii origin for the bursts . each of the seven bursts is followed by a dip in the persistent emission flux , which recovers exponentially with a characteristic time - scale @xmath212 s. we find an average missing fluence in the dip of about @xmath210 erg @xmath15 , consistent with the fluence emitted in the bursts . this indicates that the energy emitted during the burst is compensated for in the dip , and that the long - term accretion rate is constant . the pulse - profiles of the persistent and the dip intervals are nearly sinusoidal with only weak contribution from the second harmonic . the pf increases from about 10% at 4 kev to 15% at 13 kev , and remains constant thereafter . the bp persistent and dip broadband spectra are identical and well fit with a bb with @xmath213 kev , a _ cutoffpl _ with an index @xmath214 and an energy rolloff @xmath215 kev , a 10 kev feature assumed to be the result of inadequate modeling of the _ cutoffpl _ , and a 6.7 kev emission feature , all modified by neutral absorption . phase - resolved spectroscopy shows that the bb and the _ cutoffpl _ components show variations at the pulse period of the source , both getting harder at pulse maximum , whereas no significant changes are seen in the 10 kev and the 6.7 kev feature . assuming that the bb is reprocessing of the non - thermal emission in the inner regions of the accretion disk , we derive an inner disk radius @xmath216 cm . this radius translates into a dipole magnetic field of @xmath5 g. the _ chandra_/hetg spectrum resolved the 6.7 kev feature into ( quasi-)neutral and highly ionized fe xxv and fe xxvi narrow emission lines . modeling the highly ionized lines with xstar places the emitting region at a distance of about @xmath217 cm from the neutron star , consistent with an accretion disk corona origin . using a similar xstar grid to model the ( quasi- ) neutral fe , we find that it originates from a distance @xmath218 cm , most likely the outer regions of an accretion disk . the broadband burst spectrum , with a peak flux more than an order of magnitude higher than eddington , is well fit with a _ cutoffpl _ and a 10 kev feature , with similar fit values compared to the persistent and dip spectra . the burst spectrum , however , lacks a thermal component ( bb ) and fe features . if the burst emission were anisotropic ( beamed ) , the lack of the bb component is expected since no reflection of the burst photons on the inner disk would take place . similarly the fe xxv , fexxvi , and the neutral fe lines would remain at the flux levels detected in the persistent and dip emission and , therefore , are too weak to be detected above the strong burst continuum . if , on the other hand , the burst emission is isotropic , we show that the disk region where the fe xxv and fexxvi lines would be produced is now fully ionized ; the neutral iron line could still be at very low levels and masked by the continuum . in that case , however , we would expect a strong bb component , which is not detected . we conclude that , as suggested by @xcite and @xcite , the burst emission is highly beamed . this work was supported under nasa contract no . nng08fd60c , and made use of data from the _ nustar _ mission , a project led by the california institute of technology , managed by the jet propulsion laboratory , and funded by the national aeronautics and space administration . we thank the _ nustar _ operations , software and calibration teams for support with the execution and analysis of these observations . this research has made use of the _ nustar _ data analysis software ( nustardas ) jointly developed by the asi science data center ( asdc , italy ) and the california institute of technology ( usa ) .
we report on a 10 ks simultaneous _ chandra_/hetg@xmath0_nustar _ observation of the bursting pulsar , , during its third detected outburst since discovery and after nearly 18 years of quiescence . the source is detected up to 60 kev with an eddington persistent flux level . seven bursts , followed by dips , are seen with _ chandra _ , three of which are also detected with _ nustar_. timing analysis reveals a slight increase in the persistent emission pulsed fraction with energy ( from @xmath1% to @xmath2% ) up to 10 kev , above which it remains constant . the @xmath3 kev spectra of the persistent and dip emission are the same within errors , and well described by a blackbody ( bb ) , a power - law with an exponential rolloff , a 10 kev feature , and a 6.7 kev emission feature , all modified by neutral absorption . assuming that the bb emission originates in an accretion disc , we estimate its inner ( magnetospheric ) radius to be about @xmath4 cm , which translates to a surface dipole field @xmath5 g. the _ chandra_/hetg spectrum resolves the 6.7 kev feature into ( quasi-)neutral and highly ionized fe xxv and fe xxvi emission lines . xstar modeling shows these lines to also emanate from a truncated accretion disk . the burst spectra , with a peak flux more than an order of magnitude higher than eddington , are well fit with a power - law with an exponential rolloff and a 10 kev feature , with similar fit values compared to the persistent and dip spectra . the burst spectra lack a thermal component and any fe features . anisotropic ( beamed ) burst emission would explain both the lack of the bb and any fe components .
the standard model of work extraction is based on an engine composed of a working medium , a piston and two heat baths that are at equilibrium with different temperatures . its importance stems from its success to set an universal bound to any work extraction process , the carnot bound , which shows , in agreement with the kelvin s formulation of the second law of thermodynamics , two different temperatures are needed for work extraction . besides , there is a complementary picture of an engine as a self - oscillating system _ focusing on their ability to convert energy inputted at one frequency ( usually zero ) into work outputted at another , well - defined frequency " _ . this picture seems to apply to all types of turbines and motors @xcite and one can expect that it is equally correct for engines powered by a flux of photons . the standard description of the photovoltaic cell involves the following processes @xcite : + i ) generation of the charge carriers due to the absorption of photons , + ii ) subsequent separation of the photo - generated charge carriers in the junction . as noticed in @xcite the often used explanation of the second process as caused by the emerging electric field in p - n junction can not be correct . charge separation is supposed to produce a dc current which , on the other hand , can not be driven in a closed circuit by a purely electrical potential difference . a standard thermodynamical explanation of electric current ( work ) generation in photovoltaic and thermoelectric heat devices is the following : + electrons gain energy in a form of heat current @xmath0 from the hot bath , then flow against potential difference @xmath1 producing useful power @xmath2 , where @xmath3 is an electric current . a part of the heat described by the heat current @xmath4 is dumped to the cold bath . + the laws of thermodynamics put the following constraints @xmath5 @xmath6 adding kinetic equations describing the processes of electron - hole creation , thermalization to the ambient temperature and recombination one obtains the correct formulas for the open circuit voltage and voltage - current relation . however , this picture does not explain the mechanism of persistent steady work extraction . similarly , for a steam engine the net pressure due to the temperature difference obviously provides the net force acting on the piston but to explain the permanent periodic action of this engine we have to understand the details of operation of a piston linked to a flywheel and valves . this is exactly the place where the mechanism of self - oscillation supported by the external constant energy flow enters into the game . in the following we propose a model in which plasma oscillations play a role of the periodic motion of a piston " and show that , indeed , under realistic assumptions a positive power is supplied to this essentially classical macroscopic oscillator . subsequently , the collective charge oscillations at thz frequencies are rectified by a p - n junction diode ( valve " ) to the output dc current . the mathematical formalism is based on the quantum markovian master equations for slowly driven open quantum systems studied in @xcite ( compare @xcite,@xcite for the fast driving case ) , and consistent with thermodynamics . we consider a model of heat engine which consists of a working medium " called simply a system , two baths at different temperatures , and a work reservoir " called often a piston " which is a system supplying to or extracting work from the working medium . because work , in contrast to heat , is an ordered and deterministic form of energy we expect that a piston should be a macroscopic system operating in the semi - classical regime . therefore , within the reasonable approximation can be replaced by external deterministic driving @xcite . for the readers convenience we briefly present the formalism of quantum master equations ( qme ) based on the davies weak coupling limit @xcite , the lindblad - gorini - kossakowski - sudarshan generators and their extension to slowly varying external driving . then the thermodynamical consequences are discussed and a special generic class of models with _ diagonal , weak driving _ , applicable to the theory of solar cells , is presented . . the total system consisting of a system @xmath7 weakly coupled to a bath @xmath8 . the total hamiltonian is a sum of three terms : + i ) a `` bare '' hamiltonian of the system - @xmath9 replaced in the final formulas by @xmath10 - a physical renormalized hamiltonian " containing the lowest order corrections , + ii ) hamiltonian of the bath - @xmath11 , + iii ) a system - bath ( elementary ) coupling : @xmath12 where @xmath13 and @xmath14 are observables of the system and the bath , respectively . @xmath15 denotes the average with respect to the stationary state of the bath . two main ingredients enter the qme derived using davies weak coupling limit procedure @xcite : + a ) the spectral density of the bath @xmath16 b ) fourier components of the coupling operator @xmath17 introducing system hamiltonian spectral decomposition and bohr frequencies + @xmath18 one obtains the relations @xmath19 = -\hbar{\omega } a({\omega})\ , \ a(-{\omega } ) = { a^{\dagger}}({\omega } ) . \label{transition}\ ] ] where @xmath20 called _ transition operators _ or _ lindblad operators _ correspond to energy exchange of @xmath21 . the standard derivation yields the qme in the schrdinger picture @xmath22+\mathcal{l}\rho(t ) \label{}\ ] ] where @xmath23 + [ a({\omega}),\rho a^{\dagger}({\omega})]\bigr ) \label{}\ ] ] for general interactions @xmath24 @xmath25 + [ a_{\alpha}({\omega}),\rho a_{\beta}^{\dagger}({\omega})]\bigr ) \label{}\ ] ] where @xmath26 $ ] is a positively defined _ relaxation matrix_. the qme s obtained in the weak coupling limit possess the following properties @xcite , @xcite : \1 ) @xmath27 possesses the _ lindblad - gorini - kossakowski - sudarshan _ structure what implies that the solution of qme is a _ completely positive , trace preserving , one - parameter semigroup_. \2 ) hamiltonian part commutes with the dissipative one , i.e. @xmath28 \3 ) diagonal ( in @xmath29 - basis ) and off - diagonal density matrix elements evolve independently . \4 ) the stationary state @xmath30 satisfying @xmath31=0 $ ] and @xmath32 , always exists for finite - dimensional hilbert spaces . \5 ) for a single heat bath ( i.e. a reservoir in the thermal equilibrium state ) : \a ) spectral density satisfies the kms condition @xmath33 \b ) the gibbs state is stationary @xmath34 \c ) the following quadratic form @xmath35 is negatively defined @xcite,@xcite , where @xmath36 is the heisenberg picture generator given by @xmath37 + \bigl([a_{\beta}^{\dagger},x ] a_{\alpha}({\omega})\bigr ) . \label{dbalance}\ ] ] it means that the heisenberg picture generator @xmath38 can be treated as a hermitian , negatively defined operator acting on the space of ( complex ) observables equipped with the scalar product @xmath39 . we identify the physical entropy with the von neumann entropy of the reduced density matrix @xmath40 and use also the _ relative entropy _ @xmath41 . for the solution @xmath42 of mme , and the stationary state @xmath43 , ( @xmath44 ) @xmath45 where @xmath46[\ln\rho(t ) - \ln \bar{\rho}]\bigr)\geq 0 $ ] is interpreted as an _ entropy production_. positivity of the entropy production follows from the fact that for any completely positive and trace - preserving map @xmath47 , @xmath48 @xcite . for many independent heat baths one obtains the second law in the following form @xmath49 where @xmath50 is a heat current flowing from the @xmath51-th bath . in order to define work we introduce the time - dependent hamiltonian @xmath52 with slowly varying and typically periodic perturbation @xmath53 which gives a semi - classical description of a piston . we combine now the weak coupling assumption concerning the interaction of the system with several baths with a kind of adiabatic approximation concerning the time - dependent driving . the former condition means , practically , that the relaxation rates are much smaller than the corresponding bohr frequencies . the later is valid for the case when the time scale of driving is much slower than the time scale determined by the relevant bohr frequencies . this is a similar situation to standard adiabatic theorem in quantum mechanics and implies that in the derivation of master equation we can put the temporal values of bohr frequencies @xmath54 and transition operators @xmath55 satisfying with @xmath10 replaced by @xmath56 . under the conditions of above one obtains the following form of qme @xmath57 + \sum_j \mathcal{l}_j(t)\rho(t ) , \label{}\ ] ] where @xmath58 is a lgks generator obtained by a weak coupling to the @xmath59-th bath and for a fixed @xmath56 . the properties of generators @xmath58 imply the _ zero - th law of thermodynamics _ ( @xmath60 ) @xmath61 using the definitions @xcite : @xmath62-work provided by @xmath7 , @xmath63 - heat absorbed by @xmath7 , @xmath64 - internal energy of @xmath7 @xmath65 @xmath66 @xmath67 where @xmath68 is a heat absorbed by @xmath7 from @xmath69 , one obtains the _ first law of thermodynamics _ @xmath70 the _ second law of thermodynamics _ follows again from spohn innequality @xmath71 where @xmath72 is an entropy production caused by @xmath69 and given by @xmath73\bigr)\geq 0 . \label{senprod}\ ] ] we consider a generic case of oscillating weak driving @xmath53 which in the lowest order approximation can be replaced by the diagonal ( in the basis of @xmath74 ) operator @xmath75=0 , \label{m - perturbation}\ ] ] with the small coupling constant @xmath76 . in this case all hamiltonians @xmath56 commute . the unitary part of the dynamics @xmath77 governed by @xmath56 commutes with @xmath74 and @xmath78 . one can write @xmath79 $ ] where @xmath80 $ ] is computed with the system hamiltonian @xmath81 and @xmath82 . again for different @xmath83 s the super - operators @xmath80 $ ] commute with the hamiltonian part @xmath84 $ ] and one can use their lowest order expansion with respect to @xmath83 @xmath85 = \mathcal{l}[0 ] + \xi \mathcal{l}'[0 ] + \mathcal{o}(\xi^2 ) , \label{slowest_order}\ ] ] in the next step we use the lowest order expression for the dissipative part of the super - propagator @xmath86 ds\bigr\ } \\ & \simeq e^{t\mathcal{l}[0 ] } + g\int_0^t ( \sin \omega s)\ , e^{(t - s)\mathcal{l}[0 ] } \mathcal{l}'[0]e^{s\mathcal{l}[0]}\ , ds . \nonumber\end{aligned}\ ] ] applying now the definition of work , using the commutation properties of the generator @xmath80 $ ] , and the fact that @xmath87= 0 $ ] , one obtains the formula for the stationary average power output @xmath88 inserting the lowest order expression into and using the fact that all super - operators commute one can compute the super - operator - valued integral like a usual one . then , we apply the obvious assumption that the unperturbed dynamics @xmath89 } $ ] drives , asymptotically , any initial state @xmath90 to the stationary state denoted by @xmath91 $ ] . finally , the limit @xmath92 in can be performed leading to the second order approximation for the average output power @xmath93)^2}\mathcal{l}'[0]\bar{\rho}[0]\bigr ) . \label{spower1}\ ] ] @xmath80 $ ] also possesses the stationary state @xmath94 $ ] , i.e. @xmath80 \bar{\rho}[\xi ] = 0 $ ] and hence we can use the identity @xmath95\bar{\rho}[\xi ] = - \mathcal{l}[\xi]\bar{\rho}'[\xi ] \label{identity}\ ] ] where prime " denotes the derivative with respect to @xmath83 . then , replacing the schroedinger picture generator @xmath96 $ ] by the heisenberg picture one @xmath97 $ ] we can transform the formula into @xmath98 \frac{\omega^2}{\omega^2 + ( { \mathcal{l}^*}[0])^2}\mathcal{l}^*[0 ] m\bigr ) . \label{spower2}\ ] ] for the case when the decay rate of @xmath78 is much lower than the modulation frequency we can neglect @xmath99)^2 $ ] in to obtain the simplified expression @xmath98 \mathcal{l}^*[0 ] m\bigr ) . \label{spower3}\ ] ] the compact formula of above will be used to derive the specific expression for the solar cell power . the obtained lowest order formulas for power , are still consistent with thermodynamics . namely , assuming that the reservoir is a thermal equilibrium bath at the temperature @xmath100 we have the following properties : 1 . @xmath94 $ ] is the gibbs state with respect to the hamiltonian @xmath101 , 2 . the heisenberg picture generator @xmath97 $ ] is a negatively defined operator on the hilbert space equipped with the scalar product @xmath102 x^{\dagger } y \bigr)$ ] ( compare with , ) . using the first property one can compute @xmath103 $ ] and rewrite the formula as @xmath104 m \frac{\omega^2}{\omega^2 + ( { \mathcal{l}^*}[0])^2}\mathcal{l}^*[0 ] m\bigr)\\ & = \frac{g^2}{2 k_b t } \langle m , \frac{\omega^2}{\omega^2 + ( { \mathcal{l}^*}[0])^2}\mathcal{l}^*[0 ] m\rangle_{eq } \leq 0 \label{spower4}\end{aligned}\ ] ] which is obviously negative as well as its simplified version . therefore , as expected , one can not extract power from a single heat bath . for a reservoir composed of two equilibrium ones at different temperatures a positive output power @xmath105 can be obtained and in this case the mechanism of self - oscillations works . treating the external perturbation as caused by a coupling of the system to a macroscopic oscillator we see that the positive power is supplied to the oscillator increasing the amplitude of its oscillation ( positive feed - back ) up to the moment when the net energy supply from the reservoir is compensated by the load attached to the oscillator . a solar cell is an engine which produces work from heat exchanged with a non - equilibrium bath . the bath consists of the photonic non - equilibrium reservoir characterized by the local state population @xmath106 $ ] and the basically phononic heat bath at the temperature @xmath100 of the device . the typical semiconductor solar cell consists of a moderately doped p - type absorber , on both sides of which a highly doped layer is formed , n - type on the top side and p - type on the back side , respectively . the electronic states in the valence band and in the conduction band are labeled by the index @xmath107 which corresponds to the quasi - momentum @xmath108 ( spin can be easily added ) with the energies @xmath109 and @xmath110 . we assume a direct band structure with vertical optical transitions which preserve quasi - momentum ( see fig . [ fig : figure4 ] ) . a. fast intraband thermalization processes mediated by phonons and described by @xmath111 . b. optical transitions between the valence and conduction band , which create or annihilate electron - hole pairs , described by @xmath112 , c. non - radiative electron - hole recombination which is neglected in our idealized model . + the fundamental question in the presented approach to work generation in solar cell is the origin of periodic oscillations which can be seen as classical . the frequency @xmath113 is assumed in our derivations to be much smaller than the frequency @xmath114 , but much larger than the recombination rate in order to justify . the only phenomenon which satisfies all these requirements is plasma oscillation visible for p - n junctions . their appearance is due to the fact that a p - n junction creates an interface between regions of different electron concentrations which can oscillate in space producing collective macroscopic electric field oscillations . in several experiments such oscillations have been observed @xcite , with typical frequencies @xmath115 , much lower than @xmath116 corresponding to the energy gap @xmath117 . on the other hand @xmath113 is much higher than the recombination rate @xmath118 what justifies the transition from to . in the eq .. for small @xmath83 plasma self - oscillation is described by the quantum harmonic oscillator coherently pumped by the feed - back mechanism . due to the asymmetric diode - type potential at the edges , charge oscillations are converted into a dc current . ] in the final step of cell operation the thz plasma oscillations must be converted into a direct current . a qualitative picture of this mechanism is shown on fig . [ fig : figure5 ] . the plasmonic degree of freedom is represented by the quantum levels in the asymmetric potential which is harmonic for lower energies . asymmetry is due to the p - n junction which defines an easy " direction for the carrier flow ( to the left ) . the work supplied to the oscillator drives the unidirectional electric current . + the electrons in a semiconductor occupying the conduction and valence bands are described by the annihilation and creation operators @xmath119 , @xmath120 and @xmath121 , @xmath122 , respectively , subject to canonical anticommutation relations . the unperturbed hamiltonian reads @xmath123 in the p - n junction a non - homogeneous free carrier distribution created in a self - consistent build - in potential can be perturbed producing collective plasma oscillations with the frequency @xmath113 . the associated time - dependent perturbation added to the electronic hamiltonian has a mean - field form ( @xmath124 - number of atoms in the sample ) @xmath125 where @xmath83 is a small dimensionless parameter describing the magnitude of deformation , @xmath126 is the relevant energy scale , and @xmath127 , @xmath128 are number operators of free electrons and holes , respectively . to apply the formulas derived in the previous sections we notice first that the driving perturbation depends only on the total numbers of both types of carriers and hence does not interfere with intraband transitions . therefore , the relevant bohr frequency is associated with the gap @xmath126 yielding the time scale @xmath129 , much faster than the modulation period @xmath130 this justifies the adiabatic approximation . the weak coupling assumption is for sure satisfied for very slow radiation recombination processes . among the basic irreversible processes the intraband thermal relaxation is the fastest ( thermalization time @xmath131 ) and therefore , the stationary state of the electronic systems with the total hamiltonian @xmath81 is , within a reasonable approximation , a product of grand canonical ensembles for electrons in conduction and valence band with the same temperature @xmath100 of the device and different electro - chemical potentials @xmath132 and @xmath133 , respectively . the associated density matrix has form @xmath134 = & \frac{1}{z[\xi ] } \exp\left\ { -\frac{1}{k_b t } \sum_{{\mathbf{k } } } \left [ \left(e_c({\mathbf{k}})+ \frac{\xi e_g}{\sqrt{n } } -\mu_c\right ) c^{\dagger}_{{\mathbf{k}}}c_{{\mathbf{k } } } \right . \nonumber \\ & \left . \left . - \left(e_v({\mathbf{k } } ) - \frac{\xi e_g}{\sqrt{n}}-\mu_v\right ) v_{{\mathbf{k}}}v^{\dagger}_{{\mathbf{k}}}\right]\right\ } . \label{grand}\end{aligned}\ ] ] the electro - chemical potentials are determined by the numbers of carriers and hence by doping and radiative and non - radiative processes of electron - hole creation and recombination . because the intraband thermalization to the ambient temperature @xmath100 does not change the number of free electrons and holes , i.e. @xmath135 , the generator @xmath136 does not enter the formula for power . here , one can doubt whether for such fast relaxation the weak coupling condition and hence the validity of the markovian approximation leading to @xmath111 holds . however , intraband relaxation does not contribute to work generation but only determines the structure of the stationary state . the form of this state expressed in terms of fermi - dirac distributions is generally accepted in the literature @xcite and the accuracy of the markovian approximation for the thermalization process is not very relevant . finally , the contribution which remains in the eq . describes the quasi - momentum preserving ( vertical ) transitions and reads @xmath137 \rho & = \frac{1}{2 } \sum_{{\mathbf{k}}}\bigl\{\gamma_{rec}({\mathbf{k}})\bigl ( [ c_{{\mathbf{k } } } v_{{\mathbf{k}}}^{\dagger } , \rho \,v_{{\mathbf{k } } } c_{{\mathbf{k}}}^{\dagger } ] + [ c_{{\mathbf{k } } } v_{{\mathbf{k}}}^{\dagger}\ , \rho , v_{{\mathbf{k } } } c_{{\mathbf{k}}}^{\dagger } ] \bigr ) \nonumber \\ & + \gamma_{ex}({\mathbf{k } } ) \bigl ( [ c_{{\mathbf{k}}}^{\dagger } v_{{\mathbf{k } } } , \rho \ , v_{{\mathbf{k}}}^{\dagger } c_{{\mathbf{k } } } ] + [ c_{{\mathbf{k}}}^{\dagger } v_{{\mathbf{k}}}\ , \rho , v_{{\mathbf{k}}}^{\dagger } c_{{\mathbf{k } } } ] \bigr)\bigr\ } , \label{me_solar } \end{aligned}\ ] ] @xmath138 , \gamma_{ex}({\mathbf{k } } ) = \frac{1}{\tau_{se } } n ( \omega_{{\mathbf{k}}})\ ] ] where @xmath139 is the spontaneous emission time , @xmath140 , and @xmath141 denotes a number of photons occupying a state with the frequency @xmath142 . one can insert all elements computed in the previous section into the expression for power . then we use the properties of the quasi - free ( fermionic gaussian ) stationary state which allow to reduce the averages of even products of annihilation and creation fermionic operators into sums of products of the fermi - dirac distribution functions @xmath143 with @xmath144 . the leading order contribution to power possesses a following form @xmath145 f_{v}({\mathbf{k } } ) \\ & - \gamma_{rec}({\mathbf{k } } ) \bigl [ 1 -f_{v}({\mathbf{k}})\bigr ] f_{c}({\mathbf{k}})\bigr ) , \nonumber \label{power3a}\end{aligned}\ ] ] where @xmath146 $ ] is the total number of free charge carriers . one can introduce the local temperature of light @xmath147 $ ] defined by @xmath148 } = \frac{n ( \omega)}{1 + n ( \omega ) } . \label{local_temp}\ ] ] for the incident sunlight on earth a rough approximation holds @xmath149 where @xmath150 is the temperature of the sun surface and @xmath151 ^ 2\simeq 2\times 10^{-5 } $ ] is the geometrical factor ( @xmath152 - sun radius ) which takes into account the photon density reduction at large distance from the source . in particular , for the typical value of the energy gap @xmath153 the effective temperature of sunlight @xmath154 \simeq 1000 k $ ] . because the product of population numbers for free carriers given by @xmath155 f_{c}({\mathbf{k}})$ ] is essentially concentrated on the interval @xmath156 $ ] , and @xmath157\simeq 1000 k \ll \hbar\omega_0/k_b \simeq 12 000 k$ ] , the expression can be approximated by @xmath158}\bigr]\hbar\omega_0 - e\phi\bigl)\bigr\ } -1\bigr)\ ] ] where @xmath159\bigl [ 1 -f_{v}({\mathbf{k}})\bigr ] f_{c}({\mathbf{k } } ) > 0 $ ] is independent of the size of a cell , and the _ voltage _ @xmath1 is identified with the difference of electro - chemical potentials , i.e. @xmath160 . + the condition for work generation by the solar cell reads @xmath161 } \label{positive_work}\ ] ] what implies that @xmath162 is an _ open - circuit voltage _ of the cell for the idealized case @xcite . the presence of the carnot factor @xmath163 suggests also the interpretation of the eq . in terms of thermodynamical efficiency . indeed , the incident photon of the frequency @xmath164 produces an excitation of the energy close to @xmath126 in the process of electron - hole creation followed by the fast thermalization of an electron to the bottom of the conduction band , and a hole to the top of the valence one . then , a part of energy @xmath165 is transformed into useful work , equal at most @xmath166 per single electron flowing in the external circuit . the maximal efficiency @xmath167 under the conditions that each photon with the energy higher than the gap produces an electron - hole pair and non - radiative recombination processes are neglected , is given by the product @xmath168 where @xmath169 is the so - called _ ultimate efficiency _ computed under the assumptions : + a ) _ ... photons with energy greater than @xmath126 produce precisely the same effect as photons of energy @xmath126 , while photons of lower energy will produce no effect " _ @xcite , + b ) the whole @xmath126 is transformed into work . under standard illumination conditions the ultimate efficiency of a solar cell can reach @xmath170 and the carnot factor is about @xmath171 what yields @xmath172 - the shockley s _ detailed balance limit _ @xcite . actually , photons are absorbed along their path in the absorber and @xmath173 decays exponentially with the penetration distance . taking a more realistic average value @xmath174 + t)/2 \simeq 650 k$ ] one obtains @xmath175 , much closer to the real performance of standard @xmath176 solar cells . the presented model based on the idea of self - oscillations explains the dynamical origin of work generation in photovoltaic cells which is not present in the standard static " picture . the main new ingredient is the role of plasma oscillation as a piston " which transforms the steady heat input from the photon flux into periodic motion . this model provides a bridge between the theory of driven quantum open systems applied to heat engines and the theory of photovoltaic devices . the formulas and explain in a simple way the meaning of the light temperature " , the carnot bound , and the linear relation between the open circuit voltage and the device temperature . the experimental verification of this model should provide the evidence of thz plasma oscillation in the device with the amplitude square proportional to the power output . such oscillations produce a weak thz radiation which , in principle , could be detected . the similar ideas can be applied to other types of heat engines with hidden self - oscillations " . it seems that thermoelectric devices based either on bimetallic or semiconductor p - n junctions can be described by the very similar models . plasma oscillation remains a piston and sunlight is replaced by the hot bath . for organic photovoltaic systems , proton pumps or photosynthesis there exists quite strong evidence of the important role of coherent molecular oscillations played in the energy and charge transfer ( see e.g. @xcite ) . it is plausible that those oscillations can play the role of a piston in the work extraction mechanism as well .
solar cells are engines converting energy supplied by the photon flux into work . all known types of macroscopic engines and turbines are also self - oscillating systems which yield a periodic motion at the expense of a usually non - periodic source of energy . the very definition of work in the formalism of quantum open systems suggests the hypothesis that the oscillating piston " is a necessary ingredient of the work extraction process . this aspect of solar cell operation is absent in the existing descriptions and the main goal of this paper is to show that plasma oscillations provide the physical implementation of a piston .
often in astronomical research , absolute photometric accuracy better than @xmath17 , is required . in many cases , the method of choice is to observe photometric standards ( e.g. , landolt 1992 ) . however , this requires photometric observing conditions , and additional observations . the sloan digital sky survey ( sdss ; york et al . 2000 ) provides an excellent photometric calibration in @xmath18 bands , with accuracy better than @xmath19 ( adelman - mccarthy et al . 2008 ) . however , this is available only for about a quarter of the sky . other all - sky visible - light catalogue , like the usno - b1 ( monet et al . 2003 ) and the us naval observatory ccd astrograph catalogue ( zacharias et al . 2004 ) provide relatively poor photometric accuracy . the magnitudes of individual stars in these catalogue are accurate to about @xmath20mag , and even with a large number of stars there is still a considerable field - to - field systematic errors . in this paper , i calculate the sdss @xmath6 magnitudes of tycho-2 stars over the entire sky . in case @xmath21 of these stars are visible in a camera field of view , these stars can be used to photometrically calibrate an astronomical image to accuracy of better than @xmath22mag . the only overhead is that typically a shorter exposure in which the tycho-2 catalog stars will not be saturated is required . from the tycho-2 catalog ( hog et al . 2000 ) , i selected all stars with @xmath23 mag and @xmath24 mag . this choice was made in order to remove stars with magnitude uncertainties @xmath25 mag . i cross correlated this list with the 2mass catalog ( skrutskie et al . 2006 ) , and selected only stars which have a single 2mass match within @xmath26 . the final list contains 1,560,980 stars . next , i fitted the tycho-2/2mass magnitudes of these stars with synthetic photometry , in the @xmath27 bands , of 131 stellar spectral templates ( pickles 1998 ) . the synthetic photometry was calculated using the code of poznanski et al . ( 2002 ) . for each star i preformed two types of fits : ( i ) a @xmath28 fit ; ( ii ) a least square , of residuals , fit ignoring the tycho-2 and 2mass photometric errors . i note that the spectral templates were not corrected for galactic extinction . a table listing the 1.56 million stars , along with their coordinates , observed @xmath27 magnitudes , fitted @xmath6 magnitudes for both kinds of fits , and corresponding root - mean - square ( rms ) and @xmath28 , is available in the electronic version of this paper , and via the vizier service . in table [ tab - catdesc ] i describe the content of this table . the third column in the catalog is a flag indicating if the star is recommended for use as a standard star ( i.e. , flag@xmath29 ) . these stars satisfy all of the following criteria : @xmath30mag and @xmath31mag and @xmath32mag ; error(@xmath1)@xmath33mag ; error(@xmath2)@xmath33mag ; and rms of the best rms - fit template smaller than 0.15mag . there are 992,235 stars with flag@xmath29 . lll 1 & right ascension j2000.0 & deg + 2 & declination j2000.0 & deg + 3 & flag for good standards & + 4 & @xmath1 magnitude & mag + 5 & @xmath1 magnitude error & mag + 6 & @xmath2 magnitude & mag + 7 & @xmath2 magnitude error & mag + 8 & @xmath3 magnitude & mag + 9 & @xmath3 magnitude error & mag + 10 & @xmath4 magnitude & mag + 11 & @xmath4 magnitude error & mag + 12 & @xmath5 magnitude & mag + 13 & @xmath5 magnitude error & mag + 14 & best rms - fit template & + 15 & rms for the best rms - fit template & mag + 16 & rms - fit @xmath12-band magnitude & mag + 17 & rms - fit @xmath13-band magnitude & mag + 18 & rms - fit @xmath14-band magnitude & mag + 19 & rms - fit @xmath15-band magnitude & mag + 20 & best @xmath28-fit template & + 21 & @xmath28 for the best @xmath28-fit template ( d.o.f.=4 ) & + 22 & @xmath28-fit @xmath12-band magnitude & + 23 & @xmath28-fit @xmath13-band magnitude & + 24 & @xmath28-fit @xmath14-band magnitude & + 25 & @xmath28-fit @xmath15-band magnitude & [ tab - catdesc ] the constructed catalog star density as a function of galactic latitude is shown in fig . [ fig : tyc2_starden ] . ( i.e. , good for photometric calibration ) . [ fig : tyc2_starden],width=321 ] for flag@xmath29 stars , the sky density ranges from 55deg@xmath34 at the galactic plane to about 10deg@xmath34 at the galactic poles . in order to test the accuracy of the derived synthetic magnitudes , i constructed a catalog of all the photometric measurements available from the sdss photometric telescope `` secondary patches '' fields ( tucker et al . 2006 ) . from this catalog i selected all the non - saturated stars brighter than 12th magnitude in both the @xmath12- and @xmath13-bands . davenport et al . ( 2007 ) analyzed the systematic offset between the sdss magnitude system and the sdss photometric telescope bands . they found that the magnitudes of very red stars are different in the two systems . using the transformations given by davenport et al . ( 2007 ) i converted all the sdss photometric - telescope magnitudes to the sdss system . next i cross - correlated this list with the catalog of @xmath6 synthetic magnitudes presented in [ cat ] , and selected stars which have @xmath35 magnitudes fainter than 6 ( i.e. , not saturated ) ; @xmath1 and @xmath2 magnitude errors smaller than 0.15 ; and rms of the best rms - fit template of less than 0.15mag . for each of the 3714 stars satisfying these criteria , i compared the corrected magnitudes ( i.e. , davenport et al . 2007 ) as measured by the sdss photometric telescope with its best - fit synthetic magnitude . in figure [ fig : mag_sdss_vs_fit ] , i show histograms of the fitted synthetic magnitudes minus the sdss magnitudes for the @xmath6-bands . the median value , along with half the range containing @xmath36 of the stars , are indicated in each panel . the upper row is for the best - fit magnitude by minimizing the rms , while the lower row is for the @xmath28 fit . this plot suggests that the @xmath28 fit is marginally better . therefore , it should be preferred over the rms - fit . in all the bands there are small offsets , listed in fig . [ fig : mag_sdss_vs_fit ] , between the sdss photometric telescope magnitudes and the derived synthetic magnitudes . the magnitudes in the catalog ( table [ tab - catdesc ] ) are corrected for these small offsets . i note that i repeated this test using 42 sdss photometric standardsdtucker / ugriz / tab08.dat ] ( smith et al . 2002 ) , and found similar results . the field of view of large format cameras may contain only several suitable tycho-2 stars ( see fig . [ fig : tyc2_starden ] ) . to estimate the uncertainty in magnitude calibration when using several stars , i have carried out the following simulations : i randomly selected @xmath37 stars , for any @xmath37 between 2 and 100 , out of the 3714 sdss standard stars that i used for the comparison of the derived magnitudes with the actual sdss magnitudes . for each set of randomly selected @xmath37 stars i calculated the median of differences between the @xmath28-fitted synthetic magnitudes and the sdss magnitudes . next , i repeated this procedure 10,000 times ( for each @xmath37 ) , and calculated the standard deviation of the 10,000 median of differences . in figure [ fig : error_vs_nstars ] i show the standard deviation of the distribution of the median of differences between the synthetic magnitudes and the measured sdss magnitudes , as a function of @xmath37 . the plot suggests that , for example , by using five stars the errors in photometry reduce to 0.07mag , 0.04mag , 0.03mag and 0.03mag in the @xmath12 , @xmath13 , @xmath14 and @xmath15-bands , respectively . when using ten stars the errors reduce to about 0.04mag , 0.03mag , 0.02mag and 0.02mag in the @xmath12 , @xmath13 , @xmath14 and @xmath15-bands , respectively . in order to remove outliers it is important to use the median of differences ( i.e. , and not mean of differences ) . to summarize , i suggest an alternative method for photometric calibration that may work to accuracy of about several percents . this method relies on selected stars in the tycho-2 catalog for which i fitted spectral templates to their @xmath27 magnitudes . two major limitation of this method is that several tycho-2 stars are needed in the field of view , and that shorter exposures , in which these stars are not saturated , are required . given the catalog star density as a function of galactic latitude ( fig . [ fig : tyc2_starden ] ) , this method is applicable for instruments with large field of view or for low galactic latitude observations . i would like to thank the referee for constructive comments . i thank shri kulkarni for expressing the need for an all - sky catalog of standard stars , and to avishay gal - yam , dovi poznanski , orly gnat , and andrew pickles for valuable discussions . this work is supported in part by grants from nsf and nasa .
photometric calibration to @xmath0 accuracy is frequently needed at arbitrary celestial locations ; however , existing all - sky astronomical catalogs do not reach this accuracy and time consuming photometric calibration procedures are required . i fit the hipparcos @xmath1 , and @xmath2 magnitudes , along with the 2mass @xmath3 , @xmath4 , and @xmath5 magnitudes of tycho-2 catalog - stars with stellar spectral templates . from the best fit spectral template derived for each star , i calculate the synthetic sdss @xmath6 magnitudes and constructed an all - sky catalog of @xmath6 magnitudes for bright stars ( @xmath7 ) . testing this method on sdss photometric telescope observations , i find that the photometric accuracy , for a single star , is usually about @xmath8 , @xmath8 , @xmath9 and @xmath10mag ( 1@xmath11 ) , for the @xmath12 , @xmath13 , @xmath14 , and @xmath15-bands , respectively . however , by using @xmath16 such stars , the typical errors per calibrated field ( systematic + statistical ) can be reduced to about 0.04 , 0.03 , 0.02 , and 0.02mag , in the @xmath12 , @xmath13 , @xmath14 , and @xmath15-bands , respectively . therefore , in cases for which several calibration stars can be observed in the field of view of an instrument , accurate photometric calibration is possible .
the principal role of chaperones is to assist in the resolution of the multitude of alternative misfolded structures that rna readily adopts so that sufficient yield of the native material is realized in biologically viable time less than @xmath1 . because spontaneous yield of the native state of large ribozymes even at high mg@xmath18 concentrations is small @xcite , it is likely that _ in vivo _ rna chaperones are required to boost the probability of reaching the folded state within @xmath1 . unlike the well - studied bacterial groel - groes , a well - identified `` one - fit - all '' chaperonin system for processing cytosolic proteins @xcite , protein - cofactors that act as rna chaperones vary from one rna to the other @xcite . based on a number of experiments ( see @xcite for reviews ) we classify the client rna molecules into two classes depending on the need for the rna chaperones to utilize the free energy of atp hydrolysis in facilitating folding . 1 . folding of class i rna molecules is greatly aided by the interactions with protein cofactors while their assistance may not be strictly required . these rna molecules are not stringent rna substrates . for example , the splicing reaction of mitochondrial bi5 group i intron is activated in 50 mm or greater mg@xmath19 concentration at room temperature but interactions with cytochrome b pre - mrna processing protein 2 ( cbp2 ) or _ neurospora crassa _ mitochondrial tyrosyl trna synthetase ( cyt-18 ) enables splicing at physiological level ( @xmath57 mm ) of mg@xmath19 by enhancing folding of the bi5 core . tetrahymena _ ribozyme and other group i introns belong to stringent class ii substrates . spontaneous folding , even at high counterion concentration , occurs too slowly with low yield of the native state to be biologically viable . at high temperatures folding of the misfolded _ tetrahymena _ ribozyme is aided by formation of ribonucleoprotein ( rnp ) assembly with the promiscuously interacting cyt-18 , which in essence follows the mechanism of passive assistance @xcite . however , an atp - dependent helicase activity associated cyt-19 produces functionally competent states that can splice efficiently at normal growth temperature . although not firmly established , it is suspected that rna chaperones bind to single stranded regions of the misfolded structures , which upon release places the rna in a different region of the folding landscape , giving it a new opportunity to fold just as anticipated by the iterative annealing model ( iam ) @xcite . the fundamental difference between class i and class ii rna substrates is in the apparent time scale of catalysis ( @xmath20 ) by the ribonucleprotein ( rnp ) complex formed between rna and the rna chaperone . if this time scale is smaller than @xmath1 ( @xmath21 ) , the formation of rnp alone is sufficient to produce functionally competent rna molecules . in the opposite case ( @xmath22 ) , the conversion of misfolded rna into folding competent form needs assistance from a specially designed action of rna chaperone that can transduce the free energy of atp hydrolysis . below we will describe a mathematical model for the two scenarios . the tertiary structure capture model @xcite accounts for the passive action of rna chaperones in the folding of mitochondrial bi5 group i intron without atp . explicit mechanisms of recognition by passive rna chaperones by the collapsed rnas may differ for different systems , and might also depend on whether the rna collapse is specific or non - specific . in a majority of cases ribozymes undergo an extended to a collapsed transition even at modest ion concentration producing a heterogeneous population of compact structures whose affinity for the protein cofactors could vary greatly . for example , cbp2 could bind to these compact structures with partially folded cores ( p5-p4-p6 and p3-p7-p8 ) of group i intron ( fig.[fig : passive_rnp]a ) with differing specificity , and promote the subsequent assembly of 5 domain of bi5 core . in contrast , cyt-18 binds to rna and forms a stable cyt-18-bi5 complex at an early stage of rna folding and promotes the splicing competent states @xcite . if the association between the cofactor and compact rna is too weak then large conformational fluctuations can produce long - lived entropically stabilized metastable kinetic traps for rna . in this case , the protein cofactor would have little effect on rna folding . in the opposite limit , when the cofactor interacts strongly with collapsed rna , transient unfolding in the rna conformations , which are needed for resolving misfolded structures to the native state , would be prohibited . thus , for the chaperone - assisted folding of class i rna substrates , an optimal stability of the rna - cofactor intermediate is needed to efficiently produce an assembly - competent rnp complex . the physical picture of passive assistance of rna chaperones described above , encapsulated in the weeks - cech tertiary capture mechanism , can be translated into the kinetic scheme shown in fig.[fig : passive_rnp ] . after rna collapses rapidly to an ensemble of collapsed intermediate structures \{c } ( @xmath23 ) consisting of a mixture of specifically and non - specifically collapsed structures , promiscuous binding of chaperone ( blue spheres in [ fig : passive_rnp ] ) to the conformations in \{c } produces a fluctuating ensemble of tightly and loosely bound intermediate rnp complex @xmath24 . this process is conceptually similar to the encounter complex in protein - protein interaction @xcite . only a fraction ( @xmath0 ) of states among the tightly bound ensemble of rnp , @xmath25 , is viable for producing functionally competent rnp state . thus , @xmath25 is partitioned roughly into @xmath26 , where @xmath27 denotes the intermediate ensemble that can fold into the competent rnp while @xmath28 can not . since transitions among the states in @xmath25 is non - permissible on viable time scales , the only way for a molecule trapped in @xmath28 to reach the competent rnp state is to visit a transiently unbound ( or loosely bound ) intermediate ensemble @xmath29 and explore the states belonging to @xmath27 . once the rna is in \{@xmath30 } ensemble , the rate of rnp formation is given by @xmath31 , \end{aligned}\ ] ] which can be quantified by assuming steady state production of @xmath32 $ ] , i.e. , @xmath33/dt\approx 0 = -(k_{cat}+k_{ub})[i_b^{cat}]_{ss}+k_b\phi_f[i_{ub}]_{ss}$ ] . defining a constant for rapid pre - equilibration ( @xmath34 , @xmath35 ) between the two collapsed intermediate ensembles @xmath36_{ss}/[i_b]_{ss}=k_{ub}/k_b$ ] with a total concentration of collapsed intermediate state @xmath37+[i_b]$ ] , we obtain the rate of rnp formation at steady state : @xmath38 a change in the strength of binding between rna and protein cofactor would affect the values of @xmath39 by modulating @xmath40 or the stability of @xmath25 ensemble , while keeping other rate constants ( @xmath41 and @xmath42 ) unchanged ( the inset of fig.[fig : passive_rnp ] ) . it can be argued that there be an optimal stability for @xmath25 in order to maximize the rate of rnp formation . if @xmath25 is too stable compared with @xmath43 then @xmath25 becomes a dead end with negligible probability of reshuffling its population into non - productive ensemble of @xmath28 into @xmath27 through conformational fluctuations . in contrast , if @xmath43 is more stable than @xmath25 the production of competent rnp would be inefficient . it is clear from eq.[eqn : rss ] that the limiting condition of @xmath44 , either @xmath45 or @xmath46 , leads to a vanishing value of @xmath47 ; hence it follows that there is an optimum value @xmath48 that maximizes the rate of rnp production . the maximum rate is obtained using @xmath49 : @xmath50 where @xmath51 . the presence of @xmath52 that maximizes @xmath47 is indicative of an optimal unbinding rate ( @xmath53 ) , for the formation of competent rnp , that satisfies @xmath54 . as long as @xmath55 remains less than the biologically viable time scale @xmath1 ( @xmath56 ) , rna chaperone promotes rna molecule to reach the functionally competent form by merely providing a suitable molecular interface on which rna could interact and anneal its conformation . physically , this situation is not that dissimilar to the role mini chaperone ( apical domain of groel ) plays in annealing certain non - stringent substrates @xcite . if sufficient yield of the folded rna is not realized on the time scale @xmath1 , i.e. , @xmath57 ( the partition factor , @xmath0 , in the kpm is small ) , then a more active role including atp consumption is required to resolve the misfolded states . for kinetically trapped misfolded rna molecules , transient unfolding of misfolded elements by rna chaperones is needed to increase the yield of rna since it provides another chance for refolding into a functional state . in experiments involving cyt-18 and the dead - box protein cyt-19 on _ neurospora crassa _ group i intron @xcite , it was shown that the two protein cofactors ( cyt-18 and cyt-19 ) work in a coordinated fashion by utilizing atp hydrolysis . atp - dependent activity of cyt-19 was required for efficient splicing at the normal growth temperature ( 25 @xmath58c ) while cyt-18 alone could rescue the misfolded rna at high temperatures . in this sense , the active participation shares features of many biological processes including motility of molecular motors @xcite , and steps in signal transduction pathways @xcite . in the absence of rna chaperone , the kpm predicts that the initial pool of unfolded rna ribozymes are partitioned into folded and misfolded conformations , described by the following set of rate equations @xcite . @xmath59 for a given ribozyme concentration @xmath60+[m]+[n]$ ] , a fraction @xmath0 of ribozyme folds into the native state directly at a rate @xmath61 , and the remaining fraction ( @xmath62 ) is misfolded upon undergoing non - specific collapse transition . the solution of eq.[eqn : basic ] , in terms of the probability of not being folded , is given by @xmath63 , which follows from the kpm @xcite . in the presence of chaperone ( @xmath64 ) , an additional set of equations involving the actions of chaperone on the different species should be included in eq.[eqn : basic ] : @xmath65 here , @xmath66 is the rate associated with the capture process of the misfolded rna by the rna chaperone producing the rnp , @xmath67 ; @xmath68 is the rate of the release process . note that rna captured by the chaperone is released in the form of unfolded state , which provides rna with an another chance to fold @xcite . the corresponding rate equations for eqs.[eqn : basic ] and [ eqn : chaperone ] are : @xmath69}{dt}&=-k_f[u]+k_r[cs]\nonumber\\ \frac{d[n]}{dt}&=\phi_f k_f[u]+k_s[m]\nonumber\\ \frac{d[cs]}{dt}&=\lambda_m[m][c]-k_r[cs ] \label{eqn : rate}\end{aligned}\ ] ] with @xmath60+[n]+[m]$ ] and @xmath70+[cs]$ ] . in the context of chaperone - assisted rna folding , @xmath71 . fig.[fig : chaperone_action]a displays a numerical solution for each chemical species ( @xmath73 , @xmath74 , @xmath75 , @xmath64 , @xmath67 ) starting from an initial condition of @xmath76=0 $ ] , @xmath77=1 $ ] , @xmath78=0.1 $ ] . note that all the states in misfolded ensemble are converted to the native state through the reactions @xmath79 , followed by @xmath80{k_f}n$ ] , resulting in 100 % yield of the native state if the chaperone cycle is iterated multiple times . if there are @xmath81 iterations , the yield of the native state ( @xmath82 ) becomes @xmath83 ; thus the population of the folded state will increase as @xmath84 where @xmath85 is the atp concentration dependent cycling time of the chaperone . for the case of group i intron , with @xmath86 @xcite , the cycle should be iterated @xmath87 times to obtain the yield of more than 90 % ( @xmath88 ) . in the iam for folding of proteins it is assumed that groel does not recognize the folded state of proteins @xcite because the recognition sites are sequestered in the folded state . however , bhaskaran and russell @xcite have reported that cyt19 can unfold ( at least partially ) the native state of group i intron . the plausible lines of evidence that cyt-19 can interact with the native ribozymes and destabilize it @xcite are : ( i ) although cyt-19 does not significantly alter the cleavage activity of group i intron ribozymes below 5 mm - mg@xmath19 , the ribozymes incubated with cyt19 under less stabilizing conditions of 1 mm mg@xmath19 are accompanied by a reduction in the cleavage efficiency . ( ii ) a finite steady state values of native and misfolded ribozymes are reached in the presence of cyt19 regardless of their initial population . ( iii ) refolding of ribozymes kinetically trapped in the cbas , unfolded by cyt-19 from both native and misfolded species , follows the same pathway that predominates in the absence of cyt-19 . ( iv ) unfolding efficiency of cyt-19 depends on the stability of rna . as indicated by the cyt-19 data @xcite , rna chaperone can interact with both native and misfolded states under certain conditions and promote their folding by unfolding ( at least partially ) them . incorporation of this finding in our model requires modification of eq.[eqn : rate ] along the following lines : @xmath69}{dt}&=-k_f[u]+k_r[cs]\nonumber\\ \frac{d[n]}{dt}&=\phi_f k_f[u]-\lambda_n[c][n]\nonumber\\ \frac{d[cs]}{dt}&=(\lambda_n[n]+\lambda_m[m])[c]-k_r[cs ] \label{eqn : general_rate}\end{aligned}\ ] ] with @xmath60+[n]+[m]$ ] and @xmath70+[cs]$ ] . the terms involving @xmath89 account for the recognition of the native state by chaperones ( @xmath90 ) . typically , @xmath91 , implying that the stability of rna chaperone in complex with the native state is less than the complex between misfolded states and the chaperones . of particular note is that the numerical solution of eq.[eqn : general_rate ] from the same initial condition used in fig.[fig : chaperone_action]a shows different behavior . as shown in fig.[fig : chaperone_action]b , @xmath73 and @xmath74 reach steady state values with @xmath76\neq 1 $ ] and @xmath92\neq 0 $ ] . for a given set of parameters with @xmath93 , the steady state values are : @xmath94_{ss}&=\frac{k_r}{k_f}[cs]_{ss}\nonumber\\ [ n]_{ss}&=\frac{\phi_f k_f}{\lambda_n}\times\frac{[cs]_{ss}}{(c_0-[cs]_{ss})}\nonumber\\ [ m]_{ss}&=\frac{(1-\phi_f)k_r}{\lambda_m}\times\frac{[cs]_{ss}}{(c_0-[cs]_{ss})}. \label{eqn : steadystate}\end{aligned}\ ] ] the deterministic kinetic schemes presented above do not account for the population of discrete misfolded states explicitly . this can be accomplished by casting the rna chaperone activity using a chemical master equation ( cme ) formalism , which implicitly accounts for fluctuation effects due to noise . the corresponding chemical langevin equation can be derived from the cme under certain approximations @xcite . within the cme inclusion of a finite probability of leakage of flux from the native state facilitated by rna chaperones requires a generalization of a formalism developed by previously @xcite to quantitatively predict groel - assisted folding of proteins . in the previous study it was assumed that groel does not recognize the folded state , which does not apply to rna chaperones . in what follows we include this possibility explicitly within the cme formalism . let us assume that the structural ensemble of rna can exist in a number of discrete free energy states @xmath95 @xmath96 . we assume that @xmath97 is native and all other states with @xmath98 are non - native or misfolded corresponding to the cbas . the occupation probability of each state @xmath99 obeys the following master equation : @xmath100 \(i ) to simplify the above master equation greatly , we formulate the cme for the native state as @xmath101 assuming that the folding transition rates from the @xmath81 misfolded states to the native state are all similar leads to @xmath102 ; and the transition from the native to misfolded states is @xmath103 . with @xmath104 , the steady state solution of eq.[eqn : simple_master ] is given by @xmath105 . \label{eqn : ss_solution}\ ] ] note that when rna chaperone recognizes the native state 100 % yield of the folded state is not achieved even at @xmath106 and @xmath107 . at an irreversible limit @xmath108 , which is typically assumed in groel assisted protein folding , the yield of native state increases as @xmath109 , consistent with the experimental finding @xcite . \(ii ) to develop a more sophisticated solution , we assume that detailed balance is satisfied so that @xmath110 with @xmath111 at equilibrium . because the transition rate between two states depends on the height of free energy barrier , it is not possible to uniquely determine the transition rate from the free energy of each state alone . nevertheless , for concreteness we choose " @xmath112 as @xmath113}. \label{eqn : detailed_balance}\end{aligned}\ ] ] in eq.[eqn : detailed_balance ] @xmath114 sets the timescale of the problem . the master equation for a state @xmath115 can be written as @xmath116 by inserting eq.[eqn : detailed_balance ] in eq.[eqn : other_master ] , we get @xmath117\nonumber\\ & = \tau_0^{-1}\left[a_a^{-1}p_{t}(t)-\lambda a_ap_a(t)\right]\end{aligned}\ ] ] where @xmath118 , @xmath119 , and @xmath120 . the formal solution of this ode is : @xmath121 where @xmath122 , the initial probability for state @xmath115 . in laplace domain , @xmath123 , @xmath124 where @xmath125 . insertion of eq.[eqn : other_laplace ] into @xmath126 and rearrangement with respect to @xmath127 leads to @xmath128 thus , the probability of native state ( @xmath97 ) in laplace domain is @xmath129 the native probability @xmath130 is obtained by the inverse laplace transform : @xmath131 where @xmath132 and @xmath133 are the poles and residues of @xmath134 . the behavior of @xmath130 is determined by the pole structure of @xmath134 and the corresponding residues @xmath133 . the residues change depending on the initial conditions specified by @xmath135 ( @xmath136 ) satisfying @xmath137 . note that @xmath138 is not a pole but a regular point of @xmath134 for all @xmath139 because @xmath140 . @xmath134 has poles at @xmath141 that satisfies @xmath142 . the pole structure from the algebraic solution of @xmath143 can be visualized by plotting @xmath144 . @xmath145 since @xmath146 . in fact , @xmath147 is the largest pole of @xmath134 with its residue @xmath148 where @xmath149 . to be specific , we use a model landscape with multiple minima in fig.3(b ) , and calculate @xmath150 and the time dependent behavior of @xmath130 whose behavior depends on the initial condition @xmath135 ( @xmath151 ) . in fig.3(c ) the poles ( @xmath152 , @xmath153 , @xmath154 ) due to @xmath155 are found at @xmath156 . thus , @xmath157 here note that @xmath158 for all @xmath159 and that @xmath130 converges to @xmath160 independent of any initial condition specified by @xmath135 ( @xmath151 ) . the sign of @xmath161 changes depending on the value of @xmath135 ( see fig.3(d ) ) , indicating that the population of the native state adjusts to the equilibrium value regardless of the initial conditions . the non - unity steady state value @xmath162 of @xmath130 is consistent with eq.[eqn : ss_solution ] , @xmath163 for @xmath164 as well . as long as destabilization of the native state is permitted in the chaperone - assisted folding of rna , complete recovery of the native population can not occur at @xmath106 , a result that is not only consistent with experiments @xcite but also follows from mass action kinetic equations . finally , it is worth noting that the value of @xmath165 is the equilibrium population of the native state ( @xmath166 ) , i.e. , @xmath167 , which is also confirmed to be identical to the value 0.864 using the parameter set in fig.3(b ) . therefore , the iterative action of the rna chaperone results in the annealing " the ensemble of molecules that is kinetically trapped in a multitude of metastable states to reach _ equilibrium _ by assisting the system to overcome otherwise insurmountable high free energy barriers separating the cbas and the nba ( fig.3(a ) ) . only when the stability of native state is far greater than all other metastable intermediates in the cbas can molecular chaperones bring the yield of native state close to the unity . the tendency of large rna molecules to misfold readily _ in vitro _ strongly suggests that rna chaperones must be involved in assisting their folding under cellular conditions . in this paper we discussed two general mechanisms of chaperone assisted rna folding . in both scenarios , interactions with protein cofactors facilitates rna to escape from the kinetic traps , in the process annealing misfolded states into functionally competent folded states . whether the rescue of substrate rnas occurs by passive ( as envisioned in the tertiary capture model ) or active mechanism could depend on the nature of rnas . for the class i rna substrates it suffices that the misfolded conformations in the cbas interact transiently with the protein cofactors . as long as optimal interaction ( not too weak or too strong ) between rna and the protein cofactor is achieved the misfolded states can reach the folded state without requiring atp . on the other hand , for the more stringent class ii substrates atp hydrolysis in the presence of rna chaperones is coupled to conformational changes in rna places the rna molecules in a different region of the folding landscape from which it can fold with probability @xmath0 . by repeating this process multiple times sufficient yield of the native material is generated . we show using mass action kinetics and chemical master equation that as long as the chaperone system does not discriminate between the folded and misfolded states @xcite the total yield of the native fold is less than 100% , which accords well with experimental findings . it is interesting to estimate the work that a typical rna chaperone system could perform on the rna . given that a typical binding free energy between two macromolecules in the cell is @xmath168 @xmath14 , which we estimate by using an estimated lower bound of dissociation constant @xmath169 nm ( calculated based on one bound complex in a _ e. coli _ cell ) , it could be argued that the upper bound of energy or mechanical work stabilizing the rnp complex is @xmath170 @xmath14 . in the active model , rna chaperones can facilitate folding to the native state by performing work on the rna , thus redistributing the population of native and misfolded rnas on the folding landscape @xcite . using the typical dimensions of the misfolded rna ( @xmath171 nm ) the conformational changes in rna should generate a force of @xmath172pn , which is large enough to partially unfold compact rna molecules @xcite . it is intriguing to note that the free energy associated with atp hydrolysis is @xmath173 @xmath14 per one atp molecule , which lies at the borderline of the maximum binding free energy that is associated with typical protein - protein or protein - rna interactions . the individual monomers in muti - subunit molecular chaperones , which have their own catalytic sites for atp hydrolysis , typically form a ring - like structure or act with other cofactors @xcite to further increase the efficiency of free energy transduction by tightly interacting with the target structure and enhancing the generation of mechanical work . it is also worth emphasizing that full conversion of atp hydrolysis free energy to mechanical work is not always realized . for some chaperones functioning through an active mechanism , conformational cycle of chaperone could remain futile due to inefficient coupling to the structures of the substrate molecules . as a result , multiple rounds of chaperone cycle are often needed to convert one misfolded molecule to the folded state . the variation in the efficiency of free energy transduction from one specific molecular chaperone to the other should lead to a rescaling of the number of iterations @xmath81 to @xmath174 where @xmath175 is a machine - dependent efficiency of unwinding ( unfolding ) of the substrate in one chaperone cycle . thus , the yield of the native state ( @xmath82 ) after @xmath81 cycles should be @xmath176 , leading to @xmath177 , and the time scale for complete annealing of misfolded biomolecules due to a single molecular chaperone is estimated to be @xmath178 . theses estimates can be experimentally verified provided the stoichiometry of atp consumption in the active rna chaperone machinery is measured . we thank the korea institute for advanced study for providing computing resources . dt acknowledges support from the national science foundation through grant che 09 - 14033 . and @xmath43 . @xmath43 is an ensemble of transiently unbound intermediate . binding and transient release anneal the misfolded rna , leading to the assembly competent rnp containing the folded rna . the final step is the release of the native rna . [ fig : passive_rnp ] , width=480 ] _ t$ ] , @xmath92_t$ ] , @xmath77_t$ ] , @xmath78_t$ ] and @xmath179_t$ ] ( a ) when chaperone does not recognizes the native state , @xmath180 @xmath181 and ( b ) when chaperone recognizes the native state , @xmath182 @xmath181 with @xmath183 , @xmath184 , @xmath76_0=0 $ ] , @xmath78_0=c_0 $ ] , @xmath185 , @xmath186 @xmath181 , @xmath187 @xmath181 , @xmath188 @xmath181 . [ fig : chaperone_action],width=384 ] ) and 5 misfolded states corresponding to the cbas in ( a ) ( @xmath189 ) . the free energy value of each state is assigned as @xmath190 , @xmath191 , @xmath192 , @xmath193 , @xmath194 , and @xmath195 , which leads to @xmath196 . ( c ) the plot of @xmath197 ( black solid lines ) , graphically showing the solutions of @xmath198 , i.e. , the pole structure of @xmath134 due to @xmath155 : @xmath147 , @xmath199 , @xmath200 , @xmath201 , @xmath202 , @xmath203 . ( d ) the time evolution of the fraction of native state ( @xmath130 ) with different initial conditions ( i ) @xmath204 , @xmath205 ( black , solid line ) ( ii ) @xmath206 , @xmath207 ( red , dashed line ) ( iii ) @xmath208 , @xmath209 , @xmath210 , @xmath211 ( blue , dot - dashed line ) . note that the fraction of native state in the steady state is @xmath212 due to the flux out of native to non - native states . the steady state value is independent of the initial conditions , suggesting that the rna chaperones redistribute the population of folded and misfolded states till equilibrium is reached . [ fig : ot ] , width=499 ]
as a consequence of the rugged landscape of rna molecules their folding is described by the kinetic partitioning mechanism according to which only a small fraction ( @xmath0 ) reaches the folded state while the remaining fraction of molecules is kinetically trapped in misfolded intermediates . the transition from the misfolded states to the native state can far exceed biologically relevant time . thus , rna folding _ in vivo _ is often aided by protein cofactors , called rna chaperones , that can rescue rnas from a multitude of misfolded structures . we consider two models , based on chemical kinetics and chemical master equation , for describing assisted folding . in the passive model , applicable for class i substrates , transient interactions of misfolded structures with rna chaperones alone are sufficient to destabilize the misfolded structures , thus entropically lowering the barrier to folding . for this mechanism to be efficient the intermediate ribonucleoprotein ( rnp ) complex between collapsed rna and protein cofactor should have optimal stability . we also introduce an active model ( suitable for stringent substrates with small @xmath0 ) , which accounts for the recent experimental findings on the action of cyt-19 on the group i intron ribozyme , showing that rna chaperones does not discriminate between the misfolded and the native states . in the active model , the rna chaperone system utilizes chemical energy of atp hydrolysis to repeatedly bind and release misfolded and folded rnas , resulting in substantial increase of yield of the native state . the theory outlined here shows , in accord with experiments , that in the steady state the native state does not form with unit probability . since the ground breaking discovery of self - splicing catalytic activity of group i intron ribozymes @xcite numerous and growing list of cellular functions have been shown to be controlled by rna molecules @xcite . these discoveries have made it important to determine how rna molecules fold @xcite , and sometimes switch conformations in response to environmental signals @xcite to execute a wide range of activities from regulation of transcription and translation to catalysis . at a first glance , it may appear that rna folding is simple because of the potential restriction that the four different are paired as demanded by the watson - crick ( wc ) rule . however , there are several factors that make rna folding considerably more difficult than the more thoroughly investigated protein folding problem @xcite . the presence of negative charge on the phosphate group of each nucleotide , participation of a large fraction of nucleotides in non wc base pairing @xcite , the nearly homopolymeric nature of purine and pyrimidine bases , and paucity of structural data are some of the reasons that render the prediction of rna structures and their folding challenging @xcite . despite these difficulties considerable progress has been made in understanding how large ribozymes fold _ in vitro _ @xcite . these studies have shown that the folding landscape of rna is rugged consisting of many easily accessible competing basins of attraction ( cbas ) in addition to the native basin of attraction ( nba ) , which implies that the stability gap @xcite separating the cbas and the nba is modest relative to proteins . as a consequence of the rugged folding landscape , only a small fraction of initially unfolded molecules reaches the nba rapidly while the remaining fraction are kinetically trapped in a number of favorable alternative low energy misfolded cbas , as predicted by the kinetic partitioning mechanism ( kpm ) @xcite . the free energy barriers separating the cbas and the nba is often high . consequently , the transitions times to the nba from the cbas could exceed biologically relevant time scale ( @xmath1 ) . the upper bound for @xmath1 should be no greater than tens of minutes given the typical cell cycle time . because of the modest stability gap even simple rna molecules could misfold at the secondary as well as tertiary structure levels . in structural terms , secondary structure rearrangements , which are observed in the folding of p5abc @xcite and riboswitches @xcite induced by metal ions and metabolites , respectively , are one cause of the high free energy barriers separating cbas and nba in rna . the free energy barrier associated with melting of @xmath2 base pairs is @xmath3 where @xmath4 , the free energy stabilizing a base pair , is @xmath5 @xmath6 kcal / mol / bp @xcite . the average length of a duplex in rna structure is estimated to be @xmath7 bp from the ratio of nucleotides participating in the duplex formation @xmath8 @xcite , where @xmath9 @xcite is the average length of a single stranded chain in native rnas . using these estimates , we surmise that a typical free energy barrier associated with secondary structure rearrangement is @xmath10 @xmath11 kcal / mol ( @xmath12 @xmath13 @xmath14 ) . by assuming that the prefactor for barrier crossing is @xmath151 @xmath16 @xcite the time scale for spontaneous melting of a hairpin stack could be as large as @xmath17 sec @xmath5 1 day ! indeed , several _ in vitro _ experiments have shown that _ tettrahymena _ ribozyme does not reach the folded state with unit probability even after hundreds of minutes @xcite . the sluggish rna folding kinetics _ in vitro _ is reminiscent of that observed in glasses due to the presence of multiple metastable states ( cbas ) @xcite . because of trapping in long - lived cbas , it is practically impossible for a large ribozyme to spontaneously make a transition to the native state with substantial probability within @xmath1 . these considerations suggest _ in vivo _ folding would require rna chaperones @xcite . the goal of this paper is to produce a quantitative framework for understanding the function of rna chaperones , which are protein cofactors that interact with the conformations in the cbas and facilitate their folding . we classify rna chaperones as passive and active . passive chaperones transiently interact with rna molecules and reduce the entropy barrier to folding without requiring an energy source . on the other hand , active chaperones function most efficiently by lavish consumption of atp in the presence of dead - box proteins . the need for passive or active chaperones depends on the client molecules and the extent of misfolding ( see below ) . we formulate a general kinetic model to describe both passive ( no atp required ) and active ( requires atp hydrolysis ) roles rna chaperones play in rescuing misfolded states . the resulting theory accounts for experimental observations , and should be useful in quantitatively analyzing future experiments .
binary systems consisting of a massive star and a compact object are known to be sources of high - energy gamma - rays . the nature of compact objects in these systems is not exactly known ; there are models , which either assume they are black holes ( e.g. , @xcite ) or pulsars ( e.g. , @xcite ) . in the present paper , we consider gamma - ray binaries with pulsars , where one naturally expects collision of stellar wind with ultrarelativistic pulsar wind . in all such models , the pulsar wind is assumed to be the source of observed gamma - rays through inverse compton scattering of stellar photons either directly , by electrons moving with extreme bulk lorentz factor @xmath0 ( e.g. , @xcite ) , or from the wind termination shock , where electrons are accelerated well beyond their initial ( bulk ) lorentz factor @xmath1 ( e.g. , @xcite ) . a system , where there is a dense radiation field ( here stellar photons ) and an ultrarelativistic plasma flow ( here the pulsar wind ) can launch a runaway - like particle acceleration ( converter acceleration ) , as been already discussed in application to gamma - ray bursts and active galactic nuclei ( @xcite ) . in gamma - ray binaries the feedback from downstream of the pulsar wind to its upstream is realized through two - photon pair production . below we show that it is possible to construct a model , which can reproduce main observed properties of gamma - ray binaries without involving any physical processes apart from the two - photon pair production feedback itself and consequent cooling of the energetic pairs through synchrotron and inverse compton radiation . obviously , the physics of gamma - ray binaries is much more complex than this simple model , but our model demonstrates that the pair feedback and subsequent converter acceleration has to be included as an essential , if not the most important , component in any advanced model for gamma - ray binaries with pulsar winds . for our model , we assume the following setup ( base on the parameters of ls 5039 and ls i + 61 303 , which can be considered to be typical representatives of gamma - ray binary population ) . the massive star has the luminosity @xmath2 erg / s and the mass - loss rate @xmath3/yr . the mass outflow is due to stellar wind with velocity @xmath4 . the pulsar orbits the massive star at the distance @xmath5 cm and produces a magnetized ultrarelativistic wind with power @xmath6 erg / s ) , which consists of electron - positron pairs and has the lorentz factor @xmath7 ( we will show later that @xmath8 ) and magnetization parameter @xmath9 . the pulsar wind termination shock is located at the distance @xmath10 from the pulsar , where the dynamical pressures of the two winds are equal . at this distance , the energy density in the pulsar wind is @xmath11 , what constitutes a fraction @xmath12 of the energy density of the background thermal photons supplied by the massive star , @xmath13 . thus , neither comptonization of stellar radiation nor the synchrotron and self - compton radiation is an a priori dominant emission mechanism . the shock - bounded pulsar wind region is filled with stellar photons and the high - energy radiation produced at the termination shock , which forms ( inside the shock ) an isotropic and uniform photon field . interacting with each other , these photons occasionally produce electron - positron pairs , which are picked up by the pulsar wind , becoming much more energetic , and then lose energy for radiation on their way to the shock and , to a greater extent , after crossing the shock and entering the shocked plasma . the rate of secondary pair production @xmath14 does not depend on the angular distribution of the parent photons as soon as at least one of the interacting photon fields is isotropic ( the shock - generated photons are always isotropic ) , and the velocities of secondary pairs are uncorrelated with the wind velocity . when pairs with such velocity distribution are picked up by the pulsar wind , their energy increases on average by the factor @xmath15 . energetic secondary electrons and positrons readily cool : a fraction of their energy is radiated before they reach the termination shock , the rest is radiated in the post - shock region . the cooling is more efficient in the post - shock region both because the particles spend more time there and because synchrotron emission is possible in the shocked plasma in addition to inverse compton emission . the synchrotron emission from secondary pairs supplies additional low - energy target photons for production of new pairs , forming a feedback loop . the overall contribution of this feedback to the radiative efficiency of the shocked pulsar wind can be estimated as @xmath16 where @xmath17 is the average total energy of a secondary pair at the moment of birth . should the pulsar wind become heavily loaded with secondary pairs , its lorentz factor decreases to the value , which ensures the feedback efficiency @xmath18 is less than unity . consider two isotropic populations of photons with energies around @xmath19 and @xmath20 ( @xmath21 for definiteness ) , and energy densities @xmath22 and @xmath23 . the pair production rate approximately equals @xmath24 where @xmath25 @xmath26 is the angle - averaged pair production cross - section , @xmath27 and @xmath28 are the energy densities in two populations of interacting photons in units of the background stellar radiation energy density . with increase of @xmath19 , the optimal energy of target photons @xmath29 decreases and , for a given radiation energy density , the number of target photons ( hence opacity ) goes up . so , all other things being equal , the larger is @xmath19 ( which determines the energy of electron - positron pair at birth , @xmath17 ) , the larger is the feedback efficiency . even though the feedback may be controlled by inverse compton radiation of primary pairs , the overall radiated power is mainly due to secondary pairs , which have energies @xmath30 and upscatter stellar photons inefficiently because of the klein - nishina suppression in the scattering cross - section . their main energy - loss channel is synchrotron radiation in the post - shock region , where secondary pairs form the cooling distribution and their synchrotron spectrum has the photon index -3/2 . it extends in energy up to @xmath31 where @xmath32 the magnetic field strength in the post - shock region . when the pair feedback starts to develop , secondary pairs are born in interaction of comptonized stellar photons ( produced by the primary electrons ) with energy density @xmath33 where @xmath34 is the cooling timescale , @xmath35 the thomson cross - section , and @xmath36 is the lorentz factor of the pulsar wind as it would be in absence of the pair feedback . the target radiation field is synchrotron radiation of the secondary pairs . given the photon index @xmath37 , its energy density at @xmath29 is @xmath38 combining equations eq . ( [ effic ] , [ pair_rate ] , [ spec_end ] , [ w1 ] , [ w2 ] ) we find the self - consistent lorentz factor : @xmath39 the feedback efficiency enters eq . ( [ scgamma ] ) inexplicitly , through the ratio @xmath40 . with this ratio much less than unity the feedback efficiency is @xmath41 . the lorentz factor , found in this way , depends on the distance to the massive star , @xmath42 . then , the synchrotron cutoff energy is very sensitive to this distance : @xmath43 . if this volatile synchrotron cut - off approaches @xmath44 , then the pair feedback enters a different , more robust mode , where the synchrotron photons from secondary pairs get absorbed on themselves , so that the feedback becomes over - efficient and saturates , keeping @xmath45 somewhat below @xmath46 . the self - consistent lorentz factor now becomes @xmath47 where we substituted @xmath45 for @xmath46 , keeping in mind that dependence of the equilibrium lorentz - factor @xmath7 on @xmath45 is very weak . under typical conditions , eq . ( [ sy_eq_gamma ] ) gives @xmath48 . the pair feedback model predicts that the radiation from pulsar wind and the shock region contains several spectral components of different origin . first , it is emission from ( slowly cooling ) primary pairs inverse compton from stellar radiation , peaked at @xmath49 and synchrotron , peaked at @xmath50 ev , which is overwhelmed by the radiation of the massive star . next , there is emission from secondary pairs , both synchrotron , peaked at @xmath51 , and ( less efficient due to the klein - nishina effect ) inverse compton , peaked at @xmath52 tev . the energy of comptonized photons is limited by the energy of secondary pairs .
we analyze the model of gamma - ray binaries , consisting of a massive star and a pulsar with ultrarelativistic wind . we consider radiation from energetic particles , accelerated at the pulsar wind termination shock , and feedback of this radiation on the wind through production of secondary electron - positron pairs . we show that the pair feedback limits the lorentz factor of the pulsar wind and creates a population of very energetic pairs , whose radiation may be responsible for the observed gamma - ray signal . address = institute of applied physics ras , 46 ulyanov st , 603950 nizhny novgorod , russia , altaddress = lobachevsky state university of nizhni novgorod , 23 gagarin av , 603950 nizhny novgorod , russia address = school of cosmic physics , dublin institute for advanced studies , 31 fitzwilliam place dublin 2 , ireland , altaddress = max planck institute fr kernphysik , saupfercheckweg 1 , d-69117 heidelberg , germany
in 1987 anderson @xcite suggested the strange behavior of cuprate materials between the superconducting dome and the magnetically ordered insulating phase could be described by a resonating valence bond ( rvb ) state in which preexisting magnetic singlet pairs of the insulating state become charged superconducting pairs when the insulator is doped . just one year later appeared the first effective model in which the magnetic degrees of freedom are disregarded in favor of the more pertinent singlet degrees of freedom @xcite . this is nothing but the quantum dimer model ( qdm ) . however , it was soon realized@xcite that the groundstate of the undoped @xmath1 heisenberg antiferromagnet on a square lattice exhibits a long - range nel order in contrast to the initial expectation based on the rvb picture . furthermore , the qdm on a square lattice was also found to have only gapped crystalline phases but no evidence of an rvb spin liquid phase in a finite region of the phase diagram.@xcite it is still possible to argue that , even though the undoped antiferromagnet has the nel order , the rvb picture gives a better theoretical starting point once the system is doped with holes . however , it would be natural to ask if the rvb spin liquid phase can be realized in undoped magnetic system with only short - range interaction . one may expect that magnetic frustration would favor rvb state over the nel phase . thus , over the time , the main interest in qdm was shifted from the original motivation of the application to high @xmath2 superconductivity , to the effects of frustration . however , clear confirmation of a rvb phase remained elusive for a rather long time . a breakthrough in the study of qdm was due to moessner and sondhi @xcite who showed that a simple qdm defined in the triangular lattice exhibit a disordered phase which , recalling that these dimer models are supposed to be effective models for frustrated magnets , can be considered as an explicit example of the rvb spin liquid phase . it was also recognized that , the rvb spin liquid phase is a topologically ordered phase with a nontrivial topological degeneracy of the groundstates.@xcite in fact , the rvb spin liquid phase is essentially identical to the @xmath3 topological phase which was introduced in a completely different context of quantum information processing.@xcite the qdm is generally not exactly eqivalent to an antiferromagnetic hamiltonian defined in terms of quantum spins . however , the projection from a magnetic system to a qdm was performed successfully in heisenberg antiferromagnets on frustrated lattice , such as the square lattice with strong enough second and/or third neighbors couplings or the kagome lattice . these suggest that the qdm may well represent phases without magnetic order in antiferromagnets . in fact , very recently , frustrated heisenberg antiferromagnets on kagome and other lattices are reported to be in the rvb spin liquid phase ( @xmath3 topological phase ) by several authors.@xcite now that the existence of the rvb spin liquid phase appears to be confirmed in qdms as well as in antiferromagnets , the issue of superconductivity in doped spin liquids becomes a more pressing question . this isssue started in fact to be investigated shortly after the apearence of qdm @xcite doping of an rvb spin liquid is expected to induce a novel type of elementary excitations called holon . a holon , carrying electric charge @xmath0 but no spin , appears as a result of fractionalization , namely deconfinement of fractionalized excitations . indeed , topological degeneracy of the undoped rvb spin liquid is known to be intimately connected to the fractionalization phenomenon@xcite . superconductivity may be realized if the holons condense . at least naively , one may expect that the resulting superconductor is an exotic one due to condensation of charge-@xmath0 holons , instead of usual charge-@xmath4 cooper pairs . a fundamental issue in this problem is the statistics of the holon . for the holons to condense without forming pairs , they must be bosons . however , it should be noted that transmutation of the statistics@xcite is possible . namely , the statistics of holons as elementary excitations appearing in the low - energy limit can be different from the statistics assigned to holons in the microscopic model . in this paper we address the issue of the statistics of holes and its interplay with possible superconducting phases in doped qdms . in a recent work@xcite it was shown that a qdm with fermionic ( at microscopic level ) holes is equivalent to another qdm with bosonic holes . because of the equivalence , the statistics of the holon as a physical elementary excitation must be the same for either representation . this proves the existence of a dynamical statistical transmutation in the system . in this paper we study in more details the statistical transmutation in qdm and give a simple and efficient method to obtain the relation between the qdms with fermionic and bosonic representation of the holes . in section [ sec.gauge ] we introduce a second quantization notation for qdm hamiltonians and show the gauge symmetry associated with them . in section [ sec.composite ] we present the composite particle representation of qdm hamiltonians which is the key ingredient to show the exact equivalence between a qdm with bosonic and another qdm with fermionic holes . this equivalence is shown for a generic flipping term defined in any kind of lattices . the result , which relies in an orientation prescription of the bonds in the lattice considered , is totally generic and can then be applied to any qdm defined in the most common lattices . it is important to stress that the method used here differs considerably with the one used in ref . where a two - dimensional version of the jordan - wigner ( jw ) transformation was used . as we explain in the conclusion , it is important to point out the numerous advantages of this new representation compared to the jw transformation . in section [ sec.classification ] we argue how the modification of the orientation prescription can be interpreted as a simple gauge transformation in the qdm hamiltonian . we then apply the general result of the statistical transmutation obtained in section [ sec.composite ] to generic qdm hamiltonians defined on the square , triangular , hexagonal and kagome lattices . section [ sec.numerical ] is devoted to numerical investigation of four inequivalent qdm defined on the triangular lattice . in particular , we identify an exotic superconductor phase due to condensation of holons with charge @xmath0 , measuring the gauge - invariant green s function of a single holon . in section [ sec.bh ] we discuss an explicit realization of one of the qdm discussed in section [ sec.numerical ] . it is obtained as a low energy strong interaction limit of a bose - hubbard model on the kagome lattice . the number of bosons is directly related to the doping , or number of holes , in the resulting qdm on the triangular lattice . section [ sec.conclusions ] is devoted to the discussion of our results . we also include as an appendix the derivation of the statistical transmutation for a generic qdm on the kagome lattice using the jordan - wigner transformation . of course the result is consistent with the one obtained with the composite particle representation obtained in section [ sec.composite ] , but allows a better understanding of the connection between these two different methods . we start with a doped quantum dimer model on a two - dimensional lattice . to fix the ideas , we work here with the hamiltonian defined on the square lattice but all the arguments remain valid for any two dimensional lattice . we write the hamiltonian as h = h_j+h_v+h_t [ eq : h1 ] with h_j&=&-j _ + h_v&=&v _ + h_t&=&-t_\ { ||+ || . + & + & . ||+ ||+ } , where the sums are over all the smallest resonant plaquettes on the lattice ( for the square lattice these are the squares ) . in a second quantized formalism we assume that dimer configurations are created by spatially symmetric dimer operators @xmath5 and holes are created by bosonic operators @xmath6 . then , we can re - write the hamiltonian as : h_j&=&-j_\ { b^_i , jb^_k , l b_j , kb_l , i+ } + h_v&=&v_\ { b^_i , jb^_k , l b_i , jb_k , l + b^_j , kb^_l , i b_j , kb_l , i } + h_t&=&-t_i\ { b^_i , jb_j , k a^_ka_i + } . [ hoping_t ] in the last equation , the indices correspond to the labeling of the sites of a square plaquette as in figure [ fig : index ] . in our previous conventions , dimer configurations are represented by spatially symmetric operators @xmath5 satisfying : [ eq : commut_b ] & = & _ i , k_j , l+_i , l_j , k + & = & = 0 . the boson operator @xmath7 creates a hole in the site @xmath8 and satisfies [ eq : commut_a ] & = & _ i , j + & = & = 0 . the operators @xmath9 and @xmath10 commutes one with each other [ eq : commut_ab ] = = = 0 . we introduce in the model a constraint on the number of dimers and holes which warrants that at each site of the lattice there is either one and only one hole or one and only one dimer arriving to it : [ eq : constraint ] a^_ia_i+_z=_1,_2b^_i , i+zb_i , i+z=1 . of course this constraint implies , among others , that the holes have to be considered as hard core bosons . it is important to notice that the hamiltonian has the following @xmath11 gauge symmetry : [ eq : gauge1 ] a_j & & e^i_ja_j + [ re : gauge2 ] b_j , k & & e^i(_j+_k)b_j , k where @xmath12 is an angle . this invariance can be exploited to prove the statistical transmutation symmetry in some two dimensional systems by means a jordan - wigner transformation on the holon operators@xcite . in the following we present an alternative description for the doped qdm using composite operators which allows us to understand in a different way the equivalence between a model with bosonic holes and one with fermionic holes . for doing this , we have first to make a choice of a given orientation prescription for the bonds in the lattice . ( color online ) schematic snapshot of a doped dimer liquid " . each site is occupied by either a ( single ) dimer or a hole ( empty site ) . , scaledwidth=40.0% ] in order to prove the equivalence between a qdm hamiltonian with bosonic holes and another hamiltonian with fermionic holes , we propose a different formulation for the qdm . this formulation is done in terms of composite particles by defining the operator b_i , j = b_i , ja^_ia^_j . this operator destroys a dimer between sites @xmath8 and @xmath13 and creates two holes at the same sites . let s call @xmath14 the subspace of states that satisfies the constraint ( [ eq : constraint ] ) . for a given state @xmath15 we have that @xmath16 is also a vector in @xmath14 . a ) indexes corresponding to each square plaquette in the hamiltonians @xmath17 and @xmath18 . b ) indexes corresponding to each hopping process in @xmath19 . , scaledwidth=45.0% ] one can easily notice that the operator @xmath20 is invariant with respect to the gauge transformation a_j & & e^i_ja_j + b_j , k & & e^i(_j+_k)b_j , k this @xmath11 gauge symmetry was exploited in ref . to represent doped qdm as theories in which a matter field is coupled to a gauge field . more importantly , one can check that _ within the subspace _ @xmath14 , the set of operators @xmath21 form a closed algebra similar to the one of @xmath22 . the hamiltonian can entirely be written in terms of these @xmath20 operators making its gauge invariance a trivial statement . its precise form is given by @xmath23 with : h_j&=&-j_\ { b^_i , jb^_k , l b_j , kb_l , i+ } + h_v&=&v_\ { b^_i , jb^_k , l b_i , jb_k , l + b^_j , kb^_l , i b_j , kb_l , i } + h_t&=&-t_i\ { b^_i , jb_j , k + } it is evident that , within this formulation , the basic building blocks of the model are created by @xmath24 which corresponds to composite particles of charge @xmath4 . namely , the model is completely defined in terms of the constituent particle with charge @xmath4 . this has several important consequences . in particular , the gauge invariance requires that the energy spectrum of the system on a torus , as a function of the magnetic flux @xmath25 through the `` hole '' of the torus , is invariant under @xmath26 . ( for early discussions on the @xmath27-flux periodicity in the qdms , see refs . and references therein . ) this periodicity corresponds to the unit flux quantum for charge @xmath4 objects . however , this does not necessarily mean that the physical elementary excitations of the system have minimum charge @xmath4.@xcite the system can have a topological order which leads to _ fractionalization _ ; elementary excitations can have fractions of the charge @xmath4 of the constituent particle of the microscopic hamiltonian . if the charge-@xmath0 holons are deconfined as a result of fractionalization , they could condense to form an exotic superconductor . the apparent contradiction between the periodicity of the energy spectrum in @xmath27 flux and the expected flux quantization in the unit of @xmath28 in the condensate of charge-@xmath0 holons is resolved by the existence of the topological vortex excitation called vison . insertion of the @xmath27 flux corresponds to trapping of a vison . although the flux periodicity of the groundstate energy does not distinguish an exotic charge-@xmath0 condensate from the usual superconductor , an experimental detection scheme of the charge-@xmath0 condensation , based on a `` vortex memory effect '' , was proposed.@xcite an actual experiment@xcite on the high-@xmath2 superconductor did not find such a signature of charge-@xmath0 condensation . nevertheless , the exotic supercondutivity due to condensation of charge-@xmath0 objects is possible in principle , and is an interesting subject to pursue theoretically and experimentally . later in this paper , we will introduce a quantity which detects a charge-@xmath0 condensation , and study it numerically in several qdms . the main advantage of the formulation in terms of composite operators presented above is that one can prove an equivalence between a hamiltonian where holes are hard core bosons and another one where the holes are fermions . let us consider a qdm hamiltonian with bosonic holes , where their creation and annihilation operators @xmath7 and @xmath29 satisfy bosonic commutation relations . let us also consider another qdm hamiltonian with fermionic holes , created and annihilated by the set of operators @xmath30 and @xmath31 which now satisfy fermionic anti - commutation relations . we then build the respective composite operators : b_i , j&=&b_i , ja^_ia^_j + _ i , j&=&b_i , jf^_if^_j as for the operators @xmath32 , all quantities in the rest of the paper with a tilde correspond to operators and coupling constants of the _ fermionic _ representation for the holes . before proceeding , there is an important statement to make . again , one can show that within the subspace @xmath14 , both set of operators @xmath33 and @xmath34 form the _ same _ closed algebra of _ bosonic _ dimer operators . another important point is that the definition of the composite operators in terms of fermions is more subtle because it is necessary to take a prescription for the orientation of the dimers ( which determines the order of the fermions in the endpoints of each dimer ) . in the following we will call _ `` even prescription '' _ of a given plaquette an ordering prescription for the bonds such that all the bonds are oriented in a clockwise direction or an even number of bonds are oriented anti - clockwise . by contrast we call _ `` odd prescription '' _ the prescriptions obtained from the clockwise ordering by flipping an odd number of bonds . notice that , since the resonance plaquettes containing @xmath35 dimers have necessary @xmath36 bonds , the anti - clockwise prescription(where all bonds are oriented anti - clockwise ) is always an _ even prescription_. [ [ theorem ] ] theorem : + + + + + + + + _ given a resonant plaquette of arbitrary length with an even prescription for the bonds , then , for the kinetic term of the dimers in the plaquette , we have the equivalence : @xmath37 . in other words , the resonance term of dimers in the plaquette is invariant under a simultaneous change of `` statistics '' of the holes in the system ( i.e. bosonic into fermionic or vice - versa ) and the sign of the dimer resonance loop amplitude @xmath38 . _ proof : consider a resonance loop containing @xmath36 sites ( @xmath35 dimers ) numbered from 1 to @xmath36 in the clockwise direction as in fig . [ fig : loop]-a and fig . [ fig : loop]-b . the kinetic hamiltonian for dimers belonging this loop can be written in terms of bond operators @xmath39 as h^n_j&=&j + & & + where , the index @xmath35 indicates the number of dimers in the loop . now , we add one dimer ( 2 sites ) to the loop , obtaining a resonance loop with @xmath40 dimers . in this case , the hamiltonian can be written as h^n+1_j&=&j + & & + since all the operators are acting on different bonds they all commute and we can rearrange them in the following way h^n_j&=&j + & & + h^n+1_j&=&j + & & + or in a compact notation + & & j ( _ j=1^n b^_2j-1,2j ) ( _ j=1^n-1 b_2j,2j+1 ) b_2n,1 + h^n+1_j = j ( _ j=1^n b^_2j-1,2j ) ( _ j=1^n-1 b_2j,2j+1 ) b^ _ , b_2n , b_,1 + now , we insert on the right of the hamiltonian the string of operator @xmath41 , where the index @xmath8 correspond to the sites on the resonance loop . this operator acts as the identity operator on the sites belonging the loop because @xmath42 in the absence of holes . we start with the hamiltonian @xmath43 h^n_j&=&h^n_js^f_n + & = & j ( _ j=1^n b^_2j-1,2j ) ( _ j=1^n-1 b_2j,2j+1 ) b_2n,1 + & & f_1f_1^f_2nf_2n^ + now , we move the fermions to the left in order to form the composite operators @xmath44 corresponding to the dimer operators inside the brackets . commutation of the fermions gives a global sign . h^n_j&=&(-1)^j ( _ j=1^n ^_2j-1,2j ) ( _ j=1^n-1 _ 2j,2j+1 ) + & & b_2n,1 f_1^f_2n^ + we can follow exactly the same procedure in the loop with @xmath40 dimers , the global sign resulting from the commutation of fermion operators to write the products in terms of composite particles is the same that in the @xmath35 dimers case . we can write for the @xmath40 case h^n+1_j&=&(-1)^j ( _ j=1^n ^_2j-1,2j ) ( _ j=1^n-1 _ 2j,2j+1 ) + & & b^_,b_2n , b_,1 f_1^f_2n^f_f_^f_f_^ + now we can determine the change in the sign of @xmath38 when a dimer is added in the loop . first we commute the operators @xmath45 and @xmath46 in @xmath43 to form the operator @xmath47 . this commutation gives another sign to complete the global phase in the hamiltonian . then we can write for the @xmath35 dimers case h^n_j=(-1)^+1j ( _ j=1^n ^_2j-1,2j ) ( _ j=1^n-1 _ 2j,2j+1 ) _ 2n,1 + in the hamiltonian corresponding to @xmath40 dimers we have to define 3 composite operators . we have that b^_,b_2n , b_,1 f_1^f_2n^f_f_^f _ f_^ & = & ( b^_,f_f _ ) b_2n , b_,1 f_1^f_2n^f_^ f_^ + & = & ( b^_,f_f _ ) ( b_2n , f_2n^f_^ ) b_,1 f_1^ f_^ + & = & ( -1)(b^_,f_f _ ) ( b_2n , f_2n^f_^ ) ( b_,1f_^f_1^ ) + & = & ( -1)^ _ , _ 2n , _ , 1 finally , the hamiltonian corresponding to @xmath40 dimers reads h^n+1_j&=&(-1)^+1j ( _ j=1^n ^_2j-1,2j ) ( _ j=1^n-1 _ 2j,2j+1 ) + & & ^_,_2n,_,1 + we have proved by induction that , if for a @xmath35 dimers loop , the kinetic term acquires a given sign when the hamiltonian is written in terms of fermionic composite particles , then the kinetic amplitude corresponding to @xmath40 dimers acquires the same sign . to complete this mathematical induction proof we need check that the statement holds for the lowest value of @xmath35 . the smallest possible resonance loop is given by a loop with only 2 dimers . it is easy to check that in this case h^2_j&= & j b^_1,2b^_3,4b_2,3b_4,1 + + & = & j b^_1,2b^_3,4b_2,3b_4,1 f_1 f^_1 f_2f^_2 f_3f^_3 f_4f^_4 + + & = & ( -1 ) j ^_1,2^_3,4 _ 2,3_4,1 + then we have proved that the kinetic term for a resonance loop of arbitrary length oriented in a clockwise direction , the amplitude @xmath38 in the hamiltonian written using dimer operators @xmath39 changes to @xmath48 when we write the hamiltonian in terms of fermionic composite operators @xmath49 . a trivial verification shows that the amplitude remains unchanged when we write the hamiltonian in terms of bosonic operators @xmath50 . the result above can be re - written in the more appealing way : h_j(j,)h_j(-j , b ) the proof can easily be extended to the potential term @xmath18 . in this case is easy to see that the bosonic and fermionic versions give the same sign in the amplitude @xmath51 . h_v(v,)h_v(v , b ) the equivalence proved above is valid for any even prescription on the plaquette . starting from the clockwise prescription where the results above has been proved , if we flip two bonds this induce the commutation of two fermionic operators and the sign remains unchanged . but if we flip an odd number of loops we must to commute an odd number of extra fermionic commutations in order to form the composite operators . these permutations gives an extra sign in the hamiltonian . then is easy to prove the following corollary : _ corollary 1 : _ + _ given a resonant plaquette of arbitrary length with an odd prescription for the bonds , then , for the kinetic energy of the dimers in the plaquette , we have the equivalence : @xmath52 . _ the equivalence in the potential term do not change if we take an odd or even prescription . using this property of the potential term and the corollary 1 we can derive the following corollary . _ corollary 2 : _ + _ note that we have actually proved that , if in a given lattice we can take an even prescription for all the plaquettes involved in the hamiltonian . then the equivalence h(j , v , bosons)h(-j , v , fermions ) is valid for the hamiltonian in the complete lattice . whereas if we can take an odd prescription for all the plaquettes in the hamiltonian . we have the equivalence h(j , v , bosons ) h(j , v , fermions ) _ in order to complete the panorama for the doped qdm we study the fermionic and bosonic representation of the hamiltonian @xmath53 corresponding to the hopping of holes . consider 3 nearest - neighbors sites of the lattice as in fig . [ fig : index]-b . in terms of bosonic holes and dimer operators we can write a general hoping term as h^(t)_i , j , k = b^_i , jb_j , ka^_ka_i if there is no hole in the intermediate site @xmath13 we can add on the right the identity as @xmath54 . we then obtain h^(t)_i , j , k&=&-t b^_i , jb_j , ka^_ka_ia_ja_j^ + & = & -t ( b^_i , j a_ja_i ) ( b_j , ka_j^a^_k ) + & = & -t b^_i , jb_j , k in the bosonic case we do nt need to worry about the prescription in the lattice but it is important when we study the fermionic description . in this case we take the prescription @xmath55 . starting from the hamiltonian ^(t)_i , j , k = b^_i , jb_j , kf^_kf_i we insert on the right the operators @xmath56 ^(t)_i , j , k&=&-t b^_i , jb_j , kf^_kf_if_jf_j^ + & = & -t ( b^_i , j f_jf_i ) ( b_j , kf_j^f^_k ) + & = & -t ^_i , j_j , k using the prescription @xmath57 , the amplitude in the hopping term for the holes is the same if we use the fermionic or bosonic versions of the composite operators . flipping two arrows we have the prescription @xmath58 . it is a simple matter to see that with this prescription the hopping amplitude is also the same for the two cases . then , the hopping of the holes written in terms of bosonic and fermionic composite operators have the same amplitude @xmath59 , provided that we use one of the two prescriptions satisfying that the intermediate site has one incoming and one out - coming arrow . if we take this @xmath60 prescription in all the sites of the lattice the arrows follow a sort of kirchhoff s first rule : see fig . [ fig : pres_square]- ( a ) , ( b ) , ( d ) . we will call this kind of prescriptions as `` zero - current '' prescriptions . of course it is only possible to satisfy this prescription in all the sites if the coordination number of the lattice is even . an example where this is not possible is the honeycomb lattice ( with @xmath61 ) . in this lattice it turns out that it is not possible to take a prescription with the same number of incoming and out - coming arrows in each site . [ fig : pres_square]-(c ) . as we saw in the last section , in order to prove the equivalence between hamiltonians built with bosonic and fermionic operators , one needs a bond orientation prescription for the fermionic case . of course , this prescription is totally arbitrary and before proceeding it is important to clarify the issue of a different choice of prescription . let us imagine a generic lattice for which we have chosen two different prescriptions , a and b. to clarify the ideas , imagine that the orientation of all the bonds in prescription b are the same that in prescription a , except for one single bond , which is connecting points @xmath8 and @xmath13 . then , starting from a bosonic hamiltonian , by doing the transmutation , we end up with two different hamiltonians @xmath62 and @xmath63 which have the same signs for all the flipping and hopping terms except for the ones that contain the bond @xmath64 . let us illustrate this with the following example : consider the square lattice in which prescription a is the one given in fig . [ fig : pres_square ] . then , imagine a prescription b where only the arrow between sites @xmath8 and @xmath13 is reversed , as shown in fig . [ fig : arrow_reversed ] . starting from the same bosonic hamiltonian , after the statistical transmutation , we get the hamiltonians @xmath62 and @xmath63 . what is the difference between @xmath62 and @xmath63 ? they have the same signs for all the flipping terms , except the the ones of plaquettes @xmath65 and @xmath66 , which are the only two containing this reversed bond . also , all hoping terms are the same except those containing the link @xmath64 . although one could naively think that these two resulting hamiltonians are not equivalent , in fact they are , as can be easily seen by performing the following gauge transformation : b^_n , m & & -b^_n , m n = i , m = j n = j , m = i + b^_n , m & & b^_n , m most generally , it is easy to convince oneself that different choices of prescriptions give rise to apparently different hamiltonian which in fact are equivalent under a certain gauge transformation . we are now going to consider each lattice in detail and justify for each of them the choice of prescription we have made . the change of prescription corresponding to reversing the orientation of one single bond ( @xmath64 in the figure ) corresponds to a gauge transformation where only configurations containing a dimer in the @xmath64 bond have their sign changed . this in turn has the effect of reversing the sign of the flipping and hoping terms containing the bond @xmath64 , as for example the flipping of plaquettes @xmath65 and @xmath66 . , scaledwidth=45.0% ] for the square lattice , we consider the prescription given in fig . [ fig : pres_square]-(a ) . using this prescription , the hopping amplitude ( @xmath59 ) remains equal for the bosonic and fermionic representation @xmath67 and @xmath32 . on the other hand , the kinetic amplitude corresponding to dimers ( @xmath68 ) changes its sign if an even prescription is induced in the plaquette of length @xmath65 . in fig . [ fig : pres_square ] the plaquettes of lowest order are shown . light - blue areas correspond to even prescriptions induced in the plaquettes while green areas correspond to odd prescriptions . the relative sign between the couplings in the fermionic and bosonic representations corresponding to the 8 smallest plaquettes are presented in table [ tab : square_plaq ] after the transmutation , the hamiltonian corresponding to the hoping of holes is given by _ t=^(t)_+^(t ) _ where in @xmath70 dimers are also created by operators @xmath71 and the holes are fermions created by the operators @xmath30 : ^(t)_&=&-_\ { | | + | | + | | . + [ eq : h_t_tilde_kagome ] & + & . } , where @xmath72 . in order to obtain the original value of the hoping constant for the holons ( @xmath73 ) , we can perform a gauge transformation on the dimers . this is possible , for example , by mean the following gauge transformation | & & | + ^(t)_&=&-t_\ { | | + | | + | | . + [ eq : h_t_tilde_kagome_2 ] & + & . } while for the resonance terms the gauge transformation does not change the couplings @xmath74 and @xmath51 . then we obtain _ j&=&j_6 - j_8^(a ) + j_8^(b ) -j_8^(c ) + & + & j_10^(a ) - j_10^(b ) + j_10^(c ) -j_12 finally we can use another gauge transformation to get a more simple hamiltonian . we change the dimers on up triangles as | & & e^i_1 | + @xmath75 , @xmath76 and @xmath77 . the amplitudes @xmath68 change as _ j&=&-j_6 - j_8^(a ) - j_8^(b ) -j_8^(c ) + & - & j_10^(a ) - j_10^(b ) -j_10^(c ) -j_12 then finally we obtain a hamiltonian with a global change of sign in the amplitudes of the kinetic term . after the transformation , the terms corresponding to the hoping of holons becomes ^(t)_&=&-t_\ { e^-i/2| | . | | + e^i/2 | | } + & + & this hoping hamiltonian can be transformed by mean a simple gauge transformation on the holes , @xmath78 with @xmath79 , to recover the original form of the hamiltonian . with this transformation we have that f^_j+p_1f_j & & e^i f^_j+p_1f_j + f^_j+p_2f_j & & e^i f^_j+p_2f_j , and we recover the original form ( [ eq : h_t_tilde_kagome_2 ] ) for the hamiltonian corresponding to the holes . then , finally we have that changing the statistics of holons together with the sign of all the kinetic amplitudes we obtain a completely equivalent hamiltonian . by using the j - w transformation we can recover the equivalences presented in previous section , but the procedure is more laborious . we have pointed that starting from different prescriptions for the bonds we can prove different equivalences . a change i the lattice prescription used in the composite operator representation is equivalent to a gauge transformation on on the dimers . one last very important point concerns the issue of boundary conditions within the j - w approach . the composite operator approach which we have extensively used in the paper is valid independently of the boundary conditions used for the system . in contrast , the j - w approach is valid only for infinite systems or finite systems with open boundary conditions . let us come back to equation ( [ eq : jordan_wigner ] ) assuming that the bosonic @xmath9 operators are well defined in a system in periodic boundary conditions . this means that for example the operators @xmath80 and @xmath81 are forced to be the same if sites @xmath8 and @xmath82 correspond to the same point in the system because of the periodic boundary conditions . however , because of the very nature of the non - local j - w transformation , the relation between the operators @xmath83 and @xmath84 must contain a twist . in the one - dimensional version of the j - w transformation , it is easy to see that this twist is a sign which depends only on the total number of particles in the system . then , in one dimension , if one restricts to the subspace of a fixed number of particles it is possible to implement consistent way periodic boundary conditions . here in two dimensions , the twist one should force for the fermionic operators depends not only on the number of particles but also on their relative positions with respect to the points @xmath8 and @xmath82 . that means that , even restricting to a fixed number of particles , starting from periodic boundary conditions for the bosons operators , it is not possible to implement boundary conditions for the fermionic operators which are consistent with all the possible particle configurations . n. read and s. sachdev , phys . lett . * 62 * , 1694 ( 1989 ) ; s. sachdev , phys . rev . b * 40 * , 5204 ( 1989 ) ; a. ralko , d. poilblanc , and r. moessner , phys . rev . lett . * 100 * , 037201 ( 2008 ) ; see also e. fradkin , _ field theories of condensed matter systems _ ( addison - wesley , redwood city , ca , 1991 ) .
quantum dimer models ( qdm ) arise as low energy effective models for frustrated magnets . some of these models have proven successful in generating a scenario for exotic spin liquid phases with deconfined spinons . doping , i.e. the introduction of mobile holes , has been considered within the qdm framework and partially studied . a fundamental issue is the possible existence of a superconducting phase in such systems and its properties . for this purpose , the question of the statistics of the mobile holes ( or holons " ) shall be addressed first . such issues are studied in details in this paper for generic doped qdm defined on the most common two - dimensional lattices ( square , triangular , honeycomb , kagome , ... ) and involving general resonant loops . we prove a general `` statistical transmutation '' symmetry of such doped qdm by using composite operators of dimers and holes . this exact transformation enables to define duality equivalence classes ( or families ) of doped qdm , and provides the analytic framework to analyze dynamical statistical transmutations . we discuss various possible superconducting phases of the system . in particular , the possibility of an exotic superconducting phase originating from the condensation of ( bosonic ) charge-@xmath0 holons is examined . a numerical evidence of such a superconducting phase is presented in the case of the triangular lattice , by introducing a novel gauge - invariant holon green s function . we also make the connection with a bose - hubbard model on the kagome lattice which gives rise , as an effective model in the limit of strong interactions , to a doped qdm on the triangular lattice .
from a classical result by ore @xcite it is well - known that if a simple graph @xmath0 with @xmath1 vertices is such that @xmath2 for each pair @xmath3 of non - adjacent vertices of @xmath0 , then @xmath0 contains a hamiltonian path . a _ leaf _ of a tree @xmath4 is a vertex of @xmath4 with degree one . a natural generalisation of hamiltonian paths are spanning trees with a small number of leaves . in this direction , ore s result was generalised by broersma and tuinstra @xcite to the following theorem . @xmath5}$ ] let @xmath6 and @xmath1 be integers . if @xmath0 is a connected simple graph with @xmath7 vertices such that @xmath8 , for each pair @xmath3 of non - adjacent vertices , then @xmath0 contains a spanning tree with at most @xmath9 leaves . further related results have been obtained by egawa _ et al _ @xcite and by tsugaki and yamashita @xcite . see also @xcite for a survey on spanning trees with specific properties . in this note we consider spanning trees with small degrees as well as with a small number of leaves . our result is the following . let @xmath7 , @xmath10 and @xmath11 be integers with @xmath12 and @xmath13 . if @xmath0 is a @xmath14-connected simple graph with vertex set @xmath15 such that @xmath16 for any non - adjacent vertices @xmath17 and @xmath18 of @xmath0 , then @xmath0 has a spanning tree @xmath4 with at most @xmath19 leaves and such that @xmath20 for @xmath21 . let @xmath4 be a largest subtree of @xmath0 with at most @xmath22 leaves and such that if @xmath23 , then @xmath24 . since @xmath0 is @xmath10-connected and @xmath1 , it contains a path with at least @xmath25 vertices . therefore , we may assume that tree @xmath4 has at least @xmath25 vertices . if @xmath4 is not a spanning tree , there is a vertex @xmath26 of @xmath0 not in @xmath4 . by menger s theorem , there are @xmath10 internally disjoint paths @xmath27 in @xmath0 joining @xmath26 to @xmath10 different vertices @xmath28 of @xmath4 . let @xmath29 denote the number of leaves of @xmath4 . we claim @xmath30 , otherwise there is a vertex @xmath31 such that @xmath32 where @xmath33 . then @xmath34 is a subtree of @xmath0 with more vertices than @xmath4 such that @xmath35 for each @xmath36 and with at most @xmath37 leaves , which contradicts our assumption on the maximality of @xmath4 . because of ore s theorem , we can assume @xmath38 for some @xmath39 . since @xmath4 has @xmath40 leaves , as shown above , there is a vertex @xmath41 of @xmath4 such that @xmath42 . suppose there are vertices @xmath43 and @xmath44 of degree one in @xmath4 such that @xmath45 . since @xmath4 is not a path , there is an edge @xmath46 in the unique @xmath47 path contained in @xmath48 with @xmath49 . let @xmath50 and notice that @xmath51 is a subtree of @xmath0 with @xmath52 , with less than @xmath53 leaves and such that @xmath54 for each @xmath36 . as above , this is a contradiction and therefore no leaves of @xmath4 are adjacent in @xmath0 . notice that @xmath55 , otherwise @xmath56 would be a tree larger than @xmath4 , with the same number of leaves and with @xmath57 for each vertex @xmath58 of @xmath51 . let @xmath17 and @xmath18 be any two leaves of @xmath4 with the property that the vertex @xmath59 lies in the unique @xmath60 path @xmath61 , contained in @xmath4 . orient the edges of @xmath4 in such a way that the corresponding directed tree @xmath62 is outdirected with root @xmath17 ( see fig . [ one ] . ) because of the way the tree @xmath4 was chosen , all vertices of @xmath0 adjacent to @xmath69 or to @xmath18 lie in @xmath4 and therefore @xmath70 . let @xmath71 and @xmath72 be vertices of @xmath4 adjacent to @xmath17 in @xmath0 , if @xmath73 for some vertex @xmath74 of @xmath4 , let @xmath75 . since @xmath76 and @xmath77 are edges of @xmath78 , @xmath79 @xmath80 and @xmath51 is a subtree of @xmath0 with @xmath81 , with less than @xmath82 leaves and such that @xmath35 for each @xmath83 . again , this is a contradiction , therefore @xmath84 . let @xmath107 and @xmath11 be integers with @xmath108 and @xmath109 and @xmath110 be sets of vertices . the complete bipartite graph@xmath111 with bipartition @xmath112 is @xmath10-connected , has @xmath113 vertices and is such that @xmath114 for any vertices @xmath17 and @xmath18 of @xmath0 . nevertheless , if @xmath4 is a spanning tree of @xmath0 , then @xmath115 for some @xmath116 . this shows that the condition in theorem 2 is tight .
we give an ore - type condition sufficient for a graph @xmath0 to have a spanning tree with small degrees and with few leaves . spanning tree . bounded degree . few leaves .
recently , a new member of fullerenes , c@xmath0 , was synthesized by arc - discharge method and purified in bulk quantities@xcite . up to now , it is the smallest fullerene ever discovered . the c@xmath0 molecule is more curved than c@xmath2 because of the adjacent pentagonal rings and the small number of carbon atoms@xcite . this feature suggests stronger electron - phonon interaction and possible higher superconducting transition temperatures in the alkali - doped c@xmath0 solids than those of c@xmath2@xcite . the solid - state nuclear magnetic resonant experiment suggested that the most favorable configuration of c@xmath0 molecule has @xmath1 symmetry @xcite . this confirmed the results of the early _ ab initio _ pseudopotential density functional calculations@xcite , which indicated that the @xmath1 structure is one of the most energetically favorable structures among several possibilities . looked in a direction perpendicular to the six - fold axis(fig.1 ) , the molecule is composed of six parallel layers . on each layer , six carbon atoms lie at the vertexes of a hexagon . the edges of the hexagons of the 1st , 2nd , 5th , 6th layers are parallel , while those of the two hexagons in the middle two layers are turned by an angle of @xmath3 . in addition to this special structure , the six - fold principal axis is distinct form the ordinary five - fold principal axis in the cases of c@xmath2 and c@xmath4 and little attention has been paid to it . in this paper , we discuss both the electronic and structural properties of c@xmath0 molecule , emphasizing the interesting properties brought by this unusual @xmath1 symmetry . we employ a simple and elegant model , extended ssh model , which is successful in treating c@xmath2 and c@xmath4@xcite . the bdeg formalism is performed to obtain the stable lattice configuration and the corresponding electronic states self - consistently . the electronic part of hamiltonian without coulomb interaction can be solved analytically by methods of group theory . we find that the electron densities of homo(@xmath5 ) and lumo(@xmath6 ) are zero at the two middle layers . furthermore , the homo and lumo are mainly confined in the 2nd and 5th layers with a possibility of @xmath7 . the gap between them is considerably small , compared to that of c@xmath2 and c@xmath4 , because the splitting of homo and lumo is due to a long - distance hopping in our case . we find the large splitting of the triplet and singlet excitons due to more localized homo an lumo in comparison with c@xmath2 and c@xmath4 . as a result , the triplet exciton s energy is very small and possible experiments are suggested to test this phenomenum . the geometrical figures of c@xmath0 molecule and polarons are discussed and bond lengths and angles are calculated . the charge density on each layer are given . we find that the differences of charge densities among polarons and molecule are mainly in the 2nd and 5th layers . the following sections are arranged as : in sec.@xmath8 , the extended ssh model is introduced ; in sec.@xmath9 , we analytically solve the electronic part of the hamiltonian without coulomb interaction ; in sec.@xmath10 , the whole hamiltonian is solved self - consistently ; in sec.@xmath11 , results and discussions are presented ; and conclusions are made in the final section . the hamiltonian of the extended ssh model to the c@xmath0 molecule is written as : @xmath12 the first term @xmath13 of eq.(1 ) is the hopping term of @xmath14-electrons . @xmath15(c_{i\sigma}^\dagger c_{j\sigma}+h.c . ) -{\sum_{\langle i , j \rangle}}^\prime t_1(c_{i\sigma}^\dagger c_{j\sigma}+h.c . ) \nonumber\\ & -&{\sum_{\langle i , j \rangle}}^{\prime\prime } t_2(c_{i\sigma}^\dagger c_{j\sigma}+h.c.)\end{aligned}\ ] ] where the three terms describe hopping between the nearest neighbors , the next nearest neighbors and the third neighbors respectively , with the corresponding hopping integrals @xmath16 and @xmath17 . the influence of electron - phonon coupling is only included in the first term of @xmath13 , since those of the last two terms are much smaller than @xmath18 . the third term is important here because it changes the accidental degeneracy of homo and lumo states , which will be explained in the next section . the larger distance terms are neglected , as they are small and do not bring any new effects . the second term of eq.(1 ) is the screened coulomb interaction expressed in the hubbard model . @xmath19 where @xmath20 is the strength of the on - site interaction , and @xmath21 is that between the nearest neighbors . the third term of eq.(1 ) is the elastic energy of the lattice . this term is composed of three parts , @xmath22 where the first part describes the spring energy of bond - length terms , and the next two terms describe the spring energy of the angular terms . @xmath23 are the elastic constants for these different kinds of lattice vibrations . @xmath24 are bond angle deviations from the original angle @xmath25 . e.g. @xmath26 . the first summation is taken up over all nearest pairs . the second is of all interior angles of pentagons . and the third is of all interior angles of hexagons . since there are no enough experiment data to determine the semiempirical parameters , we set the values in the scope of fullerenes such as c@xmath2 and c@xmath4 which can produce reasonable results . we adjust them to give the gap between homo and lumo and the bond length consistent with those of the pseudopotential density functional approach . we take @xmath27 ev , @xmath28 ev / , @xmath29 , @xmath30 ev/ @xmath31 , @xmath32 ev / rad@xmath31 , @xmath33 ev / rad@xmath31 , @xmath34 . in fact , the particular values , except for the value of @xmath17 , would not affect the physics too much . based on the calculation of the third - nearest @xmath14-orbit integral , @xmath17 was estimated to be 0.111@xmath18 . .the analytical solution of @xmath13 without @xmath35 and @xmath17 terms . [ cols=">,>,^,^,^",options="header " , ] the configurations and energy of ground and low excited states are given below , @xmath36 we can see that the ground state @xmath37 is a spin singlet rather than triplet after the long distance hopping @xmath17 and electron - electron interaction are taken into account , which is an improvement of the prediction of simple hckel theory , and there is no need to add the hybridizing of @xmath38 and @xmath14 bonds to investigate the qualitative physics of c@xmath0 as discussed in ref.@xcite . the lowest excited states @xmath39 are spin triplet excitons , which are only about @xmath40ev above the ground states . the energy of singlet exciton @xmath41 is very high , and the splitting between the triplet and singlet excitons is about 0.63ev , which is much larger than 0.2ev in the cases of c@xmath2 and c@xmath4 . this is because the electrons of homo and lumo are more localized , @xmath7 in the 2nd and 5th layers and @xmath42 in the 1st and 6th layers . compared to the c@xmath0 case , the electron densities of homo and lumo in c@xmath2 and c@xmath4 are distributed more uniformly over all sites , so the splitting of the singlet and triplet exciton due to the coulomb interaction is relatively small . the fact that the triplet exciton has low energy can be verified through experiment observation : when illuminated by external light , the c@xmath0 can be excited to the state of singlet exciton and then may transit to triplet exciton through nonradiative decay . the transition rate to ground state is slow because of the different spin configuration and small energy splitting . we predict that the triplet excitons are metastable states , and the phosphorescence phenomena is possible to be observed . the triplet exciton is paramagnetic while the ground state is a diamagnetic singlet . so the magnetic susceptibility is increased upon illumination of the external light . electron spin resonance(esr ) experiment can be performed to detect the triplet exciton states . when esr is performed to a solution sample , the usual unimportant magnetic dipole interaction between electrons becomes crucial because its anisotropy can smear the resonance peak . however , this interaction is decayed rapidly with distance as @xmath43 . the electrons in homo and lumo are mainly uniformly distributed on 12 carbon atoms in two layers , the possibility of their short distance is considerably small compared to other small organic molecule , such as naphthalene . so the resonance peak is possible to be observed . the phone absorption and luminescence spectra experiments can be perform to test the lowest singlet exciton s energy . the shape of the c@xmath0 molecule is an ellipsoid with high aspect ratios . our calculation shows that the distance between the 1st and 6th layers is @xmath44 , and that between the parallel vertical hexagon planes is @xmath45 . the charge density , bond lengths and bond angles of c@xmath0 molecule and ions are calculated and shown in table 2 , 3 and 4 . we note that because of the @xmath1 symmetry of the molecule , the 1st and 6th layers are regular hexagons , with each internal angle @xmath46 and the shortest bonds ( @xmath47 ) @xmath48 . the six vertical hexagons around the equator are slightly deformed due to the drag of the 1st and 6th layers from each side . @xmath49 is @xmath50 longer than @xmath47 and @xmath51 than @xmath52 . the @xmath53 bonds are relatively weaker than those in hexagons , because they couple the hexagons in the two ends with the hexagons around the equator . @xmath53 is @xmath54 longer than @xmath47 . this result is consistent with that of lda except that the @xmath49 is as long as @xmath52 in lda , and we think that our result is more reasonable . the bond lengths of @xmath52 and @xmath49 of ions differ from those of molecule very slightly . but the negative polarons @xmath47 is longer and positive polaron is shorter than the neutral molecule s , while the @xmath53 changes in the opposite way . the reason is that the wavefunctions of the electron in @xmath5 and @xmath6 has opposite sign between two nearest sites connected by @xmath47 , same sign by @xmath53 , and have zero amplitude on the 3rd , 4th layers . according to the contribution f these two states form hopping term , the sites connected by @xmath47 are repulsive , but those connected by @xmath53 are attractive . the angles in the pentagons and hexagons around the equator deviate from @xmath55 and @xmath46 slightly , which justifies the validity of @xmath56 . the bond angles in polarons differ from those of molecule very slightly , so we omit them in table 4 . form c@xmath57 polaron to c@xmath58 , the adding electrons distribute mostly in the 2nd and 5th layers . this is because of the particular charge density distribution of the 18th and 19 t levels . in this paper , we have carefully studied the effect of @xmath1 symmetry on c@xmath0 s electronic properties under the extended ssh model . a small gap between homo(@xmath59 ) and lumo(@xmath60 ) is obtained due to long distance hopping . the large splitting of the spin triplet and singlet lowest excitons , the differences of bond lengthes and electron density between molecule and polarons are discussed as results brought by the more localized homo and lumo and their special symmetries . possilbe experiments are suggested .
the extended ssh model and bogoliubov - de gennes(bdeg ) formalism are applied to investigate the electronic properties and stable lattice configurations of c@xmath0 . we focus the problem on the molecule s unusual @xmath1 symmetry . the electronic part of the hamiltonian without coulomb interaction is solved analytically . we find that the gap between homo and lumo is small due to the long distance hopping between the 2nd and 5th layers . the charge densities of homo and lumo are mainly distributed in the two layers , that causes a large splitting between the spin triplet and singlet excitons . the differences of bond lengths , angles and charge densities among the molecule and polarons are discussed . : c@xmath0 , ssh model , @xmath1 symmetry * contact author * : xiaoqing yu * e - mail * : [email protected]
the price of a european option is typically expressed in terms of the black&scholes _ implied volatility _ @xmath0 ( where @xmath1 denotes the log - strike and @xmath2 the maturity ) , cf . @xcite . since exact formulas for a given model are typically out of reach , an active line or research is devoted to finding asymptotic expansions for @xmath0 , which can be useful in many respects , e.g. for a fast calibration of some parameters of the model . explicit asymptotic formulas for @xmath0 also allow to understand how the parameters affect key features of the volatility surface , such as its slope , and what are the possible shapes that can actually be obtained for a given model . let us mention the celebrated lee moment s formula @xcite and more recent results @xcite . a key problem is to link the implied volatility _ explicitly _ to the distribution of the risk - neutral log - return @xmath3 , because the latter can be computed or estimated for many models . the results of benaim and friz @xcite are particularly appealing , because they connect directly the asymptotic behavior of @xmath0 to the _ tail probabilities _ @xmath4 their results , which are limited to the special regime of extreme strike @xmath5 with fixed maturity @xmath6 , are based on the key notion of _ regular variation _ , which is appropriate when one considers a single random variable @xmath3 ( since @xmath2 is fixed ) . this leaves out many interesting regimes , notably the much studied case of small maturity @xmath7 with fixed strike @xmath1 . in this paper we provide a substantial extension of @xcite : we formulate a suitable generalization of the regular variation assumption on @xmath8 , @xmath9 which , coupled to suitable moment conditions , yields the asymptotic behavior of @xmath0 in essentially _ any regime of small maturity and/or extreme strike _ ( with bounded maturity ) . we thus provide a unified approach , which includes as special cases both the regime of extreme strike @xmath5 with fixed maturity @xmath6 , and that of small maturity @xmath7 with fixed strike @xmath1 . mixed regimes , where @xmath1 and @xmath2 vary simultaneously , are also allowed . this flexibility yields asymptotic formulas for the volatility surface @xmath0 in open regions of the plane . in section [ ch2:sec : examples ] we illustrate our results through applications to popular models , such as carr - wu finite moment logstable model and merton s jump diffusion model . we also discuss heston s model , cf . [ ch2:sec : heston ] . in a separate paper @xcite we consider a stochastic volatility model which exhibits multiscaling of moments , introduced in @xcite . the key point in our analysis is to connect explicitly the asymptotic behavior of the tail probabilities @xmath10 to call and put option prices @xmath11 ( cf . theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] ) . in fact , once the asymptotics of @xmath11 are known , the behavior of the implied volatility @xmath0 can be deduced in a model independent way , as recently shown gao and lee @xcite . we summarize their results in [ ch2:sec : main1 ] ( see theorem [ ch2:th : main1 ] ) , where we also give an extension to a special regime , that was left out from their analysis ( cf . also @xcite ) . the paper is structured as follows . * in section [ ch2:sec : main ] we set some notation and we state our main results . * in section [ ch2:sec : examples ] we apply our results to some popular models . * in section [ ch2:sec : pricetovol ] we prove theorem [ ch2:th : main1 ] , linking option price and implied volatility . * in section [ ch2:sec : probtoprice ] we prove our main results ( theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] ) . * finally , a few technical points have been deferred to the appendix [ ch2:sec : app ] . we consider a generic stochastic process @xmath12 representing the log - price of an asset , normalized by @xmath13 . we work under the risk - neutral measure , that is ( assuming zero interest rate ) the price process @xmath14 is a martingale . european call and put options , with maturity @xmath6 and a log - strike @xmath15 , are priced respectively @xmath16 \ , , \qquad p(\kappa , t ) = { \ensuremath{{\mathrm}e}}[(e^\kappa - e^{x_t})^+ ] \,,\ ] ] and are linked by the _ call - put parity _ relation : @xmath17 as in @xcite , in our results _ we take limits along an arbitrary family ( or `` path '' ) of values of @xmath18_. this includes both sequences @xmath19 and curves @xmath20 , hence we omit subscripts . without loss of generality , we assume that all the @xmath1 s have the same sign ( just consider separately the subfamilies with positive and negative @xmath1 s ) . to simplify notation , we only consider positive families @xmath21 and give results for both @xmath1 and @xmath22 . our main interest is for families of values of @xmath18 such that @xmath23 whenever this holds , one has ( see [ ch2:sec : cpexplained ] ) @xmath24 we stress that gathers many interesting regimes , namely : [ ch2:it : a ] @xmath25 and @xmath26 ( in particular , the case of fixed @xmath27 ) ; [ ch2:it : b ] @xmath25 and @xmath7 ; [ ch2:it : c ] @xmath7 and @xmath28 ( in particular , the case of fixed @xmath29 ) ; [ ch2:it : d ] @xmath7 and @xmath30 . remarkably , while regime needs to be handled separately , regimes -- will be analyzed at once , as special instances of the case `` @xmath1 is bounded away from zero '' . we stress the requirement of _ bounded _ maturity @xmath2 in . some of our arguments can be adapted to deal with cases when @xmath31 , but additional work is needed ( for instance , we assume the boundedness of some exponential moments @xmath32 $ ] , cf . - below , which is satisfied by most models if @xmath2 is bounded , but not if @xmath31 ) . we refer to @xcite for results in the regime @xmath33 . given a model @xmath34 , the _ implied volatility @xmath0 _ is defined as the value of the volatility parameter @xmath35 that plugged into the black&scholes formula yields the given call and put prices @xmath36 and @xmath37 in ( see [ ch2:sec : bs]-[ch2:sec : th : mainmain ] below ) . to avoid trivialities , we focus on families of @xmath18 such that @xmath38 and @xmath39 ( in fact , note that @xmath40 if @xmath41 and , likewise , @xmath42 if @xmath43 ) . _ notation . _ throughout the paper , we write @xmath44 to mean @xmath45 . let us recall a useful standard device ( _ subsequence argument _ ) : to prove an asymptotic relation , such as e.g.@xmath46 , it suffices to show that from every subsequence one can extract a further sub - subsequence along which the given relation holds . as a consequence , _ in the proofs we may always assume that all quantities of interest have a ( possibly infinite ) limit _ , e.g. @xmath47 $ ] and @xmath48 , because this is true along a suitable subsequence . we first focus on families of @xmath18 such that @xmath49 a regime that we call _ atypical deviations_. this is the most interesting case , much studied in the literature , since it includes regimes , and described on page , and also regime provided @xmath50 sufficiently slow . when @xmath25 with fixed @xmath6 , benaim and friz @xcite require the _ regular variation _ of the tail probabilities , i.e.there exist @xmath51 and a slowly varying function is slowly varying if @xmath52 for all @xmath53 . ] @xmath54 such that @xmath55 it is not obvious how to generalize when @xmath2 is allowed to vary , i.e. which conditions to impose on @xmath56 . however , one can reformulate the first relation in simply requiring the existence of @xmath57 for any fixed @xmath53 , by ( * ? ? ? * theorem 1.4.1 ) , and analogously for the second relation in . this reformulation ( in which @xmath56 is not even mentioned ! ) turns out to be the right condition to impose in the general context that we consider , when @xmath2 is allowed to vary . we are thus led to the following : [ ch2:ass : rv ] the family of values of @xmath18 with @xmath58 , @xmath6 satisfies , and for every @xmath59 the following limit exists in @xmath60 $ ] : @xmath61 where limits are taken along the given family of values of @xmath18 . moreover @xmath62 depending on the regime of @xmath1 , we will also need one of the following _ moment conditions_. * given @xmath63 , the first moment condition is @xmath64 < \infty \,,\end{gathered}\ ] ] along the given family of values of @xmath18 . when @xmath65 , it is enough to require that @xmath66 < \infty \,,\end{gathered}\ ] ] because @xmath67 is a submartingale and hence @xmath32 \le { \ensuremath{{\mathrm}e}}[e^{(1+\eta)x_t}]$ ] . * given @xmath63 , the second moment condition is @xmath68 < \infty \,,\end{gathered}\ ] ] along the given family of values of @xmath18 . note that for @xmath69 this simplifies to @xmath70 \le 1 + c \kappa^2 \,.\ ] ] we are ready to state our main results , which express the asymptotic behavior of option prices and implied volatility explicitly in terms of the tail probabilities . due to different assumptions , we first consider right - tail asymptotics . [ ch2:th : main2b ] consider a family of values of @xmath18 with @xmath58 , @xmath6 such that hypothesis [ ch2:ass : rv ] is satisfied by the right tail probability @xmath8 . let the moment condition hold for _ every @xmath71 _ , or alternatively let it hold only for _ some @xmath71 _ but in addition assume that @xmath72 then @xmath73 special case : if @xmath74 , assumption can be relaxed to @xmath75 and relations - simplify to @xmath76 let the moment condition hold for _ every @xmath71 _ , or alternatively let it hold only for _ some @xmath71 _ but in addition assume . then @xmath77 next we turn to left - tail asymptotics . the assumptions in this case turn out to be sensibly weaker than those for right - tail . for instance , the left - tail condition @xmath78 < \infty$ ] required in ( * ? ? ? * theorem 1.2 ) is not needed , which allows to treat the case of a _ polynomially decaying _ left tail , like in the carr - wu model described in section [ ch2:sec : examples ] . [ ch2:th : main2bl ] consider a family of values of @xmath18 with @xmath58 , @xmath6 such that hypothesis [ ch2:ass : rv ] is satisfied by the left tail probability @xmath79 . * with no moment condition and no extra assumption on @xmath80 , one has @xmath81 special case : if @xmath82 , relations - simplify to @xmath83 * let the moment condition hold for _ every @xmath71 _ , or alternatively let it hold only for _ some @xmath71 _ but in addition assume that @xmath84 then @xmath85 we prove theorems [ ch2:th : main2b ] and [ ch2:th : main2bl ] in [ ch2:sec : main2bproof ] below . the key step is to link the option prices @xmath36 , @xmath37 to the tail probabilities @xmath8 , @xmath79 , exploiting hypothesis [ ch2:ass : rv ] . once this is done , the asymptotic behavior of the implied volatility @xmath0 can be deduced using the model independent results of @xcite , that we summarize in [ ch2:sec : main1 ] . [ ch2:rem : moment ] the `` special case '' conditions @xmath86 are automatically fulfilled in the small maturity regime @xmath7 with fixed strike @xmath87 . in this case , one can use the simplified formulas - and - . next we focus on the regime when @xmath88 and @xmath30 sufficiently fast , so that the tail probability @xmath8 , resp . @xmath79 has a strictly positive limit and condition is violated . we call this regime _ typical deviations_. this includes the basic regime of fixed @xmath89 and @xmath90 . mixed regimes , when @xmath30 and @xmath7 simultaneously , are also interesting , e.g. to interpolate between the at - the - money @xmath91 and out - of - the - money @xmath92 cases , which can be strikingly different as @xmath7 ( see @xcite ) . we make the following natural assumption . [ ch2:ass : smalltime ] there is a positive function @xmath93 with @xmath94 such that @xmath95 converges in law as @xmath90 to some random variable @xmath96 : @xmath97{d } y \,.\ ] ] we refer to remark [ rem : equivalent ] below for concrete ways to check hypothesis [ ch2:ass : smalltime ] . let us stress that is a condition on the tail probabilities , since it can be reformulated as @xmath98 for all @xmath99 with @xmath100 . if the support of the law of @xmath96 is unbounded from above and below ( as it is usually the case ) , the limits in are strictly positive for every @xmath101 . the appropriate moment condition in this regime turns out to be with @xmath102 , i.e. @xmath103 < \infty \,.\ ] ] lastly , we introduce some notation . denote by @xmath104 and @xmath105 respectively the density and distribution function of a standard gaussian ( see below ) , and define the function @xmath106 as we show in [ ch2:eq : backg ] below , @xmath107 is a smooth and strictly decreasing bijection from @xmath108 to @xmath108 . its inverse @xmath109 is also smooth , strictly decreasing and satisfies @xmath110 we can finally state the following result , linking option prices and implied volatility to tail probabilities in the regime of typical deviations . [ ch2:th : main2a ] assume that hypothesis [ ch2:ass : smalltime ] is satisfied , and moreover the moment condition holds . then the random variable in satisfies @xmath111=0 $ ] . fix @xmath112 with @xmath113 , resp . @xmath114 . for any family of @xmath18 with @xmath115 the asymptotic behavior of option prices is given by @xmath116 \ , , \qquad \text{resp . } \qquad p(-\kappa , t ) \sim \gamma_t \ , { \ensuremath{{\mathrm}e}}[(y+a)^- ] \,,\end{aligned}\ ] ] and correspondingly the implied volatility is given by @xmath117}{a } \big ) } & \text { if \ $ \displaystyle a > 0 $ } \ , , \\ \rule{0em}{1.7em}\displaystyle \sqrt{2\pi } \ , { \ensuremath{{\mathrm}e}}[y^\pm ] & \text { if \ $ \displaystyle a = 0 $ } \ , . \end{cases}\end{aligned}\ ] ] [ rem : equivalent ] hypothesis [ ch2:ass : smalltime ] can be easily checked when the characteristic function of @xmath3 is known , because , by lvy continuity theorem , the convergence in distribution is equivalent to the pointwise convergence @xmath118 \to { \ensuremath{{\mathrm}e}}[e^{i u y}]$ ] for every @xmath119 . we will see concrete examples in subsections [ ch2:sec : cw ] ( carr - wu model ) and [ ch2:sec : merton ] ( merton s model ) . another interesting case is that of _ assume that @xmath120 , where @xmath121 evolves according to the stochastic differential equation @xmath122 where @xmath123 is a brownian motion and @xmath124 is a positive adapted process , representing the volatility , possibly correlated with @xmath125 . under the mild assumption that @xmath126 a.s . , where @xmath127 is a _ constant _ , one can show that hypothesis [ ch2:ass : smalltime ] holds with @xmath128 and @xmath129 , see appendix [ app : diffusions ] . interestingly , plugging @xmath129 into yields @xmath130 ( see appendix [ app : diffusions ] ) . consequently , if the moment condition holds , we can apply theorem [ ch2:th : main2a ] , getting @xmath131 along any parabolic curve @xmath132 . this result is consistent with recent results by pagliarani and pascucci @xcite , who go beyond first - order asymptotics . the proof of theorem [ ch2:th : main2a ] is given in [ ch2:sec : proofth : main2a ] below . the asymptotic behavior of option prices follows easily from the convergence in distribution , because the needed uniform integrability is ensured by the moment condition . the asymptotic behavior of implied volatility can again by deduced from the option prices asymptotics in a model independent way , that we now describe . whenever the option prices @xmath36 or @xmath133 vanish , they determine the asymptotic behavior of the implied volatility through _ explicit universal formulas_. these are summarized in the following theorem ( of which we give in section [ ch2:sec : pricetovol ] a self - contained proof ) , which gathers results from the recent literature . [ ch2:th : main1 ] consider an arbitrary family of values of @xmath18 with @xmath134 and @xmath135 , such that @xmath136 , resp . @xmath137 . * case of @xmath1 bounded away from zero ( i.e. @xmath138 ) . @xmath139 * case of @xmath50 , with @xmath140 . @xmath141 where the function @xmath142 is defined in - . * case of @xmath89 . @xmath143 we stress that theorem [ ch2:th : main1 ] allows to derive immediately all the asymptotic relations for the implied volatility @xmath144 appearing in theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] from the corresponding relations for the option prices @xmath36 and @xmath133 . the main part of theorem [ ch2:th : main1 ] is equation , which was recently proved by gao and lee @xcite extending previous results of lee @xcite , roper and rutkowski @xcite , benaim and friz @xcite and gulisashvili @xcite . as a matter of fact , gao and lee prove much more than , providing explicit estimates for the error beyond first order asymptotics . equation is a new contribution of the present paper . in fact , @xcite assume that @xmath145 ( cf . equation ( 4.2 ) therein ) , which excludes the regimes with @xmath30 `` fast enough '' . the relevance of such regimes has been recently shown in @xcite , where the special case @xmath146 is considered ( see ( * ? ? ? * theorem 3.1 ) ) . [ ch2:rem : simple ] relation provides an _ interpolation between the at - the - money and out - of - the - money regimes _ , described by and . let us be more explicit . using , formula can be rewritten as follows : @xmath147{0em}{3.5em}\displaystyle \frac{\kappa}{d^{-1}(a ) \ , \sqrt{t } } & \text { if \ $ \displaystyle\frac{c(\kappa , t)}{\kappa } \to a \in ( 0,\infty)$ } \ , ; \\ \rule[-1.4em]{0em}{3.5em}\displaystyle \sqrt{2\pi } \ , \frac{c(\kappa , t)}{\sqrt{t } } & \text { if \ $ \displaystyle\frac{c(\kappa , t)}{\kappa } \to \infty$ , \ or if \ $ \kappa = 0 $ } \ , , \end{cases}\end{aligned}\ ] ] and analogously for @xmath148 , just replacing @xmath36 by @xmath133 . note that the last line in matches with the at - the - money regime . in order to see how matches with the out - of - the - money regime , it suffices to note that whenever @xmath149 , resp . @xmath150 , formula can be rewritten as @xmath151 and this coincides with the first line of when @xmath30 slowly enough , namely @xmath152 theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] are useful because their assumptions , involving asymptotics for the tail probabilities @xmath8 and @xmath79 , can be directly verified for many concrete models ( see section [ ch2:sec : examples ] for some examples ) . the difference between the regimes of typical and atypical deviations can be described as follows : * for typical deviations , the key assumption is hypothesis [ ch2:ass : smalltime ] , which concerns the _ weak convergence _ of @xmath3 , cf . - ; * for atypical deviations , the key assumption is hypothesis [ ch2:ass : rv ] , which concerns the _ large deviations _ properties of @xmath3 , cf . - . in particular , it is worth stressing that hypothesis [ ch2:ass : rv ] requires sharp asymptotics only for _ the logarithm of the tail probabilities @xmath153 and @xmath154 _ , and not for the tail probabilities themselves , which would be a considerably harder task ( out of reach for many models ) . as a consequence , hypothesis [ ch2:ass : rv ] can often be checked through the celebrated _ grtner - ellis theorem _ * theorem 2.3.6 ) , which yields sharp asymptotics on @xmath153 and @xmath154 under suitable conditions on the moment generating function of @xmath3 . in this section we show the relevance of our main theoretical results , deriving asymptotic expansions of the implied volatility for carr - wu finite moment logstable model ( [ ch2:sec : cw ] ) and merton s jump diffusion model ( [ ch2:sec : merton ] ) . the case of heston s model is briefly discussed in [ ch2:sec : heston ] . our results can also be applied to a stochastic volatility model , recently introduced in @xcite , which exhibits multiscaling of moments . even though no closed expression is available for the moment generating function of the log - price , the tail probabilities can be estimated _ explicitly _ , as we show in a separate paper @xcite . this leads to precise asymptotics for the implied volatility , thanks to theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] . carr and wu @xcite consider a model where the log - strike @xmath3 has characteristic function @xmath155= e^{t\left[iu\mu - |u|^{\alpha}\sigma^{\alpha } \left(1 + i \ , { \mathrm{sign}}(u ) \tan ( \frac{\pi \alpha}{2 } ) \right ) \right ] } \,,\ ] ] where @xmath156 , @xmath157 $ ] and we fix @xmath158 to work in the risk - neutral measure , cf . * proposition 1 ) . the moment generating function of @xmath3 is @xmath159 = \begin{cases } \displaystyle e^{[\lambda \mu - \frac{(\lambda\sigma)^{\alpha}}{\cos ( \frac{\pi \alpha}{2 } ) } ] \ , t } & \text{if } \lambda \ge 0 \ , , \\ + \infty & \text{if } \lambda < 0 \ , . \end{cases}\ ] ] note that as @xmath160 one recovers black&scholes model with volatility @xmath161 , cf . [ ch2:sec : bs ] below . applying theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] , we give a _ complete characterization _ of the volatility smile asymptotics with bounded maturity . this includes , in particular , the regimes of extreme strike ( @xmath5 with fixed @xmath6 ) and of small maturity ( @xmath7 with fixed @xmath1 ) . [ th : cw ] the following asymptotics hold . * atypical deviations . consider any family of @xmath18 with @xmath21 , @xmath6 such that @xmath162 ( this includes the regimes , , on page , and part of regime . ) then one has the right - tail asymptotics @xmath163 the corresponding left - tail asymptotics are given by @xmath164 which can be made more explicit distinguishing different regimes : @xmath165 * typical deviations . for any family of @xmath18 with @xmath166 one has @xmath167}{a } \big ) } & \text{if } a > 0 \ , , \\ \displaystyle \rule{0pt}{1.6em } \sqrt{2\pi } \ , \sigma \,{\ensuremath{{\mathrm}e}}[y^\pm ] & \text{if } a = 0 \ , . \end{cases}\ ] ] [ ch2:rem : joint ] the fact that relations and hold for _ any _ family of @xmath18 satisfying yields interesting consequences . we claim that , for any @xmath168 and @xmath169 , there exists @xmath170 such that the following inequalities hold _ for all @xmath18 in the region @xmath171 _ : @xmath172 similar inequalities can be deduced from - and . relation gives a _ uniform approximations _ of the volatility surface @xmath0 in open regions of the plane @xmath18 . the proof of is simple : assume by contradiction that there exist @xmath173 such that for every @xmath174 relation _ fails _ for some @xmath175 . extracting a subsequence , the family @xmath176 satisfies but not , contradicting theorem [ th : cw ] . let @xmath96 denote a random variable with characteristic function @xmath177 = e^{-|u|^\alpha\left(1+i \ , { \mathrm{sign}}(u ) \tan ( \frac{\pi \alpha}{2 } ) \right ) } \,,\ ] ] i.e. @xmath96 has a strictly stable law with index @xmath178 and skewness parameter @xmath179 , and @xmath111 = 0 $ ] . if we set @xmath180 it follows by that @xmath181 has the same distribution as @xmath96 , because @xmath182 = { \ensuremath{{\mathrm}e}}[e^{i u y } ] = e^{-|u|^\alpha\left(1+i \ , { \mathrm{sign}}(u ) \tan ( \frac{\pi \alpha}{2 } ) \right ) } \,.\ ] ] it follows by that @xmath183{d } \sigma y \,,\ ] ] hence hypothesis [ ch2:ass : smalltime ] is satisfied with @xmath184 . note that @xmath113 and @xmath114 for all @xmath185 , because the density of @xmath96 is strictly positive everywhere . the right tail of @xmath96 has a super - exponential decay : as @xmath186 @xmath187 cf . * property 1 and references therein ) . on the other hand the left tail is polynomial : there exists @xmath188 such that @xmath189 recalling that @xmath190 and @xmath191 , by we can write @xmath192 hence we can transfer the estimates and to @xmath3 . henceforth we consider separately the regimes of atypical deviations , and that of typical deviations . note that it is easy to check that is equivalent to . let us fix an arbitrary family of values of @xmath18 satisfying . then @xmath193 ( because @xmath194 ) , hence @xmath195 by , and we then obtain @xmath196 let us now check the assumptions of theorem [ ch2:th : main2b ] . * the first relation in shows that hypothesis [ ch2:ass : rv ] is satisfied by the right tail @xmath8 , with @xmath197 . note that @xmath198 for all @xmath199 , since @xmath194 , hence also condition is satisfied . * condition is satisfied because holds for all @xmath200 and @xmath71 , by . * it remains to check condition . as we prove below , for all @xmath201 and @xmath202 there are constants @xmath203 , depending on @xmath204 and on the parameters @xmath205 , such that for all @xmath206 and @xmath21 the following inequality holds : @xmath207 \le a \bigg ( \bigg ( \frac{t^{1/\alpha}}{\kappa } \bigg)^{b } + c \bigg ) \,.\ ] ] in particular , since @xmath208 by assumption , condition is satisfied . applying theorem [ ch2:th : main2b ] , since @xmath74 by the first relation in , the asymptotic behavior of @xmath0 is given by , which by coincides with . next we want to apply theorem [ ch2:th : main2bl ] . by the second relation in , hypothesis [ ch2:ass : rv ] is satisfied by the left tail @xmath79 , with @xmath209 . if @xmath1 is bounded away from zero , the asymptotic behavior of @xmath0 is given by , which by yields precisely . if @xmath30 we can not apply directly theorem [ ch2:th : main2bl ] , because the moment condition is satisfied only for some @xmath71 , and condition is not satisfied , since @xmath209 . however , we can show that still holds by direct estimates . by @xmath210 \ge { \ensuremath{{\mathrm}e}}[(e^{-\kappa } - e^{x_t } ) { { \sf 1}}_{\{x_t < -2\kappa\ } } ] \ge ( e^{-\kappa } - e^{-2\kappa } ) f_t(-2\kappa ) \,,\ ] ] and since @xmath211 , we can write by ( recall that @xmath50 ) @xmath212 next we give a matching upper bond on @xmath133 . since @xmath213 eventually ( recall that @xmath214 , hence @xmath215 ) , by and we obtain , for all @xmath216 @xmath217 for some @xmath218 . then by fubini s theorem @xmath219 = { \ensuremath{{\mathrm}e}}\bigg [ \int_\kappa^\infty e^{-x } { { \sf 1}}_{\{x < -x_t\ } } { \text{\rm d}}x \bigg ] = \int_\kappa^\infty e^{-x } \ , f_t ( -x ) \ , { \text{\rm d}}x \\ & = \kappa \int_1^\infty e^{-\kappa y } \ , f_t(-\kappa y ) \ , { \text{\rm d}}y \le c ' \kappa \ , \frac{t}{\kappa^\alpha } \int_1^\infty \frac{1}{y^\alpha } { \text{\rm d}}y = : c '' \kappa \ , \frac{t}{\kappa^\alpha } \ , , \end{split}\ ] ] hence @xmath220 this relation , together with , yields @xmath221 since @xmath208 , this shows that we are in the regime when @xmath30 and @xmath222 . we can thus apply equation in theorem [ ch2:th : main1 ] , which recalling remark [ ch2:rem : simple ] simplifies as the first line in ( with @xmath133 instead of @xmath36 ) , yielding @xmath223 hence is proved also when @xmath50 ( cf . ) . let us fix an arbitrary family of values of @xmath18 satisfying . relation for @xmath224 shows that condition is satisfied , and hypothesis [ ch2:ass : smalltime ] holds by . we can then apply theorem [ ch2:th : main2a ] , and relation gives precisely . since @xmath225 if @xmath226 and @xmath227 if @xmath228 , we have @xmath229 for all @xmath230 . if @xmath231 are such that @xmath232 , young s inequality @xmath233 yields @xmath234 noting that @xmath235 for @xmath236 , by hlder s inequality , and denoting by @xmath237 a suitable constant depending only on @xmath238 , we can write @xmath239 given @xmath240 , we fix @xmath241 such that @xmath242 . ( note that @xmath243 depends only on @xmath244 . ) moreover , it follows by that @xmath245 = ( \sigma t^{1/\alpha})^b \ , { \ensuremath{{\mathrm}e}}[|y|^b ] \ , \big ( 1+{{\ensuremath{\mathcal o } } } ( t^{b(1 - 1/\alpha ) } ) \big ) \,,\ ] ] and note that @xmath246 < \infty$ ] , because @xmath96 has finite moments of all orders strictly less than @xmath178 , cf . * property 1 ) . since for @xmath65 one has @xmath247 \le { \ensuremath{{\mathrm}e } } [ e^{q(1+\eta)x_t } ] < \infty$ ] , by , relation is proved . consider a model @xcite where the log - return @xmath3 has an infinitely divisible distribution , whose moment generating function is given by @xmath248 = \exp \bigg ( t \left\{z\mu+\frac{1}{2}z^2\sigma^2 + \lambda\left(e^{z\alpha+z^2\frac{\delta^2}{2}}-1\right)\right\ } \bigg ) \ , , \qquad \forall z \in { \mathbb{c}}\,,\ ] ] where @xmath249 and @xmath250 are fixed parameters . the asymptotic behavior of @xmath0 has been studied by many authors . the case of fixed @xmath135 and @xmath25 was derived by benaim and friz @xcite using saddle point methods ( for the detailed computation see @xcite , @xcite ) . the case of fixed @xmath58 and @xmath7 follows by @xcite , while the mixed regime of @xmath7 , @xmath30 with @xmath146 was considered in @xcite . applying our results , we can complete the picture , providing general formulas which interpolate between all these regimes , cf . theorem [ ch2:th : mertonsmile ] . let us define two functions @xmath251 and @xmath252 as @xmath7 as follows : @xmath253 which will separate different behaviors as @xmath7 . we stress that @xmath254 is precisely the scaling considered in @xcite . let us also define @xmath255 by @xmath256 note that @xmath257 is continuous and piecewise quadratic : more precisely , by explicit computation , @xmath258 for all @xmath259 , with @xmath260 . it follows that @xmath261 the role of the function @xmath257 is explained by the following lemma , proved in appendix [ sec : lemma ] . [ th : lemma ] for every fixed @xmath262 , as @xmath7 one has @xmath263 moreover , if either @xmath7 and @xmath264 , or if @xmath26 and @xmath25 , one has @xmath265 we are now ready to state our main result for merton s model ( see figure [ ch2:fig : merton ] ) . for merton s model ( with parameters @xmath266 , @xmath267 , @xmath268 , @xmath269 ) in the regime @xmath270 for @xmath271 . asymptotic volatility _ is given by our formula , while the _ simulated volatility _ is obtained using standard computational packages . ] [ ch2:th : mertonsmile ] consider a family of values of @xmath18 with @xmath272 and @xmath6 . 1 . [ svm : a ] if @xmath7 and @xmath273 , then @xmath274 which can be rewritten more explicitly as follows : @xmath275 2 . [ svm : e ] if @xmath276 and @xmath277 , or if @xmath278 and @xmath279 , then @xmath280 we have to prove relation ( which is equivalent to , by extracting subsequences ) and relation . we distinguish different subcases . assume first that @xmath88 with @xmath281 . by extracting sub - sequences , assume that @xmath282 . note that @xmath283{\ , d\ , } n(0 , \sigma^2)$ ] , because @xmath284 \to e^{-\frac{u^2}{2\sigma^2}}$ ] for every @xmath285 , as one checks by . we then apply theorem [ ch2:th : main2a ] with @xmath128 , because the moment condition for @xmath69 follows by ( see also ) . relation yields @xmath286 since @xmath287 ( see in appendix [ app : diffusions ] ) . this matches with the first line of . next we assume that @xmath88 with @xmath288 . applying ( * ? ? ? * proposition 2.3 ) , we can write @xmath289 \,+\ , c \ , t \,,\ ] ] with @xmath290 , where @xmath291 denotes the lvy measure of @xmath292 . the first term is the usual black&scholes price of a call option with @xmath293 : applying below with @xmath294 and @xmath295 , together with and , we get @xmath296 writing @xmath297 as @xmath298 , we get @xmath299 the inequalities @xmath300 yield @xmath301 ( the additive constant @xmath302 is irrelevant , since @xmath303 ) hence @xmath304 it is easy to check that the asymptotic equality @xmath305 holds when @xmath306 . it follows that : * in case @xmath307 we have @xmath308 * in case @xmath309 we have @xmath310 ( note that when @xmath306 both relations and hold . ) we can deduce the asymptotic behavior of @xmath0 applying relation ( note that @xmath311 , since @xmath293 ) which , by remark [ ch2:rem : simple ] , reduces to the first line of , i.e.@xmath312 plugging - into this relation yields the first and second line of . next we assume that @xmath7 with @xmath313 , for some fixed @xmath71 . we claim that @xmath314 which plugged into proves the second line of ( since @xmath315 in this regime ) . when @xmath58 is fixed , ( a stronger version of ) relation was proved by figueroa lpez and forde in @xcite . for the general case , we fix @xmath262 and we apply relation , which yields the lower bound @xmath316 \ge ( e^{a \ , { \boldsymbol{\kappa}}_2(t)}-e^{\kappa } ) \ , { \ensuremath{{\mathrm}p}}(x_t > a \ , { \boldsymbol{\kappa}}_2(t ) ) & \sim e^{a \ , { \boldsymbol{\kappa}}_2(t ) - f(a ) \log\frac{1}{t}(1+o(1 ) ) } \\ & \sim e^{- f(a ) \log\frac{1}{t}(1+o(1 ) ) } \ , , \end{split}\ ] ] because @xmath317 . for an upper bound , we recall that @xmath318 and @xcite @xmath319 then , for every fixed @xmath320 , using , and caucy - schwarz we can write @xmath321 \\ & \le e^{b \ , { \boldsymbol{\kappa}}_2(t ) - \log \frac{1}{t}(1+o(1 ) ) } + { \ensuremath{{\mathrm}e}}[e^{2x_t}]^{\frac{1}{2 } } \ , { \ensuremath{{\mathrm}p}}(x_t > b \ , { \boldsymbol{\kappa}}_2(t))^{\frac{1}{2 } } \\ & \sim e^{- \log \frac{1}{t}(1+o(1 ) ) } + c_1 \ , e^{- \frac{1}{2}f(b ) \log \frac{1}{t}(1+o(1))}\ , , \end{split}\ ] ] where in the last step we used @xmath322^{\frac{1}{2 } } \le c_1 $ ] for some constant @xmath323 , since @xmath322 \to 1 $ ] as @xmath7 by . by , we can fix @xmath324 large enough so that @xmath325 , hence @xmath326 , and for every @xmath169 we can choose @xmath327 small enough such that @xmath328 , hence @xmath329 by . altogether , relation is proved . let us proceed with the regime @xmath7 and @xmath330 for some @xmath262 . relation shows that such a family of @xmath18 satisfies hypothesis [ ch2:ass : rv ] with @xmath331 ( we stress that @xmath262 is fixed throughout this argument , hence @xmath332 can depend on @xmath333 ) . since the moment condition is clearly satisfied by , we can apply theorem [ ch2:th : main2b ] : relation , coupled with , proves the third line of . finally , it remains to prove , hence we assume that either @xmath7 and @xmath264 , or @xmath26 and @xmath25 . relation shows that hypothesis [ ch2:ass : rv ] holds with @xmath334 . by theorem [ ch2:th : main2b ] , relation yields , completing the proof of theorem [ ch2:th : mertonsmile ] . given the parameters @xmath335 and @xmath336 $ ] , the heston model @xcite is a stochastic volatility model @xmath121 defined by the following sdes @xmath337 where @xmath338 and @xmath339 are standard brownian motions with @xmath340 . note that @xmath341 displays explosion of moments , i.e.@xmath342 = \infty$ ] for @xmath343 large enough . in general , for any fixed @xmath344 one can define the explosion moment @xmath345 as @xmath346 < \infty\}\,,\ ] ] so that @xmath347 < \infty$ ] for @xmath348 while @xmath347 = \infty$ ] for @xmath349 ( in the case of heston s model , one has @xmath347 = \infty$ ] also for @xmath350 ) . the behavior of the explosion moment @xmath345 is described in the following lemma , proved below . [ ch2:th : lemmaheston ] if @xmath351 , then @xmath352 for every @xmath353 . + if @xmath354 , then @xmath355 for every @xmath356 . moreover , as @xmath90 @xmath357 where @xmath358 the asymptotic behavior of the implied volatility @xmath0 is known in the regimes of large strike ( with fixed maturity ) and small maturity ( with fixed strike ) . * in @xcite , benaim and friz show that for fixed @xmath6 , when @xmath359 @xmath360 based on the estimate ( cf . also @xcite ) @xmath361 * in @xcite , forde and jacquier have proved that for any fixed @xmath58 , as @xmath90 @xmath362 where @xmath363 is the legendre transform of the function @xmath364 given by @xmath365 where @xmath366 is the constant in . their analysis is based on the estimate @xmath367 obtained by showing that the log - price @xmath12 in the heston model satisfies a large deviations principle as @xmath90 , with rate @xmath368 and good rate function @xmath369 . we first note that the asymptotics and follow easily from our theorem [ ch2:th : main2b ] , plugging the estimates and into relations and , respectively . we also observe that the estimates and match , in the following sense : if we take the limit @xmath7 of the right hand side of ( i.e. we first let @xmath370 and then @xmath371 in @xmath372 ) , we obtain @xmath373 if , on the other hand , we take the limit @xmath374 of the right hand side of ( i.e. we first let @xmath371 and then @xmath370 in @xmath372 ) , since @xmath375 , as @xmath376 , hence the slope of @xmath369 converges to @xmath366 as @xmath25 . ] we obtain @xmath377 which coincides with . analogously , also the estimates and match . it is then natural to conjecture that , for _ any _ family of values of @xmath18 such that @xmath378 and @xmath90 jointly , one should have @xmath379 where @xmath366 is the constant in . if this holds , applying theorem [ ch2:th : main2b ] , relation yields @xmath380 providing a smooth interpolation between and . [ ch2:rem : jointh ] if holds for any family of values of @xmath18 with @xmath25 and @xmath7 , it follows that for every @xmath169 there exists @xmath381 such that the following inequalities hold : @xmath382 _ for all @xmath18 in the region @xmath383 _ , as it follows easily by contradiction ( cf . remark [ ch2:rem : joint ] for a similar argument ) . given any number @xmath384 we define the explosion time @xmath385 as @xmath386 < \infty \ } \,.\ ] ] note that if @xmath387 then @xmath388 . by @xcite ( see also @xcite ) @xmath389 where @xmath390 observe that if @xmath351 , then @xmath391 and @xmath392 , which implies @xmath393 for every @xmath384 , or equivalently @xmath352 for every @xmath6 . on the other hand , since @xmath394 we observe that if @xmath395 , then @xmath396 as @xmath397 , which implies @xmath398 in particular this leads to the conclusion that , if @xmath399 , then @xmath400 where @xmath366 was defined in . it remains to study the case @xmath401 , in which @xmath402 for every @xmath403 . we have two possibilities : if @xmath404 then @xmath405 when @xmath406 , and so by @xmath407 on the other hand , if @xmath408 , then @xmath409 when @xmath410 and so @xmath411 finally if @xmath412 , @xmath413 , and so @xmath414 in all the cases we obtain @xmath415 , in agreement with . in this section we prove theorem [ ch2:th : main1 ] . we start with some background on black&scholes model and on related quantities . we let @xmath416 be a standard gaussian random variable and denote by @xmath417 and @xmath418 its density and distribution functions : @xmath419 the mills ratio @xmath420 is defined by @xmath421 the next lemma summarizes the main properties of @xmath422 that will be used in the sequel . [ ch2:th : mills ] the function @xmath422 is smooth , strictly decreasing , strictly convex and satisfies @xmath423 since @xmath424 and @xmath417 is an analytic function , @xmath422 is also analytic . since @xmath425 , one obtains @xmath426 recalling that @xmath427 , these relations already show that @xmath428 and @xmath429 for all @xmath430 . for @xmath431 , the following bounds hold ( * ? ? ? * eq . ( 19 ) ) , ( * ? ? ? 1.5 ) : @xmath432 applying yields @xmath429 and @xmath433 for all @xmath431 , hence . we recall that the smooth function @xmath142 was introduced in . since @xmath434 @xmath435 is a strictly decreasing bijection ( note that @xmath436 and @xmath437 ) . its inverse @xmath438 is then smooth and strictly decreasing as well . writing @xmath439 , it follows by that @xmath440 as @xmath441 , hence @xmath442 it follows easily that @xmath443 satisfies . let @xmath444 be a standard brownian motion . the black&scholes model is defined by a risk - neutral log - price @xmath445 , where the parameter @xmath156 represents the volatility . the black&scholes formula for the price of a normalized european call is @xmath446 , where @xmath1 is the log - strike , @xmath2 is the maturity and we define @xmath447 = \begin{cases } ( 1-e^\kappa)^+ & \text{if } v = 0 \ , , \\ \rule{0pt}{1.2em}\phi(d_1 ) - e^\kappa \phi(d_2 ) & \text{if } v > 0 \ , , \end{cases}\ ] ] where @xmath418 is defined in , and we set @xmath448 note that @xmath449 is a continuous function of @xmath450 . since @xmath451 , for all @xmath452 one easily computes @xmath453 hence @xmath449 is strictly increasing in @xmath454 and strictly decreasing in @xmath1 ( see figure [ ch2:fig : bs ] ) . it is also directly checked that for all @xmath15 and @xmath455 one has @xmath456 in the following key proposition , proved in appendix [ ch2:sec : app : bs ] , we show that when @xmath21 the black&scholes call price @xmath457 vanishes precisely when @xmath458 or @xmath459 ( or , more generally , in a combination of these two regimes , when @xmath460 ) . we also provide sharp estimates on @xmath457 for each regime ( weaker estimates on @xmath461 could be deduced from theorems [ ch2:th : main2b ] and [ ch2:th : main2bl ] ) . [ ch2:th : bs ] for any family of values of @xmath450 with @xmath21 , @xmath452 , one has @xmath462 that is , @xmath463 if and only if from any subsequence of @xmath450 one can extract a sub - subsequence along which either @xmath459 or @xmath458 . moreover : * if @xmath464 , then @xmath465 * if @xmath458 , then @xmath466 where @xmath104 and @xmath467 are defined in and . since the function @xmath468 is a bijection from @xmath469 to @xmath470 , it admits an inverse function @xmath471 , defined by @xmath472 by construction , @xmath473 is a strictly increasing bijection from @xmath470 to @xmath469 . we will mainly focus on the case @xmath21 , for which @xmath474 . , for @xmath475 $ ] and @xmath476 $ ] . ] consider an arbitrary model , with a risk - neutral log - price @xmath12 , and let @xmath36 be the corresponding price of a normalized european call option , cf . . since @xmath477 is a convex function , one has @xmath478 - e^\kappa)^+ = ( 1-e^\kappa)^+$ ] by jensen s inequality ; since @xmath479 , one has @xmath480 = 1 $ ] . consequently , by , we have the following relation between the _ implied volatility _ @xmath0 ( defined in [ ch2:sec : setting ] ) and @xmath481 : @xmath482 relation allows to reformulate theorem [ ch2:th : main1 ] more transparently in terms of the function @xmath483 . inspired by , we define @xmath484 by @xmath485 consider an arbitrary family of values of @xmath486 , such that either @xmath21 , @xmath487 and @xmath488 , or alternatively @xmath489 , @xmath490 and @xmath491 ( with @xmath403 as in ) . then , in light of , we can write the following : * if @xmath1 bounded away from zero @xmath492 , relation is equivalent to @xmath493 * if @xmath1 is bounded away from infinity @xmath494 , relations and are equivalent to @xmath495 where @xmath443 is the inverse of the function @xmath435 defined in , and satisfies . the proof of theorem [ ch2:th : main1 ] is now reduced to proving relations and . we first show that we can assume @xmath21 , by a symmetry argument . recalling and , for all @xmath496 and @xmath497 we have @xmath498 where @xmath403 is defined in . as a consequence , in the case @xmath489 , replacing @xmath1 by @xmath22 and @xmath499 by @xmath500 in the first line of , one obtains the second line of . performing the same replacements in the first line of yields @xmath501 which is slightly different with respect to the third line of . however , the discrepancy is only apparent , because we claim that @xmath502 . this is checked as follows : if @xmath30 , then @xmath503 ; if @xmath504 , since @xmath491 by assumption , the first relation in yields @xmath505 , as required . ( see the lines following below for more details . ) we fix a family of values of @xmath486 with @xmath488 and @xmath1 bounded away from zero , say @xmath506 for some fixed @xmath507 . our goal is to prove that relation holds . if we set @xmath508 , by definition we have @xmath509 . let us first show that @xmath510 . by proposition [ ch2:th : bs ] , @xmath511 implies @xmath460 , which means that every subsequence of values of @xmath486 admits a further sub - subsequence along which either @xmath512 or @xmath458 . the key point is that @xmath458 implies @xmath459 , because @xmath513 ( recall that @xmath506 ) . thus @xmath459 along every sub - subsequence , which means that @xmath459 along the whole family of values of @xmath514 . since @xmath459 , we can apply relation . taking @xmath515 of both sides of that relation , recalling the definition of @xmath417 and the fact that @xmath516 , we can write @xmath517 we now show that the last term in the right hand side is @xmath518 and can therefore be neglected . note that @xmath519 eventually , because @xmath459 , hence @xmath520 since @xmath521 is decreasing for @xmath522 , in case @xmath523 one has @xmath524 on the other hand , recalling that @xmath513 , in case @xmath525 one has @xmath526 , which can be rewritten as @xmath527 and together with @xmath525 yields @xmath528 in conclusion , yields @xmath529 , that is there exists @xmath530 such that @xmath531 , and since @xmath532 we can write @xmath533 this is a second degree equation in @xmath534 , whose solutions ( both positive ) are @xmath535 \,.\ ] ] since @xmath459 , eventually one has @xmath536 : since @xmath537 , it follows that @xmath538 , which selects the `` @xmath539 '' solution in . taking square roots of both sides of and recalling that @xmath540 yields the equality @xmath541 as one checks squaring both sides of . finally , since @xmath542 , it is quite intuitive that relation yields . to prove this fact , we observe that by we can write @xmath543 where for fixed @xmath544 we define the function @xmath545 by @xmath546 by direct computation , when @xmath547 ( resp . @xmath548 ) one has @xmath549 ( resp . @xmath550 ) for all @xmath551 . since @xmath552 , it follows that for every @xmath228 one has @xmath553 if @xmath547 , while @xmath554 if @xmath548 ; consequently , for any @xmath555 , @xmath556 which yields @xmath557 _ uniformly over @xmath228_. by , relation is proved . we now fix a family of values of @xmath486 with @xmath488 and @xmath1 bounded away from infinity , say @xmath558 for some fixed @xmath174 , and we prove relation . we set @xmath508 so that @xmath509 , cf . . ( note that @xmath452 , because @xmath559 by assumption . ) applying proposition [ ch2:th : bs ] we have @xmath460 , i.e. either @xmath459 or @xmath458 along sub - subsequences . however , this time @xmath459 implies @xmath458 , because @xmath560 ( recall that @xmath561 ) , which means that @xmath458 along the whole given family of values of @xmath514 . since @xmath458 , relation yields @xmath562 let us focus on @xmath563 : recalling that @xmath564 and @xmath565 , we first show that @xmath566 by a subsequence argument , we may assume that @xmath567 $ ] , and we recall that @xmath565 : * if @xmath568 , @xmath563 and @xmath569 converge to @xmath570 , hence @xmath571 ; * if @xmath572 , @xmath573 and @xmath574 diverge to @xmath575 and yields @xmath576 . the proof of is completed . next we observe that , again by @xmath565 , @xmath577 we can thus rewrite as @xmath578 if @xmath89 , recalling we obtain @xmath579 , which is the second line of . next we assume @xmath58 . by , and , for all @xmath431 we can write @xmath580 hence can be rewritten as @xmath581 for some @xmath530 . recalling that @xmath540 , we have shown that @xmath582 we now claim that @xmath583 by a subsequence argument , we may assume that @xmath584 $ ] and @xmath585 $ ] . * if @xmath63 , then @xmath586 ( recall that @xmath488 ) hence @xmath587 ; then both sides of converge to @xmath588 , hence their ratio converges to @xmath589 . * if @xmath590 , then again @xmath586 , hence @xmath591 : since @xmath592 as @xmath593 , cf . , it follows immediately that holds . * if @xmath594 , then @xmath595 : since @xmath596 as @xmath597 , cf . , @xmath598 because @xmath599 while @xmath600| \to \frac{1}{2 } \bar\kappa \in [ 0,\frac{m}{2}]$ ] , hence holds . having proved , we can plug it into , obtaining precisely the first line of . this completes the proof of theorem [ ch2:th : main1 ] . in this section we prove theorems [ ch2:th : main2b ] , [ ch2:th : main2bl ] and [ ch2:th : main2a ] . we stress that it is enough to prove the asymptotic relations for the option prices @xmath36 and @xmath133 , because the corresponding relations for the implied volatility @xmath144 follow immediately applying theorem [ ch2:th : main1 ] . we prove theorem [ ch2:th : main2b ] and [ ch2:th : main2bl ] at the same time . we recall that the tail probabilities @xmath8 , @xmath79 are defined in . throughout the proof , we fix a family of values of @xmath18 with @xmath58 and @xmath601 , for some fixed @xmath168 , such that hypothesis [ ch2:ass : rv ] is satisfied . extracting subsequences , we may distinguish three regimes for @xmath1 : * if @xmath25 our goal is to prove , resp . ; * if @xmath28 our goal is to prove , resp . , because in this case , plainly , one has @xmath74 , resp . @xmath82 , by ; * if @xmath30 , our goal is to prove , resp . . of course , each regime has different assumptions , as in theorem [ ch2:th : main2b ] and [ ch2:th : main2bl ] . _ step 0 . preparation . _ it follows by conditions and that @xmath602 therefore for every @xmath169 one has eventually @xmath603 where the inequality is `` @xmath604 '' instead of `` @xmath605 '' , because both sides are negative quantities . we stress that @xmath606 , resp . @xmath607 , by , hence @xmath608 moreover , we claim that in any of the regimes @xmath25 , @xmath28 and @xmath30 one has @xmath609 this follows readily by if @xmath30 or @xmath610 . if @xmath25 we argue as follows : by markov s inequality , for @xmath71 @xmath611 e^{-(1+\eta)\kappa } \,,\ ] ] hence @xmath612 \,.\ ] ] since in the regime @xmath25 we assume that the moment condition holds for some or every @xmath71 , the term @xmath613 $ ] is bounded from above , hence eventually @xmath614 which proves relation . the rest of the proof is divided in four steps , in each of which we prove lower and upper bounds on @xmath36 and @xmath133 , respectively . lower bounds on @xmath36 . _ we are going to prove sharp lower bounds on @xmath36 , that will lead to relations , and . by and , for every @xmath169 we can write @xmath615 \ge ( e^{{\varrho}_{\varepsilon}\kappa}-e^\kappa ) { \overline f}_t({\varrho}_{\varepsilon}\kappa ) \ , , \end{split}\ ] ] and applying we get @xmath616 if @xmath25 , since @xmath617 eventually , we obtain @xmath618 where in the last inequality we have applied . it follows that @xmath619 where the @xmath620 is taken along the given family of values of @xmath18 ( note that @xmath621 and @xmath622 are negative quantities , cf . , hence the reverse inequality with respect to ) . since @xmath169 is arbitrary and @xmath71 is fixed , we have shown that @xmath623 that is we have obtained a sharp bound for . if @xmath28 , since @xmath624 is bounded while @xmath625 , relation gives @xmath626 since @xmath169 is arbitrary , we have shown that when @xmath28 @xmath627 obtaining a sharp bound for . finally , if @xmath30 , since for @xmath21 by convexity @xmath628 , relation yields @xmath629 again , since @xmath630 is constant and @xmath625 , and @xmath169 is arbitrary , we get @xmath631 proving a sharp bound for . _ step 2 . lower bounds on @xmath133 . _ we are going to prove sharp lower bounds on @xmath133 , that will lead to relations , and . recalling and , for every @xmath169 we can write @xmath632 \ge ( e^{-\kappa}-e^{-{\varrho}_{\varepsilon}\kappa } ) f_t(-{\varrho}_{\varepsilon}\kappa ) \ , , \end{split}\ ] ] and applying we obtain @xmath633 if @xmath25 , since @xmath634 , eventually one has @xmath635 and we obtain @xmath636 since @xmath169 is arbitrary , it follows that @xmath637 which is a sharp bound for . if @xmath28 , since @xmath638 is bounded while @xmath639 , and @xmath169 is arbitrary , relation gives @xmath640 which is a sharp bound for . finally , if @xmath30 , since @xmath641 by convexity , since @xmath21 , one has eventually @xmath642 because @xmath643 while @xmath639 relation then yields , eventually , @xmath644 since @xmath169 is arbitrary , we have shown that @xmath645 obtaining a sharp bound for . upper bounds on @xmath36 . _ we are going to prove sharp upper bounds on @xmath36 , that will complete the proof of relations , and . we first consider the case when _ the moment assumptions and hold for every @xmath71_. let us look at the regimes @xmath25 and @xmath28 ( i.e. @xmath1 is bounded away from zero ) , assuming that condition holds _ for every @xmath71_. by hlder s inequality , @xmath646 \le { \ensuremath{{\mathrm}e}}[e^{x_t } { { \sf 1}}_{\{x_t > \kappa\ } } ] \le { \ensuremath{{\mathrm}e}}[e^{(1+\eta)x_t}]^{\frac{1}{1+\eta } } \ , { \overline f}_t(\kappa)^{\frac{\eta}{1+\eta } } \,.\ ] ] let us fix @xmath169 and choose @xmath647 large enough , so that @xmath648 . by assumption , for some @xmath649 one has @xmath650^{\frac{1}{1+\eta } } \le c \,,\ ] ] hence eventually , recalling that @xmath625 , by , @xmath651 since @xmath169 is arbitrary , this shows that @xmath652 which together with completes the proof of , if @xmath28 . if @xmath25 and condition holds for every @xmath71 , then @xmath653 by , which yields @xmath654 , hence together with completes the proof of . we then consider the regime @xmath30 , assuming that condition holds _ for every @xmath71_. we modify as follows : since @xmath655 , @xmath656 \le \kappa \ , { \ensuremath{{\mathrm}e}}\bigg [ \bigg|\frac{e^{x_t}-1}{\kappa}\bigg|^{1+\eta } \bigg]^{\frac{1}{1+\eta } } \ , { \overline f}_t(\kappa)^{\frac{\eta}{1+\eta } } \,.\ ] ] let us fix @xmath169 and choose @xmath647 large enough , so that @xmath648 . by assumption , for some @xmath649 one has @xmath657^{\frac{1}{1+\eta } } \le c \,,\ ] ] hence relation yields eventually @xmath658 since @xmath169 is arbitrary , we have proved that @xmath659 which together with completes the proof of . it remains to consider the case when the moment assumptions and holds _ for some @xmath71 _ , but in addition conditions ( if @xmath25 or @xmath28 ) or ( if @xmath30 ) holds . we start with considerations that are valid in any regime of @xmath1 . defining the constant @xmath660 where the @xmath620 is taken along the given family of values of @xmath18 , we claim that @xmath661 . this follows by if @xmath30 or if @xmath28 ( in which case , plainly , @xmath662 ) , while if @xmath359 it suffices to apply to get @xmath663 . it follows by that eventually @xmath664 next we show that , for all fixed @xmath169 and @xmath665 , eventually one has @xmath666 } e^{\kappa y } \ , { \overline f}_t(\kappa y ) \bigg ) \le ( 1-{\varepsilon } ) \big ( \log { \overline f}_t(\kappa ) + \kappa \big ) \,,\ ] ] which means that the @xmath667 is approximately attained for @xmath668 . this is easy if @xmath30 or if @xmath28 : in fact , since @xmath669 is non - increasing , we can write @xmath670 } e^{\kappa y } \ , { \overline f}_t(\kappa y ) \bigg ) & \le \log \big ( e^{\kappa m } { \overline f}_t(\kappa ) \big ) = \kappa m + \log { \overline f}_t(\kappa ) \\ & = \big ( \log { \overline f}_t(\kappa ) + \kappa \big ) \,+\ , ( m-1 ) \kappa \ , , \end{split}\ ] ] and since @xmath671 by , while @xmath672 is bounded , follows . to prove in the regime @xmath186 , we are going to exploit the assumption . first we fix @xmath507 , to be defined later , and set @xmath673 and @xmath674 for @xmath675 , so that @xmath676 \subseteq \bigcup_{n=1}^{\bar n } [ a_{n-1 } , a_n]$ ] . for all @xmath677 $ ] one has , by , @xmath678 having used that @xmath198 , by , hence eventually @xmath679 \,.\ ] ] recalling that @xmath680 , we can write @xmath681 , because @xmath682 by construction , and since @xmath683 for @xmath677 $ ] , it follows that @xmath670 } e^{\kappa y } \ , { \overline f}_t(\kappa y ) \bigg ) & \le \max_{n=1 , \ldots , \bar n } \big ( a_n \kappa + ( 1-\delta ) a_{n-1 } \log { \overline f}_t(\kappa ) \big ) \\ & = \max_{n=1 , \ldots , \bar n } \big((1-\delta ) a_{n-1 } \big ( \log { \overline f}_t(\kappa ) + \kappa \big ) + \delta(1+m ) \kappa \big ) \ , . \end{split}\ ] ] plainly , the @xmath684 is attained for @xmath685 , for which @xmath686 . recalling , we get @xmath687 } e^{\kappa y } \ , { \overline f}_t(\kappa y ) \bigg ) \le ( 1-\delta(1+a+am ) ) \big ( \log { \overline f}_t(\kappa ) + \kappa \big ) \,.\ ] ] choosing @xmath688 , the claim is proved . we are ready to give sharp upper bounds on @xmath36 , refining . for fixed @xmath689 , we write @xmath690 + { \ensuremath{{\mathrm}e}}[(e^{x_t}-e^\kappa ) { { \sf 1}}_{\{x_t > \kappa m\ } } ] \,,\ ] ] and we estimate the first term as follows : by fubini - tonelli s theorem and , @xmath691 & = { \ensuremath{{\mathrm}e}}\bigg [ \bigg ( \int_\kappa^{\infty } e^x \ , { { \sf 1}}_{\{x < x_t\ } } \ , { \text{\rm d}}x \bigg ) { { \sf 1}}_{\{\kappa < x_t \le \kappa m\ } } \bigg ] \\ & = \int_\kappa^{\kappa m } e^x \ , { \ensuremath{{\mathrm}p}}(x < x_t \le \kappa m)\ , { \text{\rm d}}x \le \int_\kappa^{\kappa m } e^x \ , { \overline f}_t(x ) \ , { \text{\rm d}}x \\ & = \kappa \int_1^m e^{\kappa y } \ , { \overline f}_t(\kappa y ) \ , { \text{\rm d}}y \le \kappa \ , ( m-1 ) \ , e^{(1-{\varepsilon } ) ( \log { \overline f}_t(\kappa ) + \kappa ) } \ , . \end{split}\ ] ] to estimate the second term in , we start with the cases @xmath25 and @xmath28 , where we assume that holds for some @xmath71 , as well as , hence we can fix @xmath692 such that @xmath693 . bounding @xmath694 , hlder s inequality yields @xmath695 \le { \ensuremath{{\mathrm}e}}[e^{(1+\eta)x_t}]^{\frac{1}{1+\eta } } \ , { \overline f}_t(\kappa m)^{\frac{\eta}{1+\eta } } = c \ , { \overline f}_t(\kappa m)^{\frac{\eta}{1+\eta } } \,,\ ] ] where @xmath649 is an absolute constant , by . applying relation together with @xmath693 we obtain @xmath696 hence eventually @xmath697 \le ( 1-{\varepsilon } ) \log { \overline f}_t(\kappa ) \le ( 1-{\varepsilon } ) \big(\log { \overline f}_t(\kappa ) + \kappa \big ) \,.\ ] ] recalling and , eventually @xmath698 , hence by @xmath699 \le ( 1 - 2{\varepsilon } ) \big(\log { \overline f}_t(\kappa ) + \kappa \big ) \,.\ ] ] looking back at , since @xmath700 by , and again one has eventually @xmath701 since @xmath169 is arbitrary , this shows that @xmath702 which together with completes the proof of , if @xmath25 . since @xmath703 if @xmath28 , by , we can rewrite in this case as @xmath704 which together with completes the proof of . it remains to consider the case when @xmath30 , where we assume that relation holds for some @xmath63 , together with . as before , we fix @xmath692 such that @xmath693 . since @xmath705 \le { \ensuremath{{\mathrm}e}}\bigg [ \bigg|\frac{e^{x_t}-1}{\kappa } \bigg|^{1+\eta}\bigg ] \le c \,,\ ] ] for some absolute constant @xmath649 , by , the second term in is bounded by @xmath706 \le \kappa { \ensuremath{{\mathrm}e}}\bigg [ \bigg|\frac{e^{x_t}-e^\kappa}{\kappa } \bigg|^{1+\eta } \bigg]^{\frac{1}{1+\eta } } \ , { \overline f}_t(\kappa m)^{\frac{\eta}{1+\eta } } \le \kappa \ , c \ , { \overline f}_t(\kappa m)^{\frac{\eta}{1+\eta } } \,.\ ] ] in complete analogy with - , we obtain that eventually @xmath707}{\kappa } \le ( 1-{\varepsilon } ) \log { \overline f}_t(\kappa ) \,.\ ] ] by , eventually @xmath708 , hence by @xmath709}{\kappa } \le ( 1 - 2{\varepsilon } ) ( \log { \overline f}_t(\kappa ) + \kappa ) \,.\ ] ] recalling and , we can finally write @xmath710 because @xmath30 and @xmath711 . since @xmath169 is arbitrary , we have proved that @xmath712 which together with completes the proof of . upper bounds on @xmath133 . _ we are going to prove sharp upper bounds on @xmath133 , that will complete the proof of relations , and . by we can write @xmath713 \le e^{-\kappa } \ , f_t(-\kappa ) \,,\ ] ] therefore @xmath714 which together with completes the proof of , if @xmath25 . on the other hand , if @xmath28 , since relation implies ( recall that @xmath134 ) @xmath715 in view of , the proof of is completed . it remains to consider the case @xmath30 . if relation holds _ for every @xmath63 _ , we argue in complete analogy with -- , getting @xmath716 which together with completes the proof of . if , on the other hand , relation holds only _ for some @xmath63 _ , we also assume that condition holds , hence we can fix @xmath692 such that @xmath717 . let us write @xmath718 + { \ensuremath{{\mathrm}e}}[(e^{-\kappa } - e^{x_t } ) { { \sf 1}}_{\{x_t \le -\kappa m\ } } ] \,.\ ] ] in analogy with , for every fixed @xmath169 , the first term in the right hand side can be estimated as follows ( note that @xmath719 is decreasing ) : @xmath720 & \le \int_{-\kappa m}^{-\kappa } e^x \ , f_t(x ) \ , { \text{\rm d}}x = \kappa \int_{1}^{m } e^{-\kappa y } \ , f_t(-\kappa y ) \ , { \text{\rm d}}y \\ & \le \kappa ( m-1 ) \ , f_t(-\kappa ) \le \kappa \ , e^{(1-{\varepsilon } ) \log f_t(-\kappa ) } \ , . \end{split}\ ] ] the second term in is estimated in complete analogy with -- , yielding @xmath721}{\kappa } \le ( 1-{\varepsilon } ) \log f_t(-\kappa ) \,.\ ] ] recalling , we obtain from @xmath722 and since @xmath169 is arbitrary we have proved that relation still holds , which together with completes the proof of , and of the whole theorem [ ch2:th : main2b ] . by skorokhod s representation theorem , we can build a coupling of the random variables @xmath12 and @xmath96 such that relation holds a.s .. since the function @xmath723 is continuous , recalling that @xmath724 , for @xmath725 we have a.s . @xmath726{\,a.s.\ , } ( y - a)^+ \,,\ ] ] and analogously for @xmath727 @xmath728{\,a.s.\ , } ( -a - y)^+ = ( y+a)^- \,.\ ] ] taking the expectation of both sides of these relations , one would obtain precisely . to justify the interchanging of limit and expectation , we observe that the left hand sides of and are uniformly integrable , being bounded in @xmath729 . in fact @xmath730 and the second term in the right hand side is uniformly bounded ( recall that @xmath725 by assumption ) , while the first term is bounded in @xmath729 , by . recall from [ ch2:sec : setting ] that @xmath12 denotes the risk - neutral log - price , and assume that @xmath731 in distribution as @xmath7 ( which is automatically satisfied if @xmath292 has right - continuous paths ) . for an arbitrary family of values of @xmath18 , with @xmath6 and @xmath21 , we show that condition implies . assume first that @xmath7 ( with no assumption on @xmath1 ) . since @xmath21 , one has @xmath732 in distribution , hence @xmath733 by and fatou s lemma . with analogous arguments , one has @xmath734 , hence is satisfied . next we assume that @xmath25 and @xmath2 is bounded , say @xmath735 $ ] for some fixed @xmath200 . since @xmath736 is a convex function and @xmath737 is a martingale , the process @xmath738 is a submartingale and by we can write @xmath739 = { \ensuremath{{\mathrm}e}}[(e^{x_t } - e^\kappa){{\sf 1}}_{\{x_t > \kappa\ } } ] \le { \ensuremath{{\mathrm}e}}[e^{x_t}{{\sf 1}}_{\{x_t > \kappa\ } } ] \,.\ ] ] it follows that , if @xmath359 , then @xmath733 . with analogous arguments , one shows that @xmath734 , hence condition holds . let @xmath121 be a positive process which solves . by ito s formula , the process @xmath740 solves @xmath741 assuming @xmath742 a.s . as @xmath7 , we want to show that @xmath743{\ , d \ , } y \sim n(0,\sigma_0 ^ 2 ) \,.\ ] ] let us define @xmath744 by @xmath742 a.s . it follows that @xmath745 a.s . , by the fundamental theorem of calculus . moreover , since @xmath746 a.s . , @xmath747{a.s . } 0 \,.\ ] ] we now use the inequality @xmath748 , cf . * problem 5.25 ) . sending first @xmath7 for fixed @xmath507 , and then @xmath749 , we see by that @xmath750 in probability as @xmath7 . since @xmath751 in distribution as @xmath7 , is @xmath752 _ for all @xmath6_. ] by slutsky s theorem @xmath753{d } 0 + y = y \sim n(0,\sigma_0 ^ 2 ) \,,\ ] ] hence relation holds . next we show that , plugging @xmath129 into , we obtain @xmath754 for any @xmath101 . since @xmath96 has a symmetric law , it suffices to focus on @xmath755 . then @xmath756 & = \sigma_0 \,{\ensuremath{{\mathrm}e}}\left[\left(n(0,1)-\frac{a}{\sigma_0}\right)^+\right ] = \sigma_0 \bigg[\int_{\frac{a}{\sigma_0}}^{\infty } x \ , \frac{e^{-\frac{x^2}{2}}}{\sqrt{2\pi } } \ , { \text{\rm d}}x \,-\ , \frac{a}{\sigma_0 } \int_{\frac{a}{\sigma_0}}^{\infty } \frac{e^{-\frac{x^2}{2}}}{\sqrt{2\pi } } \ , { \text{\rm d}}x \bigg]\\ & = \sigma_0\bigg(\phi\bigg(\frac{a}{\sigma_0}\bigg ) \,-\ , \frac{a}{\sigma_0}\phi\left(-\frac{a}{\sigma_0}\right)\bigg ) = a \ , d\bigg(\frac{a}{\sigma_0}\bigg)\ , , \end{split}\ ] ] where we used the density @xmath417 and distribution function @xmath418 of a @xmath757 random variable , cf . , and the definition of @xmath107 . looking back at , we obtain @xmath758 . we start with some estimates . it follows by that @xmath759 with @xmath760 and @xmath761 ( and we agree that the sum equals @xmath762 in case @xmath763 ) . by chernoff s bound = e^{-m \alpha + \lambda t(e^\alpha - 1)}$ ] and optimize over @xmath764 . ] @xmath765 , hence @xmath766 where @xmath767 denotes a gaussian random variable with mean @xmath333 and variance @xmath324 . we recall the standard estimate @xmath768 as @xmath769 . then we can write : @xmath770 in particular , we get from that , for fixed @xmath771 , @xmath772 for a lower bound , restricting the sum in to a single value @xmath260 , we get @xmath773 we now prove relation . we fix a family of @xmath774 with @xmath7 and @xmath330 for some @xmath262 . to get an upper bound , we drop the term @xmath775 in ( since @xmath776 ) and plug @xmath777 , getting @xmath778 let us denote by @xmath779 the value of @xmath260 for which the minimum in the definition of @xmath780 is attained . choosing @xmath781 large enough , so that @xmath782 , the middle term in is @xmath783 and is the dominating one , provided @xmath784 and @xmath785 , so that the third term is @xmath786 . for an analogous lower bound , we apply with @xmath787 : since @xmath788 ( recall that @xmath7 ) , we get @xmath789 we have thus proved relation . it remains to prove relation . we fix a family of @xmath774 such that either @xmath7 and @xmath264 , or @xmath26 and @xmath25 . since @xmath790 , @xmath791 by direct computation , the infimum is attained at @xmath792 which yields @xmath793 we now choose @xmath794 in , so that @xmath795 where we have absorbed @xmath796 inside the @xmath797 term in the exponential , because @xmath798 by ( recall that @xmath25 ) . the dominant contribution to is given by the middle term ( note that @xmath799 , always by ) . for a corresponding lower bound , we apply with @xmath800 : since @xmath801 and @xmath802 ( because @xmath803 ) , we get @xmath804 we have thus shown that @xmath805 completing the proof of relation and of lemma [ th : lemma ] . let us first prove and . since @xmath806 , cf . and , recalling we can rewrite the black&scholes formula as follows : @xmath807 if @xmath459 , applying we get @xmath808 and is proved . next we assume that @xmath458 . by convexity of @xmath467 ( cf . lemma [ ch2:th : mills ] ) , @xmath809 hence to prove it suffices to show that @xmath810 . to this purpose , by a subsequence argument , we may assume that @xmath811 . since @xmath812 for @xmath21 , when @xmath458 necessarily @xmath813 $ ] . if @xmath814 , i.e.@xmath815 , then @xmath816 and @xmath810 follows by . on the other hand , if @xmath817 $ ] then both @xmath563 and @xmath818 converge to @xmath819 , by continuity of @xmath820 , hence @xmath821 , i.e. @xmath810 as requested . let us now prove . assume that @xmath460 , and note that for every subsequence we can extract a sub - subsequence along which either @xmath459 or @xmath458 . we can then apply and to show that @xmath463 : * if @xmath459 , the right hand side of is bounded from above by @xmath822 ; * if @xmath21 and @xmath458 , then @xmath823 and consequently @xmath824 is uniformly bounded from above , hence the right hand side of vanishes ( since @xmath565 ) . finally , we assume that @xmath825 and show that @xmath826 . extracting a subsequence , we have @xmath827 for some fixed @xmath174 , i.e. both @xmath828 and @xmath829 , and we may assume that @xmath830 $ ] and @xmath831 $ ] . consider first the case @xmath832 , i.e. @xmath833 : by one has @xmath834 , hence @xmath835 ( because @xmath417 is bounded ) , and recalling relation yields @xmath836 next consider the case @xmath837 : since @xmath812 , we have @xmath838 and again by we obtain @xmath839 . in both cases , @xmath840 . we thank fabio bellini , stefan gerhold and carlo sgarra for fruitful discussions .
we provide explicit conditions on the distribution of risk - neutral log - returns which yield sharp asymptotic estimates on the implied volatility smile . we allow for a variety of asymptotic regimes , including both small maturity ( with arbitrary strike ) and extreme strike ( with arbitrary bounded maturity ) , extending previous work of benaim and friz @xcite . we present applications to popular models , including carr - wu finite moment logstable model , merton s jump diffusion model and heston s model .
galaxies have no edge . with rare exceptions , the stellar spatial distribution in normal galaxies shows little sign of a sharp outer cutoff , and is reasonably well approximated by density laws that extend smoothly to arbitrarily large radius . extrapolations of the inner luminosity profile , however , suggest that little light comes from regions of surface brightness much fainter than those traditionally used to define the luminous radii of galaxies ( @xmath4 mag / asec@xmath5 ) . perhaps because of this reason , together with the obvious observational difficulties inherent to studying regions of low - surface brightness , the outer luminous halos of external galaxies have in the past been regarded as a topic of little more than academic interest . this impression , however , is rapidly changing , as new datasets start to unveil some unexpected properties of the stellar component that populates the outer confines of galaxies . these developments have been made possible by the development of panoramic digital cameras able to map the light distribution of external galaxies down to unprecedented surface brightness levels , complemented by efficient observational techniques designed to identify and measure radial velocities of outer halo tracers in external galaxies , such as planetary nebulae ( pne , romanowsky et al . finally , the advent of efficient spectrographs in 10m - class telescopes have enabled the measurement of radial velocities of large samples of giant stars throughout the local group ( ibata et al 2004 ) , and dedicated spectroscopic campaigns targeted on giant stars have dramatically increased the sample of tracers in the outer halo of our own milky way ( morrison et al 2003 , battaglia et al 2005 , clewley et al 2005 ) . likewise , the use of stacking techniques in wide - field surveys such as the sloan digital sky survey ( sdss ) have led to clear detections of a distinct low surface brightness component around both isolated spiral galaxies ( zibetti , white & brinkmann 2004 ) as well as central cluster galaxies ( zibetti et al 2005 ) . the surface brightness profile of these outer luminous halos deviate significantly from straightforward extrapolations of the laws that describe the main body of the galaxy , and suggest that estimates of the amount of light in the intergalactic or intracluster medium might need to be revised upward . in the case of clusters , for example , zibetti et al ( 2005 ) conclude that of order 11% of the light in a typical abell cluster might be in the form of an intracluster luminous component with structural properties distinct from those of the galaxy cluster population and of the central cluster galaxy . these studies have brought about several unexpected results that challenge some of the accepted premises of the currently accepted galaxy formation paradigm . for example , the kinematics of pne in the vicinity of several normal ellipticals has been found to be consistent with simple models where such galaxies have no surrounding dark matter halo ( romanowsky et al . 2003 , but see dekel et al 2005 ) . this result , if confirmed by further studies , would be very difficult to accommodate in the @xmath6cdm scenario , where _ all _ galaxies are envisioned to form as a result of the dissipative collapse of baryons within massive dark matter halos . in the case of the local group , wide field photometric surveys ( and their follow - up spectroscopic campaigns ) have modified radically the traditional view of the outer stellar spheroid of the milky way and m31 . giant structures interpreted as tidal streams of disrupted or disrupting dwarfs have been identified in both galaxies ( ibata et al 1994 , helmi et al 1999 , majewski et al 2003 , ibata et al 2004 ) , and have solidified the notion that accretion events may play a substantial role in the shaping of individual galaxies . these streams are expected to be short - lived as readily identifiable structures on the sky ( johnston et al 1995 , helmi & white 1999 ) , and , although impressive , they are thought to contribute a relatively small fraction of stars in the halo of these systems . in broad terms , the main characteristics of the outer stellar halo are probably adequately captured by a simple model of a reasonably well - mixed spheroid of stars . extensive spectroscopic surveys of giant stars in the outskirts of local group galaxies have unraveled some intriguing dynamical properties for the outer halo component . for example , just like pne around ellipticals , the velocity dispersion of these halo tracers drops significantly in the outer regions of the milky way ( battaglia et al 2005 , clewley et al 2005 ) . this is a somewhat unexpected result if these galaxies are embedded in extended dark matter halos , and has prompted renewed interest in the extent and spatial distribution of dark matter in the outskirts of galaxies . from a theoretical point of view , the interest elicited by outer luminous halos stems from the fact that , in the currently accepted paradigm , stars are envisioned to form only in the collapsed , high - density regions near the center of dark halos which we identify with the main body of individual galaxies . this prejudice is supported by observations that indicate the need for a threshold gas density below which stars do not form ( kennicutt 1989 , martin & kennicutt 2001 , schaye 2004 ) , and imply that outer halo stars did _ not _ form _ in situ _ but have been shed from protogalaxies during the merger events that characterize the assembly of galaxies in a hierarchically clustering universe . in simple words , stars found as far away as , say , 100 kpc from a galaxy s center originate in satellites whose orbital apocenter was about that large before they spiraled in to merge with the main galaxy . we therefore expect a clear connection between the orbital properties of stars in the outer halo and those of the progenitors that merged to form the present - day galaxy . unraveling the structure of the outer luminous halo of galaxies may thus be seen as an important step toward unraveling the merging history of individual galaxies . in this paper , we analyze the origin and structure of the luminous halos of galaxies simulated in the @xmath6cdm scenario . we describe briefly in section [ sec : numexp ] the numerical simulations . the main results of our analysis are presented in section [ sec : results ] and discussed in [ sec : disc ] . we conclude with a brief summary in section [ sec : conc ] . we analyze a suite of eight numerical simulations of the formation of galaxies in the @xmath6cdm scenario . these simulations have been analyzed in earlier papers , which may be consulted for details on the code used as well as on the numerical setup ( steinmetz & navarro 2002 , abadi et al 2003a , b , meza et al 2003 , 2005 ) . in brief , each simulation follows the evolution in a @xmath6cdm universe of a small region surrounding a target galaxy , excised from a large periodic box and re - simulated at higher resolution preserving the tidal fields from the whole box . the simulation includes the gravitational effects of dark matter , gas and stars , and follows the hydrodynamical evolution of the gaseous component using the smooth particle hydrodynamics ( sph ) technique ( steinmetz 1996 ) . we adopt the following cosmological parameters for the @xmath6cdm scenario : @xmath7 km / s / mpc , @xmath8 , @xmath9 , @xmath10 , @xmath11 , with no tilt in the primordial power spectrum . all simulations start at redshift @xmath12 , have force resolution of order @xmath13 kpc , and mass resolution so that each galaxy is represented , at @xmath14 , with @xmath15 star particles . dense , cold gas in collapsed regions is allowed to turn into stars at rates consistent with the empirical schmidt - like law of kennicutt ( 1998 ) . the energetic feedback of evolving stars is included mainly as a heating term on the surrounding gas , but the short cooling times in these regions reduce significantly the effectiveness of feedback in curtailing star formation . the transformation of gas into stars thus tracks closely the rate at which gas cools and condenses at the center of dark matter halos . as discussed in the references listed in the above paragraph , this results in an early onset of star - forming activity in the many progenitors of the present - day galaxy present at high redshift . this leads to the formation of a prominent spheroidal component as these progenitors coalesce and merge to form the final galaxy . gas accreted after the merging activity abates leads to the formation of a centrifugally - supported gaseous and stellar disk component clearly present in most of our simulations , although they make up typically only about @xmath16 of the final stellar mass of each galaxy . it is likely that improvements to our feedback algorithms may lead to revisions in the efficiency and timing of star formation in these galaxies . as we discuss below , outer halo stars originate mainly in the merger of the early progenitors , so reducing the efficiency of early star formation should have some impact on the actual prominence of the outer stellar halo component . we expect , however , that reasonable modifications of the star formation algorithm will affect mainly the number , ages , and chemical composition of outer halo stars , rather than its dynamical and structural properties , which we expect to depend mainly on the orbital properties of the merging progenitors and on the structure of the dark matter halo host . these properties are less sensitive to the complex astrophysics of star formation and feedback , and we therefore focus our analysis on the structure and dynamics of the outer stellar halo in our eight simulations . these target a small range in mass chosen so that at @xmath14 the galaxies have luminosities and circular speeds comparable to the milky way . we summarize the main structural parameters of the dark matter and stellar components of the simulated galaxies in table [ tab : table1 ] . kpc ( shown with arrows ) , they contribute @xmath17 of the stellar mass . outside @xmath18 the stellar component consists almost exclusively of accreted stars ; indeed , fewer than @xmath19 of stars in the outer ( @xmath20 ) halo formed in situ . [ figs : rfinsitu ] ] ] figure [ figs : xypanel ] shows , at @xmath14 , one of our simulated galaxies ( kia3 , see table [ tab : table1 ] ) projected onto a box of @xmath21 kpc on a side . the top panels show the dark matter particles ( left ) and stars ( right ) within the virial radius ( @xmath22 kpc , shown by the outer green circle ) , defined to encompass a region of mean density @xmath23 times the critical density for closure . dark matter particles are coloured by their local density , while stars are coloured by their age , as described in the label . the bottom panels in figure [ figs : xypanel ] separate the stars in two components : `` in situ '' stars that formed in the most massive progenitor ( left ) and `` accreted stars '' that formed in progenitors that merged with the main galaxy ( right ) . stars labelled as `` accreted '' exclude those associated with self - bound satellite systems that survive until the present . these can be seen clearly in the top - right panel of figure [ figs : xypanel ] , but are largely absent in the bottom - right panel , except for a sprinkling of stars associated with a young stream recently torn from a disrupting satellite , visible at @xmath24 kpc . roughly @xmath25 of stars ( by mass ) formed `` in situ '' in this galaxy , compared with @xmath26 which make up the accreted component . satellites contribute a rather small fraction ( @xmath27 ) of all stars within @xmath28 . these numbers are typical of our simulations , as may be seen from the numbers listed in table [ tab : table1 ] . -band surface brightness profile of stars , split between the `` in - situ '' and `` accreted '' components ( run kia3 , see table [ tab : table1 ] ) . the accreted stars dominate outside the luminous radius and appear as an excess of light over a bulge+disk fit to the inner surface brightness profile , which is shown as a solid line . the dashed line shows a sersic - law fit to the outer profile of accreted stars . note that a single sersic - law ( with parameters given in table [ tab : table2 ] ) reproduces very well the radial distribution of all accreted stars . [ figs : sbprof ] ] figure [ figs : xypanel ] illustrates a number of properties of the stellar component common to all of our simulations . in particular , it is important to note that ( i ) stars spread as far out as the virial radius of the system ( outer circle in the panels of figure [ figs : xypanel ] ) , although they are more highly concentrated than the dark matter halo ; ( ii ) _ in situ _ stars ( bottom left panel ) , are responsible for most of the young stars in the main body of the galaxy ( note the prominence of the young disk in the bottom - left panel ) , and are practically absent from the outer halo ; whereas ( iii ) accreted stars ( bottom right panel ) make up preferentially the spheroidal component and dominate the stellar budget in the outer regions of the galaxy . the relative importance of the `` accreted '' vs `` in situ '' components at various radii is shown in figure [ figs : rfinsitu ] . beyond a radius @xmath29 kpc ( marked by arrows in figure [ figs : rfinsitu ] ) accreted stars dominate in all our simulations . we shall define @xmath30 as the `` luminous radius '' of the galaxy and refer to stars beyond @xmath31 as the `` outer luminous halo '' or `` outer galaxy '' , for short . our analysis focuses on the structure and dynamics of the outer halo and , consequently , mainly on the properties of the accreted component . it is also clear from this figure that the outer luminous halo contains a relatively small fraction of the stellar mass of the galaxy ; fewer than 15% of all stars are found between @xmath30 and the virial radius . we note as well that accreted stars make up a non - negligible fraction of stars in the inner galaxy ( i.e. , inside @xmath30 ) ; we shall therefore distinguish between `` accreted '' and `` outer halo '' stars in what follows . figure [ figs : rhoprof ] shows the density profile of the stellar and dark matter components of the simulation shown in figure [ figs : xypanel ] . the stellar contribution is split between in - situ and accreted stars . note that stars are significantly more concentrated than the dark matter ; indeed , they contribute a significant fraction of the _ total _ mass within the luminous radius ( @xmath32 on average for all runs in our series ) , but they make up a negligible amount of the total mass in the outer regions . note as well the sharp truncation of in - situ stars at @xmath20 ; essentially no stars in the outer halo were formed in the main progenitor . this is a natural consequence of the fact that in - situ star formation proceeds efficiently only in regions of high - density . the few in - situ stars found beyond @xmath30 at @xmath14 have been torn from the main progenitor during mergers , although figure [ figs : rfinsitu ] ( and the data in table [ tab : table1 ] ) shows that this is rather inefficient , and that the `` accreted '' component dominates the stellar component of the outer halo . the radius beyond which accreted stars begin to dominate is signaled by an abrupt change in the surface brightness profile of the galaxy , as shown in figure [ figs : sbprof ] . inside the luminous radius , where @xmath33 is brighter than about @xmath34 mag / asec@xmath5 , the surface brightness profile of this galaxy ( projected in this case face - on and studied in detail by abadi et al 2003a , b ) is well approximated by a de vaucouleurs bulge plus an exponential disk ( shown as a solid line ) . this bulge+disk model is , however , unable to fit the structure of the outer luminous halo , which deviates from the inner profile and appears as a outer luminous `` excess '' beyond @xmath35 kpc , where the @xmath36-band surface brightness drops to values fainter than about @xmath37 mag / asec@xmath5 . the outer halo surface brightness profile steepens gradually outwards and is well approximated by a sersic - law , @xmath38}\ ] ] where @xmath39 and @xmath40 are , respectively , the radius containing half the light and the surface brightness level at that radius . the surface brightness profile of the outer regions of our simulated galaxies is shown in figure [ figs : sersic]a , together with sersic - law fits to the outer portion of the profiles . the sersic fit parameters are listed in table [ tab : table2 ] . on average we find @xmath41 ; @xmath42 kpc . the outer luminous halo s surface brightness profile steepens from @xmath43 at about the luminous radius to @xmath44 at @xmath45 kpc and to @xmath46 or steeper around the virial radius . the gradually steepening profile of the outer halo characterized by the sersic law is reminiscent of the distribution of dark matter , whose density profile also steepens monotonically outward ( navarro et al 2004 ) . the dark matter profile is , however , much less centrally concentrated and its slope is shallower than the stellar halo at all radii . as discussed by merritt et al ( 2005 ) , the dark matter density profile may also be approximated by a sersic law , but with a characteristic value of @xmath47 and much larger effective radii . figure [ figs : sersic]b shows the surface brightness profiles _ of the accreted stellar component only _ for all eight simulations in our series . the profiles have been scaled so that they would all line up along a line of slope -1 if they followed accurately a sersic law . all profiles are indeed well approximated by eq . 1 , with the possible exception of one case where a `` bump '' of stars associated with an ongoing satellite disruption event is seen just outside @xmath39 ( open starred symbols in figure [ figs : sersic]b ) . this raises the interesting prospect of estimating the _ total _ fraction of stars accreted throughout the history of a galaxy simply by fitting its outer profile . as the last two columns of table [ tab : table2 ] demonstrate , the total light contributed by accreted stars is well approximated ( to within a factor of @xmath48 ) by the total light of a sersic - law fit to the outer luminous halo . the similarity between profiles suggest that the violent relaxation associated with mergers endows the accreted stellar component with a simple , roughly self - similar structure that is well approximated by a sersic law . this process has been studied extensively in the literature , and previous studies have consistently found that the structure of n - body merger remnants is quite reminiscent to that of bright ellipticals ( see , e.g. , barnes & hernquist 1992 and references therein ) , which are indeed well approximated by a sersic law with @xmath49 ( see , e.g. , graham & guzman 2003 and references therein ) . one may think of the accreted stellar component as formed by the overlap of the many `` tidal tails '' stripped from the merging progenitors ; in this interpretation the outer luminous halo would just be the superposition of the outer tails stripped from each progenitor during the merger process . this characterization of the accreted stellar component helps to explain the remarkable kinematics of the outer halo shown in figure [ figs : sigmar ] . this figure shows the velocity dispersion profile of stars and dark matter , averaged over all simulations after scaling the positions and velocities of all particles to the virial radius and virial velocity of each system . the right panels in figure [ figs : sigmar ] show that the velocity dispersion of the dark matter declines gradually from the center outwards , and is characterized by a mild radial anisotropy , going from almost isotropic near the center to @xmath50 in the outer regions . ( we use @xmath51 to denote the tangential velocity dispersion ; @xmath52 . ) the outer luminous halo , on the other hand , shows a much more pronounced radial bias : from @xmath53 just outside the luminous radius of the galaxy ( @xmath30 is on average of order @xmath54 ) to about @xmath55 near the virial radius . this substantial and monotonically increasing radial anisotropy may also be understood as a direct consequence of the tidal disruption process of formation described above . outer halo stars are typically stripped from merging progenitors during pericentric passages . however , because stars form in high - density regions the stripping of stars operates efficiently only when pericenters are small ( and tidal effects are greatest ) . this occurs typically after dynamical friction has eroded the orbit sufficiently to bring the pericenter of the accreting satellite close to the central galaxy . as a result , outer halo stars are `` launched '' into highly energetic orbits with , characteristically , the ( small ) pericentric radii that accompany the most disruptive tides . stars able to reach farther are thus typically on more eccentric orbits , leading to the increasing radial anisotropy in the orbits of the outer halo stars seen in figure [ figs : sigmar ] . the formation process of the accreted stellar component is thus qualitatively similar to that of the dark matter halo , where mergers also play a substantial role . it is thus perhaps not surprising that they settle onto similar , mildly triaxial structures , as shown in figure [ figs : cbba ] . for the accreted stars , the average intermediate - to - major axis ratio is @xmath56 , whereas the average minor - to - intermediate axis ratio is @xmath57 ( @xmath58 ) . this is not too different from the dark matter halos , which have on average @xmath59 and @xmath60 ( @xmath61 ) . we note that the dark matter halos in these simulations are much rounder than typically found in n - body simulations of cdm halo formation : @xmath62 and @xmath63 ( see , for example , bailin & steinmetz 2004 ) . this is the result of the response of the halo to the dissipative assembly of the baryonic component of the galaxy at its center , which steepens the potential well and reduces the triaxiality of the halo ; see , e.g. , kazantzidis et al ( 2004 ) , bailin et al ( 2005 ) , and references therein . the in - situ stellar component , on the other hand , often sports a well defined rotationally - supported disk ( see , e.g. , lower left panel of figure [ figs : xypanel ] ) and its shape is thus well described by an oblate structure with @xmath64 and @xmath65 . although rotation plays little role supporting the accreted component ( the mean rotation velocity , @xmath66 , is typically less than one tenth of the circular velocity at each radius ) , it is interesting to note that the direction of its angular momentum is well aligned with that of the inner galaxy . this is shown in figure [ figs : jjdist ] , which show the distribution of the cosine of the angles between the angular momentum of the dark matter halo ( @xmath67 ) , the in - situ stars ( @xmath68 ) , and the accreted stellar halo ( @xmath69 ) . although the good agreement between @xmath69 and @xmath68 could have perhaps been anticipated ( after all , they are both part of the same stellar system ) it is still interesting to note the strong correlation between the rotational properties of the inner galaxy and the dark matter halo . this implies that , although mergers lead to substantial transfer of angular momentum from the stars to the dark halo ( navarro & steinmetz 1997 , abadi et al 2003a ) , this does not alter radically its orientation , so that the spin of the inner galaxy generally ends up aligned with the rotation axis of the surrounding dark matter halo . it is also of interest to characterize the orientation of the rotation axis relative to the principal axes of the system . figure [ figs : jcdist ] shows the alignment between the galaxy s angular momentum and the minor axis of the dark matter and luminous halos . no obvious alignment is seen in these panels , suggesting that the rotational properties of the galaxy are , at best , weakly correlated with the shape of the surrounding dark matter ( or luminous ) halo . we shall return to possible interpretations of this result in the following section . we end the characterization of the outer luminous halos of simulated galaxies by comparing the age distribution of stars in the inner and outer galaxy , as well as that of satellites enclosed within the virial radius ( figure [ figs : agedist ] ) . the ages of stars in the outer galaxy ( @xmath70 ) differ significantly from those in higher density regions ( like the inner galaxy or the surviving satellites ) , where star formation may proceed . the outer halo is populated mainly by older stars , reflecting the fact that the mergers responsible for its formation are more common at earlier times . interestingly , the distribution of ages of stars in the outer halo is also fairly distinct from that of stars in satellites orbiting within the virial radius . this shows that relatively few stars in the outer halo originate in the `` harassment '' of satellites that have survived as self - bound entities until the present ( moore et al 1996 ) . most stars in the outer halo come from merger events whose progenitors have long been fully disrupted , suggesting that the properties of the satellite population may be quite distinct from that of the smooth outer halo . we explore the consequences of these results for the interpretation of observational data in the following section . in these units . note that the dark matter velocity dispersion exhibits a mild radial bias ; the radial bias is much more pronounced , and monotonically increasing , in the case of the accreted stellar component . [ figs : sigmar ] ] the results presented in the previous section have a number of implications regarding the interpretation of observations of stars in regions far removed from the normal boundaries of typical galaxies . our simulations suggest that these `` intergalactic stars '' ( or , more properly , `` extra - galactic stars '' ) are a relict of the merging history of each individual galaxy . as such , all galaxies with a past of active merging activity ( the majority in a hierarchically clustering scenario such as @xmath6cdm ) should have an outer stellar halo component with properties similar to those described in the previous section . the outer luminous halo should appear as an excess of light over the extrapolated profile of the inner galaxy . its structure is characterized by a gradually steepening density profile and by a clearly defined radial anisotropy in the velocity distribution . outer halos are rather difficult to detect , since they are by definition confined to regions of very low surface brightness difficult to probe observationally . however , the presence of outer luminous components detached from the properties of the inner galaxy have been recently reported by a number of authors . following the pioneering work of sackett et al ( 1994 ) , morrison et al ( 1994 ) , and zheng et al ( 1999 ) among others , zibetti , white & brinkmann ( 2004 ) , have recently reported the statistical detection of a @xmath0 halo of stars in the outskirts of a sample of edge - on disk - dominated galaxies selected from the sloan digital sky survey ( sdss ) . encouragingly , this outer halo component is detected as excess light over an extrapolation of the inner disk ( see figure [ figs : sbprof ] ) ; the halo dominates the minor axis light profile beyond about @xmath71 exponential scalelengths , in regions where the @xmath72-band surface brightness drops to values fainter than @xmath73 . a qualitatively similar result has been derived for m31 through detailed stellar counts by guhathakurta et al ( 2005 ) and by irwin et al ( 2005 ) , who find that the surface brightness profile along the minor axis `` flattens '' relative to an extrapolation of the de vaucouleurs fit to the inner spheroid at radii beyond @xmath74 kpc . for all particles of each component within the virial radius of the system . note that the shape of the dark matter and accreted components are similar and only mildly triaxial , but that the in - situ stars are predominantly oblate . this reflects the fact that in - situ stars have a higher proportion of newly formed stars , which tend to arrange themselves in a disk - like structure ( see figure [ figs : xypanel ] ) . [ figs : cbba ] ] our simulations provide a compelling interpretation where outer stellar halos are made up predominantly of stars accreted during previous merger events and , in particular , by the subset which were propelled into highly energetic ( and highly eccentric ) orbits during the disruption process that accompanies the mergers . mergers and phase mixing lead then to the formation of a tenuous , distinct stellar component that fills the halo of the galaxy out to the virial radius . although we expect them to be ubiquitous , the prominence of these outer luminous halos is difficult to assess from a theoretical point of view , as it depends critically on the number and timing of merger events , as well as on the fraction of stars present at the time the mergers occur , all of which remain highly uncertain . a number of general inferences are , however , still possible . late major mergers should propel a proportionally larger fraction of stars into the outer halo , and therefore the relative prominence of this component should be higher in spheroid - dominated galaxies , which are widely regarded as merger remnants . this process should be especially important in systems that have undergone repeated mergers , leading to the expectation that the relative importance of the `` intracluster '' stellar component should be highest in clusters with prominent cd galaxies . in disk - dominated galaxies , where in - situ star formation is typically still ongoing and mergers are less important , the outer halo is likely to dominate the profile only at large radii , where star formation thresholds prevent the efficient formation of in - situ stars ( kennicutt 1989 , ferguson et al 1998 , schaye 2004 ) . the accreted stellar component to which no such threshold applies appears then as a distinct outer component , `` flattening '' the surface brightness profile in the outer regions , as reported by zibetti et al ( 2004 ) , guhathakurta et al ( 2005 ) and irwin et al ( 2005 ) . , the accreted stars , @xmath69 , and the dark matter , @xmath67 . each of the three different permutations are shown by different shaded histograms , as indicated in the labels of each panel . [ figs : jjdist ] ] , accreted stars , @xmath75 , and the dark matter , @xmath67 , and the minor axis of the dark matter . as in figure [ figs : jjdist ] , histograms of different shades refer to the various permutations , as indicated in the labels . _ ( b)-bottom : _ same as ( a ) , but for the minor axis of the accreted stellar component . [ figs : jcdist],title="fig : " ] , accreted stars , @xmath75 , and the dark matter , @xmath67 , and the minor axis of the dark matter . as in figure [ figs : jjdist ] , histograms of different shades refer to the various permutations , as indicated in the labels . _ ( b)-bottom : _ same as ( a ) , but for the minor axis of the accreted stellar component . [ figs : jcdist],title="fig : " ] quantitatively , we expect the slope of the surface brightness profile of the outer luminous halo to steepen gradually outwards , as dictated by the sersic - law fits summarized in table [ tab : table2 ] . given that the average effective radii of the accreted component is @xmath76 kpc and @xmath77 , we expect its surface brightness profile to steepen from from @xmath78 at the luminous radius ( @xmath79 kpc ) to @xmath80 at @xmath45 kpc . this is in reasonable agreement with the results of zibetti et al ( 2004 ) , irwin et al ( 2005 ) , and guhathakurta et al ( 2005 ) . the latter authors , in particular , report that the m31 outer halo may be approximated by @xmath81 in the range @xmath82 kpc @xmath83 kpc , slightly shallower than in our simulations , but not inconsistent given the substantial observational uncertainty . we note as well that the distribution of globular clusters in the milky way and m31 is consistent with that of the accreted stellar component in the simulations . with the caveat of small number statistics ( only @xmath84 globular clusters around the galaxy and @xmath85 around m31 are currently confirmed ; see harris 1996 and galleti et al 2004 ) , it is interesting that a sersic law with @xmath86 is consistent with the ( projected ) number density profile of galactic globulars . this is shown in figure [ figs : gc ] , where the profile of m31 and galactic globulars is compared with the sersic - law that describes the average accreted stellar component in our simulations . we emphasize that the curves shown are _ not _ fits , and therefore the general agreement is quite suggestive . a simple interpretation , of course , is that the globular cluster population shares the same `` external '' origin as that of accreted stars in our simulations , and that they are thus relicts of the accretion history of each galaxy . ( guhathakurta et al ( 2005 ) note as well that the smooth halo of m31 red giant stars they detect out to @xmath84 kpc mimics the distribution of its globular cluster population . ) taken together , the evidence linking globular clusters with accretion events , as envisioned in the classic model of searle & zinn ( 1978 ) , seems quite strong . in terms of space density , the slope of the accreted stellar component is close to @xmath87 near the edge of the luminous galaxy . this is in reasonably good agreement with that inferred from tracers of the stellar halo of the milky way in the vicinity of the sun ( see , e.g. , zinn 1985 , saha 1985 , morrison et al 2000 ) . our simulations imply that the halo profile should steepen gradually with radius and become significantly steeper than @xmath88 in the outer regions . it is interesting to explore the consequences of this gradual steepening for the interpretation of dynamical data in the outskirts of the milky way . battaglia et al ( 2005 ) report that the velocity dispersion of halo tracers such as blue horizontal branch stars and red giants declines from @xmath89 km / s at the edge of the luminous disk to @xmath90 km / s at @xmath89 kpc . these authors argue that the data favour a dark matter halo model sharply truncated in the outer regions over the extended mass distributions predicted for cdm halos , such as the navarro , frenk & white ( 1996 , 1997 , nfw ) profile . neither halo model , however , is completely satisfactory . extreme radial velocity anisotropy ( @xmath91 ) is required to reconcile the data with an nfw halo model although the opposite ( i.e. , a tangentially anisotropic velocity distribution , @xmath92 ) is required for the truncated halo model . the latter seems an unlikely result given that the outer regions of systems assembled hierarchically are typically dominated by radial motions ( see , e.g. , hansen & moore 2004 and references therein ) . we can use our results to explore the consistency of the velocity dispersion data with cdm halos . we write the spherical jeans equation as , @xmath93 where we have used a @xmath94 subscript to denote quantities associated with the stellar halo . the terms in the left - hand side of eq [ eq : jeans ] may be evaluated by combining the battaglia et al data with the assumption that the stellar halo profile may be approximated by a sersic - law with the ( average ) parameters given in table [ tab : table2 ] . the result of this exercise is shown in figure [ figs : vchalo ] . interestingly , this simple model predicts a circular velocity near the center consistent with that measured at the solar circle ( @xmath95 km s@xmath96 ) without tuning ( solid squares in the bottom panel of figure [ figs : vchalo ] ) . further out , the sustained drop in @xmath97 leads to much lower circular velocities , approaching @xmath84 km s@xmath96 at @xmath45 kpc . this drop in the circular velocity of the system illustrates why battaglia et al s analysis favours a truncated halo . we argue , however , that the data are consistent with an extended cdm halo , and that the velocity dispersion drop places interesting constraints on its total mass . this is shown in the bottom panel of figure [ figs : vchalo ] by the top thick solid line , which corresponds to an ( adiabatically contracted ) nfw halo of virial velocity @xmath98 km / s and concentration @xmath99 ( the average for halos of this mass in the @xmath6cdm cosmogony ) . the halo contribution before and after contraction is shown by the dotted and dashed lines , respectively . the relatively good agreement with the @xmath100 profile derived from the data implies that , rather than being truncated , the halo of the milky way may just be less massive than commonly assumed . the virial mass of the fit shown in figure [ figs : vchalo ] is only @xmath101 , and its virial velocity is only one half of the circular speed at the solar circle . this result is similar to that of klypin , zhao & somerville ( 2002 ) , who analyzed constraints placed by dynamical tracers within the luminous radius of the milky way and found that reconciling such data with cdm halos requires the virial velocity of the dark matter halo to be substantially lower than the rotation speed of the disk . ( their favourite milky way halo model has @xmath102 , which corresponds to @xmath103 kpc , and @xmath104 km s@xmath96 . ) one may rephrase this result by stating that the relatively small dark matter contained within the luminous radius of the galaxy is easiest to accommodate with low - mass halo models , which have the virtue of predicting directly a sharp drop in the velocity dispersion of outer halo tracers . a similar argument was developed by navarro & steinmetz ( 2000 ) , although note the erratum presented in eke , navarro & steinmetz ( 2001 ) . , solid blue line ) , in the outer galaxy ( @xmath20 , dotted red line ) and in satellites ( dashed green line ) enclosed inside the virial radius @xmath105 and averaged over our set of 8 simulations . [ figs : agedist ] ] arcmin@xmath106 kpc at the distance of m31 . we have averaged the distribution of galactic globulars over several random lines of sight to construct the projected number density profile . each bin contains 15 and 40 globulars for the galaxy and m31 , respectively , and has poisson error bars . the two curves are _ not _ fits to the data ; rather , they illustrate the distribution of accreted stars in our simulations . parameters for the two sersic - law curves are chosen to be the average of the data presented in table [ tab : table2 ] and are vertically scaled to match the data for each galaxy . [ figs : gc ] ] it is unclear whether the mass of the halo of the milky way is unusually low compared with other spiral galaxies of similar luminosity , but galaxy - galaxy weak lensing data seem to favour halo masses of the order we find here for late - type @xmath107 galaxies ( see , e.g. , guzik & seljak 2002 ) . also , there is now convincing evidence that the velocity dispersion of satellite galaxies drops at large projected radii so the drop in @xmath100 in the outer regions may actually be a common feature in milky way - like galaxies ( see brainerd 2004 for a review ) . turning our attention now to the velocity structure of the outer halo , we recall the pronounced radial anisotropy shown in figure [ figs : sigmar ] , and argue that such effect should not be ignored when interpreting the dynamics of halo tracers in the outskirts of galaxies . as shown by dekel et al ( 2005 ) , accounting for such anisotropies may be enough to reconcile the steeply declining velocity dispersion profiles of planetary nebulae around bright ellipticals ( romanowsky et al . 2003 ) with the presence of massive dark halos around these objects . the radial anisotropy differentiates dynamically the stars in the outer halo from the dark matter ( figure [ figs : sigmar ] ) . it also distinguishes the stellar halo from the population of surviving satellites , which are found to trace rather faithfully the main properties of the dark halo . indeed , the average anisotropy of the satellite population in our simulations is found to be @xmath108 ( sales et al , in preparation ) , closer to that of the dark matter , and certainly less anisotropic than the accreted stellar component . this dynamical distinction is not surprising , as we find that the smooth stellar halo has little relation with the population of surviving satellites ( see figure [ figs : agedist ] ) . indeed , few stars in the outer halo may be traced to satellites that survive as self - bound entities until the present . the properties of the stellar halo seems more closely linked to the progenitors of early mergers than to the `` harassment '' of the satellite population that survives until the present day . this result has interesting consequences when applied to the interpretation of the origin of the intracluster light ( icl ) component . zibetti et al ( 2005 ) , for example , find that the properties of the icl in their sample of clusters selected from the sdss are quite different from those of the cluster galaxy population as a whole . in particular , the icl is found to be more centrally concentrated than the cluster galaxies and more significantly aligned with the ( brightest cluster galaxy ( bcg ) than cluster galaxies . in other words , the icl seems more intimately related to the central galaxy ( bcg ) than to the cluster galaxy population . all of these properties are consistent with a scenario where the origin of the icl is traced to the mergers that led to the formation of the bcg , with a minor contribution from stars stripped from surviving galaxies . a final item for discussion is the weak alignment between dark halo shapes and angular momentum shown in figure [ figs : jcdist ] . at first , this seems in disagreement with previous studies , many of which report significant alignment between the rotation axis and the minor axis of cdm halos ( see , e.g. , bailin & steinmetz 2004 and references therein ) . the issue may be resolved by noting that the shapes of the dark halos change dramatically when a dissipative baryonic component is included ; indeed , in most cases the dark halos are so nearly spherical that the precise directions of the principal axes are poorly determined . furthermore , as discussed by bailin et al ( 2005 ) , it appears as if the _ inner _ regions of the dark matter halo respond to the assembly of the galaxy by aligning its minor axis with that of the luminous component , but at the same time decouple from the orientation of the halo in the outer regions . interestingly , the weak correlations between halo shape and spin that our simulations predict could be tested directly with upcoming weak lensing surveys . hoekstra et al ( 2004 ) report a tantalizing correlation between the shape of the luminous galaxy and that of their surrounding matter , but their sample contains a broad mixture of morphological types and a relatively strong contribution from spheroid - dominated galaxies , where the relationship between shape and spin is less clear than in spiral galaxies . weak lensing analysis applied to spiral - dominated samples should provide further constraints on this topic . this is currently being addressed using large , morphologically - segregated samples from the sdss ( mandelbaum et al , in preparation ) . , shown by the thin line through the data points . the three other curves ( scale at right ) illustrate the radial dependence of the right - hand side terms in jeans equation ( eq . [ eq : jeans ] ) , derived assuming a sersic profile for the tracers . _ b - bottom : _ solid squares show the _ total _ circular velocity of the milky way derived from the top panel . the top solid curve shows a fit assuming an ( adiabatically contracted ) nfw dark matter halo and an exponential disk . the contribution of the dark matter halo ( before and after contraction ) , as well as that of the disk , are also shown . see text for further details . [ figs : vchalo ] ] we investigate the origin of stars that populate regions well beyond the traditional luminous boundaries of normal galaxies using numerical simulations of galaxy formation in the @xmath6cdm cosmogony . our simulations shows that such components are ubiquitous and owe their presence to the many mergers that characterize the early formation history of galaxies assembled hierarchically . these faint stellar halos extend out to the virial radius of the system , and consist mainly of the overlap and mixing of the many `` tidal tails '' shed by merging progenitors during the assembly of the galaxy . such origin leads to robust predictions for the dynamics and structure of the outer spheroid that may be contrasted with observation . * the density profile of the accreted stellar component which dominates the outer luminous halo is well approximated by a sersic - like model where the logarithmic slope steepens monotonically with radius ; from @xmath109 near the edge of the galaxy s traditional luminous boundary to @xmath1 or steeper near the virial radius . the shape and concentration of the accreted component is in reasonable agreement with the galactic and m31 globular cluster population , lending support to the classic searle - zinn scenario for the formation of the globular cluster population . * the accreted stellar component is reasonably well approximated with a mildly triaxial spheroid with average axis ratios @xmath110 and @xmath111 . rotation plays a negligible role in the support of the accreted stellar component , which is characterized by a strong radial anisotropy in its velocity distribution . this anisotropy grows from the inside out , from @xmath112 in the luminous outskirts of the galaxy to @xmath113 near the virial radius . * the accreted stellar component is distinct from the dark matter halo as well as from the satellite population , which are typically less concentrated and more isotropic . most stars in the outer halo formed in progenitors that have since merged with the central galaxy . only a small fraction of the outer halo stars are contributed through `` harassment '' of satellites surviving as self - bound entities until the present . these properties are in broad agreement with recent observations of the outskirts of spiral galaxies as well as of giant stars in the outer regions of the milky way and of m31 . in particular , they show that the outer light `` excess '' over extrapolations of the inner luminous profile recently reported for m31 and other spirals is a generic feature of galaxies formed hierarchically , and that the decline in velocity dispersion seen in the outer milky way halo tracers is consistent with the presence of an extended , massive halo of dark matter ( albeit one of perhaps lower mass than commonly assumed ) . they also imply that the intergalactic stellar component of galaxy clusters may be more intimately related to the central galaxy than to the galaxy cluster population as a whole . these results illustrate ways to unravel the clues to the tumultuous accretion history of individual galaxies contained in the stars ejected beyond their borders . we acknowledge useful discussions with simon white , stefano zibetti , and vincent eke . we thank laura sales for allowing us to quote some of her results in advance of publication and giuseppina battaglia for providing the data used in figure [ figs : vchalo ] in electronic form . jfn acknowledges support from nserc , the canadian foundation for innovation , as well as from the alexander von humboldt and leverhulme foundations .
we use numerical simulations to investigate the origin and structure of the luminous halos that surround isolated galaxies . these stellar structures extend out to several hundred kpc away from a galaxy , and consist of stars shed by merging subunits during the many accretion events that characterize the hierarchical assembly of galaxies . such origin suggests that outer luminous halos are ubiquitous and that they should appear as an excess of light over extrapolations of the galaxy s inner profile beyond its traditional luminous radius . the mass profile of the accreted stellar component is well approximated by a model where the logarithmic slope steepens monotonically with radius ; from @xmath0 at the luminous edge of the galaxy to @xmath1 or steeper near the virial radius of the system . such spatial distribution is consistent with that of galactic and m31 globular clusters , suggesting that many of the globulars were brought in by accretion events , in a manner akin to the classic searle - zinn scenario . luminous halos are similar in shape to their dark matter counterparts , which are only mildly triaxial and much rounder than dark halos formed in simulations that do not include a dissipative luminous component . the outer stellar spheroid is supported by a velocity dispersion tensor with a substantial and radially increasing radial anisotropy ; from @xmath2 at the edge of the central galaxy to @xmath3 at the virial radius . these properties distinguish the stellar halo from the dark matter component , which is more isotropic in velocity space , as well as from some tracers of the outer spheroid such as satellite galaxies . most stars in the outer halo formed in progenitors that have since merged with the central galaxy ; very few stars in the halo are contributed by satellites that survive as self - bound entities at the present . these features are in reasonable agreement with recent observations of the outer halo of the milky way , of m31 , and of other isolated spirals , and suggest that all of these systems underwent an early period of active merging , as envisioned in hierarchical models of galaxy formation . [ firstpage ] galaxy : disk galaxy : formation galaxy : kinematics and dynamics galaxy : structure
oxides in the @xmath2o@xmath0 perovskite family present a multitude of functional properties and are widely renowned for their potential in technological applications . construction of heteroepitaxial thin films is being actively explored as a route to further enhance and expand on the existing oxide functionalities . the presence of an interface between an oxide and a substrate can dramatically affect material properties , particularly if a film is grown coherently so that its in - plane lattice constant is forced to match that of the substrate . the resulting heteroepitaxial strain has been credited , for example , with dramatically enhancing the ferroelectric polarization and curie temperature in thin film batio@xmath0 @xcite and inducing ferroelectricity in usually non - polar srtio@xmath0.@xcite in addition to ferroelectric distortions , we increasingly find repercussions of strain in the patterns of rigid rotations of the corner - sharing @xmath3o@xmath1 octahedral units . these changes in rotation have been invoked to account for such phenomena as the strain - dependence of magnetic properties , as a result of the corresponding changes in the electronic bandwidth.@xcite in this work , we use density functional theory to examine the effect of epitaxial strain on the rotational instabilities in laalo@xmath0 ( lao ) . lao is an ideal model system in which to isolate the interplay between strain and rotations because it exhibits only rotational distortions in its ground state with no indications of ferroelectric or jahn - teller instabilities . we are also motivated by a need to understand the influence of octahedral rotations in oxide heterostructures such as lao / srtio@xmath0 ( sto ) , which has been reported to form a highly conductive electron gas at the interface in spite of the insulating nature of the two parent compounds.@xcite the propagation of octahedral rotations across the interface is likely to be relevant to its electronic properties , but microscopic techniques have limited access to the positions of oxygen ions.@xcite an improved understanding of the effect of strain on the parent compound lao is therefore highly desirable . = 5.0 cm = 8.5 cm here we investigate the effect of biaxial strain in lao . we find that the bulk @xmath4 structure is destabilized by the constraint of coherent epitaxy and that compressive and tensile biaxial strains greater than @xmath50.2% stabilize previously unidentified phases of lao . in addition , we find that uniaxial relaxation of the biaxial epitaxial constraint stabilizes a third phase . the primary structural differences between these strain - stabilized phases and the parent phase are the patterns of octahedral rotations , and our analysis of the transition between them provides insight into the coupling between strain and rotations in lao and similar @xmath2o@xmath0 perovskites . for all calculations we use the local density approximation ( lda ) of density functional theory as implemented in the vienna _ ab - initio simulation package , vasp.@xcite we use the projector augmented wave method @xcite with the default vasp lda potentials ( la , al , o ) and a plane wave energy cut - off of 800 ev . we use a 10-atom rhombohedral unit cell with a 5@xmath65@xmath65 _ k_-point sampling . to determine ground state structures we relax ionic positions to a force tolerance of 1 mev / . space group determinations are performed with the findsym symmetry analysis software.@xcite _ strain is calculated as @xmath7 , where @xmath8 is the lda equilibrium lattice parameter . we use the notation @xmath9 to denote the pseudocubic ( pc ) lattice constants ( except where noted ) , which correspond to the crystallographic directions [ 100]@xmath10 , [ 010]@xmath10 , and [ 001]@xmath10 . our computational unit cell , shown in fig . [ fig : uc ] , is defined by lattice vectors ( @xmath11 ) , ( @xmath12 ) , and ( @xmath13 ) . lattice distortions comprised of rigid octahedral rotations are described by the angles of rotation around the pseudocubic axes , @xmath14 , and @xmath15 , as shown in fig . [ fig : strain - e ] . for completeness we also give the glazer notations for our structures , where , e.g. , @xmath16 is written as @xmath17 if consecutive octahedra along @xmath18 rotate in the same direction ( in - phase ) or @xmath19 if consecutive octahedra rotate in the opposite directions ( out - of - phase).@xcite multiple subscripts indicate a compound rotation , e.g. @xmath20 denotes rotation around the @xmath21 axis ( crystallographic direction [ 110]@xmath10 ) . ( note that the @xmath22 are not strictly independent , however a rigorous decomposition made using irreducible representations yields nearly identical results . ) the bulk crystal structure of lao deviates from the ideal @xmath23 perovskite by out - of - phase rotations around the crystallographic [ 111 ] axis ( @xmath24 , glazer system @xmath25 ) that lower the symmetry to space group @xmath4 , and an accompanying rhombohedral distortion . we reproduce this structure by relaxing lattice parameters and atom positions in our 10-atom unit cell , and calculate a rhombohedral angle of 60.2@xmath26 and an equilibrium lattice constant of @xmath27 ( pseudocubic @xmath28 ) , which underestimates the experimentally determined value by @xmath291% , a common artifact of lda calculations . we calculate @xmath24 = 5.98@xmath26 , close to the experimentally determined value.@xcite = 7.0 cm m2.2 cm m2.4 cm m2.2 cm m1.8 cm m2 cm m4.5 cm & rotation angles & & rotation axis & glazer notation & conditions for stability + @xmath4 & & [ 111 ] & @xmath25 & unconstrained + @xmath30 & @xmath31 & @xmath32 & [ 111]@xmath10&@xmath33 & -0.2% @xmath34 0.1% , biaxial + @xmath35 & @xmath36 & @xmath32 & [ 001]@xmath10&@xmath37 & @xmath38 -0.2% , biaxial + @xmath39 & @xmath31 & @xmath40 & [ 110]@xmath10&@xmath41 & @xmath42 0.1% , biaxial + @xmath43 & @xmath44 & @xmath45 & [ 100]@xmath10&@xmath19 & small uniaxial strain , @xmath46 + to mimic clamping to a substrate , we enforce a square lattice in the plane of epitaxy ( @xmath18 = @xmath47 ) and constrain the value of in - plane lattice parameter @xmath18 while relaxing the out - of - plane parameter @xmath48 and atom positions . we also allow a [ 110]@xmath10 shear of the unit cell . the resulting structure , for @xmath49 3.746 has space group @xmath30 ; it retains the @xmath24 rotations of the bulk and has a relaxed monoclinic angle @xmath50 90.3@xmath26 . we then investigate the effect of biaxial strain by adjusting the value of @xmath18 . for strains of -2% to + 2% we identify two competing phases with space groups @xmath35 and @xmath39 . the resulting energy - versus - strain phase diagram is shown in fig . [ fig : strain - e ] , and details of the transition region are shown in fig . [ fig : order - params ] . to facilitate fine - sampling near the transition we do not relax @xmath48 in these calculations but instead maintain the equilibrium volume . later we relax this constraint and find no qualitative changes to the phase diagram . we find that the shear distortion is only energy - lowering for the @xmath30 phase and all other phases retain @xmath50 90@xmath26 . at very small strains of -0.2% to 0.1% , the ground state is @xmath30 . the structure exhibits the @xmath24 rotations of the bulk and has a relaxed monoclinic angle @xmath50 90.3@xmath26 . as strain is increased , the @xmath30 phase becomes rapidly unstable and the system transitions to @xmath39 or @xmath35 , depending on the sign of the strain . compressive strain greater than -0.2% stabilizes the @xmath35 phase , in which the lattice distortion consists of @xmath15 rotations ( glazer system @xmath37 ) , whereas tensile strain greater than 0.1% stabilizes the @xmath39 phase , comprised of @xmath20 rotations ( @xmath41 ) , as shown in fig . [ fig : structures ] . thus , the bulk - like @xmath30 phase is found only in a very narrow range of strain near the equilibrium lattice parameter . these results are summarized in table [ table ] and structural parameters for each phase at a representative strain are given in table [ table2 ] . ( note that the precise range of stability for the three ground - state phases is difficult to identify from total energies . thus we obtain our predicted range of the @xmath30 phase by calculating the zone - center phonons of each phase as a function of strain and comparing the frequencies of the softest non - trivial modes . ) to examine the coupling between strain and rotations , we decompose the distortion from a high - symmetry parent phase with space group @xmath51 in terms of irreducible representations ( or _ irreps _ ) using the software isodisplace.@xcite we find that the dominant irreps are @xmath52 and @xmath53 , which correspond to @xmath15 and @xmath20 rotations , respectively . the evolution of the irrep amplitudes in the lowest - energy structures is shown in fig . [ fig : rot_angle ] . we also calculate @xmath15 and @xmath20 from the angles between neighboring octahedra ( in the manner of ref . ) to provide an approximate correspondence between irrep amplitude and rotation angle ( fig . [ fig : rot_angle ] ) . within @xmath35 and @xmath39 we find a roughly linear dependence of rotation angles on lattice parameter . our results suggest a simple model to describe the response of lao to biaxial strain . in bulk lao the @xmath24 rotations reduce the bonding distances around the small la cation and minimize the sum of coulombic and repulsive energies , while the dimensions of the octahedra are determined by the strongly covalent al - o bonds.@xcite within this lattice , strain - related changes in lattice spacing are accommodated by changes in either the rotations or the al - o bond lengths . under biaxial strain , we find two distinct responses within the range of strains investigated , as seen in the evolution of al - o distances and octahedral rotations , shown in fig . [ fig : order - params ] . at very small values of strain ( -0.2% to 0.1% ) the changes in lattice dimensions are accommodated primarily by rotations while al - o distances remain nearly constant , and the system remains in the @xmath30 phase . under compressive ( tensile ) strain , transformation to @xmath35 ( @xmath39 ) occurs when @xmath20 ( @xmath15 ) goes to zero . increased strain is then accommodated by changes to al - o distances , while the rotations remain nearly constant . we note that the precise values of strain at which the phase transformations are predicted to occur are dependent on very small energy differences and are therefore somewhat sensitive to the computational parameters used . for example , test calculations performed using the generalized gradient approximation for the exchange - correlation energy indicated small variations in the quantitative details of the phase diagram but the qualitative model does not change significantly . = 8.0 cm similar strain - induced variations in tilt patterns have recently been reported in a number of other perovskite oxides , including srruo@xmath0,@xcite bifeo@xmath0,@xcite and lanio@xmath0.@xcite the behavior of lao is distinct from these systems , however , in the extremely narrow window in which the parent phase is stable and the dramatically different structures stabilized by compressive and tensile strain . we next examine the effect of a uniaxial relaxation of tensile strain in lao . we compare the energetics of @xmath20 rotations ( @xmath39 phase ) to a phase with only @xmath16 rotations ( space group @xmath43 ) . under biaxial strain , the @xmath43 phase is @xmath290.5 mev higher in energy than @xmath39 in the range of strains investigated , but a uniaxial relaxation of the strain , such that @xmath54 , alters the balance of energy between the two . starting with atom configurations from the relaxed structures for @xmath55 , we vary the ratio @xmath56 and calculate the total energy for the two phases . we maintain a constant in - plane area @xmath21 and do not relax @xmath48 . we first explore the system with @xmath57 ( 3.746 )@xmath58 and find that a 0.5% distortion of @xmath56 stabilizes @xmath16 rotations relative to @xmath20 by @xmath291 mev . we compare this to the system with @xmath57 ( 3.85 )@xmath58 , @xmath293% tensile strain , and find that the uniaxial relaxation is no longer energetically favorable , and the ground - state retains @xmath20 rotations and @xmath55 . our results indicate that lao films on substrates with a lattice mismatch of @xmath290.5% may , in theory , lower their energy through a partial relaxation such that @xmath54 . the resulting structure exhibits @xmath16 rotations ( for @xmath59 ) in the space group @xmath43 . in practice , no common growth substrate provides these conditions , but the result is likely to be a general phenomenon in similar materials . the data suggest an explanation for the experimental observation of @xmath60 in some coherent films grown on square substrates . for example , bifeo@xmath0 films thinner than 50 nm grown on lao substrates are reported to have @xmath613.84 and @xmath623.76 , compared to the substrate parameters @xmath63 , indicating a uniaxial relaxation of the epitaxial constraint.@xcite our results for lao at 3% tensile strain , which do not predict a uniaxial strain relaxation , are consistent with experimental data for films of lao on sto ( a mismatch strain of 3% ) . such films are reported to have square in - plane lattice parameters within an accuracy of 0.01 , indicating that any distortion of @xmath56 must be smaller than 0.5%.@xcite m3.0 cm m1.2 cm m1.2 cm m1.2 cm m1.5 cm m1.7 cm m1.5 cm m1.5 cm m1.2 cm & a ( ) & b ( ) & c ( ) & @xmath64 & wyckoff pos . & _ x _ & _ y _ & _ z _ + @xmath30 & 9.15 & 5.2976 & 5.2976 & 54.927@xmath26 & la ( 4e ) & 0.0 & 0.2500 & 0.25 + _ ( 0% strain ) _ & & & & & al ( 4d ) & 0.25 & 0.25 & 0.5 + & & & & & o ( 8f ) & -0.2652 & 0.4829 & 0.2823 + & & & & & o ( 4e ) & 0.0 & -0.2804 & 0.25 + @xmath35 & 5.2712 & 5.2712 & 7.5634 & - & la ( 4b ) & 0.0 & 0.5 & 0.25 + _ ( -0.5% strain ) _ & & & & & al ( 4c ) & 0.0 & 0.0 & 0.0 + & & & & & o ( 8h ) & -0.2212 & 0.2788 & 0.0 + & & & & & o ( 4a ) & 0.0 & 0.0 & 0.25 + @xmath39 & 7.4642 & 5.3156 & 5.3156 & - & la ( 4e ) & 0.0 & 0.25 & -0.2491 + _ ( 0.5% strain ) _ & & & & & al(4c ) & 0.25 & 0.25 & 0.25 + & & & & & o ( 8f ) & -0.2298 & 0.0 & 0.0 + & & & & & o ( 4e ) & 0.0 & 0.25 & 0.2894 + @xmath43 & 7.4800 & 7.4502 & 7.5099 & - & la ( 8i ) & 0.0 & 0.0 & -0.2501 + _ ( @xmath290% strain ) _ & & & & & al ( 8e ) & 0.25 & 0.25 & 0.0 + & & & & & o ( 8f ) & 0.25 & 0.25 & 0.25 + & & & & & o ( 8 g ) & -0.2770 & 0.0 & 0.0 + & & & & & o ( 8h ) & 0.0 & 0.2231 & 0.0 + finally , we address the effect of manually disabling octahedral rotations . for this we treat a five - atom tetragonal unit cell in which rotation of the oxygen octahedra is forbidden by symmetry . at each value of epitaxial strain we relax the positions of the ions and calculate the electric polarization using the berry phase method.@xcite we find that , while the unstrained system is non - polar , an abrupt transition to a state with large polarization occurs at -3% strain . the polarization is out - of - plane and results from a structural distortion within @xmath51 symmetry in which al and la displace along [ 001 ] . a compressive strain of 4% results in a polarization of 38 @xmath65c @xmath66 relative to a centrosymmetric reference structure . these results indicate the existence of an incipient ferroelectric mode that is usually suppressed by the dominant antiferrodistortive rotational modes . in summary , we find that several different phases of lao , characterized by distinct patterns of octahedral rotations , can be stabilized by varying the epitaxial constraints over a range that is readily accessible experimentally . this remarkable structural softness indicates that it is unlikely that the bulk - type @xmath24 rotations will be observed in heteroepitaxial thin films of lao on any substrate . this has important implications for the interpretation of structural data for lao films , with particular relevance to investigations of the properties of lao / sto interfaces . we find that biaxial strain of -0.2% to 0.1% results in large changes in octahedral tilt angles in the bulk - like @xmath30 phase of lao . larger strains , however , induce a tilt - driven transition to one of three phases in which at least one tilt component is zero ( @xmath35 , @xmath39 , or @xmath43 ) . within these strained phases , we find that changes in lattice parameters related to epitaxial strain are largely accommodated by changes in the al - o bond lengths , with only small changes to tilt angles , and without any significant effect on electronic properties . these findings provide insight to the growing list of complex oxide perovskites in which the rotational distortions are observed to depend on lattice parameters . given the coupling between rotations and strain in lao , and its non - polar structure , we propose it as a model system for exploring new non - linear optical techniques for probing octahedral rotations.@xcite finally , for small values of tensile strain we find that a uniaxial relaxation of the strain , such that @xmath67 , stabilizes @xmath16 rotations over @xmath20 . our results suggest a route to selectively stabilize different tilt patterns via the substrate geometry . conversely , these results also suggest an explanation for the unequal in - plane lattice parameters observed in some epitaxial thin films of perovskites , in which similar energetics of @xmath16 and @xmath20 rotations may drive a distortion of @xmath56 . this work was supported by the national science foundation ( nsf ) , grants no . nirt-0609377 ( nas ) and dmr-0820404 ( ajh ) . computational support was provided by the mrl central facilities under the nsf mrsec program , award no . dmr05 - 20415 , and the cnsi computer facilities at uc santa barbara , nsf grant no . we thank j. m. rondinelli and m. stengel for helpful discussions .
we investigate the effect of epitaxial strain on [ 001]-oriented laalo@xmath0 using first - principles density functional calculations . we find a series of structural phase transitions between states characterized by different patterns of tilting of the alo@xmath1 octahedra . by tuning the biaxial strain from compressive to tensile , we induce an evolution in the crystal structure in which the tilt axis changes from out - of - plane to in - plane , corresponding to space groups _ i4/mcm and _ imma . we also study the effect of uniaxial relaxation of the usual biaxial epitaxial constraint and explore this as a mechanism for selectively stabilizing different patterns of octahedral tilts . _ _
during the last decades several observational efforts have been carried out to determine the characteristics of the molecular gas and dust associated with massive star forming regions . surveys of molecular emission in high density tracers ( plume et al . 1992 ; juvela 1996 ; plume et al . 1997 ; shirley et al . 2003 ) and of dust continuum emission ( beuther et al . 2002 ; mueller et al . 2002 ; fandez et al . 2004 ; williams et al . 2004 ; garay et al . 2007 ) , made with single dish telescopes , show that the maternities of high - mass stars are regions of molecular gas with distinctive physical parameters , with typical radii of 0.4 pc , densities of @xmath22 @xmath14 and masses of @xmath23 @xmath24 . the dust observations also show that these regions have dust temperatures of typically @xmath25 k , indicating that a luminous energy source has been already formed inside them . these structures , also refereed as massive and dense cores , have typically column densities of @xmath26 @xmath12 ( garay et al . 2007 ) which make them very dark at optical wavelengths and even dark at infrared wavelengths ( menten et al . 2005 , rathborne et al . the iras luminosity of most regions in these studies are in the range @xmath27 @xmath28 . in order to investigate possible differences , in the physical and chemical characteristics as well as in the stellar population , between massive and dense cores of different luminosities , we have carried out a multi - wavelength study of four highly luminous iras sources in the southern hemisphere selected from the list of iras point sources with colors of uc hii regions and detected in cs(2@xmath11 ) emission compiled by bronfman et al . the observed objects are listed in table [ tbl - obssources ] 2 , 3 , and 4 give , respectively , the name , right ascension , and declination of the associated iras source . the kinematic distance to the objects is given in col . the four objects are located in the tangent region of the norma spiral arm and therefore there is little ambiguity in the kinematic determination of their distances ; their @xmath29 s are nearly the maxima allowed by pure circular motion in their lines of sight . the total far infrared luminosities , @xmath30 , are given in col.6 . the objects have @xmath30 greater than @xmath31 @xmath28 and their cs(2@xmath11 ) line profiles show broad molecular line wing emission suggesting the presence of powerful molecular outflows . here we report molecular line and dust continuum observations towards these four regions . the aim of these observations is to undertake a detailed study of the molecular gas and dust content of luminous massive and dense cores in order to determine their physical conditions , kinematics , and chemistry . molecular emission was mapped in a number of species that are commonly used as diagnostics of the conditions of massive and dense cores : carbon monosulfide , silicon monoxide , and methanol ( e.g. , zinchenko et al . 1995 ; juvela 1996 ) . thermal emission of sio is both a good probe of high velocity gas in regions of massive star formation ( downes et al . 1982 ) and a tracer of high - temperature chemistry ( ziurys et al . the methanol abundance is highly enhanced in the hot gas surrounding infrared sources ( menten et al . 1988 ) as well as in shocked gas ( liechti & walmsley 1997 ) , thus thermal methanol emission is an excellent probe of the environments of high - mass yso s . the cs and dust continuum observations are intended to determine the density structure of the cores . in a subsequent paper ( chavarra et al . 2009 ; paper ii ) we will present the characteristics of the stellar content within these regions . the molecular line and dust continuum observations were carried out using the 15-m swedish - eso submillimetre telescope ( sest ) located on la silla , chile . the molecules , transitions , and frequencies observed and the instrumental parameters are summarized in table [ tbl - obspar ] . the cs(@xmath32 ) observations were made between december 1990 and january 1992 . during this period the telescope was equipped with a cooled schottky mixer and an acousto - optical spectrometer with 2000@xmath3343 khz channels . the typical system temperature above the atmosphere was 500 k. the cs(@xmath32 ) line emission was mapped with spacings of 45 within regions of at least 135@xmath33135 . the observations were done in the frequency switch mode with a frequency throw of 15 mhz . the sio(@xmath32 ) , sio(@xmath34 ) , ch@xmath2oh(@xmath5 ) , ch@xmath2oh(@xmath35 ) , cs(@xmath34 ) and c@xmath6s(@xmath34 ) observations were made during march of 1997 . in this epoch the telescope was equipped with sis receivers , and it was possible to simultaneously observe lines at 2 and 3 mm wavelengths . single - sideband receiver temperatures were typically 120 k for both receivers . as backend we used high resolution acousto - optical spectrometers providing a channel separation of 43 khz and a total bandwidth of 43 mhz . the emission in the above lines was mapped within regions of at least 90@xmath33 90 , with 30 spacings . the observations were performed in dual beam - switching mode , with a beam separation of 1147 in azimuth . the integration times on source per position and the resulting rms noise in antenna temperature are given in table [ tbl - tinrms ] . within the available bandwidths , three rotational lines of ch@xmath2oh could be observed at 2 mm ( a@xmath36 , e , and e lines ) and four rotational lines at 3 mm ( e , e , a@xmath36 , and e lines ) . during march 2001 we further observed the sio ( 217104.935 ghz ) and ch@xmath2oh(@xmath37 ) ( 84521.21 ghz ) lines toward the center of the g330.949 - 0.174 core . the receivers and backends were the same as described above . single - sideband receiver temperatures were 1200 k at 217 ghz and 190 k at 85 ghz . the cs(@xmath38 ) observations of g324.201 , g329.337 , and g330.949 were made between september 1991 and january 1992 . in this epoch the telescope was equipped with a cooled schottky mixer , and an acousto - optical spectrometer with 2000@xmath3343 khz channels ; however , a setup of 1000@xmath3386 khz resolution elements was used . the typical system temperature above the atmosphere was 1850 k. the cs(@xmath38 ) line emission was mapped with spacings of 15 within regions of at least 60@xmath3345 . the observations were done in the frequency switch mode with a frequency throw of 15 mhz . the cs(@xmath38 ) observations of g328.307 were made in may 2001 , with an sis receiver and the acousto - optical spectrometer with 2000@xmath3343 khz channels . typical system temperature above the atmosphere was of 450 k. observations were done in dual beam switching mode , with a beam separation of 1147 in azimuth , in a grid of 9 positions with spacings of 45 . observations of the co , @xmath7co , and c@xmath8o emission in the j=1@xmath10 lines were made during may , 2000 and june , 2001 . spectra were obtained only toward the peak position of the cs emission in each region . these observations were performed in the position switched mode , with off positions taken from bronfman et al . the goal of these observations was to obtain the data needed to determine co column densities , which in turn will permit to derive abundance enhancements relative to co. the 1.2-mm continuum observations were made using the 37-channel sest imaging bolometer array ( simba ) during october , 2001 and august , 2003 . the passband of the bolometers has an equivalent width of @xmath39 ghz and is centered at @xmath40 ghz . the hpbw of a single element is 24 and the separation between elements on the sky is 44 . we observed in the fast mapping mode , using a scan speed of 80 s@xmath41 . each observing block consisted of 50 scan lines in azimuth of length 800and separated in elevation by 8 , giving a map size of 400 in elevation . this block required @xmath42 15 minutes of observing time . we observed @xmath43 blocks per source , achieving rms noise levels of typically @xmath44 mjy beam@xmath41 . the data were reduced according to a standard procedure using the software package mopsi , which included baseline subtraction and rejection of correlated sky - noise . flux calibration was performed using a sky - opacity correction and a counts - to - flux conversion factor derived from maps of uranus . uncertainties in the absolute calibration and pointing accuracy are estimated at @xmath45 and 3 , respectively . figs . [ fig - specmapg324 ] to [ fig - specmapg330 ] show spectral maps of the cs(2@xmath11 ) , cs(3@xmath12 ) , c@xmath6s(3@xmath12 ) , sio(2@xmath11 ) , sio(3@xmath12 ) , ch@xmath2oh(3@xmath4 ) , and ch@xmath2oh(2@xmath46 ) line emission observed toward g324.201 + 0.119 , g328.307 + 0.423 , g329.337 + 0.147 and g330.949 - 0.174 , respectively . the spacing is 30 for all transitions , except cs(2@xmath11 ) for which the spacing is 45 . offsets are from the reference position given in columns 3 and 4 of table [ tbl - obssources ] . table [ tbl - linepar ] gives the parameters of the line emission observed at the peak position ( cols . 2 to 4 ) and of the spatially averaged emission ( cols . 5 to 7 ) , determined from gaussian fits to the respective line profiles . carbon monosulfide was detected in all observed transitions towards all sources . the cs(2@xmath11 ) and cs(3@xmath12 ) line profiles indicate the presence of two emission components : ( i ) a bright , spatially extended , and relatively narrow velocity feature ; and ( ii ) a weaker and broader velocity feature , which is usually spatially confined to the central region . the line widths of the narrow component emission , which most likely arises from the dense ambient gas , are typically about 5 , a value considerably larger than the thermal width expected for kinetic temperatures of @xmath4240 k , of @xmath47 , implying that the line widths are dominated by non - thermal supersonic motions . whether the non - thermal motions are due to the effect of stellar winds or are intrinsic to the initial conditions of the formation process of massive stars is still an open question . the velocity range of the broad wing emission ( full width at zero power ) in different transitions are given in table [ tbl - wingpar ] . the broad emission , which has typically full widths at zero power of @xmath48 , is mainly detected toward the peak position of the massive cores . although this emission most likely arises from outflowing gas , the coarse angular resolution ( @xmath49 ) of our observations does not allow to determine its spatial distribution and degree of collimation . the nature of the high velocity gas giving rise to the pedestal feature is thus difficult to assess . higher angular resolution observations are needed to investigate whether the high velocity emission arises from a collimated bipolar outflow driven by the recently formed massive stars or from an spherically symmetric structure possible driven either by stellar winds or by the expansion of the hii region onto the ambient medium . silicon monoxide was detected toward the four observed high - mass star forming regions , the strongest emission being observed toward the g330.949 - 0.174 core . sio lines are tracers of both the warm ambient gas surrounding young luminous stars and of molecular outflows associated with newly formed massive stars . harju et al . ( 1998 ) have already shown that sio emission is relatively easy to detect toward regions of high - mass star formation , with the detection rate increasing with the fir luminosity . the profiles of the sio emission show a broad component of high velocity gas superposed on a narrower component , whose central velocity is close to the velocity of the ambient gas . the full widths at zero power ( emission above a @xmath50 level ) of the broad component range from 17.4 to 36.6 , with a median of 26 . the relative contribution to the observed intensity is different from source to source . in g324.201 the sio emission from the broad component dominates the spectrum , whereas in g330.949 the narrower emission is stronger . we suggest that , on the one hand , the narrow sio emission arises from the warm ambient gas surrounding the recently formed massive stars , and that this component results from the evaporation of silicon compounds from grain mantles produced by the stellar radiation . on the other hand , the broad sio emission most likely indicates mass outflows from a star or stars embedded within the massive core , and that this component results from grain disruption by shock waves . hence , the sio emission traces both high kinetic temperature ambient gas and outflowing shocked gas . the shape of the profiles are different from molecule to molecule depending on the relative intensities between the narrow and broad components . to illustrate in more detail the differences between the line profiles , figs . [ fig - specpeakg324 ] to [ fig - specpeakg330 ] show the spectra of all molecular lines , except methanol , observed at the peak position of each region . the cs(5@xmath14 ) spectra corresponds to the average spectra over the 30@xmath3330 mapped region , and therefore it is directly comparable to the peak spectra observed in the other transitions with @xmath4230 beams . [ fig - csmaps ] shows contour maps of the velocity integrated ambient gas emission in the cs(2@xmath11 ) line toward all four regions . the 50% contour level of the emission is marked with dark lines . in all cases the emission , at the resolution of 52 , arises form a single source exhibiting a simple morphology . the deconvolved major and minor fwhm angular sizes determined by fitting a gaussian profile to the emission are given in 3.3 . [ fig - dustsimba ] presents maps of the 1.2-mm continuum emission , within regions of typically @xmath51 in size , towards the four observed sources . emission was detected toward all four regions . the extent and morphology of the 1.2-mm dust continuum emission is similar to that of the cs line emission , indicating that these two probes trace the same physical conditions . the observed parameters of the 1.2-mm sources are given in table [ tbl - obspardust ] . cols.(2 ) and ( 3 ) give their peak position . cols.(4 ) and ( 5 ) give , respectively , the peak flux density and the total flux density , the later measured directly from the maps using the aips tasks imean . col.(6 ) gives the deconvolved major and minor fwhm angular sizes determined by fitting a single gaussian profile to the whole spatial distribution , except for g329.337 for which two gaussian components were fitted . in three of the four regions the morphology of the dust emission can be described as arising from a bright compact peak surrounded by an extended envelope of weaker emission , which is characteristic of sources with centrally condensed density profiles . this is illustrated in fig . [ fig - intcuts ] which shows slices of the observed 1.2-mm intensity across the two sources with nearly circular morphology ( g324.201 and g330.949 ) . we find that the observed radial intensity profiles can be well fitted with single power - law intensity profiles of the form @xmath52 , where @xmath53 is the distance from the center , properly convolved with the beam of 24 . we derive @xmath54 indices of 1.6 and 1.9 for the g324.201 and g330.949 cores , respectively . for the g328.307 core an @xmath54 index of 1.5 has been already derived by garay et al . ( 2007 ) . three of the high - mass star forming regions investigated here , g324.201 , g329.337 and g330.949 , are in the list of @xmath422000 massive young stellar objects ( myso ) candidates compiled by the rms survey based on a colour selection criteria of msx point sources ( hoare et al . radio continuum observations have , however , shown that all four norma cores are associated with compact radio sources , indicating that they are in a more advanced stage of evolution in which an uchii region has already developed ( walsh et al . 1998 ; garay et al . 2006 ; urquhart et al . 2007 ) . in all cases the compact radio sources are found projected at the peak position of the 1.2-mm dust continuum emission , suggesting that massive stars are located at the center of the massive and dense cores . in what follows we discuss the characteristics of the molecular gas and dust emission , as well as of the radio emission , toward each of the iras sources , individually . _ g324.201 + 0.119_.@xmath55 the g324.201 + 0.119 massive core is associated with the iras source 15290 - 5546 , which has a total luminosity of @xmath56 . the cs profiles across the core show emission from a nearly gaussian component , with central velocities of @xmath57 and fwhm widths of @xmath58 . the spatial distribution of the emission in the cs(@xmath32 ) line from the ambient gas is slightly elongated , with fwhm deconvolved major and minor axis of 43 and 30 , respectively . the morphology of the 1.2-mm dust emission also shows the presence of a single structure with fwhm deconvolved major and minor axis of 35 and 28 , respectively . a power - law fit to the 1.2-mm intensity profile gives an index of 1.6 ( see fig . [ fig - intcuts ] ) . the total flux density of the region at 1.2 mm is 17.3 jy . toward the center of the core the gaussian component is superposed on a broader and weaker component . in the cs(@xmath34 ) line the wing emission extends up to a flow velocity of @xmath59 in the blue side and 11.4 in the red side . the flow velocity , @xmath60 , is defined as @xmath61 , where @xmath62 is the ambient cloud velocity . the sio line profiles are , on the other hand , dominated by the emission from the broad component , covering a flow velocity range from @xmath63 to @xmath64 . radio continuum observations towards g324.201 + 0.119 show the presence of two compact radio sources : a cometary - like component , with an angular diameter of @xmath424 `` , located at the center of the core and an unresolved component located @xmath426 '' north of the former ( urquhart et al . if excited by single stars , the rate of uv photons needed to produce the cometary and unresolved regions of ionized gas implies the presence of o7 and o8.5 zams stars , respectively . _ g328.307 + 0.423_.@xmath55 the g328.307 + 0.423 massive core is associated with the iras source 15502 - 5302 , which has a total luminosity of @xmath65 . the spatial distribution of the cs(@xmath32 ) emission from the ambient gas shows an elongated morphology , with fwhm deconvolved major and minor axis of 65 and 37 , respectively . the elongated morphology is more clearly seen in the observed spatial distribution of the 1.2-mm emission ( see fig . [ fig - dustsimba ] ) , which has fwhm deconvolved major and minor axis of 43 and 26 , respectively . the total flux density of the region at 1.2-mm is 24.1 jy . the co(@xmath66 ) spectra observed at the center of the core show a double peaked profile , with a strong blueshifted peak at the velocity of @xmath67 and a weaker redshifted peak at the velocity of @xmath68 . on the other hand , the spectra of the optically thin c@xmath8o(@xmath66 ) line show a single gaussian profile with a peak velocity of -92.6 . these spectral characteristics indicate the presence of infall motions ( leung & brown 1977 , walker , narayanan , & boss 1994 , myers et al . 1996 ) . using expression ( 9 ) of myers et al . ( 1996 ) we derive an infall velocity of @xmath69 . the profiles of the cs and c@xmath6s emission observed across the core are generally asymmetric , showing a peak at velocities of @xmath70 and a shoulder toward redshifted velocities . the observed shapes are consistent with the infall motions suggested by the co(@xmath66 ) emission but for transitions with smaller optical depths . in addition to infalling gas , outflowing gas is clearly detected toward the central region . in the cs(3@xmath12 ) line , blueshifted and redshifted emissions are detected up to flow velocities of @xmath71 and 12.2 , respectively . the sio emission , which is detected in the velocity range from @xmath72 to @xmath73 , arises mainly from the outflow component , but a narrower component from the ambient core gas is also present . radio continuum observations towards g328.307 + 0.423 show the presence of a complex region of ionized gas , consisting of a handful of low brightness , extended components and a bright , compact component located at the center of the core ( garay et al . the flux density of the later component increases with frequency whereas its angular size decreases with frequency . these characteristics suggest the presence of a steep gradient in the electron density . _ g329.337 + 0.147_.@xmath55 the g329.337 + 0.147 massive core is associated with the iras source 15567 - 5236 , which has a total luminosity of @xmath74 . the cs profiles across the core show emission from a bright component with a central velocity of @xmath75 and a fwhm width of @xmath76 . strong high velocity wing emission is detected toward the central region in all observed species ( see fig . [ fig - specpeakg329 ] ) . the blueshifted and redshifted emissions extend up to flow velocities of @xmath77 and @xmath78 in the co(1@xmath10 ) line , @xmath79 and @xmath64 in the cs(3@xmath12 ) line , and @xmath80 and 17.4 in the sio(3@xmath12 ) line . the cs(@xmath32 ) emission from the ambient gas arises from a structure that is slightly elongated in the se direction , with deconvolved fwhm major and minor axis of 44 and 39 , respectively . the peak of the cs emission is coincident with a methanol maser ( walsh et al . the spatial distribution of the silicon monoxide and methanol emission show notable differences with respect to that of the emission in carbon monosulfide . the morphology of the sio and ch@xmath2oh emission shows a double peaked structure , with the main peak being displaced from the peak of the cs emission by @xmath42 60 toward the southeast ( see fig . [ fig - mapsg329 ] ) . the secondary sio peak coincides with the peak of the cs emission . the double peaked morphology is also clearly seen in the map of the dust continuum emission . radio continuum observations towards g329.337 + 0.147 show the presence of a single source , with an angular diameter of @xmath81 " , exhibiting an irregular multi - peaked morphology ( walsh et al . 1998 ; urquhart et al . 2007 ) . if excited by a single star , the rate of uv photons required to ionize this region implies the presence of an o6.5 zams star . _ g330.949 - 0.174_.@xmath55 the g330.949 - 0.174 massive core is associated with the iras source 16060 - 5146 , which has a total luminosity of @xmath82 . the cs(2@xmath11 ) line exhibit an asymmetric profile , with a peak at the velocity of @xmath83 and a shoulder toward redshifted velocities . the cs(3@xmath12 ) and cs(5@xmath14 ) profiles are double peaked , with peaks at @xmath84 and @xmath85 . the blue peak is stronger than the red peak by a factor of @xmath42 1.5 . the profiles of the c@xmath6s and sio emission are , on the other hand , more symmetric , with peak velocities lying between the two velocity peaks of the cs emission . this result indicates that the complex cs profiles are produced by optical depth effects rather than being due to two velocity components along the line of sight . the presence of a high velocity gas toward the central region is clearly indicated by the profiles of the cs and sio emission . in co lines the outflow emission is blended with emission from less dense molecular clouds in the line of sight . the emission in the cs(@xmath34 ) line extends up to a velocity of @xmath86 in the red side and up to a velocity of @xmath87 in the blue side . the spatial distribution of the cs(@xmath32 ) emission from the ambient gas is nearly circular , with fwhm deconvolved major and minor axis of 43 . the circular morphology is also seen in the 1.2-mm dust emission , which exhibits a single structure with fwhm deconvolved major and minor axis of 33 . the total flux density of the region at 1.2 mm is 47.2 jy . a power - law fit to the 1.2-mm intensity profile gives an index of @xmath88 ( see fig . [ fig - intcuts ] ) . the peak of the molecular and dust emissions are displaced about 30 east of the iras position , but are coincident with the position of a methanol maser and a compact radio continuum source ( walsh et al . 1997 , 1998 ) . the radio continuum observations of urquhart et al . ( 2007 ) towards g330.949 - 0.174 show the presence of two compact sources : a bright cometary - like component , with an angular diameter of @xmath423 `` , located near the center of the core and an unresolved , weak component , located @xmath4212 '' northeast of the former . if excited by single stars , the rate of uv photons needed to produce the cometary and unresolved regions of ionized gas implies the presence of o7 and b0 zams stars , respectively . [ fig - sed ] shows , for each of the 4 sources , the spectral energy distribution ( sed ) from 12 @xmath89 m to 1.2 mm . in this wavelength range the emission is mainly due to thermal dust emission . the flux densities at 12 , 25 , 60 and 100 correspond to iras data , and those at 8.3 , 12.1 , 14.7 and 21.3 correspond to _ midcourse space experiment ( msx ) _ data ( price et al . the iras and msx fluxes were obtained from the ipac database . we fitted the sed with modified blackbody functions of the form @xmath90\omega_s~ , $ ] where @xmath91 is the dust optical depth , @xmath92 is the planck function at the dust temperature @xmath93 , and @xmath94 is the solid angle subtended by the dust emitting region . the opacity was assumed to vary with frequency as @xmath95 , i.e. @xmath96 , where @xmath97 is the frequency at which the optical depth is unity . a single temperature model produced poor fits , underestimating the emission observed at wavelengths smaller than 25@xmath89 m , and therefore we used a model with two temperature components . since the regions are likely to present temperature gradients , this is indeed a coarse simplification . however , this model allows us to determine the average dust temperatures representative of each wavelength range . the parameters derived from the fits , namely dust temperature , opacity power - law index , wavelength at which the opacity is unity , and angular size ( assuming a gaussian flux distribution ) , for the colder dust component are given in fig . [ fig - sed ] . the parameters derived from the 1.2-mm observations are summarized in table [ tbl - deriveddust ] . the dust temperatures given in col . ( 2 ) correspond to the colder temperature determined from fits to the spectral energy distribution ( sed ) . the radius given in col . ( 3 ) were computed from the geometric mean of the deconvolved major and minor angular sizes obtained from single gaussian fits to the observed spatial structure . the masses of the cores , given in col.(4 ) , were derived using the expression , ( e.g. , chini , krgel , & wargau 1987 ) @xmath98 where @xmath99 is the observed flux density at 1.2 mm , @xmath100 is the mass absorption coefficient of dust with a value of 1 @xmath101 g@xmath41 ( ossenkopf & henning 1994 ) , @xmath102 is the dust - to - gas mass ratio ( assuming 10% he ) with a value of @xmath103 , and @xmath104 is the planck function at the dust temperature @xmath93 . this expression implicitly assumes that the observed emission at 1.2 mm corresponds to optically thin thermal dust emission and that the source is isothermal . the masses derived in this way range from @xmath105 to @xmath106 @xmath24 . cols . ( 5 ) and ( 6 ) give , respectively , the average molecular densities and average column densities derived from the masses and radius assuming that the cores have uniform densities . clearly , this is a rough simplification , since as discussed in 3.2 the massive and dense cores are likely to have steep density gradients . finally , the continuum optical depth at 1.2-mm is given in col . ( 7 ) . the optical depth in the cs(3@xmath12 ) line , @xmath107 , can be derived from the ratio of observed main beam brightness temperatures in the cs(3@xmath12 ) and c@xmath6s(3@xmath12 ) lines using the expression , @xmath108 } \over{[j_{32}^{\small c^{34}s } ( t_{ex})-j_{32}^{\small c^{34}s } ( t_{bg } ) ] } } { { t_{mb}(c^{34}s)}\over{t_{mb}(cs ) } } , ~ \label{eqn - optdepth}\ ] ] where the subscript 32 refers to the @xmath34 transition , @xmath109 is the excitation temperature of the transition , @xmath110 is the background temperature , @xmath111 and @xmath53 is the @xmath112 optical depth ratio , given by @xmath113 where @xmath114 is the [ c@xmath115s / c@xmath6s ] isotopic abundance ratio , and @xmath89 and @xmath116 are the permanent dipole moment and the rotational constant of the molecule , respectively . if the excitation temperature is known , the above expressions allow the optical depth to be determined . we avoid the usual approximation that @xmath117 , and solve equation ( [ eqn - optdepth ] ) using an interpolation procedure , assuming a [ c@xmath115s / c@xmath6s ] abundance ratio of 22.5 ( blake et al . 1994 ) . the optical depth in the co(@xmath66 ) line , @xmath118 , can be derived from the ratio of observed main beam brightness temperatures in the co(@xmath66 ) and @xmath7co(@xmath66 ) lines in a similar manner as described above ( see bourke et al . 1997 , for the appropriate expressions ) . we assumed a [ co/@xmath7co ] abundance ratio of 55 , corresponding to the average value across the galactic disk ( wannier et al . 1982 ) . further , the observations of the j=@xmath66 line emission in the three isotopic species of carbon monoxide ( co , @xmath7co and c@xmath8o ) allow to determine the [ @xmath7co]/[c@xmath8o ] isotopic abundance ratio . using a [ co]/[@xmath7co ] abundance ratio of 55 , we derive [ @xmath7co]/[c@xmath8o ] abundance ratios in the range 8 to 10 . the peak optical depths at the peak position of the massive cores in the cs(3@xmath12 ) , c@xmath6s(3@xmath12 ) , co(1@xmath10 ) and @xmath7co(1@xmath10 ) lines , are given in cols . 2 to 5 of table [ tbl - opacities ] . they were computed assuming @xmath109 = 40 k. we find that the emission is optically thick in the cs lines and optically thin in the c@xmath6s line for all cores . we note that the dependence of the optical depth with excitation temperature is weak . for instance , assuming @xmath109 = 100 k the derived optical depths are only 0.4% larger . the total column density @xmath119 of a linear , rigid rotor , molecule can be derived from the optical depth and excitation temperature of a rotational transition at frequency @xmath120 , using the expression ( e.g. garden et al . 1991 ) @xmath121 where @xmath122 is the rotational quantum number of the lower state . this expression assumes that all energy levels are populated according to local thermodynamic equilibrium ( lte ) at the temperature @xmath109 . in particular the total column density of the cs molecule is given , in terms of the opacity and excitation temperature of the @xmath34 transition , by @xmath123 where @xmath124 is measured in . 2 of table [ tbl - colden ] gives the derived cs column density through the center of the cores . the co column density through the center of the cores are given in col . 3 of table [ tbl - colden ] . they were determined in the same way as the cs column densities using the [ co,@xmath7co ] pair of observations . from the column density in cs , n(cs ) , we can estimate the mass of the cores using the expression , @xmath125 \mu_m~ m_h~ \pi r^2~~,\ ] ] where @xmath126 $ ] is the h@xmath18 to cs abundance ratio , @xmath127 is the mean molecular weight , and @xmath128 is the core radius . the radii , given in column 2 of table [ tbl - derivedpar ] , were computed from the geometric mean of the deconvolved cs(2@xmath11 ) angular sizes using the distances given in table [ tbl - obssources ] . the core masses computed from expression ( 6 ) , using @xmath126= 3\times10 ^ 7 $ ] ( see 4.3 ) , are given in col . 3 of table [ tbl - derivedpar ] . they range from @xmath129 to @xmath130 @xmath24 . these masses are in good agreement with those derived from the dust continuum observations . the rotational temperature , @xmath131 , and the total column density , @xmath132 , of methanol and silicon monoxide can be computed using a rotational diagram analysis ( see e.g. , linke , frerking & thaddeus 1978 ; blake et al . 1987 ) , with the assumptions of optically thin conditions and local thermodynamical equilibrium ( lte ) , which relates the integrated line intensity , rotational temperature , and column density via @xmath133 where @xmath89 , @xmath120 , and @xmath134 are the transition dipole moment , frequency , and line strength of the transition , respectively , @xmath135 is the velocity integrated main beam brightness , obtained directly from the observations , @xmath136 is the upper state energy , and @xmath137 is the rotational partition function . [ fig - trotmet ] shows rotational diagrams for the ch@xmath2oh emission observed at the peak position of the massive cores . the column densities and rotational temperatures , derived from a linear least squares fit to the data , are given in cols . 4 and 6 of table [ tbl - colden ] . the rotational temperatures of methanol range from 12.0 k to 22.6 k. these temperatures are significantly smaller than the temperatures derived from the dust observations , of @xmath42 40k , indicating that the methanol populations are sub - thermally excited ( bachiller et al . 1995 ; avery & chiao 1996 ) . from rotational diagram analysis of the sio emission observed at the peak position of the massive cores we derived column densities and rotational temperatures of both the ambient gas , given in cols . 5 and 7 of table [ tbl - colden ] , and flowing molecular gas , given in cols . 8 and 9 the rotational temperatures of the flowing gas are typically two times larger than those of the ambient gas . this is illustrated in fig . [ fig - trotsio ] , which presents a rotational diagram for the sio emission observed at the peak position of the g330.949 massive core . in this case the gas giving rise to the narrow emission has a rotational temperature of 11.2@xmath138 k and a column density of @xmath139 @xmath12 , whereas the outflowing gas producing the broad emission has a rotational temperature of 24.9@xmath140 k and a column density of @xmath141 @xmath12 . the chemical composition of a gas cloud is usually characterized by the fractional abundance of molecules relative to molecular hydrogen , the main constituent of the interstellar gas . the abundance of h@xmath18 is however not directly measured . our observations allow us , however , to directly determine abundances relative to co. we compute the abundance of species x relative to co , [ x / co ] ( where x = cs , sio , or ch@xmath2oh ) , as the ratio of the column density of species x and the co column density in the same velocity range . for sio and ch@xmath2oh we use the column densities derived from rotational analysis ( see x.x ) whereas for cs and co we use the column densities derived from the observations of their main and isotopic species ( see 4.3.1 ) . table [ tbl - abundances ] summarizes the abundances of cs , sio and ch@xmath2oh relative to co. to convert [ x / co ] to fractional abundances relative to h@xmath18 we will assume [ co / h@xmath18]=@xmath143 . * silicon monoxide*. we find that the [ sio / co ] abundance ratio in the ambient gas range from @xmath144 to @xmath145 . assuming that [ co / h@xmath18 ] is @xmath143 , then the [ sio / h@xmath18 ] abundance ratio in the ambient gas ranges between @xmath146 and @xmath147 . chemical models predict that both high temperature and shocks can produce enhancements in the gas phase sio abundance . the abundance of sio in the ambient gas is likely due to the gas - phase high temperature chemistry that takes place in the hot regions surrounding the recently formed massive stars . models of gas - phase chemistry of hot molecular cores that takes into account grain mantle evaporation by stellar and/or shock radiation ( mackay 1995 ) , predict [ sio / h@xmath18 ] abundances of @xmath146 for gas kinetic temperatures of 100 k and densities of @xmath148 @xmath14 . we also find that the [ sio / co ] abundance ratios in the outflowing gas are similar to those derived in the ambient gas . the abundance of sio in the high velocity gas is likely to be due to gas phase chemistry behind a shock ( iglesias & silk 1978 ; hartquist et al . 1980 ; mitchell & deveau 1983 ) . the enhancement of the sio abundance in the high velocity material is probably the result of grain destruction by shocks which releases silicon into the gas phase , allowing sio to form ( seab & shull 1983 ; schilke et al . 1997 ; caselli et al . 1997 ) . * methanol*. we find that the [ ch@xmath2oh / co ] abundance ratio in the ambient gas is similar for all observed cores , having a mean value of @xmath149 . assuming an [ h@xmath18/co ] of @xmath150 , it implies that the [ ch@xmath2oh / h@xmath18 ] abundance ratio in the ambient gas is typically @xmath20 . at temperatures smaller than 100 k , chemistry models predict that the production of ch@xmath2oh in the gas phase yield abundances of only 10@xmath151 relative to h@xmath18 ( lee et al . 1996 ) . the high abundance of methanol in the cores suggests that the radiation from their embedded luminous stars are producing a substantial evaporation of icy grain mantles . van der tak et al . ( 2000a ) found a [ ch@xmath2oh / h@xmath18 ] ratio of a few @xmath152 for hot cores and @xmath153 for the cores with the lower temperatures . the values obtained for the cores investigated here are in between the values found for dark clouds and hot cores , and may correspond to an average of the values appropriate for the cold envelope mass and hot core mass . * carbon monosulfide*. we find that the [ cs / co ] abundance ratio in the ambient gas is similar for all observed cores , having a mean value of @xmath154 . assuming an [ h@xmath18/co ] of @xmath150 , it implies that the [ h@xmath18/cs ] abundance ratio in the ambient gas is typically @xmath155 . this value is typical of molecular cores associated with uc hii regions and hot cores ( van der tak et al . 2000b ) . in the previous analysis we have implicitly assumed that the density and temperature are uniform within the massive cores . this is obviously a simplification , since the physical conditions are likely to vary with distance from the center of the core . in fact , the intensity profiles of the dust continuum emission indicate that the cores are highly centrally condensed . assuming that cores have density and temperature radial distributions following power laws , then for optically thin dust emission the intensity index @xmath156 is related to the density index @xmath157 ( @xmath158 ) and the temperature index @xmath159 ( @xmath160 ) , by the expression @xmath161 ( adams 1991 ; motte & andr 2001 ) , where q is a temperature and frequency correction factor with a value of @xmath162 at 1.2-mm and 30 k ( beuther et al . 2002 ) and @xmath163 is a correction factor to take into account the finite size of the cores . the massive and dense cores investigated here are heated by a cluster of luminous sources located at their central position ( chavarra et al . 2009 ; paper ii ) . for this type of objects van der tak et al . ( 2000b ) found that the temperature decreases with distance following a power - law with an index of 0.4 . adopting this value for @xmath159 and a value of 0.1 for @xmath163 , we then infer , using the above expression , that massive and dense cores have density distributions with power - law indices in the range @xmath164 . several dust continuum studies have already shown the presence of density gradients within massive and dense cores ( e.g. , van der tak et al . 2000b , mueller et al . 2002 , beuther et al . 2002 , williams et al . in particular , beuther et al . ( 2002 ) observed a large sample of high - mass protostellar candidates and found that the steeper indices are associated with the more luminous and more massive objects . the steep indices derived for the norma cores are characteristics of the subgroup of _ strong molecular sources _ within beuther s sample , which have a mean density index of 1.9 . the steep profiles are thought to indicate objects in the collapse and accretion phase . given the large mass of the massive and dense cores investigated here , of typically @xmath105 m@xmath10 , their overall collapse process is likely to produce a protostar cluster . the questions of how an individual massive star forms and how the cluster forms are therefore closely related . the derived properties of massive and dense cores are then important boundary conditions to be taken into account for models and simulations . for dynamical considerations , the relevant mass of the cores is their total mass , including gas and stars . we have already shown that the mass of gas in the cores determined from two independent methods , one based on dust observations and the other on molecular line observations , are in very good agreement . the determination of the total mass of the cores is not straightforward and is usually estimated assuming that the cores are in virial equilibrium . masses computed under this assumption , using the observed velocity dispersion and size of the cs(2@xmath11 ) emission , are given in column 4 of table [ tbl - derivedpar ] . the virial masses are similar , within the errors , to the gas masses , implying that the bulk of the core mass is in the form of molecular gas . we note that the virial masses are in average a factor of 1.5 smaller than the dust or molecular masses which might also indicate that the cores do not have enough support and may be contracting . in paper ii we show that the massive cores investigated here are associated with clusters of embedded stars which exhibit clear mass segregation . the high - mass stars are found concentrated near the center of the core while intermediate mass stars are spread over much larger distances from the core center . garay et al . ( 2006 ) have already reported that massive stars are typically located near the center of the massive and dense cores . the question arises as to whether the massive stars are formed in situ or if their spatial location is a result of dynamical effects . the characteristics of the cores derived here , namely their high densities , steep density profiles and sizes , together with the relatively low fraction of the total mass in the form of stars , suggest that dynamical friction ( also called gravitational drag ) produced by the gaseous background onto the stars may play an important role in the orbital evolution of stars ( ostriker 1999 , sanchez - salcedo & brandenburg 2001 , kim & kim 2007 ) . the role of gravitational drag on a gaseous background in the evolution and migration of stars within dense gaseous star forming clouds has been investigated using numerical smooth particle hydrodynamics simulations ( escala et al . 2003 , 2004 ) . simulations have been carried out with initial conditions similar to those of the dense cores investigated here , namely steep cloud density profile ( @xmath165 ) and perturber having masses of 1 % of the total gas mass ( @xmath166 @xmath24 ) . for a typical cloud mass of @xmath167 @xmath24 and initial distance of the stars to the center of the cloud of 0.4 pc , the migration timescale for high - mass stars is @xmath168 yr . this value is smaller than the estimated ages of the young clusters within the cores of @xmath42 1 myr ( see paper ii ) . since the migration timescales for dynamical friction are proportional to the inverse of the perturber s mass ( binney & tremaine 1987 ) , intermediate mass stars have migration timescales typically a factor @xmath42 10 longer than those of high mass stars and greater than the estimated age of the clusters . therefore , gravitational drag is considerably less effective for intermediate mass stars than for high mass stars . this suggest that in the young embedded stellar clusters only the massive stars have been able to efficiently migrate towards the center due to gravitational drag . we conclude that dynamical interactions of cluster members with the gaseous component of massive and dense cores can explain the observed mass segregation , indicating that the gas play an important role in the dynamic of young cluster members and originating the observed mass segregation . we can not rule out , however , the hypothesis that the cradles of massive stars are formed by direct accretion at the center of centrally condensed massive and dense cores ( mckee & tan 2003 ) . in their turbulent and pressurized dense core accretion model , the collapse of a massive and dense core is likely to produce the birth of a stellar cluster , with most of the mass going into relatively low - mass stars . the high - mass stars are formed preferentially at the center of the core , where the pressure is the highest , and in short time scales of @xmath169 yrs ( osorio et al . 1999 ; mckee & tan 2002 ) . we undertook molecular line observations in several species and transitions and dust continuum observations , both made using sest , towards four young high - mass star forming regions associated with highly luminous iras point sources with colors of uc hii regions and cs(2@xmath11 ) emission . these are thought to be massive star forming regions in early stages of evolution . the objectives were to determine the characteristics and physical properties of the dust clouds in which high - mass stars form . our main results and conclusions are summarized as follows . we find that the luminous star forming regions in norma are associated with molecular gas and dust cores with radii of typically 0.5 pc , masses of @xmath9 m@xmath10 , molecular hydrogen densities of typically @xmath13 @xmath14 , column densities of @xmath11 @xmath12 , and dust temperatures of @xmath15 k. all the derived physical parameters of the norma cores , except the last , are similar to those of cores harboring uc hii regions ( hunter et al . 2000 ; fandez et al . 2004 ) . for a sample of about 150 iras point sources with colours of uc hii regions and average luminosity of @xmath170 @xmath28 , fandez et al . ( 2004 ) derived an average mass of @xmath105 m@xmath10 , an average size of 0.4 pc , and an average density of @xmath171 @xmath14 . the dust temperature of the norma cores are however larger than of the fandez et al . cores , for which they find an average temperature of 32 k. the larger values derived for the norma cores are consistent with them being centrally heated by more luminous objects ( average luminosity of @xmath172 @xmath28 ) . we find that the observed radial intensity profiles of the dust continuum emission from the norma cores are well fitted with single power - law profiles , with intensity indices in the range 1.5 to 1.9 . this in turn indicates that the cores are highly centrally condensed , having radial density profiles with power - law density indices in the range @xmath173 . these steep profiles are thought to indicate objects in the collapse and accretion phase of evolution . we find that under the physical conditions of the norma cores , high - mass stars can migrate towards the central region due to gravitational drag in time scales of @xmath174 yr , smaller than the estimated ages of the clusters , providing an explanation for the observed stellar mass segregation within the cores ( paper ii ) . lccccc g324.201 + 0.119 & 15290 - 5546 & @xmath175 & @xmath176 & 6.0 & @xmath177 + g328.307 + 0.423 & 15502 - 5302 & 15 54 06.3 & @xmath5553 11 38 & 5.6 & @xmath178 + g329.337 + 0.147 & 15567 - 5236 & 16 00 33.4 & @xmath5552 44 45 & 7.0 & @xmath179 + g330.949 - 0.174 & 16060 - 5146 & 16 09 49.1 & @xmath5551 54 45 & 5.4 & @xmath180 + lcrccc cs & & 97980.968 & 52 & 0.73 & 0.132 + & & 146969.049 & 34 & 0.66 & 0.085 + & & 244935.606 & 22 & 0.45 & 0.053 + c@xmath6s & & 144617.147 & 34 & 0.66 & 0.087 + sio & & 86846.998 & 57 & 0.75 & 0.144 + & & 130268.702 & 40 & 0.68 & 0.098 + ch@xmath2oh & a@xmath36 & 96741.42 & 52 & 0.73 & 0.129 + & e & 145097.470 & 34 & 0.66 & 0.086 + co & & 115271.204 & 45 & 0.70 & 0.111 + @xmath7co & & 110201.353 & 47 & 0.71 & 0.116 + c@xmath8o & & 109782.160 & 47 & 0.71 & 0.116 + lccccccccccc cs & 3 & 0.12 & & 3 & 0.14 & & 6 & 0.064 & & 3 & 0.094 + cs & 4 & 0.065 & & 4 & 0.071 & & 4 & 0.065 & & 4 & 0.057 + cs & 5 & 0.47 & & 3 & 0.11 & & 6 & 0.41 & & 5 & 0.44 + c@xmath6s & 4 & 0.074 & & 4 & 0.068 & & 4 & 0.068 & & 4 & 0.049 + sio & 20 & 0.018 & & 12 & 0.028 & & 12 & 0.027 & & 8 & 0.026 + sio & 12 & 0.033 & & 22 & 0.016 & & 22 & 0.021 & & 4 & 0.035 + ch@xmath2oh & 8 & 0.037 & & & & & 4 & 0.054 & & 4 & 0.038 + ch@xmath2oh & 8 & 0.043 & & 8 & 0.045 & & 8 & 0.044 & & 4 & 0.049 + co & 1 & 0.276 & & 1 & 0.276 & & 1 & 0.254 & & 1 & 0.254 + @xmath7co & 3 & 0.072 & & 3 & 0.071 & & 3 & 0.068 & & 3 & 0.076 + c@xmath8o & 6 & 0.057 & & 6 & 0.055 & & 5 & 0.060 & & 5 & 0.063 + lccccccc cs & @xmath1810.12 & @xmath182.04 & @xmath1830.3 & & @xmath184.02 & @xmath185 & @xmath186 + cs & @xmath1870.07 & @xmath188.01 & @xmath1830.1 & & @xmath189.02 & @xmath190 & @xmath191 + cs & @xmath1920.47 & @xmath193.05 & @xmath1940.1 & & @xmath1950.16 & @xmath196 & @xmath197 + c@xmath6s & @xmath1980.074 & @xmath199.05 & @xmath2000.1 & & @xmath201 & @xmath202 & @xmath203 + sio & @xmath2040.018 & @xmath2050.3 & @xmath2061.1 & & @xmath207 & @xmath208 & @xmath209 + sio & & & & & @xmath210 & @xmath211 & @xmath212 + ch@xmath2oha@xmath36 & @xmath2130.037 & @xmath214.05 & @xmath2150.2 & & @xmath216 & @xmath217 & @xmath218 + ch@xmath2oha@xmath36 & @xmath2190.043 & @xmath220.03 & @xmath2210.1 & & @xmath222 & @xmath223 & @xmath191 + cs & & @xmath224 & & & @xmath225 & @xmath226 & @xmath227 + cs & @xmath228 & @xmath229.02 & @xmath2300.1 & & @xmath2310.02 & @xmath232 & @xmath233 + cs & @xmath2340.11 & @xmath235.04 & @xmath2000.2 & & & & + c@xmath6s & @xmath236 & @xmath237.09 & @xmath1940.2 & & @xmath2380.02 & @xmath239 & @xmath240 + sio & @xmath2410.028 & @xmath242.13 & @xmath2430.4 & & & @xmath244 & + sio & @xmath245 & @xmath246.10 & @xmath2060.3 & & & & + ch@xmath2oha@xmath36 & @xmath2470.045 & @xmath248.05 & @xmath2490.1 & & @xmath250 & @xmath251 & @xmath252 + cs & @xmath2530.06 & @xmath254.02 & @xmath2550.1 & & @xmath2560.01 & @xmath257 & @xmath258 + cs & @xmath2590.07 & @xmath260.01 & @xmath2610.1 & & @xmath2620.02 & @xmath263 & @xmath264 + cs & @xmath2650.41 & @xmath266.08 & @xmath2670.3 & & @xmath2680.14 & @xmath269 & @xmath270 + c@xmath6s & @xmath2710.07 & @xmath272.03 & @xmath2000.1 & & @xmath2730.02 & @xmath274 & @xmath275 + sio & @xmath2760.027 & @xmath277.10 & @xmath2780.2 & & @xmath2790.007 & @xmath280 & @xmath281 + sio & @xmath282 & @xmath283.13 & @xmath2780.3 & & & & + ch@xmath2oha@xmath36 & @xmath2840.054 & @xmath285.04 & @xmath2860.1 & & @xmath2870.014 & @xmath288 & @xmath289 + ch@xmath2oha@xmath36 & @xmath2900.044 & @xmath291.02 & @xmath2300.1 & & @xmath292 & @xmath293 & @xmath264 + cs & & @xmath224 & & & @xmath294 & @xmath295 & @xmath296 + cs & & @xmath224 & & & @xmath297 & @xmath298 & @xmath296 + cs & & @xmath224 & & & @xmath299 & @xmath300 & @xmath301 + c@xmath6s & @xmath3020.049 & @xmath303.06 & @xmath2300.5 & & @xmath304 & @xmath305 & @xmath306 + sio & @xmath3070.026 & @xmath308.03 & @xmath3090.2 & & @xmath310 & @xmath311 & @xmath312 + sio & @xmath313 & @xmath314.01 & @xmath2060.2 & & & & + ch@xmath2oha@xmath36 & @xmath315 & @xmath316.03 & @xmath2060.1 & & @xmath317 & @xmath311 & @xmath318 + ch@xmath2oha@xmath36 & @xmath319 & @xmath320.02 & @xmath3210.1 & & @xmath322 & @xmath323 & @xmath324 + llccc g324.201 + 0.119 ... & cs & @xmath55101.5 & @xmath5577.0 & 24.5 + & sio & @xmath55101.0 & @xmath5571.6 & 29.4 + & sio & @xmath55101.4 & @xmath5572.0 & 29.4 + g328.307 + 0.423 ... & cs & @xmath55101.0 & @xmath5580.6 & 20.4 + & sio & @xmath55100.0 & @xmath5582.6 & 17.4 + g329.337 + 0.147 ... & cs & @xmath55122.5 & @xmath5590.9 & 31.6 + & sio & @xmath55114.0 & @xmath5592.2 & 21.8 + & sio & @xmath325 & @xmath5590.3 & 36.6 + g330.949 - 0.174 & cs & @xmath55106.0 & @xmath5575.7 & 30.3 + & sio & @xmath55107.0 & @xmath5570.4 & 36.6 + & sio & @xmath55103.4 & @xmath5575.2 & 28.2 + lcccccc g324.201 + 0.119 & 15 32 52.91 & -55 56 10.8 & & 5.28 & 17.3 & @xmath326 + g328.307 + 0.423 & 15 54 06.33 & -53 11 43.1 & & 6.35 & 24.1 & @xmath327 + g329.337 + 0.147a & 16 00 33.10 & -52 44 46.6 & & 4.35 & 11.7 & @xmath328 + g329.337 + 0.147b & 16 00 38.68 & -52 45 31.7 & & 1.45 & 4.58 & @xmath329 + g330.949 - 0.174 & 16 09 52.41 & -51 54 58.5 & & 21.75 & 47.2 & @xmath330 + lcccccc g324.201 + 0.119 & 37 . & 0.45 & @xmath331 & @xmath332 & @xmath333 & 0.016 + g328.307 + 0.423 & 42 . & 0.46 & @xmath334 & @xmath332 & @xmath335 & 0.017 + g329.337 + 0.147a & 40 . & 0.53 & @xmath336 & @xmath337 & @xmath338 & 0.010 + g329.337 + 0.147b & 40 . & 0.76 & @xmath339 & @xmath340 & @xmath341 & 0.002 + g330.949 - 0.174 & 37 . & 0.44 & @xmath342 & @xmath343 & @xmath344 & 0.038 + lccccccccc g324.201 + 0.119 & @xmath347 & @xmath348 & @xmath349 & & 12.0 & & & @xmath350 & 31.0 + g328.307 + 0.423 & @xmath351 & @xmath352 & @xmath353 & @xmath354 & 10.3 & 6.3 & & @xmath355 & 24.3 + g329.337 + 0.147 & @xmath356 & @xmath352 & @xmath357 & @xmath358 & 17.0 & 16.7 & & @xmath359 & 45.6 + g330.949 - 0.174 & @xmath360 & @xmath361 & @xmath362 & @xmath139 & 22.6 & 11.2 & & @xmath141 & 24.9 + lcccc g324.201 + 0.119 & @xmath369 & & @xmath370 + g328.307 + 0.423 & @xmath371 & @xmath372 & @xmath373 + g329.337 + 0.147 & @xmath374 & @xmath375 & @xmath376 + g330.949 - 0.174 & @xmath374 & @xmath377 & @xmath378 +
we report molecular line and dust continuum observations , made with the sest telescope , towards four young high - mass star forming regions associated with highly luminous ( @xmath0 ) iras sources ( 15290 - 5546 , 15502 - 5302 , 15567 - 5236 and 16060 - 5146 ) . molecular emission was mapped in three lines of cs ( j=2@xmath11 , 3@xmath12 and 5@xmath14 ) , two lines of sio ( j=2@xmath11 and 3@xmath12 ) , two rotational transitions of ch@xmath2oh ( j@xmath3=3@xmath4 and @xmath5 ) , and in the c@xmath6s(j=3@xmath12 ) line . in addition , single spectra at the peak position were taken in the co(j=1@xmath10 ) , @xmath7co(j=1@xmath10 ) and c@xmath8o(j=1@xmath10 ) lines . we find that the luminous star forming regions are associated with molecular gas and dust structures with radii of typically 0.5 pc , masses of @xmath9 m@xmath10 , column densities of @xmath11 @xmath12 , molecular hydrogen densities of typically @xmath13 @xmath14 and dust temperatures of @xmath15 k. the 1.2 mm dust continuum observations further indicate that the cores are centrally condensed , having radial density profiles with power - law indices in the range @xmath16 . we find that under these conditions dynamical friction by the gas plays an important role in the migration of high - mass stars towards the central core region , providing an explanation for the observed stellar mass segregation within the cores . the cs profiles show two distinct emission components : a bright component , with line widths of typically 5 ( fwhm ) , and a weaker and wider velocity component , which extends up to typically @xmath1713 from the ambient cloud velocity . the sio profiles also show emission from both components , but the intensity of the pedestal feature relative to that of the bright component is stronger than for cs . the narrow sio component is likely to trace warm ambient gas close to the recently formed massive stars , whereas the high velocity emission indicates mass outflows produced by either the expansion of the hii regions , stellar winds , and/or collimated outflows . we find that the abundances of cs , ch@xmath2oh and sio , relative to h@xmath18 , in the warm ambient gas of the massive cores are typically @xmath19 , @xmath20 , and @xmath21 , respectively .
electrical capacitance tomography ( ect ) is an attractive method for imaging multiphase flows , as it is noninvasive , fast , safe and low cost @xcite . a typical ect system consists of three main parts : a multi - electrode sensor , an acquisition hardware and a computer for hardware control and image processing . specifically , the multi - electrode hardware in ect typically has @xmath1 electrodes surrounding the wall of the process vessel . the number of independent capacitance measurements in such a configuration is @xmath2 due to the independent number of sensor pairs with @xmath1 electrodes . the final objective is to recover the cross section or even 3d images of the permittivity distribution by using these measurements to solve an inverse problem . however , the inverse problem is underdetermined , since the number of measurements is far fewer than the number of pixels in the reconstructed image @xcite . furthermore , the governing equations to be considered are non - linear @xcite . various reconstruction algorithms have been developed to cope with these difficulties . direct , or single step , algorithms include the classic linear back projection ( lbp ) , approaches based on singular value decomposition ( svd ) , and tikhonov regularization @xcite . indirect , or iterative , algorithms include landweber iterations ( li ) @xcite and iterative tikhonov methods @xcite . these algorithms all inherently assume a smooth permittivity distribution within the sample , but for many systems this assumption is poor . in recent years , concepts from compressive sensing ( cs ) theory@xcite have been shown to permit the reconstruction of sharp changes in permittivity @xcite . cs can not be applied strictly to ect image reconstruction due to the non - linear nature and because the sensitivity matrix does not satisfy the restricted isometry property ( rip ) . however , several researchers have extended the ideas of cs to non - linear systems @xcite . in this paper we adapt ideas from cs to propose a comprehensive reweighted total variation iterative shrinkage thresholding ( tv - ist ) algorithm for non - linear ect image reconstruction . after explaining the ect physical model , we modify the conventional tv - ist to develop the tv - ist for ect and its fast version with auxiliaries . we introduce adaptive weights @xcite to approximate the @xmath3-norm solution closely . finally , we combine this reweighting approach with a method to minimize the non - linearity of the reconstruction by updating the sensitivity matrices within tv - ist . the algorithm has been examined using simulated measurements of phantoms to show its superiority compared with other existing algorithms . in ect , the permittivity distribution inside a pipe or vessel of interest , corresponding to the material distribution , is calculated from measured capacitances between all pairs of sensors located around the pipe s periphery . the total electric flux over all the electrodes surfaces is equal to zero , hence the potential and permittivity are obtained from a form of poisson s equation : @xmath4 = 0,\ ] ] where @xmath5 is the spatial permittivity distribution , and @xmath6 the electric potential distribution . the boundary conditions are @xmath7 for the excited electrode and @xmath8 for other electrodes . for the two - dimensional case @xmath9 , the relationship between the capacitance and permittivity distribution can be expressed by the following equation : @xmath10 where @xmath11 is the total charge , @xmath12 denotes the closed line of the electrical field , @xmath13 is the permittivity distribution in the sensing field , and @xmath14 is the potential difference between two electrodes forming the capacitance . in ( [ eq : c ] ) @xmath15 is also a function of @xmath16 . therefore the capacitance between electrode combinations can be considered as a function of permittivity distribution @xmath13 : @xmath17 where @xmath18 is a non - linear function , and elements of @xmath19 are the non - redundant capacitance values obtained from the electrode pairs @xmath20 $ ] . if we descretise the permittivity and express it as a vector , we can estimate the changes in the capacitance values from a taylor s series expansion : @xmath21 where @xmath22 is the sensitivity of the capacitance with respect to changes in the permittivity distribution , and @xmath23 represents the higher order terms of @xmath24 . because @xmath25 is usually small , the high order terms are often neglected . ( [ eq : trianglec ] ) can be linearized in a matrix form : @xmath26 where @xmath27 , @xmath28 is a jacobian / sensitivity matrix denoting the sensitivity distribution for each electrode pair , and @xmath29 . as a result , the non - linear forward problem has been reformulated to a linear approximation . generally in ect , ( [ eq : linearc ] ) is written in a normalized form @xmath30 where @xmath31 is the normalized capacitance vector , @xmath32 is the jacobian matrix of the normalized capacitance with respect to the normalized permittivities , which gives a sensitivity map for each electrode pair , and @xmath33 is the normalized permittivity vector , which can be visualized by the colour density of the image pixels . the conventional optimization problem of ect becomes @xmath34 because there are @xmath1 electrode pairs , @xmath35 should be @xmath36 . the objective of the reconstruction algorithm of ect is to recover @xmath13 from measured capacitance vector @xmath19 . while in the discrete linear model , it is to estimate @xmath37 given @xmath38 , and @xmath39 is seen as a constant matrix determined in advance for simplicity . there are several difficulties with the reconstruction problem . firstly , ( [ eq : lambda ] ) is under - determined so the solution is not unique , and it is very sensitive to disturbances of @xmath40 . secondly , owing to the non - linearity in eq . ( [ eq : c ] ) , @xmath39 is not constant but varies for different permittivity distributions . in this paper , we propose a non - linear reweighted total variation image reconstruction algorithm to overcome these difficulties . to recover the permittivity distribution image @xmath41 , many reconstruction algorithms for ect have been developed . generally the reconstruction algorithms can be categorized in two groups : direct algorithms and iterative algorithms . among them , landwater iteration and steepest descent method ( lwsdm ) is considered as one of the best algorithms with good efficiency . it minimizes the cost function @xmath42 , e.g. to minimize @xmath43 the gradient of @xmath44 is @xmath45 we iteratively update the image @xmath37 in the direction that @xmath44 decreases most quickly . therefore the new image will be @xmath46 where @xmath47 is a positive value determining the step size . in fact , lwsdm can be derived from the iterative shrinkage thresholding algorithm ( ista ) as a special case with ect constraints . here we introduce and explain the general model of ista on which our algorithm also is based . the ista is to solve a class of optimization problems with convex differentiable cost functions and convex regularization . [ ther : ista ] @xcite consider the general formulation : @xmath48 and the following assumptions are satisfied : * @xmath49 : a smooth convex function which is also continuously differentiable with lipschitz continuous gradient @xmath50 : @xmath51 where @xmath52 is the lipschitz constant of @xmath53 . * @xmath54 : a continuous convex function mapping @xmath55 * problem @xmath56 is solvable . then basic ista converges to its true solution by running iteration @xmath57 , where its iteration : @xmath58 for example , lwsdm is actually a special instance of problem @xmath59 by substituting @xmath60 and @xmath61 as a smooth quadratic minimization problem with the lipschitz constant of the gradient @xmath53 being @xmath62 . then according to ( [ eq : p_l ] ) we have @xmath63 which is equivalent to the lwsdm , where @xmath64 . theorem [ ther : ista ] provides the theoretical convergence for algorithms . total variation ( tv ) norm of the image has been used widely to penalize the cost function @xcite . it also has been verified that the tv norm can be utilized to address the under - determined image reconstruction and reproduce ect @xcite or other tomography images @xcite with sharp transitions in intensity . therefore , unlike conventional techniques for iterative reconstruction , we assume that there are sharp changes in intensity that can be sparsely represented by their spatial gradients . in this case the cost function is to minimize the least squares error and the sparsity of intensity changes : @xmath65 where @xmath66 is the discrete isotropic tv of the two dimensional @xmath67 defined by @xcite : @xmath68 with the boundary conditions @xmath69 and @xmath70 . ( [ eq : tv1 ] ) belongs to linear inverse problems with nonquadratic regularizers . nonquadratic regularizers include wavelet representations @xcite , sparse regression @xcite and total variation , etc . these problems can be solved by a signal processing technique called compressive sensing ( cs ) in the literature @xcite . ista is very convenient to solve cs problems with @xmath0 norm regularization . the non - linear shrinkage operation , or so called soft thresholding , is @xmath71 for instance , the ista and its derivative versions along with the shrinkage operation have been verified to solve wavelet - based reconstruction for magnetic resonance imaging ( mri ) efficiently @xcite . in the next section we will explain how to implement ista to ect image reconstruction using tv regularization and prove its effectiveness . in this section we will present an iterative reconstruction technique for ect . as in ist , the iterative soft thresholding is applied to penalize the total variation of the ect image . some of the contents have been introduced in our conference paper @xcite . here we provide the full theoretical analysis of this algorithm and its convergence rate . in the ect model , the permittivity distribution inside the pipe can usually be formulated as the 2d image / matrix @xmath72 . set @xmath37 is @xmath72 expressed as a column vector . @xmath73 denotes the pixel of the position @xmath74 in the imaging region . its magnitude is proportional to the permittivity difference @xmath25 and is @xmath75 outside of the imaging region . we use @xmath76 to represent the gradients of the image respectively which correspond to the horizontal and vertical finite differences . in detail , the gradient transforms are used to calculate the gradients : @xmath77 where @xmath78 are transform matrices . each element of @xmath79 corresponds to the same @xmath80th element in @xmath37 . likewise , given @xmath81 , an inverse transform can be carried out by solving a least squares ( ls ) problem of @xmath82 using linear algebra we can obtain the standard ls solution : @xmath83 @xmath84 approximates the laplacian operator for the image . it is an approximation of the fourier transform version @xcite but only considers the pixels within the imaging region . ( a ) ( b ) to consider an isotropic form of tv , a single vector @xmath85 is used to represent the gradient magnitude , where elements of @xmath85 are given by @xmath86 . then ( [ eq : tv1 ] ) can be reformulated to @xmath87 equation ( [ eq : tvist1 ] ) is different from the conventional @xmath0 minimization problem . however , we can still use the iterative update idea to pursue the solution . instead of updating @xmath88 in iterations , here we calculate @xmath89 and @xmath90 by updating it to their steepest descent . following the second step in ( [ eq : x_k+1_f1 ] ) , the residuals are calculated and projected to the @xmath89 and @xmath90 directions , respectively @xmath91 where @xmath92 due to the symmetry of matrix @xmath84 . as a result , according to the ist algorithm the gradients @xmath81 can be updated : @xmath93 where @xmath94 due to the requirements of lipschitz continuous , and @xmath95 are the iterative gradient solution that can be derived from @xmath96 if we only consider the least squares error . the next step is to optimize the @xmath85 with @xmath97 where @xmath98 is equal to @xmath99 multiplied by some constant . similar to ( [ eq : shrinkage ] ) , a shrinkage operator can be used . while here the difference to the conventional shrinkage is that rather than set a soft thresholding on @xmath100 directly , we decrease the magnitudes of @xmath100 by making it proportional to the magnitudes of @xmath101 after a soft thresholding on @xmath101 to reduce the total variation . the magnitude vector @xmath101 can be calculated element - wise by @xmath102 so for the ect tv - ist , the soft thresholding process can be carried out @xmath103 where @xmath104 is the shrinkage operator defined in ( [ eq : shrinkage ] ) , and this equation is calculated element - wise . by using this new soft thresholding process we are able to eliminate small variation and meanwhile reduce the large variation in @xmath100 directions . finally , the new reconstructed image can be updated by ( [ eq : ls ] ) , and the new gradients of the image are updated by multiplying the transform matrices @xmath105 which returns to the beginning of the section and completes one iteration in the algorithm . algorithm 1 sums up the total variation - iterative soft thresholding algorithm . + -0.1 cm [ cols="<",options="header " , ] similar to the weighted @xmath0 minimization , it is natural to incorporate the reweighting technique in the total variation constraints . then the tv optimization problem for ect is transformed to @xmath106 in the following steps , ( [ eq : g12 ] ) and ( [ eq : hat_g12 ] ) remain the same for reweighted tv - ist in each iteration . the difference occurs in the soft thresholding step . because we want to pursue the minimal @xmath0 norm of the weighted tv , the thresholding process needs to be changed adaptively : @xmath107 where this operation should be done element - wise . it is a weighted version of the 2d soft threshold update . by using the weighted @xmath108 for thresholding , the @xmath0 norm behaves more like the @xmath3 norm . all the non - zero entries of @xmath85 above the threshold will be calculated more equally in the weighted norm , similar to the definition of the @xmath3 norm which see all non - zero entries contribute equally . finally , the weights @xmath109 vary depending on @xmath85 . specifically , the weights can be updated element by element as the second step of algorithm 3 : @xmath110 where @xmath111 is a parameter that is set slightly smaller than the expected nonzero magnitudes of @xmath85 . the value for @xmath112 can be determined from experience , but in general should be small . moreover , in practice since the reconstructed result @xmath37 and its gradient @xmath85 evolves gradually after each iteration , we insert the weights updating step into the tv - ist algorithm every @xmath113 iterations . hence the weights can be updated every @xmath113 iterations and we can use the parameter @xmath113 to make a tradeoff between calculation speed and weights update . meanwhile , the auxiliary vectors @xmath114 can also be adopted to accelerate the convergence and the reweighted tv - ist becomes reweighted tv - fist . the faster implementation is always used here and henceforth the reweighted tv - fist will be referred to as reweighted tv - ist for simplicity . before demonstrating the reweighted tv - ist algorithm , we explain two techniques that can be used to compensate for the non - linearity in our algorithm . in the ect model , the non - linear effects can be resolved in two aspects . the first one is the approximation of the linear model in ( [ eq : linearc ] ) . the second non - linear effect lies on the accuracy of the measurements of the sensitivity matrix @xmath39 . two techniques are introduced here to address the non - linear effect , respectively @xcite . however , only the first of these is implemented in the simulations as it is computationally simpler . from ( [ eq : trianglec ] ) to ( [ eq : linearc ] ) the quadratic and higher order terms have been neglected to reduce the ect model to a linear model . however this approximation causes errors due to the higher order terms that have been neglected . to offset this bias , a fitting curve has been proposed @xcite : @xmath115 where @xmath116 denotes the measurement at electrode @xmath80 when electrode @xmath117 is under the voltage @xmath118 while other electrodes are grounded , @xmath119 . this setting is to make sure that @xmath120 when @xmath25 tends to infinity and the slope is @xmath121 at @xmath122 . this approach may reduce the non - linear error by around @xmath123 @xcite . the non - linear sensitivity matrix @xmath124 can be defined as @xmath125 combining ( [ eq : m ] ) and ( [ eq : s ] ) we have @xmath126 which adjusts @xmath39 to a non - linear sensitivity matrix @xmath124 , where the permittivity of the area of interest is assumed to vary from @xmath16 to @xmath25 , whose values can be determined before the experiments . an adaptive sensitivity matrix model has also been proposed for use with landweber iterations @xcite . we introduce this feedback iteration to our reweighted tv - ist algorithm . the sensitivity map for an electrode pair can be calculated from the potential distribution @xmath127 where @xmath128 are the gradient values of the potential with electrode @xmath80 in the row and column vectors , respectively ; and the potential value of each pixel can be computed after iterations using the finite difference method ( fdm ) depending on the potential values of the surrounding four pixels : @xmath129 where @xmath130 are the location indexes . in summary , the reweighted tv - ist algorithm for non - linear ect uses the tv penalties in the cost function to pursue the optimal solution iteratively and meanwhile makes use of the superiority of updated reweighted norms and the auxiliary method s fast convergence . it is distinct from conventional fista for total variation minimization and designed to be suitable for ect reconstruction specifically . compared to the conventional linear tv - ist , our non - linear reweighted tv - ist has two differences . firstly the reweighted term @xmath109 has been adopted in the cost function to pursue a more sparse total variation in the optimization process . this should produce clearer edges between areas with different permittivities . secondly , two methods are introduced to reduce the non - linear effects . the methods add an extra step in the algorithm ( step 2 in algorithm 4 ) to update the sensitivity matrix during the calculation . the two methods of non - linearity correction introduced in this section have similar effects . one is derived from the second order terms of the taylor series expansion , while the other represents the non - linearity from a potential distribution perspective . in algorithm 4 , either ( [ eq : ss ] ) or ( [ eq : s_ij ] ) can be used to compensate the non - linearity in ( [ eq : c ] ) . herein we only consider the correction obtained using ( [ eq : ss ] ) as this implementation is faster . to test the proposed algorithm , numerical simulations were performed on an ect model . the results of reweighted tv - fist algorithm are compared with the performances of several widely used algorithms in practice , which include lbp , art ( relaxed kacmarz iteration ) and sirt ( relaxed cimmino iteration ) . all reconstructions are carried out on a standard desktop pc with an amd phenom(tm ) @xmath131 ghz processor and @xmath132 gb ram . the simulations are run in matlab 2009b and @xmath133 iterations were run throughout for each of these algorithms . the ect system ( normally with @xmath134 or @xmath135 electrodes here ) was modeled using the comsol multiphysics software package , and the sensitivity matrix was generated from comsol for all reconstructions . as in @xcite , the normalised capacitance was used to help minimise the effect of non - linearity introduced by the wall of the sensor @xcite . firstly , we implement various algorithms on a phantom consisting of an arc - shaped part and a circular object in a @xmath136 pixel image , as shown in fig . [ fig:1 ] . in the ect system we use @xmath134 electrodes , which can provide @xmath137 independent capacitance measurements . the smaller round object has image intensity ( a.u . ) of @xmath138 and the larger object has an intensity of @xmath139 ; black and white in fig . [ fig:1 ] correspond to intensities of 0 and 1 , respectively . from fig . [ fig:1 ] ( b ) one can see that the two objects can be recovered approximately by the lbp reconstruction , however the shape is significantly smoothed and broadened compared with the true image in ( a ) . the landweber , art and sirt methods show a similar recovered result in ( c - e ) . errors in the permittivity distribution make precise identification of the boundary of the objects challenging . fig [ fig:1 ] ( f ) shows the reconstruction using reweighted tv - ist . the boundaries of both objects are clearly resolved with the correct intensity . the only significant error occurs at the wall of the system . the error at the wall is likely caused by non - linearity at the wall , or the use of the isotropic form of tv which can introduce smoothing at sharp points in the image . in the second simulation , the ect system consists of @xmath135 electrodes , which can provide @xmath140 independent capacitance measurements . the tested permittivity distribution was the ` two bubbles ' image , as shown in fig . [ fig : wwobubble](a ) . it is a phantom image consisting of a circular pipe containing two circular objects in a @xmath136 pixel image . in the simulation , the relative permittivity of the cylindrical wall and the background was set to @xmath141 ; the relative permittivity of the two circular objects were @xmath139 and @xmath142 for the large and small objects , respectively . [ fig : wwobubble](a ) is different from the normal ect permittivity distributions considered since the background has a high permittivity while the two bubbles have low permittivity . the lbp reconstruction of this image , shown in fig . [ fig : wwobubble](b ) , fails as both bubbles blur into a single object . the poor reconstruction arises from the close proximity of the two bubbles and the use of a high permittivity background . landweber iterations , fig . [ fig : wwobubble](c ) , gives a better result with the two bubbles resolved , but the bubbles still appear heavily smoothed . similar results were obtained for the art and sirt reconstructions . the linear tv - ist algorithm , [ fig : wwobubble](d ) , recovers the sharp boundaries around the two bubbles . however , a high permittivity `` bridge '' is seen connecting the two bubbles and the permittivity of the smaller bubble is over estimated . the proposed reweighted tv - ist result is shown in fig . [ fig : wwobubble](g ) . the outline of the two bubbles is recovered fairly accurately , with only a slight tendency of the two bubbles to merge together and the size of the two bubbles overestimated by @xmath143% . the `` bridge '' seen using the standard tv - ist algorithm has been eliminated . the permittivity is also recovered fairly well with the permittivity of the large bubble found to be @xmath139 and the permittivity of the small bubble @xmath144 , which compare with the input permittivities of @xmath139 and @xmath142 , respectively . the non - linear reweighted reconstruction is shown in fig . [ fig : wwobubble](h ) . the recovered bubble shapes are slightly more `` square '' than the input bubble shapes , but otherwise the outline of both bubbles is recovered well . the size of each bubble is accurate to within @xmath145% of the true bubble size . the permittivity in the large and small bubbles was @xmath139 and @xmath146 , respectively , in good agreement with the true values . the reconstruction quality is sensitive to the choice of the parameters @xmath147 , @xmath98 , and @xmath112 , as well as the number of iterations performed . however , overall these results demonstrate that the introduction of the reweighted tv - ist algorithm , including non - linearity correction , significantly improves the quality of the reconstructed images for piecewise smooth input permittivity distributions . the re - weighting approach enables the solution to approach the true @xmath3-norm solution closely , while the updates to the sensitivity matrix during image reconstruction help mitigate against the non - linearity effects . in this paper , a non - linear reweighted total variation algorithm for reconstruction of images obtained from ect measurements has been proposed and analyzed . the proposed algorithm penalises the @xmath0-norm of the spatial finite differences of the image ( total variation ) by using an iterative thresholding approach . a varying weight calculated in each iteration is used to make sure that the result converges towards the desired @xmath3-norm . in addition , the non - linearity of the governing equations was considered and a straightforward approach to update the sensitivity matrix was introduced accordingly . the proposed algorithm was verified on two simulated permittivity distributions . it is shown that the reweighting significantly increases the quality of the reconstructed images recovering sharper boundaries with fewer artefacts than existing algorithms including lbp , art , sirt and our previous implementation of tv - ist . the incorporation of the updated sensitivity matrix to approximate the non - linearity of the ect sensor further increased the accuracy of the reconstructed images , most notably in recovering quantitative permittivity values in each domain . the new algorithm here promises to increase the quality of ect imaging . we anticipate even greater benefits if the algorithm can be combined with recently proposed enhanced sensing strategies @xcite . this work was partially supported by the epsrc grant reference : ep / k008218/1 . the authors would like to thank t.c . chandrasekera and yi li for assisting with the comparison to existing image reconstruction algorithms . q. marashdeh , w. warsito , l .- s . fan , and f. l. teixeira , `` nonlinear forward problem solution for electrical capacitance tomography using feed - forward neural network , '' _ ieee sensors journal _ , vol . 6 , pp . 441449 , 2006 . m. b. haddadi and r. maddahian , `` a new algorithm for image reconstruction of electrical capacitance tomography based on inverse heat conduction problems , '' _ ieee sensors journal _ , vol . 16 , pp . 17861794 , 2016 . w. yang , d. spink , t. york , and h. mccann , `` an image - reconstruction algorithm based on landweber s iteration method for electrical - capacitance tomography , '' _ measurement science and technology _ , vol . 10 , pp.1065 , 1999 . l. h. peng , h. merkus , and b. scarlett , `` using regularization methods for image reconstruction of electrical capacitance tomography , '' _ particle and particle systems characterisation _ , vol . 96104 , 2000 . t. c. chandrasekera , y. li , j. s. dennis , and d. j. holland , `` total variation image reconstruction for electrical capacitance tomography , '' in _ ieee international conference on imaging systems and techniques ( ist ) , _ , jul . 2012 , pp . 584589 . j. ye , h. wang , and w. yang , `` image reconstruction for electrical capacitance tomography based on sparse representation , '' _ ieee transactions on instrumentation and measurement _ , vol . 64 , pp . 89102 , 2015 . w. xu , m. wang , j. f. cai , a. tang , `` sparse error correction from nonlinear measurements with applications in bad data detection for power networks , '' _ ieee transactions on signal processing _ 61 , pp.61756187 , 2013 . m. soleimani and w. r. b. lionheart , `` nonlinear image reconstruction for electrical capacitance tomography experimental data using experimental data , '' _ measurement science and technology _ , vol . 16 , pp . 19871996 , 2005 . y. liu , j. ma , y. fan , and z. liang , `` adaptive - weighted total variation minimization for sparse data toward low - dose x - ray computed tomography image reconstruction , '' _ physics in medicine and biology _ , vol . 57 , pp . 79237956 , 2012 . m. guerquin - kern , m. haberlin , k. pruessmann , and m. unser , `` a fast wavelet - based reconstruction method for magnetic resonance imaging , '' _ ieee transactions on medical imaging _ , vol . 30 , pp . 16491660 , 2011 . d. l. donoho and m. elad , `` optimally sparse representation in general ( nonorthogonal ) dictionaries via @xmath0 minimization , '' _ proceedings of the national academy of sciences _ , vol . 21972202 , 2003 . a. beck and m. teboulle , `` fast gradient - based algorithms for constrained total variation image denoising and deblurring problems , '' _ ieee transactions on image processing _ , vol . 18 , pp . 24192434 , 2009 . g. villares , l. begon - lours , c. margo , y. oussar , j. lucas , and s. hole , `` a non - linear model of sensitivity matrix for electrical capacitance tomography , '' in _ proceedings of the electrostatics joint conference _ , jun . 2010 . t. c. chandrasekera , y. li , d. moody , m. schnellmann , j. s. dennis , and d. j. holland , `` measurement of bubble sizes in fluidised beds using electrical capacitance tomography , '' _ chemical engineering science _ , vol . 679 687 , 2015 .
a new iterative image reconstruction algorithm for electrical capacitance tomography ( ect ) is proposed that is based on iterative soft thresholding of a total variation penalty and adaptive reweighted compressive sensing . this algorithm encourages sharp changes in the ect image and overcomes the disadvantage of the @xmath0 minimization by equipping the total variation with an adaptive weighting depending on the reconstructed image . moreover , the non - linear effect is also partially reduced due to the adoption of an updated sensitivity matrix . simulation results show that the proposed algorithm recovers ect images more precisely than existing state - of - the - art algorithms and therefore is suitable for the imaging of multiphase systems in industrial or medical applications . electrical capacitance tomography ( ect ) , iterative reconstruction , reweighted total variation , non - linear effect .
in recent years , the interest in describing slow and fast dynamics of many - body fermionic systems has strongly increased . this renewal of interest is largely due to the advances in experimental studies that allow to form and manipulate a limited set of fermions in the laboratory @xcite . dedicated studies have been recently made on the fast dynamics of systems after a quantum quench @xcite . most , if not all , quantum transport theories dedicated to interacting fermions use the mean - field theory as a starting point . deterministic approaches , based on density matrix or non - equilibrium green function ( negf ) evolution , that aim at including the effect of correlations turn out to be rather involved numerically and have been applied with varying predictive powers @xcite . in parallel , several stochastic methods where noise is added to the mean - field , have been proposed ( for a recent review see @xcite ) . among these approaches , one has recently attracted more interest due to its simplicity and its rather surprisingly good predictive power . this approach , called stochastic mean - field ( smf ) was historically introduced in the nuclear physics context @xcite . in recent years , it was firstly validated in schematic model in ref . @xcite , extended in ref . @xcite to describe small superfluid systems and shown to be competitive with most recent negf theories applied to lattice systems in ref . @xcite . the smf approach starts from the following observation . time - dependent mean - field is generally adequate for describing the evolution of one - body collective observables but not their fluctuations especially in finite systems where zero point fluctuations can be large . in addition , even if fluctuations are accounted for at initial time , their effect on one - body evolution is poorly treated . as a matter of fact , tdhf can indeed be considered as a quasi - classical approximation for many - body fermionic systems where quantal fluctuations due to two - body correlations are poorly treated . the smf approach is based on the fact that a quantum system can sometimes be accurately simulated by considering an ensemble of classical trajectories with initial values sampled from the initial phase - space @xcite . the use of such a strategy , depicted in fig . [ fig : schem ] , is far from being straightforward in many - body systems . in the pioneering work of @xcite , the collective space for fermions was assumed to correspond to the full set of one - body observables . then , four crucial assumptions have been made : ( i ) the initial phase - space has been obtained by imposing that the quantal average of first and second moments of one - body observables identifies with classical average . ( ii ) consistently with the hypothesis ( i ) , the sampling from the initial phase - space can be performed using a set of eventually correlated gaussian random variables . ( iii ) the mean - field dynamics is used to propagate each initial conditions . ( iv ) at each time step , quantal fluctuations can be estimated using classical average . in ref . @xcite , some formal arguments were given to support these ideas . afterwards , it was shown in different test cases that the smf concept @xcite can be more predictive and versatile than anticipated . in the present article , we analyze in more details the hypothesis made for the initial sampling , i.e. ( i ) and ( ii ) . in particular , we show that such hypothesis seems to be rather effective if the initial state is uncorrelated , i.e. if it is a pure slater determinant state . the accurate description of correlated systems requires to go beyond the gaussian approximation for the initial phase - space . we show in that case , that smf can be combined with the standard husimi method to further increase its predictive power especially in the strong coupling case . our discussion is illustrated with the lipkin - meshkov - glick ( lmg ) model that has been used as the first sensitive test of smf @xcite . in this section , we will recall the minimal ingredients of the lmg model that are useful for the present discussion . more details can be found in refs . @xcite . the lmg model was originally invented to represent the interaction of nucleons in two n - fold degenerate levels ( one above and the other below the fermi level ) of a nucleus @xcite . it can also describe a system of n interacting spins in the xy plane under an influence of transverse magnetic field @xcite . the lmg hamiltonian is given by @xmath0 where @xmath1 is the energy gap between the two states and @xmath2 is the interaction strength . we set @xmath3 . the quasispin operators are defined as @xmath4 with @xmath5 where @xmath6 and @xmath7 are creation operators of the upper and lower single - particle levels , respectively . the quasispin operators satisfy the usual su(2 ) commutation relation , @xmath8=ij_z,\end{aligned}\ ] ] and its cyclic permutations . the hamiltonian in eq . ( [ ham ] ) commutes with the total quasispin operator @xmath9 which has an eigenvalue @xmath10 . hence the spin quantum number @xmath11 is conserved . using the standard eigenvectors @xmath12 of @xmath9 and @xmath13 operators , exact solution of the lmg model can be obtained by forming the matrix representation of the hamiltonian in the @xmath12 basis and diagonalizing the resulting @xmath14 matrix . the mean - field approximation of the lmg model can be obtained by using the su(2 ) spin coherent states @xcite . the spin ( bloch ) coherent states are defined by @xmath15 where the two angles cover the parameter space @xmath16 and @xmath17 @xcite . the coherent states are normalized , @xmath18 , and the closure relation is given by @xmath19 with @xmath20 . the coherent state algebra , discussed in detail in refs . @xcite , can be used to get any moments of the spin operators . the expectation values of the first and second moments of quasispin operators are given by @xmath21 and @xmath22 where @xmath23 is the symmetrical ordering of operators , @xmath24 and @xmath25 . by using the ehrenfest theorem and eqs . ( [ firstmom ] ) and ( [ secondmom ] ) , the mean - field equations @xcite @xmath26 can easily be obtained where the expectation values given by eq . ( [ firstmom ] ) are replaced with c - number quasispins @xmath27 . @xmath28 is a relative interaction strength parameter . the mean - field equations , eq . , can be derived either starting from a coherent state or from a statistical ensemble of coherent states . then , the mean - field equations are solved with the initial values @xmath29 where @xmath30 is the initial density matrix . if the initial state is a coherent state , we have @xmath31 . as we will see below , the smf approach allows to also make use of the mean - field equations , even if the initial state is correlated . in the mean - field approximation , time evolution of expectation values of the second and higher moments of collective observables depends only on the expectation values of the first moments , therefore quantum correlations of collective variables are mostly neglected leading to a classical picture . as demonstrated recently @xcite , the smf approach can partially cure this deficiency . the assumption of smf is that the dynamics of a many - body system can be approximated by an ensemble of independent mean - field evolutions starting from a statistical ensemble of initial values @xmath32 instead of the deterministic initial values given by eq . . here , @xmath33 labels a specific configuration of the ensemble . in the original formulation of smf @xcite as well as in recent applications @xcite , it was proposed to sample the initial values using a gaussian stochastic process in such a way that the first and second quantum moments of selected one - body observables , here @xmath34 , are exactly reproduced at initial time . then , each event in the resulting ensemble of initial values @xmath35 is evolved by the standard mean - field equations given by eq . ( [ mf ] ) . the final results are obtained by averaging over the ensemble . this approach was shown to describe accurately the short - time evolution of many different quantum fermionic many - body systems @xcite . since the mapping between the quantal and classical problem is made by imposing solely the first two moments of a selected set of collective variables , the gaussian assumption is a simple practical solution to the initial phase - space sampling problem and seems quite natural . however , in the following , we show that it might be sometimes advantageous to relax this hypothesis . below , we first test the validity of the gaussian assumption by considering alternative initial distributions . let us assume a simple distribution function , denoted generically by @xmath36 , for the stochastic quantities @xmath32 that can reproduce the initial quantum expectation values of the first few moments of the observables . hence we have @xmath37_d&=&\langle \psi_0|j_i|\psi_0\rangle,\nonumber \\ \left[j_i^\lambda(0)j_j^\lambda(0)\right]_d&=&\langle\psi_0|\widetilde{j_ij_j}|\psi_0\rangle,\nonumber\\ & \vdots & \label{vdots}\end{aligned}\ ] ] where @xmath38_d$ ] stands for averaging over the assumed ( predefined ) distribution function . in general , it is possible to define infinitely many distributions that satisfy the expectation values of the same set of finite moments . therefore , smf approach is expected to depend on what distribution is chosen and how many expectation values are satisfied by the chosen distribution . for this purpose , we consider three distribution functions for the observables @xmath27 , normal ( gaussian ) , uniform , and a bimodal distribution . figure [ fig : d ] shows the shapes of these distribution functions . the explicit forms and some properties of these distributions are given in appendix a. the parameters of the all three distribution functions are arranged so that the same quantum means and ( co)variances are reproduced . . all distributions have zero mean and unit variance . the kurtosis for the bimodal distribution is arbitrarily chosen to be equal to 1.1.,width=283 ] in all the following computations , time is measured in units of @xmath39 . the number of events for stochastic sampling is @xmath40 . the number of particles is @xmath41 , hence @xmath42 . for the sake of brevity , only one observable that best demonstrates the differences for compared conditions is plotted . the plots of the other observables , such as means and variances of quasispins , have similar differences for the compared conditions . in fig . [ fig : jz1 ] , the exact variance of the observable @xmath13 is compared with the variances computed with smf approach using the three distributions . the initial state is the bloch coherent state @xmath43 which is defined by eq . ( [ bloch ] ) . it is worth to mention that any bloch coherent state corresponds to a slater determinant in the lmg model @xcite . hence , slater determinant and coherent state terminologies are interchangeably used in the text . similarly to what has been observed in previous applications , the smf approach applied to an initial coherent state turns out to be almost identical to the exact solution for weak two - body interactions . while a careful look to the smf evolution shows that the use of gaussian approximation for the initial distribution is slightly better than the use of uniform or bimodal distribution , we see that the evolution is overall insensitive to the assumed distribution in the weak - coupling regime . this is quite surprising in view of the completely different shapes of the distributions ( see fig . [ fig : d ] ) . in contrary , in the strong coupling case , the evolution is quite sensitive to the assumption for the initial distribution . in that case , a much better agreement with the exact evolution is obtained with the normal distribution . from this analysis , it is tempting to conclude that the gaussian assumption is validated . however , the better agreement directly stems from the nature of the initial state . indeed , are plotted versus time for the initial coherent state @xmath44 . mf stands for mean - field and b , u , and n signify bimodal , uniform , and normal distributions , respectively . a ) strong coupling ( @xmath45 ) case . b ) weak coupling ( @xmath46 ) case.,width=283 ] coherent states have gaussian ( or gaussian - like ) shapes that give gaussian - like distributions for observables . this conclusion can be drawn by analyzing the relative third and fourth centered moments generally called skewness and kurtosis of the initial state ( see appendix a for details ) . the quantum skewnesses and kurtoses can be easily obtained from eq . ( [ qusk ] ) as @xmath47 for the initial coherent state @xmath44 . the quantum skewnesses are close to zero and the quantum kurtoses are very close to 3 which is the kurtosis of normal distribution . for any initial bloch coherent state @xmath48 , the quantum kurtoses are very close to the kurtosis of normal distribution , therefore normal distribution is the best choice for the initial distribution of collective observables when the initial state is a coherent state . this will be further illustrated below by a direct focus on the initial phase - space . the smf approach has been originally formulated for initial states that are either pure independent particle states or statistical independent particle states . here , we show that smf can also be applied starting from a correlated state using exactly the same strategy . in that case , the initial phase - space is obtained by mapping the quantum average including correlation effect to classical average and then performing independent mean - field evolutions . are plotted versus time for the initial non - coherent state @xmath49 . b , u , and n signify bimodal , uniform , and normal distributions , respectively . the mean - field solution is constant and equal to 0 , hence it is not shown . a ) strong coupling ( @xmath45 ) case . b ) weak coupling ( @xmath46 ) case.,width=283 ] figure [ fig : mixed1 ] is the same with figure [ fig : jz1 ] except that initial state is the correlated ( non - coherent ) state @xmath50 . this state , being an eigenstate of the two - body operator @xmath9 , is strongly entangled in terms of the coherent states . using the closure relation eq . ( [ eq : closure ] ) , this state can be written as @xmath51 can be eventually obtained by inverting eq . ( [ bloch ] ) . starting from this correlated state , the mean - field solutions for quasispins are constant and equal to their initial values . contrary to all previous applications where a simple slater determinant was considered initially @xcite , it is observed that smf approach for all three distribution functions fails in the weak coupling regime . in this regime , it is also observed that the dynamics is very sensitive to the assumption made for the distribution . similar conclusions are drawn in the strong coupling regime . in particular , the gaussian approximation appears to be less accurate . this can be phenomenologically understood from the quantum skewnesses and kurtoses which are obtained as @xmath52 for the correlated initial state @xmath49 . in particular , the kurtosis for the correlated state is far from the gaussian limit . one of the motivations to introduce the bimodal distribution is that it is slightly more flexible than the others . in particular , not only the first two moments can be imposed to match the quantal ones , but also the third and fourth moments can be imposed simultaneously . in fig . [ fig : mixed1 ] , the kurtosis of the bimodal distribution is chosen to be equal to the quantum kurtosis given in eq . ( [ bimkur ] ) . it is seen that even though the bimodal distribution satisfies the quantum expectation values up to the fourth moment of quasispin operators @xmath53 , the agreement between smf ( b ) and the exact solution is poor for weak and strong coupling cases . this is due to the fact that , in general , non - coherent states lead to complicated distribution functions for observables that can not be guessed . as a matter of fact , the disagreement observed in fig . [ fig : mixed1 ] directly stems from the inadequacy of any assumed initial distribution to account for the complexity of the real distribution of the observables . this observation points out the necessity to improve the description of the initial phase - space without necessarily relying on a specific assumption for the initial distribution . in the present section , we show that standard techniques generally used to obtain the phase - space of a quantum objects can be advantageously combined with the smf approach to significantly improve the description of complex fermionic many - body systems . the phase - space formulation of quantum many - body systems is generally done by using one of the glauber - sudarshan p , wigner , or husimi q quasiprobability distribution functions ( see for instance @xcite ) . a stochastic extension of the mean - field description of bosonic systems , especially the bose - einstein condensate ( bec ) , is introduced by what is called truncated wigner approximation ( twa ) ( see @xcite and references therein ) . in this approach , the mean - field evolution is given by the time - dependent gross - pitaevskii equation and stochastic sampling from the initial wigner distribution function is performed @xcite . however , in general , wigner as well as glauber - sudarshan functions can assume negative values while the husimi function is strictly positive - definite hence it can be easily used for stochastic simulations . for that reason , we use the husimi function in the present study . in ref . @xcite , the initial husimi distribution function for the bose - hubbard model is used for constructing the initial quantum fluctuations . the gross - pitaevskii equation is again evolved with an ensemble of initial values that are sampled from the husimi function . our approach , while having a similar strategy , is different than the mentioned studies for two reasons : ( i ) the approach we proposed can be applied to fermionic as well as bosonic systems that can be expressed in terms of some set of collective one - body observables . ( ii ) we provide a method to obtain initial phase - space distribution of any set of one - body observables and evolve them with the corresponding mean - field equations . as a result , a connection with the previous discussion on the smf approach with a predefined distribution function is clearly provided . we have seen in the previous discussions that the initial distribution of some desired observables ( quasispins @xmath54 in here ) can not be guessed in general situations . then , the only way to obtain the shapes of distributions of these observables seems to be through a quasiprobability distribution function of conjugate phase - space variables . before we explain how the initial distribution of observables is obtained with the husimi function , we provide some necessary background information . the husimi function is defined as @xmath55 where @xmath56 is the initial density operator . the expectation value of an arbitrary operator @xmath57 in terms of the husimi function is given by @xmath58 where @xmath59 is the antinormally ordered weyl symbol of operator @xmath57 . it is defined by @xmath60 the weyl symbols of quasispin operators read as @xmath61 where the expectation values are given by eq . ( [ firstmom ] ) . one of the main subtlety with the husimi function is the fact that it is a completely positive - definite distribution although it is representing a quantum object . indeed , the price to pay to use a positive distribution is the necessity to use a renormalized quantity @xmath62 instead of directly @xmath63 in eq . ( [ expa ] ) . as we will see below this subtle aspect has to be taken with care when coupling to smf is performed . the first advantage of the husimi method is that it can directly give access to the probability distribution of any states without assuming a priori its shape . the initial distribution of the desired observables can be written in terms of the conjugate phase - space variables by using the properties of the husimi function . in lmg model , the conjugate phase - space variables are @xmath64 and @xmath65 which are the normalized relative population difference between the upper and lower single - particle states and the relative phase between the upper and lower states , respectively . the husimi function , eq . provides the distribution of these conjugate phase - space variables , that is the distribution of angles @xmath66 and @xmath67 . these distributions are shown in the top panels of figs . [ fig : hus1 ] and [ fig : hus2 ] respectively for the pure coherent state and correlated state considered previously . to make contact with the distribution of quasispins used in smf and discussed previously , it is necessary to transform the two angles into the @xmath27 components . this transformation turns out to be less straightforward than it could be anticipated due to the use of the weyl symbols in quantum averages . the following strategy has been used here . first , an ensemble of stochastic angles @xmath68 that are sampled according to the husimi function can be formed by using the rejection method . at this point , a method that transforms the ensemble of stochastic angles @xmath69 to an ensemble of stochastic quasispins @xmath35 is required . the direct approach of defining the quasispins as @xmath70 where the weyl symbol is obtained by substituting the stochastic angles into eq . ( [ correc1 ] ) will give the correct quantum means when averaging over the ensemble is performed but the variances will be different than the quantum ones , since @xmath71 . this problem can be resolved by defining the quasispins in an effective way . let us write the quasispin components as an average and fluctuating part , @xmath72_h + \delta j_i^\lambda(0),\end{aligned}\ ] ] where @xmath38_h$ ] means averaging over the husimi distribution @xmath73 . guided by the smf approach , the properties of the average and fluctuating part in the above equation can directly be deduced imposing that they match the quantum ones associated to the considered state . the first moment directly gives @xmath37_h&=&\langle\psi_0|j_i|\psi_0\rangle = \left[p_{j_i}(\omega^\lambda)\right]_h,\end{aligned}\ ] ] while the second moment imposes the constraint @xmath74_h&=&\langle\psi_0|j_i^2|\psi_0\rangle-\langle\psi_0|j_i|\psi_0\rangle^2 \nonumber \\ & = & \left[p_{j_i^2}(\omega^\lambda)\right]_h-\left[p_{j_i}(\omega^\lambda)\right]_h^2 , \end{aligned}\ ] ] where eq . has been used . assuming that the fluctuations in the new effective quasispins are directly proportional to the fluctuations in the weyl symbols , finally gives @xmath75_h-\left[p_{j_i}(\omega^\lambda)\right]_h^2}{\left[\left(\delta p_{j_i}(\omega^\lambda)\right)^2\right]_h}}\;\delta p_{j_i}(\omega^\lambda ) , \label{husqs}\end{aligned}\ ] ] where @xmath76_h$ ] . in summary , the quasispin phase - space associated to a given initial state can be obtained using the numerical scheme : @xmath77 where the angles are sampled according to the husimi function , eq . . then , the associated set of weyl symbols are computed from eqs . and . finally , the set of quasispins are deduced from eq . ( [ husqs ] ) . the great advantage of the present method is that it provides automatically the distribution of quasispins with first and second moments that match the exact quantum ones , hence removing the need for guessing the shape of the initial probability distribution . we illustrate here the strength of the approach for the two initial states considered previously . the husimi functions for an initial coherent state @xmath48 and for an initial correlated state @xmath12 are given by @xmath78^{2j } , \label{hus1}\end{aligned}\ ] ] and @xmath79 respectively . ) for the initial coherent state @xmath80 is shown . in this plot , husimi function is normalized so that its maximum value is 1 . figures b ) and c ) show distributions of @xmath81 and @xmath82 , respectively . the filled areas indicate distributions that are obtained according to the husimi function whereas black lines indicate the normal distributions.,width=283 ] ) is shown . in this plot , husimi function is normalized so that its maximum value is 1 . b ) the distribution of @xmath81 is shown . the filled area indicates the distribution that is obtained according to the husimi function . the three distribution functions are also indicated . the kurtosis for the bimodal distribution is the value given by eq . ( [ bimkur]).,width=283 ] figure [ fig : hus1]a shows the shape of the husimi function for the initial coherent state @xmath80 . it is seen that the shape is close to that of a gaussian . figures [ fig : hus1]b and [ fig : hus1]c indicate the distributions of quasispins @xmath81 and @xmath82 , respectively . the distribution of @xmath83 is the same with that of @xmath81 . the quasispin distributions obtained from the husimi function are close to gaussian shapes . this confirms the skewness and kurtosis analysis made previously and also explains why in this case , the gaussian assumption made at initial stage of smf was more predictive than the other distributions . in fig . [ fig : hus2]a , the shape of the husimi function for the initial correlated state @xmath84 is shown . the husimi function is independent of @xmath67 , hence @xmath67 is uniformly distributed . figure [ fig : hus2]b shows the distribution of @xmath81 . the distribution of @xmath83 is the same with that of @xmath81 . the @xmath85 component is not distributed since its quantum variance is zero , hence it has a single initial value . note that even though the first four moments of quasispins obtained from the bimodal ( b ) distribution ( black line ) and husimi function ( red filled area ) are the same , the shapes of these two distributions are completely different . from fig . [ fig : hus2 ] , it is clear that ( i ) none of the assumed initial distributions is close to the real phase - space , ( ii ) this phase space could have hardly been guessed without the present analysis . we now return to one of the main motivation of the present work , and show that a realistic description of the initial phase - space can extend the domain of validity of smf and enable to describe the dynamics of an initially correlated state . the sampling method of initial quasispins based on the husimi function is now used to obtain the set of initial configuration for the mean - field dynamics . in fig . [ fig : husdyn1 ] , the exact variances of @xmath13 as well as the variances computed by using smf approach with normal ( n ) distribution and with distribution provided by the husimi ( h ) function are shown . it is seen that both smf ( n ) and smf ( h ) provide good agreement with the exact variances at weak and strong coupling regimes for the initial coherent state @xmath80 . systematic analysis was performed and it was observed that , for any initial coherent state , observables computed by using smf ( n ) and smf ( h ) approaches provide good agreement with the exact observables . figure [ fig : husdyn2 ] shows the time evolution of the variances of @xmath13 for the initial non - coherent state @xmath49 . in the weak coupling regime , it is observed that smf approach with distribution obtained by the husimi ( h ) function provides a very good approximation to the exact evolution and therefore considerable improvement over the results of fig . [ fig : mixed1 ] where the distributions have been predefined is achieved . therefore , we see that a more realistic description of the initial phase - space has a direct impact on the predictive power of smf approach . we performed systematic studies with different non - coherent states and observed that smf ( h ) always gives a better description of the time evolution of observables than smf ( n ) approach . are plotted versus time for the initial coherent state @xmath44 . the exact variances are compared with the approximate ones obtained with smf ( n ) and smf ( h ) methods . a ) strong coupling ( @xmath45 ) case . b ) weak coupling ( @xmath46 ) case.,width=283 ] are plotted versus time for the initial non - coherent state @xmath49 . the mean - field solution is constant and equal to 0 , hence it is not shown . a ) strong coupling ( @xmath45 ) case . b ) weak coupling ( @xmath46 ) case.,width=283 ] it is interesting to note from fig . [ fig : husdyn2 ] as well as fig . [ fig : mixed1 ] that the smf theory applied with gaussian approximation leads to an overdamping of the fluctuation compared to the exact case in the strong coupling regime . this has been misleadingly interpreted in refs . @xcite as a generic defect of smf . these two figures clearly demonstrate that this overdamping can disappear simply by changing the initial distribution . as an additional illustration of the importance of initial sampling , we consider a situation where the initial density can not be considered as a pure state density . we consider an initial density written as a mixing of several states , @xmath86 the corresponding husimi function is obtained as @xmath87 and is shown in fig . [ fig : mix1]a . as we see in this figure , the @xmath67 component is uniformly distributed while the @xmath66 component has a fermi shape . therefore none of the two angles resembles a gaussian distribution . in fig . [ fig : mix2 ] , the smf evolution starting from the mixed density using either a gaussian assumption or the sampling based on the husimi technique is compared with the exact solution . while the former again fails to provide the right evolution , the latter gives very good results both in the weak and strong coupling regime . ) is shown . in this plot , husimi function is normalized so that its maximum value is 1 . figures b ) and c ) show distributions of @xmath81 and @xmath82 , respectively . the distribution of @xmath83 is the same with that of @xmath81 . the filled area indicates the distribution that is obtained according to the husimi function . the normal distributions are also indicated.,width=283 ] are plotted versus time for the initial mixed state eq.([mixedst ] ) . the mean - field solution is constant and equal to 0 , hence it is not shown . a ) strong coupling ( @xmath45 ) case . b ) weak coupling ( @xmath46 ) case.,width=283 ] in the present work , we investigate the validity and sensitivity of the gaussian approximation made in the recently proposed stochastic mean - field approach @xcite . this approach , where the initial phase - space of a selected collective variable set is sampled to get a set of initial conditions followed by standard mean - field dynamics , has been shown to tremendously improve the independent particle picture by including the effect of quantal fluctuations beyond mean - field . using the lipkin - meshkov - glick model as an illustration , we show here that indeed , the gaussian assumption is a good approximation if the initial state is not correlated as was always the case in all applications up to now @xcite . however , when the initial state is correlated and corresponds to a mixing of slater determinants , smf dynamics with the gaussian approximation for the initial distributions can fail even in the weak coupling regime . we explicitly demonstrate that the failure comes from the complexity of the initial phase - space that can not be assumed to be gaussian anymore . this finding motivates the search for more accurate methods to describe initially correlated states . it is proposed here to combine the smf approach with more elaborate technique to obtain the initial phase - space of a quantum state . an illustration is given here using the husimi distribution that leads to a more realistic description of the initial conditions . a methodology is proposed to use this technique within the smf . we show that a more realistic description of the initial phase - space greatly improves the description of correlated systems . application to the lmg model gives perfect results in the weak coupling regime while in the strong coupling regime , the time over which the smf dynamics is in agreement with the exact solution is increased compared to the simple gaussian sampling assumption . the present study demonstrates that the predictive power of the smf approach can further be improved if a special attention is made on the initial phase - space and approximation beyond the gaussian sampling is made . this might in particular be crucial , when correlated initial quantum states are considered . here , we show how the husimi technique can be used to get a more realistic phase - space . this technique can be very helpful in situations where the coherent state algebra is well under control . such husimi approach has been recently applied in ref . note that other techniques like wigner transform that allow to make connection between quantum and classical mechanics can be used . such wigner functions have been , for instance , used in the nuclear context for a different purpose @xcite . the normal distribution function with means @xmath88 and ( co)variances @xmath89 , is given by @xmath90,\end{aligned}\ ] ] where @xmath91 and @xmath92 are vectors and @xmath93 is the variance - covariance matrix given by @xmath94 @xmath95 is the determinant of the matrix @xmath93 . the uncorrelated uniform distribution function is given by @xmath96 where @xmath97 we also arbitrarily define a bimodal distribution that is given by @xmath98 with @xmath99 for this distribution , the variances can be easily obtained in terms of @xmath100 and @xmath101 as @xmath102 each of the normal and uniform distribution functions depends on two sets of parameters @xmath88 and @xmath89 which can be fixed by choosing them to be equal to the quantum means and ( co)variances , that is @xmath104 and @xmath105 . on the other hand , the bimodal distribution depends on three sets of parameters @xmath88 , @xmath100 , and @xmath101 . therefore , next to the quantum means and ( co)variances , in principle , the expectation values of higher moments of observables can also be reproduced . the third and fourth moments of a distribution are generally expressed in terms of the skewness @xmath106_{d}/(\sigma^2_{ii})^{3/2}$ ] and the kurtosis @xmath107_{d}/(\sigma^2_{ii})^{2}$ ] . the skewnesses and kurtoses of the three distributions are given by @xmath108 the kurtosis of the bimodal distribution depends on the parameters @xmath100 , and @xmath101 . it is obtained as @xmath109 for any given variance and kurtosis which is within the interval provided in eq . ( [ gambet ] ) , the parameters @xmath100 and @xmath101 can be calculated by solving eqs . ( [ bivar ] ) and ( [ bikur ] ) . hence , the bimodal distribution introduces some freedom for the choice of kurtosis . the quantum skewness and kurtosis of the observables @xmath53 can be similarly defined for the initial state @xmath110 as @xmath111 where @xmath112 . the quantum skewnesses and kurtosis can be compared to the ones for the normal , uniform , and bimodal distributions . this will allow us to understand which distribution is closer to the actual distribution for a given initial state . correlated random variables can be produced by using uncorrelated random variables . let @xmath113 represent the vector of the desired correlated random variables with means represented by the vector @xmath114 and ( co)variances represented by the matrix @xmath115 . if @xmath116 is a vector of uncorrelated random variables with zero mean and unit variance , it is possible to obtain @xmath113 from @xmath116 by the equation @xmath117 for @xmath118 random variables . it is obvious that @xmath119_d=\mu$ ] since @xmath120_d$ ] is zero . the variance of the vector @xmath113 reads as @xmath121_d&=&\mu\mu^t+c\left[zz^t\right]_dc^t\nonumber\\ & = & \mu\mu^t+\sigma , \end{aligned}\ ] ] where @xmath122_d$ ] is unit matrix and @xmath123 is a given variance - covariance matrix . the matrix @xmath124 in eq . ( [ chol ] ) can be obtained from the variance - covariance matrix by using cholesky decomposition or eigendecomposition . 99 i. bloch , j. dalibard , and w. zwerger , rev . * 80 * , 885 ( 2008 ) . l. hackermller , j. p. ronzheimer , s. will , s. braun , t. best , i. bloch , e. demler , s. mandt , d. rasch , and a. rosch , nat . phys . * 8 * , 213 ( 2012 ) . d. stadler , s. krinner , j. meineke , j .- brantut , and t. esslinger , nature ( london ) * 491 * , 736 ( 2012 ) . n. strohmaier , d. greif , r. jrdens , l. tarruell , h. moritz , t. esslinger , r. sensarma , d. pekker , e. altman , and e. demler , phys . lett . * 104 * , 080401 ( 2010 ) . bernier , r. citro , c. kollath , and e. orignac , phys . lett . * 112 * , 065301 ( 2014 ) . u. schneider , l. hackermller , j. p. ronzheimer , s. will , s. braun , t. best , i. bloch , e. demler , s. mandt , d. rasch , and a. rosch , nat . phys . * 8 * , 213 ( 2012 ) . m. bonitz , _ quantum kinetic theory _ ( teubner , stuttgart / leipzig , 1998 ) . n. e. dahlen , and r. van leeuwen , phys . rev . lett . * 98 * , 153004 ( 2007 ) . sebastian hermanns , niclas schlnzen , and michael bonitz , phys . b * 90 * , 125111 ( 2014 ) . d. lacroix and s. ayik , eur . j. a * 50 * , 95 ( 2014 ) . s. ayik , phys . b * 658 * , 174 ( 2008 ) . d. lacroix , s. ayik , and b. yilmaz , phys . c * 85 * , 041602(r ) ( 2012 ) . d. lacroix , d. gambacurta and s. ayik , phys . c * 87 * , 061302(r ) ( 2013 ) . denis lacroix , s. hermanns , c. m. hinz , and m. bonitz , phys . b * 90 * , 125112 ( 2014 ) . m. f. herman and e. kluk , chem . phys . * 91 * , 27 ( 1984 ) . k. g. kay , j. chem . phys . * 100 * , 4432 ( 1994 ) ; * 101 * , 2250 ( 1994 ) . h. j. lipkin , n. meshkov , and a. j. glick , nucl . a * 62 * , 188 ( 1965 ) . p. ring and p. schuck , _ the nuclear many - body problem _ ( springer - verlag , new york , 1980 ) . a. p. severyukhin , m. bender , and p .- h . heenen , phys . c * 74 * , 024311 ( 2006 ) . wei - min zhang , da hsuan feng , and robert gilmore , rev . mod . phys . * 62 * , 867 ( 1990 ) . r. botet , r. jullien , and p. pfeuty , phys . lett . * 49 * , 478 ( 1982 ) . j. vidal , g. palacios , and r. mosseri , phys . a * 69 * , 022107 ( 2004 ) . j. vidal , g. palacios , and c. aslangul , phys . a * 70 * , 062304 ( 2004 ) . m. trindade dos santos , m. c. nemes , physica a * 259 * , 291 ( 1998 ) . f. trimborn , d. witthaut , and h. j. korsch , phys . a * 79 * , 013608 ( 2009 ) . eva - maria graefe , hans jrgen korsch , and astrid elisa niederle , phys . a * 82 * , 013629 ( 2010 ) . kan , p. c. lichtner , m. dworzecka , and j. j. griffin , phys . c * 21 * , 1098 ( 1980 ) . jean - pierre gazeau , _ coherent states in quantum physics _ , ( wiley - vch verlag , weinheim , 2009 ) . v. r. vieira and p. d. sacramento , ann . phys . * 242 * , 188 ( 1995 ) . w. gardiner and p. zoller , _ quantum noise _ , 2nd ed . ( springer - verlag , berlin , 2000 ) . n. p. proukakis , b. jackson , j. phys . b * 41 * , 203002 ( 2008 ) . n. p. proukakis , s. a. gardiner , m. j. davis and m. h. szymanska , _ quantum gases : finite temperature and non - equilibrium dynamics ( vol . 1 cold atoms series ) _ , pp.203 - 214 , ( imperial college press , london , 2013 ) ; arxiv:1009.1073 [ cond-mat.quant-gas ] . anatoli polkovnikov , phys . a * 68 * , 053604 ( 2003 ) ; phys . rev . a * 68 * , 033609 ( 2003 ) . n. loebl , a. s. umar , j. a. maruhn , p .- reinhard , p. d. stevenson , and v. e. oberacker , phys . c * 86 * , 024608 ( 2012 ) . n. loebl , j. a. maruhn , and p .- reinhard , phys . c * 84 * , 034608 ( 2011 ) . b. yilmaz , s. ayik , d. lacroix , and o. yilmaz , phys . rev . c * 90 * , 024613 ( 2014 ) .
in the stochastic mean - field ( smf ) approach , an ensemble of initial values for a selected set of one - body observables is formed by stochastic sampling from a phase - space distribution that reproduces the initial quantum fluctuations . independent mean - field evolutions are performed with each set of initial values followed by averaging over the resulting ensemble . this approach has been recently shown to be rather versatile and accurate in describing the correlated dynamics beyond the independent particle picture . in the original formulation of smf , it was proposed to use a gaussian assumption for the phase - space distribution . this assumption turns out to be rather effective when the dynamics of an initially uncorrelated state is considered , which was the case in all applications of this approach up to now . using the lipkin - meshkov - glick ( lmg ) model , we show that such an assumption might not be adequate if the quantum system under interest is initially correlated and presents configuration mixing between several slater determinants . in this case , a more realistic description of the initial phase - space is necessary . we show that the smf approach can be advantageously combined with standard methods to describe phase - space in quantum mechanics . as an illustration , the husimi distribution function is used here to obtain a realistic representation of the phase - space of a quantum many - body system . this method greatly improves the description of initially correlated fermionic many - body states . in the lmg model , while the gaussian approximation failed to describe these systems in all interaction strength range , the novel approach gives a perfect agreement with the exact evolution in the weak coupling regime and significantly improves the description of correlated systems in the strong coupling regime .
( a ) the density of states @xmath52 at 4 different pressures . the cutoff @xmath69 is identified as the point where the hessian matrix with ( solid line ) and without the stress term ( dashed line ) diverge . ]
we report the first measurements of the effect of pressure on vibrational modes in emulsions , which serve as a model for soft frictionless spheres at zero temperature . as a function of the applied pressure , we find that the density of states @xmath0 exhibits a low - frequency cutoff @xmath1 , which scales linearly with the number of extra contacts per particle @xmath2 . moreover , for @xmath3 , @xmath4 ; a quadratic behavior whose prefactor is larger than what is expected from debye theory . this surprising result agrees with recent theoretical findings @xcite . finally , the degree of localization of the softest low frequency modes increases with compression , as shown by the participation ratio as well as their spatial configurations . overall , our observations show that emulsions are marginally stable and display non - plane - wave modes up to vanishing frequencies . due to their translational symmetry , the vibrational modes of crystals are plane waves that exhibit the debye scaling in the density of states , @xmath5 , where @xmath6 is the spatial dimension . by contrast , amorphous materials , such as granular materials and glasses , exhibit peculiar elastic properties . in particular , they display an excess of low frequency vibrational modes in the density of states compared to crystalline solids , known as the `` boson '' peak @xcite . in simple model systems , such as sphere packings , the boson peak diverges in amplitude as the pressure is lowered towards the unjamming transition @xcite . the associated soft modes are responsible for anomalous elastic and transport properties @xcite and play an important role in plasticity and activated events @xcite . as the applied pressure increases above the jamming transition , their spatial localization increases @xcite . theoretically , it was argued that these properties result from the fact that packings near jamming are _ marginally stable _ : their structure is such that they lie at the threshold of a linear elastic instability controlled by pressure and coordination @xcite . very recently , a mean - field approximation @xcite and exact calculations in infinite dimensions @xcite indicated that for marginally stable structures at low frequencies @xmath4 in any spatial dimension , where @xmath7 is a characteristic frequency of the soft modes . in three dimensions , this quadratic behavior matches the debye result , but with a much larger prefactor . this result remains untested in real materials . experimentally , the density of states of colloidal glasses in 2d has been obtained from the thermal motion of the particles as a function of the global density below the glass transition @xcite . however , experimental tests on athermal particulate packings and in 3d are lacking in the literature , and are appropriate to accurately test if materials are indeed marginally stable . in this letter , we experimentally investigate the pressure dependence of the density of states and the localization of vibrational modes in compressed emulsions . because emulsions are effectively at zero temperature , the present experiment appears close enough to the jamming transition to precisely probe its properties . we find that the density of states can be rescaled by some frequency @xmath8 where @xmath9 is the packing fraction and @xmath10 its ( preparation dependent ) value at the jamming threshold , and find that it is consistent with the recently proposed scaling behavior @xmath11@xcite . finally , we examine the microscopic configuration of the soft modes by measuring the participation ratio of the particles and their spatial distribution . we find delocalized soft modes near jamming , and localized soft modes far above jamming . ( color online ) confocal image of an emulsion at a pressure of @xmath12 kpa and @xmath13 ( greyscale ) . for each reconstructed droplet @xmath14 in the middle , the low - frequency vibrational amplitude in the color map ( dim to bright ) is obtained as @xmath15 , where @xmath16 is the polarizaton vector of the @xmath17th eigenmode , summing over the lowest @xmath18 modes . ] our experimental system is an oil - in - water emulsion , described in references @xcite , which is a model for soft , frictionless , athermal sphere packings with an average droplet radius @xmath19 and a polydispersity of @xmath20 . the refractive index matched emulsion is transparent and the nile red dye reveals the 3d packing of droplets using a confocal microscope ( leica tcs sp5 ii ) . due to a density difference between the droplets and the continuous phase , the lowest compression rate is achieved by creaming under gravity , which gives a packing fraction of @xmath21 due to polydispersity , while centrifugation for 20 minutes at an acceleration rate of 3000 g leads to highly compressed structures with @xmath22 @xcite . allowing the emulsion to relax to its uncompressed state over a period of several days probes a broad range of intermediate packing densities @xcite , an example of which is shown in fig . [ fig1 ] . the packings are analyzed using a fourier transform algorithm to identify the particle positions and radii @xcite . we then extract the repulsive force @xmath23 exerted by particle @xmath24 on particle @xmath14 via the princen model @xcite : @xmath25 where @xmath26 and @xmath27 denote the particle positions , @xmath28 is the area of droplet deformation , or the geometric area of overlap between the reconstructed spheres @xcite , @xmath29 is the weighted mean radius of the droplets , and @xmath30 is the interfacial tension . we therefore sum over the force moment tensor for all contacts @xmath31 and then divide by the volume of the box to obtain the total pressure on the system , @xmath32 from the forces , we construct the interaction potential between contacting particles @xmath33 , where @xmath34 . this potential is harmonic for small deformations . to ensure mechanical equilibrium , we numerically quench the system into the nearest energy minimum to satisfy force balance using the method of steepest descent @xcite . this procedure does not move any of the particles beyond the resolution of the droplet finding technique , which is one third of a voxel size ( i.e. @xmath35 nm , or @xmath36 of the average droplet diameter ) . this correction does not have a measurable effect on the calculation of the density of states done below and gives validity to the force model in eq . . we then calculate the corresponding hessian matrices , defined for every pair @xmath14 , @xmath24 of particles in contact as @xmath37 , where the first term stems from the contact stiffness and the second from the contact force : @xmath38 and @xmath39 where the notation @xmath40 indicates that particle @xmath41 is in contact with particle @xmath14 . the polarization vectors @xmath42 and the eigenfrequencies @xmath43 can now be calculated via solution of the eigenvalue problem @xmath44 where @xmath45 denotes the mass of particle @xmath14 . note that in our system inertia is negligible , and dynamically modes would be overdamped . here we have included the masses to allow for a closer comparison with the literature . an alternative choice is to set @xmath46 for all @xmath14 . then our analysis would correspond to the normal modes of the hessian , which characterize the amplitude of thermal fluctuations ( independently of the damping mechanism ) . we have checked that both choices lead to essentially identical results . the amplitudes of the polarization vectors @xmath42 are visualized in fig . [ fig1 ] . since each particle must have all its neighbors and forces identified , we use @xmath47 particles in the middle of the experimental packing for each compression and repeat the experiment three times to collect statistics . we thus obtain a histogram of eigenmodes , i.e. the density of states @xmath0 , where @xmath48 is shown in units of natural frequency in fig . [ fig2](a ) for different levels of compression . the microscopic frequency is given by @xmath49rad / s , where stiffness @xmath50 . as expected , the log - log representation reveals a cut - off frequency at high values , and a progressive increase in the number of low - frequency modes as we approach the jamming transition under gravity , in agreement with numerical simulations @xcite . excess soft modes are generally characterized by the ratio @xmath51 , which often presents a maximum in simulations and in experiments on molecular glasses @xcite . our results do not reveal a maximum in @xmath52 , as shown in the supplemental material , similar to previously observed data from numerical packings of soft spheres @xcite . this absence of a peak can be understood in terms of the effect of the experimental control parameter , the pressure , on the system . pressure affects vibrational properties in two distinct ways : first , it affects the packing fraction , which in turn increases the mean number of contacts between particles , or coordination @xmath53 , and tends to stabilize the system further . this is illustrated in the inset of fig.[fig2](b ) , from previous measurements in the same set - up @xcite . second , pressure directly has a destabilizing effect that can lead to a `` buckling '' phenomenon where soft modes become unstable @xcite . these effects can be disentangled by considering the density of states with and without the applied stress term , i.e. by using @xmath54 or @xmath55 , respectively in eq.[007 ] . in the latter case , we keep the geometric configurations that were formed due to the applied pressure , but we effectively replace our droplets by point particles connected with relaxed springs . as predicted @xcite , we find that @xmath0 is affected by pressure only below a certain crossover frequency , @xmath1 , which is extracted as the frequency below which the densities of states with and without the pressure term begin to deviate ( see fig . s1 in si ) . note that @xmath52 has a maximum near @xmath1 if @xmath56 alone is considered . marginal stability @xcite indicates that as pressure increases , the stabilizing effect of increasing coordination ( apparent when considering @xmath56 alone ) precisely matches the direct destabilizing effect of the pressure ( induced by the term @xmath57 in the hessian ) . in other words , the packing just makes enough contacts to remain stable . as a result , it can be proven that @xmath58 @xcite . we confirm this prediction in fig.[fig2](b ) . these findings agree quantitatively with theory and numerical simulations @xcite , unlike recent colloidal experiments probing the jamming transition @xcite . for the rescaled @xmath59 for all pressures shows good agreement with the predicted slope of @xmath60 in eq.[dw ] . ] another prediction of marginal stability is that there is no debye regime at all . instead , the theory of marginal stability leads to @xmath61 below @xmath1 @xcite , with a factor larger than the theoretical prediction for the debye model @xcite , which would predict that @xmath62 . rescaling our data in fig.[fig2](a ) by @xmath59 reveals a satisfying data collapse for all pressures in fig.[fig.3 ] , consistent with eq.[dw ] . this result is in agreement with recent numerical simulations in high dimensions @xcite . since @xmath1 decreases to zero at the jamming transition , the regime below @xmath1 becomes progressively smaller , and the experimental resolution may not be high enough to definitively rule out the debye result . however , together with the softening of modes with compression , our results strongly support marginal stability . next , we study the degree of localization of vibrational modes by computing the participation ratio @xmath63 of a mode @xmath17 as a function of the applied pressure : @xmath64 the value of @xmath63 ranges from @xmath65 to @xmath66 , from modes localized to one particle to extended modes , in which all particles participate . fig.[fig3](a ) shows that the higher the compression and the lower the frequency , the lower the participation ratio , suggesting increasing localization . this observation is consistent with 2d experiments @xcite , three dimensional @xcite and higher dimensional simulations @xcite . the participation ratio does not report on the spatial configuration of the participating particles . it is therefore useful to visualize the increase in the degree of localization using a color map ( from light to dark blue ) for the vibrational amplitudes of the lowest frequency mode in the packing under gravity and that of the highest compression , @xmath67kpa , as shown in fig.[fig3](b , c ) , respectively . these configurations clearly show the distinction between extended and localized modes . at intermediate frequencies and pressures , the modes exhibit a coexistence between localized and disordered modes . in conclusion , we have characterized experimentally the effect of pressure on the spectrum of the vibrational modes for compressed emulsions . our central results are that both the scaling of the characteristic frequency @xmath1 as well as the behavior of @xmath0 for @xmath68 are consistent with the notion that emulsion are marginally stable . in the range that can be probed experimentally , our observations agree with recent quantitative mean - field theories describing such a marginal behavior @xcite , giving empirical support for these views . in the future , it would be very interesting to study if marginal stability is robust to different system preparations , and if it survives at very large compression @xcite . we thank eric degiuli and eric vanden - eijnden for insightful discussions . this work was supported primarily by the materials research science and engineering center ( mrsec ) program of the national science foundation under award number dmr-1420073 . 29ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop _ _ , edited by , , vol . ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreve.72.051306 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.107.108301 [ * * , ( ) ] link:\doibase 10.1063/1.3157261 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.110.048302 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
the easiest to understand and to manipulate , and the most natural states of the quantum harmonic oscillator are the number states @xmath0 . number states are eigenstates of the hamiltonian and , of course , are also eigenstates of the number operator @xmath1 , where @xmath2 and @xmath3 are the well known creation and annihilation operators , respectively . however , for any @xmath4 , no matter how big , the mean field is zero ; i.e. , @xmath5 , and we know that a classical field changes sinusoidally in time in each point of space ; thus , these states can not be associated with classical fields [ 1 , 2 ] . in the first years of the sixties of the past century , glauber [ 3 ] and sudarshan [ 4 ] introduced the coherent states , and it has been shown that these states are the most classical ones . coherent states are denoted as @xmath6 , and one way to define them is as eigenstates of the annihilation operator ; that is , @xmath7 . an equivalent definition is obtained applying the glauber displacement operator @xmath8 to the vacuum : @xmath9 ; we see then coherent states as vacuum displaced states . coherent states also have the very important property that they minimize the uncertainty relation for the two orthogonal field quadratures with equal uncertainties in each quadrature [ 1 , 2 ] . since then , other states have been introduced . in particular , squeezed states [ 5 ] have attracted a great deal of attention over the years because their properties allow to reduce the uncertainties either of the position or momentum , while still keeping the uncertainty principle to its minimum . because of this , they belong to a special class of states named minimum uncertainty states . once produced , for instance as electromagnetic fields in cavities , they may be monitored via two level atoms in order to check , or measure , that such states have been indeed generated [ 6 , 7 ] . based on the above properties , we can think on the eigenstates of the position as limit cases of the squeezed states . as the squeezed states are minimum uncertainty states , we can reduced to zero the uncertainty in the position , while the uncertainty in the momentum goes to infinity , so that we keep the uncertainty principle to its minimum . of course , there is also the option to reduce to zero the uncertainty in the momentum , while the position gets completely undefined , obtaining that way the possibility to define momentum eigenstates . in sections 1 and 2 , we analyze the possibility of define the position eigenstates as the limit of extreme squeezing of the squeezed states . in what follows , we will use a unit system such that @xmath10 . there are two equivalent forms to define the squeezed states . in the first one , introduced by yuen [ 8 ] , squeezed states are obtained from the vacuum as @xmath11 where @xmath12\ ] ] is the so - called squeeze operator . in this view , squeezed states are created displacing the vacuum , and after , squeezing it . note that when the squeeze parameter @xmath13 is zero , the squeezed states reduce to the coherent states . in this work , we will consider only real squeeze parameters , as that is enough for our intentions . in the way followed by caves [ 9 ] , the vacuum is squeezed and the resulting state is then displaced ; that means , that in this approach @xmath14 both definitons of the squeezed states agree when the squeeze factor is the same , @xmath15 , and when the modified amplitude @xmath16 of the caves approach is given by @xmath17 being @xmath18 and @xmath19 to analyze the uncertainties in the position and in the momentum of the squeezed states , we introduce , following loudon and knight [ 5 ] , the quadrature operators @xmath20 and @xmath21 where @xmath22 is the position operator and @xmath23 the momentum operator . note that the quadrature operators are essentially the position and momentum operators ; this definition just provides us with two operators that have the same dimensions . in order to show that really the squeezed states are minimum uncertainty states , we need to calculate the expected values in the squeezed state ( 1 ) of the quadrature operators ( 7 ) and ( 8) , and its squares . using ( 7 ) and ( 1 ) , we get @xmath24 the action of the squeeze operator on the creation and annihilation operators is obtained using the hadamard s lemma [ 10 , 11 ] , @xmath25 such that @xmath26 therefore , as @xmath7 and @xmath27 , it is easy to see that @xmath28 and that @xmath29 so , we obtain for the uncertainty in the quadrature operator @xmath30 , @xmath31 proceeding in exactly the same way for the quadrature operator @xmath32 , we obtain @xmath33 as we already said , we can then think in the position eigenstates and in the momentum eigenstates as limit cases of squeezed states . indeed , when the squeeze parameter @xmath13 goes to infinity , the uncertainty in the position goes to zero , and the momentum is completely undetermined . of course , when the squeeze parameter goes to minus infinity , we have the inverse situation , and we can think in define that way the momentum eigenstates . in the two following sections , we use the yen and the caves definitions of the squeezed states to test this hypothesis . from equation ( 14 ) above , we can see that in the limit @xmath34 the uncertainty for position vanishes and so a position eigenstate should be obtained ( from now on , we consider @xmath35 real ) , @xmath36 we have written a sub index @xmath37 in the position eigenstate in order to emphasis that fact . following the yuen definition @xmath38 , so @xmath39 we now write the squeeze operator as [ 12 ] @xmath40 where , as we already said , @xmath41 and @xmath42 . so , @xmath43 now , we develop the first operator ( from right to left ) in power series , we use the definition of the coherent states , @xmath44 , to obtain @xmath45 as @xmath34 , @xmath46 , which means that the only term that survives from the sum is @xmath47 , and then @xmath48 that would give an approximation for how to obtain a position eigenstate from the vacuum . however , note that the above expression does not depend on @xmath49 and therefore can not be correct . we now squeeze the vacuum and after we displace it . thus , in this case , @xmath50 we use again expression @xmath51 for the squeeze operator [ 12 ] , where @xmath52 and @xmath53 are defined in ( 5 ) and ( 6 ) , and we write the displacement operator as @xmath54 [ 12 ] , to obtain @xmath55 as @xmath56 and @xmath57 , we cast the previous formula as @xmath58 inserting two times the identity operator , written as @xmath59 , we get @xmath60 , and using the hadamard s lemma [ 10 ] , it is easy to prove that @xmath61 for any well behaved function @xmath62 ; thus @xmath63\exp \left[-\frac{\nu } { 2\mu } \left(\hat{a}^{\dagger } -\frac{x}{\sqrt{2}}\right)^2\right]|0\rangle .\ ] ] after some algebra , @xmath64\exp \left[-\frac{\nu } { 2\mu } \hat{a}^{\dagger^{2}}+\frac{x}{\sqrt{2}}\left(1+\frac{\nu } { \mu } \right)\hat{a}^{\dagger } \right]|0\rangle .\ ] ] we take now the limit when @xmath34 , or @xmath65 , so @xmath66 we get an expression that gives us the position eigenstates as an operator applied to the vacuum . unlike the yuen case , expression ( 21 ) , now we have an @xmath49 dependence and it looks like a better candidate to be the position eigenstate . in fact , in the next section , we will show that this really is an eigenstate of the position . we will try now an alternative approach to the eigenstates of the position . we can write a position eigenstate , simply by multiplying it by a proper unit operator @xmath67 therefore the position eigenstate @xmath68 may be written as [ 13 ] @xmath69 with @xmath70 ; such that @xmath68 may be re - written as @xmath71 that may be added via using the generating function for hermite polynomials [ 14 ] @xmath72 to give @xmath73 the above expression allows us to write the position eigenstate as an operator applied to the vacuum . note that this expression is the same as the one obtained using the caves definition for the squeezed states , formula ( 28 ) . we prove now that indeed ( 32 ) is an eigenvector of the postion operator ; for that , we write the position operator as @xmath74 , thus @xmath75 inserting the identity operator in the above expression as @xmath76 , we get @xmath77 as @xmath78 , @xmath79 , and @xmath56 , we obtain @xmath80 as we wanted to show . we can write ( 32 ) in terms of coherent states . we have @xmath81 thus @xmath82 with the expressions obtained , it is easy to show that the squeezed states have the form of a gaussian wave packet . to confirm this , we use the above expression to state that @xmath83 we write @xmath84 as @xmath85 , where we have just inserted the identity operator @xmath86 , and we use that @xmath87 , for any well behaved function @xmath88 , to obtain @xmath89}{\pi ^{1/4}}\langle \alpha |e^{-\frac{\hat{a}^{\dagger^{2 } } } { 2}}e^{\frac{r}{2}\hat{a}^2-r\hat{a}^{\dagger } \hat{a}}|\sqrt{2}x\rangle .\ ] ] as the coherent states @xmath6 are eigenfunctions of the annihilation operator @xmath3 , it is very easy to show that @xmath90 , so @xmath91}{\pi ^{1/4}}\langle \alpha |e^{\frac{r}{2}\hat{a}^2-r\hat{a}^{\dagger}\hat{a}}|\sqrt{2}x\rangle .\ ] ] in the appendix , we disentangle the operator @xmath92 as @xmath93 , and we get @xmath91}{\pi ^{1/4}}\langle \alpha |e^{-r\hat{a}^{\dagger } \hat{a}}e^{\frac{1-e^{2r}}{4}\hat{a}^2}|\sqrt{2}x\rangle .\ ] ] it is very easy to see that @xmath94 , and that @xmath95 , thus @xmath96 \right\rbrace \left\langle \alpha|\sqrt{2}e^{-r}x\right\rangle .\ ] ] finally , as @xmath97 , we have @xmath98\right\},\ ] ] as we wanted to show . we can now find the wave function of a coherent state as a function of the position [ 15 ] . we use equation ( 32 ) , that express the eigenstates of the position as an operator acting on the vacuum , and get that @xmath100{\pi } } e^{-\frac{\beta^{*^{2 } } } { 2}+\sqrt{2 } \beta ^ * x } \left\langle \beta|0\right\rangle = \frac{e^{-\frac{x^2}{2 } } } { \pi^{1/4 } } e^{-\frac{\beta^{*^{2 } } } { 2}-\frac{\left| \beta \right| ^2}{2}+\sqrt{2 } \beta ^ * x}\ ] ] as @xmath101 and @xmath102 . the husimi @xmath99-function [ 16 ] can be calculated from ( 45 ) simply as @xmath103 that after some algebra , can be re - written as @xmath104\ ] ] in figures 1 and 2 , we plot the husimi @xmath99-function for different values of @xmath49 . -function for @xmath105 and for @xmath106.,title="fig : " ] -function for @xmath105 and for @xmath106.,title="fig : " ] -function for @xmath107 and for @xmath108.,title="fig : " ] -function for @xmath107 and for @xmath108.,title="fig : " ] we have found an operator that applied to the vacuum gives us the eigenstates of the position . we did that by two ways ; first , using the caves definition of the squeezed states , we took the limit of extreme squeezing in the position side , to get the position eigenstate . second , we used the expansion of an arbitrary wave function in the base of the harmonic oscillator ; i.e. , we wrote an arbitrary wave function in terms of hermite polynomials . the expressions obtained allows us to show certain properties of the squeezed states , and also allow us to write in a very easy way the husimi _ q_-function of the position eigenstates . the same procedure can be followed to find the eigenstates of the momentum , but taken the limit when the squeeze parameters goes to @xmath109 . + we can also conclude that from the point of view of this work , the caves approach to the squeeze states is more adequate , since it gives the correct eigenstates of the position ; while the yuen definition , formula ( 1 ) , gives an expression that is incorrect . so , we must first squeeze the vacuum , and after , we displace it . in this appendix , we show how to disentangle the operator @xmath110 . we define @xmath111 and we suppose that ( 48 ) can be rewritten as @xmath112\exp \left[g(r)\hat{a}^2\right],\ ] ] where @xmath113 and @xmath114 are two unknown well behaved functions ; as @xmath115 , being @xmath116 the identity operator , these functions most satisfy the conditions @xmath117 . at first sight , one can think that in the proposal ( 45 ) should be a term of the form @xmath118 $ ] ; however , this is not the case because @xmath119=2\hat{a}^2 $ ] . we differentiate with respect to @xmath13 , to find @xmath120\exp \left[g\hat{a}^2\right]+\frac{dg}{dr}\exp \left[f\hat{a}^{\dagger } \hat{a}\right]\hat{a}^2\exp \left[g\hat{a}^2\right],\ ] ] where for simplicity in the notation , we have dropped all @xmath13-dependency ; we write the identity operator as @xmath121\exp \left[f\hat{a}^{\dagger } \hat{a}\right]$ ] in the second term , to obtain @xmath120\exp \left[g\hat{a}^2\right]+\frac{dg}{dr}\exp \left[f\hat{a}^{\dagger } \hat{a}\right]\hat{a}^2\exp \left[-f\hat{a}^{\dagger } \hat{a}\right]\exp \left[f\hat{a}^{\dagger } \hat{a}\right]\exp \left[g\hat{a}^2\right].\ ] ] using the hadamard s lemma [ 10 , 11 ] , it is very easy to prove that @xmath122\hat{a}^2\exp \left[-f\hat{a}^{\dagger } \hat{a}\right]=e^{-2f}\hat{a}^2,\ ] ] so @xmath123 equating this equation to the one obtained differentiating the original formula for @xmath124 , equation ( 44 ) , we get the following system of first order ordinary differential equations @xmath125 the solution of the first equation , that satisfies the initial condition @xmath126 , is the function @xmath127 . substituting this solution in the second equation and solving it with the initial condition @xmath128 , we obtain @xmath129 . thus , finally we write @xmath130 c. gerry and p. knight , _ introductory quantum optics_. cambridge university press ( 2005 ) . j.c . garrison and r.y . chiao , _ quantum optics_. oxford university press ( 2008 ) . glauber , _ coherent and incoherent states of the radiation field _ , _ phys . _ , 2766 ( 1963 ) . sudarshan , _ phys . _ , 277 ( 1963 ) . r. loudon and p.l . knight , special issue of _ j. of mod . opt . _ , 709 ( 1987 ) . satyanarayana , p. rice , r. vyas , and h.j . carmichael , _ j. opt . soc.am._ b , 228 ( 1989 ) . h. moya - cessa and a. vidiella - barranco , _ j. of mode . _ , 2481 ( 1992 ) . yuen , _ phys . rev . _ a , 2226 ( 1976 ) . caves , _ phys . _ d , 1693 ( 1981 ) . h ' ector manuel moya - cessa and francisco soto - eguibar , _ introduction to quantum optics_. rinton press , ( 2011 ) . h ' ector manuel moya - cessa and francisco soto - eguibar , _ differential equations . an operational approach_. rinton press , ( 2011 ) . werner vogel and dirk - gunnar welsch , _ quantum optics _ , third , revised and extended edition . schleich , _ quantum optics in phase space . _ wiley - vch , ( 2001 ) . g. arfken , _ mathematical methods for physicists_. academic press , inc . , 3rd edition , ( 1985 ) . u. leonhardt , _ measuring the quantum state of light . _ cambridge university press , ( 1997 ) . k. husimi , _ proc . _ , 264 ( 1940 ) .
the squeezed states are states of minimum uncertainty , but unlike the coherent states , in which the uncertainty in the position and the momentum are the same , these allow to reduce the uncertainty , either in the position or in the momentum , while maintaining the principle of uncertainty in its minimum . it seems that this property of the squeezed states would allow you to get the position eigenstates as a limit case of them , doing null the uncertainty in the position and infinite at the momentum . however , there are two equivalent ways to define the squeezed states , which lead to different expressions for the limit states . in this work , we analyze these two definitions of the squeezed states and show the advantages and disadvantages of using such definition to find the position eigenstates . with this idea in mind , but leaving aside the definitions of the squeezed states , we find an operator applied to the vacuum that gives us the position eigenstates . we also analyze some properties of the squeezed states , based on the new expressions obtained for the eigenstates of the position .
waveform / code design , as one of the major problems in radar signal processing @xcite , active sensing @xcite , and wireless communications @xcite , has attracted significant interests over the past several decades @xcite . in radar signal processing and active sensing applications , waveform design plays an essential role because `` excellent '' waveforms can ensure higher localization accuracy @xcite , enhanced resolution capability @xcite , and improved delay - doppler ambiguity of the potential target @xcite . moreover , designing waveforms with robustness or adaptiveness is also required for the scenarios with harsh environments that include heterogeneous clutter and/or active jammers @xcite . in addition , with the advance of multiple - input multiple - output ( mimo ) radar @xcite , the problem of joint multiple waveform design is gaining even more importance and tends to grow to large scale . in order to obtain waveforms with desired characteristics , existing approaches usually resort to manipulations with correlation properties , such as the auto- and cross - correlations between different time lags of waveforms , which serve as the determinant factors for evaluating the quality of designed waveforms @xcite . perfect auto- and cross - correlation properties indicate that the emitted waveforms are mutually uncorrelated to any time - delayed replica of them , meaning that the target located at the range bin of interest can be easily extracted after matched filtering , and the sidelobes from other range bins are unable to attenuate it . for example , in the applications such as the spot and barrage noise jamming suppression @xcite and synthetic aperture radar imaging @xcite , waveforms with deep notches towards the time lags ( or equivalently , frequency bands ) , where the jamming or clutter signals are located , are highly desired . on the other hand , it is preferred from the hardware perspective that the waveforms maintain the constant - modulus property , which can reduce the cost of developing advanced amplifiers . there is a number of existing waveform design methods based on consideration of the correlation properties @xcite . the integrated sidelobe level ( isl ) , which serves as an evaluation metric for correlation levels of waveforms , or equivalently , the accumulated sidelobes at all time lags , is typically used . if the receiver is fixed to be the matched filter , the waveform design methods are focused on the waveform quality itself . corresponding waveform designs use the fact that the matched filter can be implemented in terms of the correlation between the waveform and its delayed replica . for example , the method of @xcite has proposed to design unimodular waveform in frequency domain using a cyclic procedure based on iterative calculations . a surrogate objective which is minimized by a cyclic algorithm has been introduced , and the methods associated with the isl and weighted isl ( wisl ) minimization therein have been named as can and wecan , respectively . these methods have been later extended to multiple waveform design in @xcite . if the receiver is not fixed and therefore has to be jointly optimized with the transmitted waveforms , the focus typically shifts to the so - called mismatched filter ( also called instrumental variable filter @xcite ) design at the receiver . such designs add flexibility as they enable to consider constraints which are difficult to address otherwise . the receive filter is generally mismatched because it trades off the signal - to - noise ratio in order to improve the signal - to - interference - plus - noise ratio . the corresponding design techniques are typically based on alternating optimization where minimum variance distortionless response ( mvdr ) filter design is involved . given the waveforms , finding the optimal mvdr receive filter is typically a computationally simpler problem than the waveform design itself . therefore , our focus here is the development of computationally efficient algorithms for addressing the core problem of waveform design when the optimal receive filter is the matched filter . the computational complexity of algorithms is of crucial importance for the isl and wisl minimization - based unimodular waveform design problems . indeed , the corresponding optimization problems can quickly grow to large scale with increasing the code length and number of waveforms . however , such problems are non - convex , while classical large - scale optimization approaches are developed for convex problems with relatively simple objective functions and constraints @xcite . the isl and wisl objective functions as well as the unimodular constraint to the desired waveforms are in fact complex to deal with and the required accuracy of waveform design is high . the aforementioned can and wecan @xcite use a cyclic procedure based on iterative calculations . although large code length up to several thousands is allowed by can and wecan , the cost in terms of time for these algorithms can reach several hours or even days when the code length and required number of waveforms grow large . this is a significant limitation that restricts the design of waveforms in real time . in large - scale optimization , the targeted computational complexity per iteration of an algorithm is linear in dimension of the problem or at most quadratic @xcite . to reduce the computational complexity to a reasonable one , many relevant works resort to the majorization - minimization ( mami ) technique @xcite , which is the basic technique for addressing large - scale and/or non - convex optimization problems with complex objectives @xcite . for example , @xcite have dealt with multistatic radar waveform design , where an information - theoretic criterion has been utilized , while @xcite have been concerned with single- and @xcite multiple - waveform designs . in addition to the computational complexity , another important characteristic of large - scale optimization algorithms is the convergence speed / rate @xcite . although the analytic bounds on the convergence rate may be hard / impossible to derive even for some existing large - scale convex optimization algorithms , the design of algorithms with provably faster convergence speed to tolerance than that of the other algorithms is possible even for non - convex problems considered here . in this paper , we focus on the isl and wisl minimization - based unimodular waveform designs for the matched filter receiver , aiming at developing fast algorithms that reduce the computational complexity and have faster convergence speed than the existing algorithms . the paper is based on a more detailed study of inherent algebraic structures of the objective functions , and concerning mami , also designing better majorization functions . the principal goal is to enable the real time waveform design even when the code length and number of waveforms are large . although our work also employs the mami approach , it differs from the previous works in many ways . different from @xcite , we formulate the isl minimization - based unimodular waveform design problem as a non - convex quartic problem by transforming the objective into frequency domain and rewriting it as a norm - based objective . moreover , we find out and use inherent algebraic structures in wisl expression that enable us to derive the corresponding quartic form into an alternative quartic form which in turns allows to apply the quartic - quadratic transformation . this equivalent form is based on eigenvalue decomposition , which we prove to be unnecessary to compute in our corresponding algorithm . then the isl and wisl minimization problems in the form of non - convex quartic optimization are simplified into quadratic forms . it allows us to utilize the mami technique where the majorization functions also differ from those of @xcite and @xcite . our algorithms have lower or comparable computational burden , faster convergence speed , and demonstrate better correlation properties than the existing state - of - the - art algorithms . the paper is organized as follows . in section [ sec : sigmodel ] , the signal model and the isl and wisl minimization - based unimodular waveform design problems are presented . in section [ sec : optviamm ] , new algorithms for the isl and wisl minimization problems are detailed . simulation results are presented in section [ sec : simulation ] , while the paper is concluded in section [ sec : conclu ] . _ notations _ : we use bold uppercase , bold lowercase , and italic letters to denote matrices , column vectors , and scalars , respectively . notations @xmath0 , @xmath1 , and @xmath2 are used for euclidean norm of a vector , frobenius norm of a matrix , and absolute value , respectively . similarly , @xmath3 , @xmath4 , and @xmath5 stand for conjugate , transpose , and conjugate transpose operations , respectively , while @xmath6 , @xmath7 , and @xmath8 respectively denote column - wise vectorization of a matrix , largest eigenvalue of a matrix , and maximization operations . notations @xmath9 and @xmath10 stand respectively for the floor function and modulo operation with the first argument being the dividend , while @xmath11 denotes the operation of constructing a hermitian toeplitz matrix from a vector that coincides with the first column of a matrix and @xmath12 is the operator that picks up diagonal elements from a matrix and writes them into a vector ( for matrix argument ) or forms a diagonal matrix with main diagonal entries picked up from a vector ( for vector argument ) . in addition , @xmath13 stands for the matrix trace , @xmath14 stands for the real part of a complex value , @xmath15_{i , j } $ ] denotes the @xmath16th element of a matrix , @xmath17 and @xmath18 respectively denote kronecker and hadamard product operations , @xmath19 is the @xmath20 identity matrix , and @xmath21 denotes an @xmath22 vector with all elements equal to 1 . consider a radar ( or communication ) system which emits @xmath23 unimodular and mutually orthogonal waveforms . each waveform is of code length @xmath24 . then the whole waveform matrix @xmath25 of size @xmath26 is defined as @xmath27 $ ] . here the @xmath28th column @xmath29 corresponds to the @xmath28th launched waveform . let the @xmath30th element of @xmath31 be @xmath32 where @xmath33 is an arbitrary phase value ranging between @xmath34 and @xmath35 . when the number of waveforms @xmath23 reduces to one , the waveform matrix @xmath25 shrinks to a column vector . the isl for the set of waveforms @xmath36 can be expressed as @xcite @xmath37 where @xmath38 is the cross - correlation between the @xmath28th and @xmath39th waveforms at the @xmath30th time lag . the first term on the right - hand side of is associated with the auto - correlations , while the second term represents the cross - correlations of the waveforms . likewise , the wisl for the waveforms @xmath36 can be expressed as @xcite @xmath40 where @xmath41 are real - valued symmetric weights , i.e. , @xmath42 , used for controlling the sidelobe levels corresponding to different time lags . if @xmath43 takes zero value , it means that the sidelobe level associated with the @xmath30th time lag is not considered . if all the controlling weights @xmath41 take the value @xmath44 , then @xmath45 in coincides with @xmath46 in . the basic unimodular waveform design problem is then formulated as the synthesize of unimodular and mutually orthogonal waveforms @xmath36 which have as good as possible auto- and cross - correlation or weighted correlation properties . using , the wisl minimization - based unimodular waveform design problem can be formally expressed as @xmath47 where the constraints ensure the modularity of waveforms , while the orthogonality between waveforms is guaranteed by the objective . obviously , if all the controlling weights @xmath41 take the value @xmath44 , the problem becomes the isl minimization - based unimodular waveform design problem . in this section , we develop fast algorithms for the isl and wisl minimization - based unimodular waveform designs . the algorithms make use of the mami technique and exploit inherent algebraic structures in the objective function , which allows to reduce the computational complexity . the isl @xmath46 in can be rewritten in the matrix form as @xmath48 where @xmath49 is the following @xmath50 waveform correlation matrix @xmath51 \label{eq : rp}\end{aligned}\ ] ] and @xmath52 is the kronecker delta function . transforming into frequency domain and performing some derivations , the isl @xmath46 can be expressed as @xcite @xmath53 where @xmath54 with @xmath55 being the @xmath56th row of the waveform matrix @xmath25 , i.e. , @xmath57^ { \mathrm t } $ ] . expanding the norm in , after some elementary algebraic computations , the isl @xmath46 can be rewritten as @xmath58 moreover , introducing the @xmath59 vectorized version of the waveform matrix @xmath60 as @xmath61^ { \mathrm t}$ ] and the @xmath62 matrix @xmath63 with @xmath64 defined as @xmath65^ { \mathrm t}$ ] where @xmath66 , and using the facts that @xmath67 and @xmath68 , the isl expression can be further rewritten as @xmath69 noticing that @xmath70 and using the fact that the desired waveforms are orthogonal and have constant modulus , i.e. , @xmath71 , we can find that @xmath72 using and excluding the immaterial optimization terms from , the optimization problem can be rewritten as @xmath73 where the objective function takes a quartic form with respect to @xmath74 . introducing the @xmath75 and @xmath76 , respectively , matrices @xmath77 and @xmath78 and using the property that @xmath79 , which follows from the elementary properties of the trace and vectorization operations , the objective function in can be transformed from quartic into quadratic form as follows @xmath80 therefore , the problem can be further rewritten as [ eq : optcanii ] @xmath81 since the objective function takes a quadratic form , a proper majorized function can be applied . before applying the majorant to , we present the following general result that will be used later . [ th : th1 ] if a real - valued function @xmath82 with complex variable @xmath83 is second - order differentiable , and there is a matrix @xmath84 satisfying the generalized inequality @xmath85 for all @xmath83 , then for each point @xmath86 , the following convex quadratic function @xmath87 majorizes @xmath82 at @xmath86 . appears in @xcite as theorem 3.1 . ] using taylor s theorem , the second - order expansion of @xmath82 at the point @xmath86 is given as @xmath88 where @xmath89 is a point on the line connecting @xmath90 and @xmath83 . due to the fact that @xmath91 , the inequality @xmath92 also holds true , where @xmath93 is given by . if @xmath94 is a quadratic form , i.e. , @xmath95 , as it is the case for the objective function in , by substituting @xmath96 in , the majorant can be obtained as @xmath97 let @xmath98 be the @xmath76 identity matrix magnified by the largest eigenvalue of the matrix @xmath99 , i.e. , @xmath100 . for such selection of @xmath98 , the generalized inequality @xmath101 is guaranteed to hold . then using , the function can be majorized by the following function @xmath102 where the matrix @xmath103 is obtained at the @xmath104th iteration with @xmath105 being the vectorized version of the waveform matrix @xmath106 at iteration @xmath107 . using the elementary properties of the kronecker product and vectorization operations , we can find that @xmath108 furthermore , using and the fact that the desired waveforms are orthogonal and unimodular , we obtain @xmath109 moreover , using the definition of the matrix @xmath99 , the maximum eigenvalue of @xmath99 can be found as @xmath110 returning to and using the facts , we can see that the first two terms on the right hand of are constant and therefore immaterial for optimization . thus , ignoring these two terms , the majorization problem for can be written as @xmath111 using the definition and the properties and also @xmath112 the objective function in , denoted hereafter as @xmath113 , can be expanded as @xmath114 applying the mixed - product property of the kronecker product to the right - hand side of the expansion , the objective in can be further derived as @xmath115 it is straightforward to check that the equality @xmath116 holds . applying this equality to , the objective in can be rewritten as @xmath117 where the @xmath118 matrix @xmath119 and the @xmath120 matrix @xmath121 are defined as @xmath122 $ ] and @xmath123 , and the @xmath124 vector @xmath125 is defined as is applied to a matrix argument , which means that the magnitude is found for each element of the matrix , that is , the element - wise magnitude . ] @xmath126 via the @xmath127 matrix @xmath128 $ ] . using , the problem can be rewritten as @xmath129 where the objective function takes a quadratic form , to which the majorant can be applied again . substituting the @xmath120 matrix @xmath98 , defined as @xmath130 , into , we find that can be majorized by the following function @xmath131 where @xmath132 is the largest element of @xmath133 , equivalently , @xmath134 . this scaling factor guaranties that the generalized inequality @xmath135 holds . noticing that @xmath136 and using the fact that the desired waveforms are orthogonal and have constant modulus , i.e. , @xmath137 , we can see that the first two terms in are constant , and hence , immaterial for optimization . ignoring these terms , the optimization problem can be further majorized by the following problem @xmath138 using again the fact that the desired waveforms have constant modulus , the problem can be equivalently rewritten as @xmath139_{m , p } { \aftergroup\egroup\originalright}| = 1 , \ ; m = 1 , \ldots , m ; \ , p = 1 , \ldots , p \label{eq : optcaneqv}\end{aligned}\ ] ] where the @xmath140 matrix @xmath141 is a hermitian toeplitz matrix constructed from the @xmath142 vector @xmath143 . the problem has the following closed - form solution @xmath144_{m , p } = \exp { \mathopen{}\mathclose\bgroup\originalleft}\ { j \cdot \mathrm { arg } { \mathopen{}\mathclose\bgroup\originalleft } ( { \mathopen{}\mathclose\bgroup\originalleft } [ \mathbf { t } ^ { ( k ) } \mathbf{y}^{(k ) } { \aftergroup\egroup\originalright}]_{m , p } { \aftergroup\egroup\originalright } ) { \aftergroup\egroup\originalright}\ } , \ ; \forall m , \ , \forall p. \label{eq : solcan}\end{aligned}\ ] ] = < .75 < .25 @xmath145 , @xmath146 unimodular sequence matrix with random phases . = < .75 < .25 @xmath147 = < .75 < .25 @xmath148 = < .75 < .25 @xmath149 = < .75 < .25 @xmath150_{m , p } = e^ { j \cdot \mathrm { arg } { \mathopen{}\mathclose\bgroup\originalleft } ( { \mathopen{}\mathclose\bgroup\originalleft } [ \mathbf { t } ^ { ( k ) } \mathbf{y}^{(k ) } { \aftergroup\egroup\originalright}]_{m , p } { \aftergroup\egroup\originalright } ) } , \quad \forall m , \ , \forall p \end{aligned}$ ] = < .75 < .25 @xmath151 convergence finally , according to the mami procedure and using the closed - form solution to the majorization problem , the isl minimization - based unimodular waveform design algorithm is summarized in algorithm [ canmm ] . there exist accelerated schemes for mami , such as the squared iterative method ( squarem ) of @xcite , which can be straightforwardly applied to speed up algorithm [ canmm ] . the squarem scheme is an extension of the scalar steffensen type method @xcite , @xcite to vector fixed - point iteration empowered with the idea of `` squaring '' @xcite . it is an `` off - the - shelf '' acceleration method that requires nothing extra to the parameter updating rules of an original algorithm , except possibly the computationally cheap projection to feasibility set , and it is guaranteed to converge @xcite . different stopping criteria can be employed in algorithm [ canmm ] . for example , it can be the absolute isl difference between the current and previous iterations normalized by the initial isl , or it can be the norm of the difference between the waveform matrices obtained at the current and previous iterations . in terms of the per iteration computational complexity of algorithm [ canmm ] , the straightforward calculation of @xmath125 according to requires @xmath152 operations , the calculation of @xmath153 costs @xmath154 operations , while the computational burden of the matrix to matrix product @xmath155 in is @xmath156 operations . therefore , the total computational complexity is @xmath157 operations . however , @xmath125 and @xmath158 can be computed by means of the fast fourier transform ( fft ) at the order of complexity @xmath159 and @xmath160 , respectively . similarly , using the toeplitz structure of @xmath161 , the product @xmath155 can also be calculated at a reduced complexity @xmath162 , which is the highest in algorithm [ canmm ] . thus , the order of complexity of algorithm [ canmm ] is @xmath162 , which is nearly linear in the dimension of the problem , as required in large - scale optimization . the wisl in can be written in a matrix form as @xmath163 where @xmath164 are defined in . in the frequency domain , can be expressed as @xcite @xmath165 where @xmath166 is defined in and @xmath167 is the weighted spectral density matrix . let us also define the @xmath140 toeplitz matrix constructed by the weights @xmath41 as follows @xmath168 . \label{eq : gamatrix}\end{aligned}\ ] ] then the matrix @xmath169 in can be rewritten in the vector - matrix form as @xmath170 substituting into , we arrive to the following wisl expression @xmath171 expanding the squared norm in the sum of yields @xmath172 using the facts that the desired waveforms are orthogonal and unimodular , i.e. , @xmath173 , and also that @xmath174 , we find that @xmath175 therefore , the second and third terms of are constant and immaterial for optimization . with this observation , the wisl minimization problem can be rewritten as [ eq : wislfreopt ] @xmath176 the hadamard product of two matrices appears under the frobenius norm in , and the resulting matrix there is complex . as a result , we can not arrive to a proper quartic form with respect to @xmath74 by directly expanding the squared norm of . instead , we need to convert it into a proper one . towards this end , let us consider the eigenvalue decomposition of @xmath177 , which in general may be indefinite and can be expressed as @xmath178 where @xmath179 and @xmath180 are the @xmath107th eigenvalue and eigenvector , respectively , @xmath181 , @xmath182 equals @xmath183 when @xmath184 is negative , otherwise it is the same as @xmath185 , and @xmath186 is the rank of @xmath177 . substituting into and expanding the frobenius norm , the objective function , called hereafter as @xmath187 , can be rewritten as @xmath188 applying the property @xmath189 ( also holds when @xmath185 is replaced by @xmath190 ) to together with the mixed - product property of the kronecker product , the objective function can be rewritten as @xmath191 where the @xmath75 hermitian matrices @xmath192 and @xmath193 are defined as @xmath194 and @xmath195 . substituting to , the wisl minimization problem becomes [ eq : optwecaniconstr2nd ] @xmath196 [ eq : optwecani ] the objective function takes a proper quartic form with respect to @xmath74 that enables us to design an algorithm based on the mami approach . by means of the trace and vectorization operations for matrices , and similar to the previous subsection , we can transform , denoted for brevity as @xmath197 , into the following form @xmath198 where @xmath199 has been defined before , @xmath200 is the @xmath201 matrix defined as @xmath202 with @xmath203 replacing the objective function with , the optimization problem can be rewritten as [ eq : optwecanii ] @xmath204 where takes a quadratic form , to which a majorant can be applied . yet before applying the majorization procedure , we present the following result that will be used later . [ lemmaii ] given a set of @xmath205 arbitrary complex vectors @xmath206 and an @xmath207 arbitrary hermitian matrix @xmath208 , the following generalized inequality @xmath209 holds , where @xmath210 . let @xmath211 and @xmath212 be respectively the sets of eigenvalues ( in descending order ) and corresponding eigenvectors of the matrix @xmath213 , i.e. , @xmath214 . using this expression and elementary properties of the hadamard product , the inequality can be derived as @xmath215 the proof is complete . applying lemma [ lemmaii ] by taking @xmath216 , @xmath217 , and @xmath218 , we obtain the following inequality @xmath219 note that for a given matrix @xmath220 in , the largest eigenvalue of @xmath221 in , i.e. , @xmath222 , is fixed , and it can be found that @xmath223 . moreover , the diagonal elements of @xmath224 take values either zero or @xmath225 . therefore , we can replace the matrix @xmath226 in with an identity matrix magnified by @xmath225 without disobeying the inequality . using with @xmath227 @xmath228 that satisfies @xmath229 , the objective function can be majorized by the following function @xmath230 due to the property , the first and second terms in are constant and therefore immaterial for optimization . ignoring these terms , can be majorized by the problem [ eq : optwecaniii ] @xmath231 to further simplify , we will need the following result that relates hadamard and kronecker products . [ lemmaiii ] given two matrices @xmath232 and @xmath233 of the same size @xmath207 and the @xmath234 selection matrix @xmath235 $ ] with @xmath236 being the @xmath237th @xmath207 block matrix composed of all zeros except the @xmath56th element on the main diagonal equalling one , i.e. , @xmath238_{n , n } = 1 $ ] , the following equality @xmath239 holds . under the condition that @xmath240 is an integer , @xmath236 can be decomposed as @xmath241 where the matrices @xmath242 and @xmath243 are constructed in the same way as @xmath244 but have the reduced size @xmath245 , and @xmath246 are respectively the column and row indices of the element in the @xmath247 matrix with linear ( column - wise ) index @xmath56 . the proof of appears in lemma 1 of @xcite . the remaining results are the elementary properties of the selection matrix . applying lemma [ lemmaiii ] by taking @xmath248 , @xmath249 , and @xmath250 , and substituting into , the objective function , denoted for brevity as @xmath251 , can be rewritten as @xmath252_{n , n ' } \mathbf { \bar{e } } _ n \boldsymbol{\bar { \phi } } \mathbf { \bar{e}}_{n'}^{\mathrm{h } } { \aftergroup\egroup\originalright } ) \mathrm { vec } \big ( \mathbf { \tilde{y}}^{(k ) } \big ) \nonumber \\ & \quad\ ; - \tfrac{\lambda_{\boldsymbol{\tilde{\phi}}}}{2 } \big ( \mathrm{vec } \big ( \mathbf{\tilde{y } } \big ) \big)^{\mathrm{h } } \mathrm{vec } \big ( \mathbf { \tilde{y}}^{(k ) } \big ) \label{eq : longeqii}\end{aligned}\ ] ] where the latter expression in is obtained by expanding the kronecker product in the prior expression for the objective . using and , and applying the properties and , the objective can be further rewritten as @xmath253 _ { n , n ' } \mathbf { y } ^ { \mathrm{h } } \big ( \mathbf { y } ^ { \mathrm { t } } \otimes \mathbf { i } _ { m p } \big ) \big ( \mathbf { \hat{e } } _ { u{\mathopen{}\mathclose\bgroup\originalleft } ( n { \aftergroup\egroup\originalright } ) } \nonumber \\ & \otimes \mathbf { \hat{e } } _ { v { \mathopen{}\mathclose\bgroup\originalleft } ( n { \aftergroup\egroup\originalright } ) } \big ) \big ( \mathbf { a } _ p^ { \mathrm { t } } \otimes \mathbf { i } _ { mp } \big)^ { \mathrm { h } } \mathrm{vec } \big ( \mathbf { a } _ p \big ) { \mathopen{}\mathclose\bgroup\originalleft } ( \mathrm{vec } \big ( \mathbf { a } _ p \big ) { \aftergroup\egroup\originalright})^{\mathrm{h } } \big ( \mathbf { a } _ p^ { \mathrm { t } } \nonumber \\ & \otimes \mathbf { i } _ { mp } \big ) \big ( \mathbf { \hat{e } } _ { u{\mathopen{}\mathclose\bgroup\originalleft } ( n ' { \aftergroup\egroup\originalright } ) } \otimes \mathbf { \hat{e } } _ { v { \mathopen{}\mathclose\bgroup\originalleft } ( n ' { \aftergroup\egroup\originalright } ) } \big)^ { \mathrm { h } } \big ( ( \mathbf { y } ^{(k ) } ) ^ { \mathrm { t } } \otimes \mathbf { i } _ { m p } \big)^ { \mathrm { h } } \mathbf { y } ^{(k ) } \nonumber \\ & - \tfrac{\lambda_{\boldsymbol{\tilde{\phi}}}}{2 } \mathbf{y}^{\mathrm h } \big ( \mathbf{y}^ { \mathrm { t } } \otimes \mathbf{i}_{mp } \big ) \big ( ( \mathbf{y}^{(k ) } ) ^ { \mathrm { t } } \otimes \mathbf{i}_{mp } \big)^ { \mathrm { h } } \mathbf{y}^{(k)}. \label{eq : longeqiii}\end{aligned}\ ] ] applying the mixed - product property of the kronecker product together with the property @xmath254 to , we obtain @xmath255 _ { n , n ' } \mathbf { \hat{e } } _ { v{\mathopen{}\mathclose\bgroup\originalleft } ( n { \aftergroup\egroup\originalright } ) } \mathbf { a } _ p { \mathopen{}\mathclose\bgroup\originalleft } ( \big ( \mathbf { y } ^{(k ) } \big)^ { \mathrm { h } } \mathbf { \hat{e } } _ { u{\mathopen{}\mathclose\bgroup\originalleft } ( n ' { \aftergroup\egroup\originalright } ) } { \aftergroup\egroup\originalright}. { \aftergroup\egroup\originalright}. \nonumber \\ & \quad \times \!\ ! { \mathopen{}\mathclose\bgroup\originalleft}. \vphantom{\sum_{l=1}^{m^2 p^2 } } { \mathopen{}\mathclose\bgroup\originalleft}. \mathbf { a } _ p \mathbf { a } _ p^ { \mathrm { h } } \mathbf { \hat{e } } _ { v{\mathopen{}\mathclose\bgroup\originalleft } ( n ' { \aftergroup\egroup\originalright } ) } \mathbf { y } ^{(k ) } { \aftergroup\egroup\originalright } ) \mathbf { a } _ p^ { \mathrm { h } } \mathbf { \hat{e } } _ { u{\mathopen{}\mathclose\bgroup\originalleft } ( n { \aftergroup\egroup\originalright } ) } \ ! - \tfrac{\lambda_{\boldsymbol { \tilde { \phi } } } } { 2 } \mathbf { y } ^ { ( k ) } ( \mathbf { y } ^ { ( k ) } ) ^ { \mathrm { h } } \big ) \!\ ! { \aftergroup\egroup\originalright } ) \mathbf { y } \nonumber \\ & = \mathbf { y } ^ { \mathrm{h } } { \mathopen{}\mathclose\bgroup\originalleft } ( \mathbf{b}^{(k ) } - \tfrac{\lambda_{\boldsymbol { \tilde { \phi } } } } { 2 } \mathbf { y } ^ { ( k ) } ( \mathbf { y } ^ { ( k ) } ) ^ { \mathrm { h } } { \aftergroup\egroup\originalright } ) \mathbf { y } \label{eq : optwecanivobj}\end{aligned}\ ] ] where @xmath256 and @xmath257 is an @xmath75 hermitian matrix composed of @xmath258 block matrices , i.e. , @xmath259 \label{eq : bkmatrix}\end{aligned}\ ] ] with the @xmath260th block @xmath261 being a @xmath262 toeplitz matrix whose first row and column coincide with the @xmath142 vectors @xmath263 and @xmath264 , respectively . here , the @xmath265th ( @xmath266 ) elements of @xmath267 and @xmath268 are respectively given by @xmath269 where @xmath270 , @xmath271 is the set of non - negative indices associated with the non - zero isl controlling weights ( always including zero index for the sake of simplicity ) , and @xmath272 is the complementary set of @xmath273 with the full set defined as @xmath274 $ ] . the meanings of and are that the non - zero elements of @xmath267 and @xmath275 are expressed by the sum of the off diagonal elements in the upper and lower triangular parts of @xmath276 magnified by @xmath277 , respectively . using and , we can avoid calculations for the zero elements . note that @xmath278 , therefore , only the upper ( or lower ) triangular part of @xmath279 needs to be determined . the objective function takes a quadratic form , to which the majorant of can be applied again . let @xmath280 , so that the generalized inequality @xmath281 is guaranteed for . here we can use any matrix norm of @xmath282 for @xmath283 because any matrix norm serves as an upper bound of the largest eigenvalue . thus , the objective function can be majorized by the following function @xmath284 similar to the majorant , the first two terms of are constant and therefore immaterial for optimization . ignoring these two terms , the problem can be majorized by @xmath285 due to the constant modulus property of @xmath74 , the problem is equivalent to the following optimization problem @xmath286 where @xmath287 . the problem can be then solved in closed form as @xmath288 finally , reshaping the so - obtained vector @xmath74 into a @xmath289 matrix , we obtain the designed waveform matrix @xmath60 . the wisl minimization based unimodular waveform design algorithm is summarized in algorithm [ wecanmm ] . = < .75 < .25 @xmath145 , @xmath290 unimodular sequence with random phases . = < .75 < .25 @xmath291 = < .75 < .25 @xmath292 = < .75 < .25 @xmath293 = < .75 < .25 construct @xmath294 via = < .75 < .25 @xmath295 = < .75 < .25 @xmath296 = < .75 < .25 @xmath297 = < .75 < .25 @xmath151 convergence to find the computational complexity of algorithm [ wecanmm ] , we assume that the set of @xmath273 consists of @xmath298 elements . it can be seen that both @xmath263 in and @xmath268 in can be calculated with at most @xmath299 operations if @xmath276 is given . the calculation of the covariance matrix @xmath276 costs @xmath300 operations . note that we only need to do calculations for the subscripts @xmath301 and @xmath302 , and then repeat the above summarized calculations @xmath303 times . finally , the calculation of the vector @xmath304 needs @xmath305 operations . consequently , the total number of operations is upper bounded by @xmath306 . in other words , the computational complexity of algorithm [ wecanmm ] is at most @xmath307 that is smaller than quadratic in the problem size , and therefore suitable for large - scale optimization .. ] the accelerated version of algorithm [ wecanmm ] is obtained by a straightforward application of the squarem acceleration scheme @xcite as in the case of algorithm [ canmm ] . we evaluate here the performance of the proposed isl and wisl minimization - based waveform design algorithms ( algorithms [ canmm ] and [ wecanmm ] ) by comparing them with existing isl and wisl minimization based algorithms . to be specific , our algorithm [ canmm ] for isl minimization ( named hereafter as islnew ) is compared with the can of @xcite and the ( third ) algorithm in @xcite ( named hereafter as islsong ) , while our algorithm [ wecanmm ] for wisl minimization ( named hereafter as wislnew ) is compared with the wecan of @xcite and the ( second ) algorithm in @xcite ( named hereafter as wislsong ) . the accelerated versions of the mami - based algorithms , including the islnew , wislnew , islsong , and wislsong algorithms , are also tested , where the squarem scheme @xcite is used for mami acceleration . we generate sets of unimodular sequences with random phases as the initialization for each algorithm tested , and apply the same set of sequences to all algorithms for the purpose of fair comparison . all comparisons are conducted based on the same hardware and software platforms . throughout our simulations , two stopping criteria are employed : ( i ) the absolute isl or wisl difference between the current and previous iterations normalized by the initial isl or wisl , whose tolerance is set to be @xmath308 ; and ( ii ) the norm difference between the waveform matrices ( or vectors ) obtained at the current and previous iterations , whose tolerance is set to be @xmath309 . the isl and wisl values in dbs are defined as @xmath310 and @xmath311 , respectively . in the first example , we study the convergence properties of the waveform design algorithms ( can , islsong , islnew , accelerated islsong , and accelerated islnew ) in terms of the number of conducted iterations for a problem of relatively small size . specifically , a single waveform ( @xmath312 ) of the code length @xmath313 is designed in this example . in figs . [ fig : islvsite ] and [ fig : islvsite ] , the isl performance versus the number of conducted iterations is displayed for the aforementioned algorithms , where the stopping criteria ( i ) and ( ii ) are used , respectively . the isl values for different algorithms obtained at each iteration are normalized by the isl value associated with the initial set of sequences . it can be seen from figs . [ fig : islvsite ] and [ fig : islvsite ] that for all the algorithms tested , the isl decreases monotonically as the number of iterations increases . among the isl minimization - based algorithms tested , the accelerated islnew algorithm shows the best convergence speed , i.e. , it requires the smallest number of iterations to converge to a solution that satisfies the pre - set tolerance parameter for both stopping criteria used . the accelerated islsong algorithm shows the second best convergence speed . this demonstrates the superiority of applying accelerated mami techniques to the isl minimization - based waveform design . the proposed islnew algorithm without acceleration shows a little slower convergence speed , but achieves around @xmath314 db better isl than that of the can algorithm . the islsong algorithm without acceleration shows the worst convergence speed among all the algorithms tested . the same convergence behavior can also be seen in figs . [ fig : islvsite ] and [ fig : islvsite ] independent on the stopping criteria used . max width= = 7pt [ tab : isl_t1 ] [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] min . : obtained minimum wisl value ( in db ) . ave . : obtained average wisl value ( in db ) . number of conducted iterations . in the fourth example , we compare the performance of the wisl minimization - based algorithms ( wecan , accelerated wislsong , and accelerated wislnew ) in terms of the following characteristics : the minimum and average wisl after convergence ( in dbs ) , the average consumed time ( in seconds ) , and the average number of conducted iterations . the number of waveforms and the code lengths are taken the same as in the second example , and all results are averaged over @xmath315 independent trials . moreover , the isl controlling weights are the same as in the previous example . table [ tab : wisl_t1 ] shows the results when the stopping criterion ( i ) is used , while table [ tab : wisl_t2 ] shows the results for the stopping criterion ( ii ) . it can be seen from table [ tab : wisl_t1 ] that the accelerated wislnew algorithm outperforms the other two algorithms tested in this example for all code lengths . the accelerated wislsong algorithm shows the second best performance in terms of all evaluation characteristics , and the wecan algorithm performs the worst . among all code lengths , the smallest average wisl values after convergence by the wecan , accelerated wislsong , and accelerated wislnew algorithms are respectively @xmath316 db , @xmath317 db , and @xmath318 db , while the largest average wisl values are @xmath319 db , @xmath320 db , and @xmath321 db ( all at @xmath322 ) , respectively . the wecan algorithm generally consumes significantly more time , requires more iterations , and achieves higher wisl values than the other two algorithms . it is manly because it does not deal with the original wisl objective function but instead a surrogate one . the advantages of the wislnew and wislsong algorithms over the wecan algorithm become a lot more obvious when the code length @xmath323 is larger than @xmath324 , which verifies the fact that the wecan algorithm is suitable only for the wisl minimization - based waveform design with short code length . the wecan algorithm may converge very slowly when the set of isl controlling weights is not sparse . focusing on the comparisons between the accelerated wislsong and accelerated wislnew algorithms , we can see from table [ tab : wisl_t1 ] that the accelerated wislnew algorithm is superior to the accelerated wislsong algorithm when the code length @xmath323 is larger than @xmath324 , and the biggest differences ( occurring at @xmath325 ) of the minimum and average wisl values between these two algorithms reach @xmath326 db and @xmath327 db , respectively . the accelerated wislnew algorithm always consumes less time and requires smaller number of iterations than the accelerated wislsong algorithm . the larger the code length @xmath323 is , the more obvious the superiority of the accelerated wislnew algorithm becomes . for example , the ratio of the consumed time between these two algorithms decreases from about @xmath328 ( that is , the accelerated wislnew algorithm requires only @xmath328 time required by the accelerated wislsong algorithm ) to @xmath329 , and the corresponding ratio of the number of conducted iterations for these two algorithms decreases from about @xmath330 to @xmath331 as the code length increases from @xmath332 to @xmath325 . thus , the proposed wislnew algorithm is better suited for large - scale waveform design problems . in addition , for the waveform design with smaller code length and larger number of non - zero isl controlling weights ( corresponding to the @xmath333 case in this example ) , the minimum and average wisl values after convergence for the wislnew and wislsong algorithms are close to each other , and are slightly better than those for the wecan algorithm . however , the accelerated wislnew algorithm is still superior in terms of the other characteristics . the above discussed advantages of the proposed wislnew algorithm over the wecan and wislsong algorithms are also verified by table [ tab : wisl_t2 ] where the experiment is conducted under the stopping criterion ( ii ) . it can be seen that the data therein follow the same trends as in table [ tab : wisl_t1 ] . to be explicit , the accelerated wislnew algorithm takes around @xmath334 times less time and @xmath335 times less number of iterations compared to the accelerated wislsong algorithm , and it achieves significantly lower minimum and average wisl values after convergence , especially for large code length . on the contrary , the wecan algorithm always demonstrates the worst performance in terms of all evaluation characteristics . finally in the fifth example , we present the auto- and cross - correlations of @xmath336 waveforms designed by the accelerated wislsong and accelerated wislnew algorithms with code length @xmath337 . we set the isl controlling weights to the same values as those in the previous two examples . the stopping criterion ( i ) is utilized in this example . the four sub figures in fig . [ fig : correvaluation ] stand for the auto- and cross - correlations of the two sets of waveforms generated by the two aforementioned algorithms . here the correlation levels , defined as @xmath338 , are shown ( in dbs ) . to better display the results , we only show the auto- and cross - correlations for the time lags within the range @xmath339 $ ] . the wecan algorithm for the large code length in this example costs significantly more time and shows the worst auto- and cross - correlations , and therefore , is not shown in fig . [ fig : correvaluation ] . it can be seen from fig . [ fig : correvaluation ] that the auto - correlations associated with the time lags @xmath340 \cup [ 1 , 19]$ ] and cross - correlations associated with the time lags @xmath341 $ ] for both generated sets of waveforms are well controlled , while the waveform correlations associated with other time lags are not controlled . therefore , the latter results in much higher correlation levels . under the condition of using the same tolerance parameter , the correlation levels corresponding to the time lags of interest obtained by the proposed wislnew algorithm are significantly better than those obtained by the wislsong algorithm . the largest gap between the obtained correlation levels by these two algorithms has reached about @xmath342 db . moreover , the proposed wislnew algorithm needs significantly shorter time than the wislsong algorithm as it has been discussed above . in this paper , we have developed two ( one based on isl and the other based on wisl minimization ) new fast algorithms for designing single / multiple unimodular waveforms / codes with good auto- and cross - correlation or weighted correlation properties . since the corresponding optimization problems are non - convex and may be large - scale , the proposed algorithms are based on the mami framework and utilize a number of newly found inherent algebraic structures in the objective functions of the corresponding optimization problems . these properties have enabled us to reduce the computational complexity of the algorithms to the level which is suitable for large - scale optimization and at least similar to or lower than that of the existing algorithms . moreover , the proposed algorithms also show faster convergence speed to tolerance and provide waveforms of better quality than those of the existing state - of - the - art algorithms . y. yang and r. s. blum , `` mimo radar waveform design based on mutual information and minimum mean - square error estimation , '' _ ieee trans . _ , vol . 43 , no . 1 , 330343 , jan . 2007 . a. de maio , s. d. nicola , y. huang , z .- q . luo , and s. zhang , `` design of phase codes for radar performance optimization with a similarity constraint , '' _ ieee trans . signal process . _ , vol . 57 , no . 2 , pp . 610621 , 2009 . h. hao , p. stoica , and j. li , `` designing unimodular sequences sets with good correlations including an application to mimo radar , '' _ ieee trans . signal process . _ , vol . 57 , no . 11 , pp . 43914405 , nov . a. de maio , y. huang , m. piezzo , s. zhang , and a. farina , `` design of optimized radar codes with a peak to average power ratio constraint , '' _ ieee trans . signal process . _ , vol . 59 , no . 6 , pp . 26832697 , jun . a. aubry , a. de maio , a. farina , and m. wicks , `` knowledge - aided ( potentially cognitive ) transmit signal and receive filter design in signal - dependent clutter , '' _ ieee trans . _ , vol . 49 , no . 1 , 93117 , jan . 2013 . q. he , r. s. blum , and a. m. haimovich , `` noncoherent mimo radar for location and velocity estimation : more antennas means better performance , '' _ ieee trans . signal process . _ , vol . 58 , no . 7 , pp . 36613680 , jul . 2010 . l. zhao , j. song , p. babu , and d. p. palomar , `` a unified framework for low autocorrelation sequence design via majorization - mimimization , '' _ ieee trans . signal process . _ , vol . 65 , no . 2 , pp . 438453 , jan . 2017 . m. m. naghsh , m. modarres - hashemi , s. shahbazpanahi , m. soltanalian , and p. stoica , `` unified optimization framework for multi - static radar code design using information - theoretic criterion , '' _ ieee trans . signal process . _ , vol . 61 , no . 21 , pp . 54015416 , nov . a. khabbazibasmenj , f. roemer , s. a. vorobyov , and m. haardt , `` sum - rate maximization in two - way af mimo relaying : polynomial time solutions to a class of dc programming problems , '' _ ieee trans . signal process . _ , vol . 60 , no . 10 , pp . 54785493 , oct . m. soltanalian , a. gharanjik , b. shankar , and b. ottersten . ( 2017 ) grab - n - pull : a max - min fractional quadratic programming framework with applications in signal processing . submitted to _ ieee trans . signal process . _ [ online ] . available : http://msol.people.uic.edu/papers/gnp_pre.pdf y. li , s. a. vorobyov , and z. he , `` design of multiple unimodular waveforms with low auto- and cross - correlations for radar via majorization - minimization , '' in _ proc . 24th european signal process . ( eusipco ) _ , budapest , hungary , aug.sep . 2016 , pp . 22352239 . y. li and s. a. vorobyov , `` efficient single / multiple unimodular waveform design with low weighted correlations , '' in _ proc . conf . acoust . , speech , , signal process . ( icassp ) _ , new orleans , usa , mar .
in this paper , we develop new fast and efficient algorithms for designing single / multiple unimodular waveforms / codes with good auto- and cross - correlation or weighted correlation properties , which are highly desired in radar and communication systems . the waveform design is based on the minimization of the integrated sidelobe level ( isl ) and weighted isl ( wisl ) of waveforms . as the corresponding optimization problems can quickly grow to large scale with increasing the code length and number of waveforms , the main issue turns to be the development of fast large - scale optimization techniques . the difficulty is also that the corresponding optimization problems are non - convex , but the required accuracy is high . therefore , we formulate the isl and wisl minimization problems as non - convex quartic optimization problems in frequency domain , and then simplify them into quadratic problems by utilizing the majorization - minimization technique , which is one of the basic techniques for addressing large - scale and/or non - convex optimization problems . while designing our fast algorithms , we find out and use inherent algebraic structures in the objective functions to rewrite them into quartic forms , and in the case of wisl minimization , to derive additionally an alternative quartic form which allows to apply the quartic - quadratic transformation . our algorithms are applicable to large - scale unimodular waveform design problems as they are proved to have lower or comparable computational burden ( analyzed theoretically ) and faster convergence speed ( confirmed by comprehensive simulations ) than the state - of - the - art algorithms . in addition , the waveforms designed by our algorithms demonstrate better correlation properties compared to their counterparts . correlation , majorization - minimization , mimo radar , waveform design .
yb - based intermetallic compounds are of interest due to their anomalous physical properties concerning to valence fluctuation and/or heavy - fermion behavior @xcite . yb ion can have two different valence state : divalent(yb@xmath3 ) and trivalent(yb@xmath5 ) states . the former has a closed 4@xmath6 shell and thereby is nonmagnetic , whereas the latter has one hole in the 4@xmath6 shell , leading to a total angular momentum of @xmath7 = 7/2 . valence - fluctuating yb - compounds are characterized by nearly trivalent yb ions at temperatures sufficiently higher than the valence - fluctuation temperature , @xmath8 . with decreasing temperature , valence admixture with the divalent state occurs , resulting in a nonmagnetic ground state with a large density of states . since divalent yb has a larger ionic radius than trivalent yb , lattice expansion of yb - based compounds can occur on cooling . the most prominent example may be ybincu@xmath9 , which shows a first - order valence - transition at @xmath10 = 40 k @xcite , at which the lattice abruptly expands more than 0.1% on cooling @xcite . if such volume expansion occurs moderately to cancel the normal lattice shrinkage with decreasing temperature , the total volume change can become zero . such zero thermal - expansion has recently been reported by salvador et al . in the intermetallic compound ybgage @xcite . it has been reported that the volume change is almost negligible in a wide temperature range from 300 k to 100 k @xcite . the authors have also presented magnetic susceptibility data that suggest a yb valence changes from yb@xmath5 at high temperatures ( @xmath11150 k ) to nearly divalent state at low temperatures @xcite . from these facts , they have concluded that the zero thermal - expansion is due to the valence change of yb . lately , however , controversial results have been reported for ybgage by several authors @xcite . margadonna et al . have reported the absence of zero thermal - expansion . they have reported that the lattice volume shrinks similarly to normal substances below 700 k @xcite . instead , these authors have found a sudden volume change at 5 k without symmetry change , possibly due to a valence transition of yb @xcite . on the other hand , muro et al . and bobev et al . have shown that the magnetic susceptibility of ybgage corresponds to an almost nonmagnetic state , suggesting the yb@xmath3 state in whole the temperature range @xcite . these results are indicative of large sensitivity of magnetic and thermal properties of ybgage to the chemical composition or stoichiometry . to clarify these discrepancies , we have studied the composition dependence of the physical properties of the ybgage system . in this paper , we report the solubility range of ybga@xmath0ge@xmath1 system , and their magnetic susceptibility and the thermal expansion . polycrystalline samples were prepared by argon arc melting from pure metals . to compensate the loss of volatile yb , a slightly richer amount of yb ( @xmath12% ) was charged . the samples were subsequently annealed in evacuated silica tubes at 750@xmath13c for 4 days . for samples with ge - rich compositions , a higher annealing temperature ( 850@xmath13c ) for 1 day was also examined . eds analysis revealed that the samples thus prepared were almost of ideal chemical composition . structure of samples were studied by powder x - ray diffraction at room temperature using cu k@xmath14 radiation with a rigaku diffractmeter . the magnetic susceptibility was measured with a squid magnetometer ( quantum design , mpms5s ) at 1000 oe . the temperature dependence of the lattice constants was investigated by means of powder x - ray diffraction using another rigaku diffractmeter . ge@xmath1 . patterns of ybga@xmath15 and yb@xmath16ge@xmath17 are calculated ones . peaks marked by @xmath18 are due to the ybga@xmath15 phase , and those marked by @xmath19 are due to yb@xmath16ge@xmath17 . , width=377 ] at first , we have studied the stoichiometry for yb@xmath20gage . sem and eds analysis have shown that the solubility range is within 3% , indicating that the chemical composition ratio of yb:(ga , ge ) is almost 1 : 2 . next , we have studied the system ybga@xmath0ge@xmath1 . in fig.1 , powder x - ray diffraction patterns of ybga@xmath0ge@xmath1 are shown . the pattern of @xmath4=1.0 , i.e. , ybgage , well agrees with a calculation ( not shown ) based on the structure parameter reported in ref @xcite . it is seen in the figure that the ybga@xmath0ge@xmath1 compounds have the same crystal structure up to @xmath4 = 1.5 , i.e. , ybga@xmath22ge@xmath23 . for @xmath24 , additional peaks are found to develop , which are attributed to the ybga@xmath15 phase with cain@xmath15-type structure @xcite , as is seen in the figure , where the peaks from the ybga@xmath15 phase are marked by asterisks . for @xmath25 the solubility range has been found to be narrow . in fig.1 , the patterns for @xmath4=0.9 and 0.8 are shown . one can see that peaks due to the yb@xmath16ge@xmath17 phase appear for @xmath26 . these data are those of the samples annealed at 850@xmath13c and subsequently quenched . for samples annealed at 750@xmath13c , the yb@xmath16ge@xmath17 phase is seen even for @xmath4 = 0.9 , suggesting that the solubility range is narrower at low temperatures . from these results , it is concluded that the single - phase region of ybga@xmath0ge@xmath1 is @xmath2 . in fig.2 , the lattice constants of ybga@xmath0ge@xmath1 are shown . these lattice constants were determined by a least square method using the peak position of the ( 3 0 0 ) , ( 1 1 12 ) , ( 2 1 8) , and ( 2 0 12 ) reflections . it is found that with increasing ga concentration @xmath4 , @xmath27 increases monotonically . on the other hand , @xmath21 does not show a monotonic composition dependence . the origin of this is not yet clear . nevertheless , the unit cell volume varies monotonically with @xmath4 in the single - phase region ( @xmath28 ) . we have found that the standard deviation for the estimation of lattice constants becomes extremely large for @xmath29 when we use the ( @xmath30 ) peaks with odd @xmath31 numbers , such as ( 2 1 7 ) and ( 2 0 11 ) . this is probably due to stacking faults along the @xmath21-direction for ga - rich compositions , as is shown below . in fig.3 , x - ray diffraction patterns of ybga@xmath0ge@xmath1 at high angles are shown . one can see that the ( 3 0 0 ) peak monotonically shifts to lower angles with increasing @xmath4 , indicating the monotonic increase of @xmath27 . on the other hand , other reflections do not show such simple @xmath4-variations . for the ( 2 1 7 ) and ( 2 0 11 ) reflections , the peaks shift to lower angles with increasing @xmath4 . especially for @xmath29 , the peaks shift rapidly . in contrast , the ( 2 1 8) and ( 2 0 12 ) reflections do not shift rapidly to lower angles , or even shift to higher angles for @xmath29 . it is also noticeable that the intensities of the peaks with odd @xmath31 show a remarkable decrease for @xmath29 , whereas those with even @xmath31 are almost unchanged . these @xmath4 variations can not be explained by an @xmath4-dependence of the lattice parameters . it should be noted that for ybgage , the ga - ge layers are stacked along the @xmath21 direction with the sequence @xmath32 @xcite . on the other hand , for ybga@xmath15 , the ga layers stack as @xmath33 @xcite . we speculate for ybga@xmath0ge@xmath1 with large @xmath4 , that the stacking of @xmath34 is inserted in the regular @xmath35 stacking . for the case of complete @xmath34 stacking , ( @xmath36 ) reflection with odd @xmath31 should vanish . thereby the insertion of @xmath34 in the regular @xmath35 stacking can lead to an incommensurate modulation of the @xmath37-value of these plains , giving rise to an @xmath4-dependence of the peak positions different from that expected from the lattice constants . having determined the single phase region to be @xmath2 , we show the physical properties of the ybga@xmath0ge@xmath1 system . in fig.4 , the magnetic susceptibility @xmath39 of ybga@xmath0ge@xmath1 is shown together with that reported by salvador et al . in ref . @xcite . in a good contrast to the data of ref . @xcite , our data show that the susceptibility of ybga@xmath0ge@xmath1 is extremely small in whole the range of @xmath4 . we have fitted the data between 150 and 300 k using a curie - weiss function : @xmath40 , where @xmath41 , @xmath42 , and @xmath43 are the curie constant , weiss temperature , and temperature - independent susceptibility , respectively . here @xmath41 is written as @xmath44 , where @xmath45 is the concentration of yb@xmath5 per formula unit , @xmath46 the avogadro number , @xmath47 the effective magnetic moment , and @xmath48 the boltzmann constant . @xmath47 should be 4.54@xmath49 for the case of yb@xmath5 . the results of fitting have yielded a value of @xmath50 , which corresponds to @xmath51 . this indicates that most of the yb ions are nonmagnetic , i.e. , in the yb@xmath3 state . this result is consistent with the results of x - ray photoemission spectroscopy measurement on polycrystalline samples @xcite . the low temperature increase in @xmath39 is then attributed to a small amount of impurities . since the yb@xmath3 state is consistently dominant for all the @xmath4 measured , we conclude that the valence of the yb ions in ybga@xmath0ge@xmath1 is divalent , independent of the chemical composition . in addition , no magnetic transition was observed down to 3 k. in several samples , a very tiny transition was observed at 2.3 k , and this is attributed to the antiferromagnetic transition in a trace of yb@xmath15o@xmath16 , of which the nel temperature is @xmath52 = 2.2 k @xcite . this contrasts with the valence transition around 5 k suggested in ref . @xcite . since the ionic radius of yb@xmath3 is larger than that of yb@xmath5 , a volume expansion with decreasing temperatures or zero thermal - expansion should not occur for these systems , as far as the valence - change mechanism is concerned . to confirm this expectation , we have studied the thermal expansion of ybga@xmath0ge@xmath1 , of which results are shown in fig . 5 . for ybgage , both @xmath27 and @xmath21 decrease monotonically with decreasing temperatures as in normal substances . as a result , the lattice volume also shows a monotonical decrease on cooling . this result for @xmath4 = 1.0 is consistent with that reported for ybgage by several authors @xcite . moreover , a similar lattice expansion is seen for the other samples , @xmath4 = 0.9 , 0.94 , and 1.2 , where @xmath27 , @xmath21 , and the lattice volumes show a conventional decrease on cooling . this behavior well agrees with the divalent yb state in these systems , because yb@xmath3 ions can no longer expand through a valence transition mechanisms . in summary , we have determined the solubility range of the ybga@xmath0ge@xmath1 system . in the composition range of @xmath2 , it is found that single phase samples isostructural to ybgage are formed . the magnetic susceptibility of these samples reveals that the yb ions are consistently divalent . this should lead a normal lattice shrinkage with decreasing temperature . in fact , x - ray diffraction shows normal lattice shrinkage on cooling in these systems . these results lead us the conclusion that the valence state of yb in ybga@xmath0ge@xmath1 is not sensitive to the composition @xmath4 but is robustly divalent , and that a zero thermal expansion does not occur in ybga@xmath0ge@xmath1 for whole the composition range ( @xmath2 ) , at least not via a valence transition mechanism . the ybgage phase with yb@xmath5 ions may be obtained at much smaller @xmath4 , because yb@xmath5 favors smaller lattice volumes . however , such samples were not obtained by conventional arc melting or annealing technique . other method such as high pressure synthesis may be useful . 00 e. bauer , adv . 40 ( 1991 ) 417 . m. laewenhaupt and k.h . fischer , in _ handbook on the physics and chemistry of rare earth _ , vol.16 , p.1 , edited by k.a . gschneidner jr and l. eyring , 1993 , elsevier . i. felner and i. nowik , phys . b 33 ( 1986 ) 617 . i. felner , i. nowik , d. vaknin , u. potzel , j. moser , g.m . kalvius , g. wortmann , g. schmiester , g. hilscher , e. gratz , c. schmitzer , n. pillmayr , k.g . prasad , h. de waard , and h. pinto , phys . b 35 ( 1987 ) 6956 . salvador , fu guo , t. hogan , and m.g . kanatzidis , nature 425 ( 2003 ) 702 , _ ibid _ ( corrigenda ) , 426 ( 2003 ) 584 . s. margadonna , k. prassides , a.n . fitch , j.r . salvador , m.g . kanatzidis , j. am . ( 2004 ) 4498 . y. muro , t. nakagawa , k. umeo , m. itoh , t. suzuki , and t. takabatake , j. phys . 73 ( 2004 ) 1450 . s. bobev , d.j . williams , j.d . thompson , and j.l . sarrao , solid state commun . 131 ( 2004 ) 431 . u. schwarz , r. giedigkeit , r. niewa , m. schmidt , w. schnelle , r. cardoso , m. hanfland , z. hu , k. klementiev , and y. grin , z. anorg . , 627 ( 2001 ) 2249 . k. yamamoto et al . , klaasse , f.r . de boer , and p.f . de chtel , physica 106b ( 1981 ) 178 .
the composition dependence of the yb valence and of the thermal expansion have been studied in the ybga@xmath0ge@xmath1 system . x - ray diffraction reveals that single - phased samples isostructural to ybgage are obtained in the range of @xmath2 . the magnetic susceptibility shows that the systems are almost nonmagnetic below room temperature , indicating a stable yb@xmath3 state for the @xmath4 range @xmath2 . the lattice constants as well as the lattice volume of these systems are found to decrease monotonically with decreasing temperature , suggesting the absence of zero thermal - expansion previously reported for ybgage . intermetallics , ybgage , x - ray diffraction , magnetic measurements , thermal expansion
b physics studies are exploiting a unique laboratory for testing the standard model in the fermion sector , for studying the qcd in the non - perturbative regime and for searching for new physics through virtual processes . + in the last decade , the lep and sld experiments played an important role in the study of b hadrons . at the start of the lep and slc accelerator in 1989 , only the @xmath9 and the @xmath10 hadrons were known and their properties were under study . new weakly decaying b hadrons have been observed ( @xmath1 , @xmath2 , @xmath3 ) for the first time and their production and decay properties have been measured . new strongly decaying hadrons , the orbitally ( l=1 ) excited b ( @xmath4 ) mesons have been also observed and their mass and production rates measured . + in this paper we will focus on the measurements of the ckm matrix elements : @xmath11 and @xmath12 through b decays and @xmath13 and @xmath14 using @xmath15 oscillations . on the other hand many additional measurements on b meson properties ( mass , branching fractions , lifetimes ... ) are necessary to constrain the heavy quark theories ( operator product expansion ( ope ) /heavy quark effective theory ( hqet ) /lattice qcd ( lqcd ) ) to allow for precise extraction of the ckm parameters . we finally show how these measurements constrain the standard model in the fermion sector , through the determination of the unitarity triangle parameters . + in this paper we try to compare the lep / sld results with those obtained from other collaborations ( cleo at cornell , cdf at tevatron and the asymmetric b - factories : babar and belle ) and to present , when available , the world average result . a detailed description of the results and of the averaging techniques can be found in @xcite . at the @xmath16 resonance , b hadrons are produced from the coupling of the @xmath16 to a @xmath17 quark pair . the production cross section is of @xmath18 6 nb , which is five times larger than at the @xmath19 . because of the specific ( v - a ) behaviour of the electroweak coupling at the @xmath16 pole , hadronic events account for about 70@xmath20 of the total production rate ; among these , the fraction of @xmath21 events is @xmath22 . because of the energy available only @xmath10 and @xmath23 mesons can be produced at the @xmath19 . the b particles are produced almost at rest ( the average momentum is of about 350 mev / c ) , with no accompanying additional hadrons , and the decay products of the two b particles are spread isotropically over the space . at the z pole , the primary @xmath17 pair , picks up from the vacuum other quark - antiquarks pairs and hadronizes into b hadrons plus few other particles . therefore not only @xmath24 and @xmath9 mesons are produced , but also @xmath25 mesons or @xmath26 baryons can be present in the final state . the @xmath0 and the @xmath27 hadronize almost independently . @xmath0 quarks fragment differently from light quarks , because of their high mass as compared with @xmath28 . b hadrons carry , on average , about 70@xmath20 of the available beam energy , whereas the rest of the energy is distributed among the other fragmenting particles . as a consequence , the two b hadrons fly in opposite directions and their decay products form jets situated in two opposite hemispheres . + the hard fragmentation and the long lifetime of the b quark make that the flight distance of a b hadron at the z pole , defined as @xmath29 , on average of the order of 3 mm . as decay products have a mean charged multiplicity of 5,-quark fragmentation and from b decay . ] it was possible to tag b hadrons using a lifetime tag . most of the precision measurements in b physics performed at lep / slc , tevatron and b factories , would not have been possible without the development of silicon micro vertex detectors . in practice the averaged flight distance of the b hadrons becomes measurable thanks to the precision of silicon detectors , located as close as possible to the beam interaction point . these detectors determine with a precision better than 10 @xmath30 , the position of a charged particle trajectory . in particular the separation between @xmath0 quarks and other quarks is mainly based on the use of vertex detectors . charged particles produced at the b vertex ( secondary vertex ) can be separated from those produced at the interaction point ( primary vertex ) using the precise tracking information . in spite of the relatively small statistics collected by the sld experiment , it gave very important and competitive contributions to b physics , because of its silicon vertex detectors , which is located very close to the interaction point . a typical lep @xmath31 event is shown in figure [ fig : aleph_event ] . , @xmath32 and @xmath33 which are obtained in correlation with an opposite sign lepton . these events are attributed mainly to the semileptonic decays of @xmath34 , @xmath35 and @xmath36 hadrons , respectively . the bottom - right figure shows the possibility of distinguishing the charged and neutral b mesons based on inclusive techniques.__,width=415 ] because of the large b mass , b hadrons are expected to decay into several decay modes with branching ratio of the order of a per mil . according to the registered statistics , at lep , inclusive or semi - exclusive @xmath0-hadron decays had to be studied in place of exclusive channels for which very few events are expected could be accessed . ] . + semileptonic decays benefit of a large branching ratio ( of the order of 10@xmath20 ) and of clean and easily distinguishable final states . semileptonic decays allow also to distinguish between different types of @xmath0 hadrons , by reconstructing charmed hadrons . as an example , a @xmath37 accompanied by a lepton with negative electric charge , in a jet , signs a @xmath0-baryon . for baryons , it is not even necessary to completely reconstruct the @xmath37 charmed baryon , correlations as @xmath38 or @xmath39 are sufficient . similarly , @xmath40 or @xmath41 events in a jet , provide event samples enriched in @xmath42 and @xmath43 mesons respectively . + an overview of the signals used to study these new states is given in figure [ fig:4figures ] . meson in 1992 , seen in the semileptonic decay : @xmath44 , whereas the right plot shows the same signal few years later . _ , title="fig:",width=226 ] meson in 1992 , seen in the semileptonic decay : @xmath44 , whereas the right plot shows the same signal few years later . _ , title="fig:",width=302 ] we take the example of the @xmath45 meson to illustrate how our knowledge on the properties of b hadrons has evolved during the last ten years . in 1992 , 7 events @xmath44 , constituted the first evidence for the @xmath25 meson . a few years later the same signal consists of more than 200 events . + in the mean time our knowledge has much improved : the fraction of @xmath1 mesons in b jets is precisely measured as well as the @xmath1 mass and lifetime . * the @xmath46 rate in @xmath0-jets amounts to : @xmath47 , * the @xmath25 meson mass is @xmath48 mev ( cdf mainly ) * the lifetime is @xmath49 . * the studies on @xmath50 oscillations give @xmath51 * the ratio @xmath52 measurements of b lifetimes test the decay dynamics , giving important information on non - perturbative qcd corrections induced by the spectator quark ( or diquark ) . decay rates are expressed using the ope formalism , as a sum of operators developed in series of order @xmath53 . in this formalism , no term of order @xmath54 is present and spectator effects contribute at order @xmath55 . would appear if in this expansion the mass of the heavy hadron was used instead of the mass of the quark . the presence of such a term would violate the quark - hadron duality . ] non - perturbative operators are evaluated , most reliably , using lattice qcd calculations . since the beginning of the lep / sld data taking an intense activity has been concentrated on the studies of b hadron lifetimes . + results are given in table [ table : life ] @xcite . [ table : life ] @ll b hadrons & lifetime [ ps ] + @xmath56 & 1.540 @xmath57 0.014 ( 0.9 @xmath20 ) + @xmath58 & 1.656 @xmath57 0.014 ( 0.8 @xmath20 ) + @xmath59 & 1.461 @xmath57 0.057 ( 3.9 @xmath20 ) + @xmath60 & 1.208 @xmath57 0.051 ( 4.2 @xmath20 ) + + + + + figure [ fig : liferatio ] gives the ratios of different b hadron lifetimes , as compared with theory predictions ( dark(yellow ) bands ) . the achieved experimental precision is remarkable and lep results are still dominating the scene . the fact that charged b mesons live longer than neutral b mesons is now established at 5@xmath61 level and is in agreement with theory . the @xmath62 and @xmath1 lifetimes are expected ( at @xmath631@xmath20 ) and found ( at @xmath634@xmath20 ) to be equal . a significant measurement in which this ratio differs from unity will have major consequences for the theory . the lifetime of the b - baryons is measured to be shorter than the @xmath62 lifetime , but the size of this effect seems to be more important than predicted ( 2 - 3@xmath61 ) . recent calculations of high order terms give an evaluation of the b - baryon lifetime in better agreement with the experimental result@xcite . + new results are expected from b - factories ( which could decrease the relative error on the lifetimes of the @xmath62 and @xmath10 to 0.4 - 0.5@xmath20 ) and mainly from tevatron ( run ii ) which could precisely measure all b hadron lifetimes , including those for the @xmath3 , @xmath64 and the @xmath65 . + in figure [ fig : life_story ] the improvement on the precision of the measured b hadron lifetimes over the years is shown . the @xmath5 element of the ckm matrix can be accessed by studying the rates of inclusive and exclusive semileptonic @xmath0-decays . the first method to extract @xmath5 makes use of the inclusive semileptonic decays of b - hadrons and of the theoretical calculations done in the framework of the ope . the inclusive semileptonic width @xmath66 is expressed as : @xmath67 from the experimental point of view the semileptonic width has been measured by the lep / sld and @xmath19 experiments with a relative precision of about 2@xmath20 @xcite : @xmath68 the precision on the determination of @xmath5 is mainly limited by theoretical uncertainties on the parameters entering in the expression of @xmath69 in equation [ eq : vcbtheo ] . moments of the hadronic mass spectrum , of the lepton energy spectrum and of the photon energy in the @xmath70 decay are sensitive to the non perturbative qcd parameters contained in the factor @xmath69 of equation [ eq : vcbtheo ] and in particular to the mass of the @xmath0 and @xmath71 quarks and to the fermi motion of the heavy quark inside the hadron , @xmath72 . and @xmath73 parameters are used , which can be related to the difference between hadron and quark masses and to @xmath72 , respectively . ] + results from delphi collaboration are shown in figure [ fig : moments ] . + similar results ( and with comparable precision ) have been obtained by cleo ( which did a pioneering work in this field ) and by the babar coll . @xcite . plane obtained : by the delphi coll . using the measured values of the first two moments of the hadronic mass and lepton energy spectra . the bands represent the 1@xmath61 regions selected by each moment and the ellipses show the 39@xmath20 , 68@xmath20 and 90@xmath20 probability regions of the global fit.__,width=359 ] using the experimental results on @xmath74 and @xmath73 : @xmath75 this result corresponds to an important improvement on the determination of the @xmath5 element . part of the theoretical errors ( from @xmath76 and @xmath72 ) is now absorbed in the experimental error and the theoretical error is reduced by a factor two . the remaining theoretical error could be further reduced if the parameters controlling the @xmath77 corrections are extracted directly from experimental data . an alternative method to determine @xmath5 is based on exclusive @xmath79 decays . using hqet , an expression for the differential decay rate can be derived @xmath80 @xcite.__,width=377 ] @xmath81 is the relative velocity between the b ( @xmath82 ) and the d systems ( @xmath83 ) . g(@xmath81 ) is a kinematical factor and f(@xmath81 ) is the form factor describing the transition . at zero recoil ( @xmath81=1 ) f(1 ) goes to unity . the strategy is then to measure @xmath84 , to extrapolate at zero recoil and to determine @xmath85 . + the experimental results are summarised in figure [ fig : vcb_ave ] . using f(1 ) = 0.91 @xmath57 0.04 @xcite , it gives @xcite : @xmath86 combining the two determinations of @xmath5 ( a possible correlation between the two determinations has been neglected ) it gives @xcite : @xmath87 measurements @xcite.__,width=377 ] the ckm matrix element @xmath6 has been measured at lep using semileptonic @xmath0 to @xmath88 decays . this measurement is rather difficult because one has to suppress the large background from the more abundant semileptonic @xmath0 to @xmath71 quark transitions . by using kinematical and topological variables , the lep experiments have succeeded in measuring the semileptonic @xmath0 to @xmath88 branching ratio @xcite , and obtain : @xmath89 using models based on the operator product expansion , a value for @xmath6 is obtained : @xmath90 prior to this analysis , the @xmath12 matrix element was firstly obtained , by cleo and argus collaborations , by looking at the spectrum of the lepton in b semileptonic decays . the difference between d meson and @xmath91 masses is reflected in the momentum of the lepton from the b decays . this analysis has been recently revised by the cleo coll .. an alternative method to determine @xmath6 consists in the reconstruction of the charmless semileptonic b decays : @xmath92 . this analysis has been performed by the cleo coll . and now by the b - factories . + figure [ fig : vub_ave ] shows the full set of results on @xmath12 @xcite . the probability that a @xmath94 meson oscillates into a @xmath95 or remains a @xmath96 is given by : @xmath97 where @xmath98 is the proper time , @xmath99 the lifetime of the @xmath100 meson , and @xmath101 the mass difference between the two physical mass eigenstates . is usually given in ps@xmath102 : 1 ps@xmath102 corresponds to 6.58 10@xmath103ev . ] to derive this formula the effects of cp violation and lifetime differences for the two states have been neglected . + integrating expression [ eq : oscillation ] , over the decay time , the probability to observe a @xmath104 meson starting from a @xmath105 meson is given by @xmath106 , where @xmath107 . at z energies , both @xmath23 and @xmath1 mesons are produced with fractions @xmath108 and @xmath109 . the average mixing parameter @xmath110 is defined as : @xmath111 . it has to be noted that for fast @xmath1 oscillations @xmath112 takes values close to 0.5 and @xmath112 becomes very insensitive to @xmath113 . even a very precise measurement of @xmath112 does not allow a determination of @xmath114 . + it is then clear that only the time evolution of the @xmath115 oscillations allow to measure @xmath116 and @xmath114 . + a time dependent study of @xmath117 oscillations requires : * the measurement of the proper time t , * to know if a @xmath118 or a @xmath95 decays at time t ( decay tag ) * to know if a @xmath0 or a @xmath119 quark has been produced at t = 0 ( production tag ) . in the standard model , @xmath15 oscillations occur through a second - order process - a box diagram - with a loop of w and up - type quarks . the box diagram with the exchange of a @xmath120 quark gives the dominant contribution : @xmath121f^2_{{b}_d } { b}_{b_d } & \nonumber \\ \delta { m}_s & ~\propto ~v_{ts}^2 f^2_{{b}_s } { b}_{b_s } ~\propto ~ v_{cb}^2 f^2_{{b}_s } { b}_{b_s } & \nonumber \\ \frac{\delta m_d}{\delta m_s } & ~\propto 1/\xi^2|\frac{v_{td}}{v_{ts}}|^2 ~\propto ~ 1/\xi^2 \lambda^2 [ ( 1 - \bar{\rho})^2 + \bar{\eta}^2 ] & \label{eq : dmddmsxi}\end{aligned}\ ] ] where @xmath122 . thus , the measurement of @xmath116 and @xmath114 gives access to the ckm matrix elements @xmath7 and @xmath8 respectively . the difference in the @xmath123 dependence of these expressions ( @xmath124 ) implies that @xmath125 . it is then clear that a very good proper time resolution is needed to measure the @xmath126 parameter . on the other hand the measurement of the ratio @xmath127 gives the same constraint as @xmath116 but this ratio is expected to have smaller theoretical uncertainties since the ratio @xmath128 is better known than the absolute value of @xmath129 . analyses using different events sample have been performed at lep . a typical time distribution is shown in figure [ fig : dmd ] . @xmath131 oscillations with a frequency @xmath130 are clearly visible . this can be a textbook plot ! the present summary of these results on @xmath130 , is shown in figure [ fig : dmdsummary ] . combining lep , cdf and sld measurements it follows that @xcite : @xmath132 @xmath130 has been first measured with high precision by the lep / sld / cdf experiments . the new and precise measurements performed at the b - factories confirmed these measurements and improved the precision by a factor two . the combined result is now : @xmath133 . the evolution , over the years , of the combined @xmath116 frequency measurement is shown in figure [ fig : dmd_story ] . frequency measurement over the years.__,width=302 ] the search for @xmath134 oscillations is more difficult because the oscillation frequency is much higher . in the standard model one expects @xmath125 . the proper time resolution will therefore play an essential role . five different types of analyses have been performed at lep / sld . an overview is given in table [ tab1 ] . .__characteristics of the different analyses are given in terms of statistics ( n ) , @xmath135 purity [ @xmath136 , tagging purities - i.e. the fraction of correctly tagged events - at the production and decay time @xmath137 and average time resolution within the first picosecond . _ _ [ cols="^,^,^,^,^,^",options="header " , ] different constraints can be used to select the allowed region for the apex of the triangle in the @xmath138-@xmath139 plane . five have been used so far : @xmath140 , @xmath141 , @xmath116 , the limit on @xmath114 and sin 2@xmath142 from the measurement of the cp asymmetry in @xmath143 decays . these constraints are shown in figure [ fig : bande ] @xcite . these measurements provide a set of constraints which are obtained by comparing measured and expected values of the corresponding quantities , in the framework of the standard model ( or of any other given model ) . in practice , theoretical expressions for these constraints involve several additional parameters such as quark masses , decay constants of b mesons and bag - factors . the values of these parameters are constrained by other measurements ( e.g. the top quark mass ) or using theoretical expectations . + different statistical methods have been defined to treat experimental and theoretical errors . the methods essentially differ in the treatment of the latter and can be classified into two main groups : frequentist and bayesian . the net result is that , if the same inputs are used , the different statistical methods select quite similar values for the different ckm parameters @xcite . the results in the following are shown using the bayesian approach . + central values and uncertainties taken for the relevant parameters used in these analyses are given in table [ tab : inputs ] @xcite . + the most crucial test is the comparison between the region selected by the measurements which are sensitive only to the sides of the unitarity triangle and the regions selected by the direct measurements of the cp violation in the kaon ( @xmath144 ) or in the b ( sin2@xmath142 ) sector . this test is shown in figure [ fig : testcp ] . and @xmath145 ( contours at 68% , 95% ) as selected by the measurement of @xmath146 , @xmath147 , the limit on @xmath148 are compared with the bands ( at 1 and 2@xmath61 ) selected from cp violation in the kaon ( @xmath144 ) and in the b ( sin2@xmath142 ) sectors.__,width=453 ] it can be translated quantitatively in the comparison between the value of sin2@xmath142 obtained from the measurement of the cp asymmetry in @xmath143 decays and the one determined from triangle measurements @xcite,@xcite : @xmath149 the spectacular agreement between these values shows the consistency of the standard model in describing the cp violation phenomena in terms of one single parameter @xmath150 . it is also an important test of the ope , hqet and lqcd theories which have been used to extract the ckm parameters . + including all five constraints the results are @xcite,@xcite : @xmath151 the ranges within parentheses correspond to 95@xmath20 probability . + the results on @xmath114 and @xmath152 are predictions for those quantities which will be measured in near future . during the last ten years , our understanding of the flavour sector of the standard model improved . lep and sld played a central role . + at the start of lep and sld , only the @xmath9 and the @xmath10 hadrons were known and their properties were under study . today b hadrons have been carefully studied and many quantities have already been measured with good precision . the hadron lifetimes are now measured at the one / few percent level . lep experiments are the main contributors for the measurement of @xmath5 , which is known with a relative precision better than 2@xmath20 . in this case , not only , the decay width has been measured , but also some of the non - perturbative qcd parameters entering in its expression . it is a great experimental achievement and a success for the theory description of the non - perturbative qcd phenomena in the framework of the ope . + lep experiments have been pioneering in determining @xmath6 using inclusive methods and reaching a precision of about 10@xmath20 , defining a road for future measurements at b - factories . + the time behaviour of @xmath153 oscillations has been studied and precisely measured in the @xmath23 sector . the new and precise measurements performed at the b - factories confirmed these measurements and improved the precision by a factor two . the oscillation frequency @xmath116 is known with a precision of about 1@xmath20 . @xmath154 oscillations have not been measured so far , but this search has pushed the experimental limit on the oscillation frequency @xmath114 well beyond any initial prediction for experimental capabilities . sld experiment has played a central role in this search . today we know that @xmath25 oscillates at least 30 times faster than @xmath23 mesons . the frequency of the @xmath154 oscillations will be soon measured at the tevatron . nevertheless the impact of the actual limit on @xmath114 for the determination of the unitarity triangle parameters is crucial . + the unitarity triangle parameters are today known within good precision . the evolution of our knowledge concerning the allowed region in the @xmath155-@xmath145 plane is shown in figure [ fig : storia ] . the reduction in size of the error bands , from the year 1995 to 2000 , is essentially due to the analyses described in this paper and to the progress in the ope , hqet and lqcd theories . the reduction between 2000 and 2002 is also driven by the precise measurements of sin 2 @xmath142 at the @xmath0-factories . + a crucial test has been already done : the comparison between the unitarity triangle parameters , as determined with quantities sensitive to the sides of the unitarity triangle ( semileptonic b decays and oscillations ) , with the measurements of cp violation in the kaon ( @xmath144 ) and in the b ( sin2@xmath142 ) sectors . this agreement tells us that the standard model is also working in the flavour sector and it is also an important test of the ope , hqet and lqcd theories which have been used to extract the ckm parameters . on the other hand , these tests are at about 10@xmath20 level accuracy , the current and the next facilities can surely push these tests to a 1@xmath20 level . i would like to thank the organisers for the invitation and for having set up a very interesting topical conference in a stimulating and nice atmosphere during and after the talks . + thanks to all the lep and sld members which have made all of it possible ! i would also like to remember the important work made from the members of the heavy flavour working groups who prepared a large fraction of the averages quoted in this note . they are all warmly thanked . + thanks to p. roudeau for the careful reading of the manuscript . aleph , cdf , delphi , l3 , opal , sld collaborations , cern - ep/2001 - 050 a. stocchi , plenary talk given at the xxxi@xmath156 ichep , amsterdam 24 - 31 july 2002 , hep - ph/0211245 . lep b lifetime working group : + http:@xmath157lepbosc.web.cern.ch/lepbosc/lifetimes/lepblife.html e. franco , v. lubicz , f. mescia , c. tarantino , nucl . b633 ( 2002 ) 212 . ( hep - ph/02030890 ) . @xmath11 working group : http:@xmath157lepvcb.web.cern.ch/lepvcb/ l. lellouch , plenary talk given at the xxxi@xmath156 ichep , amsterdam 24 - 31 july 2002 ; a. s. kronfeld , p. b. mackenzie , j. n. simone , s. hashimoto , s. m. ryan , proceedings of fpcp , may 1618 , philadelphia , pennsylvania ( hep - ph/0110253 ) ; phys . rev . d66 ( 2002 ) 014503 . results presented at the ckm unitarity triangle workshop , cern feb . 2002 : + http:@xmath157ckm-workshop.web.cern.ch/ckm-workshop/ in particular see also : + lep working group on @xmath5 : http:@xmath157lepvcb.web.cern.ch/lepvcb/ winter 2002 averages . + lep working group on @xmath6 : + http:@xmath157battagl.home.cern.ch/battagl/vub/vub.html . m. battaglia , talk given at the xxxi@xmath156 ichep , amsterdam 24 - 31 july 2002 . lep working group on oscillations : + http:@xmath157lepbosc.web.cern.ch / lepbosc / combined@xmath158results / amsterdam@xmath1582002/. h.g . moser and .a . roussarie , nucl . . meth . a384 ( 1997 ) 491 . n. cabibbo , phys . ( 1963 ) 531 . m. kobayashi and k. maskawa , prog . 49 ( 1973 ) 652 . l. wolfenstein , phys . * 51 * ( 1983 ) 1945 . buras , m.e . lautenbacher and g. ostermaier , phys . rev . * d50 * ( 1994 ) 3433 . for details see : m. ciuchini et al . , jhep 0107:013 , 2001 hep - ph/0012308 . m. artuso and e. barberio , hep - ph/0205163 . et al . , _ cdf collaboration , phys . rev . lett . * 74 * ( 1995 ) 2626 . + s. abachi _ et al . , _ d0 collaboration , phys . * 74 * ( 1995 ) 2632 . l. lellouch , plenary talk given at the xxxi@xmath156 ichep , amsterdam 24 - 31 july 2002 . average from y. nir see these proceedings based on : r. barate et al . , ( aleph collaboration ) _ phys . lett . _ * b492 * ( 2000 ) , 259 - 274 ; k. ackerstaff et al . , ( opal collaboration ) _ eur . phys . _ * c5 * ( 1998 ) 379 ; t. affolder at al . , d61 ( 2000 ) 072005 ; b. aubert et al . , ( babar collaboration ) hep - ex/0207042 ; k. abe at al . , ( belle collaboration ) hep - ex/0207098 . f. parodi , talk given at the xxxi@xmath156 ichep , amsterdam 24 - 31 july 2002 . results presented at the ckm unitarity triangle workshop , cern feb 2002 . + for the description of the methods see : + m. ciuchini _ et al . _ , jhep * 0107 * ( 2001 ) 013 ( hep - ph/0012308 ) . + a. hcker , h. lacker , s. laplace , f. le diberder , eur . j. * c21 * , 225 ( 2001 ) . + g. p. dubois - felsmann d. g. hitlin , f. c. porter and g. eigen , calt 68 - 2396 june 2002 .
in the last decade , the lep and sld experiments played a central role in the study of b hadrons ( hadrons containing a @xmath0 quark ) . new b hadrons have been observed ( @xmath1 , @xmath2 , @xmath3 and @xmath4 ) and their production and decay properties have been measured . in this paper we will focus on measurements of the ckm matrix elements : @xmath5 , @xmath6 , @xmath7 and @xmath8 . we will show how all these measurements , together with theoretical developments , have significantly improved our knowledge on the flavour sector of the standard model .
we present the methodology used to derive structural measurements for the galaxies investigated in the _ zurich environmental study ( zens ) _ ( @xcite , hereafter paper i ) . the resulting measurements are provided in the zens global catalog that we have published electronically with paper i. zens is designed to address the question of which specific environment is most relevant for influencing the properties of different galaxy populations . several definitions of environment have been commonly employed in the literature to study the relation between environment and galaxy evolution : the density of galaxies calculated out to a fixed or an adaptive distance ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , the mass of the host group or cluster @xcite , the distance from the group / cluster center @xcite or the location into larger structures such as cosmic filaments or superclusters @xcite . recently , the ability to separate galaxies into centrals and satellites within their host group halos has produced mounting evidence that the environmental influence on the star - formation properties of galaxies may peak for satellite galaxies @xcite . there is cross - talk however between different definitions of environment , which may also relate to one another from a physical perspective . a key question is therefore to identify what is the relative importance of different environmental conditions for well - defined galaxy populations of different masses , star formation activity levels , and structural / morphological properties ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in zens we aim at helping clarifying which of the many environments that a galaxy experience has a larger impact on its evolution . we do so by using the same sample of suitably selected nearby galaxies to investigate the dependence of their properties , at fixed stellar mass , on four environment measurements : the host group halo mass , the radial segregation within the group , the large scale density field on which the group halos reside , and the galaxy rank within its group halo , i.e. , whether it is the central or a satellite galaxy within the gravitational potential of its host group . in computing our proxies for these different environments , we have attempted to minimize cross - talks between their definitions , in order to better disentangle one from another of the physical conditions that galaxies experience ( see paper i ) . this paper focuses on the quantification of robust galaxy structural and morphological properties , which provide key information on the life histories of galaxies . the presence and properties of massive disks and spheroids highlight the occurrence of relatively slow and dissipative gas accretion @xcite or mergers ( e.g * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , respectively . inner cores or cusps @xcite and tidal debris @xcite also trace the degree of dissipation involved in the evolution of galaxies ( not surprisingly with some debate , @xcite ) . bars and pseudo - bulges are smoking guns for either secular evolution processes ( @xcite ) , or possibly for early bulge formation through instabilities in the proto - disks @xcite . a robust determination of galaxy morphology is hence essential for understanding whether this is linked to any of the environmental conditions above , how precisely it relates to the occurrence , enhancement or cessation of star formation activity , and thus for pinning down which physical processes drive galaxy evolution . determining galaxy structure is however notoriously not a trivial task . visual morphological classification is still a widely adopted method ( e.g. * ? ? ? * ; * ? ? ? * ) , despite its subjectivity and failure to provide quantitative measurements for different components , which are necessary to trace galaxy assembly over cosmic time . for these reasons numerous publications have been devoted to the development of methods and software for the automated quantification of structure on large galaxy samples . there are a number of approaches to the problem which can be broadly divided into two categories : those which employ parametric descriptors for the galaxy morphology , namely a set of analytical profiles used to model the bulge , disk , or bar component ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and those which instead use the observed properties of the light distribution , such as the degree of asymmetry , isolation of bright pixels or decompositions into a set of basis functions ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the two methods have different strengths : parametric decomposition is useful to obtain measurements of characteristic sizes and to have an estimate of the relative importance of the bar , disk and bulge components ; it also easily includes the effects of seeing . non - parametric estimators well describe the inhomogeneities in the light distributions of real galaxies , which typically display irregular , non axis - symmetric features generated by recent star - formation , dust or galaxy interactions . both parametric and non - parametric measurements suffer however from a number of observational biases , which must be corrected for in order to perform comparisons between galaxies of different properties , and observed in different conditions . in particular , in ground - based surveys , the effect of atmospheric seeing is one of the major complications . several studies have shown the strong impact of the seeing on the photometric and structural properties of galaxies , not only in the inner regions of galaxies , but also out to radii corresponding to several fwhm of the point spread function ( psf ) ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? another factor which affects the derivation of structural parameters is the inclination angle at which a galaxy is observed . the overlap , in projection , of multiple subcomponents , as well as physical factors such as the non - uniform distribution of inter - stellar dust which causes a higher attenuation of short - wavelength light in the central regions of edge - on galaxies than in similar face - on galaxies ( e.g. * ? ? ? * ; * ? ? ? * ) can substantially bias the measurements of sizes , bulge - to - disk ratios , concentration and even stellar masses ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? finally , the background sky makes the detection of faint components difficult , a fact which introduces severe biases in the measurements of the galaxy properties , especially magnitudes and sizes ( e.g. * ? ? ? * ; * ? ? ? the strength of the bias depends on galaxy size , inclination and stellar light profile . although a number of widely - used measurement techniques , such as the computation of kron aperture fluxes @xcite or the extrapolation of model galaxy surface brightness profiles @xcite , can help recover light below the isophotal limit , significant systematic biases remain in the low surface brightness regimes specific to each survey ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in zens we attempt _ to correct _ , when possible , all measurements of galaxy structure for systematic biases as a function of psf - size , and also galaxy magnitude , size , axis ratio and radial shape of the light profile . we also quantify the size of systematic biases in regimes of parameter space where the ( statistical ) recovery of the intrinsic information is not achievable , e.g. , at small galaxy sizes and low surface brightnesses . we also stress that in zens each galaxy is handled individually , till self - consistent and both quantitatively- and visually - checked accurate measurements are achieved . this enables zens to tackle complementary questions regarding the galaxy - environments relationship relative to larger but less detailed galaxy samples . the paper is organized as follows . in section [ sec : survey ] we briefly review the specifications and definitions for the four environments under scrutiny in zens . we then devote the first part of the paper to an overview of the structural measurements carried out on the zens galaxy sample . these measurements include isophotal analyses and bar detection / quantification ( section [ sec : isophotalanalysis ] ) , analytical surface brightness fits and bulge+disk decompositions ( section [ sec : gim2dmeasurements ] ) , and derivation of non - parametric structural indices ( concentration , gini , asymmetry , m@xmath8 , smoothness ; section [ sec : zest ] ) . in section [ sec : simulations ] we thoroughly investigate the sources of error in these measurements and derive a correction scheme that recovers the intrinsic structural parameters . in section [ sec : morphclass ] we present the morphological classification of the zens galaxies , based on a quantitative partition of the structural parameter space in regions that are associated with elliptical , bulge - dominated disks , intermediate bulge - to - total ratio disks , late - type disks and irregular galaxies . in the same section we also discuss the statistics of the structural properties for the various morphological classes . we describe in detail in appendix [ app : datareduction ] the data reduction and photometric calibration of the eso @xmath9 and @xmath10 wfi/2.2 m imaging data for the zens groups that are introduced in paper i. appendices [ sec : testgim2d]-[app : corrections_psf ] present additional details on the tests performed on the analytical surface - brightness fits and supplementary information for the derivation of the corrections for the structural parameters . stamp images for galaxies in the different morphological classes are found in appendix [ app : classdist ] . in the final part of the paper we use the corrected structural measurements to study , at constant stellar mass , the concentration of satellite galaxies as a function of hubble type and environment ( section [ sec : results ] ) . we summarize the paper in section [ sec : conclusions ] . this second zens publication is complemented by a companion paper ( @xcite , hereafter paper iii ) , in which we present the spectrophotometric properties of our galaxy sample and to which we refer for details on the derivation of , e.g. , the galaxy spectrophotometric types , star formation rates and stellar masses that we use in this and other zens papers . the following cosmological parameters are adopted in all the zens publications : @xmath11 , @xmath12 and @xmath13 . unless otherwise specified , magnitudes are in the ab system , and galaxy sizes are _ semi - major _ axis measurements . all derived luminosities are corrected for galactic extinction using the maps of @xcite . zens is based on a sample of 1484 galaxies , members of 141 galaxy groups extracted from the 2-degrees field galaxy redshift survey ( 2dfgrs ) ( @xcite ) , and specifically from the percolation - inferred galaxy group ( 2pigg ) catalogue ( @xcite ) . the 141 zens groups are a random selection of the 2pigg groups which are found in the very thin redshift slice [email protected] and have at least 5 confirmed members , down to a magnitude @xmath15 . new @xmath16 and @xmath17band images were acquired for these groups with the wfi camera mounted at the cassegrain focus of the mpg / eso 2.2 m telescope at la silla , over several observing runs between 2005 and 2009 ( see paper i for details ) . the data reduction and photometric calibration of these wfi data is reported in appendix [ app : datareduction ] . this paper describes in detail the structural analysis that we have performed on these wfi @xmath9 and @xmath10 images . for each zens galaxy , paper i discusses and publishes four estimates of environment : the group halo mass , the distance from the center of the group , the rank within the group ( i.e. , whether the galaxy is the central or a satellite ) and the location on the large scale structure ( lss ) . in particular : @xmath18 group masses @xmath1 are derived from the total group luminosities by assuming a mass - to - light ratio calibrated with mock catalogs ( @xcite ) . @xmath19 the ranking of galaxies in centrals and satellites factors in the errors on the galaxy stellar masses , and includes a test of self - consistency requiring that a bona - fide central galaxy must , simultaneously , be consistent with being the most massive galaxy of the group , be located within a projected distance @xmath20 from the mass weighted center of the group ( with @xmath21 the characteristic size of the group , as defined in paper i ) , and have a relative velocity within one standard deviation from the group velocity . @xmath22 the centers of the groups , on which the projected radial positions of satellite galaxies within the groups rely , are identified with the locations of the central galaxies . we operationally divide the zens group sample into _ relaxed _ and _ unrelaxed _ groups , depending on whether a bona - fide central galaxy can be found , according to the prescription above . in paper i we discuss the observational biases which may hamper the identification of the central galaxy in groups which are in fact dynamically relaxed . we therefore use our group classification scheme mostly to test that our results are not affected by the inclusion / exclusion of the unrelaxed groups from our studies , although we keep an eye on the possibility that there may be a physical origin , related to the dynamical evolution of the host groups , for differences observed in otherwise similar galaxies that inhabit the two classes of groups . @xmath23 the lss density at the zens group locations is defined using an nth nearest - neighbor analysis which adopts the groups ( not the galaxies ) as the tracers of the lss density . specifically , we define @xmath24 , with @xmath25 the density of 2dfgrs groups in a projected circular area defined by the comoving distance of the 5th nearest - neighboring group around the zens group , and @xmath26 is the mean projected density calculated over the global 2dfgrs area at the given redshift . our choice results in identical @xmath3 values for all galaxy members of any given group , and avoids the problems that are associated with using nth - nearest neighboring galaxy algorithms ( see paper i , and also @xcite ) . in paper iii we discuss the mass completeness limits of zens which is determined by the 2dfgrs apparent magnitude selection . the strongest constraints are set by passively evolving galaxies , which have the larger mass - to - light ratio , for which the 85@xmath27 completeness is reached above @xmath28 . this is the mass completeness threshold that we adopt for passive elliptical and s0 galaxies . progressively more actively star forming galaxies have progressively lower mass completeness thresholds , with strongly star forming galaxies mass complete at @xmath29 ( see paper iii ) . given the statistical mix of spectrophotometric types in the different morphological bins , the mass completeness thresholds of @xmath30 , @xmath31 and @xmath32 are statistically adequate respectively for the morphological classes of bulge - dominated spirals , intermediate bulge - to - total disk galaxies , and late - type disk ( or irregular ) galaxies . from top to bottom , the panels show the radial profiles of @xmath17band surface brightness , ellipticity , position angle and the high order fourier coefficient @xmath33 for three example of zens disk galaxies classified as non - barred ( left ) and barred ( right ) . different symbols and gray shades are used to distinguish the profiles for the three individual galaxies.,width=340 ] surface brightness profiles and isophotal parameters were obtained with the iraf routine . this well - tested algorithm fits the intensity at a given radius with the fourier series @xmath34 @xcite . here @xmath35 is the mean intensity within the isophote , and @xmath36 , @xmath37 the high harmonic coefficients which quantify isophotal deviations from perfect ellipticity . we truncated the series after the fourth order , so that , for each isophote , the fits return the mean intensity , position angle ( pa ) , ellipticity ( @xmath38 ) , and the amplitude of the fourth - order coefficients @xmath33 and @xmath39 . the coefficient @xmath33 has been extensively used in the past to measure ` boxiness ' ( @xmath40 ) and ` diskyness ' ( @xmath41 ) of the isophotes ( e.g. , @xcite , @xcite , @xcite ) . in running , we allowed the position angle and ellipticity to vary freely with radius , and limited wanderings of the isophotal center to within 3 pixels from the center of the innermost isophotes . the pa and ellipticity determined by the sextractor algorithm ( @xcite ) served as initial guesses for the algorithm . the semi - major axis of adjacent isophotes was increased in logarithmic radial steps of 0.1 , in order to increase the signal - to - noise ratio when measuring the external isophotes . we allowed the code to perform a two iteration 3-sigma clipping of the discrepant pixels during the fits , and terminated the fitting procedure when @xmath42 of the pixels in a given step of the calculation were flagged as discrepant . the procedure was applied independently to the @xmath10 and @xmath9 images , and returned independent surface brightness profiles in each of the two passbands . we also derived surface brightness profiles in the @xmath9 filter using the isophotal parameters derived from the @xmath10 images ; this returned more reliable surface brightness profiles also at the shorter wavelength for those galaxies whose @xmath9 light distributions were too irregular ( because of dust absorption and star formation knots ) for a reliable measurement directly on the @xmath9 images . changes in the radial profiles of pa and ellipticity were used as a first diagnostics to identify bars in the ( disk ) galaxies . the presence of a bar produces a characteristic signature on the isophotes shapes : within the bar region the ellipticity increases smoothly to a maximum value with almost constant pa , and then drops abruptly at the end of the bar , where also the pa changes substantially . we inspected the @xmath10-band ellipticity and position angle radial profiles of the disk galaxies in our sample ( s0 and later , see section [ sec : morphclass ] for the definition of the morphological types ) . we classified a disk galaxy as barred if its ellipticity grows to an absolute maximum greater than 0.2 , and shows a variation in ellipticity and pa greater than 0.1 and 10@xmath43 , respectively . in the bar region , we furthermore request that the pa profile is flat within @xmath44 . the size of the bar , @xmath45 , was defined to be the semi - major axis associated with the maximum value of the ellipticity profile , subject to the condition that it had to be at least twice the fwhm of the psf , which is typically @xmath46 . note that galaxies with bar sizes smaller than this threshold are not considered as barred systems . therefore , our final sample of barred disks provides a conservative lower limit to the total number of such galaxies in the entire sample . illustrative examples of the isophotal parameters profiles for a random selection of zens galaxies classified as barred and non - barred are shown in figure [ fig : sb_profiles ] . the above mentioned criteria are widely used in the search for bars ( e.g. @xcite , @xcite or @xcite ) and are shown to provide robust results , but they can fail in identifying a bar if , e.g. , the bar has a similar pa to that of the disk ( see e.g. , @xcite ) . for this reason , we also visually inspected all the disk galaxies and those which clearly showed a bar structure and had an ellipticity profile consistent with it , but no drop in pa were included in the barred sample as well . combining the fitting method with the visual validation we identified a clear bar signature in 148 disk galaxies . to quantify the bar strength we used the definition of @xcite , i.e. , @xmath47 \ , , \ ] ] where @xmath48 is the intrinsic axial ratio of the bar . the bar strength ranges from zero to unity for unbarred galaxies and infinitely strong bars , respectively . the intrinsic axial ratio was derived from the observed one using the transformation of @xcite , i.e. , @xmath49 with @xmath50 \ , .\end{aligned}\ ] ] here @xmath51 is the inclination angle obtained from the minor and major axis of the galaxy considered as whole ( @xmath52 ) , @xmath53 is the observed axis ratio of the bar and @xmath54 is the twist angle between the bar and galaxy semi - major axis . for our sample we computed @xmath53 at the bar semi - major axis @xmath45 and we used the values of ellipticity and pa obtained from sextractor @xcite to calculate @xmath55 and @xmath54 . figure [ fig : barred_hist ] shows the distribution of bar strength and bar sizes measured in the zens samples of bulge - dominated galaxies ( including both s0 and bulge - dominated spirals ) , intermediate - type and late - type disk galaxies . the median bar strength mildly increases with hubble type , changing from a value of @xmath56 for s0 or bulge - dominated spirals , to @xmath57 for intermediate - type disks and @xmath58 for late - type disk galaxies . we also find a decrease in median bar size from late - type to early - type disks : @xmath59 kpc for s0 and bulge - dominated spirals , @xmath60 kpc for intermediate - type disks and @xmath61 for late - type disks . for both the bar size and strength a kolmogorov - smirnov test rejects a common parent distribution , between bulge - dominated galaxies and late - type disks , with a probability of 99% . the correlation between bar properties and hubble types ( which could be a reflection of a correlation with galaxy stellar mass , ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) ) has been investigated in several works in the literature , whose results are consistent with our measurements . for example , a weakening of bars in early type galaxies is reported by @xcite , and an increase in bar size in early- relative to late - type disks is found by @xcite . two dimensional fits to the surface - brightness distributions of all zens galaxy images were carried out with the galaxy image 2d ( gim2d ) software package ( @xcite , @xcite ) . we used a single srsic profile to describe the total galaxy light distribution , i.e. : @xmath62\ } $ ] where @xmath63 is the surface intensity at the half - light radius @xmath64 . the value of the parameter @xmath65 is such to ensure that the flux within @xmath64 is half of the total flux and it is approximated to @xmath66 ( e.g. * ? ? ? * ; * ? ? ? * ) . for the zens galaxies which are not classified as ellipticals ( in section [ sec : morphclass ] ) , we also performed two - components , bulge+disk decompositions of the two - dimensional galaxy light distributions . for this purpose we assumed a srsic profile for the bulge and a perfectly exponential disk represented by @xmath67 , with @xmath68 the central surface intensity and @xmath69 the disk scale length . pure exponential models , with no bulge component , were furthermore generated for galaxies classified as late - type disks ( see again section [ sec : morphclass ] ) . table [ tab : gim2d_parameters ] summarizes the range of values between which the model parameters were allowed to vary . rrrrrrrrrr @xmath70 / @xmath71 ( mag ) & 18/16.5 & 21.5/20 & 13/12 & 18/16.5 & 21.5/20 & 13/12 & 18/16.5 & 21.5/20 & 13/12 + b / t & 0.5 & 0 & 1 & 1 & 1 & 1 & 0 & 0 & 0 + @xmath72 ( kpc ) & 2.3 & 0.5 & 20 & 2.3 & 0.5 & 30 & - & - & - + @xmath73 & 0.5 & 0 & 0.7 & 0.5 & 0 & 1.0 & - & - & - + @xmath74 ( @xmath43 ) & 45 & -180 & 180 & 45 & -180 & 180 & - & - & - + @xmath75 ( kpc ) & 2.3 & 0.5 & 20 & - & - & - & 2.3 & 0.5 & 20 + @xmath76 ( @xmath43 ) & 45 & 0 & 90 & - & - & - & 45 & 0 & 90 + @xmath77 ( @xmath43 ) & 45 & -180 & 180 & - & & - & 45 & -180 & 180 + @xmath78 ( pixels ) & 0 & -2 & 2 & 0 & -2 & 2 & 0 & -2 & 2 + @xmath79 ( pixels ) & 0 & -2 & 2 & 0 & -2 & 2 & 0 & -2 & 2 + @xmath80 & 2.1 & 0.2 & 10 & 2.1 & 0.2 & 10 & - & - & - + gim2d convolves the theoretical models with the psf before fitting them to the galaxy images . we modeled the psf of each of the zens group with a two - dimensional gaussian having fwhm equal to the mean of the full - widths measured on several unsaturated stars over the zens fields . a single psf was used for a whole zens field as the wfi psf varies only slightly across the field of view of the camera , showing in particular a mild @xmath7 increase of the psf fwhm very close to the edges of the ccds ( as tested on sextractor sources with stellarity class @xmath81 in all zens fields ) . due to statistical scatter and signal - to - noise variations , errors introduced by applying a correction for this effect are comparable to or even marginally larger than those resulting from the adoption of a single psf for each field . larger psf variations in zens are instead observed between different fields ( i.e. pointings ) ; in this case , the psf can change by as much as 0.6 - 0.7 arcseconds ( see figure [ fig : skyconditions ] ) . this effect is taken into account not only during the model fitting , but also when applying the correction scheme that we develop in section [ sec : simulations ] . during the fitting procedure , the code was allowed to re - compute the initial parameters from the image moments and to fit the sky background level in each individual postage stamp galaxy image ; the postage stamps were sized proportionally to the sextractor petrosian radius of the galaxies , and set to be equal to 3 times the petrosian radius . on a set of simulations calibrated on the gems survey @xcite , fixing the background to a locally defined value during the gim2d fit was shown to improve the performances for low surface brightness galaxies @xcite . our alternative approach was to enable the local sky subtraction and to quantify any systematic bias in the gim2d fits using a large set of artificial galaxy images which are specifically tailored to the zens observations ( section [ sec : simulations ] ) . the sky pixels were identified as those pixels in the galaxy postage stamps that were located outside an aperture equal to 1.5 times the sextractor petrosian radii of the galaxies . bright star - forming clumps were masked during the fitting procedure . finally , although gim2d offers an option to correct for the effects of the disk optical thickness which consists in adding to the disk total magnitude a geometric factor @xmath82 where @xmath83 is the axis ratio we decided not to use this feature , and disks were assumed to be optically thin in order to fit the actually observed light distribution without a prior assumption about the dust distribution . to be able to detect color gradients , for the variable-@xmath80 single srsic fits as well as for the @xmath84 pure exponential models , the @xmath16 and @xmath17band images were fitted independently without imposing the structural parameters of one filter to the other band . the position angles and ellipticity derived from the @xmath16 and @xmath17 band images agree well , with differences being limited to 15@xmath43 and 0.15 , respectively . in only @xmath85 of the galaxies we find differences which are larger than this , with a maximum change in ellipticity @xmath860.2 and a maximum change in position angle @xmath8630@xmath43 . generally speaking , we thus conclude that the single component fits are robust and strong twists between the @xmath16 and @xmath17band isophotes are not a problematic issue when performing independent fits for the two passbands . to verify the validity of the gim2d single - component fits , we ran on all the gim2d models the task , keeping the isophotes fixed at the radii , position angle and ellipticity of the fits to the real galaxy images described in section [ sec : isophotalanalysis ] . the comparison between the total magnitudes and the profiles derived for the gim2d models and those derived for the real galaxies , as well as the inspection of the residual images between the gim2d models and the real galaxies , enabled us to reject unphysical gim2d models . overall , we could obtain reliable srsic fits for @xmath87 of the @xmath16band and @xmath17band images . a similar fraction of successful fits is obtained for the pure exponential models of late - type disks ( @xmath88 ) . the distribution of stellar masses ( from paper iii ) , sizes and morphological types for the @xmath89 of zens galaxies with no single srsic fits is shown in figure [ fig : failedgim2d ] and table [ tab : gim2drejected ] of appendix [ sec : testgim2d ] . in this appendix we also present , for the subset of zens galaxies for which the information is publicly available from the nyu - vagc @xcite , a comparison of our estimates of the srsic parameters with those published in this other catalogue ; the comparison shows a good agreement between our _ uncorrected _ measurements and the previously published data . we show below however that , in some regimes of parameter space , these measurements require further corrections to be cleaned by residual observational biases , indicating that also measurements in the quoted and other public catalogues necessitate of similar attention . the comparison between the structural parameters obtained from the single srsic fits with variable-@xmath80 in the two available passbands is shown in the left and central plots of figure [ fig : gim2d_comp1 ] . the single component fits in the two different pass - bands generally provide consistent measurements with little scatter . in particular , half - light radii measured in the two pass - bands are in very good agreement with each other . the @xmath10-band fits , however , result in slightly steeper profiles : at this wavelength galaxies have @xmath80 indices which are bigger by @xmath90 with respect to those measured in the @xmath9-band . this is not unexpected in an ` inside - out ' galaxy formation process , as younger stellar populations at large radii would lead to this effect . another difference is observed between @xmath17 and @xmath16band disk scalelengths of large late - type galaxies , as derived from the pure exponential fits ( right - hand side plot in figure [ fig : gim2d_comp1 ] ) : disk scalelengths @xmath69 derived from @xmath84 fits to relatively large late - type disk galaxies are smaller at the longer wavelength by @xmath91 . we have checked that the @xmath10 magnitude distribution of these late - type disks is not biased toward faint values that would raise concerns on surface brightness detections . as we show in detail in section [ sec : correctionmaps ] , the @xmath17band gim2d sizes that we derive for relatively large , @xmath92 galaxies are not severely affected by observational biases . we thus interpret this result again as mostly due to a genuine color gradient in large late - type disks . note that the average difference in the @xmath10 and @xmath9 disk scalelengths in the pure exponential fits of late - type disk galaxies is consistent with a similar difference reported for a sample of local late - type galaxies by @xcite and @xcite . summarizing , the larger @xmath69 values from @xmath84 fits and smaller srsic indices from variable-@xmath80 fits of late - type disks in the @xmath9-band relative to the @xmath17band are consistent with being the joint result of the segregation of young stars in the galaxy outskirts and , also , of light absorption from the center of the galaxies by interstellar dust , as discussed in a number of observational and theoretical works ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the more pronounced variation of disk scalelength in the pure exponential fits with respect to the half - light radius in the srsic fits for the late - type galaxies can be readily understood : whereas in the variable-@xmath80 single srsic fits a lower central concentration of light can be modeled with a smaller value of the srsic index @xmath80 and a relatively small change in the effective radius , fixing the index to @xmath84 in the pure exponential fits forces an increase of the characteristic scalelength in order to obtain a milder radial decline in surface brightness in the @xmath9 band . this is consistent with the fact that the largest variation in @xmath80 values between the @xmath9- and @xmath10-band variable-@xmath80 srsic fits to late - type disks are observed for those galaxies which also have the largest variations in scalelength @xmath69 when fitted with an @xmath84 profile . finally we note that a self - consistent correction for dust effects would require radiative transfer simulations of the light scattered and re - emitted by the dust grains ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) ; this is beyond the scope of this paper . moreover it remains difficult to disentangle dust effects from genuine radial segregation in the stellar populations . for these reasons , instead of attempting a correction for dust absorption , we choose the empirical approach of employing the less dust - sensitive @xmath10-band data as the fiducial reference for our structural measurements , and to discuss separately possible dust - reddening effects , when relevant , e.g. , in studying color profiles and star formation rates ( see paper iii ) . double component , bulge+disk decompositions were also performed on both the @xmath9- and @xmath10-band images of all galaxies which are not classified as ellipticals or irregulars ( section [ sec : morphclass ] ) . for the @xmath10-band , our reference filter for structural measurements , no a priori constraints on the bulge and disk parameters were imposed during the fitting procedure , except for the wide limits listed in table [ tab : gim2d_parameters ] . in the @xmath9 filter , each galaxy was instead fitted in four different ways : @xmath93 by performing a separate decomposition to the @xmath9-band letting the structural parameters completely unconstrained ( unconstrained model fit , hereafter @xmath94 ) , @xmath95 by fixing the disk and bulge position angles , ellipticity and inclination to those of the @xmath10-band ( constrained model fit number one , hereafter @xmath96 ) , @xmath97 by also fixing the bulge half - light radius and srsic index @xmath80 ( constrained model @xmath98 ) and @xmath99 by keeping all the parameters tied to those of @xmath10-band except for the @xmath100 and total flux ( constrained model @xmath101 ) . in all the models we allowed a maximum wandering of 2 pixels for the disk and bulge components from the sextractor center . there are clear advantages and disadvantages in performing either independent or constrained fits to the two bands : by fixing the @xmath9-band structural parameters to the @xmath10-band , one ensures that bulge and disk colors are measured consistently over the same regions ; this however prevents the detection of structural differences and color gradients . for this reason we decided to adopt a mixed approach to determine the bulge and disk parameters to the @xmath9-band . first , not all models returned by the gim2d fits are physically - meaningful bulge+disk decompositions . we hence adopted a filtering scheme to reject unreliable or unphysical models to all gim2d models , i.e. , the fiducial @xmath17band models for each galaxy , and the four versions of the @xmath16band models . our filtering scheme is described in appendix [ sec : testgim2d ] . we then followed a quantitative procedure to select , amongst the physically - valid alternatives for the @xmath9 bulge+disk fits for each galaxy , our fiducial ( i.e. , in our judgement , the most reliable ) bulge+disk @xmath16band decomposition . in brief , we required that all @xmath16band disk and bulge fits always have bulge and disk position angles , disk inclinations and bulge axis ratios consistent , within a sensible range and @xmath10 bulge and disk position angles , disk inclination angles and bulge ellipticities were respectively @xmath102 , @xmath102 and 0.15 . ] , to those of the @xmath17band fits . when this was achieved with unconstrained fits to the @xmath9 band images , these unconstrained fits were retained as a fair description of the @xmath9 bulge+disk decompositions . this choice maximizes the detection of possible wavelength - dependent structural differences and color gradients . for galaxies in which such a consistency requirement was not achieved with the unconstrained @xmath9 fits , we adopted as fiducial @xmath16band fits those which satisfied such requirement with the minimum number of @xmath9 parameters tied to the @xmath10-band fit parameters ( i.e. , in order of priority , the @xmath96 , @xmath98 and @xmath101 fits ) . in cases where both the @xmath96 and @xmath94 models were in principle both potentially good representations of the bulge+disk properties of considered galaxy , we applied the following decision scheme : if the two models gave disk and bulge size , @xmath100 and srsic indices within @xmath103 of the dispersion measured around the identity from all the model falling in this latter category , then both models were validated as reliable , and we adopted as our fiducial bulge+disk parameter estimates the mean of the structural parameters returned by these two fits . if these two fits returned discrepant values , both were inspected and a judgement was made on which model to use , on the basis of the residual images and the difference in total magnitude between the galaxy and the model . only @xmath104 of the bulge+disk @xmath9 fits needed this further visual validation . this entire procedure is schematized in figure [ fig : btchart ] . it is important to notice that , while disk scalelength and bulge - to - total ratios are quite consistently returned by all four fits to the @xmath9 images , bulge sizes and srsic indices show much larger variations from one fit to the other . this is illustrated in figure [ fig : gim2d_bruns ] , where we compare the different measurements obtained for the @xmath9-band in the four cases @xmath94 , @xmath96 , @xmath98 , @xmath101 . the robust dispersion around the identity line in the four fits is @xmath105 kpc and 8@xmath27 for @xmath69 and @xmath100 , and @xmath860.6 kpc and 1.6 for the bulge half - light radii and srsic indices @xmath80 , respectively . especially for the bulge @xmath80 values , models with the isophote s position constrained and unconstrained can indeed provide very different results . similar conclusions on the reliability of the bulge and disk parameters are drawn from tests on simulated galaxies which we discuss in section [ sec : bulgediskbias ] . not surprisingly , the success rate for the bulge+disk decomposition is lower than the one for the single component . the single component fits return robust measurements for @xmath106 of s0 , bulge - dominated spirals and intermediate - type disk galaxies in the both the @xmath9- and @xmath10-bands ( see appendix [ sec : testgim2d ] for a detailed summary for the individual morphological classes ) . bulge+disk decompositions in both @xmath9- and @xmath10-bands are available for @xmath107 of these galaxies . for galaxies with a reliable decompositions in both bands , figure [ fig : gim2d_comp2 ] presents the parameters obtained in the two pass - bands . partly by construction , the scatter around the identity line for the bulge - to - total ratio is less than @xmath108 in the vast majority of the cases , as shown on the top left panel of the figure . given that our morphological classification is based on the @xmath17band bulge - to - total ratios , this means that our classification would not change substantially if we had used the @xmath9-band measurements instead . the disk sizes obtained in the two filters agree well with each other , although disks are more extended in the @xmath9-band than in the @xmath10-band , as also discussed above . as anticipated , the bulge radii and indices have instead a broader scatter . note that , despite the scatter , in the vast majority of cases the srsic indices for the bulge are consistent with either a ` concentrated ' or a ` diffuse ' central bulge component in both filters ; truly discrepant results are obtained in a small number of cases ( @xmath109 ) . finally , in appendix [ sec : testgim2d ] we show the comparison of our bulge+disk gim2d fits with similar obtained with the package galfit @xcite , and also with bulge+disk+bar fits ; the latter provide an estimate for the impact of a bar component on the bulge ( and disk ) parameters for disks which host such a third component . as discussed extensively in section [ sec : correctionmaps ] , the calculation of a galaxy size radius is made difficult by a number of observational biases which can lead to substantial uncertainties . here we additionally show how the estimation of the galaxy half light radius depends on the choice of the specific model to the galaxy light , by comparing the global galaxies half - light radii derived from the single srsic fits with those obtained from the bulge+disk decomposition . this is shown in figure [ fig : re_sersicbd ] for the @xmath10-band , but similar results are obtained in the @xmath9-band . it is clearly seen that for about @xmath110 of the galaxies with formally ` reliable ' bulge+disk decomposition and single component fits , half - light radii from the single srsic models are larger by more than a factor 1.5 than the sizes inferred from the bulge+disk decompositions . as illustrated in the inset in figure [ fig : re_sersicbd ] , the majority of the discrepant galaxies have steep light profiles with sersic index @xmath111 . we note that for all these galaxies both the single and double component fits were inspected to confirm their formal reliability ; furthermore , the magnitude difference between the model galaxy and the real zens image within an aperture equal to 1.5 times the petrosian radius is @xmath112 mag for both the single and double component fits . for these galaxies we decided to keep both radii estimates in our zens catalogue ( published with paper i ) but , to keep memory of the discrepancy , to add to the formal errors an uncertainty equal to half the difference between the two radii estimates . as an indication that these difference in the measured radii are not peculiar to the zens sample only or to the gim2d software , we remark that other recent publications have found evidence of an overestimation of the half - light radius when a single srsic fit is performed on a galaxies which have intrinsically a bulge and a disk component , for simulations reproducing sdss noise properties and also using different fitting algorithms @xcite . we further used the _ zurich estimator of structural types plus _ ( zest+ ) , an upgrade of the zest approach published in @xcite , to derive non - parametric measurements of structure for the zens galaxies . such estimates add further information on the structural properties of the galaxies , are useful in cases where no parametric fit could be performed , and enable comparisons with other published samples . zest+ is a -based code for the study of galaxy structure and it is designed for automated morphological classification of galaxies through either a principal component analysis or a support vector machines technique ; it features several improvements relative to the zest algorithm and implementation of scarlata et al . ( 2007 ; see for example @xcite for a first application to the cosmos field ; @xcite ) . the classification scheme uses both user supplied parameters and/or a set of non - parametric morphological coefficients computed by zest+ itself . for the zens galaxies we employed zest+ only to derive the structural coefficients , rather than using the morphological classification option , since , given the limited size of our sample , we were able to perform visual and quantitative checks on the reliability of the classification for each galaxy individually , which we mainly based on the available bulge - to - disk ratios ( see section [ sec : morphclass ] ) . the measurements provided by zest+ are galaxy concentration @xmath113 , asymmetry ( @xmath114 ) , gini and m@xmath8 coefficients , and smoothness ( @xmath115 ) . these parameters are widely used in the literature to study galaxy structure and morphology ( e.g. @xcite , @xcite , @xcite , @xcite ) . for sake of brevity in the following we will refer to the @xmath4 , @xmath114 , gini , m@xmath8 and @xmath115 set of measurements as to the @xmath116 " parameters . the concentration @xmath117 was defined as the ratio of elliptical radii containing @xmath118 and @xmath119 of the total flux of the galaxy , which is provided as input by the user . for zens , the sextractor kron flux ( ) was employed . the asymmetry index @xmath114 gives information on the degree of rotational symmetry of the galaxy light . this is parametrized through the difference between the original galaxy image and a 180@xmath43-rotated version of it . to account for the effect of background noise , zest+ uses the procedure introduced by @xcite , which involves the calculation of the asymmetry for both the original image @xmath120 and for a smoothed version of it , @xmath121 , where @xmath10 is the intensity of the image on the pixel @xmath122 and @xmath123 the intensity of the rotated image . the final asymmetry value is given by @xmath124 , where the last term corresponds to the background correction factor . the gini coefficient @xmath125 introduced by @xcite contains information on how uniformly the light is distributed within the galaxy : if the flux is equally distributed among all pixels , then @xmath125 is equal to zero , whereas if all the light is concentrated in just one pixel , the coefficient @xmath125 is equal to unity . we defined it as in @xcite : @xmath126 , where @xmath127 is the mean of the flux of the galaxy pixels , sorted in increasing order . the parameter @xmath128 is the normalized second - order moment of the brightest @xmath119 of the galaxy pixels . it describes the spatial distribution of bright substructures within the galaxy , such as spiral arms , bars or bright nuclei . the computation of @xmath128 involves the following steps : @xmath93 galaxy pixels are ordered by flux , @xmath95 for the @xmath119 brightest pixels , the sum of moments @xmath129 $ ] with respect to the light - center in the petrosian ellipse @xmath130 , @xmath131 is computed , and @xmath97 the latter is normalized by the total sum of moments to give @xmath132 , with @xmath133 . finally , the smoothness @xmath115 is a measure of the degree of clumpiness of the galaxy light distribution , and is thus useful to trace patches in the light profile such as star - forming regions . to calculate @xmath115 , a smoothed version of the original image , obtained by convolving it with a gaussian filter of fwhm equal to 0.25 times the petrosian radius @xmath134 , was subtracted from the image itself . clumpy regions were then identified from the residual image as those pixels for which the intensity @xmath135 is @xmath136 times higher than the background standard deviation in the residual image @xmath137 . we used a default threshold factor of @xmath138 . the pixels so identified were then used to calculate @xmath139 a region of radius @xmath140 from the center was masked out during the calculation to avoid including the highly concentrated centers of the galaxies which will boost the final value of the smoothness . the gini , m20 , @xmath114 and @xmath115 indices were all calculated within one petrosian ellipse . along with the structural indices , zest+ also gives an estimate for the elliptical aperture " galaxy half - light radius , based on the user - provided total flux . these are the measurements we refer to when discussing zest+ half - light radii for our zens galaxy sample . the careful inspection and filtering of the measurements thus far presented , although necessary , does not provide a quantification of systematic errors in the derived structural parameters . in zens , and more generally in ground - based imaging galaxy surveys , the major contributors to such errors can be identified in the noise of the night sky , and in the psf width . in zens , the median @xmath10-band psf fwhm is @xmath141 , which is about 30% of the median half - light radius of the galaxy sample . both analytical surface brightness fits and non parametric algorithms are affected by these observational limitation and none is completely free from pitfalls : the zest+ measurements are not psf - deconvolved and , to obtain a consistent measure of structure on the different zens fields , an homogenization to a common resolution of the @xmath116 parameters is necessary . in contrast , gim2d uses the provided psf to derive , in principle , seeing - corrected quantities , and indeed biases related to psf - blurring are less severe in gim2d - based measurements ; however , systematic uncertainties remain , especially in low surface brightness regimes , where both aperture photometry and two - dimensional fits are well known to underestimate galaxy sizes and fluxes ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? we asses here the impact of these effects specifically for our zens measurements through tests on artificial galaxy images . in particular , we derive recipes to correct the observed galaxy structural parameters ( e.g. , sizes , concentrations , ellipticities , etc . ) for biases in the observations . in areas of parameter space where we can not recover the true values of the given parameters , we provide an estimate for the systematic uncertainty that affects the measurements . we perform this analysis on the @xmath10-band , which provides a view of the intrinsic galaxy structure less affected by dust or young stars thus is used as our fiducial band to classify galaxies structurally and morphologically . the corrections in the @xmath9-band are then derived from those obtained for the @xmath10 measurements , suitably rescaled to the @xmath9-band luminosity and psf . the derivation of the corrections is done by applying the entire process of object extraction , followed by parametric and non - parametric photometric / structural measurements with sextractor / zest+ and gim2d , in exactly the same way as on the real data , to a set of artificial galaxy images for which the intrinsic structural / photometric properties are know precisely by construction . the difference between the model input and output parameters provides an estimate of the uncertainties in the measurements . a self - consistent study of the errors in the quantification of the galaxy structural properties must be a function of five parameters : the galaxy luminosity ( magnitude , @xmath142 ) , the galaxy size ( half - light radius @xmath64 ) , the inclination ( ellipticity @xmath38 ) , the steepness of the light profile ( srsic index @xmath80 or concentration @xmath4 ) and the psf under which the galaxy was observed . the effects of each of such parameters on the quantification of structure , including galaxy sizes , are tightly interconnected and hence need to be considered simultaneously . it is immediately clear that , as an example , the impact of the psf is stronger with all other parameters fixed for highly concentrated galaxies than for those with shallow light profiles , and for galaxies with sizes comparable to the seeing than for more extended ones . we thus adopt a sampling approach in which we construct many thousand artificial galaxy models to fully explore the observed parameter space of galaxies in the zens data set , and such to have , for each combination of the five parameters @xmath143(or @xmath144 , a sufficient number of models on which to test the measurements . this approach is also used in @xcite on a higher redshift sample of galaxies extracted from the cosmos survey . two cautionary remarks need to be made : first , our artificial galaxies are generated to populate _ uniformly _ a broad grid in the five - dimensional @xmath143(or @xmath144 space . real galaxies are not uniformly distributed in this parameter space , and this is not taken into account in our corrections . second , we deliberately ignore dust or stellar population segregation effects , as our models are created smooth and neglect dust attenuation . in the following we will test the robustness of our corrections towards the specific design of the simulations . to derive our correction functions for the structural parameters of zens galaxies we created both srsic models and bulge+disk artificial galaxies on which we tested the corresponding gim2d fits . in both cases , models were constructed on a grid of points in the ellipticity , magnitude , size and srsic index ( or b / t ratio ) parameter space . specifically , in the single component case galaxies were simulated around the following regions : @xmath145 $ ] kpc , @xmath146 $ ] , @xmath147 $ ] and @xmath148 $ ] . for each grid node 30 models were generated randomly to have : radii and srsic indices within @xmath149 30@xmath27 of the nominal radius and index @xmath80 at the grid point , magnitude within @xmath149 0.25 magnitudes and a difference in ellipticity equal to @xmath149 0.05 . this makes a total of @xmath150 single - component model galaxies . for the double component model galaxies , which span a wider range of combination of bulge and disk parameters and are also computationally more expensive , we used a coarser sampling of the parameter space , creating models in the following way : given a value of the magnitude for the _ entire _ galaxy randomly generated around the points @xmath151 $ ] and a disk scale length @xmath69 similarly chosen within @xmath152 from the positions @xmath153 $ ] kpc , artificial galaxies were constructed in three bins of bulge - to - total ratios centered at values of @xmath154 and of width @xmath155 , hence allowing a maximum @xmath156 . the bulge half - light radius was selected on the same grid used for the disk scale length , but imposing that @xmath157 . the bulge srsic index was allowed to have values @xmath158 with a scatter of @xmath159 and three bins of disk ellipticity ( i.e galaxy inclination ) @xmath160 were employed . we furthermore assumed that bulges can not be very elongated and hence explored only two values of the ellipticity for the bulge component , namely @xmath161 and @xmath162 , so to bracket any other value in between . all model galaxies described above were then convolved with three psf sizes , for a total of 180000 single component models and @xmath163 double - component models . the three psf sizes were taken to reproduce the best , medium and worst zens seeing in the @xmath17band ( respectively 0.7@xmath164 , 1@xmath164 , and 1.5@xmath164 ) . the psf - convolved models were inserted with poisson sampling into sky - subtracted empty regions extracted from the real zens fields . to mimic noise in the reconstruction of the psf in the real data , when performing the gim2d fits on the simulated galaxies we provided as input to the code a rotated version of the psf originally used to convolve the artificial image . for brevity , we focus in the following on the results obtained for the simulations convolved with the median zens psf , which are hence representative of the bulk of the zens observations . in appendix [ app : corrections_psf ] a full account of the results derived with the best and worst psf can be found ( see figures [ fig : gim2dcorrections_best ] , [ fig : gim2dcorrections_worst ] , [ fig : zestcorrections_best ] and [ fig : zestcorrections_worst ] ) , which shows correction matrices that are consistent with those here presented . note that the corrections to each zens galaxy were obtained through linear interpolation , at the psf size value relevant for any given galaxy , of the correction matrices describing psf sizes that bracketed the psf in question . we start the description of the resulting correction functions by focusing on those parameters which our approach can self - consistently correct without recurring to additional information , namely galaxy sizes and magnitudes . we show in section [ sec : corrdegeneracy ] how for other structural properties , such as the concentration coefficient , the recovery of the true " values is instead less straightforward , and depends on the intrinsic distribution of parameters in the simulated sample ( and requires therefore a different approach ) . our fiducial measurements of galaxy sizes in zens are those that we derive from the gim2d analytic fits , which are less prone to systematic biases then those inferred with zest+ , as we show below . it is nonetheless worthy to discuss here also the results for the zest+ measurements , given that these measurements are used for the few zens galaxies for which no reliable gim2d fit could be achieved . after the extraction / fitting process on the artificial galaxies was completed , we produced calibration maps " for sizes and magnitudes as shown in figures [ fig : gim2dcorrections ] and [ fig : zestcorrections ] ( see also carollo et al . 2013b for further details on this calibration approach ) . the arrows in the maps show , at any point of the _ observed _ @xmath166 plane , the direction and strength of the correction which is needed to recover the intrinsic " galaxy size and magnitudes from the observed values ; these correction arrows are obtained by taking the median difference between the model nominal input parameters and the parameters measured with zest+/gim2d of all artificial galaxies in the given point in the grid . the correction matrices are binned in three separate panels of concentration / srsic index and three separate panels of ellipticity . the colors of the arrows gives the amount of scatter shown by the individual models around the median correction : in green are grid points where all corrections are coherent in strength , in red are shown those which have a high scatter and hence our correction is representative on average but not for the single models . we use these maps to derive corrections for observed magnitude and sizes in the real zens galaxy sample . the choice of a discretized but dense grid allows us to pinpoint the correction / uncertainty maps at well localized positions on the considered planes . as indicated above , the corrections for any given real zens galaxy are obtained by interpolation at the position of the zens galaxy in the _ observed _ @xmath167 space . to discuss figures [ fig : gim2dcorrections ] and [ fig : zestcorrections ] it is useful to identify three regions in the @xmath168 plane : @xmath18 the region populated by models which are close to the surface brightness limit of the zens study ( indicated with the dashed black lines in the plot ) , @xmath19 the region of well - resolved , high - signal to noise ratio measurements , and @xmath22 the region close or below the size of the psf ( highlighted with a gray horizontal line ) . above the detection and resolution limit , and at low ( @xmath169 , @xmath170 ) to intermediate ( @xmath171 , @xmath172 ) concentrations , both gim2d and zest+ perform fairly well , and only small corrections are needed for both @xmath64 and magnitude . not surprisingly , below the sky noise surface brightness , basically no galaxy can be recovered by both zest+ and gim2d , as models are a priori not detected during the sextractor source extraction ( necessary to define the total galaxy flux and the initial guess for the size needed as input by zest+ and gim2d ) . for galaxies with low concentration / srsic index , the detection rate falls rapidly to zero when the surface brightness limit is reached ( no model with @xmath173 and @xmath174kpc is recovered ) ; conversely , the centrally peaked light distribution in galaxies with higher concentration pushes the detection limit to slightly fainter surface brightnesses . close to the surface brightness limit , magnitude and sizes are severely underestimated for both gim2d and zest+ fits : sizes are typically smaller by more than a factor of two ( three ) and magnitudes are dimmer by half ( one ) magnitude for gim2d ( zest+ ) measurements . at @xmath175 , the zest+ aperture " measurements suffer from a strong underestimation of sizes and magnitudes at any signal - to - noise . this is a consequence of the well known tendency to miss a substantial fraction of the flux from the faint wings in steep light profiles when performing aperture photometry measurements . integration to total light mitigates this effect in the gim2d models , which nonetheless results in sizes which are about @xmath176 smaller than the intrinsic ones for extended galaxies with steep light profiles ( @xmath177 kpc ) . when moving close to the resolution limit of the survey , we notice further differences between the performance of zest+ and gim2d . thanks to the psf deconvolution , the analytical fits are able to reliable recover the sizes also for models with sizes below the psf fwhm . systematic effects become visible only at the worst observing condition ( see figure [ fig : gim2dcorrections_worst ] in appendix [ app : corrections_psf ] ) . zest+ suffers from much stronger biases in this regime , causing an artificial increase of the size of models with @xmath178 for any value of the seeing . we note however that our approach , being based on idealized , regular galaxy light distributions produced with gim2d , may return in general an optimistically good performance of gim2d than when recovering the parameters for real galaxies with irregular , clumpy light distributions . our corrections are thus to be considered as the minimal " correction functions that must be applied to the data to put the structural measurements on a comparable grid . another clear effect observed for the zest+-based parameters is a lack of recovered parameters in the galaxy half - light radius @xmath64 vs. magnitude plane at high concentrations and small radii ( see right most panels of figure [ fig : zestcorrections ] ) ; this becomes increasingly more severe at higher ellipticities and larger psf fwhm ( see also figures [ fig : zestcorrections_best]-[fig : zestcorrections_worst ] ) . a similar trend is observed for the gim2d fits in the worst seeing conditions and at small radii ( figure [ fig : gim2dcorrections_worst ] ) . note that , consistently , the distribution of the real zens galaxy measurements based on zest+ in figure [ fig : zestcorrections ] presents the same bias . the origin of such effect is investigated in figure [ fig : zest_scatter ] , which illustrates where the artificial galaxies , created in a given region of the parameter space , are placed in the _ observed _ space . in this map , the arrows indicate the direction in which models generated with a certain @xmath38 and @xmath4 are displaced relative to their intrinsic concentration and ellipticity : an upward - pointing arrow indicates that models are observed on average as less elongated , and a left - pointing arrow indicated that they are observed at lower concentrations . intrinsic concentrations are calculated analytically from the original models srsic indices , and refer to the ratio of the radii containing @xmath118 and @xmath119 of the _ petrosian _ flux ( see e.g. @xcite ) . in the same figure , red - to - yellow squares indicate points in the grid in which at least @xmath179 ( up to 100% for full red squares ) of the artificial galaxies that are _ observed _ at that location originate from a different intrinsic concentration or ellipticity grid point . green points in the figure indicate grid nodes which are not affected by either scattering of galaxies into the grid point from different intrinsic concentration / ellipticity panels , or by scattering of galaxies out of the grid point into a different panel of observed ellipticity and/or concentration . it is clearly seen in figure [ fig : zest_scatter ] that model galaxies which were originally generated at high concentrations and ellipticities have a high probability to be scattered into lower @xmath38 and @xmath4 bins by the aperture - based zest measurements ; this bias is exacerbated at sizes @xmath180 kpc . this is caused by the effect of the psf convolution which artificially lowers a galaxy concentration and circularizes their axis ratio . consequently , this zest+ region of the @xmath168 plane out to @xmath181 psf radii is highly degenerate , being populated by both galaxies which have intrinsic parameters in that location of parameter space , and by galaxies which are scattered into it from higher ellipticity and/or concentration regions . given this degeneracy , it is important to verify to which extent the corrections for radii and magnitude that we discussed above depends on the precise way in which the simulation grid is populated . if model galaxies which are scattered in " and those which have intrinsic parameters in that location of parameter space required different corrections , then a precise modeling of the relative fractions of such galaxies in the given grid point would be needed . to test this , we created magnitude and sizes correction maps using either only artificial galaxies which were scattered into the given @xmath182 bin , or only galaxies born in situ . we show the outcome of such an experiment for the median psf in figure [ fig : zestcorrections_withnocontaminers ] in appendix [ app : corrections_psf ] . both realizations resulted in very similar corrections for size and magnitudes , indicating that these are robust independent of the intrinsic concentration / ellipticity of the galaxies . on the other hand , models which are scattered into lower @xmath4 or @xmath38 bins require by definition stronger corrections in @xmath38 and @xmath4 themselves than those which were generated within the bin . while for ellipticity it is reasonable to assume that also real ( disk ) galaxies would have an uniform distribution , and we hence can consider our corrections to be representative , the correction for concentration depends on the relative fraction of truly low concentration galaxies with respect to galaxies with a high intrinsic concentration . given that we have a no priori knowledge of the true distribution of concentrations for the real galaxies , and to avoid introducing biases associated to the choice of the simulation grid , we choose to follow a different method to correct the non - parametric structural estimators , which uses the available information on the srsic index . for data sets which do not have an as comprehensive set of measurements as zens , this approach could not be applied , and a statistical modeling of the underlying distribution of galaxies will be needed . these results should be taken as a cautionary note in using solely the concentration index as a morphological discriminant for galaxy types , especially if galaxies are close to the resolution limit of the given survey and if no correction to this parameter is attempted . conversely to the zest+ measurements , an equivalent map as in figure [ fig : zest_scatter ] , but for srsic indices and ellipticities measured by gim2d ( not shown ) demonstrates instead only a marginal contamination / scattering of galaxy models across the broad ellipticity and @xmath80 bins , thank to the psf - deconvolution performed by the gim2d fitting algorithm . for this reason , we apply to the ellipticities and srsic indices the corrections obtained by interpolating between the relevant grid points of the previously discussed vector maps , which we regard as statistically representative of the average correction . we stress however that systematic effects are observed for model galaxies convolved with the worst psf of zens : in this case , models with high srsic indices and sizes smaller than @xmath183 the psf size are scattered to lower @xmath80 . interestingly , gim2d tends to overestimate the ellipticity of such galaxies as opposed to what observed for the measurements performed with zest+ . furthermore , although under typical observing conditions the error on ellipticity and srsic index measured with gim2d are small enough to keep model galaxies within the same bin in these parameters , moving closer to the resolution or detection limit of the zens wfi images increases the randomicity in the measurements ; inferred ellipticities and srsic indices have a typical scatter of @xmath184 and @xmath185 , respectively . we highlight these problematic regions with empty symbols in figure [ fig : gim2dcorrections ] . we finally note that a psf correction is often included typically as a convolution - kernel in surface brightness fitting algorithms in structural studies based on other ground - based datasets , including the sdss ( whose typical psf width is similar to the worst zens psf ; see e.g. , @xcite ) . as we show above , it is possible however that further residual ( size- , magnitude- , ellipticity- and concentration - dependent ) corrections to the structural parameters may be needed , depending on the specific psf , signal - to - noise and redshift of the studied samples . to overcome potential biases associated with calculating a correction for the @xmath116 parameters which may depend on the distribution of intrinsic structural properties , we use the srsic indices of the galaxies from the gim2d fits as a prior . srsic indices are robustly determined in the vast majority of the models and are less prone to systematic biases , as discussed in the previous section . the corrections for concentration , gini and @xmath128 indices are thus derived by splitting the artificial and real galaxies in similar bins of magnitude , radius , ellipticity and psf as those employed for the size corrections ( see figure [ fig : zestcorrections ] ) . in this case however we characterize the galaxy structure according to the observed srsic index rather than observed concentration , dividing the samples in three broad bins of @xmath186 , @xmath187 and @xmath188 , respectively . following the same approach as for sizes and magnitudes , the corrections for non - parametric structural estimators are then defined as the median difference between the models intrinsic indices and those calculated with zest+ at the given position in the observed @xmath189 parameter space . the intrinsic concentration is computed analytically from the input srsic index , as specified above , while for the gini and @xmath128 indices we use the measurements performed on pure " models that are neither psf - convolved nor degraded with the zens typical noise . given that our analysis is based on intrinsically smooth and axisymmetric models , we do not attempt to correct the asymmetry and smoothness index . these quantities will nonetheless be affected ( the smoothness possibly by the largest amount ) , and this caveat should hence be kept in mind . for the few zens galaxies for which no gim2d fit is available , we applied an average correction obtained as the median of the correction for the zens galaxies having gim2d fits and similar observed concentrations . a similar approach as the one employed for the single srsic fits was used to test the reliability of the gim2d bulge+disk ( b+d ) decompositions . all the b+d model galaxies were processed in exactly the same way as the real galaxies : i.e. , after the decomposition , the output models were filtered to reject the unphysical fits and to identify acceptable b+d decompositions ( see appendix [ app : filterbd ] ) . consistently with the approach used for the real galaxies , only these were used for deriving the correction functions for the b+d parameters . the correction maps for the b+d parameters are shown in figure [ fig : doublecorrdisks ] and [ fig : doublecorrbulges ] , for the three psfs . for bulges , we present the results for all disk inclinations together , as no strong dependence on the model axis ratio was detected . both scale lengths and magnitudes of disks are very well recovered by the gim2d code for most of the cases . systematic deviations from the input parameters are observed only for very small disks ( i.e. , @xmath190psf ) , especially for the largest psf size , and in the low surface brightness limit , where disk scale - lengths are underestimated by a factor of about two , and the measured magnitudes are systematically fainter than the input ones , as already discussed above . globally , in only @xmath7 of the cases the normalized difference between input and output disk sizes is larger than @xmath191 , and predominantly in the low surface brightness regime . the bulge parameter are instead subject to much larger uncertainties , as illustrated in figure [ fig : doublecorrbulges ] . although in many regions of the magnitude - size relation there are no systematic corrections , the scatter of measured bulge sizes around the input ones is generally large , typically @xmath192 for the measured bulge half - light radii . the strongest differences between input and output model parameters are measured for bulges with high srsic indices ( @xmath193 ) in which case a substantial over - estimation of bulge effective radii is observed . low srsic index bulges ( leftmost panels in the figure ) suffer from smaller measurement errors . below the resolution limit and for the worst psf , sizes are over - estimated also for bulges with @xmath194 whereas the effect is less evident for the best or intermediate psf . these uncertainties in the bulges sizes reflects in ( or are possibly generated by ) a general difficulty in recovering the correct bulge srsic index . as an illustration of this fact , we plot on the top panel of figure [ fig : btscatter ] the comparison between input and measured bulge srsic indices for the median psf . it can be noticed how the measured srsic index , especially below @xmath195 , can deviate substantially from the model intrinsic values . this particularly severe for bulges with half - light radii which are close to the psf size : small bulges with low index @xmath80 can be misclassified as bulges with larger half - light radii and steeper light profiles . although the structural properties of the bulges are subject to relatively large uncertainties , the fractional contribution of the bulge to the total light is generally well measured by the gim2d decompositions , as shown on the middle panel of figure [ fig : btscatter ] . in the bottom panel we plot , for a given value of the observed bulge - to - total ratio , the fraction of input models which where originated with a @xmath100 which differed less than 0.15 from the measured one , between 0.15 and 0.3 and more than 0.3 ( black , gray and red lines respectively ) . in @xmath8680 - 85@xmath27 of the cases the bulge - to - total ratio is recovered within a scatter of 0.15 and really catastrophic failures ( @xmath196 ) happen in @xmath197 of the models . it is worth to notice that the fraction of galaxies for which the @xmath100 is robustly recovered is largely independent of the value of @xmath100 itself , and the typical scatter around the input bulge - to - total ratio is of order @xmath198 . problematic fits which results in large differences in @xmath100 are again mostly associated with flat ( @xmath199 ) and small ( @xmath200kpc@xmath201psf ) bulges , for which is difficult to disentangle the disk component from the small , disk - like bulge . in the light of these results we decided _ not _ to apply any corrections to the bulge and disks structural parameters : on the one hand disk sizes are well recovered by the gim2d fits , hence no substantial correction is needed ; on the other hand , the large scatter in the recovered bulge half - light radii and srsic indices make the derived corrections noisy and dependent on the specific sampling of parameter space with our simulations . for this reason , we prefer to use the direct output from gim2d for the bulge half - light radii , but with associated a typical uncertainty of 30@xmath27 , as estimated from our tests . in the top panel we show the comparison between the intrinsic bulge srsic indices for the set of simulated galaxies used to test the reliability of gim2d bulge+disk decompositions , and those recovered by gim2d . the variation @xmath202 corresponds to the difference between the intrinsic and measured values . with black points we highlight models for which the structural parameters are well recovered , i.e. , have relative error @xmath203 . in dark gray are the points which have a fractional error between @xmath179 and @xmath204 , and in gray ( red in the online version ) are the extreme outliers with errors which are @xmath205 . the reported fractions are the fraction of discrepant models relative to the total in the three bins of srsic index @xmath80 shown in the figure . in the central panel we show the comparison of intrinsic and measured bulge - to - total ratios ( b / t ) for the model galaxies . black shows models for which @xmath206 , dark gray those with @xmath207 and gray ( red in the online version ) is for models for which the measured b / t deviates more than 0.3 from the intrinsic value . the fraction of models within each of the three types , as a function of the _ observed _ b / t , is shown in the bottom panel . ( a color version of this figure is available in the online journal.),width=264 ] we present a number of diagnostic tests that we performed to verify the reliability of the corrections for the structural parameters that are described in section [ sec : correctionmaps ] and [ sec : ginicorr ] . as a first sanity check we run our correction scheme , in exactly the same fashion as for the real galaxies , on the measured properties for the models themselves ; the outcome of this exercise is illustrated in the upper panels of figure [ fig : zest_gim2dcorectioncomp ] for sizes obtained with gim2d and zest+ , for models convolved with the median psf . the intrinsic ( input ) and corrected radii for the model galaxies agree very well , and systematic biases that are present in the uncorrected sizes are largely cured by our correction scheme . furthermore , the large discrepancies between the zest+ and gim2d raw measurements disappear in the corrected data . the bottom panels in figure [ fig : zest_gim2dcorectioncomp ] show the comparisons of zest+ and gim2d half - light radii , ellipticities and @xmath17band magnitudes , before and after the application of our correction schemes for the real zens galaxies . the raw sizes derived by zest+ are systematically smaller than those obtained by the gim2d fits , by up to a factor of @xmath208 in the worst cases . our correction scheme nicely brings the two estimate into a very good agreement . as a consequence of the psf convolution , galaxies are also measured to be rounder in zest+ than in gim2d , especially at low radii , a bias which is largely cured by the implementation of the correction scheme . the latter also mitigates the underestimation of the total fluxes in zest+ . the comparison between uncorrected and corrected @xmath10-band @xmath4 , gini and @xmath128 indices is shown in figure [ fig : zest+_corrected ] . specifically , to test whether our corrections deal properly with psf - induced biases , we compare , for corrected and uncorrected measurements , the distributions of these parameters for two bins of psf size , i.e. , psf @xmath209 and @xmath210 , respectively . the uncorrected indices show different distributions for galaxies observed with small or large psf s fwhm . the effect is mostly evident for the concentration parameter , for which a two - sided kolmogorov - smirnov test rejects the possibility that the low and high fwhm samples are drawn from the same parent distribution at the @xmath211 confidence level . a similar effect is also noticeable for the @xmath128 index ; the gini coefficient is instead less sensitive to psf blurring . all such biases disappear once our corrections are applied , as shown in the lower panels of the same figure . the inspection of the distributions of figure [ fig : zest+_corrected ] furthermore shows that the corrections recover the peak of high @xmath4 values for the early - type galaxies , which would otherwise be absent from the raw measurements ; similarly , the peak at more negative " values is recovered in the corrected @xmath128 distributions . this is also highlighted in figure [ fig : zest_vs_sersic ] , where we show the comparison between corrected and uncorrected non - parametric structural estimators and the corrected srsic indices . the upper - left panel shows the comparison for the concentration index : the solid red line marks the expected values of the concentration inside the petrosian radius for a perfect srsic profile of a given index @xmath80 . it is clear that before correction , the measured concentration flattens rapidly at values @xmath212 for @xmath213 , lying far away from the theoretical line . after applying our correction , the measured points well match this line . a similar adjustment is also observed for the @xmath128 index , while , as already mentioned , the gini coefficient is measured quite robustly and needs virtually no corrections . the upper right inset show the dependence on ellipticity of uncorrected concentration and @xmath128 parameters for galaxies classified as disks in section [ sec : morphclass ] ; the spurious lack of high concentration galaxies at high elongation is largely ameliorated by our correction scheme . we stress in concluding that all our corrections are clearly statistical " . for example , individual models can retain an underestimation of the radius of up to a factor @xmath176 , especially models close to the surface brightness limit of the zens observations . nevertheless , in a statistical sense , our correction schemes return well - calibrated measurements that can be robustly compared with each other . we show in figure [ fig : shencomp ] a comparison between the mass - size relation obtained for the zens galaxies , both before and after applying the corrections outlined in section [ sec : correctionmaps ] , and the relation derived by @xcite in sdss . the shen s sizes are from the srsic fits of @xcite , and are thus similar to the sizes of the vagc catalogue ( see appendix [ sec : testgim2d ] for a one - to - one comparison between our measurements and the blanton et al data ) . circularized apertures are used in this comparison , to match the shen et al . definition . not surprisingly , given the global robustness of the structural parameters of disks , and more generally low srsic index galaxies , the @xcite size - mass relation for the late - type galaxies is in excellent agreement with our relation . the shen s sizes for the early - types are however systematically underestimated relative to our measurements , a difference which is further increased after the application of our corrections . in this respect , we note again that the sdss data were typically acquired with a typical @xmath214-band psf of @xmath861.5@xmath215 , which is comparable to the worst seeing in the @xmath17band zens data , and no residual corrections as a function of size , magnitude , concentration and inclination effects ( as done in this paper ) were attempted in the sdss analysis . using our uncorrected sizes , the difference in size estimates is @xmath216 at @xmath217 , and increases towards lower galaxy masses , up to @xmath218 at @xmath219 . the median size of the shen et al . early - type galaxy at this latter mass scale is @xmath220kpc , while our corresponding estimate is 2.2 ( 2.4 ) kpc before ( after ) the application of our corrections . we attribute the difference in size estimates between our analysis and shen s to the combination of three factors : @xmath93 in zens the half - light radii are obtained using elliptical apertures , in contrast with the circular apertures used in the sdss study . although we have partly corrected for this difference in a post - processing mode , by circularizing our measurements for the comparison in figure [ fig : shencomp ] , residual differences may remain ; these differences are likely more relevant at relatively smaller sizes and relatively higher srsic indices ; @xmath95 the use of different upper limits for the srsic index in the sdss and zens fits , see appendix [ sec : appnyvagc ] ; @xmath97 different morphological mixes in the two samples , since shen et al . use a simple split in srsic index to define their morphological classes ; and @xmath99 finally , of course , to the fact that the shen et al measurements were not corrected by observational biases as we do for our zens sample . the availability of a large suite of structural diagnostics for the zens galaxies enables us to perform an accurate and quantitative morphological classification , eliminating biases and dilution of signal in studies of galaxy properties as a function of morphology . we classified the zens galaxies into six morphological classes , primarily based on the prominence of the bulge component in the i - band . these classes are , respectively , ellipticals , bulge - dominated disks further divided into s0 and bulge - dominated spirals intermediate - type disks , late - type disks and irregular galaxies . the stamp images of the galaxies of different morphological types are shown in figures [ fig : stampe]-[fig : stampmergers ] in appendix [ app : classdist ] . a flow - chart schematic description of the steps applied in our morphological classification scheme is shown in figure [ pic : classflowchart ] , in which we also list the fraction of each morphological type in the zens database . in detail , our classification criteria are as follows . ellipticals are required to be well fitted by a single srsic profile with _ corrected _ srsic index @xmath221 and normalized residuals smaller than @xmath110 out to the sky level of the image ( see e.g. @xcite ) . we present the @xmath10-band surface brightness profiles for all elliptical galaxies in figure [ fig : ellipticalprofiles ] : by definition , the light distribution of such galaxies is perfectly represented by a single ( corrected ) @xmath221 srsic model . the distinction among different disk types is based on the following criteria : galaxies with @xmath222 are assigned to the late - type disk class , those with @xmath223 are classified as intermediate - type disks and those with @xmath224 as bulge - dominated galaxies . for many reasons , including a good degree of mixing of physical classes in the visual classification of the third reference catalog ( rc3 , * ) , we will refrain from using its popular nomenclature . we note however that , albeit with scatter , our bulge - to - total separation for disk galaxies roughly corresponds to a division in types s0 plus s0a / sa , sab / sbc and sc / sm , in order of decreasing b / t range . both s0 and spiral galaxies with large bulges fall in the bulge - dominated galaxies category . the distinction among the two is based on the @xmath9-band smoothness parameter @xmath225 , which is set to @xmath226@xmath227 for s0 and larger than this value for early - type spirals , except in those case where faint spiral arms not prominent enough to cause a variation of the smoothness parameter are clearly seen in the visual inspection of the images . for nearly edge - on system it is obviously impossible to determine the presence or absence of spiral arms and the difference between s0 and bulge dominated spiral galaxies becomes more challenging . given that we have no means to securely distinguish these morphological types at these high inclinations , we assigned to the bulge - dominated spiral class those galaxies which satisfy the b / t criterion above , and either present a dust lane in the disk plane or have a high @xmath16band smoothness parameter ( @xmath228 ) . as seen in figures [ fig : stamps0 ] and [ fig : stampsa ] , a large fraction of the highly - inclined galaxies are classified as bulge dominated spirals , and only those which clearly do not have a dusty disk are defined as s0 . the contamination of s0 galaxies into the bulge - dominated spiral class is hence likely non - negligible at the highest inclinations ; a comparison between the relative fractions for face - on galaxies estimates this contamination at the 33% level . the distinction between s0 and elliptical galaxies is also non trivial , this time especially in face - on systems . to attempt to distinguish between these two classes we inspected all the residuals of the single component fits and any sign of a faint disk component was used as a discriminant diagnostic . finally the @xmath9-band smoothness @xmath115 parameter was used to further validate the separation between the two types , by requiring that for s0 @xmath229 , a condition which is however secondary to the fact of s0 not being perfectly fit by a single sersic model . we note that for @xmath230 of all zens galaxies that were not classified as ellipticals or irregulars , and that would therefore be assigned a disk " classification , it was not possible to obtain a reliable @xmath10-band bulge+disk decomposition ( see appendix [ sec : testgim2d ] ) . in these cases the classification is based on the corrected srsic indices from the single component fits , as shown in the right hand side of figure [ pic : classflowchart ] . for the very few galaxies ( @xmath231 ) for which we could not obtain any reliable bulge+disk decomposition or srsic index , we used the _ corrected _ zest+ parameters ( in particular the concentration ) to define their morphological class . specifically , in such few cases , galaxies with @xmath232 were assigned to the late - type disk class , those with @xmath233 to the intermediate - type disks and galaxies with @xmath234 to the bulge - dominated spiral class . the precise values of these cuts were determined through the relation between @xmath80 and @xmath4 in figure [ fig : zestclasses ] . irregular ( i.e. , thus possibly disturbed by mergers / encounters ) galaxies are identified in the zens sample by visual inspection ; in this category we included those galaxies which have no clear disk / bulge component or have a highly disturbed morphology characterized by multiple clumps , as shown in figure [ fig : stampmergers ] . there are furthermore 26 galaxy - pairs or triplets in the zens sample which are also possible merger candidates , of which we show stamp images in figure [ fig : stampmergers ] . these galaxy pairs are identified as single objects in 2dfgrs , and appear as single members in the group catalog , i.e. , they have only one redshift measurements in the 2dfgrs catalogue . for nine of such systems , we confirmed with sdss @xcite spectral or photometric redshifts , or with spectroscopic data available in the ned database , that the two galaxies are at the same redshift , strengthening the hypothesis that they are undergoing a merger event . there are 17 remaining galaxies pairs , for which no such information is available . we note that five pairs / triplets show tidal features and disturbed morphologies , supporting the merger scenario ; the remaining 12 pairs display less clear signs of interactions , and hence for them a chance projection can not be excluded . we will include nonetheless the latter in the sample of merger galaxies and , when necessary , test our results with and without these systems ; these are flagged as ` possible projections ' in our zens catalogue . for consistency with the original 2dfgrs and 2pigg catalogs , we count the merging galaxies as a single galaxy pair system in the sample of 1455 galaxies , but we compute and provide photometric / structural properties for the individual members ( thus making the total entries in the public zens catalog equal to 1484 ) . in the normal " disk population , there are furthermore some galaxies which present clear tidal features and disk distortions ; such galaxies are flagged has having plausibly undergone a recent galaxy - galaxy interaction or merger . finally we also flag as candidate mergers those zens galaxies which , although identified as individual objects in the parent 2dfgrs/2pigg catalog , are found at a distance from another confirmed group member which is smaller or equal to the maximal separation observed in the above 26 galaxy pairs ( i.e. 48@xmath164 , roughly 50 kpc at the average zens redshift ) and have a velocity difference from it @xmath235 500km / s . the properties of merger galaxies in the zens sample and the connection with the local and large scale environment are investigated in a forthcoming publication ( pipino et al . 2013 , in preparation ) . ccccccc + e & 4.11/-/- & 0.60/-/- & -2.45/-/- & 0.005/-/- & 4.93/-/ & -/-/- + s0 & 4.24/4.19/- & 0.60/60/ - & -2.48/-2.49/- & 0.006/0.0064/- & 5.19/ 4.64 & 0.62/0.63/- + bulge - dom . spiral & 3.88/3.94/3.92 & 0.59/0.59/0.61 & -2.30/-2.37/ -2.41 & 0.033/0.043/0.096 & 3.90/3.94/3.39 & 0.59/0.58/0.60 + interm .- type disk & 3.54/3.42/3.16 & 0.56/0.55/0.56 & -2.20/-2.15/-2.11 & 0.020/0.043/0.10 & 2.91/2.52/2.01 & 0.36/0.35/0.31 + late - type disk & 2.58/2.60/2.56 & 0.48/0.49/0.51 & -1.80/-1.79/ -1.81 & 0.040/0.050/0.087 & 1.14/1.14/1.09 & 0.04/0.07/0.03 + our morphological classification scheme is primarily based on a bulge - to - total ratio separation . we inspect in figure [ fig : zestclasses ] the distributions of the _ corrected _ @xmath4 , @xmath80 , gini and @xmath128 for the various morphological classes , and we summarize in table [ tab : classmedian ] the median values of these corrected non - parametric diagnostics for each morphological class , separately in three bins of ellipticity ( i.e. , inclination for disk galaxies ) . note that the six morphological classes largely segregate in specific region of the structural planes . again we stress that this consistency between b / t and non - parametric diagnostics is reached only after correcting the latter as described in section [ sec : simulations ] : the raw measurements before the implementation of the corrections would substantially mix different galaxy populations ( e.g. , through biased concentration and @xmath128 estimates , see figures [ fig : zest+_corrected ] and [ fig : zest_vs_sersic ] ) a bias which could affect other published analyses based on ( uncorrected ) non - parametric diagnostics . there are of course some noticeable deviations which remain , even after correcting the non - parametric diagnostics . a small number of galaxies classified as intermediate- or late - type disks have low @xmath128 indices for their concentration or gini values , overlapping with the region of the plane which is mostly occupied by irregular galaxies . we visually inspected all these galaxies and verified that the @xmath128 deviations are mostly caused by the presence of bright clumps in the galaxies outskirts , but that a clear and ordered disk structure is present ( we highlight these galaxies with a * " in figures [ fig : stampsbc]-[fig : stampsd ] ) . also , some intermediate - type disk galaxies have relatively high concentration values ( @xmath236 ) . this demonstrate that concentration is _ not _ necessarily a proxy for the fractional contribution of the bulge component to the stellar light , as also pointed out by other authors ( e.g. * ? ? ? there is , of course , a correlation between @xmath100 and @xmath4 characterized by a linear correlation coefficient of @xmath237 . however , the scatter is large and a given concentration value can be associated with a broad range of b / t ratios , as illustrated in the top right panel of figure [ fig : c_vs_bt ] . we emphasize that the _ uncorrected _ values of concentration lead to a much larger scatter , which results in a weaker correlation between @xmath100 and uncorrected c with @xmath238 . we highlight with colors in figure [ fig : c_vs_bt ] two regions of parameter space which are worth noticing , i.e. , that of galaxies with @xmath239 and @xmath240 ( red points ) and that of @xmath241 and @xmath242 ( blue points ) . as illustrated in the bottom panels of the same figure , highly concentrated galaxies with low @xmath100 ratios are found below the bulk of the population in the relation between the bulge half - light radius and the disk scale length ; the srsic indices of their bulges are in most of the cases ( @xmath243 ) above @xmath244 . such objects hence host dense but small bulges , which are causing an increase in the concentration value but do not give a major contribution to the total galaxy light . the blue points at low @xmath4 and high @xmath100 , in contrast , are galaxies with typically extended bulges with low srsic index ( i.e. , pseudo - bulge like structures , see e.g. , @xcite and @xcite for a review and further references ) . in figure [ fig : inclinationc ] we furthermore show the different corrected structural indices for the morphological classes , as a function of the galaxy inclination . the distribution of corrected parameters for each individual class are fairly flat with ellipticity , as they should ( in contrast with the non - corrected parameters , which we have shown to have a bias with ellipticity ; see section [ sec : reliabilitycorrections ] and figure [ fig : zest_vs_sersic ] ) . note the galaxy smoothness maintains a substantial dependence on inclination by increasing towards higher inclinations ; this parameter has not been corrected , since we used only smooth galaxy models to determine our correction matrices . part of the dependence of @xmath115 on ellipticity might thus be a residual observational bias ; part might however be a consequence of the prominence of dust lanes in edge - on galaxies . dust absorption in the disk could also in principle artificially lower the derived bulge - to - total ratios for edge - on galaxies ( see for example @xcite ) , and hence introduce a bias in the b / t - based morphological classification . we expect this effect to be minimal in our analysis , since we use the relatively unaffected @xmath17band to derive our fiducial @xmath100 estimates , but it is nonetheless worthwhile to test whether there are indication of biases among the morphological classes given that some attenuation of the disk ( and bulge ) light is certainly occurring . in the lower right panel of figure [ fig : inclinationc ] we plot the distributions of ellipticity for disk galaxies with bulge - to - total ratio @xmath245 ( blue histogram ) , @xmath246 ( green histogram ) , and @xmath247 ( red histogram ) . for @xmath248 we find a reasonable constancy with ellipticity of the relative contribution of three disk types to the total disk population , indicating that there are no major biases in this regime . at higher axis ratios , there is a depletion of @xmath249 galaxies and an increase of the bulgeless " disk galaxies , which we interpret mostly as a consequence of a true change in the morphological mix rather than of strong dust absorption . the presence of a bulge component , even in highly - inclined disks , produces a rounding of the isophotes and thus of the measured axis ratio . therefore , the decrease in the number of bulge - dominated galaxies and intermediate - type disks at @xmath250 likely reflects this correlation between `` morphology '' and measured ellipticity . likewise , the peak of @xmath251 disk galaxies at @xmath252 can be explained by the rounding " of isophotes due to psf effects in these galaxies and the intrinsic galaxy thickness . indeed no galaxy is observed at @xmath253 and the fraction of @xmath251 disk galaxies with @xmath254 relative to the total sample of disk galaxies is @xmath8625@xmath27 , i.e. , if they were ( infinitely thin disks and ) equally distributed up to @xmath255 , each of the four 0.1-wide bins between @xmath256 would contain 6@xmath27 of the sample , which is the average number observed at lower axis ratios . to further test whether the estimates of galaxy bulge - to - total ratio are strongly affected by dust extinction , we compare in the bottom panel of figure [ fig : inclinationc ] the difference between the b / t values measured in the @xmath16band and those measured in the @xmath17band , as function of the galaxy ellipticity . in grey we show the full zens sample ; other colors are the results for galaxy samples split in the three b / t classes discussed as above . the @xmath17band b / t values are typically larger by about 5% than those in the @xmath9 filter , which is consistent with larger dust absorption effects at the shorter wavelengths . there is however no clear variation with @xmath38 of the difference @xmath257 within each of the three morphological classes . we thus conclude that dust effects are generally a modest contribution to the total error budget for the b / t measurements , which is dominated by other sources of random and systematic errors ( see figure [ fig : btscatter ] and the corresponding discussion in the text ) . we note that other authors have combined empirical disk attenuation - inclination relations with theoretical dust models in order to derive inclination - dependent corrections for dust absorption in bulge and disk fluxes ( e.g * ? ? ? * ; * ? ? ? using the @xmath10-band parameters in table 1 of @xcite together with their equations 1 and 2 , we estimate a correction of order of @xmath258 to the zens @xmath10-band b / t values , comparable with or even smaller than the typical uncertainty on the b / t measurements . while such corrections could be important in some contexts , they inherit the uncertainties on the adopted dust models and would not cause a significant shift of morphological class in the zens sample . we hence choose not to apply any such ( model - dependent ) dust correction , and to adopt the dust - uncorrected gim2d measurements of b / t as our fiducial estimates for this parameter . in summary , the above considerations , and the constancy of the median b / t ratio for the different morphological classes in three ellipticity bins shown in table [ tab : classmedian ] , lead us to conclude that our @xmath100 measurements are largely independent of axis ratio and that our morphological classification is thus largely unaffected by inclination effects . as a final remark , we note that the @xmath80 srsic indices of intermediate - type disks also show a residual ( but marginal ) dependence on ellipticity : this is of course the morphological class for which the likely similar light contribution of bulge and disk components along the line - of - sight is the most difficult to disentangle . nevertheless , the @xmath80 values spanned by these morphological class at the different ellipticities are consistently in the intermediate " range of @xmath259 ( see table [ tab : classmedian ] ) , providing a consistent characterization of these systems across the whole ellipticity range . the main goal of zens is to study galaxy properties , at fixed stellar mass and hubble type , as a function of several environmental parameters derived self - consistently for the same galaxy sample , i.e. , the host halo mass , the large scale density @xmath260 and the projected group - centric distance . we furthermore split the zens galaxy sample in central galaxies in their host group halos , and satellite galaxies , which orbit the central galaxies within those halos ( see paper i for the precise definition of central and satellite galaxies , and for the environmental parameters ) . several zens analyses are underway , which use the zens database , including the structural and morphological information derived in this paper , to investigate how galaxy structural and morphological properties vary across the different environmental regimes . in this second paper in the zens series we want to show a first utilization of the structural measurements presented above to answer , from a purely observational perspective , a simple question , namely how the concentration of _ satellite galaxies _ depends , at fixed stellar mass , on @xmath93 hubble type , and @xmath95 on group mass , group - centric distance and large - scale structure density . galaxy structure , as described by the concentration parameter , is a strong function of galaxy mass ( e.g. * ? ? ? * ; * ? ? ? * ) . as we have discussed above , concentration is however a relatively poor indicator of galaxy morphology ( when this is defined according to a more physical parameter such as the bulge - to - total ratio , as we do for the zens galaxies ) . furthermore , morphology also is a function of stellar mass ( see paper iii and also , e.g. , @xcite ) . with available robust ( corrected ) concentration values , and a b / t - based morphological classification , we thus investigate how much of the concentration vs. galaxy stellar mass correlation is to be ascribed to a variation in the hubble type of galaxies , and how much is actually driven by a real change in concentration within each individual morphological types . stellar masses are taken from paper iii . the results of this analysis will also help us interpret the analysis of satellites concentration in the different environmental parameters , for which , in order to gain statistics , we will use broader morphological classes . figure [ fig : cmass ] shows the corrected concentration as a function of galaxy stellar mass . the points connected with lines are the median concentrations for satellite galaxies , divided according to the morphological types ( note that s0 and bulge - dominated spiral galaxies are joined together into a single broad morphological bin of bulge - dominated " galaxies ) . the medians were calculated using a running box above the mass completeness of each morphological type ( the results are noisier but consistent when using independent mass bins ) . the inspection of this figure shows an increase in concentration with increasing galaxy stellar mass at fixed morphological type for disk galaxies . specifically , our elliptical satellites sample covers a small range of stellar mass ( @xmath261 ) , over which the concentration is observed to remain constant . for all types of disk galaxies where the range of masses covered is larger , a trend of concentration with stellar mass is clearly observable , separately for each morphological disk type ( the latter defined according to the galaxy bulge - to - total ratios as described in section [ sec : morphclass ] ) . we note that residual bulge - to - disk variations with stellar mass _ within _ each morphological disk class are possible , since our morphological classification was done independent of galaxy stellar mass , and each morphological bin covers a relatively broad range of bulge - to - total ratio . quantitatively , however , the a posteriori computation of the median bulge - to - total ratios vary , between the mass bin centered at @xmath262 and the mass bin centered at @xmath263 , only from 59@xmath27 to 60@xmath27 , 34@xmath27 to @xmath264 , and @xmath265 to @xmath266 , respectively for bulge - dominated , intermediate - type and late - type disk galaxies . thus , we conclude that the known concentration - mass correlation is not solely driven by a correlation between morphological type and stellar mass , but is , at least partially , driven also by a genuine increase in concentration of galaxies of similar hubble type , i.e. , of similar bulge - to - total ratio . we fit the observed trends of corrected concentration versus galaxy stellar mass with a linear relation , @xmath267 , in the mass range which is probed by all morphological type , i.e. , for those bins which are above @xmath28 for their full width . the global @xmath268 relation for the late - type disks shows a break around @xmath269 , below which mass the relation flattens ; a linear fit is however a good approximation at @xmath270 where we can make a meaningful comparison with the other types . the slopes @xmath271 for the four morphological classes above this threshold mass are , respectively , @xmath272 , @xmath273 and @xmath274 ; satellite elliptical galaxies are consistent with no dependence of concentration on the galaxy mass , i.e @xmath275 . although , as commented above , the lack of dependence of @xmath4 on stellar mass for the ellipticals is established only across a relatively limited mass range , generally speaking this finding follows the global trend , established on the disk galaxies , of a flattening of the @xmath4 vs. galaxy stellar mass relation from the later- to the earlier morphological type . at the masses of our study , there are essentially no irregular satellite galaxies ; furthermore , elliptical galaxies contribute only 5% to the total number of satellites in the zens sample at these masses . the small number of elliptical satellites does not enable us to perform a statistically significant analysis of their dependence on environment . since the vast majority ( @xmath106 ) of satellites in the sample are disk galaxies , we choose to maximize morphological `` purity '' in our analysis of the dependence of satellite concentration on the different environmental parameters by limiting it to to the sample of _ disk _ satellites only . in figure [ fig : cmassgrp ] we present the ( corrected ) concentration vs. galaxy stellar mass relation for disk satellite galaxies , split into two broad morphological bins of bulge - dominated " galaxies and disk - dominated " galaxies by adding together s0s and bulge - dominated spirals in one bin , and intermediate - type and late - type disks in the other bin . the relation is plotted separately as function of group mass @xmath1 , lss ( over)density @xmath3 , and group - centric distance @xmath276 ( from top to bottom ) . the broad bulge - dominated galaxies bin is split in three bins of galaxy stellar mass , i.e. , @xmath277[9.5,9.9 [ , [ 9.9,10.45 [ , [ 10.45,10.88 ] ; we use only two bins of galaxy stellar mass for the broad bulge - dominated galaxies bin , i.e. , @xmath277[10.,10.6[,@xmath278 . before discussing the results we present some considerations on the biases and uncertainties which may affect the analysis . first , we remind that trends with group mass that we measure are , intrinsically , possibly even stronger . as discussed in paper i , the uncertainty on our group mass estimates tends to wash out " trends with this environmental quantity ; quantitatively , real trends with group mass have slopes @xmath279 steeper on average than what we can measure in the zens sample . in paper i we furthermore discuss a classification of the zens groups into relaxed " and unrelaxed " groups , according to whether a self - consistent solution for a central galaxy can be found ( in relaxed groups ) or not ( in unrelaxed groups ) . the latter are expected to be combination of genuinely dynamically young , merging structures , contaminated by physically relaxed groups for which several observational biases prevent us to properly identify their central galaxy . in the spirit of checking that our results are not affected by the inclusion of the unrelaxed groups in our studies , in figure [ fig : cmassgrp ] we show the results for all zens groups ( left panels ) and for the subset of relaxed groups only ( right panels ) when studying the dependence of the @xmath4-mass relation on either @xmath1 or @xmath276 . we furthermore exclude from the following analysis those galaxies which are located at group - centric distances @xmath280 . in some of the zens groups relatively massive galaxies are found to lie in the outskirts of the group . as discussed above , this is partially a consequence of observational limitations but it may also be the signature of on - going accretion of sub - haloes . the sample selection at @xmath281 thus helps minimizing the contamination at these large radii of in - falling substructures and maximizing the purity " of galaxies that are satellites within a single common halo . in paper i we also comment on the fact that , at group masses above @xmath282 , group mass @xmath1 and lss ( over)density @xmath3 are correlated by nature , since massive groups are found by definition only in dense regions of the lss . in contrast , groups of masses @xmath283 are found at any value of @xmath3 . it is thus possible to disentangle the effects of @xmath1 and @xmath3 by limiting the studies as a function of @xmath3 to group masses below this threshold . we adopt this strategy here as well , and , in order to search for a dependence of the @xmath4-mass relation for satellites on @xmath3 , we test that the results obtained when including all groups in the analysis with @xmath3 ( left central panel ) hold also when restricting the group sample to groups with @xmath284 ( right central panel ) . having established that the concentration parameter is a function of both galaxy stellar mass and morphological type ( see figure [ fig : cmass ] ) , it is important to keep dependencies of both factors with the environment under control in order to proceed in this analysis . in paper iii we discuss the variation of the stellar mass of each broad morphological type with our environmental indicators and show that in our sample galaxy mass is largely independent of environment at fixed morphological type . here we note that , for the broad disk - dominated galaxies bin , biases in the median mass with any of the environments are not an issue , as for this class we can afford a fine galaxy mass bin splitting , and compare the @xmath4-mass relation for different environments within relatively small galaxy mass bins . for the broad bulge - dominated galaxies bin , however , given the fairly broad ranges covered by each of the two galaxy stellar mass bins , a difference in the median galaxy stellar mass ( and , more generally , of the galaxy stellar mass distributions ) between some of the different environments could in principle be present . this could induce a spurious dependence on such environments of the @xmath4 vs. stellar mass relations plotted in figure [ fig : cmassgrp ] . to keep this potential bias under control , we computed the median galaxy stellar masses within each environmental bin , and found that , for all three environments under study , the median stellar mass of galaxies in the broad bulge - dominated galaxies morphological bin changes by less than 0.1 dex between the environmental bins under comparison ( i.e. , low / high @xmath1 , @xmath3 and inner / outer group regions ) . we also verified if the morphological mix changes substantially with environment . the fractional contribution of intermediate - type disk galaxies to the total intermediate- plus late - type disk populations which compose the broad disk - dominated galaxies morphological bin defined above varies at most by @xmath285 between the low / high @xmath1 samples . precisely , in the three bins of galaxy stellar mass defined for the broad disk - dominated galaxies morphological bin this fractional contribution , i.e. , @xmath286 , equals , respectively , 35% , 55% , 85% for @xmath287 and 36% , 56% , 82% for @xmath288 . the fractional contribution of s0s to the broad bulge - dominated galaxies morphological bin is also almost identical between the samples of @xmath287 and @xmath288 , and equal to @xmath230 at @xmath289 and @xmath290 at @xmath291 . likewise at @xmath292 , the maximal variation in the fraction of intermediate galaxies within the disk - dominated galaxies class is 5% across the two @xmath3 bins . on the other hand , variations of @xmath293 are detected in the contributions of s0s to the broad bulge - dominated morphological bin between samples at small / large group - centric distances and low / high lss ( over)densities at @xmath294 ; hence we use caution in interpreting , for this broad morphological type , possible differences in the @xmath4 vs. galaxy stellar mass relation at these mass scales . the first thing to realize from figure [ fig : cmassgrp ] is that there are at most weak trends with the different environments of the concentration vs. stellar mass relation for disk satellite galaxies . two environmental trends are however potentially important : @xmath295 there is a global trend at and above galaxy masses of order @xmath262 , for satellites with a bulge - dominated morphology to be , albeit at the @xmath296-level , systematically more concentrated , by about 5 - 10% , at low environmental densities ( large group - centric distances and low lss densities ) than at high environmental densities ( high mass groups , small group - centric distances , high lss densities ) . this difference could be at least partially driven by a change in the mix between s0 and bulge - dominated spirals within the broad bulge dominated morphological bin . @xmath297 in contrast , the disk - dominated population shows a systematic increase in median concentration , at fixed galaxy stellar mass above @xmath28 , with increasing group mass and lss density . in this case the similar contributions of intermediate - type disks and late - type disks to the broad disk - dominated morphological bin suggest that there is a genuine environmental effect at work . the difference in concentration between low and high mass groups is @xmath298 ; furthermore , the effect is robust towards the inclusion / exclusion of the sample of unrelaxed groups . the significance of this concentration difference , as assed through a @xmath299-test on the two samples , gives a probability @xmath300 that the two populations have truly different median concentrations ( or a kolmogorov - smirnov test give a 95% probability of a common parent distribution ) . on the other hand , the trend with @xmath3 becomes weaker when only groups with @xmath287 are considered for the lss analysis . as discussed in paper i , the effect observed with @xmath3 when also the most massive groups are included is likely a spurious reflection of the dependence of concentration on halo mass . as indicated above , a full investigation of the implications of either of these trends is beyond the scope of this paper . we will study these effects in more detail , also within the framework set by other independent analyses of the satellite population , in a future zens analysis . we have presented detailed structural analyses performed on the 1455 galaxies in the 141 zens groups introduced in paper i. we remark that the corresponding zens catalogue has however 1484 galaxy with valid entries , since we measure parameters for individual galaxies that are members of 29 galaxy pairs , which are given as one single entry in the parent 2dfgrs galaxy catalogue . the parametric and non - parametric structural measurements presented here , together with the detailed environmental parameters from paper i and the photometric ( including stellar masses ) measurements presented in paper iii , set the basis for a number of forthcoming publications which use the zens data to explore which environmental scales are relevant for the evolution of morphologically different galaxy populations at different mass scales . the measurements are published in the global zens catalog with paper i. in detail , the measurements that we have presented here are : * strength of bars in disks , quantified through an isophotal analysis ( section [ sec : isophotalanalysis ] ) ; * single- and double - component ( bulge+disk ) srsic fits parameters both in the @xmath9 and @xmath10 bands , including model - based galaxy sizes ( sections [ sec : sersicfits ] and [ sec : gim2d_decomp ] ) , as well as bulge+disk+bar fits for disks with a noticeable bar component ( appendix [ app : bulgebardisk ] ) ; * non - parametric structural indices of concentration , asymmetry , smoothness of the light distribution , and gini and @xmath128 coefficients , and aperture photometric " size estimates ( section [ sec : zest ] ) ; * morphological classes , based on a quantitative bulge+disk criterion , augmented ( or supported ) by quantitative criteria regarding the non - parametric diagnostics , after correction for observational biases ( sections [ sec : gim2d_decomp ] and [ sec : morphclass ] ) . crucially , we do indeed derive correction matrices , which we apply to the relevant structural estimates , to minimize biases which , depending on psf size as well as galaxy magnitude , size , shape of light profile and ellipticity , would otherwise prevent a reliable comparison of the structural properties of galaxies observed in different seeing conditions , and lying in different regions of this four - dimensional galaxy parameter space ( see section [ sec : simulations ] ) . as expected , biases in the model - fit parameters are substantially reduced thanks to the treatment of the psf - blurring effects in these algorithms ; still , some are present even in the model - fit parameters , which may have an impact in some analyses if left uncorrected . disk properties are well measured and require very modest or no further corrections ; so are bulge - to - total ratios , which therefore offer an excellent parameter to base a quantitative morphological / structural classification . in contrast , bulge @xmath80 srsic indices and half - light radii are degenerate in some circumstances . this implies that , on global galactic scales , concentration parameter or n - srsic index , alone , are not a good proxy for bulge - to - disk ratio and thus morphology : bulges can give large contributions to the total light budget and have large half - light radii , leading to low galaxy concentrations , or , vice - versa , bulges can be very compact and lead to high galaxy concentrations despite a modest contribution to the total light . finally , we particularly warn against using uncorrected non - parametric estimators as galaxy classifiers since , understandably , they suffer from severe observational biases and introduce severe errors in the classifications ( see section [ sec : simulations ] ) . as a first application of our corrected structural measurements , we have studied the variation of concentration in _ satellite _ galaxies of fixed stellar mass ( from paper iii ) with morphological type , and with the three environments detailed in paper i , i.e. , the mass of the host group halo , the projected group - centric distance , and the density of the large - scale structure cosmic web ( section [ sec : results ] ) . we find that the known correlation of satellite concentration with galaxy stellar mass ( e.g. * ? ? ? * ) holds at a fixed morphological type . specifically , there is a genuine increase in concentration with increasing stellar mass for disk satellite galaxies within each separate bin of bulge - dominated galaxies , intermediate - type and late - type disk morphologies . the slope of the concentration vs. galaxy stellar mass relationship flattens from the later to the earlier types ( and becomes @xmath301 for satellites with an elliptical morphology , which however cover a limited range in galaxy stellar mass in our sample ) . it is not trivial to disentangle , in a physically meaningful way , the contributions of an increasing bulge - to - total ratio with increasing mass , and an increasing bulge concentration at fixed bulge - to - total ratio with increasing mass , to the increase in concentration of satellite disk galaxies with increasing stellar mass . this is true even in our study , in which we did indeed base our morphological classification on the bulge - to - total ratio , and thus bulge - to - total variations within each hubble type should be minimized . there are nevertheless some residual effects , as the median @xmath100 is found to vary in our sample from @xmath302 at @xmath28 to @xmath303 at @xmath304 for late - type disks , and from 59@xmath27 to 64@xmath27 between @xmath28 and @xmath305 for bulge - dominated disks . still , for intermediate - type disk satellites , in which the bulge - to - total ratio is tightly constrained by firm lower an upper boundaries by definition , the increase in satellite density with stellar mass can be genuinely ascribed to a @xmath306 decrease in the bulge - to - disk size ratio , i.e. , to a genuine increase in the concentration of the bulge component at fixed bulge - to - total light ratio . we tentatively assume this as the explanation for the increase of disk satellite concentration with stellar mass at fixed hubble type , an hypothesis which we will test with further analyses . when considering the galaxies environment , we find that , at galaxy stellar mass @xmath219 and above , @xmath93 bulge - dominated satellites tend to be marginally more concentrated at low lss densities and high group - centric distances , and @xmath95 in contrast , disk - dominated satellites are significantly more concentrated in high group masses . the interpretation of the first weak trend as a hint for an environmental effect is further hampered , at @xmath294 , by a higher fraction of s0s relative to bulge - dominated spiral galaxies in low @xmath3 relative to high densities ( 60% and 0% , respectively in @xmath307 groups and 58% and 35% over all groups ) and in the outskirts relative to the cores of groups ( 50% and 31% , over all groups ) . this change in morphological mix at the high - end of the b / t sequence is consistent with other observational works which have also reported an increase in the s0 fraction in the outskirts of groups and clusters , although with no distinction between central and satellite galaxies . the early study of @xcite on local cluster reveals a drop in the s0 fraction close to the cluster centers , which they interpret as the outcome of disk galaxies destruction happening at the cluster cores . the more recent analysis of the morphology - density relation at redshift @xmath308 by @xcite shows a depletion of s0 galaxies within 0.3 virial radii , and an increase in the early spirals population at the same distances , consistent with the variation we see in our zens sample . also in @xmath309 groups @xcite find hints for an excess of s0 galaxies at @xmath310 mpc with respect to the group centers , which instead host a higher fraction of late - type disks . these authors furthermore find that the fraction of s0 galaxies in groups is comparable to the one in clusters at the same redshift , suggesting that galaxy pre - processing and s0 formation is effective already at these low densities . the increase of s0 in the outer regions of the groups is particularly interesting , as it counter to the intuitive idea that gas - rich bulge - dominated spirals may become s0 galaxies as they fall deeper into their group potential wells ( see e.g. @xcite who find that gas stripping in the group environment is only effective close to the group centers ) . speculating on the above , a possible scenario is that @xmath93 bulge - dominated galaxies are transformed into s0s as soon as disk galaxies enter the group potential , and @xmath95 further stellar evolution within the groups replenished the dried - out disks of s0 galaxies of fresh gas , establishing / restoring in them a bulge - dominated morphology . note that an increase in the fraction of dusty star - forming galaxies at high densities and close to cluster centers has been observed at low redshift , e.g. @xcite . furthermore , the evidence of polar and extended hi disks around s0s e.g. @xcite may also support this `` disk - regrowth '' scenario . the second , statistically more significant trend for @xmath311 disk - dominated galaxies to have a higher concentration in high - mass ( @xmath312 ) than in lower - mass groups appears to be a genuine environmental effect which should suffer from no morphological complications . @xcite found evidence in their sdss sample that , among galaxies with @xmath313 , satellites are more concentrated than centrals with identical stellar mass . these authors interpret the variation in concentration within the framework of gradual stripping and subsequent quenching of satellite galaxies during infall into the group potential . this is also believed to cause a reddening of the satellite galaxies and a shrinking of their typical sizes . while we postpone an analysis on the subject to a forthcoming paper , here we wish to note that the sdss studies divide the early- and late - type morphological classes on the basis of a non - corrected concentration criterion ( @xmath313 for late - type and @xmath175 for early - type galaxies ) . comparing their figure 1 with our figure [ fig : c_vs_bt ] , we see that @xmath313 in the sdss system roughly corresponds to a value of @xmath314 in zens , as our concentrations show a small offset due to the corrections we applied . as illustrated in figure [ fig : cmass ] , a cut at constant concentration sub - divides the zens galaxy sample into two broad bins in which , however , individual hubble types are mixed together below the chosen @xmath4 threshold . it thus remains an open question whether the difference in concentration between centrals and satellites reported by those authors is the result of a variation in the morphological mix of the central vs. satellite populations , or rather a change in the structure of central and satellite galaxies at a fixed morphological type . we will address this question in a dedicated zens analysis . a.c . , e.c . and c.r . acknowledge support from the swiss national science foundation . this publication makes use of data from eso large program 177.a-0680 , and data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation . galex ( _ galaxy evolution explorer _ ) is a nasa small explorer , launched in april 2003 . we gratefully acknowledge nasa s support for construction , operation , and science analysis for the galex mission . abazajian , k. n. , adelman - mccarthy , j. k. , ageros , m. a. , et al . 2009 , , 182 , 543 abraham , r. g. , tanvir , n. r. , santiago , b. x. , et al . 1996 , , 279 , l47 abraham , r. g. , merrifield , m. r. , ellis , r. s. , tanvir , n. r. , & brinchmann , j. 1999 , , 308 , 569 abraham , r. g. , & merrifield , m. r. 2000 , , 120 , 2835 abraham , r. g. , van den bergh , s. , & nair , p. 2003 , , 588 , 218 aguerri , j. a. l. , mndez - abreu , j. , & corsini , e. m. 2009 , , 495 , 491 allen , p. d. , driver , s. p. , graham , a. w. , et al . 2006 , , 371 , 2 bailin , j. , & harris , w. e. 2008 , , 681 , 225 baldry , i. k. , balogh , m. l. , bower , r. g. , et al . 2006 , , 373 , 469 balcells , m. , graham , a. w. , domnguez - palmero , l. , & peletier , r. f. 2003 , , 582 , l79 balogh , m. l. , morris , s. l. , yee , h. k. c. , carlberg , r. g. , & ellingson , e. 1997 , , 488 , l75 bamford , s. p. , nichol , r. c. , baldry , i. k. , et al . 2009 , , 393 , 1324 barazza , f. d. , jogee , s. , & marinova , i. 2008 , , 675 , 1194 barden , m. , et al . 2005 , , 635 , 959 barnes , j. e. 1988 , , 331 , 699 beckman , j. e. , peletier , r. f. , knapen , j. h. , corradi , r. l. m. , & gentet , l. j. 1996 , , 467 , 175 bekki , k. , couch , w. j. , & shioya , y. 2002 , , 577 , 651 bender , r. , doebereiner , s. , & moellenhoff , c. 1988 , , 74 , 385 bentez , n. , ford , h. , bouwens , r. , et al . 2004 , , 150 , 1 bernardi , m. , shankar , f. , hyde , j. b. , et al . 2010 , , 404 , 2087 bernstein , r. a. , freedman , w. l. , & madore , b. f. 2002a , , 571 , 56 bernstein , r. a. , freedman , w. l. , & madore , b. f. 2002b , , 571 , 107 bertin , e. , & arnouts , s. 1996 , , 117 , 393 blanton , m. r. , et al . 2005 , , 129 , 2562 blanton , m. r. , hogg , d. w. , bahcall , n. a. , et al . 2003 , , 594 , 186 blanton , m. r. , & berlind , a. a. 2007 , , 664 , 791 buta , r. , vasylyev , s.,salo , h. , & laurikainen , e. 2005 , , 130 , 506 byun , y. i. , freeman , k. c. , & kylafis , n. d. 1994 , , 432 , 114 cameron , e. , & driver , s. p. 2007 , , 377 , 523 cameron , e. , & driver , s. p. 2009 , , 493 , 489 cameron , e. , carollo , c. m. , oesch , p. , et al . 2010 , , 409 , 346 caon , n. , capaccioli , m. , & donofrio , m. 1993 , , 265 , 1013 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 carollo , c. m. , franx , m. , illingworth , g. d. , & forbes , d. a. 1997a , , 481 , 710 carollo , c. m. , danziger , i. j. , rich , r. m. , & chen , x. 1997b , , 491 , 545 carollo , c. m. , stiavelli , m. , de zeeuw , p. t. , & mack , j. 1997c , , 114 , 2366 carollo , c. m. , stiavelli , m. , & mack , j. 1998 , , 116 , 68 carollo , c. m. 1999 , , 523 , 566 carollo , c. m. , stiavelli , m. , de zeeuw , p. t. , seigar , m. , & dejonghe , h. 2001 , , 546 , 216 carollo , c. m. , scarlata , c. , stiavelli , m. , wyse , r. f. g. , & mayer , l. 2007 , , 658 , 960 carollo , c. m. , cibinel , a. , lilly , s .j . , et al . 2013a ( paper i ) , , submitted carollo , c. m. , bschorr , t. j. , renzini , a. , et al . 2013 , arxiv:1302.5115 cibinel , a. , carollo , c. m , lilly , s. j. , et al . 2013 ( paper iii ) , , submitted colless , m. , dalton , g. , maddox , s. , et al . 2001 , , 328 , 1039 colless , m. , peterson , b. a. , jackson , c. , et al . 2003 , arxiv : astro - ph/0306581 combes , f. , debbasch , f. , friedli , d. , & pfenniger , d. 1990 , , 233 , 82 conselice , c. j. 2003 , , 147 , 1 cooper , m. c. , newman , j. a. , madgwick , d. s. , gerke , b. f. , yan , r. , & davis , m. 2005 , , 634 , 833 ct , p. , ferrarese , l. , jordn , a. , et al . 2007 , , 671 , 1456 courteau , s. , de jong , r. s. , & broeils , a. h. 1996 , , 457 , l73 cunow , b. 2001 , , 323 , 130 debattista , v. p. , carollo , c. m. , mayer , l. , & moore , b. 2004 , , 604 , l93 debattista , v. p. , mayer , l. , carollo , c. m. , et al . 2006 , , 645 , 209 dekel , a. , birnboim , y. , engel , g. , et al . 2009a , , 457 , 451 dekel , a. , sari , r. , & ceverino , d. 2009b , , 703 , 785 de jong , r. s. 1996 , , 118 , 557 de propris , r. , et al . 2003 , , 342 , 725 de souza , r. e. , gadotti , d. a. , & dos anjos , s. 2004 , , 153 , 411 de vaucouleurs , g. 1963 , , 8 , 31 disney , m. j. 1976 , , 263 , 573 dressler , a. 1980 , , 236 , 351 driver , s. p. , popescu , c. c. , tuffs , r. j. , liske , j. , graham , a. w. , allen , p. d. , & de propris , r. 2007 , , 379 , 1022 driver , s. p. , popescu , c. c. , tuffs , r. j. , et al . 2008 , , 678 , l101 eke , v. r. , baugh , c. m. , cole , s. , et al . 2004a , , 348 , 866 eke , v. r. , frenk , c. s. , baugh , c. m. , et al . 2004b , , 355 , 769 einasto , m. , et al . 2007 , , 464 , 815 erwin , p. 2005 , , 364 , 283 fall , s. m. , & efstathiou , g. 1980 , , 193 , 189 landolt a. u , 1992 , , 104 , 340 feldmann , r. , carollo , c. m. , porciani , c. , et al . 2006 , , 372 , 565 feldmann , r. , mayer , l. , & carollo , c. m. 2008 , , 684 , 1062 ferrarese , l. , van den bosch , f. c. , ford , h. c. , jaffe , w. , & oconnell , r. w. 1994 , , 108 , 1598 fisher , d. b. , & drory , n. 2008 , , 136 , 773 forbes , d. a. , & thomson , r. c. 1992 , , 254 , 723 franx , m. , illingworth , g. , & heckman , t. 1989 , , 98 , 538 fukugita m. et al.,1995 , pasp , 107 , 945 gadotti , d. a. 2008 , , 384 , 420 gallazzi , a. , bell , e. f. , wolf , c. , et al . 2009 , , 690 , 1883 graham , a. w. 2001 , , 121 , 820 graham , a. w. , & guzmn , r. 2003 , , 125 , 2936 graham , a. w. , erwin , p. , trujillo , i. , & asensio ramos , a. 2003 , , 125 , 2951 graham , a. w. , & driver , s. p. 2005 , pasa , 22 , 118 graham , a. w. , driver , s. p. , petrosian , v. , conselice , c. j. , bershady , m. a. , crawford , s. m. , & goto , t. 2005 , , 130 , 1535 graham , a. w. , & worley , c. c. 2008 , , 388 , 1708 goto , t. , yamauchi , c. , fujita , y. , et al . 2003 , , 346 , 601 guo , y. , et al . 2009 , , 398 , 1129 hambly n. c. et al . , 2001 , , 326 , 1279 hansen , s. m. , sheldon , e. s. , wechsler , r. h. , & koester , b. p. 2009 , , 699 , 1333 hussler , b. , et al . 2007 , , 172 , 615 hogg , d. w. , et al . 2003 , , 585 , l5 hopkins , p. f. , cox , t. j. , dutta , s. n. , et al . 2009 , , 181 , 135 knobel , c. , lilly , s. j. , kovac , k. , et al . 2012 , arxiv:1211.5607 immeli , a. , samland , m. , gerhard , o. , & westera , p. 2004 , , 413 , 547 impey , c. , & bothun , g. 1997 , , 35 , 267 janowiecki , s. , mihos , j. c. , harding , p. , et al . 2010 , , 715 , 972 jedrzejewski , r. i. 1987 , , 226 , 747 kauffmann , g. , heckman , t. m. , white , s. d. m. , et al . 2003 , , 341 , 54 kawata , d. , mulchaey , j. s. , gibson , b. k. , & snchez - blzquez , p. 2006 , , 648 , 969 kimm , t. , et al . 2009 , , 394 , 1131 knapen , j. h. , shlosman , i. , & peletier , r. f. 2000 , , 529 , 93 kormendy , j. 1979 , , 227 , 714 kormendy , j. , & kennicutt , r. c. , jr . 2004 , , 42 , 603 kormendy , j. , fisher , d. b. , cornell , m. e. , & bender , r. 2009 , , 182 , 216 kron , r. g. 1980 , , 43 , 305 lackner , c. n. , & gunn , j. e. 2012 , , 421 , 2277 landolt a. u , 1992 , , 104 , 340 lauer , t. r. , ajhar , e. a. , byun , y .- i . , et al . 1995 , , 110 , 2622 laurikainen , e. , salo , h. , & buta , r. 2005 , , 362 , 1319 laurikainen , e. , salo , h. , buta , r. , & knapen , j. h. 2007 , , 381 , 401 lewis , i. , et al . 2002 , , 334 , 673 lintott , c. j. , schawinski , k. , slosar , a. , et al . 2008 , , 389 , 1179 lotz , j. m. , primack , j. , & madau , p. 2004 , , 128 , 163 lu , t. , et al . 2012 , in prep . ( paper iv ) mahajan , s. , & raychaudhury , s. 2009 , , 400 , 687 malin , d. f. , & carter , d. 1980 , , 285 , 643 malin , d. f. , & carter , d. 1983 , , 274 , 534 maller , a. h. , berlind , a. a. , blanton , m. r. , & hogg , d. w. 2009 , , 691 , 394 marleau , f. r. , & simard , l. 1998 , , 507 , 585 meert , a. , vikram , v. , & bernardi , m. 2012 , arxiv:1211.6123 menndez - delmestre , k. , sheth , k. , schinnerer , e. , jarrett , t. h. , & scoville , n. z. 2007 , , 657 , 790 mihos , j. c. , & hernquist , l. 1994 , , 437 , l47 mihos , j. c. , & hernquist , l. 1996 , , 464 , 641 mllenhoff , c. , popescu , c. c. , & tuffs , r. j. 2006 , , 456 , 941 mosleh , m. , williams , r. j. , & franx , m. 2013 , arxiv:1302.6240 muldrew , s. i. , croton , d. j. , skibba , r. a. , et al . 2012 , , 419 , 2670 naab , t. , burkert , a. , & hernquist , l. 1999 , , 523 , l133 naab , t. , & burkert , a. 2003 , , 597 , 893 nair , p. b. , & abraham , r. g. 2010 , , 186 , 427 nair , p. b. , & abraham , r. g. 2010 , , 714 , l260 norberg p. et al . , 2002 , , 336 , 907 norman , c. a. , sellwood , j. a. , & hasan , h. 1996 , , 462 , 114 noordermeer , e. , van der hulst , j. m. , sancisi , r. , swaters , r. a. , & van albada , t. s. 2005 , , 442 , 137 oesch , p. a. , carollo , c. m. , feldmann , r. , et al . 2010 , , 714 , l47 oke , j. b. 1974 , , 27 , 21 pannella , m. , hopp , u. , saglia , r. p. , et al . 2006 , , 639 , l1 peng , c. y. , ho , l. c. , impey , c. d. , & rix , h .- w . 2002 , , 124 , 266 peng , y .- j . , et al . 2010 , , 721 , 193 peng , y. , lilly , s. j. , renzini , a. , & carollo , m. 2012 , arxiv:1106.2546 porter , s. c. , raychaudhury , s. , pimbblet , k. a. , & drinkwater , m. j. 2008 , , 388 , 1152 refregier , a. 2003 , , 338 , 35 rix , h .- w . , barden , m. , beckwith , s. v. w. , et al . 2004 , , 152 , 163 sage , l. j. , & welch , g. a. 2006 , , 644 , 850 saglia , r. p. , bertschinger , e. , baggley , g. , burstein , d. , colless , m. , davies , r. l. , mcmahan , r. k. , jr . , & wegner , g. 1993 , , 264 , 961 scarlata , c. , carollo , c. m. , lilly , s. , et al . 2007 , , 172 , 406 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 schweizer , f. 1979 , , 233 , 23 schweizer , f. , seitzer , p. , faber , s. m. , burstein , d. , dalle ore , c. m. , & gonzalez , j. j. 1990 , , 364 , l33 scodeggio , m. , gavazzi , g. , franzetti , p. , boselli , a. , zibetti , s. , & pierini , d. 2002 , , 384 , 812 scoville , n. , aussel , h. , brusa , m. , et al . 2007 , , 172 , 1 sersic , j. l. 1968 , cordoba , argentina : observatorio astronomico , 1968 , shao , z. , xiao , q. , shen , s. , mo , h. j. , xia , x. , & deng , z. 2007 , , 659 , 1159 sheth , k. , elmegreen , d. m. , elmegreen , b. g. , et al . 2008 , , 675 , 1141 shen , s. , mo , h. j. , white , s. d. m. , et al . 2003 , , 343 , 978 simard , l. , willmer , c. n. a. , vogt , n. p. , et al . 2002 , , 142 , 1 tal , t. , van dokkum , p. g. , nelan , j. , & bezanson , r. 2009 , , 138 , 1417 toomre , a. 1977 , evolution of galaxies and stellar populations , 401 trujillo , i. , aguerri , j. a. l. , cepa , j. , & gutirrez , c. m. 2001 , , 328 , 977 trujillo , i. , erwin , p. , asensio ramos , a. , & graham , a. w. 2004 , , 127 , 1917 tuffs , r. j. , popescu , c. c. , vlk , h. j. , kylafis , n. d. , & dopita , m. a. 2004 , , 419 , 821 van den bosch , f. c. , aquino , d. , yang , x. , mo , h. j. , pasquali , a. , mcintosh , d. h. , weinmann , s. m. , & kang , x. 2008 , , 387 , 79 van der wel , a. 2008 , , 675 , l13 van gorkom , j. h. , schechter , p. l. , & kristian , j. 1987 , , 314 , 457 van dokkum , p. g. 2005 , , 130 , 2647 vulcani , b. , poggianti , b. m. , aragn - salamanca , a. , et al . 2011 , , 412 , 246 weinmann , s. m. , van den bosch , f. c. , yang , x. , & mo , h. j. 2006 , , 366 , 2 weinmann , s. m. , kauffmann , g. , van den bosch , f. c. , pasquali , a. , mcintosh , d. h. , mo , h. , yang , x. , & guo , y. 2009 , , 394 , 1213 weinzirl , t. , jogee , s. , khochfar , s. , burkert , a. , & kormendy , j. 2009 , , 696 , 411 whitmore , b. c. , & gilmore , d. m. 1991 , , 367 , 64 whitmore , b. c. , gilmore , d. m. , & jones , c. 1993 , , 407 , 489 white , s. d. m. , & rees , m. j. 1978 , , 183 , 341 wilman , d. j. , oemler , a. , jr . , mulchaey , j. s. , et al . 2009 , , 692 , 298 wilman , d. j. , zibetti , s. , & budavri , t. 2010 , , 406 , 1701 woo , j. , dekel , a. , faber , s. m. , et al . 2012 , arxiv:1203.1625 wyse , r. f. g. , gilmore , g. , & franx , m. 1997 , , 35 , 637 york , d. g. , adelman , j. , anderson , j. e. , jr . , et al . 2000 , , 120 , 1579 zamojski , m. a. , schiminovich , d. , rich , r. m. , et al . 2007 , , 172 , 468 table [ tab : obsruns ] gives a log of the observing runs for the new eso/2.2 m @xmath9 and @xmath10 wfi images . these images were taken with the @xmath315 mosaic of 2k@xmath3164k coated ccds , which provides a scale of 0.238@xmath164/pix and covers a total field of view of 34@xmath317 , well suited to the typical sizes of the _ zens _ groups . each group was observed with five dithered exposures to remove inter - chip gap effects and cosmic rays events ; the single exposure times were 132@xmath318 for the 2005 runs and 144@xmath318 for all other runs . all observations were carried out under clear night conditions . the final seeing after image stacking is typically @xmath319 , reaching down for the best runs to @xmath320 and @xmath321 in the @xmath10- and @xmath9-band , respectively , and , in a few worst cases , degrading to @xmath322 and @xmath323 . figure [ fig : skyconditions ] gives a summary of the airmass and seeing conditions of the wfi observations . the _ zens _ detection limits are given by the @xmath324 background fluctuations in a uniform area of 1 arcsec@xmath325(18 pixels ) and correspond to @xmath326 mag arcsec@xmath325 and @xmath327 mag arcsec@xmath325 ( ab magnitudes ) . the depth of the wfi @xmath9 and @xmath10 data allow us to robustly quantify the structure of the _ zens _ galaxies out to at least 2 half - light radii , and to detect low surface brightness tidal tails and merger - induced features . at the given depth limits , and with a total sky coverage of 45 deg@xmath325 , _ zens _ is a relatively deep survey on a relatively large area , and thus a valuable dataset also for legacy science . clcccc 1 & jan . 5 - feb . 18 , 2005 & 131.917 & 20 & 074.b-0570 & service + 2 & may 1 - dec . 23 , 2006 & 143.917 & 50 & 177.a-0680(a / b ) & service + 3 & feb . 18 - july 14 , 2007 & 143.917 & 21 & 177.a-0680(a / b ) & service + 4 & jan . 14 - july 31 , 2008 & 143.917 & 12 & 177.a-0680(a / b ) & service + 5 & oct . 30 - nov . 5 , 2008 & 143.917 & 25 & 177.a-0680(c ) & visitor + 6 & mar . 28 , 2009 & 143.917 & 13 & 177.a-0680(d ) & visitor + + the wfi @xmath9 and @xmath10 images were reduced using standard iraf routines and in particular the noao mosaic software @xmath328 was used for most of the processing . the bias subtraction involved two steps : each frame was initially subtracted of a bias level using , for each pixels row , the median value of the overscan pixels in the same row . remaining large scale features in the bias were further eliminated by subtracting a combined bias image , which was obtained separately for each night by stacking together at least 10 ( overscan subtracted ) bias frames . dark current was not subtracted since its value is negligible ( ( 0.3 - 0.6)@xmath329adu s@xmath330 pix@xmath330 ) . the images were corrected for pixel - to - pixel variation of the quantum efficiency by dividing them with a twilight flat - field , obtained combining several tens of flat field exposures acquired throughout the different nights of each observing run . the presence of residual large - scale illumination inhomogeneities was removed with a super - flatfield obtained from the science exposures themselves , stacked together with a @xmath331-clipping algorithm to remove all bright sources . the @xmath10-band images were affected by fringing and required intermediate processing before deriving the super - flat images . all the bias - corrected and flat - fielded @xmath10-frames were combined to obtain an image of the fringing pattern . the large - scale features of the fringing image were removed by subtracting , from it , a smoothed version of itself ; this produced the final fringe template , containing only the signal arising from the fringing itself . the fringing template was then subtracted from the science exposures with an interactive task , allowing us to verify the quality of the correction . these fringing - removed images were finally used to create the super - flatfield for the red filter . summary of the observing conditions for the _ zens _ wfi data . shown are the distribution of airmass and ( instrumental plus atmospheric , i.e. , total psf ) seeing values over the targeted fields , for the @xmath9- ( left panels ) and @xmath10-band ( right panels ) , respectively.,width=377 ] further reduction steps were made to correct for detector cosmetic effects , such as bad pixels , hot or dead columns . the five individual reduced images of each group in each filter were stacked together to obtain the final science images that we used in our analysis . from the large group frames we extracted postage stamp images for each of the galaxies , of size equal to 3 petrosian radii . sky subtraction was performed locally on the cleaned stamps , so as to account for small inhomogeneities in the sky level . cleaning of the postage stamps involved removing most of the light coming from other galaxies ( or from spurious stellar spikes and other contaminations ) , and substituting to the relevant pixels an iterated value of the average sky level in the stamp . companion galaxies were identified using sextractor as sources above 1.5@xmath332 of the sky level and the pixels belonging to these objects were replaced with random blank sky regions . all stamps were furthermore visually inspected to verify the quality of the cleaning process . photometric calibration was performed using landolt ( @xcite ) standard stars . several landolt fields , with at least 5 standard stars , were observed for each night . aperture magnitudes for these stars were measured with the iraf task _ qphot _ and used to derive the photometric zero point ( zp ) for the two pass bands . we used the following calibration equations between the landolt b and i and the wfi instrumental magnitudes _ b _ , _ i _ : @xmath333 and @xmath334 , respectively . in these calibration equations @xmath335 is the airmass value at the time of observation and @xmath136 is the extinction coefficient . zero points , color terms and extinction coefficients are presented in table [ tab : phot_calib ] . since our standard stars were not observed at sufficiently differing airmasses to allow for a sensible extinction correction , magnitudes were airmass - corrected by using the extinction coefficients given on the wfi website as a reference . all magnitudes are calibrated onto an extra - atmospherical airmass of zero . while there is no clear color dependence of the calibration for the @xmath10 filter ( @xmath336 ) , it is necessary to correct for a color term ( @xmath337 ) in the case of the @xmath9 band . we also found that the zp points themselves show a shift over the different observing runs , even though the slope of the calibration remains constant ( most likely be due to different ccd / ambient conditions during the observations ) . for this reason , we calibrated each night separately , fixing the slope of the color contribution to the well - established common value for all the nights , while using the night - by - night determined zps for each set of observations . cccccccc 1 & jan . 5 - feb . 18 , 2005 & 24.75 & 23.74 & 0.215 & 0 & 0.22 & 0 + 2 & may 1 - dec . 23 , 2006 & 24.75 - 25.08 & 23.74 - 24.0 & 0.215 & 0 & 0.22 & 0 + 3 & feb . 18 - july 14 , 2007 & 24.85 - 24.96 & 23.81 - 23.90 & 0.215 & 0 & 0.22 & 0 + 4 & jan . 14 - july 31 , 2008 & 24.53 -24.99 & 23.74 - 23.92 & 0.215 & 0 & 0.22 & 0 + 5 & oct . 5 , 2008 & 24.93 - 24.99 & 23.89 - 23.92 & 0.215 & 0 & 0.22 & 0 + 6 & mar . 25 - mar 28 , 2009 & 25.00 - 24.94 & 23.88 - 23.98 & 0.215 & 0 & 0.22 & 0 the landolt system is tied to the standard johnson - cousin system which uses vega for reference ; we thus converted our magnitudes to the ab system ( @xcite ) through the following procedure . for the b - band , we found the zp corresponding to an object with zero color in ab in the @xmath338 relation , and used this value as our final calibration factor for the blue filter . as mentioned before , the color term was negligible for the @xmath10 band and thus it was not possible to apply such method . in this case the following transformation was applied : @xmath339 . we computed the conversion factor using the synphot iraf package ( see also @xcite ) and found it to be @xmath340 . the values for the zp we find are in good agreement with those reported on the wfi calibration web - page . we assessed the robustness of our calibration by comparing the magnitudes obtained for ( the stars in the _ zens _ fields , as well as for ) the _ zens _ galaxies with the others available from the literature . for the _ zens _ galaxies we used the magnitudes obtained with the sextractor ( source extractor , @xcite ) software and determined from the flux curve of growth inside a kron - like elliptical aperture ( mag@xmath341auto in the code ) . the kron ( 1980 ) aperture radius correspond to 2.5 times the radius of the first image moment . all magnitudes are corrected for galactic absorption using the dust maps of @xcite and the @xcite dust law . as reference values we consider the 2dfgrs @xmath342 , the supercosmos survey @xmath342 and @xmath343 @xcite and the sdss @xmath51 and @xmath342 magnitudes @xcite . the latter is derived as in @xcite using the relation @xmath344 . we show in figure [ fig : photaccuracy ] the comparison between the derived _ zens _ magnitudes and the published ones . we have a good agreement with the supercosmos values ( within a standard deviation of @xmath345 mag ) , which is better than with the 2dfgrs one ( @xmath346 mag ) . this is very consistent with the quoted photometric errors of the 2dfgrs ( @xmath342 magnitudes , @xmath347 mag ) and supercosmos magnitudes ( @xmath108 mag ) . we mention that our @xmath9 measurements show a small non - linearity with respect to both the supercosmos data and 2dfgrs @xmath342 magnitudes . this non - linearity is however absent when we compare with sdss photometry , which is available for a subset of our galaxies . with the exception of the expected shift between the supercosmos @xmath214 passband ( @xmath348 ) and the _ zens _ @xmath10 passband , the median magnitude offsets in the relevant comparisons are small at the level of @xmath349 . we remark that in deriving galaxy stellar masses and other photometry - inferred parameters in paper iii , the zebra+ software @xcite was first run in photometry - check mode " , to detect residual offsets from the individual passbands ; no offsets were found for the wfi @xmath9 and @xmath10 passbands . for a number of galaxies in our sample , the sdss new york university value - added galaxy catalog ( nyu - vagc ) value added catalogue @xcite provides half - light radii and indices @xmath80 from single - component srsic fits to the azimuthally averaged surface brightness profiles . we thus compare in figure [ fig : zens_sdss_sersic ] our own best fit parameters for the single component fits , both _ prior to and after _ the application of our corrections described in section [ sec : correctionmaps ] , with these previously published data . to match the vagc definition , we _ circularize _ in post - processing mode , using the galaxies ellipticities , our fiducial major - axis size estimates based on elliptical apertures . as illustrated in the figure , the agreement between the two sets of measurement is generally good . only about 5% of the galaxies show in zens larger half - light radii and srsic indices than those measured in sdss . regarding the differences in srsic index , we note that the sdss fits have a maximum @xmath80=6 whereas for the zens sample we allowed @xmath80 to vary up to @xmath80=10 . indeed , the most discrepant galaxies in the @xmath80-comparison plot are those which have reached the saturation value of @xmath350 in the sdss analysis ( left panel in the figure ) . these same galaxies are actually also many of those which show a large difference in the half - light radius comparison plot ( red filled squares in the right panel of figure [ fig : zens_sdss_sersic ] ) . part of the observed discrepancy in the radii is thus likely caused by the different @xmath80 values in the zens and sdss fits . to test this directly , we refitted these galaxies ( including other galaxies which have @xmath351 in zens but @xmath352 in the sdss fits ) imposing a maximum @xmath353 value for the srsic index in their single - component surface brightness fits , as done in the sdss analysis . the red arrows in the right panel of the figure show the differences in @xmath64 that result from constraining the fits to a maximum @xmath353 value . these differences highlight the interdependence of the derived parameters in such kind of fits , and the systematic errors that such interdependence might introduce . finally we note that many of the zens galaxies whose nominal single - component size estimates are substantially larger than the sdss size estimates are those galaxies which were already flagged in our zens catalog as having a problematic single - component size estimate in the analysis of section [ sec : largeradiidiff ] ( filled red squares in the right panel of figure [ fig : zens_sdss_sersic ] ) . these galaxies were found to have substantially larger half - light radii from their single - component fits than from their double - component fits . this evidence , and a visual inspection of the data and models , motivated our choice to adopt the double - component half - light radii as our fiducial size estimates in zens ( section [ sec : largeradiidiff ] ) . caution should be exercised in using blindly the gim2d double component results , as a number of factors can produce profiles which do not correspond to a meaningful bulge+disk decomposition ( see also @xcite for an extensive discussion on this issue ) . for this reason , all residuals images , models and profiles obtained from gim2d were visually inspected to look for possible failure of the fitting algorithm and to identify physically - reasonable double - component models . as in @xcite , we analyze the radial surface brightness profiles of bulges and disks to identify problematic models which would not be recognized as unphysical by simply looking at the fit @xmath354 or at residual images . to this purpose we analyzed the surface - brightness profiles of the bulge and disk along both the semi - major and semi - minor axis ; given that the two components can be twisted by several degrees , this allow us to make a consistent comparison . for all galaxies for which we attempted a ( srsic - profile bulge plus exponential - profile disk ) decomposition , i.e. , all galaxies which are not classified as ellipticals or irregulars , we classify the double - component model fits into the following categories : lcc + @xmath9-band & total ( 96@xmath27 ) & total ( 81@xmath27 ) + + elliptical & 98@xmath27 & - + s0 & 95@xmath27 & @xmath355 + bulge - dom . spiral & 97@xmath27 & @xmath356 + intermediate spiral & 94@xmath27 & @xmath357 + late spiral & 96@xmath27 & @xmath358 + + @xmath10-band & total ( 96@xmath27 ) & total ( 77@xmath27 ) + + elliptical & 100@xmath27 & - + s0 & 95@xmath27 & @xmath358 + bulge - dom . spiral & 97@xmath27 & @xmath359 + intermediate spiral & 94@xmath27 & @xmath360 + late spiral & 97@xmath27 & @xmath361 + 1 . bona - fide bulge+disk decompositions . these models have to satisfy the following criteria : the bulge dominates the light in the inner regions , and the disk component dominates at large radii ; furthermore , the bulge half - light radius has to be smaller than the disk half - light radius . in the @xmath9- and @xmath10-bands , precisely 51% and 61% of the galaxies in our sample fall into this class . 2 . another @xmath119 and @xmath110 @xmath9 and @xmath10 models are classified as pure " ( bulge - less ) disks ( @xmath362 ) . 3 . for the remaining @xmath306 of disk galaxies , the bulge+disk decompositions failed to match either of the above categories . we separate a few cases within this class . in about @xmath265 of such galaxies , the bulge component always dominates the surface brightness profile . these models are not necessarily unphysical decompositions : some galaxies which fall in this category have a srsic indices @xmath170 and are simply disk - dominated galaxies . gim2d hence correctly recovers the information that the structure is dominated by the disk component . nonetheless the @xmath100 ratio are not considered as valid . + we consider instead as truly unphysical those fits ( @xmath363 ) which result in a disk half - light radius that is smaller than the bulge s half - light radius and/or in which the disk and bulge profiles are inverted " , i.e. , the disk dominates in the inner regions and the bulge dominates in the outer regions . also unphysical were judged those fits in which the bulge and disk surface brightness radial profiles cross twice ( again , see also @xcite ) . + we re - fitted all these problematic cases with the software galfit @xcite , which allows for more stringent constraints on the input parameter . as initial guesses for the galaxy sizes , position angles , ellipticity and magnitudes we used the values obtained from sextractor . + we were able to recover 30% of the previously - classified unphysical fits . for a total of @xmath364 ( @xmath16band ) and @xmath365 ( @xmath17band ) of zens disk galaxies , however , also galfit did not converge to a physical bulge+disk decomposition . the fraction of successful bulge+disk decompositions for each morphological class are summarized in table [ tab : gim2drejected ] . the largest formal failure rate is observed for the late - type disks , as in this case many fits results in a dominating srsic - bulge profile with @xmath170 , as mentioned before . these are meaningful fits , but not meaningful bulge+disk decompositions ; these systems are considered to be single - component , disk - dominated galaxies . excluding these late - type disks , galaxies which do not have bulge+disk decompositions in either one or both filters sum up to @xmath8620@xmath27 of the s0- to intermediate - type disk sample . in @xmath366 of galaxies belonging to genuine bulge+disk decompositions , the gim2d fits result in rather elongated bulges ( @xmath367 ) . a closer inspection showed hat a good fraction of these models are associated to galaxies classified as barred disks . in a few other cases , a resolved bar is not observed and hence these central structures may either be unresolved bars or truly flattened bulges ( although we ourselves wonder whether such distinction is meaningful , not only from a descriptive but also from a physical perspective ) . it has been argued that neglecting the bar component when modeling the light profile can affect the best fit parameters and cause an over - estimation of the bulge - to - total ratio of a factor of two to four , and an artificial increase of bulge effective radii ( @xcite , @xcite , @xcite ) . to test the importance of such an effect in our sample , we carried out on those galaxies classified as barred in section [ sec : bars ] a bulge+disk+bar decomposition with the galfit software by modeling the bar with a srsic profile having @xmath368 . the addition of a third component can introduce further degrees of degeneracy : applying only broad constraints on the bar structural properties resulted in a high fraction of unphysical models . for this reason we used the results of the isophotal analysis to constrain the bar structural parameters , when available , and to stick to bulge+disk only fits ( without a bar component ) for galaxies without this a priori information . given possible degeneracies between bulge and bar components , when appropriate we will discuss our results with and without those gim2d fits which result in very elongated bulge components . specifically , for galaxies with an isophotal fits , we used the bar size , ellipticity and position angle measured in section [ sec : bars ] as initial guesses for the fits . furthermore we imposed the bar half - light radius to be within @xmath369 of the size measured with the method , and at the same time , to be larger than the bulge @xmath64 and smaller than the disk scale - length . this avoids the inversion of the disk and bar components or fitting a spurious nuclear " bar . likewise , we limited isophotal twists between the bulge and the bar to be within @xmath370 and also imposed a minimum srsic index , @xmath371 , and a maximum ellipticity , @xmath372 , to the bulge . extensive tests validated the effectiveness of these constraints in leading to reliable fits when the extra degrees of freedom introduced by the bar parameters are allowed in the fits . the results are presented in the top right panels of figure [ fig : galfit_gim2d ] ( red points ) . consistently with previous work , comparing the gim2d b / t ratios with those obtained from the bulge+disk+bar galfit fits , we find that the latter tend to be typically smaller , although exceptions are present . only in @xmath7 of the cases the difference is of a factor of @xmath181 or larger . for most of the galaxies the difference in the final parameters are quite modest , with an average difference of @xmath373 . in several cases , fitting also a bar brings the galfit b / t values in better agreement with those derived with gim2d , indicating that the algorithms involved in the derivation of the b / t value in gim2d are less sensitive to the presence of a bar component . we thus conclude that @xmath93 adding the bar component may no doubt be relevant in some analyses of our barred sample , for which we will utilize the bulge+disk+bar decompositions ; @xmath95 the bulge+disk gim2d fits return generally robust descriptions of these galactic subcomponents , and we thus assume these fits as our fiducial two - component description of the galaxy light distributions . we furthermore tested the general consistency between the parameters obtained with bulge+disk fits performed respectively with galfit and gim2d . specifically , we ran bulge+disk galfit decompositions on all disk galaxies with available gim2d models ( independent of the availability of the isophotal profiles ) , and compared with each other the best fit parameters obtained with the two independent codes . this is illustrated in the four panels of figure [ fig : galfit_gim2d ] ( black points ) , for galaxies with reliable galfit and gim2d fits ( see above ) . the relations shown in the figure are consistent with what we expect from the discussion presented in section [ sec : bulgediskbias ] : the disk scale length is the most robustly determined parameter , with little scatter among the two measurements , whereas the bulge structural properties are subject to the largest discrepancies . the b / t ratios obtained by the two codes agree on average reasonably well . the discrepancies observed in b / t in a minority of cases is mostly caused by the differences in the bulge structural parameters . we note that we found a general tendency to associate low srsic indices ( @xmath170 ) to the bulge component in the galfit fits . this is particularly true for disk dominated galaxies ( these galaxies have single srsic indices @xmath170 ) for which galfit often converges to a double component solution with @xmath374 but @xmath170 . such bulges have furthermore elongations and sizes similar to that of the disk . in these cases , gim2d produces instead a model which is either a pure disk ( @xmath375 ) or very close to it ( @xmath251 ) . such galfit fits were flagged as not reliable , and are not plotted in figure [ fig : galfit_gim2d ] . in our studies we will adopt as our fiducial bulge+disk decompositions those derived with gim2d ; we flag however in our catalog individual galaxies with substantially different ( and yet both formally valid ) gim2d and galfit fits , so to be able to test against these uncertainties our results concerning the properties of bulges and disks . as discussed in section [ sec : gim2dmeasurements ] , reliable srsic fits and bulge+disk decompositions are not available for some galaxies in our sample ( see also appendix [ app : filterbd ] ) . we inspected whether the galaxies that remained without either or both analytical fits were somehow a biased component of our zens sample ( apart from having on average , as expected , a more complex light distribution ) . figure [ fig : failedgim2d ] shows the distributions of galaxy masses and @xmath17band half - light radii for such galaxies with no gim2d models . galaxies without any ( single- and double - component ) size measurements , we use the zest+ radii , _ corrected _ as described in section [ sec : correctionmaps ] . to establish the presence / absence of possible biases , for comparison we also show the global distributions of masses and sizes of galaxies with reliable analytical fits . a summary of the fraction of failed models for each morphological type is given in table [ tab : gim2drejected ] . galaxies with no single srsic fits tend to populate the high mass / radius tail of the distributions and their incidence is highest for intermediate - type disks . for these galaxies we will use in our zens analyses the corrected zest+ sizes , which , on the remaining galaxies , we have shown to be in excellent agreement with the ( corrected ) sizes derived from the analytical fits ; only 4% of galaxies are in this category . we hence reckon that the impact on the final results will be in any case negligible . although the overall success of the bulge+disk decompositions is lower than the success rate of the single - component fits ( see table [ tab : gim2drejected ] ) , galaxies with no bulge and disk parameters are more uniformly distributed in mass and size relative to the global zens population . it is thus reasonable to assume that the exclusion of these galaxies from analysis based on bulge / disk parameters will lower the statistical significance of the results , but should not introduce strong biases . correction maps for sizes and magnitudes , similar to those shown in the main text in section [ sec : simulations ] for the median psf - fwhm of the zens wfi @xmath10 images , are shown here for the two bracketing " i.e. , the best and the worst psf fwhm values our dataset . a comparison of the corresponding correction maps derived for the three different psf sizes clearly shows the major impact of psf - blurring in the ( uncorrected ) structural measurements . the implementation of our correction maps rends all measurements comparable onto a consistently calibrated grid . as commented in section [ sec : simulations ] for the zest+ measurements , there are regions of the observed @xmath4-@xmath38-@xmath142-@xmath64 hyper - plane which are highly degenerate : they can be populated by models with intrinsic parameters which originate in that same region of parameter space , as well as by models with very different intrinsic " ellipticity or concentration , which are scattered into that given bin by observational errors . it is thus important to verify that the size and magnitude corrections that we infer in these regions of parameter space are not strongly dependent on the precise way in which we populate the input grid of simulated models . if this were the case and without an a priori knowledge of the relative fraction of the two populations in the real universe , the derivation of the corrections in the parameters space would be affected by biases introduced by the precise choice of the simulation grid . to assess whether our correction suffer from this problem , we calculated the corrections in these troublesome " grid points by using only those models with original , intrinsic @xmath182 values in that same bin , and , for comparison , using only models which were originated in a different @xmath182 bin and were scattered into that bin by measurement errors . in many cases we found that both model samples gave consistent results for the corrections to the magnitude and sizes , which we could then use for our real galaxies . this is illustrated in figure [ fig : zestcorrections_withnocontaminers ] for the lowest ellipticity bin , which is also the one with the highest contamination , and for models convolved with the typical @xmath10-band psf . the corrections derived with the two sub - samples are reassuringly very similar over a wide are of size - magnitude space ( and thus very similar to the global corrections presented in figure [ fig : zestcorrections ] ) . we therefore argue that our implementation of the correction scheme of section [ sec : correctionmaps ] is free from biases introduced by the choice of the simulation grid . [ cols="^ " , ] finally we show the postage - stamp images of the zens galaxies , split according to their morphological type . the relevant structural parameters characterizing each type are indicated on top of each stamp image . the @xmath221 single - component profiles that _ define _ our elliptical galaxy sample are also shown .
we present structural measurements for the galaxies in the @xmath0 groups of the _ zurich environmental study _ , aimed at establishing how galaxy properties depend on four environmental parameters : group halo mass @xmath1 , group - centric distance @xmath2 , ranking into central or satellite , and large - scale structure density @xmath3 . global galaxy structure is quantified both parametrically and non - parametrically . we correct all these measurements for observational biases due to psf blurring and surface brightness effects as a function of galaxy size , magnitude , steepness of light profile and ellipticity . structural parameters are derived also for bulges , disks and bars . we use the galaxy bulge - to - total ratios ( b / t ) , together with the calibrated non - parametric structural estimators , to implement a quantitative morphological classification that maximizes purity in the resulting morphological samples . we investigate how the concentration @xmath4 of satellite galaxies depends on galaxy mass for each hubble type , and on @xmath1 , @xmath2 and @xmath3 . at galaxy masses @xmath5 , the concentration of disk satellites increases with increasing stellar mass , separately within each morphological bin of b / t . the known increase in concentration with stellar mass for disk satellites is thus due , at least in part , to an increase in galaxy central stellar density at constant b / t . the correlation between concentration and galaxy stellar mass becomes progressively steeper for later morphological types . the concentration of disk satellites shows a barely significant dependence on @xmath3 or @xmath2 . the strongest environmental effect is found with group mass for @xmath6 disk - dominated satellites , which are @xmath7 more concentrated in high mass groups than in lower mass groups .
solar and stellar oscillations are a powerful tool to probe the interior of stars . in this paper we classify stellar oscillations into solar - like or deterministic . solar - like oscillations are stochastically excited by turbulent convection and are present in the sun and other main - sequence , subgiant , and giant stars ( see _ e.g. _ @xcite , @xcite and references therein ) . deterministic oscillations are seen in classical pulsators and have mode lifetimes much longer than any typical observational run ; one of the best studied objects in this class is the pre - white dwarf pg1159@xmath2035 also known as gw vir @xcite . in practice , observations of solar - like or deterministic pulsations always have an additional stochastic component due to instrumental , atmospheric , stellar , or photon noise . an important aspect of helio- and asteroseismology is the determination of the parameters of the global modes of oscillation , especially the mode frequencies . in the case of the sun , it is known @xcite that the measurement precision is limited by the stochastic nature of the oscillations ( realization noise ) . and have shown that realization noise is expected to scale like @xmath3 , where @xmath0 is the total duration of the observation . a common practice is to extract the solar mode parameters from the power spectrum using maximum likelihood estimation ( mle , see _ e.g. _ ; ; ; ; ) . in its current form , however , this method of analysis is only valid for uninterrupted time - series . this is a significant limitation because gaps in the data are not uncommon ( daily cycle , bad weather , technical problems ) . the gaps complicate the analysis in fourier space : the convolution of the data with the observation window leads to correlations between the different fourier components . the goal of this paper is to extend the fourier analysis of solar and stellar oscillations to time series with gaps , using appropriate maximum likelihood estimators based on the correct statistics of the data . section [ section : problem ] poses the problem of the analysis of gapped time series in fourier space . in section [ section : pdf_correlation ] we derive an expression for the joint probability density function ( pdf ) of the observations , taking into account the frequency correlations . our answer is consistent with an earlier ( independent ) derivation by . based on this pdf , we derive maximum likelihood estimators in section [ section : likelihood_estimation ] . in section [ section : fitting_nocorr ] we recall the `` old method '' of maximum likelihood estimation based on the unjustified assumption that frequency bins are statistically independent . section [ section : test_setup ] explains the set up of the monte - carlo simulations , used to test the fitting methods on artificial data sets . in section [ section : fitting_results ] we present the results of the monte - carlo simulations and compare the new and old fitting methods . for the sake of simplicity , we consider only one mode of oscillation at a time ( solar - like or sinusoidal ) . we present several cases for which our new fitting method leads to a significant improvement in the determination of oscillation parameters , and in particular the mode frequency . let us denote by @xmath4 the time series that we wish to analyse . it is sampled at times @xmath5 , where @xmath6 is an integer in the range @xmath7 , and @xmath8 one minute is the sampling time . all quantities with a tilde are defined in the time domain . the total duration of the time series is @xmath9 . by choice , all of the missing data points were assigned the value zero : this enables us to work on a regularly sampled time grid . formally , we write @xmath10 where @xmath11 is the uninterrupted time series that we _ would _ have observed if there had been no gaps and @xmath12 is the window function defined by @xmath13 if an observation is recorded at time @xmath14 and @xmath15 otherwise . the @xmath11 is drawn from a random process , whose statistical properties will be discussed later . we define the discrete fourier transform @xmath16 of @xmath17 by @xmath18 where @xmath19 is the frequency and @xmath20 . note that @xmath21 and @xmath22 , where the star denotes the complex conjugate . the fourier transform has periodicity @xmath23 or twice the nyquist frequency . our intention is not to fit the complete fourier spectrum , but a rather small interval that contains one ( or a few ) modes of stellar oscillation . thus , we extract a section of the data of length @xmath24 starting from a particular frequency @xmath25 , as shown in figure [ fig : convolution_scheme](c ) . this subset of the data is represented by the vector @xmath26^t$ ] with components @xmath27 using the above definition of the fourier transform , the vector @xmath28 is given by the convolution of @xmath29 with the window @xmath30 : @xmath31 the integer @xmath32 in equation ( [ equ : fourier_elements ] ) refers to the cutoff frequency @xmath33 beyond which the observation window has no significant power . truncating the window function at frequency @xmath33 is a simplification of the general problem . our main goal , however , is , given a known window function , to study its effects on the determination of the parameters of stellar oscillations . figure [ fig : convolution_scheme ] is a schematic representation in fourier space of the convolution of a single mode of oscillation by the window function . the observed signal is spread over some frequency range and , as we shall see later , its statistical properties are affected . we note that , in practice , one can never completely isolate one single mode of oscillation in the power spectrum . in particular , other modes with frequencies outside of the fitting range can leak into it after convolution by the temporal window function . hence , fitting one mode of oscillation is a simplification . but our first objective is to try to study the effects of gaps , independently from the complications associated with a badly specified model . with the window function @xmath30 . for the sake of simplicity , only the power spectra of the different quantities are shown here . panel ( a ) shows the window function @xmath30 and its cutoff frequency @xmath33 , panel ( b ) the unconvolved signal @xmath29 , and panel ( c ) the observed signal @xmath16 . note that the selected section of the observed signal , starting at frequency @xmath33 , is of length @xmath24 , while the unconvolved signal is of length @xmath34 . , scaledwidth=95.0% ] equation ( [ equ : fourier_elements ] ) can be rewritten in matrix form as @xmath35 where the vector @xmath36^t$ ] of length @xmath37 is defined by @xmath38 and @xmath39 $ ] is the @xmath40 rectangular window matrix with elements @xmath41 , where @xmath42 and @xmath43 : @xmath44 . \label{equ : windowmatrix}\ ] ] note that @xmath45 and that @xmath46 is of rank @xmath24 . equation ( [ equ : fourier_signal ] ) is the master equation . our goal is to extract the stellar oscillation parameters ( contained in @xmath47 ) , given the uncomplete information @xmath28 . here we describe the basic assumptions that we make about the statistics of the data in the fourier domain . the unconvolved signal [ @xmath47 ] consists of a deterministic component [ @xmath49 and a zero - mean stochastic component [ @xmath50 such that @xmath51 the deterministic component @xmath52 may include deterministic stellar oscillations that are long - lived compared to the total length of the observation . the stochastic component @xmath53 may include various sources of noise ( _ e.g. _ stellar convection , photon noise , atmospheric noise , _ etc . _ ) and stochastically excited pulsations as observed on the sun . we assume that the @xmath54 are @xmath37 independent random variables in the fourier domain . this is equivalent to saying that the stochastic component of the signal in the time domain is stationary . we further assume that @xmath53 is a gaussian random vector with independent real and imaginary parts and covariance matrix @xmath55=\sigma_i^2 \delta_{ij } , \quad i , j = 0,1,\dots , m+2p-1 , \label{equ : sigma^2}\ ] ] where @xmath56 denotes the expectation value and @xmath57 is the standard deviation of @xmath54 at frequency @xmath58 . one may invoque the central limit theorem to justify the choice of gaussian distributions . the quantity @xmath59 is the expected power spectrum at frequency @xmath58 , which may include background noise and peaks corresponding to the modes of oscillations @xcite . in terms of a complex gaussian random vector @xmath60 with unit covariance matrix , @xmath61=i_{m+2p}$ ] , we can rewrite @xmath53 as @xmath62 where @xmath63 is the @xmath64 diagonal matrix @xmath65 we emphasize that , although the @xmath54 are uncorrelated random variables , the @xmath66 are correlated because of the multiplication of @xmath47 by the window matrix [ equation ( [ equ : fourier_signal ] ) ] . in this section we derive an expression for the joint probability density function of the observed signal @xmath28 . this problem had already been solved by . we reach the same conclusion , independently and with more compact notations . we start by rewriting the master equation , equation ( [ equ : fourier_signal ] ) , as @xmath67 where @xmath68 is a @xmath69 matrix with rank @xmath24 and singular value decomposition @xcite @xmath70 here the superscript @xmath71 denotes the hermitian conjugate and @xmath72 and @xmath73 are unitary matrices of dimensions @xmath74 and @xmath64 respectively , _ i.e. _ @xmath75 and @xmath76 . the @xmath69 matrix @xmath77 can be written as @xmath78 , \quad \lambda=\mbox{diag}(\lambda_0,\lambda_1 , \dots , \lambda_{m-1 } ) , \label{equ : svd}\ ] ] where @xmath79 are the @xmath24 ( positive ) singular values of the matrix @xmath80 . thus , there exists a vector @xmath81 such that @xmath82 \xi . \label{equ : fourier_signal_xi}\ ] ] since @xmath60 has unit covariance matrix and @xmath73 is unitary , the vector @xmath83 is a complex gaussian random vector of size @xmath37 with unit covariance matrix . it is obvious from equation ( [ equ : fourier_signal_xi ] ) that there exists a lower - rank complex gaussian random vector of length @xmath24 , @xmath84^t$ ] , such that @xmath85 the variables @xmath86 are dummy variables , which do not enter in the description of @xmath28 . equation ( [ equ : signalll ] ) is an important step , as the vector @xmath28 of length @xmath24 is now expressed in terms of @xmath24 independent complex gaussian variables . this enables us to write the pdf of @xmath28 as @xmath87 where @xmath88 denotes the pdf of @xmath89 and @xmath90 is the jacobian of the linear transformation @xmath91 . since @xmath89 is a complex gaussian random vector with unit covariance , _ i.e. _ @xmath92=i_m$ ] , we have @xmath93 where we used the notation @xmath94 . since @xmath72 is unitary and @xmath95 is diagonal and real , the jacobian of the transformation is given by @xmath96 combining equations ( [ equ : pdf_transorm ] ) , ( [ eq.pdf_eta ] ) , and ( [ equ : jacobian ] ) , we get the joint pdf of the observed vector @xmath28 : @xmath97 the above expression is , perhaps , more elegantly written as @xmath98 in terms of @xmath99 , the @xmath100 moore - penrose generalized inverse of @xmath80 @xcite,@xmath101 where @xmath102 is the transpose of @xmath77 in which the singular values are replaced by their inverse . one may ask , after the fact , if the quantity @xmath103 in equation ( [ equ : pdf_transorm ] ) is always defined . the answer would appear to be yes since the moore - penrose generalised inverse of @xmath80 is perfectly well defined . it is not excluded , however , that some singular values @xmath104 could be infinitesimally small . we have not encountered any such difficulty with the test cases given in section [ section : fitting_results ] . should @xmath80 be ill - conditioned in other cases , a simple truncated svd would help avoiding a numerical problem . before discussing the implementation of the method in section [ section : likelihood_estimation ] , we should like to draw attention to a parallel between fitting data with temporal gaps and fitting data with spatial gaps . in order to understand this analogy , we refer the reader to the work of @xcite ( 1998 , section 3.3.4 ) who discuss how to interpret the spatial leaks of non - radial oscillations that arise from the fact that only half of the solar disk can be observed from earth . their approach is similar to the one developed in this paper . let us assume that the stellar oscillation model that we are trying to fit to the data depends on a set of @xmath105 parameters @xmath106 . these parameters may be the amplitude , the phase , the frequency , the line asymmetry , the noise level , _ etc_. the basic idea of maximum likelihood estimation is to pick the estimate @xmath107 that maximizes the likelihood function . the likelihood function is another name for the joint pdf [ equation ( [ equ : pdf ] ) ] evaluated for the sample data . in practice , one minimizes @xmath108 rather than maximizing the likelihood function itself . in the above expression , the quantities @xmath99 and @xmath104 all depend implicitly on the model parameters @xmath109 through the covariance matrix @xmath63 . the vector @xmath52 also depends on the model parameters in the case deterministic oscillations . the probability of observing the sample data is greatest if the unknown parameters are equal to their maximum likelihood estimates @xmath110 : @xmath111 the method of maximum likelihood has many good properties @xcite . in particular , in the limit of a large sample size ( @xmath24 large ) , the maximum likelihood estimator is unbiased and has minimum variance . what is particularly new about our work is the minimization of the likelihood function given by equation ( [ equ : likelihood ] ) . we use the direction set method , or powell s algorithm , to solve the minimization problem with a computer . in practice , the result of the fit depends on the initial guess and the fractional tolerance of the minimisation procedure ( the relative decrease of @xmath112 in one iteration ) . the dependence of the fitted parameters on the initial guess is due to the fact that the function @xmath112 may have local minima in addition to the global minimum . we will address this issue in more detail in section [ section : fitting_results ] . in the case of solar - like oscillations , there is no deterministic component and the log - likelihood becomes @xmath113 if background white noise is the only stochastic component then @xmath114 the log - likelihood function becomes @xmath115 splitting the unknowns @xmath116 into the parameters describing the oscillations , @xmath117 and the noise level @xmath118 , the minimization problem reduces to finding the most likely estimates @xmath119 where @xmath120 . the noise level is explicitly given by @xmath121 \| .\ ] ] maximum likelihood estimation has been used in the past to infer solar and stellar oscillation parameters , even in the case of gapped time series . the joint pdf of the observations was assumed to be the product of the pdfs of the individual @xmath66 , as if the frequency bins were uncorrelated . for comparison purposes , we briefly review this ( unjustified ) approach . according to equation ( [ equ : fourier_signal_2 ] ) , the pdf of @xmath66 is a normal distribution @xmath122 with mean @xmath123 and variance @xmath124 under the ( wrong ) assumption that the @xmath66 are independent random variables , the joint pdf of @xmath28 becomes @xmath125 where the superscript `` nc '' stands for `` no correlation '' . this joint pdf uses the correct mean ( @xmath126 ) and variance ( @xmath127 ) of the data , but it ignores all the non - vanishing cross - terms @xmath128 $ ] . in other words , the spread of power implied by the convolution with the window is taken care of , but not the proper statistics . under the same simplifying `` no correlation '' assumption , the log - likelihood function is @xmath129 where the @xmath126 and @xmath127 are implicit functions of the model parameters @xmath109 . if the signal has no deterministic component ( @xmath130 ) , then the power spectrum [ @xmath131 has the expectation value @xmath132= v_i$ ] . thus , in the case of purely solar - like oscillations , we recover the standard expression @xcite : @xmath133 while the above expression is perfectly valid for uninterrupted data , it is not justified when gaps are present . the parameters @xmath134 that minimize @xmath135 are not optimal , as will be shown later using monte - carlo simulations . when @xmath136 , the `` no - correlation '' log - likelihood function simplifies to @xmath137 the minimization problem becomes @xmath138 where @xmath139 depends on the oscillation parameters @xmath117 . the noise level is explicitly given by @xmath140 so far we have considered a general signal which includes a deterministic component and a stochastic component . the parametrisation of each component depends on prior knowledge about the physics of the stellar oscillations . solar - like pulsations are stochastic in nature and no deterministic component is needed in this case . on the other hand , long - lived stellar pulsations are treated as deterministic . some stars may support both deterministic and stochastic oscillations . in this section , we model the two cases separately . we want to test the fitting method [ equations ( [ equ : likelihood ] ) and ( [ equ : likelihood - estimator ] ) ] by applying it to simulated time series with gaps . for comparison , we also want to apply the old fitting method ( section [ section : fitting_nocorr ] ) to the same time series . we need to generate many realizations of the same random process in order to test the estimators for bias and precision : this is called monte - carlo simulation . in section [ section : window_function ] we discuss the generation of the synthetic window functions . we then discuss the parametrisation of the solar - like oscillations ( section [ section : model_solar_like ] ) and the deterministic oscillations ( section [ section : model_deterministic ] ) used to simulate the unconvolved signal . ] used in this paper . from top to bottom , the duty cycle is ( a ) 100% , ( b ) 66% , ( c ) 30% , and ( d ) 15% . the main periodicity of the window is 24 hours for cases ( b ) and ( c ) , and 48 hours for window ( d ) . all windows are truncated at frequency @xmath141 @xmath109hz.,scaledwidth=80.0% ] we generate three different observation windows , corresponding to different duty cycles . the observation windows are first constructed in the time domain . by definition , @xmath142 is set to one if an observation is available and zero otherwise . the total length of all time series is fixed at @xmath143 days ( frequency resolution @xmath144hz ) . a window function is characterized by two main properties : the duty cycle ( fraction of ones ) and the average periodicity . a typical window function for a single ground - based site has a 24-hour periodicity . in order to deviate slightly from purely periodic window functions we introduce some randomness for the end time of each observation block . figure [ figure : windowfunction](b)(d ) shows the power spectra of the three window functions . the first and second window functions have a main periodicity of 24 hours and duty cycles of @xmath145 and @xmath146 respectively . two side lobes occur at frequencies @xmath147hz and @xmath148hz . the non - vanishing power between the side lobes is due to the deviation from a purely periodic window . the third window function has a main periodicity of 48 hours and a duty cycle of only 15% . all of these window functions are not unrealistic . we apply a sharp low - pass filter at frequency @xmath141 @xmath109hz ( @xmath149 ) to all window functions . the power at higher frequencies corresponds to about 5% of the total power in the windows . this truncation is needed to apply the fitting algorithm , which assumes that there exists a frequency @xmath33 beyond which the power in the window vanishes , _ i.e. _ that the window function is band limited . we generate the realizations of the unconvolved solar - like oscillation signal directly in the fourier domain . we consider a purely stochastic signal ( @xmath130 ) and a single mode of oscillation . since we assumed stationarity in the time domain , the fourier spectrum of the unconvolved signal for one single mode can be written as @xmath150^{1/2 } \eta_i , \quad i=0,1,\dots , m+2p-1 , \ ] ] where @xmath151 describes the line profile of the mode in the power spectrum , @xmath152 is the mode s maximum power , @xmath153 is the variance of the background noise , and @xmath154 is a centered complex gaussian random variable with unit variance and independent real and imaginary parts . solar - like oscillations are stochastically excited and intrinsically damped by turbulent convection @xcite . the expectation value of the power spectrum is nearly lorentzian , except for some line asymmetry ( _ e.g. _ , ) . here we use a simple asymmetric line profile : @xmath155 where @xmath156 is the resonant frequency , @xmath157 is the asymmetry parameter of the line profile ( @xmath158 ) , and @xmath159 is a measure of the width of the line profile . we refer to @xmath160 as the signal - to - noise ratio in the power spectrum . as @xmath157 tends to zero , @xmath159 becomes the full width at half maximum ( fwhm ) of the power spectrum and @xmath161 the mode lifetime . there are five model parameters , @xmath162 . once the unconvolved signal @xmath47 has been generated in the fourier domain , the observed signal @xmath28 is obtained by multiplication with the window matrix @xmath46 , as explained above . in the time domain , we consider a purely sinusoidal function on top of white background noise : @xmath163 the first term describes the deterministic component of the signal , where @xmath164 is the amplitude , @xmath156 the mode frequency , and @xmath165 the phase of the mode . the second term is stochastic noise with standard deviation @xmath166 . the @xmath167 are @xmath168 normally distributed independent real random variables with zero mean and unit variance . the observed signal is obtained by multiplying @xmath169 by the window @xmath142 in the time domain . the model parameters are @xmath170 . we have defined the signal and the noise in the time domain , but a definition of signal - to - noise ratio in the fourier domain is desirable . on the one hand , the variance of the noise in the fourier domain is @xmath171 where @xmath172 is the total power in the window . on the other hand , the maximum power of the signal in fourier space is @xmath173 , where @xmath174 is the power of the window at zero frequency . thus , by analogy with the solar - like case , it makes sense to define the signal to noise ratio in the fourier domain as @xmath175 in practice we fix @xmath164 and @xmath160 and deduce the corresponding noise level @xmath166 . several hundreds of realizations are needed in order to assess the quality of a fitting method . we do not intend to test all possible combinations of mode parameters but we want to show a few cases for which the new fitting method provides a significant improvement compared to the old fitting method . figure [ fig : realization_fit_stoch ] shows one realization of a simulated mode of solar - like oscillation with input parameters @xmath176 @xmath109hz , @xmath177 @xmath109hz , @xmath178 , @xmath179 , and @xmath180 . the signal to noise ratio is @xmath181 and the window function is 30% full ( see figure [ figure : windowfunction](c ) ) . the mode lifetime is @xmath182 hours . figure [ fig : realization_fit_stoch](a ) displays the real and imaginary parts of the fourier transform @xmath28 , together with the standard deviation of the data ( nc fit in blue , new fit in red , expectation value in green ) . figure [ fig : realization_fit_stoch](b ) shows the power spectrum and the fits . notice the sidelobes introduced by the convolution of the signal with the window functions . the `` no - correlation '' fit is done on the power spectrum [ equation ( [ eq.nc ] ) ] , while the new fit is performed in complex fourier space [ equation ( [ equ : likelihood ] ) ] . hz , linewidth @xmath183hz , and @xmath181 . the window function is 30% full . panel ( a ) shows the real and imaginary parts of the fourier spectrum . panel ( b ) shows the power spectrum . the vertical dashed lines represent the width of the window function . also shown are the new fit ( red ) , the old fit ( blue ) , and the expectation value ( green ) . ] each fit shown in figure [ fig : realization_fit_stoch ] corresponds in fact to the best fit out of five fits with different initial guesses . for each realization , we use the frequency guesses @xmath184 @xmath109hz for @xmath156 . the last two frequency guesses correspond to the frequencies of the two main sidelobes of the window function ( figure [ figure : windowfunction]c ) . for the other parameters , we choose random guesses within @xmath185% of the input values . the reason for using several guesses is to ensure that the fit converges to the global maximum of the likelihood , not to a nearby local maximum , _ i.e. _ that the estimates returned by the code are the mle estimates defined by equation ( [ equ : likelihood - estimator ] ) . in some cases , the global maximum coincides with a sidelobe at @xmath186 @xmath109hz from the main peak . we note that the new fitting method requires a much longer computing time than the old nc method : typically , three hours on a single cpu core for a single realization ( five guesses , five fits ) . for the particular realization of figure [ fig : realization_fit_stoch ] , the new fit is closer to the expectation value ( _ i.e. _ is closer to the answer ) than the old nc fit . no conclusions should be drawn , however , from looking at a single realization . in order to test the reliability of each fitting method , we computed a total of 750 realizations with the same input parameters as in figure [ fig : realization_fit_stoch ] and the same window function ( 30% full ) . the quality ( bias and precision ) of the estimators can be studied from the distributions of the inferred parameters . as shown by the distributions of figure [ fig : distr_stoch_sn30_wa ] the new fitting method is superior to the old nc method . this is true for all the parameters , in particular the mode frequency @xmath156 . the distributions for the mode frequency ( figure [ fig : distr_stoch_sn30_wa](a ) ) are quite symmetric and gaussian - like , although the old fitting method leads to a significant excess of values beyond the two-@xmath187 mark . we note that , in general , the old fitting method is more sensitive to the initial frequency guess . also the estimates of the linewidth @xmath159 and the mode power @xmath152 are significantly more biased with the old fitting method than with the new one ( figures [ fig : distr_stoch_sn30_wa]b , [ fig : distr_stoch_sn30_wa]c ) . it is worth noting that the fits return a number of small @xmath159/large @xmath152 estimates away from the main peaks of the distributions , less so for the new fits . these values correspond to instances when the signal barely comes out of the noise background . the new fit returns the noise level [ @xmath153 ] with a higher precision and a lower number of underestimated outliers than the old method ( the ouliers are represented by the vertical bars in figure [ fig : distr_stoch_sn30_wa](d ) ) . although the estimation of the asymmetry parameter is unbiased with the new fitting method ( figure [ fig : distr_stoch_sn30_wa](e ) ) , the uncertainty on @xmath157 is so large that it probably could have been ignored in the model . , ( b ) linewidth @xmath159 , ( c ) mode power @xmath152 , ( d ) noise level @xmath153 , and ( e ) asymmetry parameter @xmath157 . the black lines show the results obtained with the new fitting method and the grey lines show the old `` no - correlation '' fits . the vertical dashed line in each plot indicates the input value . the horizontal lines in panel ( a ) are intervals containing 68% of the fits for the new ( black line ) and the old ( grey line ) fitting methods . the thick black and grey vertical lines in panel ( d ) give the numbers of outliers with @xmath189 . ] quantitative estimates of the mean and the dispersion of the estimators are provided in table 1 . because the distributions of the estimated parameters are not always gaussian and may contain several outliers , we compute the median ( instead of the mean ) and the lower and upper bounds corresponding to @xmath190% of the points on each side of the median ( instead of the one-@xmath187 dispersion ) . this definition has the advantage of being robust with respect to the outliers . the notation @xmath191 @xmath109hz in the first row of table [ table : freq_uncertainty ] means that the median mode frequency is @xmath192 @xmath109hz and that @xmath193 of the fits belong to the interval @xmath194 $ ] @xmath109hz . we emphasize that the subscript @xmath195 and the superscript @xmath196 do not refer to an uncertainty in the determination of the median : the median is known to a much higher precision thanks to the large number of realizations . later we relax the language and refer to the `` one-@xmath187 uncertainty '' to mean the average @xmath197 . rrrr mode parameter & input value & new fitting & old fitting + + @xmath156 [ @xmath109hz ] & @xmath192 & @xmath198 & @xmath199 + @xmath200)$ ] & @xmath201 & @xmath202 & @xmath203 + @xmath204 & @xmath205 & @xmath206 & @xmath207 + @xmath208 & @xmath209 & @xmath210 & @xmath211 + @xmath157 & @xmath212 & @xmath213 & @xmath214 + the numbers from the last two columns in table 1 confirm the analysis of figure [ fig : distr_stoch_sn30_wa ] . the mode frequency can be measured with a precision of @xmath215 @xmath109hz , which is exactly twice better with the new fitting method than with the old one . this gain in precision is very significant and potentially important . since measurement uncertainty scales like @xmath216 @xcite , one may equate the gain in using the proper fitting procedure to an effective increase in the total length of the time series by a factor of four . as seen in table 1 , the linewidth , the mode power , the background noise , and the line asymmetry parameter are all less biased and more precise with the new fitting method than the old one . notice that the larger dispersions in the old - fit case are due in part to non - gaussian distributions with extended tails . @xmath109hz , title="fig : " ] + @xmath109hz , title="fig : " ] cccccc & + duty cycle & main period & average gap & new fitting & old fitting + + 100% & & & @xmath217 & @xmath217 + 66% & 24 hours & 7.4 hours & @xmath218 & @xmath219 + 30% & 24 hours & 16.4 hours & @xmath220 & @xmath221 + 15% & 48 hours & 40.7 hours & @xmath222 & @xmath223 + ] as a function of the window duty cycle [ @xmath224 . the window functions are as defined in section [ section : window_function ] . the red curve shows the 1-@xmath187 monte - carlo mle uncertainties for the new fitting method . the black curve shows the 1-@xmath187 monte - carlo mle uncertainties for the old no - correlation fitting method . the blue curves show the mean cramr - rao lower bounds ( formal error bars ) . the square symbol with a cross at @xmath225 in the left panel is a rough estimate ( see text ) . in the left panel , the input linewidth is @xmath177 @xmath109hz ( see also numbers in table 1 ) . in the right panel , the input linewidth is @xmath226 @xmath109hz , all of the other parameters being the same as in the left panel . in both panels the signal - to - noise ratio is @xmath181 . for reference , the dashed lines have slope @xmath227 . ] here we study how bias and precision change as the window function changes , in particular as the duty cycle changes . we consider the four window functions defined in section [ section : window_function ] with duty cycles [ @xmath224 equal to 15% , 30% , 66% , and 100% . first we consider input parameters of solar - like oscillations that are exactly the same as in the previous section : @xmath176 @xmath109hz , @xmath177 @xmath109hz , @xmath178 , @xmath181 , and @xmath180 . figure [ fig : distribution_window ] shows the distributions of the inferred mode frequencies and linewidths , using the old ( figures [ fig : distribution_window]a , [ fig : distribution_window]b ) and the new ( figures [ fig : distribution_window](c ) , [ fig : distribution_window](d ) ) fitting methods . each fit is the best fit from five different @xmath156 guesses ( see section [ sec : results_solar_like ] ) . the distributions for the 100%-window are identical for the two fitting methods ; this is expected since the old and new fitting methods are equivalent in the absence of gaps . the precision on @xmath156 using the old `` no - correlation '' mle drops fast as the duty cycle decreases ( figure [ fig : distribution_window](a ) ) . this drop is much faster than in the case of the fits that take the frequency correlations into account ( figure [ fig : distribution_window](c ) ) . when the duty cycle is @xmath228 , the frequency estimate is five times better with the new than the old method . the difference is perhaps even more obvious for the linewidth . for the 15% window , it is almost impossible to retrieve @xmath159 with the old fitting method ( figure [ fig : distribution_window]b ) , while the new method gives estimates that are almost as precise as in the no - gap case ( figure [ fig : distribution_window]d ) . the estimates of @xmath159 are significantly less biased with the new method . figure [ fig : distribution_window ] confirms the importance of using the correct expression for the likelihood function . table [ table : freq_uncertainty2 ] gives the medians and half - widths of the @xmath156 distributions . the one-@xmath187 dispersions are plotted as a function of the duty cycle @xmath229 in the left panel of figure [ fig : sigma_nu0 ] . the improvement in the fits is quite spectacular . for example , when @xmath230 the dispersion on @xmath156 is five times less with the new fitting method ( @xmath231 @xmath109hz vs. @xmath232 @xmath109hz ) . with the old method , the uncertainty on @xmath156 increases much faster than @xmath227 as the duty cycle @xmath229 drops ( @xmath233 between the 30% and 15% windows ) . this steep dependence on @xmath229 is worse than `` predicted '' by . in his paper , libbrecht suggested to use the uncertainty @xmath234 where @xmath235 ^ 3 $ ] and @xmath236 is an `` effective '' noise - to - signal ratio . he suggested that the main effect of the gaps is to increase the noise - to - signal ratio @xmath237 , presumably by a factor @xmath238 , itself proportional to @xmath239 . this leads however to a dependence of @xmath240 on @xmath229 which , in our particular case , is closer to @xmath227 than @xmath241 . we suspect that the libbrecht formula underestimates the dispersion because it ignores the frequency correlations . the new fitting method returns a @xmath156 uncertainty that is much less sensitive to the duty cycle , with a variation like @xmath242 ( red curve , left panel of figure [ fig : sigma_nu0 ] ) . this is quite remarkable . that the frequency uncertainty could remain nearly constant for @xmath243 is not really surprising since the average gap ( see numbers in table 2 ) is less than the mode lifetime @xmath244 hours this regime was studied by using a gap - filling method : as long as the signal - to - noise ratio is large enough , the signal can be reconstructed . why the new fit is doing such a good job for duty cycles @xmath245 is , however , puzzling ( at first sight ) , since the average gap ( @xmath246 hours ) is larger than the mode lifetime . this can be understood as follows . for small duty cycles , the time - series is effectively a collection of nearly independent blocks of data , which , for the 30% window function , are eight - hour long on average . since mle simulations tell us that the uncertainty on the mode frequency for an uninterrupted series of eight hours is about @xmath247 @xmath109hz , we would expect for the gapped time series ( @xmath143 days , 24-hour periodicity ) to be able to reach the uncertainty @xmath248 @xmath109hz . this value , represented by the box with a cross in figure 6 , is found to be very close to the mle estimate from the new fits . hence , what matters at very low duty cycle is the number of independent blocks of continuous data . the new fitting method captures this very well , which is satisfying . by comparison , the old no - correlation fitting method does poorly ( black line ) . in order to further investigate this last point , we ran another set of simulations using a mode linewidth @xmath249hz corresponding to a mode lifetime @xmath250 hours , which is significantly smaller than the average gap lengths of the 30% and the 15% windows . the other input parameters remained the same as above . we computed and fitted 1350 realizations . the results are shown in the right panel of figure 6 . for the new fitting method , the dependence of the frequency uncertainty on the duty cycle is about @xmath251 , which is comparable to the previous simulations with @xmath177 @xmath109hz . we conclude that it is really worth solving for the correct minimization problem and that fitting for the phase information in complex fourier space is important to get a good match between the model and the data . of course , this can only be done properly when we have a perfect knowledge of the model , which is the case with these numerical simulations , but is rarely the case with real observations . @xmath109hz ( see figure 5c ) and the right panel is for @xmath226 @xmath109hz . the different curves correspond to different window functions , as indicated in the legend . the means of these distributions ( cramr - rao lower bounds ) give the blue curves plotted in figure [ fig : sigma_nu0 ] . ] monte - carlo simulations are very useful in order to assess the variance and the bias of a particular estimator . when fitting real observations , however , the variance of the estimator can not be computed directly by monte - carlo simulation since the input parameters are , by definition , not known . hopefully , the fit can return a formal error from the shape of the likelihood function in the neighborhood of the global maximum . the cramr - rao lower bound @xcite achieves minimum variance among unbiased estimators . it is obtained by expanding @xmath112 about its minimum . the formal error @xmath252 on the parameter @xmath253 is given by @xmath254 where @xmath255 is the @xmath6th element on the diagonal of the inverse [ @xmath256 of the hessian matrix with elements @xmath257 the cramr - rao formal errors have been used in helioseismology by , for example , , , and . we have computed the formal error on the mode frequency for many realizations and for all window functions . the resulting distributions are shown in figure 7 . the mean formal error from each distribution is plotted in figure 6 . overall the cramr - rao lower bound is remarkably close to the monte - carlo mle uncertainty using the new fitting method ; they are even undistinguishable when @xmath226 @xmath109hz . this is useful information as it means that , on average , the hessian method provides reasonable error estimates . it should be clear , however , that the distributions shown in figure 7 show a significant amount of scatter : the formal error from the hessian may be misleading for particular realizations . . the observation window has a duty cycle of 30% . the simulated data is the thick grey line . the thin black line shows the fit to the data using the new fitting method . the fit with the old method is not shown since it is almost identical . ] figure [ fig : realization_fit_determin ] shows the fourier spectrum of a simulated time series containing a sinusoidal mode of oscillation on top of a white noise background as described in section [ section : model_deterministic ] . in this particular case the observation window with a duty cycle of 30% is used ( see figure [ figure : windowfunction](c ) ) . the input parameters of the sinusoidal function are the mode frequency @xmath258hz , the amplitude @xmath259 , and the phase @xmath260 . the signal - to - noise ratio is @xmath261 . the fit shown in figure [ fig : realization_fit_determin ] was obtained with the new fitting method . since we found no significant difference between the old and the new fitting methods in this case , the old fitting method is not shown . differences between the data and the fit are essentially due to the noise . . the window function with a duty cycle of 30% is used . the black and the grey lines are for the new and old fitting methods respectively . the vertical dashed line in each plot indicates the input value . the parameters shown are ( a ) the mode frequency [ @xmath156 ] , ( b ) the logarithm of the mode amplitude [ @xmath263 , ( c ) the phase of the oscillation [ @xmath264 , and ( d ) the logarithm of the noise level [ @xmath265 ( see section 6.3 ) . notice that the estimate of the noise is biased when frequency correlations are ignored ( old nc fit ) , although by a very small amount . ] we computed 500 realizations of sinusoidal oscillations with the same mode parameters ( frequency , amplitude , phase ) as above , the same observation window ( 30% full ) , but with a signal - to - noise ratio @xmath266 . the resulting distributions of the inferred parameters obtained with the two fitting methods are shown in figure [ fig : distribution_deterministic ] . for this simulation , the known input values were used as an initial guess to speed up the minimization ; we checked on several realizations that it is acceptable to do so when the signal to noise ratio is large . the distributions of the inferred parameters ( figure [ fig : distribution_deterministic ] ) show that for sinusoidal oscillations , the new fitting method does not provide any significant improvement compared to the old fitting method . we emphasize that the fitting parameters can be determined with a very high precision when the noise level is small . in particular , we confirm that the uncertainty of the frequency estimator can be much smaller than @xmath1 ( see figure [ fig : distribution_deterministic](a ) ) . figure [ fig : fsn_det ] shows the median and the standard deviation of the mode frequency for different signal - to - noise ratios . each symbol and its error bar in figure [ fig : fsn_det ] is based on the computation of 500 realizations of sinusoidal oscillations with the same mode parameters as above , the same observation window ( 30% full ) , but various signal - to - noise ratios . since we did not find any significant difference between the two fitting methods , only the results obtained with the new fitting method are shown . figure [ fig : fsn_det ] illustrates that even for a relatively low signal - to - noise ratio of @xmath267 , the standard deviation of the inferred mode frequency is smaller than @xmath1 by a factor of four . for higher signal - to - noise ratios the precision is even more impressive : when @xmath268 , the standard deviation of the mode frequency is about 20 times smaller than @xmath1 . the theoretical value of the standard deviation of the mode frequency obtained by can be extended to the case of gapped data ( cuypers , 2008 , private communication ) as follows : @xmath269 where @xmath164 is the amplitude of the sinusoid in the time domain , @xmath270 is the rms value of the noise , @xmath271 is the number of recorded data points , and @xmath0 is the total observation length . this theoretical uncertainty is overplotted in figure [ fig : fsn_det ] . the match with our monte - carlo measurements is excellent . this confirms that , in this case , it is equivalent to perform the fits in the temporal and in the fourier domains . note that equation ( [ equ : freq_error_determin ] ) is only valid under the assumption that the noise is uncorrelated in the time domain , a condition fulfilled by our simulations . the main reason why the measurement precision is only limited by the noise - to - signal ratio is because perfect knowledge of the model is assumed . ] as a function of signal - to - noise ratio @xmath160 . the duty cycle is 30% . only the results obtained with the new fitting method are shown . the horizontal grey line shows the input mode frequency . the dashed grey lines show the theoretical value of frequency uncertainty , @xmath240 , given by equation ( [ equ : freq_error_determin ] ) . the vertical axis of the plot spans the interval @xmath272hz.,scaledwidth=90.0% ] in this paper we derived an expression for the joint pdf of solar or stellar oscillations in complex fourier space , in agreement with the work of . this joint pdf explicitly takes into account frequency correlations introduced by the convolution with the window function . we implemented a maximum likelihood estimation method to retrieve the parameters of stellar oscillations . both stochastic solar - like oscillations and deterministic sinusoidal oscillations were considered . in the case of solar - like oscillations , we performed monte - carlo simulations to show that the improvement provided by our fitting method can be very significant in comparison with a fitting method that ignores the frequency correlations . the results are summarized in figure 6 . in one particular example , using an observation window with a duty cycle @xmath273% and a signal - to - noise ratio @xmath181 , the new fitting method increased the precision of the mode frequency by a factor of two and the estimates of the linewidth and mode power were less biased and more precise . for a window with a duty cycle @xmath274% , the precision on the mode frequency estimate was increased by a factor of five . we also found that the cramr - rao lower bounds ( formal errors ) can provide reasonable estimates of the uncertainty on the mle estimates of the oscillation parameters . in the case of long - lived , purely sinusoidal oscillations , we did not find any significant improvement in using this new fitting method . yet , we confirm that the standard deviation of the mode frequency can be measured in fourier space with a precision much better than @xmath1 for large signal - to - noise ratios , in accordance with a previous time - domain calculation ( cuypers , 1987 ; cuypers , 2008 , private communication ) . the analysis of time series containing many gaps can benefit from our work . applications may include , for example , the re - analysis of solar oscillations from the early days of the bison network @xcite or the solar - like oscillations of @xmath229 centauri observed from the ground with two telescopes @xcite . we thank t. appourchaux for useful discussions , in particular for the suggestion to compute the cramr - rao lower bounds . t. stahn is a member of the international max planck research school on physical processes in the solar system and beyond at the universities of gttingen and braunschweig . the mle source code is available from the internet platform of the european helio- and asteroseismology network ( helas , funded by the european union ) at http://www.mps.mpg.de / projects / seismo / mle_softwarepackage/. miller , b.a . , hale , s.j . , elsworth , y. , chaplin , w.j . , isaak , g.r . , new , r. : 2004 . in : danesy , d. ( ed . ) , _ proc . soho 14/gong 2004 workshop , helio- and asteroseismology : towards a golden future _ , esa * sp-559 * , esa pub . div . , noordwijk , 571 . winget , d.e . , nather , r.e . , clemens , j.c . , provencal , j. , kleinman , s.j . , bradley , p.a . , wood , m.a . , claver , c.f . , frueh , m.l . , grauer , a.d . , hine , b.p . , hansen , c.j . , fontaine , g. , achilleos , n. , wickramasinghe , d.t . , marar , t.m.k . , seetha , s. , ashoka , b.n . , odonoghue , d. , warner , b. , kurtz , d.w . , buckley , d.a . , brickhill , j. , vauclair , g. , dolez , n. , chevreton , m. , barstow , m.a . , solheim , j.e . , kanaan , a. , kepler , s.o . , henry , g.w . , kawaler , s.d . : 1991 , _ astrophys . j. _ * 378 * , 326 .
quantitative helio- and asteroseismology require very precise measurements of the frequencies , amplitudes , and lifetimes of the global modes of stellar oscillation . it is common knowledge that the precision of these measurements depends on the total length ( @xmath0 ) , quality , and completeness of the observations . except in a few simple cases , the effect of gaps in the data on measurement precision is poorly understood , in particular in fourier space where the convolution of the observable with the observation window introduces correlations between different frequencies . here we describe and implement a rather general method to retrieve maximum likelihood estimates of the oscillation parameters , taking into account the proper statistics of the observations . our fitting method applies in complex fourier space and exploits the phase information . we consider both solar - like stochastic oscillations and long - lived harmonic oscillations , plus random noise . using numerical simulations , we demonstrate the existence of cases for which our improved fitting method is less biased and has a greater precision than when the frequency correlations are ignored . this is especially true of low signal - to - noise solar - like oscillations . for example , we discuss a case where the precision on the mode frequency estimate is increased by a factor of five , for a duty cycle of 15% . in the case of long - lived sinusoidal oscillations , a proper treatment of the frequency correlations does not provide any significant improvement ; nevertheless we confirm that the mode frequency can be measured from gapped data at a much better precision than the @xmath1 rayleigh resolution .
in this paper , we consider array - based spatially - coupled low - density parity - check ( sc - ldpc ) codes as introduced in @xcite and subsequently studied in @xcite from the perspective of _ absorbing sets_. array ldpc codes were originally proposed by fan in @xcite and are specified by two integers @xmath0 and @xmath1 , where @xmath0 is an odd prime and @xmath2 . spatial coupling was first introduced in the coding theory literature by felstrm and zigangirov in @xcite , where they proposed convolutional ldpc or sc - ldpc codes . these codes have very good belief propagation ( bp ) decoding thresholds due to the phenomenon of _ threshold saturation_. for instance , in @xcite , it was shown that for binary memoryless channels , the bp decoding threshold _ saturates _ to the maximum _ a posteriori _ decoding threshold of the underlying ensemble . since the original work by fan @xcite , several authors have considered the _ structural _ properties ( including the minimum distance @xmath8 ) of array ldpc codes ( see , e.g. , @xcite ) . for high rate and moderate length , these codes perform well under iterative decoding , and they are also well - suited for practical implementation due to their regular structure @xcite . in this work , we present several tight upper bounds on the _ optimal _ minimum distance of array - based sc - ldpc codes for coupling length at least two and @xmath4 , that are independent of @xmath0 and that are valid for all values of @xmath5 where @xmath6 depends on @xmath1 . also , these bounds ( and intermediate results in their proofs ) can in some cases be used to quickly remove large parts of the search space when searching for _ optimal _ _ cutting vectors_. for small values of @xmath0 ( @xmath4 ) we present the results of an exhaustive search over all cutting vectors , showing that a careful selection can increase the minimum distance significantly , especially for @xmath9 . we remark that the iterative decoding performance of these codes , for instance , on additive white gaussian noise channels , is typically dominated by minimum absorbing sets and their multiplicities . however , improved decoding methods and ultimately maximum - likelihood decoding will not be trapped in absorbing sets and thus overcome the shortcomings of standard iterative decoding , in which case the minimum distance ( and its multiplicity ) will become an important performance metric . the array ldpc code @xmath10 , with parameters @xmath0 and @xmath1 , has length @xmath11 and can be defined by the parity - check matrix @xmath12\ ] ] where @xmath13 is the @xmath14 identity matrix and @xmath15 the @xmath14 permutation matrix @xmath16.\ ] ] since the number of ones in each row of the matrix in ( [ eq : pcmatrix ] ) is @xmath0 and the number of ones in each column is @xmath1 , the array ldpc codes are @xmath17-regular codes . furthermore , its rank is @xmath18 , from which it follows that the dimension of @xmath10 is @xmath19 . in @xcite , a new representation for @xmath20 was introduced . in particular , since each column of the parity - check matrix @xmath20 has @xmath1 blocks and each block is a permutation of @xmath21 , where @xmath22 denotes the transpose of its argument , we can represent each column as a vector of integers between @xmath23 and @xmath24 , where @xmath25 i.e. , the @xmath26-positions are associated with the integers modulo @xmath0 . furthermore , it follows from ( [ eq : pcmatrix ] ) and the integer representation in ( [ eq : rep ] ) that any column in an array ldpc code parity - check matrix is of the form @xmath27 where @xmath28 ^ 2 $ ] and , for notational convenience , for any positive integer @xmath29 , @xmath30 \triangleq \{0,1,\dots , l-1\}$ ] . thus , a column can be specified by two integers @xmath31 and @xmath32 . also , note that since there are @xmath11 distinct columns in an array ldpc code parity - check matrix , any pair @xmath28 ^ 2 $ ] specifies a valid column . in the following , it is also convenient to consider the parity - check matrix in ( [ eq : pcmatrix ] ) as an @xmath33 array of @xmath14 permutation matrices with _ row group _ indices @xmath34 $ ] and _ column group _ indices @xmath35 $ ] , from which it follows that each column can be uniquely indexed by the pair @xmath36 , where @xmath35 $ ] is the index of the column group and @xmath37 $ ] is the index within a column group . likewise , each row can be uniquely indexed by the pair @xmath38 , where @xmath34 $ ] is the index of the row group and @xmath37 $ ] is the index within a row group . array - based sc - ldpc codes can be constructed from array ldpc codes by a special type of _ edge spreading _ or _ unwrapping _ procedure @xcite specified by a `` cutting '' vector @xmath39 , where @xmath40 . the purpose of the cutting vector is to generate two parity - check matrices @xmath41 and @xmath42 , each of size @xmath43 and initially filled with zeros , as follows : * for each @xmath44 , @xmath34 $ ] , the @xmath14 permutation matrices in row group @xmath45 and column group @xmath46 , @xmath47 , of @xmath20 are copied into the corresponding positions in @xmath41 . * similarly , for each @xmath44 , @xmath34 $ ] , the @xmath14 permutation matrices in row group @xmath45 and column group @xmath46 , @xmath48 , of @xmath20 are copied into the corresponding positions in @xmath42 . if @xmath49 , @xmath50 , and @xmath51 , the matrices @xmath41 and @xmath42 become @xmath52\ ] ] and @xmath53.\ ] ] for a given positive integer @xmath29 ( the coupling length ) and a cutting vector @xmath54 , an array - based sc - ldpc code is defined by the parity - check matrix @xmath55\ ] ] of size @xmath56 . as for the uncoupled case , it is convenient to consider the convolutional parity - check matrix in ( [ eq : scpcmatrix ] ) as a @xmath57 array of @xmath14 permutation and all - zero matrices in which @xmath58 consecutive rows ( resp . @xmath11 consecutive columns ) are referred to as a row ( resp . section_. within each row ( resp . column ) section , the rows ( resp . columns ) correspond to a row ( resp . column ) group . in summary , each column can be indexed by a triple @xmath59 , where @xmath60 $ ] is the column section index , @xmath35 $ ] is the column group index within column section @xmath61 , and @xmath37 $ ] is the column index within column group @xmath46 in column section @xmath61 . similarly , each row can be indexed by a triple @xmath62 , where @xmath63 $ ] is the row section index , @xmath34 $ ] is the row group index within row section @xmath61 , and @xmath37 $ ] is the row index within row group @xmath45 in row section @xmath61 . note that as for the uncoupled matrix in ( [ eq : pcmatrix ] ) , all columns in @xmath64 contain exactly @xmath1 ones and each row in row sections @xmath65 contains exactly @xmath0 ones . however , the rows in row sections @xmath23 and @xmath29 can have fewer than @xmath0 ones . in the following , we will denote the binary linear code defined by the coupled parity - check matrix @xmath64 in ( [ eq : scpcmatrix ] ) by @xmath66 and its corresponding minimum ( resp . stopping ) distance by @xmath67 ( resp . @xmath68 ) . it was shown in ( * ? ? ? * theorem 5 ) that the minimum distance @xmath67 of @xmath66 is bounded below by the minimum distance @xmath69 of @xmath10 for all cutting vectors @xmath54 . now , let us define the optimal minimum distance of an array - based sc - ldpc code as the highest minimum distance over all possible cutting vectors as the coupling length @xmath29 tends to infinity , i.e. , @xmath70 where @xmath71 in this section , we derive upper bounds on @xmath72 for @xmath73 and @xmath4 , that are independent of @xmath0 and that hold for all values of @xmath5 where @xmath6 depends on @xmath1 . now , define the _ range _ of a sorted ( in nondecreasing order ) sequence of column group indices @xmath74 of length @xmath75 as @xmath76 } \left ( j_{l}^{\rm gr}-\left(j_{(l+1 ) \bmod n}^{\rm gr } - q \left \lceil \frac{n-1-l}{n-1 } \right \rceil \right ) \right ) + 1.\ ] ] furthermore , define @xmath77 } \left ( \left(\zeta_{(j+1 ) \bmod m}-\zeta_{j}\right ) \bmod q \right ) \notag \\ j_{\rm max}(\boldsymbol{\zeta } ) & \triangleq \operatorname*{arg\,max}_{j \in [ m-1 ] } \left ( \left(\zeta_{(j+1 ) \bmod m}-\zeta_{j}\right ) \bmod q \right ) \notag\end{aligned}\ ] ] for a given cutting vector @xmath54 . let @xmath78 , @xmath79 , and @xmath80 . then , the range of @xmath81 is @xmath82 [ th : m3 ] the minimum distance @xmath83 , for @xmath73 and @xmath84 , is upper - bounded by @xmath85 . the proof is based on the template codeword from ( * ? ? ? * theorem 4 ) . for convenience of the reader , we restate the corresponding _ template support matrix _ below @xmath86\ ] ] where @xmath87 , @xmath88 \setminus \{0\}$ ] , and @xmath37 $ ] with @xmath89 , and where all operations are taken modulo @xmath0 . the template support matrix is obtained by extracting the columns of the parity - check matrix corresponding to the support set of the underlying codeword . when constructing this matrix , the integer representation of the columns from ( [ eq : form ] ) is used . now , a column of the general form @xmath90 , @xmath91 $ ] , in this matrix has column group index of @xmath32 . thus , the sequence of column group indices corresponding to the template matrix in ( [ eq : templatem3 ] ) is @xmath92 for @xmath93 and @xmath94 , we get the following sorted sequence @xmath95 since array ldpc codes are quasi - cyclic with period @xmath0 , _ cyclically incrementing _ a sequence of column group indices @xmath96 of length @xmath75 of a codeword in an array ldpc code by an integer @xmath97 results in the valid sequence @xmath98 of column group indices of a codeword that is obtained by cyclically shifting the given codeword by @xmath99 positions to the right . if the range of a sorted ( in nondecreasing order ) sequence of column group indices corresponding to a codeword in the uncoupled array ldpc code is at most @xmath100 , it can always be cyclically incremented so that the corresponding codeword has coordinates with column group index @xmath46 within @xmath101 when @xmath102 , and within @xmath103 when @xmath104 . consequently , there will exist a codeword in the coupled code ( for all @xmath73 ) with all coordinates within the same column section when @xmath102 , or within two consecutive column sections when @xmath104 . thus , since @xmath105 , there will exist a codeword in the sc - ldpc code of weight @xmath85 for all cutting vectors @xmath54 as long as @xmath106 since @xmath107 for @xmath108 and @xmath109 for @xmath49 , the smallest @xmath0 that satisfies this inequality is @xmath110 , and the result follows since the range of ( [ eq : groupindexseqm3 ] ) for @xmath84 is constant . consider the case of @xmath79 , @xmath50 , and the template support matrix in ( [ eq : templatem3 ] ) for @xmath93 and @xmath94 . the sorted ( in nondecreasing order ) sequence of column group indices is @xmath111 which has range @xmath112 . by cyclically incrementing this sequence by two , we get @xmath113 . the corresponding support matrices ( both of which correspond to codewords ) are @xmath114 \text { and } \left [ \begin{smallmatrix } 0 & 0 & 2 & 2 & 5 & 5 \\ 2 & 0 & 2 & 1 & 1 & 0 \\ 4 & 0 & 2 & 0 & 4 & 2 \end{smallmatrix } \right]\ ] ] respectively . for the cutting vector @xmath115 , we have @xmath116 , @xmath117 , @xmath118 , and @xmath119 = \left [ \begin{mysmallmatrix } \bm{i } & \bm{i } & \bm{i } & \bm{i } & \bm{0 } & \bm{0 } & \cellcolor{lightcyan}{\bm{0 } } & \vline & & & & & & & \\ \bm{i } & \bm{p } & \bm{p}^2 & \bm{p}^3 & \bm{p}^4 & \bm{0 } & \bm{0 } & \vline & & & & & & & \\ \bm{i } & \bm{p}^2 & \bm{p}^4 & \bm{p}^6 & \bm{p } & \bm{p}^3 & \bm{0 } & \vline & & & & & & & \\ \\[-0.5ex]\hline \\[+0.5ex ] \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{i } & \bm{i } & \color{blue}{\bm{i } } & \vline & \color{blue}{\bm{i } } & { \bm{i } } & \color{blue}{\bm{i } } & \color{blue}{\bm{i } } & \bm{0 } & \bm{0 } & \bm{0 } \\ \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{p}^5 & \color{blue}{\bm{p}^6 } & \vline & \color{blue}{\bm{i } } & { \bm{p } } & \color{blue}{\bm{p}^2 } & \color{blue}{\bm{p}^3 } & \bm{p}^4 & \bm{0 } & \bm{0 } \\ \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \color{blue}{\bm{p}^5 } & \vline & \color{blue}{\bm{i } } & { \bm{p}^2 } & \color{blue}{\bm{p}^4 } & \color{blue}{\bm{p}^6 } & \bm{p } & \bm{p}^3 & \bm{0 } \\ \\[-0.5ex]\hline \\[+0.5ex ] & & & & & & & \vline & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{i } & \bm{i } & \bm{i } \\ & & & & & & & \vline & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{p}^5 & \bm{p}^6 \\ & & & & & & & \vline & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{0 } & \bm{p}^5 \end{mysmallmatrix } \right].\ ] ] from the highlighted blue columns in the matrix above we can identify a weight-@xmath85 codeword corresponding to the sequence @xmath113 of column group indices mentioned above . for other cutting vectors like @xmath120 , we get @xmath121 , which is less than the range , and there will be no codewords of the type in ( [ eq : templatem3 ] ) with @xmath93 and @xmath94 ( there are for @xmath122 and @xmath94 ) in the coupled code . by combining lemma [ th : m3 ] with ( * ? ? ? * corollary 6 ) , we get the following theorem . [ th : m3equal ] for @xmath73 and @xmath84 , the minimum distance @xmath123 . [ th : m4 ] the minimum distance @xmath125 , for @xmath73 and @xmath126 , is upper - bounded by @xmath127 . the proof follows the same main idea of the proof of lemma [ th : m3 ] above using the template support matrix ( @xmath128 ) @xmath129\ ] ] which we have found using the algorithm from @xcite . the remaining technical details are omitted for brevity . note that the template support matrix for @xmath124 given in fig . 3 in @xcite will fail to prove the above result , since the sorted ( in nondecreasing order ) sequence of column group indices has range @xmath130 for @xmath131 . as a consequence , the template matrix in fig . 3 in @xcite can only prove an upper bound of @xmath127 for @xmath132 . there is a codeword of weight @xmath127 in @xmath133 , @xmath84 , for all positive integers @xmath73 and cutting vectors @xmath54 if one of the following conditions are true . 1 . @xmath134 and @xmath135 . 2 . @xmath136 and @xmath137 . 3 . @xmath138 and @xmath139 . 4 . @xmath140 and @xmath141 . 5 . @xmath142 . the proof is omitted due to lack of space . [ lem:1 ] the minimum distances @xmath143 and @xmath144 , for @xmath73 , are both upper - bounded by @xmath127 . for the case @xmath145 , the number of cutting vectors that satisfy none of the five conditions in proposition [ prop:1 ] is only @xmath146 , while the total number of possible cutting vectors is @xmath147 . thus , by running the algorithm from @xcite , adapted to the case of sc codes , on these @xmath146 cases we have verified that there are indeed codewords of weight @xmath127 for all cutting vectors , and the result follows . for @xmath148 , there are no cutting vectors that satisfy none of the five conditions in proposition [ prop:1 ] , from which it follows that @xmath149 for all positive integers @xmath73 . by combining lemmas [ th : m4 ] and [ lem:1 ] , ( * ? ? ? * theorem 5 ) , and ( * ? ? ? * corollary 4.2 ) , we get the following theorem . [ th : m4real ] for @xmath73 and @xmath150 , the minimum distance @xmath151 . [ th : m5 ] the minimum distance @xmath152 , for @xmath73 and @xmath153 , is upper - bounded by @xmath154 . the proof follows the same main idea of the proofs of lemmas [ th : m3 ] and [ th : m4 ] above using the template support matrix from fig . 4 in @xcite , and is omitted for brevity . note that there appears to be no equivalent to proposition [ prop:1 ] ( except for the last condition ) for @xmath7 , since the template support matrix from fig . 4 in @xcite ( which is used for the proof of theorem [ th : m5 ] ) does not have the required structure . also , the algorithm from @xcite was not able to identify other nonequivalent template support matrices with a suitable structure . this has also been `` confirmed '' by the fact that for @xmath155 we have been able to identify a cutting vector that gives a minimum distance of @xmath156 ( see table [ table : scarrayldpc ] ) . [ th : m5_1 ] the minimum distance @xmath152 , for @xmath73 and @xmath157 , is upper - bounded by @xmath158 . the proof follows the same main idea of the proofs of lemmas [ th : m3 ] and [ th : m4 ] and theorem [ th : m5 ] above using a template support matrix which was found using the algorithm from @xcite . due to lack of space , the actual template support matrix and the technical details of the proof are omitted . in this section , we present some numerical results for the cases @xmath50 , @xmath109 , and @xmath112 . by performing an exhaustive search over all cutting vectors @xmath54 the optimal minimum distance @xmath159 was determined for @xmath160 ( see table [ table : scarrayldpc ] ) . for each of these values of @xmath0 , the @xmath8-optimal cutting vector is not unique , i.e. , there are several cutting vectors that give the best minimum distance . for instance , for @xmath49 , we found ( for @xmath161 ) the @xmath162-optimal ( where @xmath162 denotes stopping distance ) cutting vectors displayed in table [ table : q5m3cutting ] . for each cutting vector , in the second and fourth columns we have tabulated the corresponding minimum and stopping distance , respectively . the corresponding multiplicities are tabulated in the third ( minimum distance ) and fifth ( stopping distance ) columns . note that the _ optimal _ cutting vector @xmath163 for @xmath164 and @xmath165 absorbing sets from @xcite is also optimal when it comes to minimum / stopping distance , since it gives the optimal minimum / stopping distance of @xmath127 . however , it gives slightly more stopping sets of size @xmath127 ( see table [ table : q5m3cutting ] ) . on the other hand , for @xmath79 , the _ optimal _ cutting vector @xmath166 for @xmath164 and @xmath165 absorbing sets from @xcite is _ not _ optimal when it comes to minimum distance , since it gives a minimum distance of only @xmath85 ( the optimal minimum distance is @xmath167 ) . for @xmath168 , @xmath8-optimal cutting vectors that also give the lowest possible minimum distance multiplicity , denoted as @xmath169-optimal cutting vectors , are displayed within the parentheses in the sixth column of table [ table : scarrayldpc ] ( first and second row , respectively ) . for @xmath84 , it follows from theorem [ th : m3equal ] that the optimal minimum distance is @xmath85 . also , all cutting vectors give a minimum distance of @xmath85 , but possibly with different multiplicities . -3.0ex ccccc + [ -2.0ex ] @xmath54 & @xmath170 & mult . & @xmath171 & mult . + + [ -2.0ex ] + [ -2.0ex ] @xmath172 & 10 & 20 & 10 & 65 + @xmath173 & 10 & 20 & 10 & 65 + @xmath174 & 10 & 19 & 10 & 59 + @xmath175 & 10 & 19 & 10 & 59 + @xmath176 & 10 & 20 & 10 & 65 + @xmath177 & 10 & 20 & 10 & 65 by performing an exhaustive search over all cutting vectors @xmath54 ( with the help of proposition [ prop:1 ] to reduce the search space when @xmath178 ) the optimal minimum distance @xmath179 was determined for @xmath180 . for each of these values of @xmath0 , the @xmath8-optimal cutting vector is not unique , i.e. , there are several cutting vectors that give the best minimum distance . for instance , for @xmath79 , we found ( for @xmath161 ) the @xmath162-optimal cutting vectors displayed in table [ table : q7m4cutting ] . for each cutting vector , in the second and fourth columns we have tabulated the corresponding minimum and stopping distance , respectively . the corresponding multiplicities are tabulated in the third ( minimum distance ) and fifth ( stopping distance ) columns . for @xmath180 , the optimal minimum distance and corresponding @xmath169-optimal cutting vectors ( displayed within the parentheses ) are given in the fourth column of table [ table : scarrayldpc ] ( the first eight rows , respectively ) . for @xmath150 , it follows from theorem [ th : m4real ] that @xmath181 . -3.0ex ccccc + [ -2.0ex ] @xmath54 & @xmath182 & mult . & @xmath183 & mult . + + [ -2.0ex ] + [ -2.0ex ] @xmath184 & 14 & 30 & 14 & 401 + @xmath185 & 14 & 30 & 14 & 695 + @xmath186 & 14 & 29 & 14 & 393 + @xmath187 & 14 & 30 & 14 & 695 + @xmath188 & 14 & 30 & 14 & 401 -3.0ex clclclc + [ -2.0ex ] @xmath0 & @xmath189 & @xmath190 & @xmath179 & @xmath191 & @xmath159 & @xmath192 + + [ -2.0ex ] + [ -2.0ex ] 5 & & & @xmath193 @xmath194 & @xmath167 & @xmath193 @xmath195 & @xmath85 + 7 & @xmath196 @xmath197 & 12 & @xmath196 @xmath198 & @xmath167 & @xmath199 @xmath200 & @xmath85 + 11 & @xmath201 @xmath202 & 10 & @xmath196 @xmath203 & @xmath127 & @xmath204 & @xmath85 + 13 & @xmath205 @xmath206 & 12 & @xmath207 @xmath208 & @xmath127 & @xmath204 & @xmath85 + 17 & @xmath209 , @xmath201 , or @xmath205 @xmath210 & 12 & @xmath207 @xmath211 & @xmath127 & @xmath204 & @xmath85 + 19 & @xmath209 or @xmath201 @xmath212 & 12 & @xmath207 @xmath213 & @xmath127 & @xmath204 & @xmath85 + 23 & @xmath209 @xmath214 & 12 & @xmath207 @xmath215 & @xmath127 & @xmath204 & @xmath85 + 29 & @xmath216 @xmath217 & 12 & @xmath207 @xmath218 & @xmath127 & @xmath204 & @xmath85 + 31 & @xmath216 @xmath219 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + 37 & @xmath216 @xmath220 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + 41 & @xmath216 @xmath221 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + 43 & @xmath216 @xmath222 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + 47 & @xmath216 @xmath223 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + 53 & @xmath196 @xmath224 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + @xmath225 & @xmath207 & 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 + @xmath226 & @xmath193 or @xmath207 & 10 or 12 & @xmath193 & @xmath127 & @xmath204 & @xmath85 for @xmath79 , we have performed an exhaustive search over all possible cutting vectors . the optimal minimum distance @xmath227 is @xmath156 and a @xmath169-optimal cutting vector is displayed within the parentheses in the second column of table [ table : scarrayldpc ] ( second row ) . for @xmath228 and @xmath229 , the optimal minimum distance is as high as @xmath230 and @xmath231 , respectively , and @xmath8-optimal cutting vectors are shown within the parentheses in the second column of table [ table : scarrayldpc ] ( third and fourth row , respectively ) . for @xmath232 and @xmath130 , upper and lower bounds on the optimal minimum distance , tabulated in the second column of table [ table : scarrayldpc ] , have been established from an exhaustive search over all cutting vectors . the displayed values are bounds since we were not able to exhaustively enumerate all codewords of weight at most @xmath233 , when @xmath234 , for a given cutting vector , for these two values of @xmath0 . within the parentheses a corresponding cutting vector ( which establishes the upper bound ) is also tabulated . the lower bounds were determined by running the algorithm from @xcite , adapted to the case of sc codes , for the specific cutting vectors displayed within the parentheses . for @xmath235 , the tabulated values ( in the second column of table [ table : scarrayldpc ] ) are again the exact values of @xmath189 and also ( within the parentheses ) @xmath8-optimal ( @xmath169-optimal for @xmath155 ) cutting vectors are displayed . for @xmath236 , @xmath237 , which follows from theorem [ th : m5 ] , ( * ? ? ? * table i ) , and theorem 5 in @xcite . for @xmath238 , it follows from theorem [ th : m5 ] , corollary 4.4 in @xcite , and theorem 5 in @xcite that @xmath189 is either @xmath127 or @xmath154 , although we conjecture it to be @xmath154 . in this work , we have studied in detail the minimum distance of array - based sc - ldpc codes . several tight upper bounds on the optimal minimum distance for coupling length @xmath73 and @xmath4 , that are independent of @xmath0 and that are valid for all values of @xmath239 where @xmath6 depends on @xmath1 , have been presented . furthermore , we have conducted an exhaustive search over all cutting vectors for small values of @xmath0 ( @xmath4 ) which shows that by carefully selecting the cutting vector , the minimum distance ( when @xmath0 is not very large ) can be significantly increased , especially for @xmath7 . an interesting topic for future work is to consider the correlation with absorbing sets . in particular , will a @xmath169-optimal cutting vector also be _ close - to - optimal _ when it comes to problematic absorbing sets , and/or vice - versa ? d. g. m. mitchell , l. dolecek , and d. j. costello , jr . , `` absorbing set characterization of array - based spatially coupled ldpc codes , '' in _ proc . ieee int . inf . theory ( isit ) _ , honolulu , hi , jun./jul . 2014 , pp . 886890 . s. kudekar , t. richardson , and r. l. urbanke , `` spatially coupled ensembles universally achieve capacity under belief propagation , '' _ ieee trans . inf . theory _ , vol . 59 , no . 12 , pp . 77617813 , dec . 2013 . l. dolecek , z. zhang , v. anantharam , m. j. wainwright , and b. nikolic , `` analysis of absorbing sets and fully absorbing sets of array - based ldpc codes , '' _ ieee trans . theory _ , vol . 56 , no . 1 , 181201 , jan . 2010 . p. bhagawat , m. uppal , and g. choi , `` fpga based implementation of decoder for array low - density parity - check codes , '' in _ proc . acoustics , speech , and signal processing ( icassp ) _ , vol . 5 , philadelphia , pa , mar . 2005 , pp . a. e. pusane , r. smarandache , p. o. vontobel , and d. j. costello , jr . , `` deriving good ldpc convolutional codes from ldpc block codes , '' _ ieee trans . inf . theory _ , vol . 57 , no . 2 , 835857 , feb . e. rosnes , . ytrehus , m. a. ambroze , and m. tomlinson , `` addendum to `` an efficient algorithm to find all small - size stopping sets of low - density parity - check matrices '' , '' _ ieee trans . theory _ , vol . 58 , no . 1 , 164171 , jan . 2012
an array low - density parity - check ( ldpc ) code is a quasi - cyclic ldpc code specified by two integers @xmath0 and @xmath1 , where @xmath0 is an odd prime and @xmath2 . the exact minimum distance , for small @xmath0 and @xmath1 , has been calculated , and tight upper bounds on it for @xmath3 have been derived . in this work , we study the minimum distance of the spatially - coupled version of these codes . in particular , several tight upper bounds on the _ optimal _ minimum distance for coupling length at least two and @xmath4 , that are independent of @xmath0 and that are valid for all values of @xmath5 where @xmath6 depends on @xmath1 , are presented . furthermore , we show by exhaustive search that by carefully selecting the _ edge spreading _ or _ unwrapping _ procedure , the minimum distance ( when @xmath0 is not very large ) can be significantly increased , especially for @xmath7 .
it is well - known that classification of objects in astronomical survey data provides a fundamental characterization of the dataset , and forms a vital step in understanding the ensemble properties of the dataset and the properties of individual objects . in this paper , we apply automated machine learning algorithms to classify the 143 million non - repeat objects in the third data release ( dr3 ; * ? ? ? * ) of the sloan digital sky survey ( sdss ; york et al . the sdss is used because the combination of the quality and accuracy of the survey ccd photometry , the approximately concurrent spectroscopy performed with the same telescope , and the large number of objects , is ideally suited to the methods we employ in this work . as our primary goal is to perform reliable star - galaxy separation , we have used three classes : _ galaxy _ , _ star _ , and _ nsng _ ( neither star nor galaxy ) . the use of this third class is novel , and is included to not only cleanly separate out those objects that clearly are neither stars nor galaxies ( e.g. , quasars ) , but to also improve the extrapolation of our classifications well beyond the nominal magnitude limits of our current training sample . the latter benefit arises since star - galaxy separation becomes less accurate at fainter magnitudes due to source confusion . in this case , our algorithm can assign sources to the third category when a clear star - galaxy classification is not feasible , thereby minimizing contamination for lower signal - to - noise sources . in this paper , the learning algorithm we employ is the axis - parallel decision tree . a decision tree is trained on the subset of data for which spectra are available and is subsequently used to classify the much larger sample for which there is just photometry . this is the first public release of objects classified using this method for the entire sdss dr3 . a natural question is `` why provide classification for all objects , most of which are near the photometric limits of the survey , where classifications are less robust ? '' we have three reasons for classifying the entire sdss dr3 . first , given the speed of applying our classifications , we have found that it is considerably easier to handle large datasets when no artificial cuts are used . with the growth in the number of large astronomical surveys that are accessible from a virtual observatory , we feel this point is becoming increasingly more important , as it greatly simplifies any data federation issues a fellow scientist may need to perform if they wish to use our classifications . second , classifying all objects allows others with fainter training data to verify the efficacy of our classifications . finally , since our classifications are probabilistic in nature , by providing the classifications for all objects , we allow others to make their own probability cuts , perhaps using combinations of classes , to perform their own science . for example , our approach simplifies the task of performing follow - up spectroscopy , as a user of our fully classified catalog has an easier task in characterizing any selection effects . decision tree learning algorithms have been in existence since the 1970s , see , e.g. , @xcite or @xcite . they are able to classify objects into discrete or continuous categories , scale well to large datasets , are fairly robust to inequitably - sampled training sets , and can cope with values from individual objects that happen to be bad or irrelevant for the training targets being considered . axis - parallel decision trees , particularly smaller ones , are also easily understood , providing clear criteria for the splitting of the data set into constituent classes or values . this is in contrast to , for example , artificial neural networks ( e.g. , * ? ? ? * ) , where the resulting weights are basically a _ black box _ , or oblique decision trees ( e.g. , * ? ? ? * ) , where linear combinations of parameters , or hyper - planes , are used to subdivide the data . given the size of our dataset , however , our trees are generally rather large , making it difficult to fully interpret the classification rules , but one can in principle follow the splitting of the population . while oblique trees may be less transparant in their operation , they do have the benefit of generally producing smaller trees . most astronomical results using neural networks and decision trees show that the two methods give comparable results an example is @xcite , described below . a number of previous efforts have used decision trees to classify objects in astronomical surveys . @xcite use decision trees to separate stars and cosmic ray hits in hubble space telescope images and are able to do so with an accuracy of over 95% . their algorithm , known as oc1 , is an oblique decision tree . @xcite use a version of oc1 to classify galaxy morphologies in the european southern observatory uppsala surface photometry catalog @xcite and show their results to be comparable to those of @xcite , who use artificial neural networks on the same data . @xcite construct a radio - selected sample of quasars using the vla - first survey @xcite and the optical apm - poss - i quasar survey of @xcite . they show that using decision trees improves on simpler methods and enables quasars to be selected with 85% efficiency . they also show the completeness and efficiency of the selection as a function of threshold probability from the decision tree that the object is a quasar . @xcite compare naive bayes , decision tree and neural network classifiers for morphological galaxy classification and show that the latter two are comparable . @xcite take advantage of the much improved data , both in quality and quantity of objects , available from the sdss . they apply the oblique decision tree classifier classx , based on oc1 , to the sdss dr2 @xcite to classify objects into star , red star ( type m or later ) , agn photometric redshift and galaxy photometric redshift bins . three representative samples , each of @xmath3 photometric objects , one of which extends 2 magnitudes fainter than the spectroscopy , are also investigated . they give a table of percentage correct classifications from which one can calculate completeness and efficiency . many studies , too numerous to list here , use other machine learning techniques to classify astronomical survey objects , including star - galaxy separation and quasar identification . these previous works , however , do not provide spectroscopically - trained classifications with probabilities on the scale of the 143 million sources given here for the sdss dr3 . as part of our work , we make publicly available the full set of our 143 million object classifications for the sdss dr3 and probability cuts to select galaxies and stars for a range of completenesses and efficiencies . the rest of the paper is as follows . in [ sec : data ] we describe the datasets used in the construction and testing of the catalog . [ sec : algorithms ] describes the decision trees , their optimization using supercomputing resources , and the probability cuts to create subsamples of particular objects . [ sec : results ] then describes the optimal learning parameters found and the resulting datasets . in [ sec : discussion ] we discuss the efficacy of our star - galaxy separation , the extrapolation of our results to fainter magnitude , and new projects we have undertaken to extend this work . we conclude in [ sec : conclusions ] . the sdss is a project to map @xmath4 steradians of the northern galactic cap in five bands ( @xmath5 , @xmath6 , @xmath7 , @xmath8 and @xmath9 ) from 35008900 . the final survey will provide photometry for approximately @xmath10 objects , of which a multifiber spectrograph will provide spectra and redshifts for around one million . the spectroscopic targeting pipeline selects targets for spectroscopy from the imaging @xcite , and a tiling algorithm @xcite optimally assigns spectroscopic fibers to the selected targets . further details of the survey are given in @xcite and in @xcite for dr3 . the datasets we use in this work consist of training , testing , blind testing , and application sets . the training and testing sets are drawn from objects in the ` specobj ` view of the dr3 catalog archive server ( cas ) public database . the blind testing sets are from the sdss dr3 , 2df galaxy redshift survey ( 2dfgrs ; * ? ? ? * ) final data release and the 2df qso redshift survey ( 2qz ; * ? ? ? * ) final data release . the application set is a locally hosted version of the ` photoprimary ` view of the dr3 cas . as with ` specobj ` , this is the view which contains the primary sdss observations with no duplicates . the raw ` specobj ` data consists of 477,081 objects and the ` photoprimary ` data 142,705,734 . the training , testing and application data consist of sdss photometric and spectroscopic attributes available in the ` photoprimary ` and ` specobj ` tables . we perform no additional image or spectral reduction . since we classify all sources in the sdss dr3 , this allows a user of our classifications to apply their own sample cuts based on sdss flags or attribute errors . the photometric attributes used for each object are the colors @xmath11 , @xmath12 , @xmath13 , and @xmath14 , corrected for galactic reddening via the maps of @xcite . the colors in each of the four magnitude types psf , fiber , petrosian and model are used . the use of the four magnitude types allows star - galaxy separation to be , to some extent , built - in . the light profiles of galaxies and stars are very different , which can be quantified by measuring the object flux in different - sized apertures . this fact is often used to perform star - galaxy separation , including within the sdss photometric pipeline where the difference between @xmath15 and @xmath16 is used ( see [ sec : algorithms ] for more information ) . the spectroscopic attribute is the target object type and is given by the ` specclass ` value in ` specobj ` assigned by the sdss spectroscopic pipeline . the ` specclass ` is discrete and takes the values ` unknown ` , ` star ` , ` galaxy ` , ` qso ` , ` qso_hiz ` or ` star_late ` . we construct our three training samples : galaxies , stars , and nsng , from these classes according to the mappings shown in table [ table : mappings ] . resulting numbers of each are 361,502 galaxies , 62,333 stars and 53,246 nsng , giving a total of 477,081 objects available for the training and testing process . given the spectroscopic data available in the sdss , the nsng training sample will predominantly consist of quasars . decision trees , however , effectively utilize all of their training data . thus , at fainter magnitudes , we can expect the nsng sample to contain sources that are not just quasars , but unknowns in conjunction with the ` unknown ` category in the ` specclass ` . on the other hand , decision trees are also robust to outlying values in their training data . for the classification training we simply apply the very broad cuts @xmath17 to remove clearly unphysical values from the data such as -9999 . these are applied to each of the colors and resulted in the removal of 12 galaxies and 1 quasar , leaving 477,068 for training and testing . for the decision tree optimization , sets of objects were randomly drawn from the remaining ` specobj ` set and used in training , with the rest for testing . thus the testing set , in a strict sense , is not independent of the training set . however , the large numbers of objects available in the training data means that the random subsamples drawn , except perhaps the smallest ones at a level of 10% or less , are large enough to still be truly representative of the training data and should thus be effectively independent . the performance results quoted in [ sec : results ] are , nevertheless , from a tree trained _ and _ tested on 80% of the training data and applied to the other 20% . the 80% is subdivided as above into non - overlapping training and testing sets , also in the ratio 80:20 . given the hypothesized easily - adequate size of the training set , this pseudo - blind test should be very similar to the test results during the optimization . for the final application to the ` photoprimary ` data , the tree was trained on all of the training data to maximize the available information . thus the results quoted in [ sec : results ] may be slightly pessimistic , but again they are likely to be very similar to the performance of the tree on ` photoprimary ` , within the spectroscopic regime . for the vast majority of objects in the ` photoprimary ` table , no spectroscopic information is available . in addition , we do not restrict the application dataset , as each object is classified independently . therefore , specific cuts , either on sdss flags , attribute errors , or the attributes themselves can be later applied by a user as appropriate to create a particular sample . for example , one might only want ` best ' photometry , as indicated by the object flags , for comparing object populations , or one might want to apply a color cut to enhance the utility of the provided classifications . the blind testing sets are drawn from the 2dfgrs and the 2qz . these test the performance of the decision tree on non - sdss data in a parameter space which extends beyond the sdss dr3 spectroscopic regime . the 2dfgrs was used to test the best tree s galaxy classifications and the 2qz to test its star and nsng classifications . the surveys were matched with the sdss using object position with a tolerance of 2 arcseconds and their spectroscopic classifications were taken as the target types . of the 2dfgrs galaxies , 50,191 were matches . for the 2qz the results were compared to the objects assigned the category _ 11 _ ( best classification and redshift ; see @xcite ) objects ( 8,739 quasars , 5,193 stars ) and those assigned _ 12 _ , _ 21 _ or _ 22 _ ( second - best ; 9,273 quasars ) . star - galaxy separation in the sdss is currently performed within the sdss photometric pipeline @xcite . an object is classified as extended , and hence a galaxy , if @xmath18 . here ` psfmag ` is the point - spread - function magnitude and ` cmodelmag ` is a combination of the best fitting de vaucouleurs and exponential profiles . the separation is done for each band individually , and for all five bands combined . ` psfmag ` is described further in @xcite and ` cmodelmag ` in @xcite . additional star - galaxy classifications have been performed on sdss data , e.g. , @xcite perform a bayesian star - galaxy separation via differences between the @xmath7 band ` psfmag ` and ` modelmag ` . in our approach , we use an axis - parallel decision tree to assign a probability that a source belongs to each of three classes . these probabilities always sum to one and reflect the relative degree of certainty about an object s type . we provide cuts in these probabilities that can be applied to generate catalogs to a required completeness and efficiency of either galaxies , stars , or neither galaxies nor stars ( i.e. , nsng ) . in general , the cuts can be a function of the probabilities , and these functions can have both minimum and maximum values . we define both completeness and efficiency as functions of the classification probability . for a given sample , therefore , the _ completeness _ is the fraction of all catalog sources of a given type found at a specific probability threshold . likewise , _ efficiency _ is the fraction of all catalog sources that are correctly classified to a specific probability threshold . in previous astronomical machine learning results , the efficiency of a sample has also been called the _ reliability _ or _ purity _ , and @xmath19__efficiency _ _ has been called the _ contamination_. the quantities are also known ( e.g. , * ? ? ? * ) as the _ recall _ and _ precision_. as well as a maximal combination of completeness and efficiency , one may also desire maximal values of these measures separately . an example of the former is to include interesting or unusual objects of a certain class , for example , candidate objects for gravitational lensing . an example of the latter is applications such as the 2-point correlation function of extragalactic objects , where one wants to minimize contamination by non - cosmological objects such as galactic stars ( e.g. , * ? ? ? our decision tree methods are applied within the framework of the data - to - knowledge toolkit @xcite , developed and maintained by the automated learning group at the national center for supercomputing applications ( ncsa ) . this is a java - based application which allows numerous modules , each of which performs a single data processing task , to be interconnected in a variety of ways . the framework allows the easy addition of further data to the training and testing process and the application of the trained decision trees to the entire photometric catalog . the trees are implemented on the xeon linux cluster _ tungsten _ at ncsa . this nationally allocated supercomputing system is composed of 1280 compute nodes . each node has available 2 gb of memory and a peak double - precision performance of 6.4 gflops . our use of the _ tungsten _ supercomputing system was facilitated by a large , peer - reviewed , national allocation submitted by the laboratory for cosmological data mining at ncsa , for which rjb is the principal investigator . execution tasks were submitted to _ tungsten _ via a shell script that created and tested one or more decision trees whose testable parameters were specified at run time . one or more decision trees would be created and tested on a single _ tungsten _ node , and when completed the test results for each decision tree was returned for subsequent analysis . as this process was embarrassingly parallel , we submitted multiple execution tasks simultaneously from this single shell script , with a maximum of 240 submitted at any one time . as discussed in the next section , we analyzed the results of over seven thousand decision trees , which spanned a range of different decision tree parameters in selecting our optimal parameter values . decision trees consist first of a root node in which the parameters describing the objects in the training set population are input , along with the classifications . the tree is usually considered upside - down , with the root node at the top . here there are three classes , so for each object we have @xmath20,[o_g , o_n , o_s])\ ] ] for the 16 features @xmath21 ( four colors in four magnitude types ) and the three output classes @xmath22 . the outputs are [ 1,0,0 ] , [ 0,1,0 ] and [ 0,0,1 ] for galaxy , nsng , and star , respectively . the outputs are in the form @xmath23,\ ] ] and the predicted probabilities of class membership always sum to one . a node population is split into population groups that are assigned to child nodes using the criterion that minimizes the classification error . this is a measure of how good the classification is when the tree that would result from the split is run on the test set . the process is repeated iteratively , resulting in a number of layers of nodes that form a tree . the iteration stops when either all nodes reach the minimum allowed population of objects in a node ( the minimum decomposition population ; mdp ) , the maximum number of nodes between the termination node and the root node ( the maximum tree depth ; mtd ) , or the population split no longer decreases the classification error by a minimum set amount ( the minimum error reduction ; mer ) . the nodes from which no further nodes branch are the leaf nodes . the computational complexity of the algorithm is @xmath24 , where @xmath25 is the number of examples , @xmath26 is the number of features , and @xmath27 is the depth of the tree . here @xmath25 is @xmath28 , whereas @xmath26 and @xmath27 are @xmath29 , giving a product of @xmath30 . the split is tested for each input feature . we use axis - parallel splits , in which , following @xcite , @xmath31 where @xmath32 is the @xmath8th attribute for example @xmath33 , and @xmath34 is a value to be tested . an alternative , which we have not utilized in this work , is the oblique split @xmath35 for @xmath27 attributes . this allows hyperplanes at arbitrary angles in the parameter space . here we allow the split point to be either the midpoint , the mean , or the median of the values of the parameters for the node population , with the best of these chosen for each individual node . with the removal of extreme non - physical outliers such as -9999 from the training data , selecting the optimal statistic from those measured at each node splitting provides better results than consistently selecting the same statistic . the classification error we used is the variance @xmath36 where @xmath8 is the output feature index , @xmath37 is the actual output feature value ( 0 or 1 ) , and @xmath38 is the probability prediction made by the decision tree . the data is modeled using the mean of all the output vectors , and the result is interpreted as a probability over class space . we used @xmath25-fold bagging ( bootstrap - aggregating ) in addition to the optimized tree and node parameters . in this procedure , the testing set is a randomly sampled fraction ( bagging fraction ) of the original testing set , and the procedure is repeated @xmath25 times , creating @xmath25 decision trees . the results are then voted on , using simple majority voting . similarly , the testing set itself is randomly subsampled from the training set , a procedure known as cross - validation . we perform 10-fold bagging , but use 1-fold cross - validation , as only the former improved the results . the parameters of the decision tree , and additional methods such as bagging , can have a substantial effect on the quality of the results produced . the massively parallel supercomputing resources we used in this work allowed extensive investigation of this parameter space , including : 1 ) the minimum decomposition population , 2 ) the maximum tree depth , 3 ) the minimum error reduction , 4 ) the method of splitting the population at a node , either half - and - half , midpoint , mean or median , 5 ) the ratio of sizes of the training set to the testing set , 6 ) the bagging fraction , and 7 ) the random seed used in selecting the subsamples for the testing set and the bagging . in addition , the construction of thousands of trees and the use of advanced visualization allowed us to investigate the interaction of these parameters . once we have constructed an optimal decision tree , given the available training data , we can characterize the best probability thresholds to construct samples from further data to the desired level of completeness or efficiency . we investigated the basic probabilities @xmath39 , @xmath40 and @xmath41 ( hereafter @xmath42 , @xmath43 , and @xmath44 ) and the ratios @xmath45 for each target type galaxy , star and nsng for binned minima and/or maxima , using the 80:20 pseudo - blind test described in [ sec : data ] . small numbers of objects in the leaf nodes resulted in discretization in the output probabilities from the decision tree . therefore , care was taken when comparing the resulting mixture of exact and floating point values to bin edges to ensure that the counts were correct and not offset due to finite floating point precision . the accuracy was achieved by using a decimal datatype , which stores numbers as exact values to a specified number of decimal places . as described in [ subsec : algorithms - dts ] , a decision tree has a number of parameters that can be adjusted . the product of these parameters defines an enormous parameter space of possible trees that must be explored to identify what we expect to be the optimal decision tree . in addition , this parameter space can potentially include many local minima of the classification error , further complicating this process . given the size of the parameter space and the computational time required to construct and test an individual decision tree , we utilized the ncsa _ tungsten _ cluster ( described in more detail in [ sec : algorithms ] ) to perform an extensive exploration of this parameter space . all decision trees were built from the same training set with the 16 input parameters described in [ sec : data ] , and the full parameter space we explored is described in table [ table : dt params ] . first , we studied the interaction of the minimum node decomposition population ( mdp ) , the maximum tree depth ( mtd ) and the minimum error reduction ( mer ) for splitting the population . the values investigated were @xmath46 , where @xmath47 , @xmath48 , and @xmath49 , where @xmath50 , and @xmath51 . negative mers do not split the population . this gave 4,480 combinations , and for each one a decision tree was generated . figure [ fig : dt params ] shows a partiview @xcite visualization of the results . the best tree had an mdp of 2 and an mer of 0 . mtd was limited to 16 by the combination of the size of the training set and the memory available on tungsten but , as shown in figure [ fig : dt params ] , a deeper tree is unlikely to give a substantial improvement . we next varied the statistic used to split the population at the nodes : halving the population , using the midpoint of each input , the mean of each input , and the median . the 16 combinations give very similar results , with the exception of those which allow population halving , which are significantly worse . as a result , we allowed the algorithm to select , when splitting each node , the optimal statistic from the midpoint , mean , and median split values . in all previous measurements , we had set the level of bagging used when building the decision tree at 20-fold and assumed a ratio of the fraction of the training set used in the bagging to the fraction of the sample used in training of 80:20 ( i.e. , 80% for training to 20% for testing ) . we conducted a number of tests to verify these assumptions . when varying the level of bagging , we found that the classification error is substantially reduced when the bagging level is increased from 1- to 10-fold , and continues toward an asymptotic value slightly below the value seen for the 50-fold case . as with our tests involving the maximum tree depth , the number of models was limited by the available memory on each node of the tungsten cluster , although given the asymptotic tendency , we do not expect that the results would change substantially for larger levels of bagging . we cross - tested the training to testing ratios with the n - fold bagging test , finding the best results were obtained using a 50-fold model with an 80% training subsample , as shown in figure [ fig : bagging ] . given memory limitations , the actual trees used in our other parameter tests were always limited to a 20-fold bagging . when performing our final classifications , however , we used the more accurate 50-fold bagging . in all subsequent tests we set the training to testing fractions to 80:20 . we also note that the bagging prevents an overfitting of the training set : for low values of bagging , such as 1-fold , the classification error has a minimum at a particular mtd , and increases substantially for deeper trees . finally , the effect of the random seed in choosing a same - sized training set was investigated and is shown in figure [ fig : random seed ] . the @xmath52 variation in the classification error is approximately 0.1% . given this variation , we quote the best classification error achieved as being @xmath53 . this is a robust result as the classification error shows a broad , approximately flat , minimum in several areas of the parameter space . see , for example , figure [ fig : bagging ] , where the flat area corresponds to approximately 3% classification error . although our exploration of the parameter space was extensive , the number of _ possible _ trees in the space quickly becomes very large with the variation of all parameters , so it is always possible that a better tree exists than the one we used . although given the widely observed 3% minimum it is unlikely to be _ significantly _ better , especially for the training set we used . once the optimal decision tree parameters were quantified , we constructed and subsequently tested this decision tree using unseen data to give a more realistic idea of the quality of our classifications . we achieved this result by building the tree with the previously described parameters , but training on 64% of the dr3 data and testing on 16% ( i.e. , maintaining the optimal 80:20 ratio ) . we tested this decision tree in a pseudo - blind fashion by using the remaining and unseen 20% of the dr3 data , which consisted of 95,413 sources . from this classification test , we compared the assigned types galaxy , star and nsng to the true types , in terms of completeness and efficiency as a function of the three probabilities @xmath42 , @xmath44 , and @xmath43 . overall , this experiment demonstrated that our axis parallel decision tree is very successful at classifying objects into these three types , and , as a result , at star - galaxy separation . the vast majority of objects are classified correctly and given high probabilities for the appropriate type . the confusion matrix is shown in figure [ fig : confusion matrix ] and tabulated in table [ table : confusion matrix ] . figure [ fig : true probability ] compares the true fraction of correct classifications as a function of assigned probability from the tree , demonstrating that these are approximately correct . we note that there is some discretization in the probabilities due to the potentially small number of objects in a leaf node . as can be seen in figure [ fig : dt params ] , however , increasing the minimum decomposition population to the point at which the probabilities would not be discretized would worsen the decision tree performance . in general the function for generating subsamples according to probability is @xmath54 in all subsequent tests , we analyze a subsample in terms of true positives ( tp ) , false positives ( fp ) , true negatives ( tn ) and false negatives ( fn ) for each of the three target types separately . we define the completeness , @xmath55 , as the number of the target type that are included in the @xmath56 cut compared to the total number of objects of that target type in the whole sample : @xmath57 where @xmath8 indicates the objects to the probability threshold and @xmath58 indicates all the objects in the test set . for example , in our pseudo - blind test this would correspond to all 95,413 sources . the efficiency , @xmath59 , is the fraction assigned a particular type that are genuinely of that type : @xmath60 the completeness varies between 0 and 1 and in general increases at the expense of efficiency . an object can be tp or fn depending on the probability threshold . we define the best sample for a specific test to be that which maximizes the metric : @xmath61.\ ] ] all the ratios , but not the simple probabilities , exclude objects for which any of the three probabilities are zero . in our pseudo - blind test , 18,989 of the 95,413 sources , or roughly 20% , were affected in this manner . we used the logarithm of a specific probability ratio , given the large dynamic range of the ratios of the the three probabilities . in all cases the classification is taken to be the object type assigned the highest probability by our decision tree . if the highest probability is less than 50% it is possible that two types are assigned an equal probability , which could introduce a systematic effect in our classification . we do not expect this to be problematic in practice , however , as this effect is small , only affecting one object in our pseudo - blind testing , and excluding these sources is easy after classification has been performed . to determine the best probability function , i.e. , @xmath56 , and associated probability values to use when assigning classifications , we computed a number of different functions involving the three probabilities assigned by our decision tree . table [ table : f(p ) cuts ] provides a compilation of these different functions and a characterization of their distribution sampled across one hundred bins . from the values in table [ table : f(p ) cuts ] , we find that by using the simple minima in the three class probabilities we obtain the most accurate classification samples in our pseudo - blind testing . these probability cuts are @xmath62 , which results in 98.9% completeness and 98.3% efficiency for galaxies , @xmath63 , which results in 88.5% completeness and 94.5% efficiency for nsng , and @xmath64 , which results in 91.6% completeness and 93.5% efficiency for stars . we find that a number of probability cuts near fifty percent produce similar classification accuracies . this is not surprising , as fifty percent is an ideal threshold , as objects below this percentage are likely to be of a different class since they can have a higher probability of being one of the other two types . in some cases , ratios of probabilities are able to select better samples when a maximum value is also used as opposed to only using a minimum value . for example , for the probability ratio @xmath65 , the sample is dominated by stars at the low end and galaxies at the high end . so the nsng are only selected by requiring a minimum and a maximum @xmath56 . in the end , however , none of the ratio cuts we have studied provided a better combination of completeness and efficiency than the simple probability cuts . we interpret this result as further evidence that our decision tree is successful , especially within the regime of spectroscopically confirmed objects used for training ( i.e. , the interpolation regime ) . the full completeness - efficiency curves , as a function of @xmath42 , @xmath43 , and @xmath44 , are shown in figure [ fig : com eff ] . we constructed our final decision tree from the full ( i.e. , 100% ) , spectroscopic training sample from the sdss dr3 . by using the data - streaming modules available within the d2k environment , we applied this decision tree to the full sdss dr3 ` photoprimary ` catalog of 142,705,734 sources . with this paper , we publicly release the classification probabilities for every one of these sources . following the classification guidelines discussed in the previous section , within the magnitude range @xmath66 , this sample consists of 38,022,597 galaxies , 57,369,754 stars and 47,026,955 nsng . the sum of these subsample sizes do not match the size of the full sample because of ( a ) the magnitude restriction , which removes 589 objects , and ( b ) source classifications that are ambiguous due to equal - highest probabilities for multiple object types , which removes 285,839 ( 0.2% ) objects . one of the most important issues with any supervised classification effort is the application of an algorithm trained on one sample of data to a different set of data . this concern is generally framed in terms of interpolation versus extrapolation , and in our case results from the application of our decision tree that is trained on sources from the sdss spectroscopic sample being applied to the sdss photometric sample , which is , in some cases , considerably fainter than our training set . given the strong dependence on the number counts of galaxies and stars with apparent magnitude , this affects the vast majority of the sources we have classified . without deeper training data , however , we feel that the best approach is to classify all sources . this approach allows anyone using our classification catalog to create a well - defined sample for further study or follow - up spectroscopy that does not have any selection effects artificially imposed due to our classification algorithm . while we can not guarantee the accuracy of our classifications across the full sdss photometric sample , we can provide some guidance on the accuracy of specific sample restrictions . historically , one of the most powerful techniques for characterizing classifications in the absence of spectroscopic identifications is by analyzing the differential number of sources as a function of apparent magnitude ( i.e. , number count plots ) . this technique is simple and straightforward ; by comparing the number count distribution of training sources with the number count distribution of classified sources , we gain a statistical estimate of the apparent magnitude limit at which we can no longer expect our extrapolated classification rules to remain robust . we present , therefore , in figure [ fig : mag counts ] , the number counts for both the training set and for the classified ` photoprimary ` catalog for our three different classifications : galaxy , nsng , and star . the sdss spectroscopy is limited to @xmath67 for galaxies @xcite , which can clearly be seen in the galaxy training set number counts . similarly , the nsng training set , which is dominated by quasars , is limited to @xmath68 @xcite , which is reflected in the decline in the @xmath7-band number counts around @xmath69 . the logarithmic slopes of the galaxies and nsng are approximately 0.5 and 0.6 respectively for the training set . stars , on the other hand , have a more heterogeneous selection scheme in the sdss survey , with many subpopulations , which is reflected in the number count distribution of the stellar training set . we see clear evidence for reliable star - galaxy separation in figure [ fig : mag counts ] , where the nsng counts remain considerably lower than the star and galaxy counts until @xmath0 . given the spectroscopic limits of the sdss , this figure suggests that the ` photoprimary ` classifications are reliable to substantially fainter magnitudes than the spectroscopic sample . the galaxy counts approximately follow the logarithmic slope for around two magnitudes fainter than the training sample , and the nsng counts extend even fainter . the inflection points in the counts at @xmath70 suggest that our classifications are reliable to this magnitude and fainter than this the tree becomes less reliable and increasingly assigns objects to nsng , which is one of the primary reasons we included this third classification in our algorithm . the sdss photometric sample does extend fainter , the 95% detection repeatability limits for point sources being @xmath71 , @xmath72 , @xmath73 , @xmath74 and @xmath75 , which is seen in the number count distributions as the abrupt turnover at @xmath76 . if , as the number counts suggest , our classifications are reliable to @xmath0 , the numbers of reliable classifications are 7,978,356 , 832,993 , and 13,945,621 , totaling 22,756,970 for the galaxy , nsng , and star classes , respectively . we must emphasize , however , that the number of sources that will actually be scientifically useful will be lower , as these samples have not been cleaned . for example , before using these data , appropriate cuts should be made on either photometric errors , detection flags , or both , as described more fully on the sdss project website . since we used colors , as opposed to magnitudes , to classify our sources , we also compared , as shown in figure [ fig : color counts ] , the @xmath77 distribution of training set sources with the ` photoprimary ` classified sources . the extrapolation in @xmath77 color in going from the testing set to the classified sources is considerably less than the corresponding extrapolation in magnitude , although it is still present , particularly in the nsng class toward redder colors . as expected for the training set , galaxies and stars show a bimodal distribution in color and the nsngs are dominated by blue objects ( e.g. , uvx - selected quasars ) . this bimodality is washed out for galaxies in our classified sample , however , presumably due to the fact that the photometric sample has a higher mean redshift than the spectroscopic sample . we note that this difference in mean redshift will not affect the apparent magnitude number count distributions , as the redshift difference is too small to allow for significant evolution to occur between the two samples . in figure [ fig : p(gns ) ] , we present a visualization of the three classification probabilities for all objects in ` photoprimary ` , demonstrating how the three values interact . since the three probabilities must always sum to unity , their distribution appears as a triangle , with the height corresponding to the density of objects in that region of probability space . one can see that galaxies and stars are assigned relatively unambiguously , but that the tree is generally less certain when an object is neither a star nor a galaxy , as the base of the triangle rises noticeably before reaching @xmath78 . discretization is also seen in this figure , but is insignificant compared to the numbers of each source type with high probability . the data used to produce the visualization are available to download with the catalog . in the pseudo - blind test results described earlier ( [ subsec : results - testing ] ) , our decision tree provided robust classifications for data that was not used to train the decision tree , nor to characterize its performance . while this pseudo - blind test did provide a good approximation to a blind test , a true blind test is to compare the decision tree classifications to sources that have been assigned a spectroscopic identification from another survey . we have , therefore , matched the sdss dr3 photometric data to the 2dfgrs and 2qz surveys , by using coordinate position with a tolerance of 2 arcseconds . the 2dfgrs resulted in 50,191 matches and the 2qz 10,259 matches . the 2dfgrs contains mainly galaxies and the 2qz stars and quasars with a small number of mainly narrow emission - line galaxies . by applying the optimal cuts of @xmath62 , @xmath63 and @xmath64 , as detailed in [ subsec : results - testing ] , we can characterize our classification accuracy in this blind test as a function of apparent magnitude . we show these results for the 2dfgrs galaxies in figure [ fig : com eff galaxy ] for the 2qz stars in figure [ fig : com eff star ] . for galaxies , the overall completeness of the 2dfgrs galaxies is 93.8% ( 47,055 of 50,191 ) , with the efficiency undefined as this is the only type of object . the completeness of the stars in the 2qz match is rather low , at 79.7% ( 4141 of 5193 ) , but the efficiency is 95.4% ( 4141 of 4359 ) . figure [ fig : com eff star ] shows that the low completeness is due to stars at @xmath79 . the low value may be due to the fact that the 2qz is uvx - selected , which causes unusual stars that lie off the main - sequence , such as sub - dwarfs , halo stars , and hot young stars , to be preferentially targeted . we also blind - tested the nsng classifications in a similar manner to the galaxy and star classes , but in this case by using quasars from the 2qz with the same matching criteria as the one we used for stars from the 2qz . as shown in figure [ fig : com eff quasar ] , our decision tree does an excellent job of assigning quasars to the nsng class , with a completeness of 94.7% ( 8278 of 8739 ) and an efficiency of 87.4% ( 8278 of 9471 ) for the type _ 11 _ best ids and with a completeness of 94.4% and an efficiency of 85% for the _ 12 _ , _ 21 _ and _ 22 _ next - best ids . the efficiency drops off from around 98% fainter than the sdss spectroscopic limit of @xmath68 , and is about 87% at @xmath70 when the counts reach their limit in the 2qz . this result is further vindication of adding the third class to our algorithm , as these sources are clearly neither stars nor galaxies , and our algorithm identifies them appropriately . overall , these blind tests support our assertion , based on the number count distributions , that our classifications remain reliable to @xmath2 . the axis - parallel decision tree algorithm we have adopted in this work is an example of supervised machine learning . one of the primary criticisms of supervised techniques is their difficulty in extrapolating past the limits of the actual data used in their construction . supervised learning algorithms can be viewed as a mapping between the training parameter space and a classification , and are not necessarily a representation of anything physical ( e.g. , a decision tree might correctly classify stars , but it would not necessarily reproduce the hertzsprung - russell diagram ) . ideally , the training data should sample , perhaps sparsely , the entire parameter space occupied by the data to be classified . in the case of astronomical datasets , such as the star - galaxy separation problem being studied in this paper , this is often impossible . we do not have a sufficiently large and faint spectroscopic survey to adequately sample either the magnitude or color spaces spanned by the photometric data from the sdss survey . yet we see that our classifications do reliably extrapolate past the training data . what is most likely driving this successful extrapolation is that the source colors do not change dramatically prior to @xmath2 . as a result , the existing training data is sufficient to classify sources in this area of parameter space . at these fainter magnitudes , the algorithm increasingly assigns sources to the nsng class , as the distinction between stars and galaxies is breaking down with decreasing signal - to - noise . we can characterize this effect from the classifications in the final catalog , where the fraction of nsng sources in the full catalog rises from approximately 1.3% for @xmath80 to approximately 3.7% for @xmath0 . this observation , together with the obvious examples of sources that are neither stars nor galaxies , strengthens the case for introducing a third class into the historical star - galaxy separation problem . to aid in the utilization of our classification catalog , we have made available classifications for all sources in the sdss dr3 , without regard for photometric errors , or object detection flags . as a result , anyone wanting to use our catalog must subsequently apply any appropriate flag or photometric error restrictions as appropriate for their science just as they would when using the sdss databases directly . we feel that this is an important step forward in the realm of astronomical source classifications . while we are certainly cognizant that a large fraction of our source classifications are not highly reliable , we feel it is important in the new era of large surveys to provide an astronomer with the greatest possible freedom in selecting their own samples . if we released only those sources that pass a potentially opaque classification cut , we force all users of our catalog to deal with an additional , implicit selection effect when creating their own subsamples from our classification catalog . in this paper , our supervised learning algorithm utilized photometry and spectra from the sloan digital sky survey . while the sdss is an incredible resource for this sort of investigation , it is limited to the optical and near - infrared wavelengths , and , consequently , physical processes outside these wavelengths that might distinguish objects otherwise indistinguishable in the optical will not be seen . given the growth in survey astronomy , a large number of sdss objects have been observed in surveys at other wavelengths , and thus an important next - step is to include these in the training set . this extension has two stages : 1 ) match the sdss photometry to spectroscopic targets from other surveys and train on the sdss photometry with these augmented training data , and 2 ) add the photometry from other surveys as additional training parameters . the latter is a substantial task as the cumulative number of photometric objects in surveys such as galex @xcite , 2mass @xcite , swire @xcite and rosat @xcite number in the millions . we are in the process of matching the sdss to these surveys , and others that may become available such as ukidss , to improve the classification performance achieved here . of course this technique also extends to subsequent data releases from the sdss itself . we also would like to investigate alternative mechanisms for selecting sdss photometric parameters for training , including the use of techniques such as forward - selection , backward - elimination , or a hybrid of the two , in which training parameters are respectively added or removed according to their effect on the quality of the classification . this approach is adopted by @xcite , who uses this technique for morphological galaxy classification . the training parameters could utilize magnitudes rather than colors , and also morphological parameters such as the @xmath81 measure . in this paper , we have only used the colors of objects through four different matched apertures , as this work is our first major effort in large - scale machine classification of astronomical datasets . in the future , we anticipate performing a more detailed exploration of different classification parameters , especially as we incorporate additional datasets at other wavelengths into our algorithms . one training parameter that was found to not be useful in the current framework is the error on an object attribute , in agreement with previous analyses ( see , e.g. , * ? ? ? fundamentally , this is because , without additional information , the decision tree can not differentiate between an error and another parameter . coupled with the different dynamic ranges of the two types of attributes , this effect essentially trivializes the error attributes . since irrelevant attributes dilute the real information content in a training dataset , they reduce the performance of a decision tree . as a result , we did not include magnitude errors in our analysis . intellectually , a better technique might be to include the error as part of the appropriate attribute when constructing a machine learning algorithm . our decision tree implementation , however , does not provide this capability , and we therefore defer this concept to a future paper . a further decision tree algorithm , supported in the data - to - knowledge framework but not yet extensively investigated , is _ boosting _ , in which the differences between classes are emphasized . this method has been shown to improve results by many workers in the data mining field and is described further in @xcite . besides decision trees , d2k supports numerous other algorithms , including nave bayes , artificial neural networks , support vector machines , k - means clustering , and instance - based learning . the latter is of particular interest because , although it has shown promising results , its full potential has not been realized , due to the computational requirements of calculating the classifications for the full application set . in this method , the distance between each test or application set object ( each instance ) and the nearest @xmath25 neighbors in the training set is measured and a weighted mean is taken to give the new object its assigned type . thus the training consists of specifying @xmath25 , the weighting , and the distance measure , typically of the form @xmath82 in each dimension where @xmath27 is the distance and @xmath83 would be the euclidean distance . hence there are a large number of distance calculations involved . typically the classification of an object takes of the order of a second on a desktop machine , rendering the process computationally intractable for the 143 million objects in the sdss dr3 . the process is , however , embarrassingly parallel , so is amenable to a significant speedup by using supercomputing resources . in which case , this technique could potentially be used to not only classify sources , but to characterize the energy generation mechanisms responsible for a source s luminosity . we anticipate applying this technique , for example , to characterize the fraction of every source s luminosity in terms of the ratio between the thermonuclear fusion powering stars and the accretion powering quasars . in many applications it has been found that a _ mixture of experts _ , that is , the combination of results from more than one learning method , is best , in an analogous way to the improvement seen here through the use of bagging . a simple way to combine results is to take a weighted sum of outputs from the different methods . preliminary tests have shown that instance - based with @xmath84 and @xmath85 in a 0.7:0.3 combination with a decision tree gives improved results . of course , one must be able to run a full instance - based classification to explore the potential of combining it with the other methods , which heretofore has been impossible due to the necessary computing resources . another promising future method is semi - supervised classification @xcite , in which the algorithm is given some predefined classes but is able to discover further classes in the data for itself . this combines prior knowledge from spectra with the large number of objects for which there is available photometry and may perform better where very few spectra are available . finally , in this work we limited our analysis to three classes as our main interest was robust star - galaxy separation . given the diversity of both stars and galaxies , and the ignorance of our approach to non - stellar luminosity , we could clearly benefit from adding additional classes to our existing decision tree work ( see , e.g. , * ? ? ? as it stands , the nsng class is dominated by quasars at bright magnitudes , and could , therefore , serve as the basis for a quasar class that we expect would provide reasonable quasar classifications for relatively bright sources with clean photometry . ideally , this approach could be extended to include other classes for different types of stars and galaxies , for example , white dwarfs , hot stars , cool dwarfs , starbursts , and active galaxies . in this paper we have classified all of the 142,705,734 non - repeat ( primary ) objects in the third data release of the sloan digital sky survey ( sdss dr3 ) . the classifications were determined by using machine learning , the algorithm of choice being the axis - parallel decision tree . the algorithm was trained on 477,068 objects for which spectroscopic classifications were available within dr3 . this training data consists of 361,490 galaxies , 62,333 stars , 49,545 quasars , and 3,700 unknown objects . collaboration with domain experts at the national center for supercomputing applications ( ncsa ) and the use of the general machine learning environment data - to - knowledge combined with supercomputing resources enabled extensive investigation of the decision tree parameter space and the associated datasets . to our knowledge , this level of investigation has not previously been published in the astronomy literature . this is the first public release of objects classified in this way for the whole survey . the objects are classified as either galaxy , star or nsng , and we provide an associated probability for each class . these three probabilities always sum to unity for every object classified . by merely assigning the classification with the highest probability , we find that our full classification sample contains 38,022,597 galaxies , 57,369,754 stars and 47,026,955 nsng in the magnitude range @xmath86 , with 589 outside this range and 285,839 ( 0.2% ) ambiguous due to equal - highest probabilities for object type . the catalog is available for download at http://quasar.astro.uiuc.edu/rml . a major issue with this method of classification is the inevitable extrapolation from the spectroscopic regime . we investigate this by examining the apparent magnitude number counts for all sources and by performing several pseudo - blind or fully blind tests by using sources spectroscopically identified in the sdss , 2df galaxy redshift , and 2df qso redshift ( 2qz ) surveys . we find that for a sample giving the optimal completeness and efficiency , the completeness values are 98.9% , 91.6% , and 88.5% for galaxies , stars , and nsngs in the sdss , and 94.7% for quasars , which are classified as nsng , in the 2qz . the corresponding efficiencies are 98.3% , 93.5% , 94.5% , and 87.4% . the number count distributions suggest that the classifications are reliable for @xmath0 , giving 7,978,356 galaxies , 832,993 nsng , and 13,945,621 stars . as we have not applied any restrictions to the classification catalog , such as limiting by photometric error or detection flags , the number of reliable sources will be lower ; however , we provide these classifications to eliminate any opaque selection effects from our classification catalog , which simplifies the task of using these classifications to supplement additional analyses , such as as defining a target sample for follow - up spectroscopy . the assignment of probabilities to each object allows one to investigate the completeness and efficiency of the classifications as a function of these probabilities . while ratios of the probabilities were investigated , we find that the samples with an optimal completeness and efficiency are yielded by the simple cuts @xmath62 , @xmath63 and @xmath64 . other values close to 0.5 also give very similar results . the full data describing the completeness and efficiency as a function of these three threshold probabilities is made available with the catalog . we feel that the application of machine learning algorithms to large , astronomical surveys is a rich area of research . we are currently augmenting our training data with both additional wavelengths and fainter spectroscopic identifications . to improve our classification results , we are also performing additional tests to determine the optimal parameters for decision trees built from these data . given the efficacy of our approach , we plan to increase the number of training classes used in our analysis , first by using just the sdss , and later by using additional photometric and spectroscopic data . finally , we are also exploring the efficacy of more powerful algorithms , such as instance - based classification , to tackle the fundamental goal of characterizing sources by the fraction of their energy that is derived from fusion and accretion . these results will presented in subsequent papers in this series . nmb , rjb and adm acknowledge support from nasa through grants nag5 - 12578 and nag5 - 12580 , microsoft research , and from the nsf paci project . the authors made extensive use of the storage and computing facilities at the national center for supercomputing applications and thank the technical staff for their assistance in enabling this work . funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s . department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england . the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions . the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpa ) , the max - planck - institute for astrophysics ( mpia ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . data to knowledge ( d2k ) software , d2k modules , and/or d2k itineraries , used by us , was developed at the national center for supercomputing applications ( ncsa ) at the university of illinois at urbana - champaign . this research has made use of nasa s astrophysics data system . ll minimum decomposition population & 32768 16384 8192 4096 2048 1024 512 256 128 64 32 16 8 4 2 1 + maximum tree depth & 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 + minimum error reduction & 999999.0 100000.0 10000.0 1000.0 100.0 10.0 1.0 0.1 0.01 0.001 0.0001 0.00001 0.000001 0.0 + one half split & false true + midpoint split & false true + mean split & false true + median split & false true + number of repetitions & 1 2 5 10 + fraction train examples & 0.01 0.1 0.3 0.5 0.7 0.8 0.9 0.99 + number of bagging models & 1 10 20 50 + bagging fraction & 0.001 0.01 0.1 0.3 0.5 0.7 0.8 0.9 0.99 0.999 + random seed subsamples & 1 ... 32 + random seed bagging & 1 ... 32 + lcccc & _ galaxy _ & 71,748 & 608 & 735 + number & _ nsng _ & 397 & 9,514 & 309 + & _ star _ & 345 & 451 & 11,305 + + + & _ galaxy _ & 75.2 & 0.637 & 0.770 + percentage & _ nsng _ & 0.416 & 9.97 & 0.324 + & _ star _ & 0.362 & 0.473 & 11.8 + ccccccc @xmath39 & galaxy & 0.49 & 1.00 & 98.9 & 98.3 & 98.6 + @xmath40 & nsng & 0.54 & 1.00 & 88.5 & 94.5 & 91.5 + @xmath41 & star & 0.46 & 1.00 & 91.6 & 93.5 & 92.6 + @xmath87 & galaxy & 0.821 & 4.82 & 98.2 & 97.4 & 97.8 + @xmath87 & nsng & -4.70 & -0.989 & 76.4 & 86.2 & 81.3 + @xmath87 & star & -2.32 & 2.06 & 94.2 & 35.0 & 64.6 + @xmath88 & galaxy & 0.770 & 4.93 & 97.6 & 97.5 & 97.6 + @xmath88 & nsng & -2.38 & 1.97 & 95.3 & 32.3 & 63.8 + @xmath88 & star & -4.32 & -0.711 & 74.3 & 83.4 & 78.9 + @xmath89 & galaxy & -1.62 & 1.85 & 96.7 & 92.6 & 94.7 + @xmath89 & nsng & 2.19 & 4.02 & 56.5 & 90.0 & 73.3 + @xmath89 & star & -4.23 & 3.15 & 100.0 & 7.74 & 53.9 +
we provide classifications for all 143 million non - repeat photometric objects in the third data release of the sloan digital sky survey ( sdss ) using decision trees trained on 477,068 objects with sdss spectroscopic data . we demonstrate that these star / galaxy classifications are expected to be reliable for approximately 22 million objects with @xmath0 . the general machine learning environment data - to - knowledge and supercomputing resources enabled extensive investigation of the decision tree parameter space . this work presents the first public release of objects classified in this way for an entire sdss data release . the objects are classified as either _ galaxy _ , _ star _ or _ nsng _ ( neither star nor galaxy ) , with an associated probability for each class . to demonstrate how to effectively make use of these classifications , we perform several important tests . first , we detail selection criteria within the probability space defined by the three classes to extract samples of stars and galaxies to a given completeness and efficiency . second , we investigate the efficacy of the classifications and the effect of extrapolating from the spectroscopic regime by performing blind tests on objects in the sdss , 2df galaxy redshift and 2df qso redshift ( 2qz ) surveys for which spectra are available . we find that , for a sample giving a maximal combined completeness and efficiency , the completeness values are 98.9% and 91.6% and the corresponding efficiencies are 98.3% and 93.5% for galaxies and stars , respectively , in the sdss . we also test our star - galaxy classification by studying the inverse of this sample , finding quasars in the nsng sample with a completeness and efficiency of 88.5% and 94.5% from the sdss and 94.7% and 87.4% from the 2qz . given the photometric limits of our spectroscopic training data , we effectively begin to extrapolate past our star - galaxy training set at @xmath1 . by comparing the number counts of our training sample with the classified sources , however , we find that our efficiencies appear to remain robust to @xmath2 . as a result , we expect our classifications to be accurate for 900,000 galaxies and 6.7 million stars , and remain robust via extrapolation for a total of 8.0 million galaxies and 13.9 million stars . the latter sample should prove useful in constructing follow - up spectroscopic surveys . characterizing the success of our classifications fainter than @xmath2 will require fainter spectroscopic data .
the question of how spiral galaxies manage to sustain star formation activity over cosmological times is key to our understanding of galaxy formation and evolution . finding an answer is not easy , given that most observations point towards very short gas depletion timescales . for example , our galaxy forms new stars at a rate of @xmath2m@xmath3yr@xmath4 ( @xcite , @xcite ) . its gas mass is about @xmath5m@xmath3 ( see @xcite and @xcite ) . if we assume a constant star formation rate ( sfr , * ? ? ? * ) , then the remaining gas should be converted into stars within about @xmath6 gyr . similar values or order of only a few billion years are reported for many nearby disk galaxies ( for observational evidence , see for example @xcite , @xcite , or @xcite ; for theoretical arguments , see e.g. @xcite ) . at higher redshifts of @xmath7 , the depletion timescale can be even shorter . values of @xmath8gyr have been inferred for the total gas ( e.g. * ? ? ? * ) , while the numbers for the molecular gas component can be as low as @xmath9gyr @xcite . in vigorously star forming merging systems , the observations suggest even shorter timescales , down to @xmath10gyr . we note , however , that these are average numbers , with the data showing significant variations , both from galaxy to galaxy and within each galaxy as a function of many parameters , not just column density ( for further discussions , see ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? ? * ; * ? ? ? if we discard the possibility that most spiral galaxies are observed at the verge of running out of gas , and instead assume that they evolve in quasi steady state , then this requires a supply of fresh gas delivered to the galaxy at a rate roughly equal to its star formation rate . indeed , there is additional support for this picture . first , galactic accretion flows are a natural outcome of cosmological structure formation calculations . for example , @xcite and @xcite argue that massive galaxies are continuously fed by steady , narrow , cold gas streams that penetrate through the accretion shock at the virial radius into the central galaxy ( see also * ? ? ? roughly three quarters of all galaxies forming stars at a given rate are fed by smooth streams ( see also * ? ? ? * ) , at least up to redshifts of @xmath11 . the details of this process are still under debate , as they seem to depend on the numerical method employed and on the way gas cooling is implemented ( e.g. * ? ? ? nevertheless , even if most of the gas shocks at the virial radius and builds up a hot ionised halo , some fraction of it may cool down again and become available for disk accretion . as proposed by @xcite this gas could condense into higher - density clumps at the interface between halo and disk and then `` rain '' down onto the disk later in the process . this transport of matter converts potential energy into kinetic energy , and constitutes a source for the observed turbulence on smaller scales and thus contributes to controlling the star formation process in the galaxy @xcite . indeed , gas accretion has been proposed to be the main source of turbulence in the extended outer disk of galaxies , where the star formation activity is low and consequently stellar feedback can not provide enough energy and momentum to explain the observed velocity dispersion in the hi gas @xcite . second , indirect support for continuous accretion comes from the fact that the observed amount of atomic gas in the universe appears to be roughly constant , from a redshift of @xmath12 till today , while the stellar content continues to increase , suggesting that overall the hi content of typical galaxies is continuously replenished from some external reservoir @xcite . third , for our galaxy at least , further evidence for an ongoing inflow of low - metallicity material stems from the presence of deuterium in the solar neighbourhood @xcite as well as in the galactic centre @xcite . as deuterium is destroyed in stars and as there is no other known source of deuterium in the milky way , it is thought to be pristine material of extragalactic origin @xcite falling on the galaxy for the first time . if gas is accreted onto a disk galaxy from great distances , the new material is most likely delivered to large galactic radii . this is a consequence of angular momentum conservation , and it finds additional support from the fact that the outer disk simply has a larger cross section than the central regions . as most of the star formation occurs in the inner parts , one might expect an inward flow of gas through the disk . it is this radial mass flux that we aim to detect in this study . in many galaxies neutral hydrogen ( hi ) is found out to radii much larger than the optical disk ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? this gas provides us with the possibility to look for infall signatures . indeed , the asymmetry commonly seen in the outer disks of galaxies are often considered to be the result of gas accretion @xcite , triggered either by recent minor mergers @xcite or by tidal interaction @xcite . some galaxies show one very prominent spiral arm extending beyond the rest of the hi disk . others have tails or a very asymmetric and lopsided disk or show substantial warps , especially in the outskirts . also hi bridges to companion galaxies are observed . @xcite argue that these observations are evidence for recent or ongoing accretion and infall of cold gas , and they claim that in at least 25% of field galaxies indications for a merger event or tidal interactions can be detected . the detection of high velocity clouds ( hvcs , @xcite ) points in a similar direction . these are extra - planar hi clouds with low metallicities @xcite , with anomalous kinematics @xcite , and with velocity gradients that are perpendicular to the galactic plane @xcite . many of them have a head - tail shape as is expected for the motion of colder and denser clouds through the tenuous hot halo gas of the milky way . their origin , however , remains unclear . they could be fresh material falling onto the galaxy for the first time . or the hvcs could be the fallback of galactic fountain flows @xcite . in this case hvc s contain enriched gas that is ejected from the disk by stellar feedback . this gas is likely to fall back at larger radii . as it is lifted from the disk midplane , it runs down a vertical pressure gradient and some fraction of it spreads to larger radii . it travels through the hot halo gas above the plane that is rotating more slowly @xcite . as a consequence , the ejected material has larger angular momentum than its environment , and as it mixes with the extraplanar gas , it falls back to the disk at larger radii than it came from . gas that accretes onto the outer disk must move inwards through this disk , before it becomes converted into stars in the star forming inner regions of the galaxy . to search for these radial gas flows , @xcite studied seven spiral galaxies based on co and hi observations . they found elliptical and bar streaming motions and could give upper limits on the magnitude of possible radial inflows . however , without the availability of deep large - scale maps , the focus was on the inner disks , where the non - circular motions detected were of the order of @xmath13km@xmath0s@xmath4 , comparable to the level of turbulence in the disk . our approach is different . we focus on the extended outer disks of nearby spiral galaxies , where the gas density is low and therefore the local streaming velocity needs to be high in order to result in an appreciable mass flow . we base our analysis on 21 cm line observations of neutral hydrogen in the hi nearby galaxy survey ( things , @xcite ) . this allows to trace gas up to very large radii , typically far beyond the optical radius of the galaxy . the survey provides two - dimensional high - resolution velocity maps , and our approach to model these velocity fields is in general known as tilted ring fit . the map is decomposed into individual rings which are circular within the plane of the disk , but inclined with respect to the line - of - sight and therefore appear elliptical . for each of them , inclination , position angle , circular velocity , and other parameters are determined . the method was first described by @xcite and developed further by e.g. @xcite and @xcite . these studies derived rotation curves for a number of galaxies based on hi maps from the westerbork synthesis radio telescope . other studies extended this concept by also fitting non - circular velocity components of different kinds to the velocity field . high - order harmonic decompositions were used by @xcite ( applied to hi observations of ngc 2403 and ngc 3198 ) and @xcite ( search for elongated potential in 19 things galaxies ) . kinematic studies assuming flat disks and models of particular streaming motions were conducted by @xcite ( search for radial inflow based on co and hi observations of seven spiral galaxies ) , @xcite ( bar - like streaming motions in ngc 2976 based on h@xmath14 and co observations ; development of the ` velfit`-codespekkens / velfit/ ] ) and @xcite ( improvement and extension of the previous study to five other galaxies based on h@xmath14 and hi measurements ) . spekkens and collaborators later developed ` diskfit ` @xcitespekkens / diskfit/ ] , an improved version of ` velfit ` which is not strictly limited to constant inclination and can also operate on photometric data . @xcite applied ` diskfit ` to h@xmath14 , co and hi kinematic data of ngc 6503 . models with constant inclination are not suitable for studies including the regime beyond the star forming disk , where the hi disks often show variable inclination . furthermore , high - order fourier decompositions would require the disk geometry to be determined in a separate , prior step by a low - order tilted ring fit . in this study , we develop a fitting scheme that circumvents the complications associated with this two - step approach and at the same time can be applied to disks with ( moderately ) varying inclinations . our study is organised as follows . in section [ data ] we describe the things data set . in section [ theory ] we outline our new method , discuss its stability properties , and compare to previous approaches . in section [ verification ] we apply our method to mock data , investigate its stability properties , and discuss its limitations . we apply the method to a sample of 10 galaxies from the things survey and present our results in section [ results ] . finally , we summarise and conclude in section [ sec : conclusions ] . the main observational basis of this study are high spatial resolution observations of atomic hydrogen for a large set of nearby galaxies . these are taken from the hi nearby galaxy survey ( things , @xcite ) . it is a large survey of 34 nearby galaxies in the 21 cm line of neutral hydrogen and was conducted at the nrao very large array ( vla ) . the survey provides a high spatial resolution of 6 `` to 12 '' and a channel width of @xmath15 . not all galaxies targeted within the things survey are suitable for our analysis . objects have to be well resolved ( large angular size ) , have intermediate inclinations ( above @xmath16 to @xmath17 ) and show a relatively regular velocity field . we applied our method to all promising candidates and in the end , our analysis yielded useful results for 10 galaxies which are listed in table [ table1 ] . our targets have distances between 3 and 15 mpc which results in a physical resolutions between 100 and 500 pc . the survey provides a detailed census of the hi distribution in a large number of nearby galaxies at high angular resolution as is thus well suited for kinematic studies . for this work we use various data products provided by the survey . the intensity ( zeroth moment ) maps reflects the total hi column density and the first moment map the hi velocity field . data cubes and moment maps are available in two version . they differ in the weighting scheme applied to the interferomeric data from individual antenna pairs during image reconstruction . the robust scheme is optimised for maximising resolution and contrast by giving long baselines a higher impact . we choose the data products assuming natural weighting . this scheme is more suitable for us since since it provides better signal - to - noise and better recovers diffuse emission . a detailed description of the survey and the data processing can be found in @xcite . additionally , we also use velocity maps derived by fitting hermite functions directly to spectra in things data cubes . they should better reflect the bulk motion of the gas but require data with slightly higher signal - to - noise ratio compared to first moment maps . [ ebhis ] there is always the possibility that interferometric observations miss flux at large spatial scales due to incomplete @xmath18-coverage . to minimise this issue , things includes observations in the compact vla - d configuration . for verification , we compare the things intensity maps to single dish observations from the effelsberg - bonn hi survey ( ebhis ) @xcite and find no significant discrepancy in total fluxes for the regions of the disks we model in this study . a more detailed analysis by daniel lenz ( aifa bonn , private communication ) came to the similar conclusion that usually no more than 10% flux is missing in the things maps , at least as long as the galaxy fits within the vla primary beam . optical observations are used to find the nearside of our target galaxies which , in combination with the velocity information , allows us to uniquely determine the direction of rotation and to break the degeneracy between inflow and outflow . the optical images are taken from the nasa / ipac extragalactic database . to compare our inflow results to the average recent star formation rate , we use galex far uv ( fuv ) images . we use the final data products available from the galex archive . in the following sections [ tiltedringanalysis ] and [ fourierdecomposition ] we introduce basic theoretical concepts regarding modelling 2d velocity fields of disk galaxies and describe the different fitting procedures we evaluated . sections [ twostepfit ] and [ simultaneousfit ] compare sequential versus simultaneous fitting approaches and their impact on derived quantities . the optimum procedure we finally adopted is described in section [ fit_procedure ] the quantity accessible to observation is the line - of - sight velocity ( @xmath19 ) . to infer the rotational and radial velocities of the gas in the disk from spatially resolved measurements of @xmath19 , one has to utilise a specific model that can be fitted to the data . the straightforward approach is to assume a velocity field which is symmetric with respect to the disk centre and projected under some viewing geometry . the measured @xmath20 then has the form : @xmath21 here @xmath22 denotes the systemic velocity of the galaxy with respect to the observer . @xmath23 is the rotation velocity and @xmath24 is a velocity component in radial direction in the plane of the disk . the other variables are the inclination @xmath25 of the disk and the azimuthal angle @xmath26 which together with the radius @xmath27 forms a polar coordinate frame within the plane of the galaxy . the disk is assumed to be thin and no motions perpendicular to the plane of the galaxy are taken into consideration . all quantities mentioned above can in general be treated as functions of the radius @xmath27 . in the analysis , the disk is basically sliced in circular annuli and the above decomposition of @xmath20 done for each ring individually which leads to the term tilted ring analysis . the line - of - sight velocity is a function of the sky coordinates while the terms on the right hand side of equation [ eq1 ] are functions of the coordinates within the plane of the disk . therefore , one needs a coordinate transformation between sky coordinates ( @xmath28 , @xmath29 ) and the disk coordinates ( @xmath27 , @xmath26 ) . to do this transformation , the geometry of the disk has to be known . it can be described by the centre coordinates ( @xmath30 , @xmath31 ) , the inclination ( @xmath25 ) and the position angle of the apparent major axis of the disk ( @xmath32 ) as measured in the plane of the sky . one representation of the transformation equations is the following : @xmath33 @xmath34 this transformation is not straightforward to solve for @xmath27 and @xmath26 since the parameters @xmath30 , @xmath31 , @xmath32 and @xmath25 implicitly depend on @xmath27 , if one describes the disk as sequence of independent rings with possibly different inclinations , position angles , and centre coordinates . in this case , the global transformation from sky coordinates to disk coordinates is in general neither unique nor continuous . this is , because the rings can overlap or gaps between them can appear . the only reason this transformation can be used in practice is , because @xmath25 and @xmath32 usually vary smooth and slowly . or to be more specific , the analysis has to be restricted to galaxies where this condition is fulfilled . . , title="fig : " ] + . , title="fig : " ] + . , title="fig : " ] an illustration of the analysis is given in figure [ fit_illustration_ngc5055 ] . it shows the observed hi velocity of ngc 5055 as a `` spider '' diagram , where the lines denote regions of constant line - of - sight velocity , and in colours it indicates the corresponding tilted ring fits . it is clearly noticeable that ngc 5055 has a warped disk , but we also see how the tilted ring fit adapts to the varying disk geometry and follows the changing inclination and position angle . usually , not all parameters are kept free in a fit . in some schemes like the one implemented in ` velfit ` @xcite it is assumed that the disk is flat , and therefore all radii have the same inclination and position angle . in such cases , @xmath25 and @xmath32 are determined once for the whole galaxy and then kept fixed for each radius . because we include the outer parts of the disks in our analysis which are often warped to some degree , we allow @xmath32 and @xmath25 to vary independently for each ring . only considering average values of @xmath25 and @xmath32 would not represent the actual geometry of the disk to the necessary precision . however , leaving all fit parameters ( @xmath32 , @xmath25 , @xmath30 , @xmath31 , @xmath23 , ... ) completely unconstrained often leads to unstable fit results . we therefore choose to keep the ring centres fixed , assuming that all parts of the galaxy rotate around a common centre - of - mass . this is a good approximation for galaxies that are not subject to strong external perturbations , such as exerted by a massive satellite galaxy . we determine the galaxy centre from the inner , mostly flat part of the disk and enforce this centre coordinates for all rings . by including the @xmath24 term , the tilted ring fit as described above , already includes non - circular motions . also all parameters especially inclination and position angle can be allowed to vary from ring to ring and therefore with radius . however , the velocity field of each ring is still restricted to rotational symmetry around the ring centre . one can generalise the model by allowing @xmath23 and @xmath24 to vary with respect to the azimuthal angle @xmath26 . again , it is not possible to allow arbitrary functions here , since not enough observables are available to determine them . the usual approach is to expand the azimuthal dependence of @xmath23 and @xmath24 in a fourier series and include terms up to a certain order : @xmath35 @xmath36 the line - of - sight velocity has to be expanded in the same way , and for notational convenience a factor @xmath37 is already pulled out of all fourier parameters : @xmath38\;.\ ] ] we denote fourier parameters corresponding to the velocities in the disk frame by capital letters , while the ones referring to the line - of - sight velocity are lower case . by inserting equations [ eq2 ] and [ eq3 ] in equation [ eq1 ] and comparing with equation [ eq4 ] , one can find relations between the fourier coefficients of the line - of - sight velocity and the fourier parameters of the velocities in the disk frame : @xmath39\;. \end{split}\ ] ] one finds that a radial or rotational velocity component of order @xmath40 contributes to order @xmath41 and @xmath42 to the line - of - sight velocity . also , there are roughly two times as many fourier coefficients in the disk frame ( @xmath43 , @xmath44 , @xmath45 , @xmath46 ) than are accessible by observations ( @xmath47 , @xmath48 ) . this makes a unique reconstruction of the actual velocities impossible . we therefore restrict our analysis to second order in @xmath20 , corresponding to first order in @xmath23 and @xmath24 . this simplifies equation [ eq5 ] to : @xmath49\;. \end{split}\ ] ] in this simplified model it is easier to solve for the fourier coefficients of @xmath23 and @xmath24 . the comparison to equation [ eq4 ] shows that the @xmath20 parameter @xmath50 directly represents the rotation curve and the @xmath51 parameter corresponds to the 0th order radial velocity , sometimes also denoted as @xmath52 . there are only two remaining degeneracies . in principle the 0th order term @xmath53 is degenerate with the systemic velocity . however , this is not a significant concern , because the systemic velocity should be the same for all rings and therefore needs to be determined only once . this can , for example , be done by directly using the velocity of the galaxy centre . a degeneracy that can not be solved concerns @xmath54 , because @xmath55 and @xmath56 only appear in this term . to make progress , we adopt @xmath57 . that is , we assume that both terms contribute equally . within a single ring , all parameters are modelled without any radial dependence and therefore with constant rotation velocity . this prohibits a perfect fit to the data in the inner part of a galaxy with roughly linearly rising rotation curve . because the focus of this study are not the galactic centres , and because spiral galaxies show a mostly flat rotation curve over most of their outer disk , this is not a concern for this study . [ solid_body ] solid - body rotation produces another problem . it leads to parallel iso - velocity contours in the observed velocity field and makes it impossible to disentangle @xmath58 and @xmath37 . therefore our analysis can not be applied to dwarf galaxies which usually have linearly rising rotation curves over most of their hi disk . the same effect causes problems in the inner parts of spiral galaxies . since these parts are not the focus of our study we disregard this regime when applying our analysis and interpreting our results . the fourier components themselves form an orthogonal base and therefore can be fitted independent of each other . however , at least the components of odd order couple to inclination and position angle . it is therefore not obvious that one can first determine the ring geometry with a tilted ring fit of low order , and then in a second step , apply a higher order fourier decomposition using the previously determined ring geometry . because this is an often - used procedure , implemented e.g. by the ` reswri ` tasks ( which uses ` rotcur ` as the first step ) within the ` gipsy ` package , we decided to test the applicability of this approach . we follow a two - step procedure : first , a tilted ring analysis of first order is performed , where we allow inclination and position angle to vary fully independently for each ring . in a second step , a fourier decomposition of the velocity field is calculated , in which inclination and position angle are fixed to the values determined in the tilted ring fit . we compare two slightly different methods . in the first one , the initial tilted ring fit already includes radial motions in the form of the @xmath59 term in equation [ eq2 ] . in the second one , it is restricted to a purely rotational velocity field . the latter is similar to the one used by @xcite and @xcite . technically , they performed the fourier decomposition not on the full velocity field , but instead on the residuals of the tilted ring fit . nonetheless , in this approach it means the ring geometry is derived from a purely rotational tilted ring fit . the results of our comparison of the two schemes applied to the ngc 2403 data are presented in figure [ ngc2403-novrad ] . the derived rotation curves turn out to be nearly identical and also the inclinations are very similar . the measured position angle is systematically higher in the case where no radial motions are permitted in the tilted ring fit . the deviation in position angle of @xmath60 appears small and may be dismissed as not significant at first glance , however , it causes the measured @xmath51 component in the fourier decomposition to change substantially . if the ring geometry is determined without taking radial motions into account , the subsequent fourier analysis is not able to recover any significant radial velocities in the second step of the analysis . the @xmath51 component varies between 0 and @xmath61 , which can be taken as a null result . this is in general agreement with the results of @xcite and @xcite , who also find little evidence for radial gas motions . if in turn radial motions are already allowed in the tilted ring fit , the conclusions change and we find clear signs of radial inflow . the @xmath51 component is always well below zero and shows a large feature between radii of 350 and 900 with a maximum inflow velocity of @xmath62 . in this scheme , the radial motions are already part of the tilted ring fit , which itself does not depend on any a - priori assumptions except the centre position . we conclude that taking the radial velocity term into account in the first analysis step makes this approach more general and allows us to recover a radial velocity flow if it is present in the data . to better quantify the differences between the two approaches , we create a synthetic velocity field , similar to the one derived from the ngc 2403 data . we choose the disk to be flat with a constant position angle of @xmath63 and an inclination of @xmath64 . for the rotation velocity , we adopt a rotation curve of brandt - type @xcite in the rising part and a constant velocity of @xmath65 outside of 500 . to this ideal disk we add an inflow component starting at a radius of 350 and linearly increasing to @xmath62 at 450 . it then declines again linearly and returns back to zero at 950 . the exact shape of the rotation curve is not crucial , since the inflow we try to recover happens only in the flat part of the rotation curve . we then fit this velocity field in the same way we treated the ngc 2403 data . we first apply a scheme that includes a radial component already in the tilted ring analysis , and then use the method that does not take it into account . noise is not added to the mock data since our focus is solely on the systematics of both methods . in addition , simple gaussian noise would largely average out due to the huge number of data points in the mock velocity map . the results are shown in figure [ mock2403 ] . the analysis with inflow in the tilted ring stage perfectly recovers all parameters . there are only small deviations of the @xmath51 component in the innermost parts of the disk and some finite - resolution effects . the solid curves representing the fit are indistinguishable from the dashed lines representing the input of the mock velocity field and most importantly , the fit recovers the correct inflow pattern . in contrast to this , the analysis method that does not include radial motions in the tilted ring fit is not able to properly recover the radial flow . it shows some signal in the @xmath51 component , but the maximum amplitude of @xmath66 km s@xmath4 is only a small fraction of the correct value . one can see that the position angle determined in the tilted ring analysis yields an incorrect value to compensate for the inflow . the deviation is roughly proportional to the radial velocity and peaks at a mismatch value of about @xmath60 . the overall picture is similar to the deviation between the two methods seen for the ngc 2403 data . since the radial velocity component is already compensated for by the disk geometry in the tilted ring stage , the subsequent fourier decomposition can not recover more than a small residual @xmath51 component , and thus disguises potential signatures of substantial radial gas motions . as demonstrated above , it is necessary to include the radial velocity component already in determining the ring geometry . the next question is , what happens to the second or higher orders in the fourier decomposition . from our experience with fits to mock velocity fields , we come to the conclusion that the second order terms do not couple to @xmath25 and @xmath32 in a significant way . it should therefore be no problem to fit them in a subsequent step , once the ring geometry was determined in a tilted ring fit . however , since they are orthogonal not only to the first order terms , but also to @xmath25 and @xmath32 , it is also only a minor additional complication to fit them all simultaneous . to avoid any issues from carrying out a fit in two distinct steps , we investigate a fitting scheme where we fit inclination , position angle and all velocity components up to second order at the same time . compared to the two - step approach including radial motions in the tilted ring analysis , the simultaneous fit yields nearly identical results . it is equally reliable and stable . before outlining the details of the simultaneous fitting approach that we apply to the things data in section [ fit_procedure ] , we comment on the prospects of including terms that are higher than second order in the fourier decomposition . we find that this is a difficult exercise . for example , the @xmath67 component is highly degenerate with the inclination . this makes a simultaneous fit of @xmath67 and @xmath25 impossible . one can not compare the results of a two - step approach to a simultaneous fit of third order . a third order fourier decomposition based on a disk geometry determined in a first or second order analysis can not yield the correct value for the @xmath67 term . the parameter @xmath67 is implicitly assumed to be zero in the first step , and so the inclination is fitted to a value that compensates for a possible non - zero @xmath67 term . one could in principle fix @xmath67 and fit everything else up to fourth order , but there is no justification for assuming the @xmath67 component to vanish or to be of any particular form while at the same time trying to measure the @xmath68 and higher components . therefore it seems there is no satisfying way to safely deduce higher than second order fourier parameters without the risk of systematic errors , at least not from the velocity field alone and in a purely kinematic approach . it is for this reason that we restrict our analysis to a simultaneous fit of second order in @xmath20 . this limits the level of detail we can derive but on the other hand reduces the necessary assumptions and related systematics to an absolute minimum while also making the model as flexible as possible since everything is fitted simultaneous . the model we finally adopt is a second order fourier decomposition of the following form : @xmath69\;. \end{aligned}\ ] ] the coordinates of the disk frame ( @xmath27 , @xmath26 ) are related to the sky coordinates ( @xmath28 , @xmath29 ) by equations [ trafo1 ] and [ trafo2 ] . the coordinate transformation itself depends on the galaxy centre , the inclination and the position angle . we fit all five fourier parameters ( @xmath70 , @xmath51 , @xmath50 , @xmath71 , @xmath72 ) as well as @xmath25 and @xmath32 for each ring simultaneously . the centre coordinates ( @xmath73 , @xmath74 ) and the systemic velocity have to be determined in advance and are kept fixed for all rings . we do that by fitting a simple tilted ring model in which @xmath22 and the centre coordinates are free to vary to the innermost area of the velocity field . in this part of the disk , the fit is stable even without assuming a fixed centre . we then take the average values of the ring fits . the choice of @xmath22 only influences the @xmath70 component in the fourier decomposition . all other parameters are unaffected by that and an incorrectly chosen @xmath22 remains transparent throughout the analysis . the ring widths are chosen for each galaxy individually to get the best compromise between good sampling and fit stability . typical widths are between 24 and 48 which is always larger than the things synthesised beam width of @xmath75 " @xcite . in the common case of warped disks , gaps naturally appear between rings and the observational data within the gaps of our tilted ring model are unused . to alleviate this effect to some degree and to make maximal use of all available data , we choose the separation of adjacent rings to be 37% smaller than the width of the rings . this closes most of the gaps in the model but also leads to some overlap of rings . the exact amount of overlap depends on the fitted ring geometry . since each ring is fitted individually and independently of all other rings , the overlap has no influence on the fit result . the only consequence is that neighbouring rings are statistically slightly correlated ( however , omitting every second ring in our analysis and the following figures leads to a statistically independent sample ) . we start the fit procedure with the central ring . after the fit for this ring has converged , the fit results are taken as initial values for the second ring . this procedure of outward forwarded initial values is used for all succeeding rings . we only have to manually set the initial values for the central ring to reasonable starting values . their exact choice has negligible impact on the overall result . in general , the question of whether the fit for a particular ring converges or not depends more strongly on the quality of the data and on possible irregularities in the velocity field than on the given initial conditions . for good quality data , the fit converges to the right value even if extremely poor guesses for the initial conditions are made . for disturbed disks , low inclinations , and especially velocity fields with linearly rising rotation curves , the rings have the tendency to flip to a very high inclination . this often happens in the innermost parts of the galaxies , and is likely caused by solid body rotation and bars . if this affects only the few innermost rings , we can simply ignore them and use the well fitted outer part , since this is where our scientific interest is focused on . linearly rising rotation curves make it impossible to disentangle i and @xmath58 . it can especially in the centre of galaxies happen that the fit converges to obviously incorrect values ( i very close to 90 ) . if this affects only a few rings in the galaxy centre while the other rings are fine , we discard the obviously wrong ones and keep the rest . after the ring fit is completed , we calculate the @xmath59 and @xmath58 parameters according to equation [ eq8 ] . as mentioned before , the degeneracy between @xmath56 and @xmath55 can not be broken and we assume @xmath57 . the final step is to convert velocities to mass flow rates . from the derived fourier parameters , we construct a two - dimensional synthetic radial velocity field by taking only fourier components of radial direction into account . to convert the things hi integrated line intensity ( zeroth moment ) maps to column densities and finally to hi mass surface density , we adopt the method and distances given in @xcite and assume the hi gas to be optically thin . the things first - moment maps represent an intensity - weighted and therefore approximately mass - weighted mean velocity at each position @xmath76 . as outlined above , we decompose these velocities into a rotational , @xmath77 , and radial , @xmath78 , component . in order to compute a 2d distribution of the radial mass flow rate , we simply need to multiply @xmath79 for each position . to obtain the net flow rate at each radius , we integrate over the azimuthal angle @xmath80 : @xmath81 here , @xmath82 denotes the radial mass flow rate and @xmath83 the hi mass surface density . we note that radial velocities can be directed inward or outward so that @xmath82 represents a net flow rate . we also note that in practice we calculate @xmath82 in radial bins of finite width @xmath84 ( compare description earlier in this section ) . we interpret these radial hi net mass flow rate profiles and compare them to sfr profiles of our galaxy sample in section [ continuity ] . for the fit procedure described in section [ fit_procedure ] we use a standard @xmath85-optimiser , which also gives formal statistical uncertainties . however , in order to obtain more realistic uncertainty estimates , we carry out additional monte - carlo modelling . that is , we add random noise with specific characteristics to our data and measure how this influences the results . we repeat this 100 times to build up a statistically significant sample . we point out that adding purely gaussian noise would not serve our purpose , because it would average out due to the large number of data points in the hi maps . we are more concerned about potential systematic patterns in the velocity field that we can not account for in our fitting routine . @xcite advertise a concept that uses the residuals of the fit as a noise source . the idea is that every residual structure not included in the model is in fact noise to the fit , and the residuals can be used as a prior to create a noise field with statistical characteristics that are well matched to the data . the authors randomise and rescramble the two - dimensional map of the residual by rotating and shifting chunks of it in radius by random amounts . this creates a statistical noise field , but it also preserves any structures and coherence on small scales . we follow the same approach , but we modify the rescrambling procedure since in our model inclination and position angle are allowed to vary throughout the disk . we therefore do not deal with one continuous velocity field but instead with an ensemble of tilted rings that can in general have gaps and overlapping regions . we therefore apply this procedure to each ring individually . the exact procedure we use is the following : after the fit , we calculate for each ring the residual after subtracting the fit from the data . each residual is deprojected according to the ring geometry determined in the fit , and it is rotated around the ring centre by an individual angle , randomly chosen between @xmath86 and @xmath87 . it is then reprojected using the same geometry . an artificial noise map is created by combining all rings , which is then added to the observed velocity field . the perturbed data cube is then fitted again . the resulting residuals are rescrambled in the same way and added to the original data to be used in the next iteration . we repeat this procedure 100 times , and so obtain an ensemble of 100 values for our fit parameters . from this , we compute mean and standard deviation which we take as our final results . we investigated the number of iterations necessary for convergence by running up to 2500 iterations and found that 100 is sufficient . the processes that transport gas within a galaxy and transform it into stars are complex . in order to compare our hi radial flux measurements to the sfr profiles , we present a simplified model that is solely based on mass continuity arguments : @xmath88 @xmath89 is the hi column density or equivalent the hi mass surface density and @xmath90 the hi mass flux density . we assume the disk to be thin and therefore only deal with a two - dimensional problem . if we include the star formation rate surface density @xmath91 as a sink term on the left hand side , focus only on the radial mass flux and use gauss s theorem to express everything in local quantities , we find : @xmath92 for this study we only deal with azimuthally averaged quantities . this justifies ignoring rotational streaming in the previous step and leads to the following relation : @xmath93 the quantity @xmath94 is our measured radial mass flux profile . we note that we do not take into account the contribution from helium when we quote hi masses or surface densities ( a factor of 1.36 ) . for an appropriate description , one probably also has to add additional sink and source terms to the left - hand side . these could for example reflect stellar winds and outflows , galactic fountains or gas that condenses from the hot halo gas or circumgalactic medium and settles to the disk . these processes are in general difficult to model and beyond the scope of this toy model . however , if we measure a violation of equation [ pile - up ] , we have an indication that some of these processes may be present . if one assumes the special case that star formation and other sink and source terms vanish and observes the radial mass flux to be constant ( but not necessarily zero ) over some radial range , then the hi column density should not change with time and the galaxy would be in a ( quasi ) stationary state . on the other hand , if @xmath95 is negative , gas should pile - up . if we assume our galaxies to be in such a stationary state and therefore @xmath96 , the mass flux at a given radius equals the total star formation within this radius : @xmath97 this is a relation we can test by comparing our radial mass flux measurement to cumulative star formation rate profiles , which we derive from galex fuv data . for this we closely follow the procedure described in @xcite and use the conversion factor from @xcite . we take the ring geometry from our kinematic analysis and for each ring compute the total sfr from the galaxy centre out to this radius . we correct for galactic extinction according to @xcite and blank bright stars but neglect extinction within the target galaxies . a detailed comparison of the various sfr calibrations using different tracers ( e.g. , h@xmath14 , uv or ir emission ) is beyond the scope of this study . to be consistent with accurate , published galaxy integrated sfrs , we scale our cumulative sfr profiles to match the literature values ( table [ table1 ] ) at large radii , which is equivalent to assuming a constant intrinsic extinction . we note that this approximation is sufficient for the purposes of this study and any more accurate treatment of internal extinction at small radii does not affect our conclusions . as we show in section [ results ] , the naive assumption of vanishing time dependence and absence of other processes is rarely appropriate , but the deviations from this simplified model might give insights on the magnitude of other processes . as we show in section [ twostepfit ] , the overall fitting scheme can be in general quite sensitive to prior assumptions . we therefore conduct a suite of tests in order to assess the correctness of our results and to determine the potential limitations of our approach . we focus mostly on the analysis of synthetic velocity fields and on testing under which conditions the input values can be recovered correctly . but we also investigate the dependence of our results on the type of velocity maps ( such as intensity - weighted first - moment or hermite ) that are used . in figure [ mock2403 ] we demonstrate that a two - step fit with radial components at the tilted ring stage is perfectly able to recover radial motions from a synthetic velocity field , at least if the disk is flat . this is also true for the second order simultaneous fit scheme introduced in section [ fit_procedure ] which is used for the rest of the paper . however , galactic disks are not flat , but instead they exhibit warps and changes of position angle . both constitute a great challenge for any tilted ring like analysis , because the deprojection of the velocity field is no longer continuous and unique . in figure [ fig : mock03 ] we show an example of a fit to a synthetic velocity field with mild warps and position angle changes . we start again with a flat disk with a brandt - type rotation curve . it is then modified by a decrease in inclination @xmath25 by @xmath98 over 300 in radius , a change of position angle @xmath32 by @xmath16 over 300 , and a decrease of rotation velocity of @xmath99 per 100 starting at 350 . in addition , a pattern of in- and outflows of different magnitude up to @xmath99 is added . all these single effects cover different radial ranges . therefore , a variety of combinations of effects is simulated at the same time . again , since we are interested in systematic effects , no noise is added to the artificial velocity field . the plots in figure [ fig : mock03 ] show that our method can fully recover the parameters of the input velocity field to a very high accuracy . problems only occur , as expected , at very small radii . in addition , a slight tendency for deviations exists at points where the slope of parameters abruptly changes . but overall , the result is highly satisfactory . so , our method is valid at least for mildly disturbed disks . in the following we will address how our fitting procedure reacts to stronger warps . generally , a decrease of inclination with radius will not be a problem , but a strong increase will be : taking two annuli with radii @xmath100 , @xmath101 , @xmath102 and inclination angles @xmath103 , @xmath104 , @xmath105 , their apparent semiminor axes have the projected length @xmath106 . if @xmath107 , the two annuli completely overlap on the minor axis and our assumption of a thin disk with unique line - of - sight velocities is no longer valid . this condition can be converted to an expression for the maximum inclination increase that does not lead to a pile - up of rings on the minor axis : @xmath108 here , @xmath109 denotes the increase in radius from one ring to the next and @xmath110 the increase in inclination between the two rings , @xmath25 and @xmath27 are inclination and radius of the inner of the two rings . while limited overlap of adjacent rings is in practice uncritical ( and even enforced by the rings chosen 60% wider ) , an inclination increase exceeding the critical rate will make even non - adjacent rings overlap . the apparent minor axis of the rings shrinks with increasing radius . in such conditions , the tilted ring approach breaks down . similar overlapping effect can also appear for changes of the position angle . since the geometry of position angle changes is more complex , we do not derive an analytic expression for this case . in the end , it would anyway be necessary to consider the combined effect of inclination and position angle changes , which complicates things even further . . obviously , the analysis is not capable anymore to fit these data correctly . the steep increase of @xmath25 with nearly twice the critical rate derived in equation [ i_crit ] causes a mismatch of all parameters . the errors in @xmath25 , @xmath32 and @xmath23 seem small but they result in a significant error in the @xmath51 term of up to @xmath111 . also , the change in @xmath32 at @xmath112 is now too fast to still allow a correct derivation of the @xmath51 term . ] our experiences with fits to synthetic velocity fields are in good agreement with the above analytic considerations . in cases where the inclination decreases ( e.g. fig . [ fig : mock03 ] ) or increases with less than the critical rate given in equation [ i_crit ] , we see no significant deviations . if the inclination is increased close to the critical rate , some noticeable deviations appear , but the derived results are still usable with less than @xmath113 maximum error for the @xmath51 component ( not shown ) . however , much larger changes of the inclination angle cause strong artefacts in the fit results . this is illustrated in figure [ fig : mock04 ] , where the inclination angle changes with twice the critical rate . at the warp , the inclination is typically fitted to a too high value . this is because data at larger radii interfere with those from smaller radii . in addition , also the position angle and the rotation velocity are fitted to a slight mismatch . at first glance , these deviations do not appear severe , however , as discussed earlier , they lead to large deviations for the important @xmath51 parameter , with errors of of @xmath15 and more over a large range in radius . this is of the same order of magnitude as potential inflow velocities , and therefore can render the fit result useless . interestingly , no deviation for the radial component is seen if the @xmath51 component in the synthetic velocity field is set to zero , although the derived values for @xmath25 and @xmath50 actually differ from the input . also visible in figure [ fig : mock04 ] is an artefact in the @xmath51 curve due to a rapid change of the position angle , which is not perfectly accounted for . the inclination and rotation curve stay quite stable with only minor deviations , but the radial velocity shows incorrect peaks of up to @xmath99 . the @xmath16 change in @xmath32 over only 75 " in radius is a very fast one . if this change is stretched over 150 we find only small deviations of less than @xmath114 . at a rate of @xmath16 over 300 as in figure [ fig : mock03 ] , any significant deviations have vanished . we also investigate the influence of an incorrect determination of the galaxy centre . our tests showed that an offset of the order of a few arcseconds away from the true centre is not critical for our analysis . it creates artefacts , but their amplitude quickly decrease towards larger radii . for example , offsets of the order of 5 and 10 are no longer noticeable outside of 100 and 200 , respectively . we conclude that the uncertainties involved in determining the galaxy centre are therefore not an issue within the context of this study . first moment maps are often employed to derive velocity maps from data cubes . they are easy to compute and do not require very high s / n levels . however , if the line profile is asymmetric or has extended wings , this method does not properly represent the bulk velocity of the gas . we therefore also include velocity maps from @xcite in our analysis . these are based on fitting gauss hermite functions to the hi spectra of each spatial resolution element . by modifying the gauss function by the third order hermite polynomial , the fitting function is capable of adapting to asymmetric line shapes and in this case better represents the bulk velocity . a detailed description can be found in @xcite . we apply our method to the galaxy ngc 2403 and compare the standard moment map provided by things to the corresponding hermite map . the result is illustrated in figure [ ngc2403_hermite ] . both velocity maps do not give exactly the same result but they are in general agreement . the most obvious difference lies in the intermediate part of the galaxy . in the moment map we find a maximum inflow velocity of @xmath62 at 450 . the inflow then linearly decreases to @xmath115 at 900 . the hermite map shows even more inflow at around 500 followed by a rapid decrease to a roughly constant inflow velocity of @xmath116 between 600 and 800 . beside this , the results are quite similar and most importantly , both maps show the same unambiguous signs of large scale inflow . unfortunately , fitting hermite functions to the data cube requires substantially higher s / n compared to the moment maps . it can therefore not be applied to the outermost parts of the hi disk . based on the results of this comparison we decide to proceed using the moment maps for our analysis . the derived values for hermite- and moment maps approximately agree with each other and the moment maps extend further out from the galaxy centres which is particularly important for our aim of detecting large - scale radial inflow . in the following section we present our results . for each galaxy we present four plots describing the fit parameters as functions of radius . radius and velocities are measured within the galactic plane . for the exact details of the analysis see section [ fit_procedure ] . the four plots contain the following information : 1 . inclination and position angle , 2 . the fourier parameters @xmath50 and @xmath51 , which directly correspond to the principal components of the rotation and radial velocity , respectively , 3 . the fourier parameters @xmath70 , @xmath72 and @xmath71 , which give information about the azimuthal dependence of the velocity field , 4 . the calculated radial hi mass flux , not accounting for helium , @xmath117 or other gas fractions and the cumulative star formation rate profile giving the total sfr within a certain radius based on galex fuv images and scaled to the more precise literature values given in table [ table1 ] . the rotation direction of the galaxies is determined from optical images , especially from the more pronounced appearance of dark filaments on the nearside of the disk . by combining this information with the velocity field , one can distinguish if the galaxy is rotating clockwise or counter - clockwise , which also determines if potential radial motion is directed inwards or outwards . in our sample of 10 things galaxies , ngc 2403 turns out to be the most suitable target for detecting radial mass inflow patterns . it has a large apparent size of roughly 16 x 10 , was observed with good signal - to - noise ratio , and most importantly , it has a very flat and smooth hi disk and regular velocity field . our fitting results are summarised in figure [ ngc2403 ] . position angle @xmath32 and inclination @xmath25 are plotted as function of radius in the top left part of the figure . the galaxy is rotating clockwise , and so the south - western part of the disk is closer to earth . the rotational velocity ( fourier component @xmath50 ) is between 120 and 140@xmath118 . as already described in section [ twostepfit ] , we find clear evidence for radial inflows . the derived radial velocity ( fourier component @xmath51 ) is always negative with values of at least a few @xmath119 . the most significant radial motion is found at radii between 400 and 900 ( top right panel of figure [ ngc2403 ] ) , and the strongest hi mass flux of 3 m@xmath120 is observed at 450 ( figure [ ngc2403 ] , bottom left panel ) . this mass flux linearly declines towards larger radii . ) . it can be seen that most of the uv emission appears in the inner part of the disk , inside the region of peak inflow . this at least allows the interpretation that the gas moves inwards where it is transformed into stars . ] the quality of the fit we obtain for this galaxy is excellent . all parameters vary smoothly and adjacent rings show only very limited scatter . the residual deviation between data and our model ( shown in figure [ ngc2403_residuals ] ) are of the order of @xmath121 and show no distinct pattern which would indicate a mismatch between fit and data . the flocculent spiral arms of ngc 2403 are just barely visible in the residual map . the terms @xmath70 , @xmath72 and @xmath71 , which indicate the deviation from a rotational symmetric velocity field , are fairly small inwards of 900 ( figure [ ngc2403 ] , bottom right part ) . in the outskirts of the galaxy , there may be some lopsidedness . one can see some correlations between the different parameters around 200 . we can not completely exclude that at this small radii a possible mismatch in e.g. the inclination induces artefacts in all parameters but these effects only appear in the inner part of the galaxy . the star formation rate given in @xcite is 0.38 m@xmath120 . this is of the same order as our minimum hi inflow rate but nearly an order of magnitude lower than our maximum inflow rate of 3.5 m@xmath120 . if we follow our highly simplified model described in section [ continuity ] , we would expect the radial mass inflow rate all radii to equal the star formation rate within that radius according to equation [ mass_conservation ] . the bottom left panel of figure [ ngc2403 ] shows these two quantities and it is obvious that at most radii both quantities do not add up to zero . also , we would expect the inflow to be monotonic with radius and to be highest at the outer edge of the hi disk . this is clearly not the case . not surprisingly , there are probably additional processes other than inflow and star formation influencing the hi budget of ngc 2403 . one possible effect could be the ejection of ionised material from stellar winds perpendicular to the plane followed by recondensation and settling of that gas onto the plane at larger radii . this is often described as the galactic fountain model and could explain the additional mass flows . for ngc 2403 some indication for this was discovered by @xcite . by taking position velocity cuts parallel to the major and minor axis , they find substantial amounts of extraplanar hi gas that rotates more slowly than the disk and is reported to show a net inward motion . the gas they detect lies roughly in the same radial range ( 200 to 700 ) as the main inflow found in our analysis . however , our method has only limited sensitivity to extraplanar gas because its presence will not substantially change the intensity weighted mean used to compute the moment maps . another possible way to deviate from our stationary model is time - variable accretion , for instance if accretion happens via discrete events ( e.g. , large gas clouds ) . the spatial correlation of inflow pattern and star formation is visualised in figure [ ngc2403_image ] . it shows a composite of the galex fuv image and the reconstructed radial velocity field . the inflow has a maximum at 460 `` and decreases strongly at smaller radii . as described in section [ continuity ] , this should lead to a pile - up of gas or enhanced star formation . the galex image shows clear clustering of star formation regions in this area . most of the star formation takes place inside of 200 '' , again a region where the magnitude of the inflow decreases . this spatial correlation may provide support , at least qualitatively , for a scenario in which hi inflows are a key parameter regulating star formation in this galaxy . however , a quantitative comparison shows that the amplitudes of sfr and inflow do not match , which may not be surprising since a variety of other factors ( time evolution , stellar winds , gas phase changes , etc . ) likely play an important role as well . our observation for ngc 2403 clearly shows a qualitative correlation between star formation and radial hi flows . this gives support to the picture that star formation may be regulated by inflows in ngc 2403 . @xcite analyse gravity - induced torques in the spiral galaxy m51 , how they drive radial gas motions and how these are related to the local star formation efficiency . they find that radial motions seem to counteract star formation in this galaxy so that regions that experience higher torques show increased gas depletion times . this is the opposite sense of our finding here . m 51 and ngc 2403 , however , have quite different properties and are not immediately comparable . the spiral arms in ngc 2403 are by far not as pronounced as in m 51 and the gravitational torques and dynamical driven inflows they might induce should be much lower and probably negligible . our residual map ( figure [ ngc2403_residuals ] ) indeed shows no signs of spiral arm structure . it is therefore not necessarily expected that the `` dynamic pressure '' @xcite describe suppresses star formation in ngc 2403 as it does in m 51 . unfortunately , due to its low inclination , we can not analyse m 51 with the method presented in this study and directly compare the results . ngc 2841 has a very extended hi envelope , several times larger than the stellar disk and rotates counter - clockwise which means the north - eastern part of the disk is the near side . our fits to this galaxy are very stable and show a flat disk with very smoothly rising inclination and position angle ( figure [ ngc2841 ] ) . the radial velocity ( fourier component @xmath51 ) is close to zero up to a radius of roughly 550 . we find a small inflow in the range from 350 to 550 which is consistent with the reported sfr of 0.74 m@xmath120 @xcite . from our sfr profile we find that most of the star formation is located within 200 `` while we detect inflow between 300 '' and 600 " . outside of 550 the radial velocity rises to @xmath122 which corresponds to an hi outflow of 2 m@xmath120 . the @xmath71 term indicates some lopsidedness of this radial flow . from the appearance of very pronounced dust lanes in visual images , it is clear that the galaxy rotates counter - clockwise and the measured radial flow is indeed directed outward . from the morphological appearance one could speculate about tidal or merger effects , however ngc 2841 is not accompanied by any major galaxy and only surrounded by several dwarf galaxies ( a ned search yields only minor companions with @xmath123 within @xmath124 and @xmath125 ) . it is interesting to compare the hi distribution with the radial velocity field . a two dimensional representation of this is given in figure [ ngc2841_vrad+mom0 ] . it shows the radial velocity field we reconstruct from the determined fourier parameters as described in section [ fit_procedure ] . overplotted are contours of the things integrated intensity map . the hi distribution shows a prominent ring at intermediate radii . if a radial mass flow causes some accumulation of gas we would expect this to happen as described in section [ continuity ] at the location where the slope of @xmath126 is the most negative . from figure [ ngc2841 ] we see that this is the case at around 360 `` . this radius is marked in figure [ ngc2841_vrad+mom0 ] with a purple ellipse which exactly coincides with the hi ring . a second , but weaker feature like this can be seen at 510 '' . this might indicate that our highly simplified model described in section [ continuity ] even without any additional sink or source terms might actually provide a reasonable description for the mass flow processes in ngc 2841 . quantitatively , however , a mass flux with an amplitude of @xmath127 would have to act for a long time period to cause an enhancement in the hi distribution of ngc 2841 , which has a total hi mass of @xmath128 . ngc 2903 is rotating counter - clockwise and has complex kinematics . visual images and the heracles co - maps @xcite show a very clear bar structure which does not appear in the integrated hi maps . the bar in conjunction with the solid body rotation in the inner regions prevents us from deducing proper fit results for radii smaller than 100 and also causes a quite complicated structure outside of this radius ( see figure [ ngc2903 ] ) . at 200 , our analysis shows a sharp minimum of the inclination , followed by a steep increase from @xmath129 to @xmath130 within 200 . this is close to the fastest inclination change rate we expect from equation [ i_crit ] to still yield reliable fit results . we take this as an indication that the fit in this region is likely to be not very reliable . this dip in inclination furthermore coincides with two regions of very low hi flux , basically holes in the hi disk . in general , due to the complex kinematic structure of this galaxy the fit results for radii below 250 could be affected by artefacts and should probably not taken at face value . the largest feature in our @xmath51 curve is a massive inflow signature around 300 . similar to ngc 2841 , we can assess whether the radial gas motions could lead to the accumulation of gas in the ring structure visible in the integrated intensity map . based on equation [ pile - up ] , we would expect a pile up of gas at a radius between 200 `` and 300 '' due to the very steep negative slope of @xmath131 but this region shows a rather low hi surface density . the ring structure is located at a radius of @xmath132 . this does not correspond to any particular feature in the radial flux profile . the peak inflow at 300 " also does not coincide , but is just inside the ring structure . in contrast to ngc 2841 , our simple model described in section [ continuity ] is obviously not able to describe the relation between radial motions and hi density in ngc 2903 . we speculate that the ring structure might instead be related to more complex internal streaming motions , probably associated with the galactic bar , the effects of which are likely not captured correctly by our 2nd second - order analysis @xcite . no further galaxy in our sample shows gas density features in sufficient detail to allow a straight forward comparison of radial motions to hi density . the absence of gas accumulation can always be explained by the low amplitude of order @xmath133 of the radial motions . further out beyond 500 `` , the kinematic structure of the disk seems less complicated and the velocity measurements indicate a quite substantial inward motion . however , due to the low gas surface densities at these radii , the corresponding hi mass flow rate is only 0.3 m@xmath120 . this is quite small but similar to the sfr reported by @xcite , who report values ranging from 0.4 m@xmath120 derived from uv measurements up to 1 m@xmath120 from h@xmath14 observations . our own sfr measurement is consistent with @xcite but also shows that most of the star formation happens inside of 150 '' . since our kinematic analysis does not yield reliable results in this region , a direct comparison is not possible . however , the data are still compatible with a scenario in which star formation inside of 150 `` is fueled by the inflow visible in the disk beyond 500 '' . ngc 3198 also rotates counter - clockwise and the south - east part is the nearside . the galaxy has a flat disk without any substantial changes in inclination or position angle outside of 150 ( figure [ ngc3198 ] ) . the radial velocity we measure is always negative outside of 200 " and we find significant inflow between 0.7 and 2 m@xmath120 . the average inflow outside @xmath134 is 1.1m@xmath120 . this agrees quite well with the star formation rate of 0.93 m@xmath120 reported by @xcite . also the radial distribution is roughly consistent . most of the star formation happens between 5 and 15 kpc . in this region we find a substantial modulation of the inflow , probably caused by the star formation process , but outside of this radius we have a roughly constant inflow with an amplitude similar to the star formation rate . therefore the overall trend fits quite well in the picture that star formation is fueled by inflows for this galaxy . if we compare our results for the @xmath51 component to the values obtained by @xcite and @xcite , we find them to be very similar . we can reproduce their results by running a two - step analysis . even if no radial flow is taken into account in the tilted ring fit , the subsequent fourier decomposition shows clear signs of radial inflow with at least 50% of the amplitude found in the simultaneous fit . this is somehow surprising since for ngc 2403 neither @xcite or @xcite or our two - step fit could recover substantial radial inflows . for a detailed discussion of different fitting schemes and validations on mock velocity fields compared to section [ twostepfit ] . clearly some property in the ngc 3198 velocity field is different compare to ngc 2403 . as we show in figure [ ngc3198 ] , the @xmath51 term is relatively small and the terms @xmath70 , @xmath72 and @xmath71 have comparable amplitude . this means , the asymmetry of the velocity field is of comparable amplitude as the overall net inflow . it may be possible that the inflow in this galaxy occurs in such a way that the tilted ring fit can not adjust @xmath25 and @xmath32 to fully compensate for the inflow . indeed , inspection of the hi intensity map ( available from the things web page ) shows an overall asymmetry with an extended north - east arm . ngc 3521 is rotating counter - clockwise and its western part is closer to us . inclination , position angle and rotation velocity are fairly constant . the radial velocity seems to be of low amplitude , as depicted in figure [ ngc3521 ] . the star formation rate reported by @xcite is 2.1 m@xmath120 . no galex fuv image was available in the data archive to derive a sfr profile ( for the sake of consistency and comparability we refrain from constructing this profile based on other sfr tracers ) . we find a rather unusual behaviour for the @xmath70 term , which starts to drop at about 200 and levels off at a value of @xmath135 . the term should represent a strong lopsidedness of the radial velocity or a significant change of the systemic velocity with radius . since the latter seems unreasonable , we conclude that this term represents lopsidedness . this in consequence would imply a strong east - west gradient in the radial velocity field . gas then would enter the galaxy on the western side with roughly @xmath136 , cross the disk and leave again on the eastern side , which altogether seem quite an unlikely scenario too . however , the overall galaxy appearance is indeed quite asymmetric . images at optical wavelength show numerous well pronounced dust lanes and a very extended north - west spiral arm . this arm is also visible in the 21 cm line as extension of the disk . in addition , also a south - east lobe exists which is even more extended . it shows very high fit residuals of up to @xmath137 . this relatively localised feature introduces the peculiar @xmath70 values to our fit . we know that ngc 3521 underwent extensive merger activity in the past . @xcite present deep surface photometry which shows several stellar streams and an extended envelope of stars around the galaxy . also , @xcite found that a substantial amount of gas is not moving within the plane of the galaxy . this clearly affects our fit . given the complex dynamic structure of ngc 3521 and its recent merger history , the results of our fitting procedure need to be considered with caution . the derived fluxes need not necessarily be interpreted as large - scale inward mass motion . deciphering the kinematics of such a disturbed system most likely is beyond the capabilities of the method , as it relies on the assumption of a relatively unperturbed disk with only mildly varying geometry . the analysis for ngc 3621 reveals a relatively simple disk with fairly constant inclination and position angle . this is somewhat surprising , because the galaxy looking clumpy and inhomogeneous in the integrated intensity map . the visual impression could be influenced by the fact , that an extra gas component in the form of a stream or a warp in the line of sight crosses the main disk and creates some foreground confusion @xcite . the galaxy rotates clockwise and the rotation curve is completely flat outside of 200 ( figure [ ngc3621 ] ) . the measured radial velocity is of low amplitude and varies in the range of @xmath138 . @xcite report a high sfr of 2 m@xmath120 . our coarse sfr estimate from galex fuv images is substantially lower ( 0.5 m@xmath120 ) but as mentioned , our aim is to reproduce the radial distribution not the galaxy - integrated amplitude of the star formation activity . figure [ ngc3621 ] therefore shows the profile we measure scaled to the total sfr reported by @xcite . some slight correlations between the radial hi flow and the sfr profile might be visible , in particular at 100 " where a substantial amount of the star formation takes place and the low - level inflow we measure decreases . however , we do not find signs of large - scale inflow and the recent star formation in ngc 3621 seems to consume the existing gas inventory of the galaxy and is at the moment not fueled from outside in substantial amounts . ngc 5055 exhibits a very complex disk geometry , as indicated in figure [ ngc5055 ] . the southern part of the disk is closer to earth and rotation occurs in a clockwise direction . in the integrated hi map as shown by @xcite one can clearly distinguish between an inner star - forming part of the galaxy ( sfr given by @xcite 2.1 m@xmath120 , for the radial distribution , compare to figure [ ngc5055 ] ) and an outer diffuse ring with a noticeably different inclination angle . the kinematic analysis yields an inclination which first rises to above @xmath139 at 380 , then declines to @xmath140 at 650 and then rises again to slightly below @xmath64 . the position angle also changes . the most significant feature is a steep rise by @xmath16 between 500 and 800 . for the radial velocity we find values around zero up to a radius of 600 . outside of this , in the region of the hi ring , it rises to @xmath141 and declines again towards the outermost part of the galaxy . this corresponds to an outward peak mass flow of 6 m@xmath120 . inside of 600 , no significant mass flow is detected . the asymmetry terms show quite high and varying values . even if the rise of inclination outside of 600 is quite steep , it is only about 50% of the critical change rate given in equation [ i_crit ] . therefore , the measured @xmath51 term should be robust and we consider the hi mass flow to be real . from dust lanes we can infer that the galaxy rotates clockwise and the radial flow is directed outward . @xcite report the detection of a stellar stream in the outskirts of ngc 5055 via deep optical surface photometry . the stream is inclined with respect to the disk , but not associated with any hi structure . ngc 5055 is part of the m51-group and accompanied by dwarf galaxies are known @xcite . the detection of the stream is a clear evidence for a recent merger event . from the geometry of the stream presented by @xcite one could estimate that it passed through the disk of ngc 5055 at a radius of roughly 600 . this matches strikingly well with the minimum of the inclination , and it is also well within the regime of the steep change in position angle . it seems that the merger has perturbed the disk of ngc 5055 and left an imprint in the hi kinematics . even if the exact dynamics are unknown , it is very likely that the complex disk geometry we find is the direct result of this recent merger event . there is also a dark dust filament visible in front of the southern half of the stellar disk . it has a quite luminous hi counterpart which is easy to identify in the integrated intensity map . from the visual impression of the velocity map , the filament seems to move out of the galactic plane and at a different orbital velocity than the rest of the disk . this probably causes the bump in the inclination at around 350 since the fit can not differentiate between the disk velocity and the motion of the filament and fits an average along the line - of - sight . for ngc 6946 we derive a fairly low inclination of @xmath142 with substantial scatter from ring to ring . similar results are reported by @xcite . the low and varying inclination makes the fit quite uncertain since the majority of parameters are scaled by @xmath37 . the fit only converges if we choose quite wide rings ( increase in 30 steps ) and we consider it quite likely that the actual uncertainties are larger than the formal error bars shown in figure [ ngc6946 ] . we determine a moderately varying @xmath32 and clockwise rotation and find the nearside of the disk to be the north - west part . the inward radial velocities are very high and in excess of @xmath143 , which is roughly one third of the rotational velocity . also , the radial velocity field shows a quite substantial lopsidedness with a difference between north - west and south - east edge of around @xmath136 . even if we do not take the two rings at 450 and 480 into account , ngc 6946 shows an extremely high hi mass inflow of 10 to 15 m@xmath120 in the range from 400 to 650 . our fit results are stable and the residuals show no obvious feature that would indicate problems with the fit beyond the large statistical errors . nevertheless , we express some concern that the derived magnitude of the inflow could be larger than what is physically reasonable and may be exaggerated due to the low inclination of the galaxy . however , ngc 6946 does exhibit exceptional properties . the star formation rate is relatively large with with a value of 3.3 m@xmath120 @xcite . the gas - to - stellar mass ratio is also high . for the atomic gas , a value of 20% is reported . for the molecular gas , the number is 12% , which is the largest value in the entire things sample . from the galex fuv images we find that a substantial part of the total star formation takes place inside of @xmath144 , well inside the region where we detect the large inflows . the amplitudes of inflow and sfr roughly match , however the uncertainty in the inflow measurements prohibits a detailed comparison . the general picture we get for ngc 6946 fits the simulations by @xcite very well . they ran eight cosmological hydrodynamical zoom simulations of milky way - sized halos using the moving mesh code arepo to study the evolution of disk galaxies . they usually find quite fast inflow with radial velocities nearly matching the circular velocity in the outer part of the disk . the gas in the simulated disks is supported by rotation only up to the edge of the luminous disk and outside of this it is highly sub - keplerian . although these high radial velocities fit qualitatively to our findings for ngc 6946 , their simulations generally predict too small hi disks compared to other spiral galaxies . the authors explain that their objects include far less gas than would be realistic and we assume that this could affect the size and dynamics of the simulated gas disks . if the magnitude of the derived inflow is indeed correct , the inflow rates would be uniquely high in our sample . however , given the low inclination of ngc 6946 and the large ring - to - ring variations in the fit , we may expect these inflow rates to be upper limits and presumably should not be taken at face value . ngc 7331 is rotating clockwise and the western part of the disk is closer to earth . our analysis yields a fairly constant inclination and position angle as well as rotation velocity ( figure [ ngc7331 ] ) . we find some inflow of hi at the level of 1 m@xmath120 outside of @xmath134 and outflow inside of it . the radial velocity is close to zero in the inner part and negative at around @xmath145 in the outer part of the disk . the asymmetry terms are of similar magnitude . in the reconstructed radial velocity field one sees that inflow predominantly happens in the northern part of the galaxy . in the southern half , the radial velocity does not substantially deviate from zero . the integrated hi map shows a fairly extended lobe on the northern end of the galaxy and a small on in the south . one possible explanation could be that a relatively localised gas stream flows inward from the circumgalactic medium and reaches the northern part of the galaxy . unfortunately , the high inclination of ngc 7331 makes it difficult to study the structure of the northern lobe in more detail , and the observations are not sensitive enough to trace the hi gas further out . the leroy sfr for ngc 7331 is nearly as high as for ngc 6946 , but the measured hi inflow rate of about 1 m@xmath120 is more than an order of magnitude lower . however , all star formation takes place inside of 25 kpc ( compare to figure [ ngc7331 ] ) and we find basically constant inflow outside of 15 kpc . despite the mismatch in amplitude this gives a roughly consistent picture of mass transport and star formation . for ngc 925 , we find that the galaxy is most likely rotating clockwise and that the northern part of the disk is closest to us . there is a nearly linear drop of inclination from above @xmath146 in the centre to @xmath140 in the outskirts . at the same time the rotation velocity shows a nearly linear rise out to 250 and then saturates at @xmath147km@xmath0s@xmath4 ( figure [ ngc0925 ] ) . optical images clearly show a bar in the galaxy centre . as described in section [ solid_body ] , solid body rotation makes it impossible to disentangle inclination and rotation , and therefore the inferred disk geometry inside of 250 may be incorrect . outside of this radius , however , our fit results should be robust . the amplitude of the radial velocity is quite high . outside of 300 it rises to a maximum of @xmath141 . this results in an hi outflow of nearly 5 m@xmath120 at around 300 . the integrated hi map clearly shows two extended spiral arms , probably related to interactions or accretion events . @xcite report the detection of an extended hi envelope with complicated streams caused by a minor merger in the halogas survey . they even find a very faint optical counterpart which could be the remnant of a gas - rich dwarf galaxy that was accreted by ngc 925 . the fact that we derive outward instead of inward motion is surprising , but the overall picture slightly resembles that of ngc 2841 ( section [ secngc2841 ] ) . in this kinematic study we analysed the hi velocity field of ten disk galaxies from the things survey and investigated the presence of radial inflows , which might supply the star forming regions in the inner parts with fresh gas and thereby allow continued star formation . given that the gas depletion timescale in most spiral galaxies is short and of the order of @xmath148 gyr it is a long standing problem of galaxy evolution to explain how star formation activity is sustained over cosmological timescales . if we discard the possibility that most galaxies are observed on the verge of running out of gas , then they must be continuously fed by fresh material . our aim is to search for signatures of this process . we started our investigation of the dynamics of hi gas in nearby disk galaxies with the analysis of existing tilted ring fitting methods and the assessment of their statistical reliability . we found that most two - step fitting schemes are unable to recover the full radial inflow in the velocity field and as a consequence developed a new fourier decomposition scheme that includes second order terms of rotational and radial velocity and simultaneously determines the position angle and inclination angle as functions of radius . by applying it to mock velocity data we demonstrated that the new method is more stable and less susceptible to systematic errors . it is well suited to reliably detect radial flows of material through the disk for galaxies that show no strong warps and no sudden changes of position angle . the method works well for galaxies with intermediate inclination angles and smooth and unperturbed disks , but gives less reliable results for galaxies with complicated kinematics and geometries , for instance if they have recently gone through a merger event that disturbed the disk . we find clear evidence for radial inflow in ngc 2403 with an amplitude of @xmath149 km s@xmath4 corresponding to an hi mass flow rate of 3 m@xmath120 . substantial inwards mass flux through the outer disk is also detected in ngc 3198 with values up to 2 m@xmath120 , and to a lower degree in ngc 2903 and ngc 7331 with values around and below 1 m@xmath120 . for ngc 6946 we infer a surprisingly large inflow rate of 15 m@xmath120 , but this measurement is affected by the low inclination of the galaxy and is likely to be overestimated . we compare our radial mass flux profiles to star formation rate profiles based on galex fuv imaging and to sfr measurements from the literature . we find that the rate at with gas gets transported through the disk is sufficient to explain the current star formation activity and if this is indeed representative of the long - term average , than the accretion of new material onto the outer disk of these galaxies and its subsequent inwards transport through the disk is sufficient to maintain a steady state over secular timescales . however , a detailed comparison of the radial distribution of star formation and mass inflow usually deviates substantially from the naive assumption of a stationary state with a one - to - one relation between inflow and star formation across galaxy disks . this is indicative of additional mass transport processes acting within the galaxies , for example gas ejection by stellar winds , often referred to as galactic fountains . for ngc 5055 as well as for ngc 2841 and ngc 925 we find that hi gas flows outwards in the outer disk . in ngc 5055 , there is evidence of a recent encounter with a dwarf galaxy . we think that this is the cause for its perturbed disk geometry , and speculate that the angular momentum delivered to the disk by this encounter drives the inferred outward motion . indications of a recent merger with a gas - rich dwarf galaxy are also reported for ngc 925 . we therefore consider the encounter with a dwarf galaxy as possible explanation for the peculiar properties of all three objects . nearly all the galaxies in our sample show some degree of asymmetry like lobes , streams , or other types of disk perturbations . complex kinematic structures thus appear quite common among large spiral galaxies and unperturbed disks are likely to be an exception . to properly model the velocity fields of the observed hi disks , in particular of their outer regions , where these perturbations seem more pronounced , the employed fitting has to be able to adapt to these asymmetries . it would therefore be highly desirable to include as many fourier components as possible in the simultaneous fit of the hi velocity field , which is a difficult task : the components of the fourier decomposition of the field are highly degenerate and going beyond the second order is not practical . clearly , more sensitive observations that cover larger areas especially of the out disk will help to improve our ability to study the hi velocity field , however , the general limitations of the purely kinematic analysis method remains . doing full justice to the complex velocity patterns observed in these disk galaxies remains a challenging task . we conclude , that there are indeed radial inflows in the disks of galaxies present which can easily be of sufficient magnitude to sustain star formation in the inner parts of galactic disks . however , the situation appears more complex than simply linking the inferred inflow rates to measured star formation rates : the inflow rates are often up to an order of magnitude larger than the sfr ( to some degree a detection bias , since an inflow rate corresponding to the average sfr of @xmath150 in our sample would only be barely detectable with our method ) and in every case highly variable throughout the disk . there are a variety of possible explanations like episodic gas accretion , probably related to minor merger events , accretion directly to intermediate radii of the disks or a substantial contribution from internal cycling of gas suggested by the galactic fountain picture . without dedicated simulations aiming at a precise modelling of gas kinematics including physical feedback schemes , it will probably be impossible to decide in favour of a particular model . while our study based on kinematic data alone could show the existence of radial inflows in disk galaxies , any further investigation will probably not only need superior ( deep , extended and good resolution ) observational data but also a much broader approach supported by simulations . we thank milan den heijer , juergen kerp and shahram faridani for supplying effelsberg single dish data and stimulating discussions . this work made extensive use of the _ the hi nearby galaxy survey _ ( things ) database @xcite , as well as the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . fb acknowledges support from dfg grant bi 1546/1 - 1 . rsk thanks the deutsche forschungsgemeinschaft ( dfg ) for funding via the sfb 881 _ the milky way system _ ( subprojects b1 , b2 , and b8 ) as well as via the spp 1573 _ the physics of the interstellar medium_. rsk furthermore acknowledges support from the european research council under the european community s seventh framework programme ( fp7/2007- 2013 ) via the erc advanced grant starlight ( project number 339177 ) . wjgdb was supported by the european commission ( grant fp7-people-2012-cig # 333939 ) . genzel r. , tacconi l. j. , gracia - carpio j. , sternberg , a. , cooper m. c. , shapiro k. , bolatto a. , bouche n. , bournaud f. , burkert a. , combes f. , comerford j. , cox p. , davis m. , schreiber n. m. frster , garcia - burillo s. , lutz d. , naab t. , neri r. , omont a. , shapley a. , weiner b. , 2010 , mnras , 407 , 2091 leroy , a. k. , walter , f. , sandstrom , k. , schruba , a. , munoz - mateos , j .- c . , bigiel , f. , bolatto , a. , brinks , e. , de blok , w. j. g. , meidt , s. , rix , h .- w . , rosolowsky , e. , schinnerer , e. , schuster , k .- f . , usero , a. , 2013 , aj , 146 , 19 martinez - delgado d. , gabany r. j. , crawford k. , zibetti s. , majewski s. r. , rix h .- w . , fliri j. , carballo - bello j. a. , bardalez - gagliuffi d. c. , penarrubia j. , chonis t. s. , madore b. , trujillo i. , schirmer m. , mcdavid d. a. , 2010 , astron . j. , 140 , 962 saintonge , a. , kauffmann , g. , wang , j. , kramer , c. , tacconi , l. j. , buchbender , c. , catinella , b. , graci - carpio , j. , cortese , l. , fabello , s. , fu , j. , genzel , r. , giovanelli , r. , guo , q. , haynes , m. p. , heckman , t.m . , krumholz , m. r. , lemonias , j. , li , c. , moran , s. , rodriguez - fernandez , n. , schiminovich , d. , schuster , k. , sievers , a. , 2011 , mnras , 415 , 61 salim s. , rich r. m. , charlot s. , brinchmann j. , johnson b. d. , schiminovich d. , seibert , m. , mallery r. , heckman t. m. , forster k. , friedman p. g. , martin d. c. , morrissey p. , neff s. g. , small t. , wyder t. k. , bianchi l. , donas j. , lee , y .- w , madore b. f. , milliard b. , szalay a.s . , welsh b. y. , yi s. k. , 2007 , apj , 173 , 267
the study of 21@xmath0 cm line observations of atomic hydrogen allows detailed insight into the kinematics of spiral galaxies . we use sensitive high - resolution vla data from the hi nearby galaxy survey ( things ) to search for radial gas flows primarily in the outer parts ( up to @xmath1 ) of ten nearby spiral galaxies . inflows are expected to replenish the gas reservoir and fuel star formation under the assumption that galaxies evolve approximately in steady state . we carry out a detailed investigation of existing tilted ring fitting schemes and discover systematics that can hamper their ability to detect signatures of radial flows . we develop a new fourier decomposition scheme that fits for rotational and radial velocities and simultaneously determines position angle and inclination as a function of radius . using synthetic velocity fields we show that our novel fitting scheme is less prone to such systematic errors and that it is well suited to detect radial inflows in disks . we apply our fitting scheme to ten things galaxies and find clear indications of , at least partly previously unidentified , radial gas flows , in particular for ngc 2403 and ngc 3198 and to a lesser degree for ngc 7331 , ngc 2903 and ngc 6946 . the mass flow rates are of the same order but usually larger than the star formation rates . at least for these galaxies a scenario in which continuous mass accretion feeds star formation seems plausible . the other galaxies show a more complicated picture with either no clear inflow , outward motions or complex kinematic signatures . [ firstpage ] galaxies : kinematics and dynamics galaxies : structure radio lines : galaxies
one of the central ideas of quantum mechanics is the uncertainty principle which was first proposed by heisenberg @xcite for two conjugate observables . indeed , it forms one of the most significant examples showing that quantum mechanics does differ fundamentally from the classical world . uncertainty relations today are probably best known in the form given by robertson @xcite , who extended heisenberg s result to two arbitrary observables @xmath3 and @xmath4 . robertson s relation states that if we prepare many copies of the state @xmath5 , and measure each copy individually using either @xmath3 or @xmath4 , we have @xmath6{|\psi\rangle}|\end{aligned}\ ] ] where @xmath7 for @xmath8 is the standard deviation resulting from measuring @xmath5 with observable @xmath9 . the essence of is that quantum mechanics does not allow us to simultaneously specify definite outcomes for two non - commuting observables when measuring the same state . the largest possible lower bound in robertson s inequality ( [ eq : heisenberg ] ) is @xmath10 , which happens if and only if @xmath3 and @xmath4 are related by a fourier transform , that is , they are conjugate observables . a natural measure that captures the relations among the probability distributions over the outcomes for each observable is the entropy of such distributions . this prompted hirschmann to propose the first entropic uncertainty relation for position and momentum observables @xcite . this relation was later improved by @xcite , where @xcite show that heisenberg s uncertainty relation is in fact implied by this entropic uncertainty relation . hence , using entropic quantities provides us with a much more general way of quantifying uncertainty . indeed , it was realized by deutsch @xcite that other means of quantifying `` uncertainty '' are also desirable for another reason : note that the lower bound in is trivial when @xmath5 happens to give zero expectation on @xmath19 $ ] . hence , it would be useful to have a way of measuring `` incompatibility '' which depends only on the measurements @xmath3 and @xmath4 and not on the state . deutsch @xcite himself showed that @xmath20 where @xmath21 , and @xmath22 is the min - entropy arising from measuring the pure state @xmath5 using the basis @xmath14 ( see section [ sec : urquantities ] for more information on the entropic quantities we use ) . if @xmath14 and @xmath15 are related by a fourier transform , then the r.h.s . of becomes @xmath23 , where the minimum is achieved by a state that is invariant under the fourier transform . since the shannon entropy obeys @xmath24 , deutsch s bound also holds for the shannon entropy . better lower bounds have since been obtained for the shannon entropy by maassen and uffink @xcite following a conjecture of kraus @xcite . their uncertainty relations are again strongest ( in the sense that the lower bound is largest ) when the bases @xmath14 and @xmath15 are conjugate , that is , the two bases are related by a fourier transform . apart from their role in understanding the foundations of quantum mechanics , these uncertainty relations play a central role in cryptography in the noisy - storage model @xcite , quantum key distribution @xcite , information locking @xcite , and the question of separability @xcite . in particular , such relations have practical interest in noisy - storage cryptography @xcite where they may enable us to prove security for a larger class quantum memories . here , we are concerned with measurements in multiple bases @xmath25 . entropic uncertainty relations provide a natural way of quantifying `` incompatibility '' of more than two measurements by lower bounding @xmath26 for all states @xmath5 . we call a state @xmath5 that minimizes the average sum of entropies a _ maximally certain state_. when @xmath27 is the shannon entropy , the largest bound we can hope to obtain for any choice of bases is @xmath28 since choosing @xmath5 to be an element of one of the bases yields zero entropy when we subsequently measure in the same basis . if is indeed a lower bound to , we will call the measurements _ maximally incompatible with respect to the shannon entropy_. note that this can only happen if @xmath5 gives us full entropy can never exceed @xmath29 , where the maximum is attained when the distribution over the outcomes is uniform @xmath30 . ] when measured in any other basis , that is , the bases are all mutually unbiased . curiously , however , it was shown that whereas being mutually unbiased is necessary , it is not a sufficient condition to obtain maximally strong uncertainty relations for the shannon entropy @xcite . in particular , there do exist large sets of up to @xmath31 mutually unbiased bases in square dimensions for which we do obtain very weak uncertainty relations @xcite . recently , ambainis @xcite has shown that for any three bases from the `` standard '' mutually unbiased bases construction @xcite in prime dimension , the lower bound can not exceed @xmath32 , for large dimensions . for dimensions of the form @xmath33 and @xmath34 no further assumption is needed , but the proof assumes the generalized riemann hypothesis for dimensions of the form @xmath35 . furthermore , for any @xmath36 , there always exist @xmath37 of these bases such that the lower bound can not be larger than @xmath38 . only if we use the maximal set of @xmath39 mutually unbiased bases that can be found for any given prime power dimension , do we obtain quite strong uncertainty relations @xcite . at present , we merely know that there do exist arbitrarily large sets of two outcome measurements that give us maximally strong uncertainty relations @xcite , and that in larger dimensions selecting a large amount of bases at random does provide us with strong relations @xcite ( for a survey see @xcite ) . indeed , it remains an intriguing open question as to whether there even exist three measurements with three outcomes in dimension @xmath40 that are maximally incompatible with respect to the shannon entropy . in the light of these questions , it is therefore natural to study the structure of mutually unbiased bases(mubs ) to see whether we can identify additional properties which are sufficient for obtaining strong uncertainty relations . in @xcite , wootters and sussman made the interesting observation that for the maximal set of @xmath39 mutually unbiased bases coming from such constructions as @xcite in dimension @xmath1 , the lower bound of the entropic uncertainty relation in terms of the collision entropy given in @xcite is tight , and the minimum is attained by a state that is invariant under a unitary that cyclically permutes the set of all @xmath39 mubs . a similar unitary was noted to exist by chau @xcite . wootters and sussman derive their transformation from phase space arguments . their unitary can in fact easily be generalized to cyclically permute @xmath41 bases , whenever @xmath41 divides @xmath39 ( see section [ sec : urdwggeneral ] ) . the results in @xcite have recently been generalized by appleby @xcite , who shows that in prime power dimensions of the form @xmath42 , there exists a unitary operation that cyclically permutes the first and second halves of the full set of mubs . this raises the pressing question of whether smaller sets of mubs also exhibit such symmetries ? and can we exploit such symmetries to obtain tight uncertainty relations ? in particular , is the minimizing state always an invariant of such a transformation as observed for two bases in ? * main result * we first show by an explicit construction that there exist sets of @xmath43 mutually unbiased bases in dimension @xmath1 with the property that there exists a unitary that cyclically permutes all bases in this set , whenever ( a ) @xmath41 is prime , and ( b ) @xmath41 divides @xmath44 or @xmath45 . more specifically , we provide an explicit construction of mubs @xmath25 with @xmath46 and a unitary @xmath47 such that @xmath48 furthermore , in dimension @xmath2 , we actually find such a unitary for any set of @xmath41 mubs , where @xmath49 . our approach exploits properties of the clifford algebra , which might yield new insights into the structure of these mubs . it is entirely distinct from the phase space approach which was used to construct such a unitary for the full set of @xmath39 mubs @xcite . note that our construction gives at most @xmath50 bases , but shows that there is indeed an additional symmetry which has previously gone unnoticed . for @xmath51 bases , @xmath47 is simply the fourier transform , and it would be interesting to investigate general properties of our transformation and whether it has applications in other areas . we then apply our transformation to the study of uncertainty relations in terms of the _ min - entropy _ ( see ) . since @xmath24 , this also provides us with bounds on uncertainty relations in terms of the shannon entropy . of course , many forms of entropy could be considered when it comes to quantifying uncertainty , and each has its merits . the min - entropy is of particular interest in cryptography , and is also related to the well studied extrema of the discrete wigner function as we will discuss in section [ sec : urdiscretewigner ] . in particular , it will be easy to see that the average min - entropy for the full set of @xmath52 mubs can be bounded as @xmath53\ , \end{aligned}\ ] ] where @xmath54 is the maximum value of the discrete wigner function at the point @xmath55 in discrete phase space . symmetries thereby play an important role in determining @xmath56 . * second result * we prove a simple min - entropic uncertainty relation for an arbitrary set of @xmath41 mutually unbiased bases . for mubs @xmath25 we obtain @xmath57 . \end{aligned}\ ] ] for the case of @xmath58 mubs in dimension @xmath13 , this bound is indeed the same as the bound in . for any small set of @xmath59 mubs , our bound matches the strongest known bound @xcite . we also prove the following alternate lower bound : @xmath60,\end{aligned}\ ] ] which is stronger than for the complete set of @xmath61 mubs in dimension @xmath13 . clearly , when @xmath62 , the two bounds are equivalent . we further show that is in fact tight for @xmath63 mubs in dimension @xmath2 stemming from our construction , where the minimum is attained for an invariant state of the transformation @xmath47 that cyclically permutes all @xmath64 bases . even though this is a somewhat restricted statement , it is the first time that a tight entropic uncertainty relation has been obtained for this case . the minimizing state here has an appealing symmetry property , just as for the case of @xmath51 bases in where the minimum is attained by a state that is invariant under the fourier transform . for the collision entropy @xmath65 , wootters @xcite previously showed that the lower bound from @xcite is attained by an invariant state when considering the full set of @xmath39 mubs . here , however , we exhibit a tight uncertainty relation for these @xmath66 bases in @xmath2 for the collision entropy @xmath65 which has an entirely different structure and the minimum is not attained by an invariant state of our transformation . nevertheless , we have for the first time a _ tight _ entropic uncertainty relation for _ all _ possible mubs in a dimension larger than the trivial case of @xmath67 where the bloch sphere representation makes the problem easily accessible . in @xmath2 , we have a tight relation for @xmath68 for @xmath69 , and tight relations for @xmath65 for @xmath70 . our result indicates that due to the different properties of the minimizing state for different numbers of bases , the problem may be even more daunting than previously imagined . yet , our work shows that in each case the minimizing state is by no means arbitrary . it has a well defined ( albeit different ) structure in each of the cases . * third result * for some sets of mubs we do obtain for the first time , significant insight into the structure of the maximally certain states . in particular , we note in section [ sec : symmetry ] that for @xmath41 mutually unbiased bases that the state that minimizes the min - entropic uncertainty relations is an invariant of a certain unitary whenever @xmath41 divides @xmath39 for @xmath1 . before explaining our construction of mutually unbiased bases for which there exists a unitary that cyclically permutes them , let us define the notions of mubs more formally and recall some known facts . let @xmath71 and @xmath72 be two orthonormal bases in @xmath73 . they are said to be _ mutually unbiased _ if @xmath74 , for all @xmath75 . a set @xmath76 of orthonormal bases in @xmath73 is called a _ set of mutually unbiased bases _ if each pair of bases is mutually unbiased . for example , the well - known computational and hadamard basis are mutually unbiased . we use @xmath77 to denote the maximal number of mubs in dimension @xmath13 . in any dimension @xmath13 , we have that @xmath78 @xcite . if @xmath79 is a prime power , we have @xmath80 and explicit constructions are known @xcite . other constructions are known that give less than @xmath39 mubs in other dimensions @xcite . however , it is still an open problem whether there exists a set of @xmath81 ( or even @xmath64 ! ) mubs in dimension @xmath82 . our construction of mutually unbiased bases makes essential use of the techniques developed in @xcite , together with properties of the clifford algebra . the clifford algebra is the associative algebra generated by operators @xmath83 satisfying @xmath84 and @xmath85 . it has a unique representation by hermitian matrices on @xmath44 qubits ( up to unitary equivalence ) that can be obtained via the famous jordan - wigner transformation @xcite : @xmath86 for @xmath87 , where we use @xmath9 , @xmath88 and @xmath89 to denote the pauli matrices . furthermore , we let @xmath90 note that in dimension @xmath67 these are just the familiar pauli matrices , @xmath91 , @xmath92 and @xmath93 . of particular importance to us will be the fact that we can view the operators @xmath83 , as @xmath94 orthogonal vectors forming a basis for @xmath95 . in particular , for any orthonormal transformation @xmath96 which when applied to the vector @xmath97 gives @xmath98 , there exists a unitary @xmath99 such that @xmath100 the orthonormal transformation that is particularly interesting to us here is the one that cyclically permutes the basis vectors . as described above we can find a corresponding unitary @xmath101 which cyclically permutes the basis vectors @xmath102 . an explicit construction can be found in the appendix . this symmetry can be extended to @xmath103 , see e.g. @xcite . it will also be useful that the set of @xmath104 operators @xmath105 forms an orthogonal basis for @xmath106 hermitian matrices in @xmath1 @xcite . to construct mutually unbiased bases , we follow the procedure outlined in @xcite , but now applied to a subset of the operators in @xmath107 . that is , we will group operators into classes of commuting operators , i.e. , sets @xmath108 of size @xmath109 such that 1 . the elements of @xmath110 commute for all @xmath111 , 2 . @xmath112 for all @xmath113 . it has been shown in @xcite that the common eigenbases of such classes form a set of @xmath41 mubs . first of all , note that no class can contain two generators @xmath114 and @xmath115 since they do not commute . when forming the classes we hence ensure that each one contains exactly one generator @xmath114 , which clearly limits us to constructing at most @xmath0 such classes . the difficulty in obtaining a partitioning that is suitable for our purpose is to ensure that the unitary @xmath47 that cyclically permutes the generators @xmath116 also permutes the corresponding bases by permuting products of operators appropriately . we show in the appendix that our general construction achieves the following : suppose that @xmath43 is prime , and either @xmath41 divides @xmath44 or @xmath45 . then in dimension @xmath117 , there exist @xmath41 mutually unbiased bases @xmath25 for which there exists a unitary @xmath47 that cyclically permutes them @xmath118 let us consider two simple examples of such classes in dimension @xmath2 . these are not obtained from our general construction , but nevertheless provide us with the necessary intuition . for @xmath66 mubs the classes are given by @xmath119 it is easy to see that the unitary @xmath47 that achieves the transformation @xmath120 , but leaves @xmath121 and @xmath122 invariant , cyclically permutes the bases given above . for the collision entropy @xmath65 the minimum is attained for an eigenstate of the _ commuting _ operators @xmath123 , @xmath124 and @xmath125 . this minimizing state also shows that the wellknown bound for @xmath65 ( see e.g. @xcite ) can be _ tight_. for @xmath63 mubs we obtain the classes @xmath126 it is easy to see that the unitary @xmath47 that achieves the transformation @xmath127 , but leaves @xmath128 invariant , cyclically permutes the bases given above . for @xmath63 classes the minimum in the entropic uncertainty relation for @xmath68 is attained for a state that is invariant under the transformation @xmath47 . however , we also know that for @xmath63 or @xmath129 classes in dimension @xmath130 no partitioning of operators is possible that satisfies our requirements . thus , for what values of @xmath41 and @xmath13 such a unitary can be found , remains an interesting open question . we now investigate the relationship between the observed symmetries and entropic uncertainty relations . before comparing different uncertainty relations , we first provide a short introduction to all the entropic quantities we will use . the expert reader may safely skip this section . in general , the _ rnyi entropy _ @xcite of order @xmath55 of the distribution obtained by measuring a state @xmath5 in the basis @xmath12 is given by @xmath131.\end{aligned}\ ] ] indeed , the shannon entropy forms a special case of the rnyi entropy by taking the limit @xmath132 , i.e. , @xmath133 , where we omit the subscript . of particular importance are the _ min - entropy _ , for @xmath134:[def : minentropy ] @xmath135 and the _ collision entropy_[def : collision ] @xmath136 we have @xmath137 and hence uncertainty relations for @xmath138 also provide us with a bound on uncertainty relations for @xmath139 whenever @xmath140 . note that intuitively , the min - entropy is determined by the highest peak in the distribution and most closely captures the notion of `` guessing '' . to see why it is a more useful quantity in cryptography than the shannon entropy , consider the following example distribution @xmath141 : let @xmath142 and let @xmath143 be the all 0 string . suppose that @xmath144 and @xmath145 for @xmath146 , i.e. , with probability @xmath147 we choose @xmath148 and with probability @xmath147 we choose one string uniformly at random . then @xmath149 , whereas @xmath150 ! if @xmath151 would correspond to an encryption key used to encrypt an @xmath44 bit message , we would certainly not talk about security if we can guess the key with probability at least @xmath147 ! yet , the shannon entropy is quite high . apart from its cryptographic applications , min - entropic uncertainty relations are appealing since the problem of determining tight uncertainty relations can be simplified considerably in the presence of symmetries . furthermore , these relations bear an interesting relation to the extrema of the discrete wigner function . first of all , note that for the min - entropy we have by jensen s inequality that @xmath152 where the inequality becomes equality if all terms @xmath153 are the same . for @xmath154 , define @xmath155 note that determining a tight lower bound to is thus equivalent to determining @xmath156 clearly , any @xmath157 such that @xmath158 thus gives us a lower bound for . for any set of bases , this makes the problem of finding a bound more approachable as it reduces the problem to finding the largest eigenvalue for any operator @xmath159 . in particular , it can be phrased as a semidefinite program to minimize @xmath157 such that holds for all @xmath160 . it is now easy to see why symmetries simplify our goal of determining tight uncertainty relations . [ lem : symmetry ] suppose that for every @xmath161 there exists a unitary @xmath162 such that @xmath163 . then there exists a @xmath164 such that the minimum in is attained for a state @xmath165 that is invariant under @xmath166 . first of all , note that @xmath167 and hence for @xmath168 @xmath169 in particular , this holds for the state @xmath170 corresponding to the eigenvector @xmath5 with the largest eigenvalue of @xmath171 . when looking for the minimizing state on the r.h.s of we can thus restrict ourselves to states which are invariant under @xmath172 . note that in this case , we furthermore have that @xmath173 meaning that the inequality is tight in case of such a symmetry which is our claim . the question of course remains , whether such unitaries do exist in general . wootters and sussman @xcite have shown that there exists a unitary @xmath47 that cyclically permutes the set of all @xmath39 mubs for @xmath1 by constructing a unitary that corresponds to a rotation around the origin in phase space . clearly , by considering the unitary @xmath174 one can trivially adapt their construction to obtain a unitary that cyclically permutes @xmath41 mubs whenever @xmath175 . by first translating any point in the phase space to the origin , then applying the transformation @xmath174 and finally translating the origin back to the original point , one can obtain the desired unitaries @xmath162 that enable us to find tight bounds for the min - entropic uncertainty relations . this is the first time we gain significant insight into the structure of the states that minimize . note that our construction only gives unitaries @xmath162 for @xmath176 for any @xmath177 . this means that our complementarity transform @xmath47 , leads to tight bounds only if the largest eigenvalue of any @xmath159 happens to occur for a @xmath160 of this form . this is for example the case for @xmath63 in @xmath2 , where we do not obtain a unitary from the phase space approach of @xcite . [ sec : urdwggeneral ] to see how finding a lower bound for min - entropic uncertainty relations for @xmath39 mubs relates to finding the extrema of the discrete wigner function , let us first recall the properties of the discrete wigner function . the discrete phase space is a two - dimensional vector space over a finite field @xmath178 , where here we focus on the case of @xmath1 . for every state @xmath165 , we can associate a function @xmath179 with every point @xmath55 in the discrete phase space , known as the discrete wigner function . for completeness , we provide a short summary on how to determine @xmath179 ; a detailed account can be found in @xcite . first of all , note that the @xmath180 points of the discrete phase space can be parititioned into @xmath13 parallel lines each of which contains @xmath13 points . any such partition is called a _ striation _ , and it is known that @xmath39 such striations can be found @xcite . one may now define the discrete wigner function by relating each striation to one of the @xmath39 possible mutually unbiased bases @xcite : let @xmath181 denote the @xmath182-th line in the striation @xmath183 . with each such line , we associate a projector @xmath184 onto the @xmath182-th element of the basis @xmath185 , in a specific order so as to satisfy certain symmetry constraints @xcite . defining the phase - space point operator @xmath186 one can now define the discrete wigner function as @xmath187 the _ extrema of the discrete wigner function _ are defined as the minimum and maximum of over quantum states @xmath165 . note that when considering @xmath188 mutually unbiased bases , each point @xmath55 in the discrete phase space can be contained in exactly one line from each basis , as all lines in a striation , i.e. , one basis are parallel . hence , there is a one - to - one correspondence between points @xmath55 in discrete phase space and vectors @xmath189 . in terms of the phase space operator this means that @xmath190 note that the maximum of the discrete wigner function @xmath191 is simply the largest eigenvalue of @xmath192 ( or @xmath193 ) up to a factor of @xmath30 . we thus have that @xmath194\ , \end{aligned}\ ] ] satisfies @xmath195 and the maximum of the discrete wigner function provides a lower bound to the min - entropic uncertainty relations as given in . the extrema @xmath54 were evaluated numerically in @xcite for small @xmath13 . note , however , that as noted in section [ sec : symmetry ] , one may use symmetries to solve the problem of determining @xmath54 directly . as mentioned in section [ sec : urmin - entropy ] , the problem of finding a lower bound for the average min - entropy reduces to the problem of finding the maximum eigenvalue of the operator @xmath159 defined in . in the appendix , we use a result due to schaffner @xcite obtained using the techniques of kittaneh @xcite , to show that for any set of @xmath41 mutually unbiased bases in dimension @xmath13 , the maximum eigenvalue of @xmath159 is bounded by @xmath196 using this , we obtain the following simple bound for the average min - entropy in the appendix [ lem : smalllbound ] let @xmath25 be a set of mutually unbiased bases in dimension @xmath13.then , @xmath197\ .\ ] ] for the case of @xmath51 mubs in dimension @xmath13 , our bound exactly matches the well known result of deutsch ( see ) . for @xmath198 , the only other known lower bound for the average min - entropy is the one obtained in @xcite , where it is shown that for a set of @xmath199 mubs in dimension @xmath200 , the following holds : @xmath201\end{aligned}\ ] ] where @xmath202 are subsets of @xmath44-bit strings . in the case of min - entropic uncertainty relations , these subsets contain only a single string , which corresponds to the peak of the probability distribution induced on the state @xmath5 by the corresponding bases @xmath203 and @xmath204 , so that , @xmath205 thus , in dimension @xmath200 , our bound in is clearly the same as . the reason we obtain a more general bound for @xmath206 mubs in any dimension @xmath13 is that we reduce the problem directly to an eigenvalue problem without going through the representation in terms of bit strings as in @xcite , via a much simpler argument using only the concavity of the @xmath207 . using an alternate approach involving a bloch sphere like representation of the basis vectors @xmath208 , we show that the maximum eigenvalue of @xmath159 can be bound differently , as follows : @xmath209 as we show in the appendix , this implies [ lem : largelbound ] let @xmath25 be a set of mutually unbiased bases in dimension @xmath13.then , @xmath210 \ .\ ] ] notice that this alternate bound on the min - entropy is stronger than when @xmath211 . in particular , for the complete set of @xmath39 mubs in dimension @xmath13 , this alternate bound in is stronger than any previously known bounds . when @xmath62 the two bounds that we derive are indeed equivalent . see figs . [ fig : dim4 ] and [ fig : dim8 ] for a comparison of our bounds in dimensions @xmath2 and @xmath130 respectively . average min - entropy for different sets of mubs in dimension @xmath2 . + the crosses denote numerically computed minima of the average min - entropy for mubs obtained using our construction . the bound in is clearly tight for both @xmath66 and @xmath63 mubs . the second analytical bound in is stronger than for @xmath212 bases . the circle denotes the average min - entropy for the invariant states given in . for @xmath64 mubs in @xmath2 the minimum of the average min - entropy is indeed attained by states invariant under @xmath47.,scaledwidth=52.0% ] finally , we provide an example of a set of mubs where lemma [ lem : smalllbound ] is tight . for the set of @xmath63 mubs in dimension @xmath2 constructed from the classes given in , our bound @xmath213 \approx 0.678\ , \end{aligned}\ ] ] is tight , and the minimum is indeed achieved by a state that is invariant under the unitary transform that cycles through the bases , as defined in . as noted in section [ sec : symmetry ] , in this case , the largest eigenvalue of @xmath159 occurs for a @xmath160 of the form @xmath176 for any @xmath214 . the states that achieve the lower bound are in fact eigenvectors of @xmath47 , which can be expressed in terms of the mub basis vectors as follows , @xmath215 average min - entropy for different sets of mubs in dimension @xmath130 . + the bound in is close to tight for @xmath66 mubs in dimension @xmath130 . the second analytical bound in is stronger than for @xmath216 bases . the circle denotes the average min - entropy for invariant states constructed in dimension @xmath130 , similar to the states described in . for @xmath217 mubs in @xmath130 , the minimum of the average min - entropy is nearly attained by states invariant under @xmath47.,scaledwidth=52.0% ] we have shown that there exist up to @xmath43 mutually unbiased bases in dimension @xmath1 for which we can find a unitary that cyclically permutes these bases , whenever @xmath41 is prime and @xmath41 divides @xmath44 or @xmath45 . this unitary is found by exploiting symmetry properties of the clifford algebra . our approach is quite distinct from the phase space approaches that were previously used to show that there exists such a unitary for the set of all @xmath39 mubs @xcite , or for two halves of the full sets of mubs when @xmath42 @xcite . our unitary can be understood as a generalization of the fourier transform , and it would be interesting to see whether it has other applications in quantum information . it is an interesting open question to generalize our result to other dimensions , or to a different number of bases . in prime dimension , one could consider the generalized clifford algebra @xcite . even though it does not have the full @xmath218 symmetry , it nevertheless exhibits enough symmetries to allow an exchange of generators . this stems from the way the ( generalized ) clifford algebra is obtained @xcite , which permits any transformation that preserves the @xmath219-norm for @xmath220 in dimension @xmath219 . yet , this is only the first step of our construction . as for generalizing our result to any @xmath41 bases in dimension @xmath1 , we note that it _ is _ indeed possible to find such classes even when @xmath41 is not prime , as our example for @xmath63 in dimension @xmath2 shows . however , we also know that for @xmath129 classes in dimension @xmath221 , no partitioning of operators can be found satisfying our requirements . it is an interesting open question as to when such a partitioning can be found in general . finally , we use our complementarity transform to obtain a tight uncertainty relation for the min - entropy for @xmath63 bases in dimension @xmath2 . no tight relations are known for this case before . we also use a slight generalization of the unitary from @xcite to show that when @xmath1 and @xmath41 divides @xmath39 , the minimizing state is an invariant of a certain unitary . this is the first time that significant insight has been obtained on the structure of the minimizing states for min - entropic uncertainty relations for mutually unbiased bases . it is an exciting open question to obtain tight relations in general , and understand the structure of the minimizing states . we are grateful to david gross for pointing us to the relevant literature for the discrete phase space construction . we also thank lukasz fidkowski and john preskill for interesting discussions . pm and sw are supported by nsf grant phy-0803371 . 44 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) . , * * ( ) . , * * , ( ) . , * * ( ) . , * * , ( ) . , * * ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , ( ) , . , , , , in _ _ ( ) , pp . . , , , , ( ) , . ( ) , . , * * , ( ) . , , , , , * * ( ) . , * * , ( ) . , * * , ( ) . ( ) , . , * * ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) , . ( ) , . , * * , ( ) . ( ) , . , ph.d . thesis , ( ) . , * * , ( ) . , ph.d . thesis , ( ) . , in _ _ ( , ) , vol . of _ _ , pp . . , in _ _ ( ) , pp . , . , * * , ( ) . , * * , ( ) . , in _ _ ( ) , pp . . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , * * , ( ) . , _ _ ( , ) . , ph.d . thesis , ( ) , . it is well known that for any orthonormal transformation @xmath96 there exists a corresponding unitary transformation @xmath99 @xcite , where we refer to ( * ? ? ? * appendix c ) for instructions on how to obtain explicit constructions . the transformation we wish to construct here , of the form @xmath222 , is thereby particularly simple to obtain . it can be build up from successive rotations in the plane spanned by only two `` vectors '' @xmath114 and @xmath115 . more specifically , we first construct a unitary that corresponds to a rotation around an angle @xmath223 in the plane spanned by @xmath114 and @xmath115 , bringing @xmath114 to @xmath115 . this is simply a reflection around the plane orthogonal to the midvector between @xmath114 and @xmath115 , followed by a reflection around the plane orthogonal to @xmath115 . using the geometric properties of the clifford algebra this corresponds to the unitary @xmath224 to obtain the desired unitary , we now compose a number of such rotations . let @xmath225 if @xmath226 is odd , and @xmath227 if @xmath226 is even . furthermore , let @xmath228 if @xmath41 is odd , and @xmath229 if @xmath41 is even . note that @xmath230 is the unitary that flips the sign of @xmath231 , but leaves all @xmath114 for @xmath232 and @xmath233 invariant . we may then write @xmath234 this unitary hence transforms @xmath235 , but leaves all other generators @xmath114 for @xmath236 invariant . a similar unitary can be found for any transformation @xmath237 @xcite , but is more difficult to construct explicitly . in and we gave examples of constructing @xmath66 and @xmath63 mubs in dimension @xmath2 , such that they are cyclically permuted under the action of a unitary @xmath47 that permutes the clifford generators in @xmath2 . here , we show by a general construction that it is always possible to construct @xmath41 such classes in dimension @xmath200 , whenever @xmath238 and @xmath41 is prime . we also outline a construction for @xmath45 classes , given a unitary @xmath47 that cycles through _ all _ @xmath0 clifford generators , when @xmath0 is prime . given the @xmath94 generators of the clifford algebra in dimension @xmath200 , we consider the set @xmath239 to generate a set of @xmath240 mubs , we seek to group the elements of @xmath241 into @xmath41 classes of commuting operators , ie . sets @xmath242 of size @xmath109 such that 1 . the elements of @xmath110 commute for all @xmath111 , 2 . the classs are all _ mutually disjoint _ , that is , @xmath243 3 . the unitary @xmath47 that cyclically permutes the generators @xmath116 , also permutes the corresponding classes by permuting products of operators appropriately . our approach in obtaining such a set of classes is to first pick @xmath244 elements for the class @xmath245 and then generate the rest of the classes by repeated application of @xmath47 to the elements of @xmath245 . this automatically ensures property ( * p3 * ) . to ensure ( * p1 * ) and ( * p2 * ) , the @xmath244 operators @xmath246 must satisfy the following : 1 . for any pair @xmath247 , @xmath248 = 0 $ ] , and 2 . the operators in @xmath245 cycle through _ mutually disjoint _ sets of operators under the action of @xmath47 . to understand condition ( ii ) better , consider an operator @xmath249 in @xmath245 . then , by construction , @xmath250 for @xmath251 , assuming we construct a total of @xmath41 classes . in addition , property ( ii ) implies @xmath252 , for any @xmath253 . in other words , given any two operators @xmath247 that cycle through the sets @xmath254 respectively under the action of @xmath47 , property(ii ) demands that @xmath255 , for all @xmath256 and @xmath257 . finally , we note that no class can contain two generators @xmath114 and @xmath115 , since they do not commute . when forming the classes we hence ensure that each one contains exactly one generator @xmath114 , which we refer to as the _ singleton _ @xmath258-operator of the class , as opposed to the rest of the elements which will be _ products _ of @xmath258-operators . the fact that each class can contain at most one singleton operator limits us to constructing a maximum of @xmath0 such classes . before proceeding to outline our construction , we establish some useful mathematical facts which will help motivate our algorithm for the construction of mutually disjoint classes . for the rest of the section , we will work with a set of @xmath219 @xmath258-operators @xmath259 that are cycled under the action of @xmath47 , as follows , @xmath260 in other words , we are given a set of @xmath258-operators whose _ cycle - length _ is @xmath219 . first , we consider sets of products of two @xmath258-operators of the form @xmath261 , which we call _ length-2 _ operators . it is convenient to characterize such pairs in terms of the _ spacing _ ( @xmath262 ) between the operators that constitute them . the spacing function @xmath262 , for a given set of @xmath219 operators , is simply defined as : @xmath263 . then , the following holds : [ lem : spacing ] the action of @xmath47 on any length-2 operator @xmath261 leaves its spacing function @xmath264 invariant . thus , length-2 operators that have unique spacings cycle through mutually disjoint sets of operators under the action of @xmath47 . recall , @xmath265 . it clearly follows that @xmath266 similar to defining length-2 operators , we refer to any product of @xmath267 @xmath258-operators as a _ length-@xmath267 _ operator . for operators of length higher than @xmath58 , it becomes convenient to refer to them using their corresponding index sets . for example , the operator @xmath268 will be simply denoted by the index set @xmath269 . in the following lemma , we obtain a condition for any set of length-@xmath267 operators to cycle through mutually disjoint sets under the action of @xmath47 . [ lem : lengthl ] suppose the length-@xmath267 operators ( for @xmath270 ) that belong to the class @xmath271 are such that they correspond to index sets @xmath272 which all sum to the same value @xmath273 then , no given index set of length @xmath267 can belong to more than one class , for prime values of @xmath219 . given the operators @xmath274 , such that the corresponding index sets @xmath275 sum to @xmath276 under the action of @xmath47 , these index sets change to @xmath277 for any index set @xmath278 the sum of the indices corresponding to the new operators @xmath279 becomes @xmath280 proceeding similarly , the corresponding operators in the class @xmath281 have index sets @xmath282 that sum to @xmath283 for all @xmath284 . thus , starting with a constraint on the length-@xmath267 operators in @xmath245 , we have obtained a constraint on the corresponding operators in a generic class @xmath285 . now , to arrive at a contradiction , suppose that an index set @xmath286 whose indices @xmath287 take values from the set @xmath288 , belongs to two different classes , @xmath285 and @xmath289 ( with @xmath290 ) . the constraint imposed by implies @xmath291 without loss of generality , let @xmath292 . since we can form at most @xmath219 classes , the difference @xmath293 can be at most @xmath294 . finally , since @xmath295 , condition can not be satisfied for prime values of @xmath219 . recall that our approach to constructing any @xmath219 classes is to first construct the class @xmath245 , and then obtain the rest by successive application of @xmath47 . therefore , the fact that any index set of a certain length @xmath267 can not belong to more than one class implies that each length-@xmath267 operator in @xmath245 cycles through a unique set of length-@xmath267 operators under @xmath47 . in other words , the length-@xmath267 operators cycle through mutually disjoint sets , as desired . lemma [ lem : lengthl ] thus provides us with a sufficient condition for the set of length-@xmath267 operators in @xmath245 to cycle through mutually disjoint sets under @xmath47 , given a set of @xmath258-operators whose cycle - length is prime - valued . we only need to ensure that the length-@xmath267 operators in the first class that we construct , @xmath245 , correspond to index sets that _ all _ sum to the same value . this condition is of course subject to the constraint that the maximum allowed length for the operators in @xmath245 ( and by extension , in any class ) is @xmath296 . as a warmup , we construct @xmath45 classes in dimension @xmath297 , when @xmath0 is prime . this case is particularly easy , and illustrates how the results of the previous sections will be used in general . [ thm:(2n+1)classes ] let @xmath298 denote the complete set of @xmath299 @xmath258-operators , and let @xmath47 be the unitary that cycles through all of them , that is , @xmath300 if @xmath0 is prime , then there exist @xmath0 classes @xmath301 satisfying properties * ( p1 ) * through * ( p3)*. we prove the existence of @xmath0 classes by construction . we first outline an algorithm to pick @xmath244 operators that constitute the class @xmath271 . the remaining classes are easily obtained by the application of @xmath47 to the elements of @xmath271 . then , we make use of lemmas [ lem : spacing ] and [ lem : lengthl ] to prove that the classes obtained through our construction do satisfy the desired properties . + * * 1 . pick one of the elements of @xmath302 , @xmath303 , as the singleton operator . 2 . pair up the remaining operators in @xmath302 to form @xmath304 length-2 operators which commute with @xmath303 , as follows , @xmath305 where @xmath306 denotes the set of length-2 operators in @xmath307 . since we have left out the pair @xmath308 in the middle , we get , @xmath309 . 3 . form higher length operators that commute with @xmath310 , by combining @xmath303 with appropriate combinations of the length-2 operators . any operator of even length @xmath311 is created by combining @xmath312 pairs in @xmath313 . and any operator of odd length @xmath314 is created by appending @xmath303 to a length-@xmath315 operator . + denoting the sets of length-3 operators as @xmath316 , length-4 operators as @xmath317 , and in general , the set of length-@xmath312 operators as @xmath318 , we have , @xmath319 putting together the operators from steps @xmath320 , @xmath321 , and @xmath322 we get the desired cardinality for the class @xmath271 as follows : @xmath323 the rest of the classes are generated by successive applications of the unitary @xmath47 to the elements of @xmath271 , so that @xmath324 . + it is easy to see that the elements of each class satisfy property ( * p1 * ) above the different length operators have been picked in such a way as to ensure that they all commute with each other . similarly , by construction , they satisfy property * ( p3)*. it only remains to prove property ( * p2 * ) , that the classes are all mutually disjoint . + the elements of @xmath306 correspond to the following set of spacings @xmath325 which are all distinct . so by lemma [ lem : spacing ] , the elements of @xmath306 cycle through mutually disjoint sets of length-@xmath58 operators . for higher length operators , we first show that our construction meets the conditions of lemma [ lem : lengthl ] . for the class @xmath245 , the elements of @xmath306 correspond to index sets that satisfy @xmath326 the length-2 operators of a generic class @xmath281 similarly satisfy @xmath327 since higher length operators are essentially combinations of length-2 operators and the singleton operator , conditions similar to hold for higher length index sets as well . since operators of even length @xmath311 contain @xmath183 pairs from @xmath306 , the corresponding index sets in @xmath245 satisfy @xmath328 similarly , since the odd length operators have @xmath123 appended to the even length operators , the index sets of length @xmath314 in @xmath245 satisfy , @xmath329 to sum up , for any @xmath330 , our construction ensures that index sets of length-@xmath267 belonging to @xmath245 sum to the same value . the conditions of lemma [ lem : lengthl ] are therefore satisfied , with the quantity @xmath331 in taking the value @xmath332 , for all @xmath333 . now , we can simply evoke lemma [ lem : lengthl ] to prove that , when @xmath0 is prime , the higher length operators in @xmath245 cycle through mutually disjoint sets of operators . next , we show that it is possible to obtain an arrangement of operators into @xmath41 classes in dimension @xmath334 , when @xmath41 is prime and @xmath238 , such that the unitary @xmath47 that cyclically permutes @xmath41 @xmath258-operators also permutes the coresponding classes . [ lclasses ] suppose @xmath47 is a unitary that cycles through sets of @xmath41 @xmath258-operators from the set @xmath335 in dimension @xmath334 , where @xmath41 is prime and @xmath238 . then there exist @xmath41 classes @xmath336 that satisfy properties * ( p1 ) * through * ( p3)*. * proof : * note that since @xmath238 we have @xmath337 for some positive integer @xmath338 . the set of @xmath94 clifford generators @xmath339 can then be partitioned into @xmath340 sets as follows : @xmath341 without loss of generality , we can assume the unitary @xmath47 is constructed such that it cyclically permutes the @xmath41 operators within each set , as follows . @xmath342 once again , we begin with an algorithm for picking @xmath244 elements for the class @xmath245 . the algorithm closely follows the one outlined in the previous section , barring some minor modifications . + * * 1 . the `` middle '' element from @xmath343 , @xmath344 , is picked as the singleton element of @xmath245 . the @xmath304 length-2 operators which commute with @xmath344 are picked as follows 3 . @xmath345 pairs are picked from @xmath346 @xmath347 leaving @xmath123 and @xmath348 unused . @xmath349 pairs are picked from each of the sets @xmath343 through @xmath350 , @xmath351 leaving the first operator in each set unused . finally , the unused @xmath258-operators from different sets are put together as specified below , to get the remaining @xmath338 length-@xmath58 operators : @xmath352 the set of length-@xmath58 operators is then given by @xmath353 which gives @xmath354 . pick higher length operators from @xmath241 that commute with @xmath344 and @xmath313 , by combining @xmath344 with appropriate combinations of the length-2 operators . as before , any even - length operator of length @xmath355 is obtained by combining @xmath312 length-2 operators from @xmath306 . any operator of odd - length @xmath356 , is created by appending @xmath344 to a length-@xmath357 operator . putting together all the operators created in steps[1]-[3 ] , we get the desired cardinality for the class ( see ) , that is , @xmath358 . + * proof of properties * ( p1 ) * through * ( p3 ) * : * the different length operators have again been picked in such a way as to ensure that they all commute with each other . since the remaining @xmath359 classes are generated by successive applications of the unitary @xmath47 to the elements of @xmath271 , we have @xmath360 . thus * ( p1 ) * and * ( p3 ) * is satisfied . it remains to prove that the classes constructed here also satisfy property ( * p2 * ) . as in the earlier case of @xmath0 classes , the operators in each of the sets @xmath361 correspond to unique values of the spacing function : @xmath362 , \nonumber\ ] ] which guarantees , by lemma [ lem : spacing ] that these operators cycle through mutually disjoint sets under @xmath47 . since the operators in @xmath363 are formed by combining @xmath258-operators from different sets @xmath364 , each of them cycles through a different set of operators under @xmath47 . thus we see that all the length-@xmath58 operators in @xmath245 cycle through mutually disjoint sets . before we proceed to discuss the higher length operators , we make one further observation about the length-@xmath58 operators . the operators in @xmath306 correspond to index sets which satisfy @xmath365 in particular , the length-2 operators in the set @xmath366 have been picked carefully so as to ensure that the above constraint is satisfied . in fact , this was the rationale behind leaving out the first operator in each of the sets @xmath364 while choosing the corresponding length-@xmath58 elements in @xmath367 . the higher length operators in @xmath245 can be of two types : 1 . those that are comprised of @xmath258-operators from a single set @xmath364 alone , and 2 . operators that comprise @xmath258-operators from more than one set . since a type-(a ) operator can not cycle into a type-(b ) operator under the action of @xmath47 , these two cases can be examined separately . + * type-(a ) : * the maximum length that an operator of type-(a ) can have , as per our construction , is @xmath359 . we have ensured this by leaving at least one operator of each of the sets @xmath364 unused in constructing the length-@xmath58 operators . furthermore , the constraint in implies that the index sets corresponding to such higher length operators in @xmath245 , sum to the same value modulo @xmath41 . more precisely , any even - length index set of length @xmath368 , where the indices are all drawn from a given set @xmath364 , satisfies @xmath369 and any index set of odd length @xmath314 satisfies @xmath370 then , invoking lemma [ lem : lengthl ] with @xmath371 for even values of @xmath267 and @xmath372 for odd values of @xmath267 , we see that no operator of type-(a ) can belong to more than one class , for prime values of @xmath41 . + * type-(b ) : * an operator of type-(b ) is a product of operators from smaller sets @xmath373 . consider a length-@xmath267 operator , @xmath374 which comprises @xmath375 @xmath258-operators from @xmath376 , @xmath377 operators from @xmath343 , and in general , @xmath378 from the set @xmath364 . @xmath379 note that by our construction , the operator @xmath374 exists in more than one class if and only if , for all @xmath380 the product of all operators in @xmath380 also belongs to more than one class . in what follows , we argue that our construction ensures that this is not possible . in particular , given a set of length-@xmath267 operators in @xmath245 which can be broken down into smaller sets as described above , we will argue that there exists at least one set @xmath380 in every such length-@xmath267 operator @xmath374 , such that the products of operators in @xmath380 corresponding to different length-@xmath267 operators cycle through mutually disjoint sets , as defined earlier . note the following two facts about the subsets @xmath381 . first , our construction ensures that any subset @xmath382 of a given size @xmath383 , satisfies either or depending on @xmath383 being even or odd . second , note that the maximum size of these subsets is @xmath384 . however , in order to invoke lemma [ lem : lengthl ] , we still require @xmath383 to be strictly less than @xmath41 . our goal is hence to show that every length-@xmath267 operator must have at least one subset @xmath381 of size @xmath385 . suppose there exists a length-@xmath267 operator such that every subset is of size @xmath41 . then , the operator itself has to be of length @xmath386 however the maximum value of @xmath267 in our construction is @xmath387 , implying that atleast one of the @xmath340 subsets must be of a size strictly smaller than @xmath41 . and , for such a subset of size less than @xmath41 , constraints and ensure that the same subset can not be found in more than one class , provided @xmath41 is prime . the min - entropy of the distribution that an orthonormal basis @xmath389 induces on a state @xmath390 is given by @xmath391\ ] ] we are looking to evaluate a lower bound on the average min - entropy of any @xmath41 mutually unbiased bases ( not necessarily coming from our construction ) in a @xmath13-dimensional hilbert space . the average min - entropy is given by - @xmath392 using jensen s inequality . the problem of finding an optimal uncertainty relation for the min - entropy , thus reduces to the problem of maximizing over all @xmath390 , the quantity @xmath393 . it is easy to see that this maximum is always attained at a pure state , so we can restrict the problem to an optimization over pure states . we can simplify the problem of finding the lower bound of by recasting it as follows . consider states of the form @xmath394 where @xmath395 denotes a string of basis elements , that is , @xmath396 . suppose we can show for all possible strings @xmath160 , @xmath397 then , since @xmath398 $ ] , the bound is simply @xmath399 we have thus reduced the problem to one of finding the maximum eigenvalue of operators of the form @xmath159 , over all possible strings @xmath160 . note that by , it is sufficient to determine @xmath157 in . to solve this eigenvalue problem we recall a result of schaffner @xcite proved using the methods of kittaneh @xcite , that for a set of @xmath41 orthogonal projectors @xmath402 , the following bound holds : @xmath403 where @xmath404 denotes the operator norm , which here is simply the maximum eigenvalue for hermitian operators . applying this result to sums of basis vectors @xmath208 , we have , @xmath405 which implies @xmath406 recall , that for all @xmath407 @xmath408 where @xmath409 denotes some phase factor . further , since the vectors @xmath208 are normalized , the cauchy - schwarz inequality gives @xmath410 combining these with gives the following bound on the maximum eigenvalue of the operator @xmath159 : @xmath411 by , this immediately proves our claim . here , we present an alternate approach to bound the maximum eigenvalue of @xmath159 , using a bloch vector like representation of the mub basis states . the bound that we obtain here , stated in lemma [ lem : largelbound ] , is stronger than the last one when @xmath211 . in particular , when we consider the complete set ( @xmath188 ) of mubs in any dimension @xmath13 , this approach yields the best known bound . first , we switch to working in a basis of hermitian operators , so that every state in @xmath413 has a parametrization in terms of vectors in a real vector space . any state @xmath390 can be written as : @xmath414 where @xmath415 are hermitian , trace - less operators that are orthogonal with respect to the hilbert - schmidt norm : @xmath416 = 2\;\delta_{ij}$ ] , and the scalars @xmath417 . thus we can parameterize any state in our @xmath13-dimensional hilbert space with a vector @xmath418 . when @xmath165 is a pure state ( @xmath419 = 1 $ ] ) , the vector @xmath420 corresponding to this pure state satisfies the following normalization condition @xmath421 & = & 1 \nonumber \\ \rightarrow \frac{1}{d } + \frac{1}{2}\sum_{i=1}^{d^2 - 1}|\alpha^{(i)}|^{2 } & = & 1 \nonumber \\ \rightarrow |\vec{\alpha}| = \sqrt{\sum_{i=1}^{d^2 - 1}|\alpha^{(i)}|^{2 } } & = & \sqrt{\frac{2(d-1)}{d}}\end{aligned}\ ] ] * ( m1 ) _ normalization _ : @xmath424 = 1 $ ] implies that @xmath425 , @xmath426 . ( by an argument similar to the one that leads to . ) * ( m2 ) _ constant inner - product _ : @xmath427 implies that @xmath428 . this is easily seen , as follows : @xmath429 & = & \frac{1}{d } + \frac{1}{2}\sum_{i}\alpha^{(i)}_{(b , j)}\alpha^{(i)}_{(\hat{b},k ) } = \frac{1}{d } \nonumber \\ \rightarrow \vec{\alpha}_{(b , j)}.\vec{\alpha}_{(\hat{b},k ) } & = & 0\end{aligned}\ ] ] now , using this representation of mub states and density operators , we can rewrite the maximization problem of as : @xmath430 = \max_{{|\psi\rangle}}\textrm{tr}\left[\frac{1}{l}\sum_{j}{|b^{(j)}\rangle}{\langleb^{(j)}|}{|\psi\rangle}{\langle\psi|}\right]\nonumber \\ & \leq & \max_{\vec{\alpha}}\frac{1}{l}\sum_{j}\textrm{tr}\left[\left(\frac{\mathbb{i}}{d } + \frac{\sum_{j}\alpha^{j}_{(b^{(j)},j)}\hat{a}_{j}}{2}\right)\left(\frac{\mathbb{i}}{d } + \frac{\sum_{i}\alpha^{(i)}\hat{a}_{i}}{2}\right)\right ] \nonumber \\ & = & \max_{\vec{\alpha}}\frac{1}{l}\sum_{j}\left(\frac{1}{d } + \frac{1}{2}\vec{\alpha}_{(b^{(j ) } , j)}.\vec{\alpha}\right ) \nonumber \\ & = & \frac{1}{d } + \max_{\vec{\alpha}}\frac{1}{2l}\sum_{j}\vec{\alpha}_{(b^{(j ) } , j)}.\vec{\alpha } \end{aligned}\ ] ] now we only need to find the real @xmath431-dimensional vector @xmath420 , that maximizes the sum @xmath432 . if we now define an `` average '' vector corresponding to each string @xmath160 , as follows @xmath433 then , it becomes obvious that the maximum is attained when @xmath420 is parallel to @xmath434 . since it is a vector corresponding to a pure state , its norm is given by , so that @xmath435 note that this maximizing vector has a constant overlap with all vectors @xmath436 , for a given string @xmath160 . in other words , for each string @xmath160 , the maximum is attained by the vector that makes equal angles with all the vectors that constitute the `` average '' vector ( @xmath437 ) corresponding to that string . note however that this vector may not always correspond to a valid state . now that we know the maximizing vector , we can go ahead and compute the value of @xmath157 in . @xmath438 & \leq & \frac{1}{d } + \max_{\vec{\alpha}}\frac{1}{2l}\sum_{j}\vec{\alpha}_{(b^{(j ) } , j)}.\vec{\alpha } \nonumber \\ & = & \frac{1}{d } + \frac{1}{2}\max_{\vec{\alpha}}\vec{\alpha}_{(\textrm{avg})}.\vec{\alpha } \nonumber \\ & = & \frac{1}{d } + \frac{1}{2}\frac{\vec{\alpha}_{(\textrm{avg})}.\vec{\alpha}_{(\textrm{avg})}}{|\vec{\alpha}_{(\textrm{avg})}|}\sqrt{\frac{2(d-1)}{d } } \nonumber \\ & = & \frac{1}{d } + \frac{1}{2}|\vec{\alpha}_{(\textrm{avg})}|\sqrt{\frac{2(d-1)}{d } } \nonumber \\ & = & \frac{1}{d } + \frac{1}{2\sqrt{l}}\frac{2(d-1)}{d } \nonumber \\ & = & \frac{1}{d}\left(1 + \frac{d-1}{\sqrt{l}}\right)\end{aligned}\ ] ] where we have used the fact that the vector @xmath437 have a constant norm which can be computed as follows : @xmath439 \nonumber \\ \rightarrow |\vec{\alpha}_{(\textrm{avg})}| & = & \frac{1}{\sqrt{l}}\sqrt{\frac{2(d-1)}{d}},\end{aligned}\ ] ] thus proving our claim . the second step follows from the fact that vectors corresponding to different mub states have zero inner product ( see property @xmath440 above ) . note that the fact that the bases are mutually unbiased was crucial in giving rise to properties @xmath441 and @xmath440 which in turn enabled us to identify the maximizing vector @xmath442 . indeed the maximizing vector corresponding to a given string @xmath160 might not always correspond to a valid state , in which case the bound we derive can not be achieved . however , there exist strings of basis elements @xmath160 , for which we can explicitly construct a state that has equal trace overlap with the states that constitute the corresponding operator @xmath159 . these are in fact states of the form @xmath443 for any @xmath444 . clearly , for the symmetric mubs that we construct , an eigenstate of the unitary @xmath47 that cycles between the different mubs has the same trace overlap with each of the states @xmath445 , for a fixed value of @xmath182 . to see this , suppose @xmath446 is an eigenvector of @xmath47 with eigenvalue @xmath447 , then for all @xmath111 and a given value of @xmath182 , @xmath448 & = & |\langle b^{(j)}|\phi\rangle|^{2 } = |\langle b^{(1)}|(u^\dagger)^{j-1}|\phi\rangle|^{2 } \nonumber \\ & = & ( |\lambda|^{2})|\langle b^{(1)}|\phi\rangle|^{2 } \nonumber \\ & = & |\langle b^{(1)}|\phi\rangle|^{2 } \ ; \end{aligned}\ ] ] this is indeed the case for @xmath63 mubs in @xmath2 , where the lower bound we derive is achieved by eigenstates of @xmath47 .
even though mutually unbiased bases and entropic uncertainty relations play an important role in quantum cryptographic protocols they remain ill understood . here , we construct special sets of up to @xmath0 mutually unbiased bases ( mubs ) in dimension @xmath1 which have particularly beautiful symmetry properties derived from the clifford algebra . more precisely , we show that there exists a unitary transformation that cyclically permutes such bases . this unitary can be understood as a generalization of the fourier transform , which exchanges two mubs , to multiple complementary aspects . we proceed to prove a lower bound for min - entropic entropic uncertainty relations for any set of mubs , and show that symmetry plays a central role in obtaining tight bounds . for example , we obtain for the first time a tight bound for four mubs in dimension @xmath2 , which is attained by an eigenstate of our complementarity transform . finally , we discuss the relation to other symmetries obtained by transformations in discrete phase space , and note that the extrema of discrete wigner functions are directly related to min - entropic uncertainty relations for mubs .
studies of thermal fluctuations in the phase of the superconducting order parameter have a long history due to their importance in understanding phase transitions in general . because of the high superfluid densities , @xmath7 , and long coherence lengths , @xmath8 , of conventional superconductors , fluctuations in three - dimensional ( 3d ) samples are significant only within a hopelessly narrow temperature range near @xmath6 @xcite , and most experimental studies involve thin amorphous films in which reduced dimensionality and superfluid density enhance fluctuations to an observable level . two points that are relevant to work on fluctuations in high-@xmath6 cuprate superconductors have been established . first , quantum mechanical effects suppress thermal phase fluctuations ( tpf s ) when @xmath9 drops below a surprisingly high temperature @xcite . second , at high temperatures where fluctuations are classical , simulations of josephson junction ( jj ) arrays @xcite are in reasonable quantitative agreement with measurements of the superfluid density in two - dimensional ( 2d ) films @xcite . in this paper we concern ourselves with fluctuations in hole - doped cuprate superconductors in zero applied magnetic field . thermal phase fluctuations were immediately expected to be significant in cuprate superconductors @xcite , which have low superfluid densities and weak interlayer coupling . exactly how important tpf s are is a matter of considerable debate . in the absence of directly comparable experimental results on quasi-2d conventional superconductors , we rely on simulations of strongly anisotropic jj arrays @xcite to provide a framework for examining experimental results . in yba@xmath1cu@xmath2o@xmath3 ( ybco ) , strong fluctuations are evidenced by very rapid changes in the _ ab_-plane penetration depth @xmath10 @xcite , specific heat @xcite , and coefficient of thermal expansion @xcite over a 5 to 10 k interval near @xmath6 . perhaps the most striking evidence that these phenomena are due to tpf s is power - law behavior with the 3d - xy exponent 2/3 in the superfluid density , @xmath11 , of ybco crystals @xcite . in the tpf interpretation , fluctuations are strong because of quasi - two - dimensionality , and interlayer coupling although weak begets a crossover from 2d to the observed 3d fluctuations @xcite . why not just accept these results at face value and move on ? the reason is that @xmath6 marks a second order phase transition from an unusual normal state to a _ d_-wave superconductor . it would not be surprising if normal - state peculiarities persisted into the superconducting state and resulted in anomalous @xmath9-dependence of @xmath7 . for example , if there is a normal - state order parameter ( e.g. , see ref . 13 ) , it may compete with the superconducting order parameter @xcite . in the present paper , we argue from measurements of @xmath7 in the literature and new measurements presented below , that tpf effects are actually small , presumably because interlayer coupling in cuprates is much stronger than indicated by transport measurements . the pivotal notions of the tpf model are that fluctuations are strong because interlayer coupling is very weak , and that interlayer coupling serves only to change the @xmath9-dependence of @xmath0 from a discontinuous kosterlitz - thouless - berezinskii ( ktb ) @xcite drop at @xmath12 to a continuous decrease to zero , with rapidly increasing downward curvature @xcite . one key quantity in comparing @xmath0 with the tpf model is the hypothetical transition temperature , @xmath12 , at which the ktb transition would occur if layers could be completely decoupled without affecting the pairing interaction . @xmath12 can be deduced from the measured _ ab_-plane superfluid density . a quantitative measure of the strength of fluctuations , and therefore of interlayer coupling , is the difference between @xmath12 and the temperature , @xmath6 , at which @xmath0 vanishes : the weaker the coupling , the smaller the difference . from numerical results in ref . 6 , @xmath13 increases roughly as @xmath14 , where @xmath15 ( @xmath16 ) is the interlayer ( intralayer ) coupling strength . the other key quantity is @xmath17 . in the tpf model , 3d - xy critical behavior develops only as @xmath7 approaches zero , i.e. , for @xmath18 . let us now take a closer look at power - law behavior in @xmath0 observed for some ybco samples , setting aside concerns that power - law behavior is not seen in other ybco samples @xcite nor in clean bi@xmath4sr@xmath4cacu@xmath4o@xmath5 ( bscco ) crystals @xcite . a recent measurement of the _ a_-axis magnetic penetration depth in a clean , optimally doped ybco crystal @xcite finds that at temperatures from @xmath19% to @xmath20% of the transition at @xmath21 k , @xmath22 is fitted to within a percent or so by @xmath23 . the critical region encompasses a remarkably large range of temperature and superfluid density : @xmath24 and @xmath25 . the argument that an excellent fit is unlikely to be accidental is taken as support for the tpf model , but in reality it is not . the reason is that in the tpf model the crossover from 3d to 2d fluctuations occurs well inside the observed `` critical region '' . to see this , let us estimate @xmath12 and @xmath26 from the data . with an effective layer thickness equal to the center - to - center spacing between cuo@xmath4 bilayers ( @xmath27 nm ) , @xmath28 nm , and the ktb prediction @xcite @xmath29 where @xmath30 , we find @xmath31 and @xmath32 . the tpf model predicts 3d - xy behavior only for @xmath33 , which for these data means @xmath34 . power - law behavior from 0.85 to 0.99 @xmath6 would have to be `` accidental '' . taking this one step further , if power - law behavior is due to physics beyond the tpf model , then tpf s must be so weak as to be unapparent in the data . another argument against the tpf model arises from comparison of ybco with bscco , which has cuo@xmath4 bilayers that are nearly identical structurally to those of ybco i.e. , the same interplanar spacing and nearly square cu - o structure except that y@xmath35 is replaced by ca@xmath36 @xcite . as expected , the areal superfluid density , @xmath37 , at @xmath9 = 0 of the bilayers in clean bscco is essentially the same as in the _ a_-axis direction ( perpendicular to cuo chains ) in ybco . however , bscco has interlayer coupling several hundred times weaker than ybco , as measured by transport @xcite , so we expect tpf s to be much stronger . let us see . @xmath6 lies 4 k above @xmath12 in ybco crystals , as found above a reasonable result when compared with simulations of jj arrays @xcite . in bscco , the difference is 10 k , indicating that fluctuations in bscco are actually weaker than in ybco . the latter number comes from clean bscco crystals @xcite in which @xmath38 decreases linearly with @xmath9 at low @xmath9 , @xmath6 is 91 k , the center - to - center spacing between cuo@xmath4 bilayers is 1.54 nm @xcite , and @xmath39 nm . evidently interlayer coupling is stronger in bscco than indicated by transport measurements of _ ab_-plane _ vs_. _ c_-axis anisotropy . we could refine the above estimations by using two interlayer coupling constants , one within bilayers and another between bilayers , per carlson _ et al_. @xcite , and use an average of the _ ab_-plane superfluid density rather than just the _ a_-axis value in ybco , but the conclusions do not change . other experimental evidence supports the proposition that interlayer coupling is stronger than expected . some corroboration is provided by a study of @xmath40 above @xmath6 in ybco crystals @xcite , which finds a critical region of only 0.6 k. more direct support is provided by the measurements presented below , which demonstrate that fluctuation effects in the 2d penetration depth , @xmath41 , of ultrathin twinned ybco films ( @xmath42 = 4 , 8 , and 10 unit cells ) are quantitatively consistent with measurements on 2d _ s_-wave films @xcite , assuming that the ybco films are effectively isotropic i.e. , that their layers are strongly coupled . the films examined in this study were grown epitaxially by pulsed laser deposition on ndgao@xmath2 substrates , as detailed elsewhere @xcite . each 1 cm @xmath43 1 cm film consists of buffer layers of semiconducting pr@xmath44y@xmath45ba@xmath1cu@xmath2o@xmath3 that are 12 unit cells thick above , and 8 unit cells thick below , the ybco film . the underlayer lessens the strain of substrate lattice mismatch on the ybco , while the capping layer protects the ultrathin ybco film from damage during handling . the three films we examined are nominally 4 , 8 , and 10 unit cells thick ; however , it is likely that the top and bottom unit - cell layers are not perfectly smooth or homogeneous , leading to some uncertainty in the film thickness @xmath42 . we therefore estimate @xmath46 , @xmath47 , and @xmath48 unit cells for the three films . we measure @xmath49 with a two - coil mutual inductance technique described in detail elsewhere @xcite . the films are centered between two counterwound coils approximately 2 mm in diameter and 1 mm long . in a typical measurement , the sample is cooled to liquid helium temperatures , and a current of 100 @xmath50a is driven at 50 khz through the coil pressed against the back of the substrate . the supercurrents created in the sample are very nearly uniform through the film thickness . thus , we observe the @xmath6 s of all layers in the film . as the temperature slowly increases , the voltage induced in the secondary coil , which is pressed against the film , is measured continuously . the complex sheet conductivity @xmath51 of the _ entire _ film is deduced directly from the measured mutual inductance . the 2d penetration depth is obtained with an accuracy of about 3% from the imaginary part of the sheet conductivity from @xmath52 , where @xmath53 is the permeability of vacuum . the uncertainty in film thickness enters only into calculation of the 3d penetration depth @xmath54 , and into the 2d penetration depth of _ each _ unit - cell layer . figure 1 displays @xmath55 as measured at 50 khz for each film at optimal doping . an accelerated decrease in @xmath56 at high temperatures is most obvious for the 4 unit cell film , but similar features are apparent very close to @xmath6 for the 8 and 10 unit - cell films as well . the transitions are slightly broader than in @xmath57-moge films @xcite , presumably due to slight inhomogeneity . the vertical dotted lines mark the location of the peak in @xmath58 khz,@xmath59 , whose temperature we define to be @xmath6 and which very nearly coincides with the predicted ktb transition temperature , @xmath60 , for an isotropic film . to further motivate association of the drop in @xmath61 with the ktb transition , we appeal to the frequency dependences of @xmath62 and @xmath56 near the transition . without going into detail , the observed shift of @xmath63 and @xmath56 to higher temperature with increasing frequency , shown in fig . 2 for the 10 unit cell film , is quite similar to behavior seen in @xmath57-moge films @xcite . the experimental @xmath6 in these ultrathin films marks a ktb transition . arrows in fig . 1 indicate @xmath12 the ktb transition temperature for uncoupled layers showing that @xmath12 lies 8 to 20 k below @xmath6 , a difference significantly larger than seen in ybco crystals @xcite and simulations of jj arrays @xcite . another indicator of fluctuation effects is a rapid increase in downward curvature of @xmath64 for @xmath65 as predicted by the tpf model . the thin solid curves in fig . 1 are extrapolations of quadratic fits to the data over @xmath66 k @xmath67 . we emphasize that the quadratics fit the data extremely well over at least 20 k below @xmath12 , highlighting the constant curvature of the data below @xmath12 . the insets to fig . 1 show that only as @xmath9 gets close to @xmath6 is a change in curvature of the data apparent . the proximity of the accelerated drops in @xmath56 to the predicted @xmath60 , and the lack of fluctuation effects ( e.g. , strong downward curvature ) at @xmath12 , indicate the presence of strong interlayer coupling . we now address an important quantitative issue : the suppression of @xmath64 below its mean - field value just below @xmath68 . taking the thin solid curves in fig . 1 as reasonable estimates of the mean - field behavior of @xmath64 , we conclude that @xmath56 is 35% to 75% of its mean - field value at the transition , in agreement with results from @xmath57-moge films @xcite and calculations for 2d jj arrays @xcite . thus , the fluctuation behavior of ultrathin ybco films is quantitatively similar to that of _ a_-moge films . within the tpf model , the only explanation is that interlayer coupling in the ybco films is much stronger than expected from transport anisotropy . the strong _ ab_-plane _ vs_. _ c_-axis anisotropy in transport coefficients could be explained phenomenologically by occasional weak links between adjacent cuo@xmath4 bilayers , with most bilayers coupled strongly to their neighbors . we have presented high precision , low frequency measurements of @xmath69 and @xmath70 in ybco films of 4 to 10 unit cells thickness . the data indicate that interlayer coupling is strong , insofar as thermal phase fluctuations are concerned . whatever the origin of the coupling , it is the presence of an additional layer or layers that lessens the effect of fluctuations on a single unit - cell layer . these results , and literature results for the _ ab_-plane superfluid density of bscco crystals , argue that fluctuations are much weaker in cuprates than one would expect based on anisotropic transport coefficients . the power - law behavior near @xmath6 seen in the superfluid density of very clean ybco crystals cited as evidence of strong thermal phase fluctuations is due to some other physics , perhaps competition between the superconducting order parameter and a normal - state order parameter that is sensitive to disorder . the idea of a normal - state order parameter has been proposed on various grounds @xcite . the theory of rokhsar @xcite suggests that the critical region may be more complicated than in the usual superconducting - to - normal transition when the normal and superconducting order parameters interact . the authors gratefully acknowledge useful discussions with c. jayaprakash and y .- b . kim . this work was supported in part by doe grant de - fg02 - 90er45427 through the midwest superconductivity consortium . lobb , phys . b * 36 * , 3930 ( 1987 ) . turneaure , t.r . lemberger , and j.m . graybeal , phys . lett . * 84 * , 987 ( 2000 ) . lemberger , a.a . pesetski , and s.j . turneaure , phys . rev . b * 61 * , 1483 ( 2000 ) . l. benfatto , s. caprara , c. castellani , a. paramekanti , and m. randeria , phys . rev . b 63 , 174513 ( 2001 ) . t. ohta and d. jasnow , phys . b * 20 * , 139 ( 1979 ) ; s. teitel and c. jayaprakash , phys . rev . lett . * 51 * , 1999 ( 1983 ) ; phys . rev . b * 27 * , 598 ( 1983 ) ; w.y . shih and d. stroud , _ ibid . _ * 32 * , 158 ( 1985 ) ; hwang and d. stroud , _ ibid . _ * 57 * , 6036 ( 1998 ) . carlson , s.a . kivelson , v.j . emery , and e. manousakis , phys . 83 * , 612 ( 1999 ) . s. kamal , d.a . bonn , n. goldenfeld , p.j . hirschfeld , r. liang , and w.n . hardy , phys . lett . * 73 * , 1845 ( 1994 ) . anlage , j. mao , j.c . booth , d.h . wu , and j.l . peng , phys . b * 53 * , 2792 ( 1996 ) . s. kamal , r. liang , a. hosseini , d.a . bonn , and w.n . hardy , phys . b * 58 * , r8933 ( 1998 ) . m. charalambous , o. riou , p. gandit , b. billon , p. lejay , j. chaussy , w.n . hardy , d.a . bonn , and r. liang , phys . lett . * 83 * , 2042 ( 1999 ) . v. pasler , p. schweiss , c. meingast , b. obst , h. whl , a.i . rykov , and s. tajima , phys . lett . * 81 * , 1094 ( 1998 ) . t. schneider and j.m . singer , physica ( amsterdam ) * 341 - 348c * , 87 ( 2000 ) . s. chakravarty , r.b . laughlin , d.k . morr , and c. nayak , phys . b * 63 * , 094503 ( 2001 ) . rokhsar , phys . lett . * 70 * , 493 ( 1993 ) . kosterlitz and d.j . thouless , j. phys . c * 6 * , 1181 ( 1973 ) ; j.m . kosterlitz , _ ibid _ * 7 * , 1046 ( 1974 ) ; v.l . berezinskii , sov . jetp * 32 * , 493 ( 1971 ) . h. srikanth , z. zhai , s. sridhar , a. erb , and e. walker , phys . b * 57 * , 7986 ( 1998 ) . paget , b.r . boyce , and t.r . lemberger , phys . b * 59 * , 6545 ( 1999 ) . lee , d.c . morgan , r.j . ormeno , d.m . broun , r.a . doyle , j.r . waldram , and k. kadowaki , phys . lett . * 77 * , 735 ( 1996 ) . hazen , in _ physical properties of high temperature superconductors ii _ , edited by d.m . ginsberg ( world scientific , singapore , 1990 ) . y. iye , in _ physical properties of high temperature superconductors iii _ , edited by d.m . ginsberg ( world scientific , singapore , 1992 ) . x.h . chen , m. yu , k.q . ruan , s.y . li , z. gui , g.c . zhang , and l.z . cao , phys . b * 58 * , 14219 ( 1998 ) . turneaure , t.r . lemberger , and j.m . graybeal , phys . b * 63 * , 174505 ( 2001 ) . c. kwon , q. li , x.x _ et al_. , appl . 62 * , 1289 ( 1993 ) ; c. kwon , q. li , i. takeuchi _ et al_. , physica ( amsterdam ) * 266c * , 75 ( 1996 ) . turneaure , e.r . ulm , and t.r . lemberger , j. appl . phys . * 79 * , 4221 ( 1996 ) ; s.j . turneaure , a.a . pesetski , and t.r . lemberger , _ ibid . _ * 83 * , 4334 ( 1998 ) .
from study of the kosterlitz - thouless - berezinskii ( ktb ) transition in the superfluid density , @xmath0 , of ultrathin _ c_-axis oriented yba@xmath1cu@xmath2o@xmath3 ( ybco ) films , we find that interlayer coupling is unexpectedly strong . the ktb transition occurs at a high temperature , as if the films were isotropic rather than quasi - two - dimensional . this result agrees with a comparison of the superfluid density of ybco with bi@xmath4sr@xmath4cacu@xmath4o@xmath5 and with numerical simulations of josephson junction arrays , and challenges the thermal phase fluctuation interpretation of critical behavior near @xmath6 in ybco .
blazars are a sub - class of active galactic nuclei , emitting from radio to tev energies . they are subdivided in two main categories : flat spectrum radio quasars ( fsrq ) and bl lacertae ( bl lac ) objects . fsrqs are characterized by a flat radio spectrum in the ghz range ( with spectral index @xmath1 , where the flux density is @xmath2 ) , and strong and broad emission lines ( with equivalent width @xmath35 @xmath4 ) . bl lac objects , on the other hand , have no or weak emission lines with equivalent width @xmath55 @xmath4 . + blazars continuum emission originates from a relativistic jet aligned with the line of sight . their spectral energy distribution ( sed ) shows a double humped shape @xcite , with a low energy peak lying between ir and x - rays , and an high energy peak in the mev - tev band . + the low energy region of blazar spectra is associated with the synchrotron emission coming from the jet relativistic electrons . the high energy region can be modeled through inverse compton emission ( leptonic models ) , with seed photons coming from an external region ( e.g. the accretion disk , the dusty torus ) , eventually reprocessed by the broad line region , or the hot corona , or from the synchrotron process itself ( synchrotron self - compton , or ssc ) . a detailed description of leptonic models can be found in @xcite . the high energy region can also be modeled with hadronic scenarios @xcite , where the very high energy protons of the jet are radiatively important . the accelerated protons produce gamma ray emission through proton synchrotron emission , the decay of neutral pions , and synchrotron emission produced by secondaries . + the location of the so - called `` blazar - zone '' , i.e. , the spatial location of the blazar sed - peaks and gamma - ray emitting region , in blazars is still a matter of debate . @xcite proposed that the blazar - zone is located at 39 pc from the central engine for the outburst of 3c 454.3 ( a bright fsrq ) occurred in 2005 . for the same flare , @xcite indicated , instead , a dissipation region at 0.50.8 pc from the central black hole ( bh ) . from the combined study of time - dependent polarimetric radio images at 43 and 86 ghz , the optical polarimetry , and radio , optical , x- , gamma - ray light curves , @xcite proposed that the low and high energy emission is located near the 43 ghz core , at a distance of the order of tens of parsecs from the bh for 3c 454.3 . a similar investigation , performed for the bl lac objects oj287 and ao 0235 + 164 @xcite , led to similar results . + @xcite established that gamma - rays emitted inside the broad line region ( blr ) are absorbed at e@xmath6 @xmath7 due to the @xmath8 interaction with the blr photons ( internal absorption ) . @xcite refined this result , and claimed internal absorption features at e@xmath9 @xmath10/(1+z ) and at e@xmath6 @xmath7 in the gamma - ray spectrum of 3c 279 , 3c 454.3 , pks 1510 - 08 , and a few other fsrq . + within the leptonic scenario , @xcite show that the contribution of external photon fields , including contributions from the blr and dusty torus to the inverse compton emission can be parametrized as a function of the accretion disk luminosity and the dissipation distance of the emitting blob from the bh @xcite . @xcite modeled the sed of the gamma - ray loudest blazars as being emitted at 0.010.5 pc from the bh . + using multiwavelength observations of the blazar , we will apply the models of @xcite to further investigate the location of the blazar - zone . + was an unidentified gamma - ray source of the virgo region ( ) , detected with low significance @xcite by the egret gamma - ray telescope ( operating in the 30 mev 30 gev energy range , see @xcite ) . in recent years the gamma - ray source has shown two episodes of remarkable high energy activity : at the beginning and at the end of 2008 , when it was detected by the _ agile _ @xcite and the _ fermi_lat @xcite gamma - ray telescopes , respectively ; then named @xcite , and @xcite . the accurate source location determined by _ fermi_lat allowed for the association of the unidentified gamma - ray source with a flat spectrum radio quasar ( fsrq ) included into the second edition of the roma - bzcat multi - frequency catalog of blazars @xcite . the optical counterpart of this source is , located at z=1.762 , and the radio counterpart is named . in the following sections , we refer to this object as , its radio source name . + here we present the results of an analysis of multifrequency data simultaneous to the _ agile _ campaign on the virgo field , and to the follow - up carried out after the _ fermi_lat detection and localization . by analysing the archival data , we estimate some fundamental physical properties of the source such as the accretion disk luminosity and the supermassive black hole ( smbh ) mass . in this way we can obtain a consistent picture of the source emission in the framework of leptonic models of blazars for periods of both low and high emission activity . in sections 2 and 3 we will describe the multi - wavelength campaigns related to this source . in section 4 we will report on the archival data . in section 5 we will present our results consisting in the determination of the bh mass , the gamma - ray light curve and spectrum , and the sed modeling . the _ agile_grid gamma - ray telescope ( operating in the 30 mev 50 gev energy range , see @xcite ) performed two observing campaigns of the virgo field . the first campaign included 3 observations from 2007 december 16 to 2008 january 8 . there were 3 simultaneous observations ( revolutions 633 , 635 , 637 ) with the wide - field instruments aboard the _ integral _ mission ( operating in optical , hard x- , and soft gamma - rays , see @xcite ) . during this campaign _ agile _ detected high gamma - ray activity from @xcite . in the following sections , we will refer to this time period as _ period a _ , and to the related multiwavelength campaign as _ campaign a_. + _ agile _ also observed the virgo field from 2009 june 4 to 2009 june 15 , but there were no simultaneous observations with wide - field instruments operating at other wavelengths . was undetected during this observation . + the _ fermi_lat gamma - ray telescope ( 20 mev 300 gev , see @xcite ) , operating in all - sky survey mode since 2008 august 4 , detected high gamma - ray activity at the end of december 2008 @xcite , and triggered optical - uv / x - ray observations with the swift satellite @xcite , starting from 2009 january 2 for a total exposure of 4.7 ks . optical v band photometry and polarization measurements were also made on - ground with the kanata telescope on 2009 january 2 and on 2009 january 3 @xcite . in the following sections , we will refer to the observations collected during this period as _ period b _ , and to the related multiwavelength campaign as _ campaign b_. a summary of the observations is provided in table [ tab : obs ] . [ cols="^,^,^",options="header " , ] the fit with the gamma - gamma absorption components results in no absorption from the heii complex , and weak absorption from the hi complex . we performed the f test on the two fits reported in table [ tab : sfit ] to test the hypothesis of the need of the absorption components . the f test gives a value of 0.15 , and an associated probability of @xmath087% , hence the absorption component is not necessary to fit the spectrum , suggesting that in the blazar zone is located in the outer low - ionization region of the blr or outside it . we constructed two spectral energy distributions referring to the high gamma - ray activity observed by _ agile _ and _ fermi _ ( campaigns a and b , respectively , as reported in table [ tab : obs ] ) with the data collected so far . the _ agile _ data are integrated for 4 days ( the duration of the last observation during the campaign a , from 2008 january 4 13:35 to 2008 january 8 11:06 ut ) . fermi_lat data for period b are integrated within 4 days around the flux peak ( 2008 december 29 16:00 ut ) to achieve acceptable statistics . these spectral energy distributions are shown in figure [ fig : sed_multiepochs ] , where we also included the archival data . + c + c the sed modeling has been performed in the framework of leptonic models . we assume that the emitting region is a spherical blob of radius @xmath11 , with bulk lorentz factor @xmath12 , an electron population distribution proportional to @xmath13 ( where @xmath14 is the lorentz factor of the electrons , ranging from @xmath15 to @xmath16 ) , and a randomly oriented magnetic field @xmath17 filling the dissipation region . the observer line of sight and the jet direction form an angle @xmath18 . we parametrize the external radiation fields as proposed by @xcite , where the key elements are the accretion disk luminosity and the distance ( @xmath19 ) of the emitting blob from the central bh . we consider a jet aperture @xmath20=0.1 , according to @xcite . in evaluating the power carried by protons , we assume one proton per emitting electron . we assumed that during the campaigns a and b , the accretion disk luminosity is the same as the one obtained from the sdss data in 2001 ( dashed curve in fig . [ fig : sed_multiepochs ] ) . we found in section 5.4 that the _ fermi_lat gamma - ray spectrum ( fig . [ fig : sed_multiepochs ] green data , one month integration ) does not show absorption from the blr . therefore , we tried to find solutions for the blob dissipating beyond the blr , without taking into account the gamma - gamma absorption from the blr . we obtained two possible solutions for the modeling . model parameters are reported in table [ tab : models ] , where we use @xmath21 ( @xmath22 ) the refer to the distance of the blr region ( of the dusty torus ) from the bh ; f@xmath23 ( f@xmath24 ) to the fraction of the disk emission that is reprocessed by the blr in lines ( by the dusty torus ) ; @xmath25 to the accretion efficiency ; @xmath26 to the electron lorentz factor for which the electron energy halves in the blob crossing time ( @xmath11/_c _ ) . + the first solution ( model 1 , top panel of figure [ fig : sed_multiepochs ] ) places the dissipation region ( r@xmath27 ) just beyond the blr ( [email protected] pc from the smbh ) , where the photon field from the dusty torus is almost equal to the blr seed photon field . at such a distance , the variability time - scale of a blob of radius r@xmath297@xmath30 cm is of the order @xmath03 days , qualitatively in agreement ( within a factor 2 ) with the 1-day binned gamma - ray light curve of the source . according to this modeling of the seed photon fields , the blr photon contribution is expected to fade over time while the torus contribution remains constant . both ssc and external compton ( ec ) with seed photons from the torus contribute to the soft x - ray emission , the hard x - ray is dominated by ec emission with seed photons from the torus , and the gev emission is dominated by ec with seed photons from the blr . in the mev up to gev range the two ec contributions are almost equal . further investigation of the variability is not possible , as no optical light - curve is available from which to trace the electron population from the synchrotron emission . moreover the evolution of the gamma - ray photon index with time ( reported in fig . [ fig : lc ] ) is statistically poor . + the second solution to the sed modeling ( model 2 ) has been found by assuming r@xmath287 pc from the smbh . such a model requires a r@xmath31 = @xmath32 cm , with a variability time - scale of the order of @xmath010@xmath33 days . we do not find such a variability time - scale in the gamma - ray light curve , but we observe that the activity period for the source lasts at least 11 weeks . we can argue that the dissipating region in the jet has a radius of @xmath32 cm . gradients of the electron energy density in the distance from the black hole , or disturbances in the jet ( see for examples @xcite ) can be responsible for the shorter variability observed in gamma - rays . with this model , the x - ray emission is due to ssc ; the gamma - ray emission is due to ec with seed photons from the torus . the direct contribution of accretion disk photons to the ec is negligible . + in the model for a blob dissipating just beyond the blr , the energy region below @xmath010@xmath34hz is self absorbed . on the contrary , in the model of a distant dissipation region , there is no self - absorption , thus allowing the direct observation of radio emission from the blob . in this case , the emission predicted by the model is slightly higher than the flux reported in fig . [ fig : sed_multiepochs ] ( bottom panel ) , but we note that no radio observations were performed during the gamma - ray flares . the mismatch in the variability time - scale for model 2 with r@xmath287 pc , results from the assumption that r@xmath31=@xmath35r@xmath27 . alternatively , we tried to model the sed assuming a blob of radius @xmath36 cm to satisfy the observed variability time - scale , and dissipating at @xmath05 pc from the smbh ( model 3 ) . model parameters are reported in the right column of table [ tab : models ] , and the model is plotted in fig . [ fig : sed_multiepochs2 ] . + a similar kind of solution has been suggested by @xcite to explain faster variability ( of the order of 10 min ) for pks 1222 + 216 . this model can correspond to a reconnection event , causing the acceleration of a small portion of the jet @xcite , or to recollimation by the interaction with the external medium @xcite . + according to this sed modeling , the x - ray emission is due to ssc , and the gamma - ray emission to ec with external photons from the torus . l | c | c | c & model 1 & model 2 & model 3 + r@xmath37 & 0.22 & 6.8 & 4.8 + blob radius ( cm ) & 6.7@xmath30 @xmath38 & 2.1@xmath39 @xmath38 & 1@xmath40 + m@xmath41 ( m@xmath42 ) & + l@xmath43 ( erg / s ) & + r@xmath23 ( cm ) & + r@xmath44 ) & + f@xmath23 & + f@xmath45 & + @xmath25 & + @xmath12 & 20 & 20 & 20 + angle of view ( deg ) & 2 & 2 & 2 + @xmath15 & 1 & 1 & 1 + @xmath16 & 3.9@xmath46 & 3.4@xmath47 & 1.3@xmath47 + @xmath48 & 0.95@xmath46 & 1@xmath46 & 1@xmath46 + density at @xmath48 & 3.0@xmath49 & 1.5@xmath50 & 9.6@xmath51 + ( @xmath52 ) + s@xmath53 & 1.1 & 0.5 & 1.3 + s@xmath54 & 3.1 & 3.3 & 2.5 + b ( gauss ) & 6.1@xmath55 & 1.1@xmath49 & 7.6@xmath49 + @xmath26 & 60 & 1.1@xmath47 & 2.4@xmath46 + electron power & 2.2@xmath56 & 4.5@xmath57 & 1.1@xmath57 + ( erg / s ) + magnetic power & 2.5@xmath56 & 7.9@xmath58 & 8.6@xmath59 + ( erg / s ) + proton power @xmath60 & 1.1@xmath61 & 1.6@xmath61 & 3.0@xmath61 + ( erg / s ) + radiated power & 3.1@xmath56 & 2.5@xmath56 & 2.1@xmath56 + ( erg / s ) + + @xmath38 r@xmath31=@xmath35r@xmath27 in this model . + @xmath60 we assume one proton per emitting electron . we rarely have the opportunity to detect the disk emission in fsrqs , which are generally overwhelmed by synchrotron jet emission ( see * ? ? ? * for example ) . however , our study here suggests that we have detected accretion - disk - dominated emission in . granted , we can not fully exclude the possibility that the archival sdss and galex observations we have reported could be interpreted as other emission mechanisms than thermal disk emission because we lack strictly simultaneous radio observations and extended radio light curves to corroborate the assumption of low jet emission . however the assumption of disk emission remains , in our opinion , the most likely explanation for the observations . + the optical observations of revealed an optical flux enhancement of a factor 1530 in 6 years , signifying a shift from accretion - disk to synchrotron , jet - dominated emission . the optical spectrum obtained in the period of faint optical emission allowed the classification of the source as a fsrq in bzcat . we made an estimate of the smbh mass of and the accretion rate from a period of low jet activity . with these estimates , we were able to study the december 2008 flare . modeling the observed flat gamma - ray spectrum and sed also allowed for an investigation into the location of the blazar - zone of the object . + as a final remark , it is worth stressing two major points . first , by definition , our estimate of r@xmath27 is model dependent . the location of the dissipation region was estimated assuming the parametrization proposed by @xcite for the blr and the torus contribution to seed photons for the ec . according to this parametrization , it was possible to derive r@xmath27 from the luminosity ratio of synchrotron to ec emission . in fact , in the parametrization by @xcite for a dissipation region outside the blr , the seed photons for ec fade with distance from the smbh . we obtained two solutions . referring to figure 2 of @xcite , the ratio @xmath62 equals the ratio of the optical to gamma - rays luminosity at the two values of the r@xmath27 ( assuming knowledge of the magnetic energy density ) . for the flares of reported in this article , one solution ( model 2 ) places the blob at r@xmath287 pc , with only the dusty torus as the origin of seed photons . the other solution ( model 1 , with [email protected] pc ) corresponds to just outside the blr , where the contribution of seed photons from both the dusty torus and the blr are relevant . the magnetic field is constrained in the sed modeling by the cutoff of synchrotron emission in the uv ( due to the last and most energetic electrons ) , and by the corresponding cutoff of ec which we can not derive directly from data ( that give non - constraining upper limits at e@xmath320 gev , see fig . [ fig : sed_multiepochs ] ) . assuming the thomson regime , and with only one external photon field contributing to the ec , the ratio _ f_@xmath63 between the synchrotron cutoff energy and the ec cutoff energy is proportional to @xmath64 , where @xmath65 is the typical seed photon field energy . hence if we have constraining data at the highest energy , and with a specific geometry in the model , we can constrain b/@xmath12 . however , the geometry in the model also constrains @xmath66 , hence in the ideal case we can obtain b/@xmath12 and r@xmath27 from the sed modeling . the ratio of synchrotron to ssc luminosity further constrains the model parameters @xmath67 , allowing one to remove the degeneracy between @xmath12 and b. in reality , we can obtain only upper limits of _ f_@xmath63 because the data at higher energies are non - constraining . so we have only upper limits on @xmath68 for each model . as a consequence , the @xmath66 could be lower than in our parametrization ( we have to maintain the ratio @xmath69 at the desired value ) . this implies that r@xmath27 could be higher than our evaluations . + the second point is that the photon field intensity is proportional to the accretion disk luminosity , and the blr and torus location is proportional to @xmath70 . in all our estimations , we assume that the disk luminosity is almost steady over time , e.g. , in the low state observed during the sloan survey in 2002 , during the galex observation in 2007 , and during the gamma - ray flares observed by _ agile _ at the beginning of 2008 and by _ fermi_lat at the end of 2008 . this assumption could be false . in this case , we observe that the parametrization of @xmath71 and @xmath72 reported in equation ( 20 ) by @xcite remains unchanged while varying l@xmath43 , provided that we scale the solution for @xmath19 with @xmath73 . therefore , variations of the disk luminosity in time , and/or systematic errors in the evaluation of disk luminosity from our sdss+galex data ( possibly biased by jet emission ) only slightly affect our estimation of @xmath19 . + the starting point of our modeling is that the emission region is far from the smbh ( at parsec scale ) , and we motivate this choice with the flat gamma - ray spectrum up to energies of 15 gev . for a different approach and results for other blazars , we refer readers to the work of @xcite . they performed a detailed study of the localization of the emission region for bright blazars making use of the variability timescale for objects showing a spectral cut - off at ( 1020)/(1+z ) gev . they obtained that the variability timescale and spectra are in agreement with a dissipation region inside the blr . + we note that , contrary to the model we used , some authors ( e.g. , * ? ? ? * for example ) model the seds of both fsrq and bl lacs with pure synchrotron + synchrotron self compton components only . + model 1 ( @xmath74 0.2 pc ) , gives a variability time scale of the order of 3 days , in agreement with the flare duration estimated from the 1-day binned gamma - ray light curve . this model , however , does not properly reproduce the flat gamma - ray spectrum . in particular , in order to reproduce the observed flux at energies @xmath3 10 gev , it overestimates the spectrum for lower energies . on the contrary , model 2 ( @xmath747 pc ) reproduces the flat - gamma - ray spectrum , but it is not clear whether the predicted variability of the order of @xmath0100 days can be associated with the duration of the gamma - ray activity period estimated from the 1-week/1-month binned gamma - ray light curves . the third model has been built by relaxing the relation @xmath11=0.1@xmath19 in order to preserve the variability time - scale estimated from the 1-day gamma - ray light curve ( we follow the solution proposed by @xcite for pks 1222 + 216 ) , and it still reproduces the gamma - ray spectrum . interestingly , the size of the emitting region for pks 1222 + 216 ( @xmath75 cm ) and for differ significantly . + for the third model , we obtain @xmath11=0.0067@xmath19 , in agreement within a factor of two with the prediction of @xcite , that gives @xmath76 , for the case of efficient conversion of bulk luminosity in radiation in the strong focusing scenario . with the same assumptions , @xcite assume that the location of the emitting region is at @xmath742.5 @xmath77erg s@xmath78 pc@xmath79 pc from the smbh , where @xmath80 is the jet power . if we invert this relation , and we make use of our result ( @xmath81 pc ) , we obtain @xmath82 . we must assume that the proton to emitting electron ratio is of the order of 0.1 in order to reproduce such a power ( in the evaluation of proton power reported in table [ tab : models ] we assumed one proton per emitting electron , instead ) . we note , however , that @xcite evaluated that efficient radiative conversion could be assumed if the product of the bulk lorentz factor by the opening angle is @xmath833 , and according to our third model this product is 2 . the authors wish to acknowledge the anonymous referee for his suggestions . we are also very grateful to j. finke for comments and corrections , and for his careful proofreading of several sections of this manuscript . + we acknowledge financial contribution from the agreement asi - inaf i/009/10/0 . + the _ agile _ mission is funded by the italian space agency ( through contract asi i/089/06/2 ) with scientific and programmatic participation by the italian institute of astrophysics ( inaf ) and the italian institute of nuclear physics ( infn ) . + this research has made use of data from the mojave database that is maintained by the mojave team @xcite . + this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . + this research has made use of the simbad database , operated at cds , strasbourg , france . + rja is supported by an appointment to the nasa postdoctoral program at the jet propulsion laboratory , administered by oak ridge associated universities through a contract with nasa . + the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis . these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden . additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france . + 10 abdo a. et al . , 2010 , apjl , 714 , l73-l78 ; adelman mccharty j. k. , et al . , 2008 , apjs , 175 , 297 ; agudo i. et al . , 2011 , apjl , 726 , 13 ; agudo i. et al . , 2011 , apjl , 735 , 10 ; assef r. j. , et al . , apj in press ; arxiv:1009.1145 ; atwood et al . , 2009 , apj , 697 , 1071 - 1102 ; bottcher m. , ap&ss , 2007 , 309 , 95 - 104 ; bromberg o. & levinson a. , 2009 , apj , 699 , 1274 - 1280 ; burrows d. n. , et al . 2005 , space sci . , 120 , 165 ; j. m. casandjian , i. a. grenier , 2008 , a&a , 489 , 849 ; cattaneo p. w. et al . , 2011 , nima , 630 , 251 - 257 ; collin s. et al . , 2006 , a&a , 456 , 75 ; dickey j. m. & lockman f. j. , 1990 , ara&a 28 , 215 - 261 ; denney k. d. , et al . , 2009 , apj 692 , 246 ; denney k. d. , et al . , 2011 , pos ( nlsy1 ) , 34 ; esposito j. a. et al . , 1999 , apjs , 123 , 203 - 271 ; finke j. , et al . , 2010 , apj , 712 , 238 - 249 ; fitzpatrick e. l. , 1999 , pasp 111 , 63 - 75 ; gehrels n. , et al . 2004 , apj , 611 , 1005 ; ghisellini g. et al . 2007 , mnras , 382 , l82-l86 : ghisellini g. @xmath84 tavecchio f. , 2009 , mnras , 397 , 985 - 1002 ; ghisellini g. et al . , 2010 , mnras , 402 , 497 ; ghisellini g. et al . , 2011 , mnras , 411 , 911 ; giannios d. et al . , 2010 , mnras , 402 , 1649 - 1656 ; giommi p. et al . , 2011 , a&a submitted ; arxiv:1108.1114v1 ; hartman r. c. , et al . , 1999 , apjs , 123 , 79 - 202 ; ikejiri y. , 2009 , atel 1892 ; ivezic . , et al . , 2007 , aspc , 364 , 165 - 175 ; jester s. , 2005 , aj 130 , 873 ; jorstad s. g. et al . , 2010 , apj , 715 , 362 - 384 ; lister m. l. , et al . , 2009 , aj , 137 , 3718 ; maraschi l. , ghisellini g. , and celotti a. , 1992 , apjl , 397 , l5-l9 ; marscher a. p. , and bloom s. d. , 1992 , proceedings of the compton observatory science workshop , 346 - 353 ; marscher a. p. , et al . , 2010 , apjl , 710 , 726 ; martin d. c. et al . , 2005 , apjl , 619 , l1-l6 ; mas - hesse j. m. , et al . 2003 , a&a , 411 , l261 ; massaro e. , et al . , 2010 ; arxiv:1006.0922 ; mattox j. r. , bertsch , d. l. , chiang , j. , et al . , 1996 , apj , 461 , 396 - 407 ; mcke a. , protheroe r. j. , 2001 , astropart . phys , 15 , 121 - 136 ; mcke a. et al , 2003 , astropart . phys , 18 , 593 - 613 ; nalewajko k. & sikora m. , 2009 , mnras 392 , 1205 ; nolan p. l. et al . , 2012 , apjs , 199 , 31 ; pacciani l. , et al . , 2009 , a&a , 494 , 49 ; pian e. et al . , 1999 , apjl , 521 , 112 ; pittori c. , , et al . , 2009 , a&a , 506 , 1563 ; poole t. s. , breeveld a. a. , page m. j. , et al . , 2008 , mnras , 383 , 627 poutanen j & stern b. , 2010 , apjl , 717 , 118 ; richards g. t. , et al . , 2002 , aj , 124 , 1 ; roming p. w. a. , et al . 2005 , space sci . , 120 , 95 ; shakura n. i. & sunyaev r. a. , 1973 , a&a , 24 , 337 ; shaw m. s. , et al . , 2012 , apj , in press ; sikora m. , begelman m. c. , and rees m. , 1994 , apj , 421 , 153 ; sikora m. , et al . , 2008 , apj , 675 , 71 - 78 ; tavani , m. , barbiellini , g. , argan , a. , et al . , 2009 , a&a , 502 , 995 - 1013 tavecchio f. @xmath84 mazin d. , 2009 , mnras , 392 , l40-l44 ; tavecchio f. et al . , 2010 , mnras , 405,l94-l98 ; tavecchio f. et al . , 2011 , a&a , 534 , a86 ; terasranta h. et al . , 2005 , a&a , 440 , 409 ; tramacere a. , et al . , 2009 , atel . 1888 ; ubertini p. , et al . , 2003 , a&a , 411 , l131-l139 ; urry c. m. , and padovani p. , 1995 , pasp , 107 , 803 ; verrecchia f. et al . , 2011 , asr in press , doi:10.1016/j.asr.2011.05.035 ; vestergaard m. , peterson b. m. , 2006 , apj , 641 , 689 ; watanabe m. , et al . 2005 , pasp , 117 , 870 - 884 ; winkler c , et al . 2003 , a&a , 411 , l1 .
in 2008 _ agile _ and _ fermi _ detected gamma - ray flaring activity from the unidentified egret source , recently associated with a flat spectrum radio quasar ( ) at z=1.762 . the optical counterpart of the gamma - ray source underwent a flux enhancement of a factor 15 - 30 in 6 years , and of @xmath010 in six months . we interpret this flare - up in terms of a transition from an accretion - disk dominated emission to a synchrotron - jet dominated one . + we analysed a sloan digital sky survey ( sdss ) archival optical spectrum taken during a period of low radio and optical activity of the source . we estimated the mass of the central black hole using the width of the civ emission line . in our work , we have also investigated sdss archival optical photometric data and uv galex observations to estimate the thermal - disk emission contribution of . + our analysis of the gamma - ray data taken during the flaring episodes indicates a flat gamma - ray spectrum , with an extension of up to 15 gev , with no statistically - relevant sign of absorption from the broad line region , suggesting that the blazar - zone is located beyond the broad line region . this result is confirmed by the modeling of the broad - band spectral energy distribution ( well constrained by the available multiwavelength data ) of the flaring activity periods and by the accretion disk luminosity and black hole mass estimated by us using archival data . [ firstpage ] galaxies : active - galaxies : quasars : general - galaxies : individual : - radiation mechanism : non thermal
understanding the nature of large - scale structure in the universe is a key component of the field of cosmology and is vital to studies of galaxy formation and evolution . the clustering of galaxies reflects the distribution of primordial mass fluctuations present in the early universe and their evolution with time and also probes the complex physics which governs the creation of galaxies in their host dark matter potential wells . since the first redshift surveys , the two - point correlation function , , has been used as a measure of the strength of galaxy clustering @xcite . is relatively straightforward to calculate from pair counts of galaxies , and it has a simple physical interpretation as the excess probability of finding a galaxy at a separation @xmath18 from another randomly - chosen galaxy above that for an unclustered distribution @xcite . locally , follows a power - law , @xmath19 , on scales @xmath20 with @xmath21 @xcite . the scale - length of clustering , , is the separation at which the probability of finding another galaxy is twice the random probability . locally is measured to be @xmath22 for optically - selected galaxies but depends strongly on galaxy morphology , color , type and luminosity @xcite . the spatial clustering of galaxies need not trace the underlying distribution of dark matter . this was first discussed by @xcite in an attempt to reconcile the different clustering scale lengths of field galaxies and rich clusters , which can not both be unbiased tracers of mass . the galaxy bias , @xmath23 , is a measure of the clustering in the galaxy population relative to the clustering in the underlying dark matter distribution . it can be defined as the square root of the ratio of the two - point correlation function of the galaxies relative to the dark matter : @xmath24 , either as a function of @xmath18 or defined at a specific scale ( see section 5.1 ) . observations of galaxy clustering have shown that the galaxy bias can be a function of morphology , type , color , luminosity , scale and redshift . using galaxy morphologies , @xcite find that early - type galaxies in the stromlo - apm redshift survey are much more strongly clustered than late - type galaxies . their early - type sample has a larger correlation length , , and a steeper slope than late - type galaxies . however , @xcite show using data from the southern sky redshift survey ( ssrs2 , da costa et al . 1998 ) that in the absence of rich clusters early - type galaxies have a relative bias of only @xmath25 compared with late - type galaxies . in their sample , red galaxies with @xmath26 are significantly more clustered than blue galaxies , with a relative bias of @xmath27 . @xcite also studied galaxy clustering as a function of color , using data from the sloan digital sky survey ( sdss , york et al . 2000 ) early data release , and find that red galaxies ( @xmath28 ) have a larger correlation length , , a steeper correlation function , and a larger pairwise velocity dispersion than blue galaxies . they also find a strong dependence of clustering strength on luminosity for magnitudes ranging from @xmath29 to @xmath30 . galaxy clustering for different spectral types in the 2df galaxy redshift survey ( 2dfgrs , colless et al . 2001 ) is reported by @xcite ; absorption - line galaxies are shown to have a relative bias @xmath31 times that of emission - line galaxies on scales @xmath32 , declining to unity on larger scales . absorption - line galaxies have a steeper correlation slope and a larger pairwise velocity dispersion . all of these results indicate that red , absorption - line , early - type galaxies are found predominantly in the more massive virialized groups and clusters in which the random velocities are large . @xcite report that the correlation length of optically - selected galaxies in the 2dfgrs depends weakly on luminosity for galaxies fainter than @xmath33 , the typical luminosity of a galaxy , but rises steeply with luminosity for brighter galaxies , with the most luminous galaxies being three times more clustered than @xmath33 galaxies . these results from local @xmath34 surveys indicate that the strength of galaxy clustering is quite sensitive to different galaxy properties . a critical test of both cosmological and galaxy evolution models is the redshift - dependence of galaxy clustering . the evolution of the _ dark matter _ two - point correlation function , @xmath35 , can be calculated readily and is strongly dependent on cosmology . in high - density models the clustering strength grows rapidly , while models show a more gradual evolution ( e.g. , jenkins et al . 1998 , ma 1999 ) . however , the evolution of the _ galaxy _ two - point correlation function , @xmath36 , depends on the evolution of both the underlying dark matter distribution and the galaxy bias , which is expected to increase with redshift . applying semi - analytic modelling of galaxy formation and evolution to dark matter simulations , @xcite present models with @xmath37 for the galaxy distribution at @xmath38 compared to @xmath39 locally . they predict a galaxy bias of @xmath40 at @xmath38 for galaxies with @xmath41 log @xmath42 but also find that the galaxy bias can be a strong function of luminosity , star - formation rate , galaxy type , and sample selection . @xcite , who also apply semi - analytic modelling to dark matter simulations , predict a bias of @xmath43 at @xmath38 for galaxies with @xmath44 log @xmath42 . they also predict a similar morphology - density relation at @xmath38 to that seen locally . previous redshift surveys which have attempted to probe intermediate redshifts from @xmath45 have been hampered by small volumes and the resulting severe cosmic variance . results from the canada - france redshift survey ( cfrs , lefevre et al . 1996 ) are based on @xmath46 galaxies covering 0.14 degrees@xmath47 , the norris redshift survey @xcite sparsely samples 20 degrees@xmath47 with a survey of @xmath48 galaxies , @xcite report on a sample of @xmath49 galaxies in two very small fields , including the hubble deep field , and @xcite present a survey of @xmath50 galaxies in a total area of 27 arcmin@xmath47 , finding that correlations found in their @xmath51-band data are generally greater than those found by optically - selected surveys . @xcite compare results from several surveys which have measured the correlation length in the range @xmath45 and illustrate well the uncertainties in and discrepancies between these results . for an open cdm cosmology , the estimates of the comoving correlation length vary from @xmath52 at @xmath53 . in particular , the cfrs survey found a much smaller correlation length at @xmath54 than the other surveys , which generally are consistent with weak evolution between @xmath38 and @xmath55 . a significantly larger survey was undertaken by cnoc2 @xcite , who obtained redshifts for @xmath56 galaxies over 1.44 degrees@xmath47 . most relevant for our purposes may be @xcite , who present clustering results for a deep @xmath57 sample of @xmath48 galaxies covering a total of 42.5 arcmin@xmath47 in 5 fields ; they quote a correlation length of @xmath58 for a cosmology , implying that their galaxy sample is an unbiased tracer of the mass at @xmath10 . however , many of these surveys cover very small fields and are likely to underestimate the true clustering . there is a well - known systematic bias towards underestimation of in volumes which are small enough that all galaxies are part of a single large - scale structure and in which the large - scale modes can not be sampled . furthermore , cosmic variance will dominate any measure of clustering in volumes which are too small to be representative samples of the universe @xcite . here we present early results on galaxy clustering in the deep2 galaxy redshift survey @xcite , an @xmath59band selected survey which was designed to study the universe at @xmath10 with a volume and sampling density comparable to local surveys . our intent in this paper is to provide an initial measure of the galaxy clustering in our survey at using the first season of data and to investigate the dependence of the clustering on galaxy properties , splitting the sample by color , spectral type , and luminosity . to constrain galaxy evolution models , we measure the galaxy bias for the sample as a whole and the relative bias between subsamples this is the first of several planned papers on galaxy clustering within the deep2 survey , and here we focus strictly on analysis of spatial correlations . discussion of redshift - space distortions will appear in a subsequent paper @xcite . in the data from the first season of observations we measured 5042 redshifts with @xmath60 in three fields with a total area of 0.72 degrees@xmath47 . the most complete field currently covers 0.32 degrees@xmath47 and includes 2219 galaxies in the redshift range @xmath0 , which we use as the primary data sample in this paper . the outline of the paper is as follows : in section 2 we briefly describe the survey , provide details of the observations , data reduction , and the data sample used here . section 3 outlines the methods used in this paper , while section 4 presents our results , both for the survey sample as a whole and for subsamples based on galaxy redshift , color , spectral type , and luminosity . in section 5 we discuss galaxy bias and the relative biases between our subsamples , and we conclude in section 6 . the deep2 galaxy redshift survey is a three - year project using the deimos spectrograph @xcite on the 10-m keck ii telescope to survey optical galaxies at @xmath10 in a comoving volume of approximately 6@xmath6110@xmath62 @xmath63 mpc@xmath64 . the completed survey will cover 3.5 degrees@xmath47 of the sky over four widely separated fields to limit the impact of cosmic variance . the `` 1-hour - survey '' ( 1hs ) portion of the deep2 project will use @xmath65 hr exposure times to measure redshifts for @xmath66 galaxies in the redshift range @xmath67 to a limiting magnitude of @xmath1 ( all magnitudes in this paper are in the ab system ; oke & gunn 1983 ) . photometric data were taken in @xmath68 and @xmath69-bands with the cfh12k camera on the 3.6-m canada - france - hawaii telescope . galaxies selected for spectroscopy must additionally meet a color selection given approximately by @xmath70 , @xmath71 , or @xmath72 . this simple color - cut was designed to select galaxies at @xmath73 ( details are given in newman et al . 2004 ) and has proven effective in doing so . as discussed in @xcite this color - cut results in a sample with @xmath7490% of the objects at @xmath73 , missing only @xmath745% of the @xmath73 galaxies . each of the four deep2 1hs fields corresponds to a volume of comoving dimensions @xmath75 in a model at a redshift of @xmath38 . to convert measured redshifts to comoving distances along the line of sight we assume a flat cosmology with @xmath76 and @xmath77 . changing cosmological models within the range allowed by recent wmap analysis @xcite has only a modest influence on our results . we use @xmath78 , and we quote correlation lengths , , in comoving dimensions of . this paper uses data from the first observing season of the 1hs portion of the deep2 survey , from august october 2002 . three of the four deep2 fields were observed with a total of 68 custom - made slitmasks . each mask has on the order of @xmath79 slitlets , with a median separation in the spatial direction between targeted galaxies of @xmath80 , and a minimum of 3 . due to the high source density of objects , we are able to obtain spectra for @xmath81 of our targets . three 20-minute exposures were taken on the deimos spectrograph with a 1200 line mm@xmath82 grating for each slitmask , covering a spectral range of @xmath83 at an effective resolution @xmath84 . the multiple exposures allow us to robustly reject cosmic rays from the data . many of the slitlets in each mask are tilted to align with the major axis of the target galaxy to enable internal kinematic studies , and as a result we do not dither the telescope between exposures . the data were reduced using a sophisticated idl pipeline developed at uc - berkeley , adapted from spectroscopic reduction programs developed for the sdss @xcite . to find the redshift of each galaxy , a @xmath85-minimization is used , where the code finds minima in @xmath85 between the observed spectrum and two templates ; one is an artificial emission - line spectrum convolved with a broadening function to mimic a 1@xmath86 slit and 60 internal dispersion . the other template is a high signal - to - noise ratio absorption - dominated spectrum which is the average of many thousands of sdss galaxies covering a rest wavelength range of 2700 - 9000 @xcite . the five most - likely redshifts are saved and used in a final stage where the galaxy redshift is confirmed by human interaction . our overall redshift success rate is @xmath8770% and displays only minor variation with color and magnitude ( @xmath88% ) , with the exception of the bluest galaxies ( @xmath89 ) for which our redshift success rate is @xmath90% . these galaxies represent @xmath91% of our targeted sample and account for @xmath92% of our redshift failures . the @xmath933727 [ oii ] doublet redshifts out of our spectral range at @xmath94 , and it is believed that all of our bluest ( @xmath89 ) targeted galaxies for which we do not measure a redshift lie beyond this range . these galaxies have similar colors and source densities as the population at @xmath95 currently studied by c. steidel and collaborators ( private communication ) . if these galaxies were in our observable redshift window , it is almost certain that we would have measured a redshift , given that these blue galaxies must have recent star - formation and therefore strong emission lines . although the instrumental resolution and photon statistics of our data would suggest that we could achieve a redshift precision of @xmath96 in the rest frame of each galaxy , we find using galaxies observed twice on overlapping slitmasks that differences in the position or alignment of a galaxy within a slit and internal kinematics within a galaxy lead to an effective velocity uncertainty of @xmath97 . here we present results from only the most complete field , centered at 02 hr 30 min + 00 deg , for which we have observed 32 slitmasks covering @xmath98 degrees by @xmath99 degrees on the sky . we use data only from masks which have a redshift success rate of 60% and higher in order to avoid systematic effects which may bias our results . figures [ spatial ] and [ windowf ] show the spatial distribution of galaxies on the plane of the sky and the window function for this field . the observed slitmasks overlap each other in two horizontal rows on the sky . six of the masks have not as yet been observed in this pointing , leading to regions with lower completeness . while we measure redshifts as high as @xmath100 , for this paper we include only galaxies with @xmath101 , a range in which our selection function is currently well defined . our sample in this field and range contains 2219 galaxies , with a median redshift of @xmath102 . at this median redshift the typical rest - frame wavelength coverage is @xmath103 . figure [ zhist ] shows the overall redshift distribution of galaxies with @xmath104 in all three of our observed fields . there is a rise between redshifts @xmath105 , the result of our probabilistic pre - selection of spectroscopic targets expected to have redshifts @xmath106 . the flux limit of our sample results in the slow decrease of the observed objects at higher redshifts ; smaller - scale variations are due to galaxy clustering . in order to compute galaxy correlation statistics , we must understand our selection function @xmath107 , defined as the relative probability at each redshift that an object will be observed in our sample . in general , the selection function can depend on redshift , color , magnitude , and other properties of the galaxy population and survey selection . ideally one would compute @xmath107 from the luminosity function of galaxies in the survey . for this initial study we estimate @xmath107 by smoothing the observed redshift histogram of all the galaxies in our sample , taking into account the change in volume with redshift . we smoothed with a boxcar of width 450 and then used an additional boxcar of width 150 to ensure that there were no residual bumps due to large - scale structure . the resulting @xmath107 is shown by the solid line in figure [ zhist ] . also shown in this figure are the normalized selection functions for the emission - line and absorption - line samples discussed later in the paper ( see section 4.4 ) . note that the redshift distribution @xmath107 is determined using galaxies in all three of our observed fields , not only in the field for which we measure , which reduces effects due to cosmic variance . use of a preliminary @xmath107 constructed from the luminosity function of our sample does not change the results presented here . using mock catalogs to test the possible systematic effects due to our estimation of @xmath107 , we find that the resulting error on is 5% , signficantly less than that due to cosmic variance . the two - point correlation function is a measure of the excess probability above poisson of finding a galaxy in a volume element @xmath108 at a separation @xmath18 from another randomly chosen galaxy , @xmath109 dv,\ ] ] where @xmath110 is the mean number density of galaxies . to measure one must first construct a catalog with a random spatial distribution and uniform density of points with the same selection criteria as the data , to serve as an unclustered distribution with which to compare the data . for each data sample we create a random catalog with initially @xmath111 times as many objects with the same overall sky coverage as the data and uniform redshift coverage . this is achieved by applying the window function of our data , seen in figure 1 , to the random catalog . our redshift completeness is not entirely uniform across the survey ; some masks are observed under better conditions than others and therefore yield a higher success rate . this spatially - varying redshift success completeness is taken into account in the window function . we also mask the regions of the random catalog where the photometric data had saturated stars and ccd defects . finally , we apply our selection function , @xmath107 , so that the random catalog has the same overall redshift distribution as the data . this results in a final random catalog which has @xmath112 times as many points as the data . we measure the two - point correlation function using the @xcite estimator , @xmath113,\ ] ] where @xmath114 , and @xmath115 are pair counts of galaxies in the data - data , data - random , and random - random catalogs , and @xmath116 and @xmath117 are the mean number densities of galaxies in the data and random catalogs . this estimator has been shown to perform as well as the hamilton estimator @xcite but is preferred as it is relatively insensitive to the size of the random catalog and handles edge corrections well @xcite . as we measure the redshift of each galaxy and not its distance distortions in @xmath118 are introduced parallel to the line of sight due to peculiar velocities of galaxies . on small scales , random motions in groups and clusters cause an elongation in redshift - space maps along the line - of - sight known as `` fingers of god '' . on large scales , coherent infall of galaxies into forming structures causes an apparent contraction of structure along the line - of - sight @xcite . while these distortions can be used to uncover information about the underlying matter density and thermal motions of the galaxies , they complicate a measurement of the two - point correlation function in real space . instead , what is measured is , where @xmath119 is the redshift - space separation between a pair of galaxies . in order to determine the effects of these redshift - space distortions and uncover the real - space clustering properties , we measure @xmath118 in two dimensions , both perpendicular to and along the line of sight . following @xcite , we define @xmath120 and @xmath121 to be the redshift positions of a pair of galaxies , @xmath122 to be the redshift - space separation ( @xmath123 ) , and @xmath124(@xmath125 to be the mean distance to the pair . we then define the separation between the two galaxies across ( @xmath126 ) and along ( @xmath127 ) the line of sight as @xmath128 @xmath129 in applying the @xcite estimator , we therefore compute pair counts over a two - dimensional grid of separations to estimate . in measuring the galaxy clustering , one sums over counts of galaxy pairs as a function of separation , normalizing by the counts of pairs in the random catalog . while is not a function of the overall density of galaxies in the sample , if the observed density is not uniform throughout the sample then a region with higher density will contribute more to the total counts of galaxy pairs , effectively receiving greater weight in the final calculation . the magnitude - limit of our survey insures that our selection function , @xmath107 , is not flat , especially at the higher redshift end of our sample , as seen in figure 2 . to counteract this , one might weight the galaxy pairs by @xmath130 , though this will add significant noise where @xmath107 is low . what is generally used instead is the @xmath131 weighting method @xcite , which attempts to weight each volume element equally , regardless of redshift , while minimizing the variance on large scales . using this weighting scheme , each galaxy in a pair is given a weight @xmath132 @xmath133 where @xmath134 is the redshift of the galaxy , @xmath135 is the redshift - space separation between the galaxy and its pair object , @xmath136 , @xmath107 is the selection function of the sample , such that the mean number density of objects in the sample is @xmath137 for a homogeneous distribution , and @xmath131 is the volume integral of . we limit @xmath138 20 , the maximum @xmath126 separation we measure , in order to not over - weight the larger scales , which would lead to a noisier estimate of . note that the weighting depends on the integral over , a quantity we want to measure . ideally one would iterate the process of estimating and using the measured parameters in the @xmath131 weighting until convergence was reached . here we use a power - law form of @xmath139 , with initial parameters of @xmath140 and @xmath141=1.5 . these power - law values are in rough accordance with as measured in our full sample . as tests show that the measured is quite insensitive to the assumed values of and @xmath141 , we do not iterate this process . we estimate @xmath116 to be @xmath142 mpc@xmath143 from the observed number density of galaxies in our sample in the redshift range @xmath144 . as with and @xmath141 , we find that the results are not sensitive to the exact value of @xmath116 used . while can be directly calculated from pair counts , it includes redshift - space distortions and is not as easily interpreted as , the real - space correlation function , which measures only the physical clustering of galaxies , independent of any peculiar velocities . to recover we use a projection of along the @xmath126 axis . as redshift - space distortions affect only the line - of - sight component of , integrating over the @xmath127 direction leads to a statistic , which is independent of redshift - space distortions . following @xcite , @xmath145 where @xmath146 is the real - space separation along the line of sight . if is modelled as a power - law , @xmath2 , then and @xmath141 can be readily extracted from the projected correlation function , , using an analytic solution to equation [ eqn ] : @xmath147 where @xmath148 is the usual gamma function . a power - law fit to will then recover and @xmath141 for the real - space correlation function , . in practice , equation [ eqn ] is not integrated to infinite separations . here we integrate to @xmath149 , which includes most correlated pairs . we use analytic calculations of a broken power - law model for which becomes negative on large scales that we are underestimating by less than @xmath742% by not integrating to infinity . when observing with multi - object slitmasks , the spectra of targets can not be allowed to overlap on the ccd array ; therefore , objects that lie near each other in the direction on the sky that maps to the wavelength direction on the ccd can not be simultaneously observed . this will necessarily result in under - sampling the regions with the highest density of targets on the plane of the sky . to reduce the impact of this bias , adjacent slitmasks are positioned approximately a half - mask width apart , giving each galaxy two chances to appear on a mask ; we also use adaptive tiling of the slitmasks to hold constant the number of targets per mask . in spite of these steps , the probability that a target is selected for spectroscopy is diminished by @xmath91% if the distance to its second nearest neighbor is less than 10 arcseconds ( see davis et al . 2004 for details ) . this introduces a predictable systematic bias which leads to underestimating the correlation strength on small scales . some previous surveys have attempted to quantify and correct for effects of this sort using the projected correlation function , @xmath150 , of the sample selected for spectroscopy relative to the that of the entire photometric sample @xcite . other surveys have attempted to correct these effects by giving additional weight to observed galaxies which were close to galaxies which were not observed or by restricting the scales on which they measure clustering @xcite . it is not feasible for us to use measures of @xmath150 , as the line - of - sight distance that we sample is large ( @xmath1511000 ) and the resulting angular correlations projected through this distance are quite small . in addition , the relation between the decrease in the 2-dimensional angular correlations and the 3-dimensional real - space correlations is not trivial and depends on both the strength of clustering and the redshift distribution of sources . in order to measure this bias , we have chosen to use mock galaxy catalogs which have similar size , depth , and selection function as our survey and which simulate the real - space clustering present in our data . we have constructed these mock catalogs from the gif semi - analytic models of galaxy formation of @xcite . as described in @xcite , we use outputs from several epochs to create six mock catalogs covering the redshift range @xmath152 . to convert the given comoving distance of each object to a redshift we assumed a cosmology ; we then constructed a flux - limited sample which has a similar source density as our data . @xcite presents the selection function and clustering properties of these mock catalogs . to quantify the effect of our slitmask target selection on our ability to measure the clustering of galaxies , we calculate and in these mock catalogs , both for the full sample of galaxies and for a subsample that would have been selected to be observed on slitmasks . the projected correlation function , , of objects selected to be on slitmasks is lower on scales @xmath153 and higher on scales @xmath154 than the full catalog of objects . we find that as measured from is overestimated by 1.5% in the targeted sample relative to the full sample , while @xmath141 is underestimated by 4% . thus our target selection algorithm has a relatively small effect on estimates of the correlation strength that is well within the expected uncertainties due to cosmic variance . we do not attempt to correct for this effect in this paper . we show the spatial distribution of galaxies in our most complete field with @xmath101 in figure [ cone ] . we have projected through the short axis , corresponding to declination , and plot the comoving positions of the galaxies along and transverse to the line of sight . different symbols show emission - line and absorption - line galaxies , classified by their spectral type as discussed in section 4.4 . large - scale clustering can be seen , with coherent structures such as walls and filaments of size @xmath155 running across our sample . there are several prominent voids which contain very few galaxies , and several overdense regions of strong clustering . the visual impression is consistent with cosmologies @xcite . an analysis of galaxy groups and clusters in the early deep2 data will be presented by @xcite . in this paper we focus on measuring the strength of clustering in the galaxy population using the two - point correlation function . first we measure the clustering for the full sample shown in figure [ cone ] . given the large depth of the sample in redshift , we then address whether it is meaningful to find a single measure of the clustering over such an extended redshift range , as there may be significant evolution in the clustering strength within our survey volume . to investigate evolution within the sample , we would like to measure the clustering in limited redshift ranges within the survey ; given the current sample size , we divide the data into only two redshift subsamples , studying the front half and back half of the survey separately . finally , we split the full sample by predicted restframe @xmath156 color , observed @xmath157 color , spectral type , and absolute @xmath158 luminosity , to study galaxy clustering as a function of these properties at @xmath159 . the survey is far from complete , and with the data presented here we do not attempt to subdivide the sample further . in future papers we will be able to investigate the clustering properties of galaxies in more detail . the left side of figure [ xisp42all ] shows as measured for all galaxies in the most complete field of our survey in the redshift range @xmath0 . all contour plots presented here have been produced from measurements of in linear bins of @xmath160 , smoothed with a @xmath161 boxcar . we apply this smoothing only for the figures ; we do not smooth the data before performing any calculations . on scales @xmath162 2 , the signature of small `` fingers of god '' can be seen as a slight elongation of the contours in the @xmath127 direction . specifically , the contours of @xmath163 and @xmath164 ( in bold ) intersect the @xmath127-axis at @xmath165 and 5 , while intersecting the @xmath126-axis at @xmath166 and 4 , respectively . we leave a detailed investigation of redshift - space distortions to a subsequent paper @xcite . in order to recover the real - space correlation function , , we compute the projected function by calculating in log - separation bins in @xmath126 and then summing over the @xmath127 direction . the result is shown in the top of figure [ wprp42 ] . errors are calculated from the variance of measured across the six gif mock catalogs , after application of this field s current window function . here deviates slightly from a perfect power - law , showing a small excess on scales @xmath167 . however , the deviations are within the 1-@xmath168 errors , and as there exists significant covariance between the plotted points , there is no reason to elaborate the fit . from we can compute @xmath169 and @xmath141 of if we assume that is a power - law , using equation [ powerlawwprp ] . fitting on scales @xmath170 , we find @xmath171 and @xmath172 . this fit is shown in figure [ wprp42 ] as the dotted line and is listed in table [ r0table ] . the errors on and @xmath141 are taken from the percentage variance of the measured and @xmath141 amongst the mock catalogs , scaled to our observed values . note that we do not use the errors on as a function of scale shown in figure [ powerlawwprp ] , which have significant covariance , to estimate the errors on and @xmath141 . the errors quoted on and @xmath141 are dominated by the cosmic variance present in the current data sample . to test that our mock catalogs are independent enough to fully estimate the effects of cosmic variance , we also measure the variance in new mock catalogs , which are just being developed @xcite . these new mocks are made from a simulation with a box size of 300 with finely - spaced redshift outputs . we find that the error on in the new mocks is 12% , less than the 16% we find in the gif mock catalogs used here . we therefore believe that the errors quoted here fully reflect the effects of cosmic variance . preliminary measurements in two of our other fields are consistent with the values found here , within the 1-@xmath168 errors ; in one field , with 1372 galaxies , we measure @xmath173 and @xmath174 , while a separate field , with 639 galaxies , yields @xmath175 and @xmath176 . we have already described above how we use mock catalogs to estimate the bias resulting from our slitmask target algorithm , which precludes targeting close pairs in one direction on the sky . another method for quantifying this effect is to calculate an upper - limit on the clustering using a nearest - neighbor redshift correction , where each galaxy that was not selected to be observed on a slitmask is given the redshift of the nearest galaxy on the plane of the sky with a measured redshift . this correction will significantly overestimate the correlations on small scales , since it assumes that members of all close pairs on the sky are at the same redshift , but it should provide a strong upper limit on the correlation length . using this correction we find an upper limit on of @xmath177 and on @xmath141 of @xmath178 . in the above analysis , we measured the correlation properties of the full sample over the redshift range @xmath0 . this is a wide range over which to measure a single clustering strength , given both possible evolutionary effects in the clustering of galaxies and the changing selection function of our survey , as the luminosity distance and the rest - frame bandpass of our selection criteria change with redshift . in addition to possibly ` washing out ' evolutionary effects within our survey in measuring a single clustering strength over this redshift range , the changing selection function makes it difficult to interpret these results . in this section we attempt to quantify the redshift dependence in our clustering measurements . we begin by estimating the effective redshift of the correlation function we have measured . the calculations of presented in the previous section used the @xmath131 weighting scheme , which attempts to counteract the selection function of the survey , @xmath107 , and give equal weight to volumes at all redshifts without adding noise . to calculate the ` effective ' redshift @xmath179 of the pair counts used to calculate , we compute the mean @xmath131-weighted redshift by summing over galaxy pairs : @xmath180 where @xmath181 runs over all galaxies and @xmath182 over all galaxies within a range of separations from @xmath181 , with @xmath183 , where @xmath135 is the redshift - space separation between the pair of galaxies . the weight @xmath184 is given by equation [ j3eqn ] , and as both galaxies in the pair are at essentially the same redshift ( to within @xmath185 or better ) we use @xmath186 instead of @xmath187 note that this effective redshift will depend on the range of separations considered . for @xmath188 and @xmath189 , @xmath190 , while for @xmath191 and @xmath192 , @xmath193 for the full sample . for this reason , assuming only one effective redshift for the correlation function of a deep galaxy sample covering a large redshift range can not accurately reflect the true redshift dependence . we do however estimate an approximate averaged value for @xmath179 for the galaxy sample presented here by summing over all pairs of galaxies with @xmath126 or @xmath127 @xmath19420 ( @xmath195 and @xmath196 ) . this yields @xmath197 for this data sample , though we caution that it is not immediately clear how meaningful this number is , given the wide redshift range of our data . all values of @xmath179 quoted in table 1 are for @xmath198 . if is calculated without @xmath131 weighting , the raw pair counts in the survey are dominated by volumes with the highest number density in our sample , namely @xmath144 . without @xmath131 weighting we find @xmath199 and @xmath200 . the effective redshift of this result is found using equation [ zeff ] , setting @xmath201 for all galaxies , yielding @xmath202 . the differences between the values of and @xmath141 derived with and without @xmath131 weighting could be the result of evolution within the survey sample , cosmic variance , and/or redshift - dependent effects of our survey selection . it is important to stress that our use of the traditional @xmath131 weighting scheme for minimum variance estimates of leads to an effective redshift of the pair counts which is a function of the pair separation ; the correlations of close pairs have a considerably lower effective redshift than pairs with large separation . this complicates the interpretation of single values of and @xmath141 quoted for the entire survey . in local studies of galaxy correlations , one assumes that evolutionary corrections within the volume studied are insignificant and that the best correlation estimate will be achieved with equal weighting of each volume element , provided shot noise does not dominate . @xmath131 weighting is intended to provide equal weight per unit volume to the degree permitted by the radial gradient in source density , but it complicates interpretation of results within a volume for which evolutionary effects are expected . it is far better to subdivide the sample volume between high and low redshift and separately apply @xmath131 weighting within the subvolumes . to this end , we divide our sample near its median redshift , creating subsamples containing roughly equal numbers of galaxies with @xmath203 and @xmath204 . the effective redshift for the lower-@xmath205 sample ( averaged over all separations ) is @xmath4 , while for the higher-@xmath205 sample it is @xmath7 . the selection function for the redshift subsamples is identical to that shown in figure [ zhist ] , cut at @xmath206 . the right side of figure [ xisp42all ] shows the measured for both subsamples . at lower redshifts the data exhibit a larger clustering scale length , as might be expected from gravitational growth of structure . the lower-@xmath205 sample also displays more prominent effects from `` fingers of god '' . the bottom of figure [ wprp42 ] shows the resulting and power - law fits for each redshift range . note that we fit a power - law on scales @xmath207 for the higher-@xmath205 sample but fit on scales @xmath208 for the lower-@xmath205 sample , as decreases significantly on larger scales . we have tested for systematic effects which could lead to such a decrease and have not found any . with more data we will be able to see if this dip persists . the lower-@xmath205 sample exhibits a larger scale length than the higher-@xmath205 sample , though the difference is well within the 1-@xmath168 uncertainties due to cosmic variance ( see table [ r0table ] for details ) . for each subsample we estimate the errors using the variance among the mock catalogs over the same redshift range used for the data . a positive luminosity - dependence in the galaxy clustering would lead to an increase in measured at larger redshifts , where the effective luminosity is greater . however , evolutionary effects could offset this effect , if grows with time . we find no significant difference in our measured value of for the lower - redshift sample . locally , significant luminosity - dependence has been seen in the clustering of data in the 2dfgrs @xcite and sdss @xcite and , if present in the galaxy population at @xmath10 , could complicate measurements of the evolution of clustering within our survey volume , given the higher median luminosity of galaxies in our sample at larger redshifts . we investigate the luminosity - dependence of clustering in our sample in section 4.5 and discuss possible evolutionary effects in section 5.2 . we now measure the dependence of clustering on specific galaxy properties . we begin by creating red and blue subsamples based on either restframe @xmath156 color or observed @xmath157 color , which is a direct observable and does not depend on modelling k - corrections . k - corrections were calculated using a subset of @xcite galaxy spectra convolved with the @xmath209 , @xmath15 and @xmath69 cfh12k filters used in the deep2 survey . these are used to create a table containing the restframe colors and k - corrections as a function of @xmath205 and @xmath157 color . k - corrections are then obtained for each galaxy using a parabolic interpolation ( for more details see willmer et al . the median k - corrections for the sample used here are @xmath210 for @xmath211 and @xmath212 for @xmath213 , where we apply corrections to our observed @xmath59band magnitudes , which are the deepest and most robustly calibrated @xcite . after applying these k - corrections , we estimate the galaxy restframe @xmath156 colors and divide the sample near the median color into red , @xmath214 , and blue , @xmath215 , subsets . we further restrict the subsamples to the redshift range @xmath216 . we fit for the selection function , @xmath107 , for each subsample separately , again using data from all three observed fields which match these color selection criteria , and find that the resulting selection functions for the red and blue subsamples are similar to each other and similar to that shown in figure [ zhist ] . for these and all subsamples we use @xmath131 weighting in measuring and estimate errors using mock catalogs with half the original sample size . in this way we attempt to replicate the error due to cosmic variance , which depends on the survey volume , and sample size . we found that the error did not increase signficantly when using half the galaxies , which indicates that cosmic variance is the dominant source of error . we find that the red galaxies have a larger correlation length and stronger `` fingers of god '' . this trend is not entirely unexpected , as previous data at @xmath217 have shown similar effects @xcite , though the volume which we sample is much larger and therefore less affected by cosmic variance . the top of figure [ wprpbr ] shows for each sample ; fits to are given in table [ r0table ] . while restframe colors are more physically meaningful than observed colors , they are somewhat uncertain as k - corrections can become large at our highest redshifts . we therefore divide our full data sample into red and blue subsets based on observed @xmath157 color . there is a clear bimodality in the distribution of @xmath157 colors of deep2 targets , leading to a natural separation at @xmath218 . however , as there are not enough galaxies with @xmath219 to provide a robust result , we instead divide the full dataset at @xmath220 , which creates subsamples with @xmath37 times as many blue galaxies as red . we again construct redshift selection functions , @xmath107 , for each sample and measure in the redshift range @xmath216 . again , the redder galaxies show a larger correlation length and a steeper slope , though the differences are not as pronounced as in the restframe color - selected samples ; see table [ r0table ] . we next investigate the dependence of clustering on spectral type . @xcite have performed a principal component analysis ( pca ) of each galaxy spectrum in the deep2 survey . they distinguish emission - line from absorption - line galaxies using the parameter @xmath221 , the distribution function of which displays a bimodality , suggesting a natural split in the sample . we use the same division employed by @xcite , who define late - type , emission - line galaxies as having @xmath222 and early - type , absorption - line galaxies with @xmath223 . our absorption - line subset includes @xmath224 galaxies in the redshift range @xmath216 , while the emission - line sample has @xmath37 times as many galaxies . in figure [ cone ] different symbols show the galaxy population divided by spectral type . the early - type subset can be seen to reside in the more strongly clustered regions of the galaxy distribution . the middle of figure [ wprpbr ] shows measured for our spectral - type subsamples , and best - fit values of and @xmath141 are listed in table [ r0table ] . absorption - line galaxies have a larger clustering scale length and an increased pairwise velocity dispersion . since @xmath221 correlates well with color , this result is not unexpected ; the bulk of the early - type galaxies have red colors , though there is a long tail which extends to the median color of the late - type galaxies . thus the subsamples based on spectral type are not identical to those based on color . the spectral - type is intimately related to the amount of current star formation in a galaxy , so that we may conclude that actively star - forming galaxies at @xmath10 are significantly less clustered than galaxies that are passively evolving . interestingly , the emission - line galaxies show a steeper slope in the correlation function than the absorption - line galaxies , which is not seen at @xmath225 @xcite . this will be important to investigate further as the survey collects more data . we also split the full sample by absolute @xmath158 magnitude , after applying k - corrections , to investigate the dependence of galaxy clustering on luminosity . we divide our dataset near the median absolute magnitude , at @xmath2265 log @xmath42 . figure [ phiz_lumin ] shows the selection function for each subsample . unlike the previous subsets , here the selection functions are significantly different for each set of galaxies ; @xmath107 for the brighter objects is relatively flat , while @xmath107 for fainter galaxies falls steeply with @xmath205 . the bottom of figure [ wprpbr ] shows for each subsample . we fit as a power - law on scales @xmath227 and find that the more luminous galaxies have a larger correlation length . on larger scales both samples show a decline in , but the brighter sample shows a steeper decline ; fits are listed in table [ r0table ] . in this early paper , using a sample of roughly 7% the size we expect to have in the completed survey , we have restricted ourselves to considering only two subsamples at a time . as a result , in our luminosity subsamples we are mixing populations of red , absorption - line galaxies , which have very different mass - to - light ratios as well as quite different selection functions , with the star - forming galaxies that dominate the population at higher redshifts . however , the two luminosity subsamples in our current analysis contain comparable ratios of emission - line to absorption - line galaxies , with @xmath228% of the galaxies in the each sample having late - type spectra . in future papers we will be able to investigate the luminosity - dependence of clustering in the star - forming and absorption - line populations separately . having measured the clustering strength using the real - space two - point correlation function , , for each of the samples described above , we are now in a position to measure the galaxy bias , both for the sample as a whole at @xmath10 and for subsamples defined by galaxy properties . we first calculate the absolute bias for galaxies in our survey and then determine the relative bias between various subsamples . using these results , we can constrain models of galaxy evolution and compare our results to other studies at higher and lower redshifts . to measure the galaxy bias in our sample , we use the parameter , defined as the standard deviation of galaxy count fluctuations in a sphere of radius 8 . we prefer this quantity as a measure of the clustering amplitude over using the scale - length of clustering , , alone , which has significant covariance with @xmath141 . we can calculate from a power - law fit to using the formula , @xmath229 where @xmath230 @xcite . note that here we are not using the linear - theory @xmath231 that is usually quoted . instead , we are evaluating on the scale of 8 in the non - linear regime , leading to the notation . we then define the effective galaxy bias as @xmath232 where is for the galaxies and @xmath233 is for the dark matter . our measurements of for all data samples considered are listed in table [ r0table ] . errors are derived from the standard deviation of as measured across the mock catalogs . the evolution of the dark matter clustering can be predicted readily using either n - body simulations or analytic theory . here we compute @xmath234 from the dark matter simulations of @xcite at the effective redshifts of both our lower-@xmath205 and higher-@xmath205 subsamples . we use two simulations where the linear @xmath235 at @xmath55 , defined by integrating over the linear power spectrum , is equal to 1.0 and 0.8 . this is the @xmath231 which is usually quoted in linear theory . for convenience , we define the parameter @xmath236 . in both of these simulations , we fit @xmath237 as a power - law on scales @xmath238 , and from this measure @xmath233 using equation [ sig8eqn ] above . for the simulation with @xmath239 , we measure @xmath240 at @xmath241 and @xmath242 at @xmath243 , while for the simulation with @xmath244 , we measure @xmath245 at @xmath241 and @xmath246 at @xmath243 . our results imply that for @xmath239 in a cosmology the effective bias of galaxies in our sample is @xmath11 , such that the galaxies trace the mass . this would suggest that there was little or no evolution in the galaxy biasing function from @xmath38 to 0 and could also imply an early epoch of galaxy formation for these galaxies , such that by @xmath10 they have become relatively unbiased . however , if @xmath244 then the effective galaxy bias in our sample is @xmath13 , which is more consistent with predictions from semi - analytic models @xcite . generally , we find the net bias of galaxies in our sample to be @xmath247 . the galaxy bias can be a strong function of sample selection . one explanation for the somewhat low clustering amplitude we find may be the color selection of the survey . our flux - limited sample in the @xmath15-band translates to bands centered at @xmath93 = 3600 and 3100 at redshifts @xmath2480.8 and 1.1 , respectively . the flux of a galaxy at these ultraviolet wavelengths is dominated by young stars , and therefore our sample could undercount galaxies which have had no recent star formation , while preferentially selecting galaxies with recent star formation . the deep2 sample selection may be similar to iras - selected low-@xmath205 galaxy samples in that red , old stellar populations are under - represented ( however , our uv - bright sample is probably less dusty than the iras galaxies ) . iras - selected samples are known to have a diminished correlation amplitude and undercount dense regions in cluster cores ( e.g. moore et al . we are accumulating @xmath51-band imaging within the deep2 fields , which we can use to study the covariance of @xmath51-selected samples with our @xmath15-selected sample in order to gain a better understanding of the behavior of @xmath249 at @xmath250 . @xcite find that their @xmath51-selected sample at @xmath251 generally shows stronger clustering than optically - selected samples at the same redshifts , and we expect the same will hold true for our sample . the deep2 survey volume is sufficiently extended in the redshift direction that we expect to discern evolutionary effects from within our sample . for example , the look - back time to @xmath252 is 6.9 gyrs in a cosmology ( for @xmath253 ) , but at @xmath254 the look - back time grows to 8.4 gyrs . as discussed in section 4.2 , measuring the clustering strength for the full sample from @xmath0 is not entirely meaningful , as there may be significant evolutionary effects within the sample , and the results are difficult to interpret given the dependence of the effective redshift on scale . we therefore divide the sample into two redshift ranges and measure the clustering in the foreground and background of our survey . however , with the data available to date , our results must be considered initial ; we hope to report on a sample @xmath255 times larger in the next few years . note also that as our sample size increases and we are better able to divide our sample into narrower redshift ranges , the dependence of @xmath179 on scale will become much less important . the decreased correlations observed in the higher - redshift subset within the deep2 sample might be considered to be the effect of an inherent diminished clustering amplitude for galaxies in the more distant half of the survey . indeed , the mass correlations are expected to be weaker at earlier times , but we expect galaxy biasing to be stronger , so that the galaxy clustering may not increase with time as the dark matter distribution does . as a complication , at higher redshifts we are sampling intrinsically brighter galaxies due to the flux limit of the survey , and there is a significant dependence of clustering strength on luminosity in our data . our lower - redshift sample has an effective luminosity of @xmath2565 log @xmath42 , while for the the higher - redshift galaxies the effective luminosity is @xmath2575 log @xmath42 . as discussed in section 4.2 , this luminosity difference would lead to an increase in measured for the higher - redshift sample , if there was no intrinsic evolution in the galaxy clustering . additionally , at higher redshifts our @xmath15-band selection corresponds to even shorter restframe wavelengths , yielding a sample more strongly biased towards star - forming galaxies . galaxies which form at high - redshift are expected to be highly - biased tracers of the underlying dark matter density field @xcite ; this bias is expected to then decrease with time @xcite . if galaxies are born as rare peaks of bias @xmath258 in a gaussian noise field with a preserved number density , their bias will decline with epoch according to @xmath259 , where @xmath260 is the linear growth of density fluctuations in the interval since the birth of the objects . this equation shows that if galaxies are highly biased tracers when born , they should become less biased as the universe continues to expand and further structure forms . this has been the usual explanation for the surprisingly large clustering amplitude reported for lyman - break galaxies at @xmath16 . they have a clustering scale length comparable to optically - selected galaxies in the local universe , but the dark matter should be much less clustered at that epoch , implying a bias of @xmath261 for a cosmology @xcite . the 2dfgrs team has shown that the bias in their @xmath262-selected sample is consistent with @xmath263 @xcite . given these observations of @xmath264 at @xmath16 and @xmath265 at @xmath225 , one might expect an intermediate value of @xmath23 at @xmath266 , assuming that all of these surveys trace similar galaxy populations . however , different selection criteria may be necessary to trace the same galaxy population over various redshifts . our subsample of star - forming , emission - line galaxies has similar selection criteria as recent studies of galaxies at @xmath16 . the lyman - break population has been selected to have strong uv luminosity and therefore high star - formation rates . the spectroscopic limit of the lyman - break sample is @xmath267 , which is roughly equivalent to @xmath268 at @xmath217 , while the deep2 survey limit is @xmath269 , so that roughly similar uv luminosities are being probed by these studies . with a sample of @xmath270 lyman - break galaxies at @xmath16 @xcite measure a correlation length of @xmath271 with a slope of @xmath272 . at @xmath197 we measure a somewhat lower correlation length of @xmath273 and a slightly steeper slope of @xmath172 , implying that star - forming galaxies at @xmath10 are not as strongly biased at those at higher redshifts . the slope of the correlation function is expected to increase with time , as seen here , as the underlying dark matter continues to cluster , resulting in more of the mass being concentrated on smaller scales . in constraining galaxy evolution models , however , it is important to note that while these are measures of similar , star - forming populations of galaxies at @xmath16 and @xmath10 , the lyman - break galaxies are not progenitors of the star - forming galaxies at @xmath10 . using the linear approximations of @xcite one would expect the lyman - break galaxies to have a correlation length of @xmath274 at @xmath10 @xcite , so that the objects carrying the bulk of the star formation at @xmath10 and @xmath16 are not the same . our population of red , absorption - line galaxies have a correlation length of @xmath275 , similar to that expected for the descendants of the lyman - break population at @xmath10 . using recent studies from both 2df and sdss , we can also compare our results to @xmath225 surveys . the two - point correlation function is relatively well - fit by a power - law in all three of these surveys on scales @xmath276 . sdss find a correlation length of @xmath277 in their @xmath278-selected sample @xcite , while 2df find @xmath279 in their @xmath262-selected survey . these values are significantly larger than our measured at @xmath10 , in our @xmath15-selected survey . the slope of the two - point correlation function may be marginally steeper at low redshifts , with 2df finding a value of @xmath280 and sdss fitting for @xmath281 , compared with our values of @xmath282 at @xmath4 and @xmath7 . while the highly - biased , star - forming galaxies seen at @xmath16 appear to have formed in the most massive dark matter halos present at that epoch @xcite and evolved into the red , clustered population seen at @xmath10 , the star - forming galaxies seen at @xmath10 are not likely to be significantly more clustered in the present universe . these galaxies are not highly - biased , and as their clustering properties do not imply that they reside in proto - cluster cores , they can not become cluster members at @xmath55 in significant numbers . having measured the absolute galaxy bias in our sample as a whole , which is largely determined by the details of our sample selection , we now turn to relative trends seen within our data , which should be more universal . using the various subsamples of our data defined above we can quantify the dependence of galaxy bias on color , type , and luminosity , and we compare our findings with other results at @xmath45 . we define the relative bias between two samples as the ratio of their : @xmath283 as the subsamples are taken from the same volume and have similar selection functions , there is negligable cosmic variance in the ratio of the clustering strengths , and therefore the error in the relative bias is lower than the error on the values of individually . to estimate the error on the relative bias , we use the variance among the mock catalogs ( neglecting cosmic variance ) which leads to a 4% error , and include an additional error of 6% due to uncertainties in the selection function , added in quadrature . we find in the restframe @xmath156 red and blue subsamples that @xmath284 . this value is quite similar to the relative biases seen in local @xmath55 samples . in the ssrs2 data @xcite find red galaxies with @xmath285 have a relative bias of @xmath27 compared to blue galaxies , while @xcite report that in the sdss early data release red galaxies ( based on a split at @xmath286 ) have a relative bias of @xmath287 compared to blue galaxies . we find a a similar value of the relative bias at @xmath217 in our red and blue subsamples , implying that a color - density relation is in place at these higher redshifts . the observed - frame @xmath157 subsamples have a relative bias of @xmath288 . this value is slightly lower than that of the restframe color - selected subsamples , as expected since the observed @xmath157 color of galaxies has a strong redshift - dependence over the redshift range we cover , @xmath216 , and is therefore less effective at distinguishing intrinsically different samples . using the pca spectral analysis we find that the absorption - line sample has a clustering length , , @xmath31 times larger than the emission - line sample , with a relative bias of @xmath289 . @xcite find using 2dfgrs data that locally , absorption - line galaxies have a relative bias about twice that of emission - line galaxies on scales of @xmath290 , but that the relative bias decreases to unity on scales @xmath15110 . the relative bias integrated over scales up to 8 is @xmath291 at @xmath225 , similar to our result at @xmath217 . our current data sample is not sufficiently large to robustly measure the scale - dependence of the galaxy bias , though this should readily be measurable from the final dataset . @xcite find in their survey ( with @xmath292 ) that galaxies with absorption - line spectra show much stronger clustering at small separations , though their absorption - line sample size is small , with 121 galaxies . @xcite also report that in the redshift interval @xmath293 , galaxies with red colors have a correlation length 2.7 times greater than bluer galaxies with strong [ oii ] emission . recently , there have been several studies which have found very large clustering strengths for extremely - red objects ( eros , @xmath294 ) at @xmath217 . using the angular correlation function , @xcite find a correlation length of @xmath295 for eros at @xmath296 , while @xcite find that the correlation length is @xmath297 . these samples are of rare objects which have extreme colors and are quite luminous ; @xcite estimate that their sample is @xmath298 magnitudes brighter than @xmath299 . we find a correlation length of @xmath300 for our absorption - line sample , which has an effective magnitude of @xmath301 log@xmath42 . given the relatively large clustering strength of the absorption - line galaxies in our sample and the luminosity difference between our sample and the ero studies , it is possible that in our absorption - line sample we are seeing a somewhat less extreme population which is related to the eros seen at @xmath217 . we find that the relative bias between luminosity subsamples is @xmath302 . these datasets have significantly different selection functions , unlike the previous samples , and the error on the relative bias due to differences in cosmic variance between the two samples results in an additional 8% error , added in quadrature . this is calculated using numerical experiments utilizing the cosmic variance in redshift bins calculated in @xcite . the bright sample has a median absolute magnitude of @xmath3035 log @xmath42 , while the faint sample has a median @xmath3045 log @xmath42 . as noted in section 4.5 , our luminosity subsamples include both star - forming galaxies as well as older , absorption - line galaxies and cover a wide range in redshift ( @xmath216 ) , possibly complicating interpretation of these results . however , both samples have the same ratio of early - type to late - type spectra . the deep2 galaxy redshift survey is designed to study the evolution of the universe from the epoch @xmath305 to the present by compiling an unprecedented dataset with the deimos spectrograph . with the final sample we hope to achieve a statistical precision of large - scale structure studies at @xmath217 that is comparable to previous generations of local surveys such as the las campanas redshift survey ( lcrs , shectman et al . as we complete the survey , our team will explore the evolution of the properties of galaxies as well as the evolution of their clustering statistics . the correlation analysis reported here is far more robust than earlier studies at @xmath217 because of our greatly increased sample size and survey volume . we find values of the clustering scale - length , @xmath5 at @xmath4 and @xmath8 at @xmath7 , which are within the wide range of clustering amplitudes reported earlier @xcite . this implies a value of the galaxy bias for our sample , @xmath11 if @xmath306 today or @xmath13 if @xmath14 today , which is lower than what is predicted by semi - analytic simulations of @xmath10 . our errors are estimated using mock catalogs and are dominated by sample variance , given the current volume of our dataset . we find no evidence for significant evolution of within our sample , though intrinsic evolutionary effects could be masked by luminosity differences in our redshift subsamples . we see a significantly - increased correlation strength for subsets of galaxies with red colors , early - type spectra , and higher luminosity relative to the overall population , similar to the behavior observed in low - redshift catalogs . galaxies with little on - going star formation cluster much more strongly than actively star - forming galaxies in our sample . these clustering results as a function of color , spectral type and luminosity are consistent with the trends seen in the semi - analytic simulations of @xcite at @xmath38 , and indicate that galaxy clustering properties as a function of color , type , and luminosity at @xmath217 are generally not very different from what is seen at @xmath55 . the overall amplitude of the galaxy clustering observed within the deep2 survey implies that this is not a strongly biased sample of galaxies . for @xmath239 ( defined as the linear @xmath307 at @xmath55 ) , the galaxy bias is @xmath11 , while for @xmath244 , the bias of the deep2 galaxies is @xmath13 . this low bias may result from the @xmath15-band selection of the survey , which roughly corresponds to a restframe @xmath308-band selected sample ; the more clustered , old galaxies with red stellar populations are likely to be under - represented as our sample preferentially contains galaxies with recent star - formation activity . however , the same selection bias applies to lyman - break galaxies studied at @xmath16 , which are seen to be significantly more biased than our sample at @xmath10 . we are undertaking studies with @xmath51-band data in our fields , which should lead to clarification of these questions . more precise determinations of the evolution of clustering within our survey and the luminosity - dependence of the galaxy bias at @xmath10 awaits enlarged data samples , on which we will report in due course . we would like to thank the anonymous referee for useful comments . this project was supported in part by the nsf grants ast00 - 71048 , ast00 - 71198 and kdi-9872979 . the deimos spectrograph was funded by a grant from cara ( keck observatory ) , an nsf facilities and infrastructure grant ( ast92 - 2540 ) , the center for particle astrophysics and by gifts from sun microsystems and the quantum corporation . the deep2 redshift survey has been made possible through the dedicated efforts of the deimos staff at uc santa cruz who built the instrument and the keck observatory staff who have supported it on the telescope . full sample & 2219 & @xmath309 & 0.99 & @xmath310 & @xmath311 & @xmath312 & @xmath313 + lower @xmath205 sample & 1087 & @xmath314 & 0.82 & @xmath315 & @xmath316 & @xmath317 & @xmath318 + higher @xmath205 sample & 1132 & @xmath319 & 1.14 & @xmath320 & @xmath316 & @xmath312 & @xmath321 + @xmath322 0.7 & 855 & @xmath323 & 0.96 & @xmath324 & @xmath325 & @xmath326 & @xmath327 + @xmath328 0.7 & 964 & @xmath323 & 0.93 & @xmath329 & @xmath330 & @xmath326 & @xmath331 + @xmath332 0.9 & 442 & @xmath323 & 0.90 & @xmath333 & @xmath311 & @xmath334 & @xmath335 + @xmath336 0.9 & 1561 & @xmath323 & 0.95 & @xmath337 & @xmath338 & @xmath334 & @xmath331 + absorption - line & 395 & @xmath323 & 0.86 & @xmath339 & @xmath340 & @xmath334 & @xmath341 + emission - line & 1605 & @xmath323 & 0.97 & @xmath342 & @xmath311 & @xmath334 & @xmath343 + @xmath344 & 899 & @xmath323 & 0.99 & @xmath345 & @xmath346 & @xmath347 & @xmath348 + @xmath349 & 1088 & @xmath323 & 0.89 & @xmath350 & @xmath351 & @xmath352 & @xmath353 +
we measure the two - point correlation function in a sample of 2219 galaxies between @xmath0 to a magnitude limit of @xmath1 from the first season of the deep2 galaxy redshift survey . from we recover the real - space correlation function , , which we find can be approximated within the errors by a power - law , @xmath2 , on scales @xmath3 . in a sample with an effective redshift of @xmath4 , for a cosmology we find @xmath5 ( comoving ) and @xmath6 , while in a higher - redshift sample with @xmath7 we find @xmath8 and @xmath9 . these errors are estimated from mock galaxy catalogs and are dominated by the cosmic variance present in the current data sample . we find that red , absorption - dominated , passively - evolving galaxies have a larger clustering scale length , , than blue , emission - line , actively star - forming galaxies . intrinsically brighter galaxies also cluster more strongly than fainter galaxies at @xmath10 . our results imply that the deep2 galaxies have an effective bias @xmath11 if @xmath12 today or @xmath13 if @xmath14 today . this bias is lower than what is predicted by semi - analytic simulations at @xmath10 , which may be the result of our @xmath15-band target selection . we discuss possible evolutionary effects within our survey volume , and we compare our results with galaxy clustering studies at other redshifts , noting that our star - forming sample at @xmath10 has very similar selection criteria as the lyman - break galaxies at @xmath16 and that our red , absorption - line sample displays a clustering strength comparable to the expected clustering of the lyman - break galaxy descendants at @xmath10 . our results demonstrate that galaxy clustering properties as a function of color , spectral type and luminosity seen in the local universe were largely in place by @xmath10 .
infrared ( ir ) spectroscopy has been a powerful tool to study the microscopic carrier dynamics and electronic structures in strongly correlated electron materials , such as rare earth ( @xmath1 electron ) , transition metal ( @xmath2 electron ) , and organic ( @xmath3 electron ) compounds @xcite . the ir spectroscopy technique has been also performed under high pressure using a diamond anvil cell ( dac ) [ 2 - 13 ] since the strongly correlated materials show many interesting physical properties under high pressure . in a dac , a pair of diamond anvils and a thin metal gasket are used to seal a sample and a pressure transmitting medium @xcite . a typical diameter of the diamond surface is 0.8 mm to reach a pressure of 10 gpa , and 0.6 mm to reach 20 gpa . therefore the sample in this experiment should have dimensions of the order of 100 @xmath4 m . to perform an infrared ( ir ) reflectance study on such a small sample under the restricted sample space in a dac , synchrotron radiation ( sr ) has been used as a bright source of both far and mid - infrared . in fact , high pressure ir spectroscopy with dac is currently one of the major applications of ir - sr [ 4 - 7,9 - 13 ] . with a dac , the reflectance is measured between the sample / diamond interface , in contrast to the usual case of sample / vacuum or sample / air interface . the normal - incidence reflectance of a sample relative to a transparent medium of ( real ) refractive index @xmath5 is given by the fresnel s formula as @xcite : @xmath6 here , @xmath7 is the complex refractive index of the sample , and @xmath5=2.4 for diamond and 1.0 for vacuum . hereafter , we denote @xmath8 at sample / diamond interface as @xmath0 , and that at sample / vacuum interface as @xmath9 . from eq . ( 1 ) , it is easily seen that @xmath0 of a sample measured in dac may be substantially different from @xmath9 . the purpose of this study is to consider the kramers - kronig ( kk ) analysis of @xmath0 data measured in dac . kk analysis has been widely used to derive optical constants such as the refractive index , dielectric function and optical conductivity from a measured @xmath9 spectrum @xcite . however , due to the difference between @xmath9 and @xmath0 discussed above , the usual kk analysis method can not be straightforwardly applied to @xmath0 @xcite . to derive optical constants from @xmath0 , therefore , previous high pressure ir studies used either a drude - lorentz spectral fitting @xcite or a modified kk transform @xcite . in this work , we propose a different method , which relies on the usual kk transform with an appropriate cutoff to the @xmath0 , as an alternative approach to obtain the infrared @xmath10 from @xmath0 . the validity of the proposed method is demonstrated with actually measured reflectance data of prru@xmath11p@xmath12 . the complex reflectivity of the electric field , @xmath13 , is expressed as @xcite @xmath14 here @xmath15 is the square root of the reflectance @xmath8 , which is actually measured in experiments . then the real and imaginary parts of @xmath16 can be expressed in terms of @xmath17 and @xmath18 as @xmath19 and @xmath20 respectively . therefore , if @xmath21 can be derived from measured @xmath15 with kk analysis even for the sample / diamond reflection case , @xmath22 and @xmath23 can also be derived simply by setting @xmath5=2.4 in eqs . ( 2 ) and ( 3 ) . then , the imaginary part of the complex dielectric function is given as @xmath24 , and the optical conductivity is given as @xmath25 @xcite . in performing kk analysis on reflectance data , usually the logarithm of @xmath13 , namely @xmath26 is regarded as a complex response function . in the case of sample / vacuum reflection , the kk relation between @xmath27 and @xmath21 is expressed as @xcite @xmath28 here , @xmath29 denotes the principal value . in deriving this relation , it is required that @xmath30 has no poles in the upper complex @xmath31 plane when @xmath32 is finite . this is correct since @xmath33 only when @xmath34 in the case of @xmath5=1 . however , when @xmath35 as in the case of sample / diamond interface , @xmath36 may be satisfied at some point on the upper imaginary axis @xcite . this point is denoted as @xmath37 , where @xmath38 is a real , positive and finite number . when @xmath36 , @xmath39 from eq . ( 2 ) and @xmath40 therefore has a pole at @xmath37 . accordingly , the kk relation in this case must be modified to @xcite @xmath41.\ ] ] namely , the presence of a medium with @xmath35 brings an extra phase shift , indicated by the square bracket in eq . ( 7 ) , into the kk relation . note that the extra phase shift is a decreasing function of @xmath42 , and that the original kk relation of eq . ( 6 ) is recovered when @xmath43 @xcite . detailed theoretical considerations on the extra phase shift in various situations have been reported @xcite . in the case of actual experimental studies , however , the precise value of @xmath38 may not be known . accordingly , the value of @xmath38 has been estimated from experimental @xmath0 data by use of a combination of dl fitting and the modified kk transform @xcite . in this method , one uses eq . ( 7 ) with @xmath44 and looks for a value of @xmath38 that well reproduces the @xmath10 given by a dl fitting of @xmath0 . note that , on the other hand , if the frequency range of interest is lower than the value of @xmath38 , effects of the extra phase may be only minor , and the usual kk transform of eq . ( 6 ) , combined with the use of @xmath5=2.4 in eqs . ( 3 ) and ( 4 ) , might give sufficiently accurate values of optical constants . we will examine the validity of such a procedure in the next section . here , we use @xmath9 spectra actually measured on prru@xmath11p@xmath12 @xcite . this compound is well known for showing a metal - to - insulator transition at about 60 k @xcite , and a clear energy gap in @xmath10 was observed in our previous work @xcite . here we use @xmath9 at 60 k ( metal ) and 9 k ( insulator ) measured over a wide photon energy range of 0.008 - 30 ev @xcite , which are shown by the blue curve in figs . 1(a ) and 2(a ) . is the reflectance spectrum of prru@xmath11p@xmath12 measured at 60 k in vacuum @xcite , and @xmath0 is that expected in a dac calculated from @xmath9 as described in the text . the green , red , and light blue curves are @xmath45 extrapolations with cutoff energies of @xmath46=1.5 , 2 , and 5 ev , respectively . ( b ) the optical conductivity ( @xmath47 ) obtained with kk analysis of @xmath9 is compared with those obtained with kk analysis of @xmath0 with a cutoff at @xmath46=1.5 , 2 , and 5 ev , and @xmath45 extrapolations above them . below 1.5 ev , @xmath10 obtained from @xmath0 with @xmath46=2 ev agrees very well with that obtained from @xmath9 . ] the procedure is the following . * the full @xmath9 spectrum is kk analyzed with eq . ( 6 ) to obtain @xmath22 , @xmath23 and @xmath10 . * the above @xmath22 and @xmath23 are substituted into eq . ( 1 ) with @xmath5=2.4 to derive @xmath0 that is _ expected _ in a dac . * the @xmath0 obtained above is used with the usual kk transform of eq . ( 6 ) and @xmath5=2.4 in eqs . ( 3 ) and ( 4 ) , to obtain @xmath22 , @xmath23 and @xmath10 . before this is done , an appropriate cutoff and extrapolation are made to the @xmath0 , as described in detail below . if the kk analysis on @xmath0 works properly , the resulting @xmath10 from ( iii ) should well agree with that given by @xmath9 and the usual kk analysis . we first examine the 60 k data . the expected @xmath0 at 60 k obtained by ( i ) and ( ii ) is indicated by the black curve in fig . 1(a ) . in carrying out the integration in eq . ( 6 ) , the @xmath9 spectrum were extrapolated below 0.008 ev and above 30 ev with the hagen - rubens and @xmath45 functions , respectively . it is seen in fig . 1(a ) that @xmath0 shows very high values above about 4 ev . this physically unrealistic feature resulted from the unphysical assumption of constant @xmath5=2.4 in the entire spectral range . in reality , of course , the refractive index of diamond can not be constant and real near and above the band gap , where it shows strong light absorption . in addition , when @xmath34 , @xmath48 and @xmath49 , and therefore @xmath50 must hold just like any other material . accordingly , before performing kk transform in ( iii ) , @xmath0 in fig . 1(a ) was cut off at some energy @xmath46 , and then it was extrapolated with @xmath45 function . several different values of @xmath46 were tried , as shown in fig . for each value of @xmath46 , the usual kk transform of eq . ( 6 ) was made to get @xmath21 , which was then used to derive @xmath22 and @xmath23 with @xmath5=2.4 in eqs . ( 3 ) and ( 4 ) , and to finally obtain @xmath10 . figure 1(b ) shows the @xmath10 spectra obtained as described above , with different values of @xmath46 which are also indicated in fig . it is seen that the obtained @xmath10 spectra strongly depend on @xmath46 . with @xmath46=2.0 ev , the resulting @xmath10 below 1.5 ev agrees very well with that derived from the original @xmath9 . ( actually , @xmath46=2.2 ev gives the best agreement , but @xmath46=2.0 ev data is shown instead . this is because the 2.2 ev data almost completely overlaps with that from @xmath9 , making it difficult to distinguish them in the figure . ) the result for the 9 k data , where the sample is an insulator ( semiconductor ) , is also shown in fig . 2 . a good agreement is again observed between the @xmath10 derived from the full @xmath9 and that from @xmath0 with @xmath46=2 ev . the spectral range of their good agreement is below 1.5 ev , which is similar to the case of 60 k data discussed above . note also that both 9 k and 60 k data show good agreement with the common value of @xmath46= 2 ev . these results show that , for any spectral change in @xmath0 ( either temperature- or pressure - induced ) below 1.5 ev , the corresponding @xmath10 can be obtained by the present method . in actual high pressure studies of strongly correlated materials with dac [ 2 - 13 ] , @xmath0 is usually measured below 1 - 2 ev . hence , above the high energy limit of @xmath0 measurement , the @xmath0 expected from @xmath9 can be connected to the measured @xmath0 , with the cutoff and extrapolation discussed above . then the connected @xmath0 may be kk transformed to obtain @xmath10 , as discussed above . an obvious condition required for this method to work properly is that the pressure- and temperature - induced changes of @xmath0 should be limited below certain energy , which is 1.5 ev for prru@xmath11p@xmath12 as seen in figs . 1(b ) and 2(b ) . this condition is actually met in the high pressure data of prru@xmath11p@xmath12 , which has enabled us to derive its @xmath10 under pressure up to 14 gpa using the present method @xcite . is the reflectance spectrum of prru@xmath11p@xmath12 measured at 9 k in vacuum @xcite , and @xmath0 is that expected in a dac calculated from @xmath9 as described in the text . the green , red , and light blue curves are @xmath45 extrapolations with cutoff energies of @xmath46=1.5 , 2 , and 5 ev , respectively . ( b ) the optical conductivity ( @xmath47 ) obtained with kk analysis of @xmath9 is compared with those obtained with kk analysis of @xmath0 with a cutoff at @xmath46=1.5 , 2 , and 5 ev , and @xmath45 extrapolations above them . ] we have also done similar simulations for other compounds , both metals and insulators , using actually measured data , and have obtained similar results . namely , when an appropriate cutoff and extrapolation are applied to @xmath0 , the usual kk transform of eq . ( 6 ) gave @xmath10 spectra which agreed very well with those directly obtained from the wide range @xmath9 . a limitation of the present method is , as already mentioned above , it can give correct @xmath10 only below certain photon energy ( 1.5 ev in the case of prru@xmath11p@xmath12 ) . hence this method is useful when the temperature and pressure dependences of @xmath0 is limited to below certain energy . in addition , to use the present method , it is required that @xmath9 is known over a wide enough photon energy range , since @xmath22 and @xmath23 must be obtained from @xmath9 with the usual kk analysis . while a mathematically rigorous justification of the proposed method is beyond the scope of this work , this method may be very useful as a simple analysis technique of reflectance spectra measured under high pressure with dac . this work has been done as a part of high pressure infrared studies of strongly correlated electron materials using synchrotron radiation at spring-8 , under the approval by jasri ( 2009a0089 through 2011b0089 ) . the @xmath9 data used in figs . 1 and 2 have been already published @xcite , in collaboration with m. matsunami , l. chen , m. takimoto , t. nanba , c. sekine , and i. shirotani .
when the optical reflectance spectrum of a sample under high pressure is studied with a diamond anvil cell , it is measured at a sample / diamond interface . due to the large refractive index of diamond , the resulting reflectance @xmath0 may substantially differ from that measured in vacuum . to obtain optical constants from @xmath0 , therefore , the usual kramers - kronig ( kk ) analysis can not be straightforwardly applied , and either a spectral fitting or a modified kk transform has been used . here we describe an alternative method to perform kk analysis on @xmath0 . this method relies on the usual kk transform with an appropriate cutoff and extrapolation to @xmath0 , and may offer a simpler approach to obtain infrared conductivity from measured @xmath0 .
theoretical interest in the rare decay @xmath6 as a test of the standard model has been renewed by the experimental results of the cleo collaboration @xcite . the viability of calculating the relevant hadronic matrix elements on the lattice was first demonstrated by bernard , hsieh and soni @xcite in 1991 . the computational details and results of this work have been described in references @xcite and @xcite . the hadronic matrix elements can be parametrised by three form factors , @xmath7 where , @xmath8 and @xmath9 is the momentum of the emitted photon . as the photon emitted is on - shell , the form factors need to be evaluated at @xmath10 . in this limit , @xmath11 and the coefficient of @xmath12 is zero in the on - shell matrix element . hence , the branching ratio can be expressed in terms of a single form factor , for example @xmath4 . we calculate with a selection of quark masses near the charm mass and extrapolate to the @xmath13-quark scale . in the heavy quark limit , heavy quark symmetry @xcite tells us that , @xmath14 where @xmath15 is the pseudoscalar mass . combining this with the relation @xmath16 constrains the @xmath17 dependence of the form factors however , it does not provide a scaling law for @xmath4 without further assumptions about the actual @xmath17 behaviour of the form factors . pole dominance ideas suggest that , @xmath18 for @xmath19 , where @xmath20 is a mass that is equal to @xmath15 plus @xmath21 corrections and @xmath22 is a power . since @xmath23 for large @xmath15 , the combination of heavy quark symmetry and the form factor relation at @xmath10 implies that @xmath24 . for example , @xmath25 could be a constant and @xmath26 a single pole , or @xmath25 could be a single pole and @xmath26 a double pole . these two cases correspond to , @xmath27 the data appear visually to favour @xmath25 constant in @xmath17 when @xmath15 is around the charm scale . however , we will consider both constant and single pole behaviours for @xmath25 below . as demonstrated in a previous paper @xcite , the evaluation of @xmath28 is relatively straightforward , and @xmath29 can be determined in a similar way . we fit @xmath26 to a pole or dipole model in order to obtain the on - shell form factor @xmath4 , @xmath30 the difference between the two models was found to be negligible . the form factor @xmath29 was fitted to a pole model or constant the ratio @xmath31 at @xmath10 for dipole / pole and pole / constant fits is shown in fig.([fig : ratio ] ) . the magnitude is found to be consistent with 1 at low masses , in accordance with the identity @xmath32 , eq.([eq : t1_t2_equal ] ) . at higher masses , the dipole / pole fits for @xmath31 deviate less than the pole / constant fits . in order to test heavy quark scaling , we also extracted the form factor @xmath29 at maximum recoil , where @xmath34 , in the same way as bernard _ et al . _ @xcite . in the heavy quark limit , @xmath5 is expected to scale as @xmath35 , analogous to the scaling of @xmath36 . higher order @xmath21 and radiative corrections will also be present . for convenience , we remove the leading scaling behaviour by forming the quantity , @xmath37 the normalisation ensures that @xmath38 at the physical mass @xmath33 . linear and quadratic correlated fits for @xmath39 were carried out with the functions , @xmath40 and are shown in fig.([fig : t2-extrapolation ] ) . taking the quadratic fit of @xmath29 at @xmath41 as the best estimate , and the difference between the central values of the linear and quadratic fits as an estimate of the sytematic error , @xmath29 was found to be , @xmath42 if the @xmath17 dependence of @xmath29 at @xmath33 were known , this result could be related to @xmath4 via the identity @xmath32 . for @xmath4 we test the two possible scaling laws in the same way as for @xmath29 , by forming the quantity , @xmath43 where @xmath44 . for @xmath45 , a similar scaling relationship has been found using light - cone sum rules by ali , braun and simma @xcite . the @xmath46 case has been suggested by other sum rules calculations @xcite . linear and quadratic fits were carried out with the same functions as for @xmath29 . the two cases @xmath44 are shown in fig.([fig : t1-extrapolation ] ) . the @xmath47 are approximately 1 for the scaling laws , indicating that the models are statistically valid in the available mass range . the final results for @xmath48 are taken from the quadratic fit for @xmath49 , with the systematic error estimated as for @xmath29 , @xmath50 further information on the @xmath17 dependence of @xmath49 and @xmath29 is required to remove the uncertainty in obtaining the form factors at the physical point @xmath10 , @xmath51 .
we present updated results on the calculation of the matrix elements for @xmath0 in the quenched approximation on a @xmath1 lattice at @xmath2 , using an @xmath3-improved fermion action . the scaling behaviours of the form factors @xmath4 and @xmath5 for the decay are examined and pole model ansatzes tested .
white dwarfs are almost completely degenerate objects that can not obtain energy from the thermonuclear sources and their evolution is just a gravothermal process of cooling . globally , the evolution of the luminosity of a white dwarf can be written as : @xmath0 where the l.h.s . of this equation represents the sinks of energy , photons , neutrinos and any additional exotic term , while the r.h.s . contains the sources of energy , the heat capacity of the star , the work due to the change of volume , the contribution of the latent heat and gravitationl settling upon crystallization ( the term @xmath1 is the rate of crystallization ) and , finally , the last term represents any additional exotic source of energy @xcite . there are two ways to test the theory of white dwarf evolution , namely , studying the white dwarf luminosity function and using the secular drift of the pulsation period of variable white dwarfs . during the cooling process , white dwarfs experience some phases of pulsational instability powered by the @xmath2- and the _ convective driven_-mechanisms @xcite . depending on the composition of the atmosphere variable white dwarfs are known as dov , dbv and dav . these stars also known as pg1159 or gw vir stars , v777 he stars and zz ceti stars respectively . variable white dwarfs of different compositions occupy different regions in the hertzsprung russell diagram . the value of the pulsation period indicates that these objects are experiencing g mode non radial pulsations , where the main restoring force is gravity . one of the main characteristics of these pulsations is that they experience a secular drift caused by the evolution of the temperature and radius . for qualitative purposes this drift can be approximated by @xcite : @xmath3 where @xmath4 and @xmath5 are constants of the order of unity that depend on the details of the model , and @xmath6 and @xmath7 are the stellar radius and the temperature at the region of period formation , respectively . this equation reflects the fact that , as the star cools down , the degeneracy of the plasma increases , the buoyancy decreases , the brunt - visl frequency becomes smaller and , as a consequence , the spectrum of pulsations gradually shifts to lower frequencies . at the same time , since the star contracts , the radius decreases and the frequency tends to increase . in general , dav and dbv stars are already so cool ( and degenerate ) that the radial term is negligible and the change of the period of pulsation can be directly related to the change in the core temperature of the star . the timescales involved are of the order of @xmath8 s / s for dovs , @xmath9 to @xmath10 s / s for dbvs and @xmath11 to @xmath12 s / s for davs . therefore , the measurement of such drifts provides an effective method to test the theory of cooling white dwarfs . this measurement is a difficult but feasible task , as it has been already proved in the case of g117b15a , a zz ceti star @xcite . these properties allow to build a simple relationship @xcite to estimate the influence of an extra sink of energy , axions for instance , on the period drift of variable white dwarfs : @xmath13 where the suffix `` model '' refers to those models built using standard physics . the white dwarf luminosity function is defined as the number density of white dwarfs of a given luminosity per unit of magnitude interval : @xmath14 where @xmath15 satisfies the condition @xmath16 and @xmath17 , @xmath18 is the mass of the parent star ( for convenience all white dwarfs are labeled with the mass of the main sequence progenitor ) , @xmath19 is the cooling time down to luminosity @xmath20 , @xmath21 is the characteristic cooling time , @xmath22 and @xmath23 are the maximum and the minimum masses of the main sequence stars able to produce a white dwarf of luminosity @xmath20 , @xmath24 is the lifetime of the progenitor of the white dwarf , and @xmath7 is the age of the population under study . the remaining quantities , the initial mass function , @xmath25 , and the star formation rate , @xmath26 , are not known a priori and depend on the properties of the stellar population under study . the computed luminosity function is usually normalized to the bin with the smallest error bar , traditionally the one with @xmath27 , in order to compare theory with observations . equation ( [ lf ] ) shows that in order to use the luminosity function as a physical laboratory it is necessary to have good enough observational data and to know the galactic properties that are used in this equation ( the star formation rate , the initial mass function and the age of the galaxy ) . fortunately , the bright branch of the luminosity function is only sensitive to the average characteristic cooling time of white dwarfs at the corresponding luminosity when the function is properly normalized . the reason is twofold @xcite , on one hand the stellar population is dominated by low mass stars and on the other the lifetime of stars increases very sharply when the mass decreases . the result is that the old galactic populations are still producing bright white dwarfs and the number of such stars at each luminosity bin is the sum of the contributions of all the different episodes of star formation . ) by a 0.6 @xmath28 white dwarf versus its bolometric magnitude , a function that monotonically increases with time . the cross represents the position of g115-b15a.,scaledwidth=60.0% ] we note that in white dwarf case only dfsz axions have to be taken into account , since electron bremsstrahlung is the dominant process in white dwarf interiors . the axion emission rate under these conditions is given by @xcite : @xmath29 where @xmath30 takes into account the coulomb plasma effects , @xmath31 is the temperature in units of @xmath32 k , @xmath33 and @xmath34 are the atomic and mass numbers of the plasma components , respectively , and @xmath35 is the strength of the axion - electron yukawa coupling . figure [ fig : lum ] shows the energy losses of a typical white dwarf star . since the neutrino emission is @xmath36 and @xmath37 the luminosity function allows to disentangle the contribution of the different mechanisms of energy loss . when this method is applied to the luminosity function of white dwarfs the value of @xmath38 that best fits the luminosity function is @xmath39 , but variations of a factor two are still compatible with the observations @xcite . finally , the most recent analysis of g117-b15a shows that the value of @xmath38 quoted here is compatible with the secular drift of the pulsation period , which gives support to the necessity to include an extra cooling term in the white dwarf models @xcite . there are two independent evidences , the luminosity function and the secular drift of dav white dwarfs , that these stars are cooling down more rapidly than predicted . the introduction of an additional sink of energy linked to the interaction of electrons with a light boson ( axion , alp , ) with an strength @xmath40 solves the problem satisfactorily . of course , the remaining uncertainties , both observational and theoretical , still prevent to claim the existence of such interaction . a direct detection under laboratory conditions , like those of the cast helioscope @xcite or the shining through the wall experiments @xcite , could provide unambiguous evidences . in this sense , if the current result is interpreted as due to axions , such particles should have a mass , @xmath41 and should be coupled with photons with @xmath42 gev@xmath43 . this work has been supported by the micinn grants aya08 - 1839/esp and aya2008 - 04211-c02 - 01 , by the esf eurocores program eurogenesis ( micinn grant eui2009 - 04170 ) , by sgr grants of the generalitat de catalunya and by the eu - feder funds . 99 j. isern , e. garca - berro , m. hernanz , & r. mochkovitch , `` the physics of white dwarfs '' , jour . phys . cond . matter , * 10 * , 11263 ( 1998 ) l.g . althaus , a.h . crsico , j. isern , & e. garca - berro , `` evolutionary and pulsational properties of white dwarfs '' , astron . & , in press ( 2010 ) d.e . winget , c. j. hansen , & h. m. van horn , `` do pulsating pg1159 - 035 stars put constraints on stellar evolution ? '' , nature , * 303 * , 781 ( 1983 ) s.o . kepler _ et al . _ , `` measuring the evolution of the most stable optical clock g117-b15a '' , astrophys . , * 634 * , 1311 ( 2005 ) j. isern , m. hernanz , & e. garca - berro , `` axion cooling of white dwarfs '' , astrophys . , * 392 * , l23 ( 1992 ) j. isern , & e. garca - berro , `` white dwarfs as physics laboratories : the axion case '' , mem . soc . astr . it . , * 79 * , 545 ( 2008 ) j. isern , s. cataln , e. garca - berro , & s. torres , `` axions and the white dwarf luminosity function '' , j. phys . : conf . ser . , * 172 * , 012005 ( 2009 ) m. nakagawa , y. kohyama , & n. itoh , `` axion bremsstrahlung in dense stars '' , astrophys . , * 322 * , 291 ( 1987 ) j. isern , e. garca - berro , s. torres , & s. cataln , `` axions and the cooling of white dwarf stars '' , astrophys . jour . , * 682 * , l109 ( 2008 ) j. isern , e. garca - berro , l.g . althaus , & a.h . crsico , `` axions and the pulsation periods of variable white dwarfs revisited '' , astron . & astrophys . , * 512 * , a86 ( 2010 ) t. papaevangelou , `` cast , recent results results and future outlook '' , http : + axion-wimp.desy.de and these proceedings . p. arias , j. jaeckel , j. redondo , a. ringwald `` improving the discovery potential of future light - shining - through - the - wall experiments '' , aerxiv:1009.1519 [ hep - ph ] , http : + axion-wimp.desy.de and these proceedings .
white dwarfs are almost completely degenerate objects that can not obtain energy from the thermonuclear sources and their evolution is just a gravothermal process of cooling . the simplicity of these objects , the fact that the physical inputs necessary to understand them are well identified , although not always well understood , and the impressive observational background about white dwarfs make them the most well studied galactic population . these characteristics allow to use them as laboratories to test new ideas of physics . in this contribution we discuss the robustness of the method and its application to the axion case .
the 4d ising model is believed to belong to the same universality class as the @xmath1 model which plays a central role in relativistic quantum field theory . the grand canonical partition function for the ising model ( which corresponds to the vacuum to vacuum transition amplitude in @xmath1 theory ) in the presence of an external magnetic field @xmath2 is @xmath3 where @xmath4 is the inverse of the boltzmann constant times the temperature , @xmath5 is a coupling constant representing the strength of the intersite interaction ( set to unity in the following ) and @xmath6 represents the total number of sites on the lattice . the value , @xmath7 , of the spin at site @xmath8 , is restricted to @xmath9 . in its generic form , only nearest neighbour interactions are considered and such a link is represented by @xmath10 . the sum runs over all @xmath11 possible configurations of the spin field , and the normalization ensures @xmath12 when @xmath13 . the partition function may be expressed as @xmath14 in which @xmath15 is the dimensionality of the system , @xmath16 is the reduced external magnetic field , and @xmath17 are the configuration energy and magnetization . in @xmath18 the sum is over the @xmath19 nearest neighbours or links of a periodic lattice . the spectral density @xmath20 denotes the relative weight of configurations having given values of @xmath18 and @xmath21 . in the absence of an odd external field a second order phase transition occurs at a critical value @xmath22 of @xmath23 . the reduced temperature @xmath24 is a measure of the distance away from criticality . the partition function @xmath25 can be written as an @xmath26 degree polynomial in the fugacity @xmath27 defined by @xmath28 as @xmath29 in which @xmath30 is an integrated density . that the partition function ( [ polynomial ] ) is analytic for finite @xmath6 establishes that no phase transition can occur in a finite size system . however as @xmath6 is allowed to approach infinity , phase transitions which manifest themselves as points of non analyticity can and do occur . in 1952 lee and yang @xcite showed that the study of the onset of criticality is equivalent to that of the scaling behaviour of the zeroes of the partition function . for a finite system , or in the thermodynamic limit but in the symmetric phase ( @xmath31 ) , the zeroes in @xmath2 are strictly complex and the free energy is analytic in a non vanishing neighbourhood of the real axis . as criticality is approached ( @xmath32 , @xmath33 ) the lee yang zeroes pinch the real @xmath2 axis , precipitating a phase transition . the lee yang theorem states that for the ising model these zeroes lie on the unit circle in the complex fugacity plane ( the imaginary axis in the complex external field plane ) . this theorem holds independent of the size , dimension and structure of the lattice . in the forty years since the ideas of lee and yang were presented , there has been continual interest in this approach to the problem of phase transitions . analytical progress has included alternative and modified proofs of the original circle theorem , extensions of the result to other systems , as well as theorems proving that no zeroes can exist in certain regions ( see @xcite for a review ) . it has been shown rigorously that for isotropic nearest neighbour interactions , and for @xmath34 sufficiently positive ( the symmetric phase ) , there exists a region around @xmath35 which is free from zeroes @xcite . this means there exists a gap @xmath36 where the density of zeroes is zero . the free energy is analytic in @xmath2 in the gap and no phase transition can occur ( as a function of @xmath2 ) . the point @xmath37 , which is a branch point of the partition function , is called the yang lee edge @xcite . one expects that this property ( the existence of a gap ) holds in fact for all @xmath38 . early numerical work on lee yang zeroes involved the exact calculation of the density of states @xmath39 and was therefore restricted to very small lattice volumes @xcite . in the 1980 s , monte carlo histogram approximations to the density of states @xcite , the invention of cluster algorithms @xcite and multi - histogram methods @xcite all provided boosts to the numerical approach . nonetheless , numerical studies are necessarily limited to finite volume . it is , however , the infinite volume limit which is of primary interest . an important analytical breakthrough which related such numerical analyses in finite volume to the thermodynamic limit came when itzykson , pearson and zuber @xcite connected the concept of partition function zeroes to the renormalization group and thereby formulated a finite size scaling theory for these zeroes . their work applies to dimensions of three or less . this was later extended to dimensions of five or more in @xcite . the upper critical dimension of the ising or @xmath1 universality class is @xmath40 . above this the scaling behaviour of the thermodynamic functions simplifies and the critical exponents are exactly those of the mean field theory . below four dimensions the scaling behaviour is of a power law type . at @xmath41 the mean field power law scaling behaviour is modified by multiplicative logarithmic corrections a circumstance intimately related to the expected triviality of the theory @xcite . logarithmic corrections to the finite size scaling of partition function zeroes in four dimensions have recently been identified from a perturbative renormalization group analysis backed up by a high precision numerical study @xcite . recently salmhofer @xcite has proved the existence of a unique density of zeroes in the thermodynamic ( @xmath42 ) limit . the scaling behaviour of the yang lee edge in the thermodynamic limit was studied by abe @xcite and by suzuki @xcite in 1967 for ising models below the upper critical dimension . they found asymptotic forms for the density of zeroes and a power law behaviour for the scaling of the edge in the symmetric phase . here we would like to add to this body of knowledge by presenting some results on lee yang zeroes in the symmetric phase ( @xmath31 ) of the four dimensional ising model . to this end we find an asymptotic form for the density of zeroes in the thermodynamic limit which is sufficient to recover the scaling forms for the specific heat and susceptibility . the finite size behaviour of the location and density of the lee yang zeroes is studied both analytically and numerically . according to the lee yang theorem @xcite the zeroes of the partition function all lie on the unit circle in the complex fugacity plane . denoting the ( @xmath34dependent ) position of these zeroes by @xmath43 the partition function may be written as @xmath44 the largest coefficient @xmath45 plays no rle in the following and we henceforth set it to unity . the free energy density , @xmath46 can be written as @xmath47 the discrete measure @xmath48 is formally given by @xmath49 the @xmath34dependent density of lee yang zeroes on the unit circle in the complex @xmath27 plane is given by @xmath50 and the cumulative density of zeroes @xmath51 is a function monotonically increasing in @xmath52 from @xmath53 to @xmath54 . the free energy is @xmath55 the thermodynamic limit is @xmath56 the coefficients @xmath57 of the polynomial ( [ polynomial ] ) are real and hence @xmath58 . therefore it is sufficient to consider only the interval @xmath59 in the integrals . lee edge @xmath60 is defined by @xmath61 integrating ( [ old2.6 ] ) by parts gives for the free energy @xmath62 the magnetization is then @xmath63 and the zero field susceptibility @xmath64 one expects the contribution of small @xmath52 to be dominant @xcite . in particular we want to study its contribution singular in @xmath34 . expanding the trigonometric functions in ( [ zerofieldx ] ) ( and dropping the constant term ) , @xmath65 in four dimensions and in the symmetric phase the perturbative renormalization group gives @xcite @xmath66 a change of variables is introduced via @xmath67 . then in the critical region where @xmath31 is sufficiently small @xmath68 following @xcite , the upper integral limit can be replaced by infinity near criticality . this leads to the requirement that @xmath69 for fixed @xmath34 the integral is bounded due to the boundedness of @xmath70 . the constancy leads to a differential equation @xcite for @xmath70 with the general solution @xmath71 @xmath72 being an arbitrary function of @xmath73 with @xmath74 . then @xmath75 where @xmath76 . from ( [ free1 ] ) ( and using the fact that @xmath77 ) , one gets the specific heat @xmath78 now the cumulative density of zeroes in four dimensions may be found from ( [ twophi ] ) . in four dimensions one expects the power law scaling behaviour characteristic of dimensions below the upper critical one to be modified by multiplicative logarithmic corrections . assume therefore that the yang lee edge has the scaling behaviour @xmath79 for small @xmath31 and with @xmath80 . this gives @xmath81 } \nonumber \\ & & \times \left\ { 2(1 - 3p+2p^2)\phi(x ) + p(3 - 4p)x\phi^\prime(x ) + p^2x^2\phi^{\prime\prime}(x ) \right\}\end{aligned}\ ] ] where @xmath82 and a prime indicates derivative with respect to @xmath73 . the specific heat is then @xmath83 \int_1^{\frac{\pi}{\theta_c } } { \frac{i(x)}{x } dx } \quad , \label{myabe72 - 3.20a}\ ] ] where @xmath84 is some function of @xmath73 . as @xmath85 ( @xmath86 ) , one has @xmath87 \label{directc}\ ] ] in the symmetric phase ( @xmath31 , @xmath35 ) and near criticality . from perturbation renormalization group analyses it is known @xcite that the zero field specific heat scales as @xmath88 in four dimensions . therefore @xmath89 and @xmath90 . from ( [ assumedlyedge ] ) the yang lee edge in four dimensions scales as @xmath91 this is the same formula as that yielded by mean field theory @xcite . the density of lee yang zeroes is given by ( [ myabe3.4 ] ) as @xmath92 in which @xmath93 is an unknown function . this form is sufficient to recover the singular behaviour of the susceptibility and of the specific heat . the behaviour of the zero at @xmath94 ( for fixed @xmath73 ) as a function of @xmath34 ( @xmath31 ) is given by ( [ densityoflyzeroes ] ) . at fixed @xmath34 , @xmath95 is an unknown function of @xmath96 . kortman and griffiths emphasized the study of the density of zeroes close to the yang lee edge @xcite . using high temperature and high field series they concluded that below the upper critical dimension and for a fixed ( strictly positive ) @xmath34 , the density of zeroes near the edge exhibits a power law behaviour @xmath97 in zero dimensions ( a single site ) @xmath98 is known to be @xmath99 @xcite . for the exactly solvable one dimensional ising model @xmath100 for all @xmath31 @xcite , i.e. , the density of zeroes diverges as the edge is approached . the ising model in the presence of an external field has not been solved in more than one dimension . nonetheless the value of @xmath98 in two dimensions has been found to be @xmath101 by dhar @xcite by mapping the two dimensional ising ferromagnet into a solvable model of three dimensional directed animals . cardy @xcite found the same result by using the conformal invariance of two dimensional systems at the critical point . using high temperature numerical methods , kurtze and fisher @xcite found @xmath102 in three dimensions . it is believed that this values hold independent of the lattice parametrization used @xcite . for the mean field theory @xmath103 @xcite . thus there seems to be a systematic increase of @xmath98 with dimensionality . at criticality @xmath104 , however , the yang lee gap vanishes and one may expect the critical exponent @xmath98 to take on a value different than that in the symmetric phase . now , the density of zeroes is proportional to the discontinuity in the magnetization m crossing the locus of zeroes @xcite @xmath105 the infinite volume behaviour of the magnetization below the upper critical dimension @xmath106 should be recovered from ( [ sig ] ) at @xmath104 and therefore , for @xmath107 , @xmath108 in four dimensions where @xmath109 , one expects the above formulae to be modified by multiplicative logarithmic corrections . there , ( [ bbb ] ) becomes @xcite @xmath110 therefore , in 4d in the thermodynamic limit @xmath111 non perturbative means of calculating thermodynamic functions in spin models are provided by stochastic techniques like monte carlo integration . these numerical methods yield exact results subject only to statistical error . they are however limited to finite lattices . one has to rely on finite size scaling ( fss ) extrapolation methods to gain information on the corresponding thermodynamic limit . let @xmath112 represent the value of some thermodynamic quantity @xmath113 at reduced temperature @xmath34 on a lattice characterized by a linear extent @xmath114 . then , if @xmath115 is the correlation length , the fss hypothesis is that @xcite @xmath116 in four dimensions the scaling behaviour of the correlation length is @xcite @xmath117 its scaling with @xmath114 is @xcite @xmath118 therefore in @xmath41 the scaling variable should include logarithmic terms @xcite , @xmath119 let @xmath120 be the position of the yang lee edge in the complex external magnetic field plane in the thermodynamic limit and let @xmath121 be its finite size counterpart ( i.e. , the position of the lowest lying zero for a system of finite linear extent @xmath114 ) . from ( [ myabe3.9 ] ) , the fss hypothesis applied to the yang lee edge gives @xmath122 fixing @xmath73 ( so that when rescaling @xmath114 , the temperature is also rescaled in such a way as to keep @xmath73 constant ) , we find @xmath123 this fss formula agrees with that derived recently by perturbative renormalization group methods @xcite . the perturbative rg analysis of the finite size @xmath0 model @xcite gives the relationship between the magnetization @xmath124 and external field @xmath2 at reduced temperature @xmath34 @xmath125 where @xmath126 and @xmath127 are constants . at @xmath104 , therefore , @xmath128 for a finite size system the position of the yang lee edge is not zero at @xmath104 and the origin of non vanishing density of zeroes has to be correspondingly shifted as in ( [ sig ] ) . one therefore expects the density of zeroes to be @xmath129 where @xmath130 is the position of the @xmath131 lee yang zero . defining the cumulative density of zeroes at the @xmath131 zero by the fractional total of zeroes up to @xmath130 , @xmath132 we find ( integrating @xmath133 in ( [ smllg ] ) to @xmath134 ) @xmath135 therefore @xmath136 eq.([recover ] ) gives the @xmath137 dependence of the lowest lying zeroes as well as recovering the fss prediction of ( [ toget ] ) . we now compare these fss results with data obtained for the 4d ising model in a high statistics monte carlo calculation . the simulation was done with the swendsen wang cluster updating algorithm @xcite applied to lattices of sizes @xmath138 for lattice sizes with linear extension @xmath139 ( details of the numerics can be found in @xcite ) . histogram approximations to the spectral density @xmath20 of ( [ gndcanising1 ] ) were determined at zero external field and at various values of @xmath23 close to the pseudocritical value ( the value of @xmath23 at which the zero field specific heat peaks ) . for the histogram in the magnetization each of the raw histograms in @xmath18 and @xmath21 were firstly binned in a @xmath140 array . for each @xmath21bin the corresponding @xmath18subhistograms were then combined to multi - histograms according to @xcite . in this way , an optimal histogram in @xmath21 for arbitrary @xmath23 was obtained . from this the partition function may be determined for not too large values of ( complex ) @xmath2 . the critical value of @xmath23 in four dimensions has been determined to @xmath141 @xcite . our data yield only three reliable lee yang zeroes for each lattice size . the reason for this is demonstrated in fig.1 where the contours along which @xmath142 and @xmath143 ( for @xmath144 and @xmath145 ) are plotted . because of the magnification of statistical errors far away from the simulation point @xmath35 these contours fail to cross the imaginary @xmath146 axis when @xmath147 is large . thus the zeroes move off the @xmath147 axis and their positions are unreliable . the remaining lattices give qualitatively similar pictures . table 1 lists the positions of the first lee yang zeroes ( the lee yang edge ) as obtained from the multihistograms for various @xmath23 values near @xmath22 and for all five lattices analyzed . as @xmath23 increases one expects the zeroes to approach the real axis in the thermodynamic limit according to ( [ myabe3.9 ] ) . fig.2 shows the corresponding behaviour for the finite size systems considered . at @xmath22 they should scale according to the fss formula ( [ toget ] ) . table 2 lists the positions of the first three lee yang zeroes as obtained from the multihistogram at our estimated value for the critical coupling in the infinite volume limit , @xmath148 . the errors in the quantities calculated from the multihistograms were estimated by the jackknife method . the data for each lattice size were cut to produce @xmath149 subsamples leading to @xmath149 different multihistograms . these @xmath149 different multihistograms give @xmath149 different results for the quantities in question , whence the variance and bias were calculated . the density of zeroes should behave according to ( [ smllg ] ) or ( equivalently ) ( [ recover ] ) . the log log plot of fig.3 gives a slope of @xmath150 . the deviation from the exponent @xmath151 in ( [ recover ] ) is presumably due to the presence of logarithmic corrections . this may be seen in fig.4 where we remove the expected leading behaviour : a negative slope is clearly identified . in fact a best fit to all ten points gives a slope @xmath152 . the shaded area is bordered by lines of this slope . we find that both leading power law scaling behaviour and multiplicative logarithmic corrections for the density of zeroes ( or equivalently for the distance between zeroes ) are identified in figs . 3 and 4 . this is complementary to our previous analysis in which the scaling behaviour of the actual positions of these zeroes was analyzed @xcite . both approaches yield quantitative agreement with the ( perturbative ) theoretical predictions . the scaling behaviour of the lee yang zeroes and in particular of the yang lee edge in four dimensions and in the thermodynamic limit has been examined . the asymptotic form for the density of zeroes in the infinite volume limit is sufficient to recover the scaling formulae for the specific heat , the magnetization and the magnetic susceptibility . this extends the work of abe and suzuki to the case of four dimensions where mean field power law scaling behaviour is modified by multiplicative logarithmic corrections which are linked to the triviality of the theory . an analytical fss study of the edge and the density of zeroes is in good quantitative agreement with a numerical analysis in the form of monte carlo simulations on finite size lattices . 1234567 t.d . lee and c.n . yang , phys . rev . * 87 * ( 1952 ) 404 ; ibid 410 . r. kenna , ph.d . thesis , universitt graz , 1993 . g. gallavotti , s. miracle sole and d.w . robinson , phys . 25a * ( 1967 ) 493 ; commun . * 10 * ( 1968 ) 311 . fisher , phys . * 40 * ( 1978 ) 1610 . s. katsura , prog . * 11 * ( 1954 ) 476 ; s. katsura , y. abe and m. yamomoto , j. phys . soc . japan * 30 * ( 1971 ) 347 . m. falcioni , e. marinari , m.l . paciello , g. parisi and b. tanglienti , phys . b * 102 * ( 1981 ) 270 ; ibid . * 108 * ( 1982 ) 187 ; ibid . * 108 * ( 1982 ) 331 ; nucl . * b190[fs3 ] * ( 1981 ) 782 ; e. marinari , nucl . phys . * b235[fs11 ] * ( 1984 ) 123 . swendsen and j.s . wang , phys . * 58 * ( 1987 ) 86 . u. wolff , phys . lett . * 62 * ( 1989 ) 361 . g. bhanot , s. black , p. carter and r. salvador , phys . b * 183 * ( 1987 ) 331 ; g. bhanot , r. salvador , s. black , p. carter and r. toral , phys . * 59 * ( 1987 ) 803 . a.m. ferrenberg and r.h . swendsen , phys . lett . * 61 * ( 1988 ) 2635 ; computers in physics , sep / oct 1989 . alves , b.a . berg and r. villanova , phys . b * 41 * ( 1990 ) 383 ; phys . rev . b * 43 * ( 1991 ) 5846 . c. itzykson , r.b . pearson and j.b . zuber , nucl . phys . * b220[fs8 ] * ( 1983 ) 415 ; c. itzykson and j.m.luck , progress in physics , critical phenomena ( 1983 brasov conference ) birkhuser , boston inc . * vol 11 * ( 1985 ) 45 . m.l glasser , v. privman and l.s . schulman , j. stat . * 45 * ( 1986 ) 451 ; phys . b * 35 * ( 1987 ) 1841 . m. aizenman and r. graham , nuclear physics * b225 * ( 1983 ) 261 . c. arago de carvalho , s. caracciolo and j. frhlich , nucl . phys . * b215[fs7 ] * ( 1983 ) 209 . t. hara , j. stat . * 47 * ( 1987 ) 57 ; t. hara and h. tasaki , ibid . 99 . r. kenna and c.b . lang , phys . lett . * 264b * ( 1991 ) 396 ; nucl . phys . * b 393 * ( 1993 ) 461 ; nucl . phys . * b * ( proc . suppl . ) * 30 * ( 1993 ) 697 ; r. kenna , preprint unigraz-26 - 6 - 93 ( 1993 ) . m. salmhofer , ubc preprint , june 1992 . r. abe , prog . * 38 * ( 1967 ) 72 ; r. abe , ibid . 568 . m. suzuki , prog . * 38 * ( 1967 ) 289 ; ibid . 1225 ibid . * 39 * ( 1968 ) 349 . e. brzin , j.c . le guillou and j. zinn - justin , in _ phase transitions and critical phenomena _ , vol . * vi * , ed . c. domb and m.s . green ( academic press : new york , 1976 ) 127 . p. kortmann and r. griffiths , phys . rev . lett . * 27 * ( 1971 ) 1439 . m. bander and c. itzykson , phys . b * 30 * ( 1984 ) 6485 . d. dhar , phys . * 51 * ( 1983 ) 853 . cardy , phys . * 54 * ( 1985 ) 1354 . kurtze and m.e . fisher , j. stat . * 19 * ( 1978 ) 205 ; phys . b * 20 * ( 1979 ) 2785 . m.n . barber , in _ phase transitions and critical phenomena , vol . viii _ , eds . c. domb and j. lebowitz ( academic press , new york : 1983 ) . m. lscher and p. weisz , nucl . * b290 [ fs20 ] * ( 1987 ) 25 ; * b295 [ fs21 ] * ( 1988 ) 65 ; * b300 [ fs22 ] * ( 1988 ) 325 ; * b318 * ( 1989 ) 705 . e. brzin , j. physique * 43 * ( 1982 ) 15 . * table 1 : * the positions of the first lee yang zeroes as obtained from the multihistograms for all five lattices and near @xmath22 . the real part of the zeroes is always zero . @xmath153 & @xmath154 & @xmath144 @xmath23 & @xmath155 & @xmath155 & @xmath155 & @xmath155 & @xmath155 0.149600 & 0.015281 & 0.004511 & 0.001958 & 0.001047 & 0.000637 0.149650 & 0.015091 & 0.004384 & 0.001860 & 0.000966 & 0.000567 0.149703 & 0.014892 & 0.004253 & 0.001761 & 0.000886 & 0.000500 0.149750 & 0.014718 & 0.004140 & 0.001677 & 0.000820 & 0.000447 0.149800 & 0.014535 & 0.004023 & 0.001593 & 0.000756 & 0.000398 0.149850 & 0.014355 & 0.003910 & 0.001512 & 0.000697 & 0.000355 0.149900 & 0.014177 & 0.003800 & 0.001437 & 0.000644 & 0.000319 0.149950 & 0.014001 & 0.003693 & 0.001366 & 0.000597 & 0.000289 0.150000 & 0.013828 & 0.003590 & 0.001299 & 0.000555 & 0.000264 0.150050 & 0.013657 & 0.003491 & 0.001237 & 0.000518 & 0.000244 0.150100 & 0.013488 & 0.003395 & 0.001180 & 0.000486 & 0.000228 0.150150 & 0.013322 & 0.003302 & 0.001127 & 0.000458 & 0.000215 0.150200 & 0.013159 & 0.003213 & 0.001077 & 0.000433 & 0.000204 0.150250 & 0.012997 & 0.003127 & 0.001032 & 0.000413 & 0.000195 0.150300 & 0.012838 & 0.003044 & 0.009909 & 0.000395 & 0.000187 0.150350 & 0.012682 & 0.002965 & 0.009530 & 0.000379 & 0.000180 0.150400 & 0.012527 & 0.002889 & 0.009185 & 0.000366 & 0.000174 * table 2 : * the positions of the first three lee yang zeroes as obtained from the jackknifed multihistograms at @xmath156 . the real part of the zeroes is always zero . @xmath157 & @xmath158 8 & 0.014892(22 ) & 0.033057(48 ) & 0.047357(174 ) 12 & 0.004253(16 ) & 0.009426(15 ) & 0.013349(71 ) 16 & 0.001761(6 ) & 0.003905(22 ) & 0.005388(24 ) 20 & 0.000886(5 ) & 0.001970(12 ) & 0.002743(29 ) 24 & 0.000500(4 ) & 0.001106(5 ) & 0.001541(12 ) * fig 1 : * contours along which @xmath142 ( dotted lines ) and @xmath143 ( full lines ) ( for @xmath144 and @xmath159 ) . + * fig 2 : * the zeroes approach the real axis as @xmath23 increases ; at @xmath22 they should scale with the lattice size @xmath114 according to ( [ toget ] ) . here the triangles , circles , diamonds , stars and crosses correspond to lattice sizes 8,12,16,20 and 24 respectively . + * fig 3 : * the fss of the density of zeroes is given by ( [ recover ] ) . the leading power law behaviour is revealed by a log log plot . here the open diamonds and triangles correspond to @xmath160 and @xmath161 respectively . this gives a slope 0.778(2 ) , the deviation away from 0.75 being due to the presence of logarithmic corrections . + * fig 4 : * data like in fig.3 , but with the leading power law behaviour removed ; we clearly identify the negative exponent in the @xmath162 behaviour . the shaded band indicates the result of a fit giving a slope value of @xmath152 .
the scaling behaviour of the edge of the lee yang zeroes in the four dimensional ising model is analyzed . this model is believed to belong to the same universality class as the @xmath0 model which plays a central role in relativistic quantum field theory . while in the thermodynamic limit the scaling of the yang lee edge is not modified by multiplicative logarithmic corrections , such corrections are manifest in the corresponding finite size formulae . the asymptotic form for the density of zeroes which recovers the scaling behaviour of the susceptibility and the specific heat in the thermodynamic limit is found to exhibit logarithmic corrections too . the density of zeroes for a finite size system is examined both analytically and numerically .
there are empirical evidences that the trading activity , the trading volume and the volatility of the financial markets are stochastic variables with the power - law probability distribution function ( pdf ) @xcite and the long - range correlations @xcite . most of proposed models apply generic multiplicative noise responsible for the power - law probability distribution function ( pdf ) , whereas the long - range memory aspect is not accounted in the widespread models @xcite.the additive - multiplicative stochastic models of the financial mean - reverting processes provide rich spectrum of shapes for pdf , depending on the model parameters @xcite , however , do not describe the long - memory features . empirical analysis confirms that the long - range correlations in volatility arise due to those of trading activity @xcite . on the other hand , trading activity is a financial variable dependant on the one stochastic process , i.e.interevent time between successive market trades . therefore , it can be modeled as event flow of the stochastic point process . recently , we investigated analytically and numerically the properties of the stochastic multiplicative point processes , derived formula for the power spectrum @xcite and related the model with the multiplicative stochastic differential equations @xcite . preliminary comparison of the model with the empirical data of the power spectrum and probability distribution of stock market trading activity @xcite stimulated us to work on the more detailed definition of the model . here we present the stochastic model of the trading activity with the long - range correlations and investigate its connection to the stochastic modeling of the volatility . the proposed stochastic nonlinear differential equations reproduce the power spectrum and pdf of the trading activity in the financial markets , describe the stochastic interevent time as the fractal - based point process and can be applicable for modeling of the volatility with the long - range autocorrelation . trades in financial markets occur at discrete times @xmath0 and can be considered as identical point events . such point process is stochastic and totaly defined by the stochastic interevent time @xmath1 . a fractal stochastic point process results , when at least two statistics exhibit the power - law scaling , indicating that represented phenomena contains clusters of events over all scales of time @xcite . the dimension of the fractal point process is a measure of the clustering of the events within the process and by the definition coincides with the exponent of the power spectral density of the flow of events . we can model the trading activity in financial markets by the fractal point process as its empirical pdf and the power spectral density exhibit the power - law scaling @xcite . in this paper we consider the flow of trades in financial markets as poisson process driven by the multiplicative stochastic equation . first of all we define the stochastic rate @xmath2 of event flow by continuous stochastic differential equation @xmath3 where @xmath4 is a standard random wiener process , @xmath5 denotes the standard deviation of the white noise , @xmath6 is a coefficient of the nonlinear damping and @xmath7 defines the power of noise multiplicativity . the diffusion of @xmath8 should be restricted at least from the side of high values . therefore we introduce an additional term @xmath9 into the eq . ( [ eq : taustoch ] ) , which produces the exponential diffusion reversion in equation @xmath10\tau^{2\mu-2}\mathrm{d}t+\sigma\tau^{\mu-1/2}\mathrm{d}w , \label{eq : taustoch2}\ ] ] where @xmath11 and @xmath12 are the power and value of the diffusion reversion , respectively . the associated fokker - plank equation with the zero flow gives the simple stationary pdf @xmath13\label{eq : taudistrib}\ ] ] with @xmath14 and @xmath15 . ( [ eq : taustoch2 ] ) describes continuous stochastic variable @xmath8 , defines the rate @xmath2 and , after the ito transform of variable , results in stochastic differential equation @xmath16n^{2\eta-1}\mathrm{d}t+\sigma n^{\eta}\mathrm{d}w , \label{eq : nstoch}\ ] ] where @xmath17 and @xmath18 . ( [ eq : nstoch ] ) describes stochastic process @xmath19 with pdf @xmath20 and power spectrum @xmath21 @xcite @xmath22 noteworthy , that in the proposed model only two parameters , @xmath23 and @xmath24 ( or @xmath7 ) , define exponents @xmath25 and @xmath26 of two power - law statistics , i.e. , of pdf and of the power spectrum . time scaling parameter @xmath27 in eq . ( [ eq : nstoch ] ) can be omitted adjusting the time scale . here we define the fractal point process driven by the stochastic differential equation ( [ eq : nstoch ] ) or equivalently by eq . ( [ eq : taustoch2 ] ) , i.e. , we assume @xmath28 as slowly diffusing mean interevent time of poisson process with the stochastic rate @xmath19 . this should produce the fractal point process with the statistical properties defined by eqs . ( [ eq : ndistr ] ) and ( [ eq : nspekt ] ) . within this assumption the conditional probability of interevent time @xmath29 in the modulated poisson point process with the stochastic rate @xmath30 is @xmath31.\label{eq : taupoisson}\ ] ] then the long time distribution @xmath32 of interevent time @xmath29 has the integral form @xmath33\tau^{\alpha}\exp\left[-\left(\frac{\tau}{\tau_{0}}\right)^m\right]\mathrm{d } \tau,\label{eq : taupdistrib}\ ] ] with @xmath34 defined from the normalization , @xmath35 . in the case of pure exponential diffusion reversion , @xmath36 , pdf ( [ eq : taupdistrib ] ) has a simple form @xmath37 where @xmath38 denotes the modified bessel function of the second kind . for @xmath39 more complicated structures of distribution @xmath32 expressed in terms of hypergeometric functions arise . we will investigate how the proposed modulated poisson stochastic point process can be adjusted to the empirical trading activity , defined as number of transactions in the selected time window @xmath40 . stochastic variable @xmath19 denotes the number of events per unit time interval . one has to integrate the stochastic signal eq . ( [ eq : nstoch ] ) in the time interval @xmath40 to get the number of events in the selected time window . in this paper we denote the integrated number of events as @xmath41 and call it the trading activity in the case of the financial market . detrended fluctuation analysis @xcite is one of the methods to analyze the second order statistics related to the autocorrelation of trading activity . the exponents @xmath42 of the detrended fluctuation analysis obtained by fits for each of the 1000 us stocks show a relatively narrow spread of @xmath42 around the mean value @xmath43 @xcite . we use relation @xmath44 between the exponents @xmath42 of detrended fluctuation analysis and the exponents @xmath26 of the power spectrum @xcite and in this way define the empirical value of the exponent for the power spectral density @xmath45 . our analysis of the lithuanian stock exchange data confirmed that the power spectrum of trading activity is the same for various liquid stocks even for the emerging markets @xcite . the histogram of exponents obtained by fits to the cumulative distributions of trading activites of 1000 us stocks @xcite gives the value of exponent @xmath46 describing the power - law behavior of the trading activity . empirical values of @xmath45 and @xmath47 confirm that the time series of the trading activity in real markets are fractal with the power law statistics . time series generated by stochastic process ( [ eq : nstoch ] ) are fractal in the same sense . nevertheless , we face serious complications trying to adjust model parameters to the empirical data of the financial markets . for the pure multiplicative model , when @xmath48 or @xmath49 , we have to take @xmath50 to get @xmath45 and @xmath51 to get @xmath47 , i.e. it is impossible to reproduce the empirical pdf and power spectrum with the same relaxation parameter @xmath23 and exponent of multiplicativity @xmath7 . we have proposed possible solution of this problem in our previous publications @xcite deriving pdf for the trading activity @xmath52 @xmath53 when @xmath54 this yields exactly the required value of @xmath55 and @xmath45 for @xmath50 . nevertheless , we can not accept this as the sufficiently accurate model of the trading activity since the empirical power law distribution is achieved only for very high values of the trading activity . probably this reveals the mechanism how the power law distribution converges to normal distribution through the growing values of the exponent , but empirically observed power law distribution in wide area of @xmath52 values can not be reproduced . let us notice here that the desirable power law distribution of the trading activity with the exponent @xmath47 may be generated by the model ( [ eq : nstoch ] ) with @xmath56 and @xmath57 . moreover , only the smallest values of @xmath8 or high values of @xmath19 contribute to the power spectral density of trading activity @xcite . this suggests us to combine the stochastic process with two values of @xmath7 : ( i ) @xmath58 for the main area of @xmath8 and @xmath19 diffusion and ( ii ) @xmath48 for the lowest values of @xmath8 or highest values of @xmath19 . therefore , we introduce a new stochastic differential equation for @xmath19 combining two powers of the multiplicative noise , @xmath59\frac{n^4}{(n\epsilon+1)^2}\mathrm{d } t+\frac{\sigma n^{5/2}}{(n\epsilon+1)}\mathrm{d}w , \label{eq : nstoch2}\ ] ] where a new parameter @xmath60 defines crossover between two areas of @xmath19 diffusion . the corresponding iterative equation for @xmath61 in such a case is @xmath62\frac{\tau_{k}}{(\epsilon+\tau_{k})^2}+\sigma\frac{\tau_{k}}{\epsilon+\tau_{k}}\varepsilon_{k } , \label{eq : tauiterat2}\ ] ] where @xmath63 denotes uncorrelated normally distributed random variable with the zero expectation and unit variance . ( [ eq : nstoch2 ] ) and ( [ eq : tauiterat2 ] ) define related stochastic variables @xmath19 and @xmath8 , respectively , and they should reproduce the long - range statistical properties of the trading activity and of waiting time in the financial markets . we verify this by the numerical calculations . in figure [ fig:1 ] we present the power spectral density calculated for the equivalent processes ( [ eq : nstoch2 ] ) and ( [ eq : tauiterat2 ] ) ( see @xcite for details of calculations ) . this approach reveals the structure of the power spectral density in wide range of frequencies and shows that the model exhibits not one but rather two separate power laws with the exponents @xmath64 and @xmath65 . from many numerical calculations performed with the multiplicative point processes we can conclude that combination of two power laws of spectral density arise only when the multiplicative noise is a crossover of two power laws as in eqs . ( [ eq : nstoch2 ] ) and ( [ eq : tauiterat2 ] ) . we will show in the next section that this may serve as an explanation of two exponents of the power spectrum in the empirical data of volatility for ` s&p 500 ` companies @xcite . averaged over 100 realisations of series with 1000000 iterations and parameters @xmath66 ; @xmath67 ; @xmath68 ; @xmath69 ; @xmath70 . straight lines approximate power spectrum @xmath71 with @xmath64 and @xmath65 : a ) @xmath21 of the flow @xmath72 with the interevent time @xmath1 generated by eq . ( [ eq : tauiterat2 ] ) , b ) @xmath21 calculated by the fast fourier transform of @xmath19 series generated by eq . ( [ eq : nstoch2]).,title="fig : " ] averaged over 100 realisations of series with 1000000 iterations and parameters @xmath66 ; @xmath67 ; @xmath68 ; @xmath69 ; @xmath70 . straight lines approximate power spectrum @xmath71 with @xmath64 and @xmath65 : a ) @xmath21 of the flow @xmath72 with the interevent time @xmath1 generated by eq . ( [ eq : tauiterat2 ] ) , b ) @xmath21 calculated by the fast fourier transform of @xmath19 series generated by eq . ( [ eq : nstoch2]).,title="fig : " ] empirical data of the trading activity statistics should be modeled by the integrated flow of events @xmath52 defined in the time interval @xmath73 . in figure [ fig:2 ] we demonstrate the probability distribution functions @xmath74 and its cumulative form @xmath75 calculated from the histogram of @xmath52 generated by eq . ( [ eq : tauiterat2 ] ) with the selected time interval @xmath76 . this illustrates that the model distribution of the integrated signal @xmath52 has the power - law form with the same exponent @xmath47 as observed in empirical data @xcite . calculated from the histogram of @xmath52 generated by eq . ( [ eq : tauiterat2 ] ) with the selected time interval @xmath76 . b ) cumulative distribution @xmath77 . other parameters are as in figure [ fig:1 ] . ] the power spectrum of the trading activity @xmath52 has the same exponent @xmath45 as power spectrum of @xmath19 in the low frequency area for all values of @xmath40 . the same numerical results can be reproduced by continuous stochastic differential equation ( [ eq : nstoch2 ] ) or iteration equation ( [ eq : tauiterat2 ] ) . one can consider the discrete iterative equation for the interevent time @xmath61 ( [ eq : tauiterat2 ] ) as a method to solve numerically continuous equation @xmath10\frac{1}{(\epsilon+\tau)^2}\mathrm{d}t + \sigma\frac{\sqrt{\tau}}{\epsilon+\tau}\mathrm{d}w . \label{eq : taucontinuous}\ ] ] the continuous equation ( [ eq : nstoch2 ] ) follows from the eq . ( [ eq : taucontinuous ] ) after change of variables @xmath2 . we can conclude that the long - range memory properties of the trading activity in the financial markets as well as the pdf can be modeled by the continuous stochastic differential equation ( [ eq : nstoch2 ] ) . in this model the exponents of the power spectral density , @xmath26 , and of pdf , @xmath25 , are defined by one parameter @xmath78 . we consider the continuous equation of the mean interevent time @xmath8 as a model of slowly varying stochastic rate @xmath30 in the modulated poisson process ( [ eq : taupoisson ] ) . in figure [ fig:3 ] we demonstrate the probability distribution functions @xmath79 calculated from the histogram of @xmath29 generated by eq . ( [ eq : taupoisson ] ) with the diffusing mean interevent time calculated from eq . ( [ eq : taucontinuous ] ) . : open circles , calculated from the histogram of @xmath29 generated by eq . ( [ eq : taupoisson ] ) with the mean interevent time calculated from eq . ( [ eq : taucontinuous ] ) ; open squares , calculated form the iterative equation ( [ eq : tauiterat2 ] ) . used parameters are as in figure 1 . straight line approximates power law @xmath80 . ] numerical results show good qualitative agreement with the empirical data of interevent time probability distribution measured from few years series of u.s . stock data @xcite . this enables us to conclude that the proposed stochastic model captures the main statistical properties including pdf and the long - range correlation of the trading activity in the financial markets . furthermore , in the next section we will show that this may serve as a background statistical model responsible for the statistics of return volatility in widely accepted geometric brownian motion ( gbm ) of the financial asset prices . the basic quantities studied for the individual stocks are price @xmath81 and return @xmath82 let us express return @xmath83 over a time interval @xmath40 through the subsequent changes @xmath84 due to the trades @xmath85 in the time interval @xmath86 $ ] , @xmath87 we denote the variance of @xmath84 calculated over the time interval @xmath40 as @xmath88 . if @xmath84 are mutually independent one can apply the central limit theorem to sum ( [ eq : return2 ] ) . this implies that for the fixed variance @xmath88 return @xmath83 is a normally distributed random variable with the variance @xmath89 @xmath90 where @xmath91 is the normally distributed random variable with the zero expectation and unit variance . , averaged over 10 intervals calculated from the series of @xmath92 generated by eqs . ( [ eq : nstoch2 ] ) and ( [ eq : return4 ] ) , all parameters are the same as in previous calculations . dashed line approximates the power law @xmath93 . ( b ) power spectral density @xmath21 of @xmath94 calculated from fft of the same series @xmath92 . straight lines approximate power spectral density @xmath95 with @xmath96 and @xmath97 . ] empirical test of conditional probability @xmath98 @xcite confirms its gaussian form , and the unconditional distribution @xmath99 is a power - law with the cumulative exponent @xmath100 . this implies that the power - law tails of returns are largely due to those of @xmath101 . here we refer to the theory of price diffusion as a mechanistic random process @xcite . for this idealized model the short term price diffusion depends on the limit order removal and this way is related to the market order flow . furthermore , the empirical analysis confirms that the volatility calculated for the fixed number of transactions has the long memory properties as well and it is correlated with real time volatility @xcite . we accumulate all these results into the assumption that standard deviation @xmath101 may be proportional to the square root of the trading activity , i.e. , @xmath102 . this enables us to propose a simple model of return @xmath103 and related model of volatility @xmath104 based on the proposed model of trading activity ( [ eq : nstoch2 ] ) . we generate series of trade flow @xmath92 numerically solving eq . ( [ eq : nstoch2 ] ) with variable steps of time @xmath105 and calculate the trading activity in subsequent time intervals @xmath40 as @xmath106 . this enables us to generate series of return @xmath83 , of volatility @xmath107 and of the averaged volatility @xmath108 . in figure [ fig:4 ] we demonstrate cumulative distribution of @xmath109 and the power spectral density of @xmath110 calculated from fft . we see that proposed model enables us to catch up the main features of the volatility : the power law distribution with exponent @xmath111 and power spectral density with two exponents @xmath96 and @xmath97 . this is in a good agreement with the empirical data @xcite . starting from the concept of the fractal point processes @xcite we proposed process driven by the nonlinear stochastic differential equation and based on the earlier introduced stochastic point process model @xcite . this may serve as a possible model of the flow of points or events in the physical , biological and social systems when their statistics exhibit power - law scaling indicating that the represented phenomena contains clusters of events over all scales . first of all , we analyze the statistical properties of trading activity and waiting time in financial markets by the proposed poisson process with the stochastic rate defined as a stand - alone stochastic variable . we consider the stochastic rate as continuous one and model it by the stochastic differential equation , exhibiting long - range memory properties @xcite . further we propose a new form of the stochastic differential equation combining two powers of multiplicative noise : one responsible for the probability distribution function and another responsible for the power spectral density . the proposed new form of the continuous stochastic differential equation enabled us to reproduce the main statistical properties of the trading activity and waiting time , observable in the financial markets . in the new model the power spectral density with two different scaling exponents arise . this is in agreement with the empirical power spectrum of volatility and implies that the market behavior may be dependant on the level of activity . one can observe at least two stages in market behavior : calm and excited . finally , we propose a very simple stochastic relation between trading activity and return to reproduce the statistical properties of volatility . this enabled us to model empirical distribution and long - range memory of volatility .
we propose a model of fractal point process driven by the nonlinear stochastic differential equation . the model is adjusted to the empirical data of trading activity in financial markets . this reproduces the probability distribution function and power spectral density of trading activity observed in the stock markets . we present a simple stochastic relation between the trading activity and return , which enables us to reproduce long - range memory statistical properties of volatility by numerical calculations based on the proposed fractal point process .
the grain - scattered halos of x - ray sources contain clues to the nature and distribution of interstellar dust grains . chandra _ x - ray observatory provides the means to achieve substantial improvements over previous observations of x - ray halos . here we present results from a _ chandra _ observation and a monte carlo simulation of the halo of the eclipsing x - ray binary pulsar 4u 1538 - 52 ( hereinafter x1538 ) , an especially interesting halo because of its comparatively high intensity , evidence in the shapes of its radial profiles for a concentration of dust in discrete clouds along the line of sight , and high variability caused by eclipses and flares . the theory of grain - scattered halos of steady sources , first discussed by @xcite , has been developed by various authors , e.g. @xcite . the effects of intensity variations of a source on the variation of its halo and the possibilities this can present for measuring the distance of the source and the distribution of the scattering grains along the sight line have been discussed by @xcite . @xcite calculated the expected profiles of halos that would be produced by grains with various assumed size distributions . @xcite carried out an analytical study of the possibilities of using halos as diagnostics of grains . @xcite discussed the possibilities and the practical limitations of such diagnostics . @xcite showed that the rayleigh - gans approximation to the mie scattering theory , commonly employed in theoretical treatments x - ray halos , yields substantial overestimates of halo intensities for photon energies less than 1 kev . early evidence of x - ray halos was reported by @xcite and by @xcite from images recorded by the _ einstein _ mauche and collaborators @xcite , employing the same observatory , found a correlation between the fractional intensity of the halo of an x - ray source and the visual extinction of its optical counterpart that was consistent with the model of solid silicate and graphite grains proposed by @xcite . @xcite carried out an extensive survey of x - ray halos with the _ rosat _ x - ray observatory . @xcite found evidence for a distribution in grain size extending to 2 @xmath5 m and possibly more in a _ observation of the halo of nova cygni 1992 . @xcite derived an estimate of the distance to cyg x-3 by an analysis of its x - ray halo . @xcite employed the non - imaging _ ginga _ observatory to observe x1538 and its halo . from a comparison of the counting rates before and after an eclipse of x1538 they derived an estimate of the fractional halo intensity that was substantially less than that expected on the assumption that the visual extinction of its optical companion is caused by solid silicate grains with a specific density near 3 . they interpreted the result as consistent with a grain structure in the form of loose aggregates of smaller particles . following its discovery in the _ uhuru _ galactic plane survey @xcite , x1538 was observed to pulse with a period of 529 s and to have an orbital period of 3.73 days with an eclipse lasting about 12 hours @xcite . an accurate position determination with _ sas 3 _ @xcite led to identification of the highly reddened bo supergiant , qv nor , as its companion @xcite . the x - ray spectrum has a prominent iron k line @xcite and a pulse phase - dependent cyclotron resonance feature at 20 kev @xcite . properties of the binary system derived from various observations are listed in table 1 . for the present study x - ray images of x1538 and its halo were recorded before and after an eclipse immersion . radial profiles of simulated halos , generated by a monte carlo ray - tracing code , were fitted to the three - peaked time - varying radial profiles of the observed halo by adjustment of the parameters of a model of the source spectrum , the dust grains , and the interstellar photoelectric absorption . the model parameters that yielded a good fit are compared with the optical extinction of qv nor and with the column density of atomic hydrogen derived from the spectrum of 21-cm radiation in the direction of the source to derive information about the scattering efficiency of the grains and the total column density of hydrogen . section 2 describes the observations and data reduction . properties of the halo are presented in section 3 . the monte carlo simulation code is described briefly in section 4 together with the model and the fitting procedure . section 5 is a summary and discussion of results . section 6 lists conclusions drawn from this study . data were obtained from the imaging spectrometer ( acis - i ) at the focal plane of the x - ray telescope of the _ chandra _ observatory during a continuous 25 ksec observation ( chandra archive obs_id 90 ) beginning 2000 april 8.95 ( @xmath6mjd 51642.94881 ) at orbital phase 0.9185 , shortly before an eclipse immersion , and ending near mid - eclipse , at orbital phase 0.001 . the received data for each detected photon consist of the transfer time , @xmath7 , of its 3.200-s data frame , its arrival direction expressed as x(ew),y(sn ) sky coordinates in units of @xmath8 , and its energy with an uncertainty of about 0.1 kev . point response functions ( psfs ) of the imaging system were derived from a combination of results from a 30 ksec observation of the x - ray bright quasar 3c 273 and a 12 ksec observation of the much fainter quasar pg1634 + 70 . we prepared arrays of the effective area per second of exposure in @xmath9 pixels as a function of sky position , averaged over the duration of the observation and expressed in units of @xmath10 per second of observation , for eighteen energies between 0.5 and 8.0 kev . in our tabulation of the received data for use in the analysis we replaced the transfer time @xmath7 by @xmath11 , and add the reciprocal of the effective area per second of exposure ( rea ) as determined according to its sky position and energy . the arrival directions of all events in the energy range from 2 to 6 kev from the approximate beginning of total eclipse at @xmath12 ksec to near mid - eclipse at 25 ksec are plotted relative to the direction of x1538 as dots in fig . the image of x1538 is in the northeast corner of the region scanned by ccd chip i3 . the halo image , which persists with gradually diminishing intensity during the eclipse , is conspicuous in a circular region spread over the four chips . regions of the sky scanned by ccd pixels near the edges of the chips have relatively low effective areas achieved operationally by dithering the telescope orientation in a pattern centered on a fixed direction with a cycle time time of 16 s. we excluded from our analysis events in these regions to avoid the large statistical fluctuations they would otherwise cause in summations of rea . we searched for other sources in the field of view by mapping all the events recorded during the interval from 1039 to 25,039 s in the energy range from 2 to 6 kev in @xmath13 pixels . we eliminated any significant effect of other sources on the halo profiles by excluding events within @xmath14 of the centers of pixels containing 5 or more events and more than 10@xmath15 from the image center . finally , we excluded events within @xmath14 of the center line of the `` transfer streak '' which is the trailed image of x1538 running parallel to the southeast edge of the region scanned by chip 3 and formed during the 0.041 s shift of the ccd data to the frame store for subsequent transmission . 1(b ) is a map of the events remaining after application of the exclusion criteria . to determine the effective solid angle on the sky corresponding to a given region of the image we generated @xmath16 events with random positions distributed uniformly over the region and computed the fraction of events accepted according to the exclusion criteria . the effective solid angle was evaluated as the product of that fraction times the geometrical solid angle of the region . the incident intensity of photons corresponding to a set of events recorded within a specified region of the image and interval of time was computed as the sum of the reas of the events divided by the product of the exposure time times the solid angle . the fractional error of an intensity was estimated as the reciprocal square root of the number of events . data from within @xmath14 of the image center recorded before the eclipse are severely degraded by pileup , i.e. , multiple occurrences of events in the same pixel during the 3.200-s frame exposure . under these circumstances the transfer streak is a useful source of data on the source flux unaffected by pileup , albeit with exposures smaller than the exposures of the frame data by the product of the fraction of the exposure time occupied by transfers ( 0.0128 ) times the fraction of the streak used . to extract the streak data , we summed the rea of events in a @xmath17-wide rectangle centered on the transfer streak and extending about half the length of the streak . background intensity was derived from events in parallel rectangles between @xmath18 and @xmath19 from the center line of the streak . during the time before the eclipse , when the incident intensity of unscattered photons from x1538 was highest , the average rate of all events in the angle range from @xmath20 to @xmath21 was less than @xmath22 s@xmath4 arcsec@xmath23 . thus the effects of pileup on measured photon intensities at angles greater than @xmath20 was negligible . we derived halo profiles from events in the angle range from @xmath20 to @xmath24 and preliminary estimates of the background intensities from events in the angle range from @xmath25 to @xmath26 . since the latter events included some halo events , background intensities were included along with the psf normalizations among the parameters adjusted in the process of fitting the profiles of the simulated halo to the observed profiles . the pre - eclipse spectrum of the source is displayed in fig . it was derived from the transfer streak during seven complete pulse periods from 1037 to 4739 s. the high value in the 6.0 - 6.5 kev bin is due to the iron k line . the sharp decline below 4.0 kev is caused by strong circumstellar absorption that occurs as the sight line grazes the limb of qv nor . the histogram , derived from the function @xmath27\right\ } \exp[-n_{\rm h}\sigma_{\rm pe}(e)-\tau_{\rm sca}(e/{\rm kev})^{-2}],\nonumber\\ \hspace{3in}1037 ~{\rm s}<t<4739 ~{\rm s},\end{aligned}\ ] ] was fitted by least squares to the data , with @xmath28 , @xmath29 kev , @xmath30 kev , and the other parameter values listed in table 2 . the quantity @xmath31 is the photoelectric cross section per hydrogen atom of @xcite , and @xmath32 is the total column density of hydrogen along the sight line from the neutron star . the optical depth for scattering is represented by @xmath33 . 2(b ) displays the source spectrum during the eclipse , derived from events within @xmath14 of the image center from 5 to 25 ksec when pileup caused only a small reduction in detection efficiency . during the eclipse , x - rays in the image core are fluorescent and thomson - scattered radiation generated in distant circumstellar matter without strong circumstellar absorption . the iron k radiation is especially prominent in the eclipse spectrum . in our analysis and simulation of the halo we restricted consideration to photons with energies in the range from 2 to 6 kev to exclude , on the high end , the iron line photons and , on the low end , low - energy photons for which the scattering cross section is less accurately represented by the rayleigh - gans approximation used in the simulations . in the upper panel of fig . 3 the incident flux of photons in the energy range from 2 to 6 kev within @xmath14 of the x1538 image center is plotted as open circles against orbital phase ( bottom abscissa ) and time from the start of the observation ( top abscissa ) . solid circles show the flux derived from events in the transfer streak . in both cases integration times were set equal to the pulse period of 529 s before the eclipse and to 4952 s after the eclipse when pulsations were no longer present . the sharp drop in the streak rate near orbital phase 0.933 indicates the eclipse edge corresponding to an eclipse half - angle of @xmath34 . the drastically lower values of the open circles before the eclipse are caused by pileup . after the eclipse , when x - rays scattered from the circumstellar matter in the x1538/qv nor system are still present in the source image , the core pileup is much less severe and yields values of greater statistical accuracy , though still systematically less than the streak values . light curves of the background - subtracted halo in two energy ranges , 2 to 4 kev and 4 to 6 kev , are shown in the lower panel of fig . the initial decline of the halo flux at the eclipse is much less than that of the direct source due to the continuing arrival after the eclipse of photons delayed by scattering . the straight lines fitted by least squares to the data from 5 to 25 ksec show the decay of the halo flux during the eclipse . 4 displays the spectra of the halo as average intensities in successive intervals of angle from the source direction plotted against energy . the spectra were derived from background - subtracted data recorded during the eclipse from 5 to 25 ksec . they exhibit the progressive softening with increasing angle of scattering expected from scattering theory . radial profiles of the halo , in two energy ranges and three time intervals before and after the eclipse , are displayed in fig . the solid circles represent values of the incident intensity of photons in thirty @xmath19 intervals of angle from @xmath20 to @xmath24 from the source , with backgrounds and psfs of the source subtracted . the error bars indicate the poisson counting uncertainties of the raw data . the profiles appear to have a three - peak shape . the outer two peaks are clearly defined by the low values of the intensities in bins 5 and 10 of panel ( 3 ) and bins 6 and 13 of panel ( 5 ) . the low values in bin 1 of all three of the 2 - 4 kev profiles , though of low statistical significance , indicate the possible presence of a third peak . a uniform distribution of dust gives rise to a halo the profile of which , after an eclipse , develops a single broad peak whose inner bound moves toward larger angles as time from eclipse ingress increases . to explain the formation of multiple peaks , it therefore appears necessary to assume that the dust is not uniformly distributed . the problem we tackled was to find a model of the non - uniform grain distribution and the variable source spectrum that can account for the profiles of a halo produced by multiple scattering near the line of sight . we developed a monte carlo ray - tracing code to circumvent the difficulties of deriving halo profiles from such a model by analytical computation . the code generates the energies , arrival directions , and arrival times of contributions to a halo from scatterings of trial photons emitted by a source with a specified variable spectrum and scattered by dust grains with a specified distribution in size and a specified distribution of the optical depth for scattering along the sight line . the code launches a specified number of trial photons from the source each second beginning far enough in the past to assure that all significant contributions to the halo during the observation are included , and ending at the termination of the observation . each trial photon has an energy selected at random from the spectrum . it is tracked in three dimensions till it passes beyond the observer or has accumulated so great a delay in transit as to arrive after the end of the observation . the trajectory of a trial photon is described in a rectangular coordinate system with its origin at the source , and the z axis along the sight line to the observer . distance from the source along the z axis is expressed as a fraction , @xmath35 , of the total distance , @xmath36 , from source to observer . the delay in the arrival time of a contribution to the halo intensity from a given scattering , relative to the arrival time of an unscattered photon , is the sum of the delays in successive path segments each of which is computed in the small angle approximation as @xmath37 , where @xmath38 is the projected position on the z axis of the @xmath39th scattering , @xmath40 is the angle that the @xmath39th segment makes with the z axis , and @xmath41 is the speed of light . the contribution to the halo intensity from each scattering is tabulated according to its energy , arrival angle , arrival time , the identification number of the dust cloud in which the scattering occurred , and the number of the scattering . we assume the photoelectric absorption and grain scattering depend only on the z component of position . at each scattering the size of the grain is randomly selected from a specified size distribution . we assume the probability per steradian that a photon of energy @xmath42 scattered by a grain of size @xmath43 will be deflected at an angle @xmath44 varies with size and energy according to the relation @xmath45^{2}\phi(u)^{2},\ ] ] where @xmath46 is the form factor for homogeneous spherical particles of radius @xmath43 , and @xmath47 where @xmath48 is planck s constant . for @xmath49 kev and @xmath50 we can , with good accuracy , set the atomic scattering factor @xmath51 @xcite , and represent the form factor by the rayleigh - gans approximation @xcite , @xmath52 the code operates on a model that specifies * the distance of the source ; * the time - dependent spectrum of x - ray photons coming directly from the x1538/qv nor system before and after the observation begins ; * the distribution in size of dust grains ; * the scattering optical depth measured from the source along the line of sight as a function of @xmath35 ; * the optical depth for photoelectric absorption ; * the background intensity ; * the psf of the imaging system . at the conclusion of the computation the accumulated data are analyzed for the properties of the simulated halo in a manner similar to that employed with real data , but with the added feature of having available for each contribution the number of the scattering and the identification of the cloud from which it came . in addition , we sum the contributions in intervals of the launch times to check whether the specified beginning launch time is sufficiently early to capture all significant contributions to the simulated halo . statistical fluctuations in monte carlo simulations and strong cross - correlations in the effects of changes in the model parameters preclude effective use of the code in an automatic curve - fitting procedure . instead , we sought a fit of simulated profiles to observed profiles in multiple trials with adjustments of the model parameters aimed at reducing the mean square deviation between the observed and simulated profile intensities . figure 14 shows a simulated halo profile that can be compared with an analytical solution computed by @xcite for a steady source of 1 kev photons scattered by 1 @xmath53 grains distributed uniformly along the sight line with a scattering optical depth of 2 . the simulated and analytical profiles of the halo components from first and second scatterings are in close agreement . small systematic differences appear in the third and fourth scattering components which may be due to a geometrical approximation in the analytical solution , namely , the use at each scattering beyond the first of a `` spherical coordinate system at the grain with a polar angle @xmath54 measured from the point source '' instead of from the previous scattering site . figure 15 shows the evolution of this model halo that would occur following an eclipse immersion at 20 ksec and emersion at 60 ksec . in many trials , good fits of simulated profiles to the observed profiles were found with widely different sets of model parameters . effects of a change in a given parameter can often be reversed by a compensating change in another . for example , increase in halo intensity caused by an increase in the source flux can be approximately cancelled by a decrease in the scattering optical depth that leaves the product of the two parameters unchanged . change in the angular position of a feature in the halo of a variable source caused by a decrease in the assumed distance , @xmath55 , of a scattering site can be offset by a decrease in the assumed distance @xmath36 of the source that leaves the quantity @xmath56 unchanged . therefore , to obtain results that might provide useful information about the distance of the source and the properties and distribution of the dust , it was necessary to fix as many of the parameters as possible at plausible values derived from this and other observations . previous studies show that the spectrum of x1538 is highly variable during a single binary orbit . to allow for this , we represent the incident spectrum at time @xmath57 and orbital phase @xmath58 by @xmath59+s_{\rm pre}(e , t)+s_{\rm ecl}(e),\ ] ] in which @xmath60 is the long - term average incident spectrum of photons that arrive directly from the neutron star , @xmath61 $ ] is a factor representing flares that may have occurred before the observation began , @xmath62 is the observed pre - eclipse spectrum of x1538 derived from the transfer streak , and @xmath63 is the spectrum of photons observed during the eclipse . we assume the latter , characterizing photons produced by x - ray illumination of the circumstellar matter in the x1538/qv nor system , is constant throughout the binary orbit . we represent the average spectrum by @xmath64.\ ] ] the quantity @xmath65 is the column density of hydrogen between x1538 and the observer , and @xmath31 is the photoelectric cross section per hydrogen atom . we assume @xmath65 is the sum of a constant interstellar component , @xmath66 , and a variable circumstellar component , @xmath67 , \hspace{3.0 cm } \pi<0.5,\\ & = & n_{\rm ecl}\exp[-(1-\pi-\pi_{\rm ecl})/\lambda_{\rm in } ) , \hspace{2.4 cm } \pi>0.5,\end{aligned}\ ] ] where @xmath68 is the orbital phase of total eclipse , and @xmath69 is the column density as the sight line grazed the edge of qv nor . the average spectrum was derived from data in the archive of the all sky monitor ( asm ) of the _ rossi x - ray timing explorer _ the archive provides a background - subtracted aspect - corrected count rate for a given source whenever it is in the field of view of one of three detectors during an orientation dwell in the course of its all - sky scan . we used the 16752 b - channel ( nominal 3 - 5 kev ) count rates in the range from -3.0 to 3.0 s@xmath4 of detectors 1 and 2 during the time interval from mjd 50088 to 52400 when the corresponding rates for the crab nebula were steady with an average of 23.08 s@xmath4 . the data were folded in 50 intervals of orbital phase derived from the orbital parameters listed in table 1 . the average rates are plotted against orbital phase in fig . the smooth curve is the function @xmath70 we set @xmath66 and @xmath71 to the values derived in fitting the simulated profiles to the observed profiles . the phase of total eclipse , @xmath68 , was set to the value indicated by the vertical line in fig . the factor @xmath72 is the effective area of the asm detectors normalized to yield a predicted 3 - 5 kev count rate for the crab nebula equal to the observed rate under the assumption that the crab spectrum is @xmath73 $ ] @xmath1 s@xmath4 kev@xmath4 . the function was fitted by least squares to the asm data by adjustment of @xmath74 , @xmath75 , @xmath76 , and @xmath77 . the quantity @xmath78 , representing the counting rate in eclipse , is a measure of the sum of rates due to photons scattered in the circumstellar matter of the x1538/qv nor system , a portion of the x1538 halo , other sources in the field of view , and possible errors in the background subtraction . as such , its presence in equation ( 9 ) makes the fitted value of @xmath79 a good measure of the average spectrum of photons that would be incident in the absence of scattering and photoelectric absorption . we ignore the small error due to halo contamination that remains because the halo contribution to @xmath78 decays during the eclipse . the fixed and fitted values of the average spectrum parameters are listed in table 3 . the observed flux of x1538 before the eclipse in the interval from 1037 to 4739 s is more than an order of magnitude larger than the value implied by the average spectrum fitted to the asm data . the column density of the spectrum fitted to the pre - eclipse streak data , though large , is much less than that of the asm average spectrum in the same orbital phase range . thus it appears that there may have been , in effect , a kind of hole in the circumstellar matter that allowed an attenuated flux from x1538 to shine through . to model this portion of the pre - eclipse spectrum , which has a substantial effect on the simulated halos at small angles , we added the fitted streak spectrum previously described , @xmath62 , to the model spectrum in the phase interval from 0.922 to 0.933 . we set @xmath80 @xmath1 s@xmath4 kev@xmath4 to represent the observed source spectrum during the eclipse displayed in fig . 2(b ) , and which is nearly flat in the energy range from 2 to 6 kev . it made only a very small contribution to the simulated halos . characteristic flare phenomena of x1538 are shown in fig . 6(b ) which is a plot against orbital phase of the average counting rates of x1538 in intervals of the pulse period recorded by the proportional counter array ( pca ) of the rxte in a pointed - mode observation @xcite . during a flare the count rates can rise above average by a factor of five or more for several pulse periods . photons from such a flare , scattering from dust in a discrete cloud , can produce a sharp feature in the halo profile which moves to larger angles as the time interval between the flare occurrence and the observation increases . to add a @xmath39th flare to the spectrum model we specified a constant value , @xmath81 , of @xmath82 beginning at a certain time @xmath83 and lasting for an orbital phase interval of 0.01 which corresponds to a time interval of 3222 s. with the incident spectrum before and during the observation specified , the spectrum that would be incident in the absence of interstellar photoelectric absorption and scattering is related to the incident spectrum by the equation @xmath84.\ ] ] in simulations the energy of a trial photon is selected at random from this source spectrum . the distribution in size of the grains was specified by the power law @xcite @xmath85 we call @xmath86 the optical depth from the source to @xmath35 for scattering photons of energy @xmath42 , and note that @xmath87 . for a clue to the distribution of the dust grains along the sight line we examined the spectrum of 21-cm radiation in the direction of x1538 ( l@xmath88=3274,b@xmath89=21 ) , kindly provided by n. mcclure - griffiths from the southern galactic plane survey @xcite and displayed in fig . the sight line is sufficiently inclined with respect to the galactic plane so that we can safely assume most of the hydrogen along it lies between earth and the tangent point at 6.7 kpc . assuming @xmath90 kpc and @xmath91 km s@xmath4 , we converted the doppler velocities to radial distance to obtain the brightness temperature as a function of distance , @xmath55 , as plotted in fig . we explored the viability of a model with dust density simply proportional to the brightness temperature , i.e. , @xmath92 , and the source spectrum described above without flares . 8 illustrates the gross failure of such a model to yield a simulated halo with a profile that conforms to the peaks and valleys of the observed profiles . a halo due to scattering from a uniform distribution of dust decays slowly outward from the center , giving rise to a profile with a peak that moves to larger angles as time from eclipse ingress increases . if the dust is concentrated in a narrow cloud , the hollowing out occurs more rapidly . the rate at which the peak moves toward larger angles depends on the location of the cloud and the distance of the source . with this in mind , and with the intent to produce multi - peaked profiles with a model of dust concentrated in several narrow clouds , we call @xmath93 the scattering optical depth of the @xmath39th cloud at distance @xmath94 for photons of energy @xmath42 with @xmath95 distributed uniformly from @xmath96 to @xmath97 . with all the grains located in such clouds , @xmath98 . we placed two clouds at the distances of the two prominent peaks of atomic hydrogen density at @xmath99 kpc and @xmath100 kp , and explored how well the resulting two peaks in the profile intensity could be made to conform to the outer two peaks of the observed profiles by adjustment of the source distance and the cloud thicknesses . with @xmath99 kpc and @xmath100 kpc , and with @xmath101 and @xmath102 adjustable , we generated halos for source distances of 4.0 , 4.5 , 5.0 , and 6.0 kpc , all within the wide range of uncertainties of published estimates based on photometric measurements of qv nor . as in the previous example , we assumed the source spectrum was as described above without flares . the results , displayed in fig . 9 , show that the peaks shift , as expected , to larger angles as the source distance is increased . for distances of 5 and 6 kpc the simulated profiles are poor matches to the observed profiles . the fit for 4.5 appears slightly better than for 4.0 kpc . though 4.5 kpc is near the lower limit of photometrically plausible distances , we adopted it for the model , with the expectation that the poor fit at angles below @xmath103 could be repaired by placing a third cloud closer to the source . the six profiles displayed as solid circles with error bars in fig . 11 were the targets of the fitting process . they were compiled from the same data as were the six profiles in fig . 5 , but without subtraction of the background and the point spread functions . to achieve a fit , we added a third cloud to the model , and reduced the rms deviation between the simulated and observed profiles as much as seemed possible by adjusting @xmath104 , @xmath102 , @xmath105 , @xmath106 , and @xmath66 , together with the background levels , and the normalization factors of the point spread functions . as expected , we could increase the intensity of a given profile peak by increasing the scattering optical depth of the corresponding cloud , but with correlated effects on the other peaks due to shifts in the relative frequencies of scattering in the other two clouds . we could increase the ratio between the intensities in the 4 - 6 and 2 - 4 kev profiles by increasing @xmath66 . we then sought values of the occurrence times and magnitudes of flares that could cause single - bin deviations to mimic those that seemed to be present , for example , in bin 18 in panel 3 and bin 15 in panel 5 . the choice of occurrence times was guided by the equation @xmath107^{1/2},\ ] ] which expresses the arrival angle of a photon emitted at time @xmath108 from the source which arrives at time @xmath57 after scattering from a grain at distance @xmath55 from the observer . we note that with three clouds the effects of one flare appear around three separate angles . thus adding a flare to the model had subtle effects that reduced the overall rms deviation of the fit , but are not all evident to the eye . table 4 lists the fixed and fitted parameters of the final model , and fig . 10 displays some of the control data in graphical form . the intensities of the resulting simulated profiles are plotted as the open circles in fig . the curves indicate the sums of the backgrounds and point spread functions that were added to the simulated profiles . plotted below the profiles are the deviations of the simulated data from the observed data divided by the one - sigma statistical errors of the observed data . the quality of the fit for each profile is indicated in the plot by the rms deviation . the rms deviation of the total fit is 0.88 . the statistical uncertainties of the simulated data were negligibly small by virtue of the large number ( 10@xmath109 ) of trial photons launched per second , which yielded @xmath110 contributions to the simulated data . the separate contributions to the simulated profiles by scatterings from the three dust clouds are plotted in fig . another test of the simulation is provided by the comparison between the spectra of the observed and simulated halos displayed in fig . 13 . both the observed and simulated spectra were compiled from data in the angle range from @xmath111 to @xmath112 and the time range from 5037 to 25037 s. although a good fit has been obtained with the listed parameters , a substantial uncertainty must be attached to each of them . a major source of that uncertainty is the use of the asm average spectrum for lack of specific information about the spectrum during the period of @xmath1132 days before the observation began . another source of uncertainty are the cross correlations among the effects of variations of the various model parameters which result in a very flat chi - square function of the overall fit . we found that a good fit can be achieved with the source distance fixed at 6 kpc , provided the distances of the two nearer dust clouds are changed to 1.42 and 3.20 kpc , well away from the distances of the atomic hydrogen concentrations derived from the 21-cm spectrum . fixing the distances of the two principal dust clouds did place a significant constraint on the choice of the adjusted parameters . nevertheless , lacking an objective procedure for evaluating the uncertainties in the fitted parameters that arise from these ambiguities , we have listed the values that yielded the fit without errors . the profiles of the grain - scattered halo of x1538 , recorded before and after an eclipse , have been fitted by the profiles of a simulated halo generated by a monte carlo ray - tracing program operating on a model of the dust grains and the source spectrum . in the model the dust is concentrated in three discrete clouds of which the two responsible for most of the scattering are fixed at the distances of two peaks in the density of atomic hydrogen derived from the 21-cm spectrum , namely 1.30 and 2.56 kpc . the source spectrum before the observation began was assumed to be an average derived from several years of observations by the all sky monitor on the it rxte x - ray observatory . the essential fitted parameters of the model were the distance of x1538 , the distance of the third cloud , the scattering optical depths of all three clouds , and the column density of hydrogen . three flares were added to the model spectrum before the observation began with occurrence times and magnitudes adjusted to mimic sharp features in the observed profiles . the model parameters of general significance are the source distance , the total scattering optical depth , the interstellar photoelectric absorption , and the upper limit on the grain size . locating the two principal dust clouds at the peaks in the density of atomic hydrogen forced a choice of the source distance in the fitting model that is substantially less than the estimates of 6.0 kpc and 6.4 kpc derived from photometry of qv nor , respectively , by @xcite and @xcite . the adopted value of 4.5 kpc is in the low end of the uncertainty range of the estimate obtained from photometry by @xcite . the ratio @xmath114 is a measure of the scattering efficiency of the grains for photons of energy @xmath42 . in this ratio , @xmath115 is the visual extinction of the optical companion , qv nor . in the fitted model , @xmath116 for @xmath117 kev . assuming @xmath118 mag for the visual extinction of qv nor , we find @xmath119 mag@xmath4 , where the stated error includes only the uncertainty range of the published values of the visual extinction . this result contradicts the much lower value derived from the _ ginga _ observation of x1538 previously cited @xcite . however , it is somewhat lower than the theoretical value of 0.0272 read from the plot computed by @xcite for solid spherical carbonaceous and silicate grains , and is consistent with the idea that interstellar dust grains are not compact . the value of @xmath32 required to match the ratio of the 2 - 4 kev halo intensities to the 4 - 6 kev intensities , like the values obtained from other observations of x1538 , is much larger than the integrated column density of atomic hydrogen derived from the 21-cm doppler spectrum in the direction of x1538 . the latter , computed according to the equation @xcite @xmath120 is plotted against radial distance in fig . if the photoelectric cross section per hydrogen atom we have used is accurate , then most of the hydrogen along the sight line to x1538 must be in molecules , presumably in the three dust clouds . from this study we conclude : \1 ) monte carlo simulation of a grain - scattered x - ray halo can provide an effective means for extracting information about the distance of the source and the properties and distribution of interstellar dust grains in circumstances involving discrete dust clouds , source variability , and multiple scattering . \4 ) the time - varying three - peaked halo profiles of x1538 before and after an eclipse can be fitted by simulated halo profiles generated by a monte carlo ray - tracing code operating on a model with the source at 4.5 kpc , three discrete dust clouds of which two are at the distances of the two greatest concentrations of atomic hydrogen derived from the 21-cm spectrum , and a third cloud closer to the source . \5 ) the total scattering optical depth of the three dust clouds in the model is 0.159 for @xmath121 kev . the ratio of this quantity to the visual extinction of the optical companion , qv nor , is @xmath3 mag@xmath4 , somewhat less than the theoretical value of 0.0272 derived for solid spherical carbonaceous and silicate grains by @xcite . i thank mark bautz , john arabadjis and the chandra support staff for help in use of the _ chandra _ data system , and ronald remillard , allan levine , and edward morgen for advice in the interpretation of the rxte data . i thank naomi mcclure - griffiths for providing the 21-cm spectrum in the direction of x1538 . llll epoch = mjd 50449.93400 & & d=@xmath122 kpc & 1 + @xmath123 lt - s & & = @xmath124 kpc & 2 + @xmath125mjd @xmath126&&=@xmath127 kpc & 4 + @xmath128 s & & @xmath129 mag & 1 + @xmath130 hz s@xmath4 & & = 7.1 mag & 2 + @xmath131 & & = 6.9 mag & 3 + @xmath132 & & & + @xmath133@xmath134 d & & & + @xmath135@xmath136 d & & & + @xmath137@xmath138 d & & & + @xmath139 & & & + @xmath140 yr@xmath4 & & & + _ tenma _ & @xmath141 & @xmath142 & @xmath143 & 0.0 & high state & 1 + _ ginga _ & @xmath144 & @xmath145 & @xmath146 & 0.0&0.28 - 0.45&2 + _ sax _ & @xmath147 & 1.12 & @xmath148 & 0.0 & 0.50 - 0.65&3 + & @xmath149 & 0.97 & @xmath150 & 0.0 & 0.65 - 0.75&3 + _ chandra _ & @xmath151 & 1.12 & @xmath152 & 1.42 & 0.921 - 0.933 & 4 + ll @xmath162 @xmath1 s@xmath4 kev@xmath4 & @xmath163 kpc + @xmath164 m & @xmath157 @xmath1 + @xmath161 & @xmath165 m + @xmath156 & @xmath166 + @xmath158 & @xmath167 + @xmath153 & @xmath168 kpc + @xmath169 @xmath1 & @xmath170 + @xmath171 @xmath1 s@xmath4 kev@xmath4 & @xmath172 + @xmath173 @xmath1 & @xmath174 ksec + @xmath175 @xmath1 s@xmath4 kev@xmath4 & @xmath176 + @xmath99 kpc & @xmath177 ksec + @xmath100 kpc & @xmath178 + & @xmath179 ksec + & @xmath180 +
properties of the grain - scattered x - ray halo of the eclipsing x - ray binary pulsar 4u 1538 - 52 are derived from a 25 ksec observation by the _ chandra _ x - ray observatory extending from just before its eclipse immersion to near mid - eclipse . profiles of the observed halo , compiled in two energy ranges , 2 to 4 kev and 4 to 6 kev , and three time intervals before and after the eclipse , exhibit a three - peak shape indicative of a concentration of the interstellar dust grains in three discrete clouds along the line of sight . the observed profiles are fitted by the profiles of a simulated halo generated by a monte carlo ray - tracing code operating on a model of three discrete clouds and a spectrum of the photons emitted by the source over a period of time extending from 270 ksec before the observation began till it ended . in the model , the spectrum before the observation began is expressed as a function of the orbital phase of the pulsar and is derived as an average over 6.3 years of data accumulated by the all sky monitor of the _ rossi x - ray timing explorer_. the distances of the two nearer dust clouds are fixed at the distances of the peaks of atomic hydrogen derived from the 21-cm spectrum in the direction of the x - ray source , namely at 1.30 and 2.56 kpc . with these constraints , a good fit is achieved with the source at a distance 4.5 kpc , the distance of the third cloud at 4.05 kpc , the total scattering optical depth of the three clouds equal to 0.159 at 3 kev , and the column density of hydrogen set to @xmath0 @xmath1 . with @xmath2 mag for the binary companion star , qv nor , the ratio of the scattering optical depth at 3 kev to the visual extinction is @xmath3 mag@xmath4 . the column density of hydrogen in the model is much greater than the column density of atomic hydrogen derived from the 21-cm spectrum , which indicates that most of the hydrogen is in molecules , probably concentrated in the three dust clouds .
many papers have been written on @xmath1-symmetric quantum - mechanical hamiltonians . in nearly all cases these hamiltonians have had rising potentials and energies that are all discrete and _ real_. an example of such a class of hamiltonians is @xcite @xmath2 in a few instances quantum - mechanical hamiltonians having real continuous spectra have been examined @xcite . for the case of @xmath1-symmetric quantum - mechanical hamiltonians having discrete spectra , the associated classical - mechanical systems have also been studied @xcite . it has been found that the classical trajectories associated with real energies are @xmath1-symmetric ( they are symmetric under reflection about the imaginary axis ) and exhibit remarkably interesting topological properties . these trajectories can visit multiple sheets of a riemann surface . however , the classical mechanics associated with quantum systems having continuous spectra have not yet been examined . thus , in this paper we explore the classical trajectories of the periodic potential analyzed in ref . @xcite . consider a classical - mechanical system consisting of a simple pendulum with a bob of mass @xmath3 and a string of length @xmath4 in a uniform gravitational field of magnitude @xmath0 ( see fig . the gravitational potential energy of the system is defined to be zero at the height of the pivot point of the string . the pendulum bob swings through an angle @xmath5 . therefore , the horizontal and vertical cartesian coordinates @xmath6 and @xmath7 are given by @xmath8 which gives velocities @xmath9 we define the potential and kinetic energies @xmath10 and @xmath11 as @xmath12 the hamiltonian @xmath13 for the pendulum is therefore @xmath14 without loss of generality we set @xmath15 and @xmath16 , and for consistency with the notation used in earlier papers on @xmath1-symmetric hamiltonians we relabel @xmath17 to get @xmath18 where @xmath19 . the classical equations of motion for this hamiltonian read @xmath20 the hamiltonian @xmath21 for this system is a constant of the motion and thus the energy @xmath22 is a time - independent quantity . for most of this paper we take @xmath22 to be real because the energy levels of the corresponding quantum - mechanical system are real . this paper is organized as follows : in sec . [ s2 ] we study the complex trajectories for @xmath21 in ( [ e6 ] ) when @xmath22 and @xmath0 are real . in sec . [ s3 ] we study the complex trajectories for @xmath21 when @xmath22 is real but @xmath0 is imaginary . in sec . [ s4 ] we examine briefly the complex chaotic motion that results when the energy @xmath22 of the pendulum is allowed to be complex and also when the pendulum is subject to a periodic external driving force . in this section we examine the solutions to hamilton s equations ( [ e7 ] ) for the case in which the gravitational field strength @xmath0 is real . without loss of generality we take @xmath23 . there are three cases to consider : ( 1 ) the complex closed trajectories that result when @xmath24 , ( 2 ) the complex open trajectories that occur when @xmath25 , and ( 3 ) the complex closed trajectories that we obtain when @xmath26 . when the energy @xmath22 of a simple pendulum lies in the range @xmath27 , the physically observed motion is that of a swinging pendulum . for the example considered in this subsection we choose @xmath28 . since a _ classical turning point _ @xmath29 is the solution to the equation @xmath30 , the classical turning points for this case satisfy @xmath31 . thus , @xmath32 the observed classical trajectory for this pendulum is represented by @xmath33 on the real-@xmath34 axis . this trajectory oscillates on the real axis between the turning points at @xmath35 . however , the full array of complex solutions to the classical equations of motion is much richer ( see fig . [ f2 ] ) . in this figure we see that all classical trajectories are closed and periodic and lie in cells of width @xmath36 that fill the entire complex-@xmath34 plane . as a consequence of cauchy s theorem , all of the closed orbits have exactly the same period ; namely , the period of a real physical pendulum that undergoes periodic motion on the real axis . this is because the period @xmath11 for a particle of energy @xmath22 for a hamiltonian of the form @xmath37 is given in terms of a closed complex contour integral : @xmath38}$ ] , where @xmath39 is the complex closed path of the particle . since the only singularity that @xmath39 encloses is the square - root branch cut joining the turning points , the value of @xmath11 is independent of @xmath39 . note that the orbits in fig . [ f2 ] that lie near the turning points are approximately elliptical and resemble the exactly elliptic complex trajectories of the simple harmonic oscillator , whose hamiltonian is @xmath40 . these orbits are displayed in fig . [ f3 ] . when the energy @xmath22 of a simple pendulum is sufficiently large ( here , @xmath25 ) , the physically observed motion is that of a rotating pendulum . the classical turning points for this case are no longer on the real axis because there are no real solutions to @xmath41 with @xmath25 . the rotating pendulum has no real turning points because its real trajectory never turns back . to find the complex turning points , we replace @xmath29 by @xmath42 and get two equations : @xmath43 for the example considered in this subsection we choose @xmath44 , and thus the turning points are situated at @xmath45 , where @xmath46 , and @xmath47 . the complex trajectories for this case are shown in fig . note that the trajectories fill the entire complex plane and that there are no closed orbits . the turning points have a strong influence on the path of the complex pendulum bob . the trajectories in the upper - half complex-@xmath34 plane , for example , go just below and veer around the turning points above the real axis . the turning points seem to attract the trajectories in much the same way as was found in ref . @xcite . note that the trajectories that terminate at turning points run vertically off to complex infinity . this feature is also a property of the hamiltonian @xmath48 , which was studied in detail in ref . the complex trajectories for this hamiltonian are shown in fig . there are three turning points , and the trajectory beginning at the turning point on the imaginary axis goes up the imaginary axis to infinity . one may ask how long it takes for a particle beginning at a turning point , say , the turning point at @xmath49 , to reach infinity . the time @xmath11 for this motion is represented by the following integral : @xmath50 a numerical evaluation of this integral gives @xmath51 . thus , the particle reaches the point at complex infinity in a finite time . there are no physically observable trajectories for this choice of the energy . however , the complex trajectories in this case can be calculated numerically ( see fig . [ f6 ] ) and we find that they are all closed orbits that lie in periodic cells of width @xmath36 . the turning points come in complex - conjugate pairs . in fig . [ f6 ] we have chosen @xmath52 and the turning points are situated at @xmath53 , where @xmath54 ( @xmath46 ) and @xmath47 . if we take the gravitational field strength @xmath0 in ( [ e6 ] ) to be imaginary , the hamiltonian becomes @xmath1 symmetric , where @xmath55 reflection consists of @xmath56 . setting @xmath57 , the hamiltonian becomes @xmath58 and the turning points are located at @xmath53 , where @xmath59 there are now two cases to consider , @xmath60 and @xmath61 . in fig . [ f7 ] we take @xmath62 and find turning points at @xmath63 ( @xmath64 ) and in fig . [ f8 ] we take @xmath65 and find turning points at @xmath66 ( @xmath67 ) . like the complex trajectories in fig . [ f4 ] , the trajectories in these two figures are not closed and trajectories beginning at turning points run vertically off to infinity . in fig . [ f7 ] one such trajectory is shown . this trajectory begins at @xmath68 . the time @xmath11 for a particle to travel from this turning point to infinity is given by @xmath69 after a simple change of variables this integral becomes @xmath70 where @xmath71 is the complete elliptic function . thus , the classical particle reaches complex infinity in finite time . there are a number of interesting ways to extend and generalize the investigations in the previous two sections . in figs . [ f7 ] and [ f8 ] the energy for the pendulum in an imaginary gravitational field was taken to be real . this is because a quantum - mechanical @xmath1-symmetric hamiltonian has the virtue of possessing real energy levels . however , one might ask what happens if one takes the energy to be complex . in this case the complex trajectories are not spatially or temporally periodic , even though the turning points _ are _ periodic and are separated horizontally by @xmath36 . the behavior of the trajectories is illustrated in fig . [ f9 ] , where we have taken the energy to be @xmath72 . another way to extend these investigations is to study what happens when there is an external periodic driving force of the form @xmath73 @xcite . we have taken @xmath74 and @xmath75 and in figs . [ f10 ] , [ f11 ] , and [ f12 ] we have examined the complex trajectories up to times @xmath76 , @xmath77 , and @xmath78 . the starting energy for these figures is taken to be @xmath79 and the value of @xmath0 is @xmath80 . the initial value of @xmath34 is taken to be @xmath81,which would lead to a complex trajectory if there were no driving term . we see that for short times the trajectory is noisy , but seems to be confined to the central cell in fig . however , after a sufficiently long time the trajectory wanders into adjacent cells , as can be seen in figs . [ f11 ] and [ f12 ] . this result suggests it might be productive to investigate complex trajectories of dynamical systems whose real trajectories exhibit chaos , especially in the @xmath1-symmetric case . one could begin with the complex trajectories associated with real homoclinic chaos , as in the present case . in addition , the rich behavior seen in the complex spatial trajectories of @xmath1-symmetric systems may be expected to be the projection onto the two - dimensional complex plane of interesting behavior in four - dimensional complex phase space . for example , the phase - space trajectories of @xmath1-symmetric complex systems whose real parts are chaotic might exhibit a complex hamiltonian analog of arnold diffusion , or some other kind of hamiltonian diffusion . we thank d. c. brody for many useful discussions . cmb thanks the theoretical physics group at imperial college , london , for its hospitality . as an ulam scholar , cmb receives financial support from the center for nonlinear studies at the los alamos national laboratory and he is supported in part by a grant from the u.s . department of energy . the work of ddh was supported by the royal society of london and by the u.s . department of energy office of science applied mathematical research . p. j. holmes and j. e. marsden , j. math . * 23 * , 669 ( 1982 ) . s. h. strogatz , _ nonlinear dynamics and chaos : with applications to physics , biology , chemistry and engineering _ ( perseus , cambridge , massachusetts , 1994 ) .
the motion of a classical pendulum in a gravitational field of strength @xmath0 is explored . the complex trajectories as well as the real ones are determined . if @xmath0 is taken to be imaginary , the hamiltonian that describes the pendulum becomes @xmath1-symmetric . the classical motion for this @xmath1-symmetric hamiltonian is examined in detail . the complex motion of this pendulum in the presence of an external periodic forcing term is also studied .
let @xmath2 be connected semi - simple lie groups and @xmath3 , @xmath4 the associated symmetric spaces . we assume that they are hermitian . an equivariant embedding is a pair @xmath5 where @xmath6 is a homomorphism , @xmath7 is a holomorphic map and @xmath8 we assume that @xmath2 have no compact factors and that @xmath9 is injective . then , as is easily checked , the kernel of @xmath10 is finite . replacing @xmath11 by its image , we will also assume @xmath10 injective and therefore identify @xmath11 with its image in @xmath12 . such maps have been classified by satake @xcite and ihara @xcite when @xmath3 is _ totally geodesic _ in @xmath4 . the purpose of this note is to show the following theorem . _ assume @xmath12 has no factors of exceptional type . then any equivariant embedding @xmath13 is totally geodesic . _ we should emphasize the rather surprising content of this result when compared with the case of compact hermitian symmetric spaces . if @xmath12 is compact , the symmetric space @xmath4 ( assumed hermitian ) is a generalized grassmanian . the natural maps of algebraic geometry between grassmanians in particular the veronese and segre embeddings are holomorphic and equivariant with respect to natural maps of the associated groups . very few are totally geodesic : in fact by duality between compact and non - compact symmetric spaces , the _ totally geodesic _ equivariant maps between compact spaces correspond to those between non - compact spaces , which are quite rare ( see @xcite ) . however , this result becomes more natural from the ` global ' point of view , i.e. , if one considers arithmetic quotients of the symmetric spaces . assume @xmath2 are semi - simple groups defined over @xmath14 , @xmath6 is defined over @xmath14 and @xmath15 is an equivariant embedding . for suitable arithmetic subgroups @xmath16 and @xmath17 , @xmath9 defines a holomorphic map @xmath18 where @xmath19 and @xmath20 . in this situation , recall that @xmath21 and @xmath22 have a remarkable family of distinguished points , the cm - points or special points @xcite . also note that @xmath21 , @xmath22 are in fact algebraic varieties over @xmath23 , and that @xmath24 is an algebraic subvariety of @xmath22 by a theorem of borel . assume @xmath24 has _ one _ cm - point . by using the action of @xmath25 on @xmath3 one easily sees that it has a _ dense _ subset of cm - points for the complex topology . a conjecture of andr @xcite and oort @xcite then implies that @xmath24 is a totally geodesic submanifold of @xmath22 ( and @xmath3 is a totally geodesic submanifold of @xmath4 ) . it is not obvious that @xmath24 should have one cm - point ; note , however , the following . the hermitian symmetric spaces are open subspaces of their compact duals generalized grassmanians . an equivariant holomorphic embedding will generally be given by a natural holomorphic map between the compact duals . given the @xmath14-structure , cm - points correspond to subspaces ( in the grassmanians ) verifying some rationality conditions . it is natural to expect these to be preserved . the embedding of the symmetric space for @xmath26 , @xmath27 , into @xmath28 where @xmath29 , @xmath30 @xcite gives a very graphic example . another strong motivation for the theorem is given by mok s rigidity results . assume for simplicity that @xmath11 is irreducible over @xmath14 and @xmath31 ( this is the _ real _ rank ) . then mok ( ch . 6 , thm 4.1 of @xcite ) see also the discussion at the beginning of ch . 9 has shown that _ any _ holomorphic map @xmath32 is totally geodesic . if @xmath33 ( we now denote the lie groups by @xmath34 , @xmath35 as we will be using rationality arguments ) is given and if @xmath10 is @xmath36-conjugate to a map defined over @xmath14 , mok s theorem implies our local assertion . more generally , assume @xmath10 given , and assume that there exists a totally real number field @xmath37 and a map @xmath38 defined over @xmath37 such that , for each real prime @xmath39 of @xmath37 ( thus @xmath40 , @xmath41 is conjugate to @xmath10 . then , again using mok s results , we deduce that @xmath10 is totally geodesic . the set of homomorphisms @xmath6 , over an algebraically closed field , and modulo @xmath12-conjugation , is _ discrete _ ( homomorphism of semi - simple groups up to conjugacy are rigid ) . thus @xmath42 is @xmath36-conjugate to a map @xmath43 defined over @xmath44 ; the @xmath12-conjugacy class of @xmath43 is an irreducible variety . if it is defined over @xmath14 , a theorem of moret bailly @xcite implies that there is a totally real number field @xmath37 , and a map @xmath45 verifying our condition . it is of course , difficult to compute the field of rationality of the class associated to @xmath10 . one may , however , pose the following : if @xmath2 be semisimple groups over @xmath14 and @xmath6 a homomorphism defined over @xmath46 , does there exist a totally real field @xmath37 and @xmath47 such that @xmath48 is @xmath36-conjugate to @xmath10 at each real prime of @xmath37 ? finally , mok has informed us that he could prove the theorem even for exceptional @xmath12 . his proof , however , is more difficult and necessitates global geometric computations . let @xmath12 be a connected semi - simple lie group , with finite center and no compact factor , associated to a hermitian symmetric space @xmath49 . fix a point @xmath50 . then @xmath51 defines a maximal compact subgroup @xmath52 and a cartan involution @xmath53 on @xmath54 . let @xmath55 be the cartan decomposition . there exists an element @xmath56 such that @xmath57 induces on @xmath58 the multiplication by @xmath59 defining the complex structure . then @xmath60 induces , by the adjoint action , the cartan involution . by construction this holomorphic structure is @xmath12-equivariant : if @xmath61 the associated data are obtained by conjugation by @xmath62 . in particular , @xmath63 is well - defined by @xmath64 since @xmath65 is @xmath66-invariant , and this family of quasi - complex structures defines the holomorphic structure on @xmath49 . now assume @xmath67 , @xmath15 verify our conditions . fix a base point @xmath68 . this defines maximal compact subgroups @xmath69 . ( we will drop indexes for the group @xmath12 ) . thus @xmath70 \mathfrak{h}&=\mathfrak{k}_{h}\oplus\mathfrak{p}_{h}\end{aligned}\ ] ] and the ( injective ) map @xmath71 has the following properties : @xmath72 & f(x)=f_{c}(x)+f_{p}(x),\\[.2pc ] & ( x\in\mathfrak{p}_{h},f_{c}(x)\in \mathfrak{k}\ , \ f_{p}(x)\in\mathfrak{p})\nonumber\\[.2pc ] & f_{p}(\iota_{h}x)=\iota_{g}\ f_{p}(x),\end{aligned}\ ] ] where @xmath73 , @xmath74 are ` multiplication by @xmath75 ' on @xmath76 , @xmath58 , given by @xmath77 , @xmath78 . conversely , if a morphism @xmath71 verifies ( 1)(3 ) , @xmath10 defines a map @xmath79 , holomorphic at @xmath80 and in fact at every point by a computation similar to that as above . note that @xmath9 is a totally geodesic immersion if and only if , @xmath81 ( see p. 47 ff . of @xcite ) identifying @xmath82 with a subalgebra of @xmath83 by ( 1 ) , we note that the two components @xmath84 and @xmath85 are @xmath82-equivariant . moreover , let @xmath86 be a decomposition of @xmath87 in simple factors . then @xmath77 or @xmath73 decomposes accordingly , so the restriction @xmath88 to @xmath89 again verifies the conditions . thus we may assume that @xmath87 is simple . in this case it is known ( see s helgason , differential geometry and symmetric spaces , ch . viii , 5 ) that the ( real ) representation of @xmath82 on @xmath76 is irreducible . the @xmath82-map @xmath90 is therefore injective or zero . assume ( changing notation ) that @xmath91 is a @xmath53-stable semi - simple subalgebra such that the injection @xmath92 is holomorphic ( for the choice of @xmath93 where @xmath94 is the obvious maximal compact subgroup of@xmath95 . it suffices then to check that @xmath96 on @xmath97 . but any hermitian lie algebra @xmath87 contains a subalgebra @xmath98 isomorphic to @xmath99 , the injection being holomorphic in the obvious sense ( in fact it contains @xmath100 where @xmath101 is the real rank ( see e.g. ch . 5 of @xcite ) . thus we are reduced to the case when @xmath102 . we can also replace @xmath12 by a larger group . by the results of satake , @xmath4 embeds into @xmath103 where @xmath104 , as a totally geodesic subvariety , via an equivariant embedding . finally we are reduced to the case when @xmath11 is locally isomorphic to @xmath105 or @xmath106 and @xmath12 to @xmath107 . ( note that this does not apply when @xmath12 has exceptionalfactors ) . in this paragraph we consider the case , to which we are reduced , when @xmath108 and @xmath109 . we try to solve the linear algebra problem of 2 find @xmath10 verifying ( 1)(3 ) . we have @xmath110 \mathfrak{g}&=\bigg\{\begin{pmatrix } a&z\\ { } ^{t}{\bar{z } } & b \end{pmatrix}\!:\hbox{tr}(a)+\hbox{tr}(b)=0\bigg\},\end{aligned}\ ] ] where the block matrices are of size @xmath111 , @xmath112 is ( complex ) arbitrary and @xmath113 are skew - hermitian . let @xmath114 , @xmath115 , @xmath116 , a basis of @xmath87 ( the empty entries are zero ) . let @xmath117 a basis of @xmath118 . we take @xmath119 given by block - diagonal matrices , so @xmath120 where @xmath121 . similarly , @xmath122 if @xmath71 verifies ( 3 ) we have @xmath123 f(v)&=\begin{pmatrix } c & iz \\-i z^{\ast}&d \end{pmatrix},\end{aligned}\ ] ] @xmath124 verifying of course ( 5 ) . let @xmath125 be the images of @xmath126 . using ( 5 ) , ( 6 ) and ( 7 ) we have @xmath127 y&=\begin{pmatrix } -e^{\ast}\\ z^{\ast } & -f^{\ast } \end{pmatrix},\\[.4pc ] h=[x , y]&=\begin{pmatrix } -[e , e^{\ast}]+zz^{\ast } & -zf^{\ast}+e^{\ast}z\\[.2pc ] fz^{\ast}-z^{\ast}e & -[f , f^{\ast}]-z^{\ast}z \end{pmatrix},\end{aligned}\ ] ] where @xmath128 and @xmath112 are arbitrary @xmath111-matrices ( with @xmath129 ) . since @xmath130 , @xmath11 is block - diagonal by ( 1 ) ; conjugating @xmath131 under @xmath132 we can assume that the block - diagonal entries of @xmath11 are diagonal matrices @xmath133 , @xmath134 . the eigenvalues of @xmath11 are integral , and constitute the eigenvalues of a representation of @xmath135 . let @xmath136 be the space of the natural representation of @xmath12 , and @xmath137 its decomposition into a positive and a negative subspace . then @xmath138 & f\!:v_{-}\rightarrow v_{-},\\[.1pc ] & z\!:v_{-}\rightarrow v_{+}.\\[-1.5pc]\end{aligned}\ ] ] let @xmath139 be the distinct eigenvalues of @xmath11 in @xmath140 and @xmath141 the eigenvalues in @xmath142 . we can write @xmath143 , the eigenvalues being even or odd in each summand ; this decomposition is preserved by @xmath49 and @xmath144 . the decomposition is orthogonal and compatible with @xmath137 . if @xmath39 belongs to the @xmath145-eigenspace of @xmath140 ( resp . @xmath146 ) , @xmath147 ( resp . @xmath148 ) belongs to the @xmath149-eigenspace of @xmath140 ( resp . @xmath150 ) consider first the odd part of @xmath151 . we can write in @xmath152 : @xmath153 & 0 & e_{2}\\[.1pc ] & & 0 & \ddots\\[.1pc ] & & & \ddots&e_{t}\\[.1pc ] & & & & 0 \end{pmatrix } , \quad e^{\ast}=\begin{pmatrix } 0 & \\[.1pc ] e_{1}^{\ast}&\ddots\\[.1pc ] & \ddots & \ddots\\[.1pc ] & & e_{t}^{\ast } & 0 \end{pmatrix}.\end{aligned}\ ] ] according to ( 10 ) , @xmath156+zz^{\ast}&=\mathrm{diag}(-e_{1}e_{1}^{\ast},e_{1}^{\ast}e_{1}-e_{2}e_{2}^{\ast},\dots , e_{t}^{\ast}e_{t})+zz^{\ast}\nonumber\\[.2pc ] & = \mathrm{diag}(\lambda_{1},\lambda_{2},\dots,\lambda_{t+1}),\end{aligned}\ ] ] where the eigenvalues are now those in @xmath152 , the last ` diagonal ' matrix including of course the multiplicities . considering the restriction of the corresponding hermitian forms to the last summand we see that @xmath157 since the representation is odd , @xmath158 . similarly in @xmath159 : @xmath160 & 0 & f_{2}\\[.1pc ] & & 0\\[.1pc ] & & & \ddots & f_{s}\\[.1pc ] & & & & 0 \end{pmatrix } , \quad f^{\ast}=\begin{pmatrix } 0 \\[.1pc ] f_{1}^{\ast } & 0\\[.1pc ] & & \ddots\\[.1pc ] & & f_{s}^{\ast}&0 \end{pmatrix},\nonumber\\[.4pc ] -[f , f^{\ast}]-z^{\ast}z & = \mathrm{diag}(-f_{1}f_{1}^{\ast},\dots , f_{s}^{\ast}f_{s}-z^{\ast}z\nonumber\\[.1pc ] & = \mathrm{diag}(\mu_{1},\mu_{2},\dots,\mu_{s+1})\end{aligned}\ ] ] whence @xmath161 . finally the only non - vanishing part of @xmath112 is a map @xmath162 ( where @xmath163 , @xmath164 denote the eigenspaces of @xmath11 ) . thus @xmath165 z^{\ast}z&=\mathrm{diag}(z_{1}^{\ast}z_{1},0,\dots,0).\end{aligned}\ ] ] this implies of course that the only eigenvalue @xmath171 is @xmath172 , so @xmath173 and @xmath174 . since @xmath128 vanish the embedding is totally geodesic ; the representation of @xmath169 or @xmath106 is a multiple of the standard representation , in conformity with satake s results . by ( 11 ) and ( 19 ) , @xmath178+zz^{\ast}\nonumber\\[.3pc ] & \hskip -4pc \quad=(-e_{1}e_{1}^{\ast},e_{1}^{\ast}e_{1}-e_{2}e_{2}^{\ast},\dots , e_{t-1}^{\ast}e_{t-1}-e_{t}e_{t}^{\ast } + z_{1}z_{1}^{\ast } , e_{t}^{\ast}e_{t}+z_{2}z_{2}^{\ast})\nonumber\\[.3pc ] & \hskip -4pc \quad=(\lambda_{1},\dots,\lambda_{t},\lambda_{t+1}),\end{aligned}\ ] ] where we assume so far that both @xmath179 and @xmath180 are eigenvalues in @xmath181 . this implies first that there are only two eigenvalues since @xmath182 for @xmath183 . furthermore , the last entry in ( 21 ) yields @xmath184 , whence @xmath185 and @xmath186 . if @xmath179 does not occur in @xmath181 , the representation on @xmath187 is trivial ; if @xmath180 does not occur @xmath188 is absent . in this case , @xmath189 z^{\ast}z & = \mathrm{diag}(z_{1}^{\ast}z_{1},0,\dots 0)\end{aligned}\ ] ] and @xmath156+zz^{\ast}&=(-e_{1}e_{1}^{\ast},e_{1}^{\ast}e_{1}-e_{2}e_{2}^{\ast},\dots , e_{t}^{\ast}e_{t}+z_{1}z_{1}^{\ast})\\[.2pc ] & = ( \lambda_{1},\dots,\lambda_{t+1})\end{aligned}\ ] ] with @xmath190 . this equality @xmath191 implies that there is only one eigenvalue @xmath192 and therefore @xmath193 . the computations being symmetric we can assume for instance that it is missing in @xmath181 ; we already know that the eigenvalue @xmath179 only occurs , so the eigenvalues in @xmath187 are @xmath198 ; moreover @xmath199 by the arguments given already , so the embedding should be totally geodesic . we know that this is impossible , by satake s results . in fact , using ( 22 ) and ( 12 ) we see that @xmath200 which is impossible .
we prove that equivariant , holomorphic embeddings of hermitian symmetric spaces are totally geodesic ( when the image is not of exceptional type ) . @xmath0 ) @xmath1 ]
as the 99.9877% of the charged pions decay into muons and muonic neutrinos , we will refer specifically to this process . our discussion will be based on the following low - energy lagrangian : @xmath1 , where @xmath2 corresponds to the lepton free lagrangian , @xmath3 being the associated lepton chemical potential . the second term in the low - energy lagrangian corresponds to the @xmath4 chiral lagrangian @xcite @xmath5 the @xmath6 and @xmath7 terms are the tree level pion decay constant and the tree level chiral condensate , respectively . the @xmath8 fields contain the pion fields as @xmath9 and the covariant derivative @xmath10 includes external right and left currents . when @xmath11 we break explicitly the chiral symmetry . in this article we are not interested in computing pion radiative corrections . therefore , higher order chiral lagrangian terms will not be considered . however , those corrections have been already calculated at finite temperature and isospin chemical potential @xcite , and therefore , if we want to incorporate these contributions , it is enough to replace the masses , decay constant and chemical potential terms by temperature and isospin chemical potential dependent dressed terms @xcite . we will use the effective fermi model for the leptonic weak coupling . the leptonic weak currents and the isospin chemical potential are introduced by setting the external currents in the chiral lagrangian as @xmath12 \nonumber\\ & & + \frac{1}{2}\mu_i\tau^3 \delta_{\alpha 0}\end{aligned}\ ] ] with @xmath13 is the isospin chemical potential and where we use the following combination of pauli matrices : @xmath14 . we will concentrate only on the normal phase where @xmath15 . in the superfluid phase , @xmath16 , one of the charged pions condenses , and therefore , another treatment is needed @xcite . two different regions in the literature have been considered when the baryon chemical potential @xmath17 is introduced in the frame of chiral perturbation theory : _ small _ @xmath18 . the baryon chemical potential can be taken as @xmath19 in the power counting in chiral perturbation theory . in this case @xmath18 appears in the wess - zumino - witten anomalous term , which turns out to be relevant only in higher order radiative corrections @xcite . _ very high _ @xmath18 . in this case an expansion in powers of @xmath20 is performed ( asymptotically infinite chemical potential ) . here , effects of color superconductivity and color flavor locked phase are present , due to the appearance of diquark pairing @xcite . we can construct an effective lagrangian including @xmath18 in the absence of diquark effects , by considering effective @xmath21 and @xmath22 constants in the chiral lagrangian in eq . ( [ lchi ] ) and introducing a lorentz symmetry breaking term : @xmath23 . results obtained in the frame of the nambu jona - lasinio model show that for @xmath24 gev the @xmath6 , @xmath7 and @xmath25 parameters do not suffer significant changes @xcite . we can neglect then baryon chemical potential effects in this work . once the chiral lagrangian has been expanded in terms of the pion fields , the canonical quantization procedure is the standard one , keeping in mind that the energy of the charged pions and lepton fields are shifted due to the chemical potentials . then , we proceed to calculate the decay width of the charged pions and the corresponding neutrino emissivity . the decay width for charged pions , including finite temperature and density effects , is given by @xmath26,\end{aligned}\ ] ] where the on - shell pions are in the rest frame , @xmath27 , and where @xmath28 is fermi - dirac distribution . the phase space measure is defined as @xmath29 where @xmath30 stands for the different particles involved : pions , fermions and antifermions . the transition probability matrix was abbreviated as @xmath31 the corresponding chemical potentials are @xmath32 , and @xmath33 where @xmath34 denotes a fermion ( antifermion ) . this definition of the decay width corresponds to the imaginary part of the thermal one - loop weak interaction corrections to the pion propagator . the decay width includes the decay of pions into leptons as well as recombination of leptons and neutrinos . by considering massless neutrinos , the decay width for charged pions is then @xmath35 \theta(m_\pi\mp\delta\mu - m_\mu ) \nonumber\\&&\qquad\times \left[\frac{1-m_\mu^2/(m_{\pi}\mp\delta\mu)^2}{1- m_\mu^2/m_{\pi}^2}\right]^2 , \label{width}\end{aligned}\ ] ] with @xmath36 ^ 2,\\ \delta\mu & \equiv & \mu_i+\mu_\mu-\mu_{\nu_\mu},\\ e_{\mu^\pm } & \equiv & \frac{(m_\pi\mp\delta\mu)^2+m_\mu^2}{2(m_\pi\mp\delta\mu ) } \pm\mu_\mu,\\ e_{\nu_\mu^\pm } & \equiv & \frac{(m_\pi\mp\delta\mu)^2-m_\mu^2}{2(m_\pi\mp\delta\mu ) } \pm\mu_{\nu_\mu},\end{aligned}\ ] ] @xmath37 being the vacuum pion decay width . for the non chemical equilibrium case , when @xmath38 , one of the charged pions remains stable since its decay width vanishes , as can be seen from the heaviside function in eq . ( [ width ] ) . we will consider then , in the non - equilibrium case , that @xmath39 mev . here we used @xmath40 mev and @xmath41 mev . stable states at zero temperature , i.e. those where their decay width vanish , will now develop a small decay width due to thermal fluctuations . from eq . ( [ width ] ) , considering that @xmath42 when @xmath43 , we find two metastable cases : @xmath44 in terms of quark degrees of freedom , the beta equilibrium condition is @xmath45 , or , in terms of the isospin chemical potential , @xmath46 . if we consider a degenerate gas , which is the case for leptons in compact stars , the condition @xmath47 arises in order to equilibrate the fermi levels @xcite . as a consequence of this leptonic chemical equilibrium , the beta - equilibrium condition will produce @xmath48 . from eqs . ( [ pimmeta ] ) and ( [ pipmeta ] ) , we can see that metastable states in beta equilibrium will occur for @xmath49 for @xmath50 mesons and @xmath51 for @xmath52 mesons , where @xmath53 note that in beta equilibrium , the @xmath50 meson condenses if the lepton chemical potential is high enough such that @xmath54 . on the other side , if the neutrino chemical potential is high enough , such that @xmath55 , then a @xmath52 meson condenses @xcite . in order to explore the phenomenological consequences of these pion metastable states , we will discuss next the neutrino emission and cooling rate of a pion - lepton gas in leptonic- and beta- equilibrium . our discussion will show in a clear way the influence of metastable states on the cooling rate . neutrino emission is perhaps the most relevant phenomena associated to the temperature evolution of compact stars . the neutrino emissivity @xmath56 is defined as the energy loss through neutrino emission per unit time and unit volume . for the decaying pions , the neutrino emissivity must include the probability of finding a pion in the media as well as the pauli blocking for the emerging muons @xmath57 is the bose - einstein distribution and the other terms were defined in the previous section . hereafter , we will consider the degenerate case where @xmath58 and also beta - equilibrium . as we discussed in the last section , these two assumptions imply @xmath48 . the neutrino emissivity , then , is given by @xmath59 , \label{emissivity}\end{aligned}\ ] ] where the limits for the muon energy integral are @xmath60.\ ] ] the temperature and the muon chemical potential tend to favor the anti - neutrino emission due to the @xmath50 decay . the @xmath61 factor in eq . ( [ emissivity ] ) , which gives the probability of finding a @xmath50 meson , grows as function of temperature and lepton chemical potential . on the other hand , the @xmath62 factor gives the probability of finding an accessible state for the emerging muon . only thermal fluctuations will conspire against the pauli blocking , allowing then the decay of @xmath50 . the probability of finding a @xmath52 meson becomes smaller for higher chemical potential , being suppressed by the bose factor . we are interested to extract the leading terms for the low temperature behavior of the emissivity . since we are considering a degenerate gas ( @xmath63 ) , the main contribution is given by the emission of antineutrinos , being the neutrino emission highly suppressed by an exponential factor as we mentioned previously . in order to extract the main contribution to the total emissivity , we expand eq . ( [ emissivity ] ) in the low temperature region at the leading order . the fermi - dirac distribution becomes then @xmath64 , obtaining @xmath65 , \label{e_expanded}\ ] ] with @xmath66 . \label{gpm}\ ] ] we can separate the antineutrino emissivity in three different regions , depending on the value of the lepton chemical potential . as a function of @xmath67 , indicating three different regions for @xmath68 ( horizontal lines ) which determine the integration regions of the @xmath67 variables for the emissivity at low temperature , ruled by the condition @xmath69 in eq . ( [ e_expanded ] ) . ] [ elp - elm ] shows the functions @xmath70 plotted as a function of @xmath67 . on the vertical axes , the three regions are indicated for specific values of the lepton chemical potential . the integration limits in equation ( [ e_expanded ] ) are determined by the condition @xmath69 . in region * i * , where @xmath71 , the argument of the theta functions is always positive , then @xmath72 in region * ii * , where @xmath73 , the function @xmath74 is positive for all values of @xmath67 . however , from eq . ( [ e_expanded ] ) , the condition @xmath75 will exclude a region in the integral : @xmath76 where @xmath77 \label{epipm } \label{mpm}\ ] ] are the solutions of the equation @xmath78 , giving as a result @xmath79 . this is an intermediate region between low and high lepton chemical potential . in region * iii * , where @xmath80 , the condition @xmath69 will exclude some values in the integration limits : @xmath81 where @xmath82 was previously defined in eq . ( [ mpm ] ) and corresponds to the solutions of the equation @xmath78 , giving @xmath83 . as we can see , the integrals above can be written in the form @xmath84 where the integrand @xmath85 in eq . ( [ gpm ] ) , and where @xmath86 stands for @xmath87 . in order to extract the leading terms in the low temperature region , if @xmath88 , we can expand the bose - einstein distribution , and through an appropriate change of variables , we find @xmath89 with @xmath90 and with @xmath91 such that the last integral remains finite in the limit @xmath43 . due to the exponential factor , the integrand in the above equation will be dominated by low @xmath92-values . if the condition @xmath93 is satisfied , we can keep only the first term @xmath94 in the series . if @xmath95 , we need to sum the whole series . as a result , the low temperature behavior of the emissivity becomes @xmath96 where @xmath97 the intermediate region @xmath98 will be a combination of terms @xmath99 and @xmath100 . when the chemical potential grows , the linear term in @xmath101 starts to dominate . the neutrino emissivity , at low temperature , is strongly suppressed by a term @xmath102 $ ] . we are interested in the high chemical potential region since it increases the emissivity in the low temperature region . our numerical analysis suggests that these approximations will be valid for temperatures less than @xmath103 mev . in order to explore some effects in the metastable region of the @xmath50 , eq . ( [ pimmeta ] ) , we will now calculate the cooling rate due to muonic neutrino emission for a pion - lepton gas , at constant volume and charge . we will consider that @xmath104 , which means that all the neutrinos will escape from the gas . we also consider @xmath105-equilibrium , lepton - equilibrium and neglect the process of neutrino emission through muon decay . in other words this model is a simplification , eventually valid as isolated bubbles inside the nuclear media of compact stars , although finite volume effects should be taken into account . the cooling time @xmath107 is defined as @xmath108 where @xmath109 and @xmath110 are the initial time and temperature , respectively . @xmath111 and @xmath56 correspond to the specific heat and the emissivity , respectively . the specific heat per unit of volume is given by @xmath112 where @xmath113 is the entropy , and @xmath114 , in our case , is the charge number . in a non - interacting degenerated gas , at low temperature and high chemical potential , the charge number is dominated by fermions , and the leading term in the temperature expansion is constant : @xmath115.\ ] ] so , for such low temperature approximation , the constant @xmath114 condition is equivalent to consider a constant @xmath116 . the specific heat per unit of volume can be written as @xmath117 where @xmath118 is the pressure . the contribution to the specific heat per unit of volume for noninteracting fermions is @xmath119 with @xmath120 . the formula for the specific heat per unit of volume , for bosons , is the same as for fermions but with @xmath121 and changing @xmath122 by @xmath61 . the relevant contribution comes essentially from electrons and muons , where the specific heat per unit of volume for a degenerated fermion gas of mass @xmath86 and chemical potential @xmath116 is @xmath123 if @xmath124 . in the case of muons , when @xmath125 , their contribution will be @xmath126 . -equilibrium with zero neutrino chemical potential . the initial temperature is 5 mev . ] [ cooling ] shows the temperature dependence , as function of the logarithm of time in seconds , for three different values of the electron chemical potential : @xmath127 ( muon degeneracy ) , @xmath128 ( pion condensation ) and another value in between : @xmath129 mev . it can be seen from fig . [ cooling ] that starting from an initial temperature , @xmath130 mev , the needed time to reach 1 mev is extremely short , fraction of seconds , in the metastable as well in the condensed case . on the other hand , it will take thousands of years to diminish the temperature from 1 mev to fractions of mev . the cooling process is not so fast if we consider values of the electron chemical potential lower than 109mev . notice that , in spite of the fact that @xmath131 mev is an average tween @xmath132 and @xmath133 , the corresponding cooling time curve is notoriously closer to the beginning of the pion superfluid phase . in this paper we have calculated the charged pions decay widths in dense matter at finite temperature , analyzing the pion metastable condition . we have calculated also the neutrino emissivity through the leptonic pion decay in @xmath105-equilibrium for a degenerated system . we obtained the main contributions in the low temperature region for pions decaying into muons and muonic neutrinos . finally , we estimated the cooling rate of a pion lepton gas in @xmath105-equilibrium for three different values of the electric chemical potential . from our results , we argue that it is possible to find stable pions , even if they are not condensed , if the lepton chemical potential reaches a value higher than the muon mass , which is the case in compact stars . under such conditions , the pions might only decay through thermal fluctuations . the muonic neutrino emissivity will grow with the lepton chemical potential , varying from @xmath99 to @xmath100 , both with an exponential suppressing term . the contribution to the cooling process in a neutron star , due to the emission of muonic neutrinos from pions , has not been much considered yet in the normal phase . in fact , from our estimation of the cooling time of the lepton pion gas , it can be seen that this process is relevant for temperatures higher than 1 mev which corresponds to the cooling process of a protoneutron star . the pion superfluid case will be explored elsewhere . the authors acknowledge support from fondecyt under grant 1095217 . m.l . acknowledges also support from _ proyecto anillos _ act119 ( utfsm ) . we thank andreas reisenegger for helpful and valuable discussions . d. t. son , m. a. stephanov , phys . lett . * 86 * , 592 - 595 ( 2001 ) . m. loewe , c. villavicencio , phys . rev . * d70 * , 074005 ( 2004 ) . m. loewe , c. villavicencio , phys . rev . * d71 * , 094001 ( 2005 ) . l. he and p. zhuang , phys . b * 615 * , 93 ( 2005 ) d. ebert , k. g. klimenko , j. phys . g * g32 * , 599 - 608 ( 2006 ) . d. ebert , k. g. klimenko , eur . j. * c46 * , 771 - 776 ( 2006 ) . z. zhang and y. x. liu , phys . c * 75 * , 035201 ( 2007 ) d. t. son , m. a. stephanov , phys . rev . * d61 * , 074012 ( 2000 ) . m. loewe , c. villavicencio , phys . rev . * d67 * , 074034 ( 2003 ) . y. jiang , y. m. shi , h. li , w. m. sun and h. s. zong , phys . rev . d * 78 * , 116005 ( 2008 ) j. a. pons , s. reddy , m. prakash , j. m. lattimer , j. a. miralles , astrophys . j. * 513 * , 780 ( 1999 ) . s. flores - tulian and a. reisenegger , mon . not . * 372 * , 276 ( 2006 ) a. w. steiner and s. reddy , phys . c * 79 * , 015802 ( 2009 ) l. b. leinson , phys . c * 79 * , 045502 ( 2009 ) c. m. pi , x. p. zheng and s. h. yang , phys . rev . c * 81 * , 045802 ( 2010 ) d. blaschke , h. grigorian and d. n. voskresensky , astron . astrophys . * 424 * , 979 ( 2004 ) s. kubis , phys . c * 73 * , 015805 ( 2006 ) w. b. ding , g. z. liu , m. f. zhu , z. yu , and e. g. zhao , a&a 506 , l13 ( 2009 ) .
pions may remain stable under certain conditions in a dense media at zero temperature in the normal phase ( non pion superfluid state ) . the stability condition is achieved when the in - media pion width vanishes . however , thermal fluctuations will change this stable regime . for low temperature pions will remain in a metastable state . here we discuss the different possible scenarios for leptonic pion decays at finite temperature , taking into account all the different chemical potentials involved . the neutrino emission due to pions in a hot - dense media is calculated , as well as the coolig rate of a pion - lepton gas . the study of pions in a superfluid state has called the attention of physicist during many years in different frameworks @xcite . in particular , compact stars may provide a natural scenario for such a state of matter . however , in the absence of a stable superfluid state , pions in the normal phase could behave as metastable particles . in fact , for high densities , an appropriate combination of chemical potentials can avoid the normal pion decay process . indeed , if in cold matter the leptons states are completely filled up to the fermi level , there will be no possible allowed final states for the decay of charged pions , unless thermal fluctuations are present . pion decay properties in a hot and/or dense media have been vastly studied @xcite . the present work is focused on the existence of possible pion metastable states in the normal phase at high densities . as a consequence , we will study how these metastable states affect the neutrino emission rate . the cooling rate for a pion - lepton gas is also determined . the temperature considered is much less than the pion mass . we will consider baryon chemical potential values lower than @xmath0 gev , in order to avoid other possible media effects like color superconductivity . the possible scenario for such a process are protoneutron stars @xcite .
the identification of active species for desired chemical reactions is a main quest in the field of heterogeneous catalysis . @xcite whether a species is active or not depends on many factors , including the nature of its interaction with the catalyst material , as well as the temperature and pressure at which the catalytic reactions take place . the main approach for gaining quantitative , microscopic understanding of various properties and processes , has been to perform experiments under standard surface science conditions , i.e. , under ultra high vacuum ( uhv ) and relatively low ( room or lower ) temperatures , with the hope that what is learnt may be relevant and/or extrapolated to the system under the `` real '' catalytic conditions of high temperatures and pressures . theoretical studies of adsorption and reactions at surfaces have , at best , been carried out by first - principles approaches , which , while highly accurate , typically do not take into account the temperature or the gas - phase environment with which the surface is in contact . a recent focus , both experimentally and theoretically , is to attempt to study systems which more closely resemble those of true catalytic ones . @xcite experimentally , this is done by performing _ in situ _ real - time measurements carried out under high temperature and pressure conditions , and by the systematic study of , e.g. , well - characterized metal clusters deposited on support materials and chemical reactions thereon ( see e.g. , refs . ) . such studies are important in particular for understanding those systems which exhibit a `` pressure- and/or materials - gap '' , i.e. , where different behavior is exhibited under low and high pressures , or when results for model single - metal catalysts do not represent those of the `` real '' catalyst , typically consisting of a dispersed metal on a support material . often the pressure- and materials - gaps are related and occur simultaneously . the oxygen silver system is an example which exhibits a pronounced pressure - dependence : despite its unique activity as _ the _ catalyst for ethylene epoxidation ( c@xmath3h@xmath4o@xmath5 c@xmath3h@xmath6o , performed at 500 - 600 k and atmospheric pressure ) and partial oxidation of methanol to formaldehyde ( ch@xmath7oh+@xmath8o@xmath9 ch@xmath3o+ h@xmath3o - `` oxydehydrogenation '' , or by direct dehydrogenation , ch@xmath7oh @xmath10 ch@xmath3o+h@xmath3 , performed at 700 - 900 k and atmospheric pressure ) , these reactions do not occur with any significant probability under uhv conditions . @xcite this behavior of the oxygen silver system might be related to a materials - gap , because for studies of the partial oxidation of methanol , drastic differences are observed in the performance for fresh silver catalysts and those which have been exposed to high temperature ( @xmath11 k ) and pressure ( atmospheric ) conditions . the temperature dependence of the conversion of the clean ag material to the active catalyst exhibits a pronounced hysteresis , i.e. , the conversion is different when raising the temperature from low to high , compared to from high to low . this was the case for high temperatures up to 873 k , as well as for lower temperatures up to 560 k and 750 k. these hysteresis effects were shown from scanning electron microscopy ( sem ) studies to be related to significant morphological changes which occur in the catalyst depending on the temperature , and that do not revert to the original structure when lowering the temperature , i.e. , the changes are irreversible . @xcite structural changes in silver also occur when in contact with a _ pure _ o@xmath3 environment at high temperatures ( @xmath12900 k ) and atmospheric pressure , namely , facets with the ag(111 ) orientation result . @xcite this is presumably because under these conditions it is unlikely that there will be any oxygen on the surface and the ( 111 ) surface of ag has the lowest surface energy . it has been suggested that the elusive sub - surface or `` bulk - dissolved '' oxygen species play an essential role in the above - mentioned catalytic reactions , which so far have escaped a precise characterization . @xcite in addition to this aspect , there are other controversial debates in the literature concerning the nature of the active species , e.g. , atomic versus molecular ( ozone - like ) oxygen . @xcite another puzzling issue is how silver , as a noble metal , can function as a good catalyst when it only binds adparticles relatively weakly on the surface compared to the transition - metal catalysts to the left of it in the periodic table . catalysis clearly involves a number of complex processes , namely the dissociation of molecules and the creation of chemically active species , and subsequent interaction and reaction between the particles to form the product , which desorbs from the surface . furthermore , it is possible ( maybe likely ) that some fragments of the reactant molecules ( e.g. , h or c ) modify the catalyst material and thereby participate in the creation of the active species . in the present paper we tackle only part of the problem , namely , the study of the interaction of silver with oxygen in order to identify and exclude possible active o species for the above - mentioned heterogeneous catalytic reactions . this is an essential prerequisite for modeling the full catalytic process . although we study explicitly the ( 111 ) surface of silver , our general findings are expected to be relevant for silver _ per se _ , and possibly to have implications for gold , also a noble metal oxidation catalyst @xcite which , for the ( 111 ) surface , exhibits a restructuring and chemisorption of oxygen atoms at elevated temperatures ( 500800 k ) and atmospheric pressure , @xcite similar to ag(111 ) . they may also relate to copper , which catalyses the oxidation of methanol to formaldehyde . @xcite in our previous publications , @xcite the interaction between oxygen and ag(111 ) was thoroughly studied by density - functional theory ( dft ) . in these works we considered almost all possibly relevant oxygen species , namely , pure on - surface chemisorbed oxygen , pure sub - surface and bulk - dissolved oxygen , surface- and bulk - substitutional adsorption , structures involving both oxygen on and under the top ag layer , as well as surface - oxides and a molecular ( ozone - like ) species adsorbed at a surface vacancy . we studied these oxygen species for a wide range of coverages where the following results were obtained : at low coverages oxygen atoms adsorb on the surface . with increasing coverages , however , strong repulsive interactions develop between the partially negatively charged o atoms and the adatoms are driven to penetrate into the sub - surface region , and/or to give rise to a reconstructed , thin surface - oxide layer , when the coverage is larger than @xmath120.25 ml . for higher oxygen contents , thicker oxide - like films are predicted to form . we also found that the presence of sub - surface oxygen when bonded to the same surface ag atoms as an oxygen atom on the surface , dramatically modifies the electronic properties of the on - surface oxygen and vice - versa . depending on the particular geometry , it can give rise to a notable stabilization or destabilization , but the energetically favorable structures induce a _ stabilization _ and a lowering of the density of states at the fermi level . in contrast to the conclusions concerning the crucial role of bulk - dissolved oxygen based on interpretations of experimental results , we found that this species is energetically _ unfavorable _ , as is bulk- and surface - substitutional oxygen , as well as the molecularly adsorbed ozone - like species at a vacancy , when the system is in thermal equilibrium . @xcite on the other hand , the binding energy of atomic oxygen adsorbed at under - coordinated silver atoms is considerably stronger than on the terraces , but the concentration of such species obviously depends on the number of defects and the morphology of the substrate . nevertheless , temperature and pressure can affect the stability of structures and this has not been investigated so far . clearly this is important for gaining an understanding of the system under catalytic conditions and this is the main topic of the present paper . some of these results were reported briefly in ref . . our study uses the approach of _ ab initio _ , atomistic thermodynamics " , that is , performing systematic dft calculations for all o species that could conceivably be relevant , and take into account the effect of the environment , which we do via the pressure and temperature dependence of the oxygen chemical potential ( see , e.g. , refs . ) . from the resulting free energies ( obtained using our previously obtained energetics @xcite ) , we derive the phase diagram of oxygen at the silver ( 111 ) surface . since our earlier studies @xcite predict that oxide - like structures form for high o concentrations , we also investigate the surface and bulk properties of the most stable bulk oxide , namely , di - silver oxide , ag@xmath3o . the paper is organized as follows : the calculation procedure , as well as the thermodynamical method are explained in sec . ii , and in sec . iii results are presented for the bulk properties of ag@xmath13o . in sec . iv , we investigate various terminations of the ag@xmath13o(111 ) surface as a function of the oxygen chemical potential . section v describes investigations into the transition from oxide - like structures ( where the positions of the ag atoms are commensurate with those of the ag(111 ) substrate ) to the true oxide film which is laterally expanded in comparison . in sec . vi , the pressure temperature phase diagram of the oxygen silver system is presented and compared to experimental results . finally , the conclusions are given in sec . the dft total - energy calculations are performed using the pseudopotential plane wave method @xcite within the generalized gradient approximation ( gga ) . @xcite the pseudopotentials are generated by the scheme of troullier and martins with the same functional . @xcite the wave functions are expanded in plane waves with an energy cutoff of 50 ry . in the @xmath14 surface unit cell , corresponding to the periodicity of clean ag(111 ) , 21 special * k*-points are used in the surface irreducible brillouin zone ( ibz ) for the brillouin - zone integration . @xcite equivalent * k*-points to these are used for all of the surface structures studied for consistency , i.e. , to maximize the accuracy when comparing the energetics of different coverages as calculated in different supercells . we employ a fermi function with a temperature broadening parameter of @xmath15 ev to improve the convergence , and the total energy is extrapolated to zero temperature . ( @xmath16 is the boltzmann constant . ) the structures are created on one side of the slab and the resulting induced dipole moment is taken into account by applying a dipole correction . @xcite for the adsorption structures on ag(111 ) , we use five layers of ag to model the surface and typically keep the bottom two or three layers fixed at the bulk - like positions and fully relax all the other atoms until the forces on the atoms are less than 0.015 ev / . the oxide surfaces of ag@xmath13o(111 ) are simulated by symmetrical slabs ( i.e. , with inversion symmetry ) . we tested thicknesses corresponding to five , seven , and nine metal layers , separated by 15 of vacuum . our results show that the systems can be safely described using five metal layers with inversion symmetry . in _ ab initio _ theory , the consideration of high temperature and high pressure can be achieved by explicitly taking into account the surrounding gas phase in terms of `` _ ab initio _ , atomistic thermodynamics '' ( cf . e.g. refs . ) . this is also an appropriate ( first ) approach to steady - state catalysis , which is often run close to thermodynamic equilibrium ( or a constrained equilibrium ) to prevent catalyst degradation . in the following we outline how the combination of thermodynamics and dft can be applied to obtain the lowest - energy surface structures with a surrounding gas phase , thus enabling us to construct a @xmath17-diagram of the stability ( or metastability ) regions of different surface phases . we consider a surface in contact with an oxygen atmosphere which is described by an oxygen pressure @xmath18 and a temperature @xmath19 . the environment then acts as a reservoir , as it can give ( or take ) oxygen to ( or from ) the substrate without changing the temperature or pressure . we calculate the surface free energy , @xmath20 where @xmath21 and @xmath22 are the number of oxygen and silver atoms , and @xmath23 is surface area . the @xmath19 and @xmath18 dependence is mainly given by @xmath24 , the oxygen chemical potential , i.e. , by the o@xmath3 gas phase atmosphere : @xmath25 , \label{eq : ochem}\end{aligned}\ ] ] where @xmath26 corresponds to atmospheric pressure and @xmath27 includes the contribution from rotations and vibrations of the molecule , as well as the ideal - gas entropy at 1 atmosphere . it can be calculated or taken from experimental values from thermodynamic tables the difference between theory and experimental values is marginal in the temperature range of interest . the silver chemical potential is @xmath28 we use small symbols ( @xmath29 , @xmath30 ) in eq . 3 to indicate that these are free energies per ag atom ( top ) and per ag@xmath3o unit ( bottom ) . these two quantities are calculated by dft - gga . in eq . 1 , we have furthermore , @xmath31 these quantities are also calculated by dft - gga . for the temperature dependence of @xmath32 , we use thermodynamic tables [ ref . ] ( i.e. , experimental data ) , while the @xmath1 k value of eq . [ eq : ochem ] is @xmath33 @xmath34 in the above equation we use the experimental value of the heat of formation , @xmath35 , because of its very small value ( 0.323 ev at the standard state ( room temperature and atmospheric pressure ) , @xcite or 0.325 ev extrapolated back to @xmath36 k ) , which implies that the error introduced in the later discussion by using the theoretical value ( 0.092 ev ) would be significant . we note that the error in @xmath35 is due to the significant overbinding of o@xmath3 , and the underbinding obtained for bulk ag and bulk ag@xmath3o . we choose the zero reference state of @xmath37 to be the total energy of oxygen in an isolated molecule , i.e. , @xmath38 . this will be called the `` oxygen - rich '' condition . the so - called `` oxygen - poor '' value of @xmath24 is taken as , @xmath39 if the oxygen chemical potential becomes lower ( more negative ) than @xmath40 , di - silver oxide starts to decompose . for oxides of different transition metals , the `` oxygen poor '' limit can vary significantly depending on the heat of formation . for example , what may be rather `` oxygen rich '' for ruo@xmath3 ( with @xmath41 ev ) is already too `` oxygen poor '' for ag@xmath3o to be stable . the typical range of the allowed oxygen chemical potential is determined and limited by the pressures and temperatures used in industry and laboratories , namely , from several to thousand kelvin , and from ultra high vacuum ( @xmath42 atm . ) to several hundreds of atmospheres . for our investigations into various terminations of ag@xmath3o(111 ) surfaces , we will also consider the energy required to remove an oxygen atom from the surface and bring it to the vacuum . we call this the `` removal energy '' , and it is defined as , @xmath43 where @xmath44 is the total energy of the surface termination of interest , containing the o atom to be removed , @xmath45 is that of the reference system , i.e. , the surface structure remaining after the o atom has been removed . @xmath46 is the total energy of a free o atom where the spin - polarization energy ( @xmath471.60 ev ) is taken into account . the removal energy reflects the bondstrength of the removed o atom to the surface . to help analyze the electronic structure , we also calculate the density of states , @xmath48 , as well as the state - resolved dos , or projected dos , @xmath49 . the definitions are given in ref . . .calculated dft - gga and experimental bulk properties of ag@xmath51o . the lattice constant , @xmath52 , the distance between ag atoms , @xmath53 and oxygen atoms , @xmath54 , and the bondlength between nearest neighbor oxygen and silver atoms , @xmath55 , are given in . the bulk modulus , @xmath56 , ( in mbar ) , and the heat of formation , @xmath35 , ( in ev per oxygen atom ) are listed as well . [ tab : ag2o ] [ cols="<,<,<,<,<,<,<",options="header " , ] the nature of the bonding of the di - silver oxide surface is similar to that of the bulk oxide : oxygen and silver hybridize mainly via the ag-4@xmath57- and o-2@xmath18-like orbitals as shown in fig . [ fig : oxide - dos ] , where the atom projected density of states of the first o - ag - o trilayer is shown . compared to oxygen chemisorbed on the clean ag(111 ) surface ( see fig . 5 in ref . ) , the ag-4@xmath57 band is significantly narrower , which is due to the increased distance between the silver atoms and the weak hybridization with o-2@xmath18 states . the pdos in fig . [ fig : oxide - dos ] is rather similar to that of the oxide - like structures with same local bonding as presented in fig . 12 of ref . . [ fig : oxide - dos ] two features , one at @xmath475.53 ev and one at @xmath470.66 ev , with respect to the fermi energy , can be seen which are well separated from the center of the @xmath57-band . this is in contrast to the case of chemisorption on the metal substrate , where there is only one peak outside of the ag-4@xmath57 band , for example at @xmath471.35 ev for oxygen in the on - surface fcc - hollow site at coverage 0.25 ml . having studied the atomic and electronic structure of bulk ag@xmath3o and determined the thermodynamically most favorable surface of ag@xmath3o(111 ) , we now wish to investigate how a transition from the oxide - like structures , with positions of ag atoms commensurate with those of ag(111 ) , and laterally compressed by 14 % compared to the true oxide ( as reported in ref . ) , to the real oxide structure might occur . such a commensurate oxide - like structure is depicted in fig . [ fig : commens+4x4 ] ( top ) for an o concentration of 1.25 ml . in order to investigate such a transition , we study the difference of the total energy ( per o atom ) between oxide - like films ( with commensurate ag positions to those of the metal substrate ) and corresponding `` real '' ( unstrained ) oxide layers as a function of thickness . because the structures can be thought of as the stacking of o - ag - o trilayers ( see fig . [ fig : oxide ] ) , we calculate the energy difference with increasing numbers of trilayers . then the number of layers at which the true oxide structure becomes energetically more favorable , defines a critical thickness for the transition to occur . in this approach , the coupling between the ( strained ) oxide - like film with the underlying metal substrate is not included . we note that this study may be somewhat `` academic '' since there is so far no evidence for such compressed oxide structures , however , this structure is a plausible intermediate and perhaps can even be stabilized under certain experimental conditions . furthermore , additional insight into the o / ag system can be obtained by these calculations , and moreover , when compared to analogous studies performed for other o / transition - metal systems , in particular for o / ru(0001 ) , @xcite various trends may be identified . the results are shown in fig . [ fig : transition ] . a negative value indicates that the true oxide structures are energetically more favorable . we note that fig . [ fig : transition ] assumes that the number of o atoms is fixed . however , in many ( most ) experiments that aim to study very high coverages , the actual thermodynamic variable may be @xmath24 . for one o - ag - o trilayer , the `` strained '' [ commensurate with ag(111 ) ] oxide film is significantly more favorable than the corresponding true oxide , as indicated by the relatively large positive value of 0.53 ev , but the preference decreases quickly to 0.13 ev at three trilayers . on increasing the thickness further , the difference in the energy between the two structures decreases more slowly , and when the number of trilayers is nine , the true oxide film is energetically more favorable , as indicated by the negative value in fig . [ fig : transition ] . because each trilayer contributes half a monolayer of oxygen , the critical coverage for the transition from the commensurate oxide - like film to the true oxide film is around 3.5 - 4.5 ml . as the number of the trilayers approaches infinity , clearly the curve approaches the value corresponding to the energy difference ( per o atom ) between the bulk oxide and the `` strained '' [ commensurate with ag(111 ) ] bulk material . the electronic properties of the `` strained '' [ commensurate with ag(111 ) ] oxide film are very similar to those of the true oxide as can be seen from comparison of fig . 12 in ref . with fig . [ fig : oxide - dos ] . this at first may be somewhat surprising since the lateral dimensions of the commensurate , oxide - like structures are smaller , by 14 % . we find , however , that much of the associated strain energy is relieved by expansion of the ag layers perpendicular to the surface to accommodate the o atoms in between . through this relaxation , the lattice volume per atom is 92 % of that of the bulk oxide . in fig . [ fig : transition ] , analogous results for the phase transition from o - ru - o trilayer structures ( commensurate with ru(0001 ) ) to the corresponding real ruo@xmath3(110 ) @xcite oxide layers are also presented for comparison . bulk ruthenium dioxide has the rutile structure . the atomic geometry of ruo@xmath3(110 ) is shown as an inset in fig . [ fig : transition ] , as is that of the ( commensurate ) trilayer and its expansion into a corresponding true ruo@xmath3(110 ) layer ( bottom , left ) . the commensurate trilayer structures are , laterally , 15 % and 35 % smaller than for ruo@xmath3(110 ) ( i.e. , in the @xmath58 and @xmath59 directions of the rectangular ( 110 ) surface unit cell ) . in this case the transformation from commensurate trilayer to corresponding oxide layer , involves not only a lateral expansion like for the di - silver oxide structures , but also a `` rotation '' of the o atoms about the ru atoms , as seen from the inset ( explained in more detail in ref . ) . in particular , every second o - ru - o component rotates to be horizontal to the surface , while the alternative components rotate so that they are perpendicular to the surface . in contrast to the ag - o system , the phase transition into ruo@xmath13(110 ) oxide happens after only one o - ru - o trilayer . also , for ru(0001 ) , each trilayer contributes two monolayers of oxygen , so the critical coverage is around 2 - 4 ml . because of the considerable energy gain due to formation of the real ruo@xmath13 oxide layers , compared to continuous of stacking of the o - ru - o trilayers , the formation of the oxide structure will certainly take place once it is kinetically possible . it is interesting to note that the energy differences of oxygen in the trilayers and the corresponding bulk metal oxide layers , exhibits a rather similar dependence on the number of layers for both o / ag and o / ru , except shifted in energy . this makes sense since we consider the energy per o atom , and as the stack becomes larger , there are more and more o atoms in the center of the stack which have atomic environments which are similar ( and thus similar energies ) , i.e. , the affect on the energy due to the surfaces of the layers , diminishes . in view of this similarity , together with the fact that the heat of formation of e.g. , the rh ( 1.19 ev ) and pd ( 0.97 ev ) oxides lie between the extremes of ru ( 1.60 ev ) and ag ( 0.33 ev ) oxides , we may expect that investigations of analogous transitions from commensurate structures to the bulk oxide phase for rh and pd would exhibit a behavior intermediate to those of ru and ag . this would imply the trend that the transition from intermediate , commensurate , two - dimensional oxide films to the true oxide phase , occurs sooner ( i.e. , for thinner films ) for the elements more to the left in the periodic table , like ru , with larger heats of formation , compared to those towards the right , which have smaller heats of formation like ag . from a recent trend study on the oxidation at the basal plane of late @xmath60 transition metals ( from middle to right in the periodic table , i.e. , ru , rh , pd , ag ) , @xcite it was found that oxygen starts to enter the sub - surface region only after a certain critical coverage has been reached on the surface . namely , at coverages beyond which the heat of oxygen adsorption on the surface becomes unfavorable compared to the heat of formation of the bulk oxide . the reported critical coverages are about 1.0 ml , 1.0 ml , and 0.75 ml , and 0.25 ml for ru , rh , pd , and ag , respectively . beyond these critical coverages , onset of bulk oxide formation was proposed to take place . clearly , however , due to kinetic hindering , bulk oxide formation may _ not _ be able to occur `` immediately '' and then it may proceed via the formation of oxide - like layers as described above . we now turn to investigate the affect of the temperature and pressure on the stability of the various oxygen / silver structures , the energetics and atomic geometries of which were reported in our previous publications ; @xcite in particular , the energies are summarized in fig . 14 of ref . . to incorporate the effect of temperature and pressure , we calculate the gibbs free energy as a function of the oxygen chemical potential ( cf . [ eq - gibbs ] and [ eq : ochem ] ) similarly to the oxide surfaces , except here the reference system is the clean metal substrate . as discussed above , the `` oxygen poor '' limit is that below which di - silver oxide is unstable . for oxygen adsorption at the ag(111 ) surface , however , this limitation does nt necessarily exist . the results are shown in fig . [ fig : pt ] , where we only show the low energy structures . the dependence of the energy on @xmath24 , namely the slope of the line , reflects the o coverage . the higher the coverage , the steeper the gradient . we also include the corresponding temperatures for two given pressures , namely , that which corresponds to uhv and the other characteristic of the atmospheric pressure employed in catalysis . it can be seen that for low values of the oxygen chemical potential , i.e. @xmath61 ev , the clean ag(111 ) surface is thermodynamically most stable . for higher values , ( i.e. , in the range of @xmath470.78 ev to @xmath470.63 ev ) , a low coverage ( 1/16 ml ) of oxygen on the surface in fcc - hollow sites is thermodynamically most stable , and for values greater than @xmath470.63 ev and less than @xmath470.54 ev , slightly higher coverages of adsorbed o ( 1/9 ml and 1/4 ml ) become most stable . increasing the oxygen chemical potential to values greater than @xmath470.54 ev leads to the formation of the reconstructed @xmath62 phase ( labeled `` 0.375 ml '' ) , which is stable up to @xmath63 ev , i.e. to the `` o - poor '' limit ( left vertical dashed line ) , where the bulk oxide becomes stable . the atomic geometry of the @xmath62 phase is illustrated in fig . [ fig : commens+4x4 ] ( bottom ) . the atomic structure is very similar to a trilayer of ag@xmath3o(111 ) , except with one ag in the @xmath62 cell missing . the basis for this proposed structure ( cf . refs . ) is the observation that the area of the @xmath64-ag@xmath3o(111 ) surface unit cell is practically four times that of clean @xmath65-ag(111 ) ; the difference in the length of the lattice vectors is only 1.2@xmath66 ( 0.3@xmath66 from the experimental lattice constants ) . although the mentioned _ surface unit cells _ are ( practically ) commensurate , the positions of the ag atoms within the @xmath64 oxide surface cell are , all except one , incommensurate with those of the ag(111 ) surface . it can be seen that for values of @xmath67 greater than @xmath68 ev , the commensurate oxide - like structures ( see e.g. , fig . [ fig : commens+4x4 ] ) ( top ) are also quite favorable ( labeled by the o coverages , 0.75 , 1.25 , 2.25 ml ) . it can be noticed that the thicker the film , the more stable the structure is ( i.e. , the steeper the curve ) . clearly , in the limit of an infinitely thick film , the corresponding line would be very close to the left vertical one . this is so because the average o binding energy in the thick commensurate oxide - like structures is very similar to that of o in the bulk oxide as described above . from fig . [ fig : pt ] it can be concluded that the bulk oxide is the thermodynamically most favorable structure once the oxygen chemical potential is higher than the `` o - poor '' condition , which is expected . however , the kinetics of oxide formation can be very slow and the thickness of the oxide film will be limited by o and metal bulk diffusion . thus , it is unlikely that full thermal equilibrium of o@xmath3 with bulk di - silver oxide will be achieved . the results discussed above can be displayed as a continuous function of temperature and pressure by constructing a two - dimensional phase diagram , if we only consider the lowest energy structures . this has been done in ref . and the results are shown again in fig . [ fig : pt2 ] . before giving a discussion , it is necessary to consider the magnitude of the error bars in the temperature and pressure . an indication can be obtained from knowledge of the stability of the bulk oxide , i.e. , that it decomposes at 460 k under atmospheric pressure . from figs . [ fig : pt ] and [ fig : pt2 ] , the calculations predict that the bulk oxide decomposes at @xmath12350 k at atmospheric pressure , i.e. , around 110 k lower than experiment . at 460 k , the bulk oxide is unstable when the pressure is lower than 10@xmath69 atm . , i.e. , compared to 1 atm . as observed experimentally . the reason for this disparity may be due to _ systematic _ errors of the dft - gga approach and/or to neglect of the entropy contribution . in particular , we note that the experimental @xmath70 contribution for bulk silver at 450 k and 1 atm . is @xmath470.48 ev per two silver atoms , and @xmath470.70 for ag@xmath3o per formula unit . @xcite thus , the entropy contribution from di - silver oxide , compared to silver bulk ( the entropy of o@xmath13 has already been included in the chemical potential ) , is greater by 0.22 ev per two silver atoms , which will shift the left vertical line in fig . [ fig : pt ] further to the left . therefore , in figs . [ fig : pt ] and [ fig : pt2 ] , we generally expect that the temperature is underestimated , but the error less than 110 k , while correspondingly , the pressure is overestimated , by less than three orders of magnitude . for the temperature values quoted below , we give an estimate of the upper limit in brackets , that is , 110 k higher than that obtained from fig . [ fig : pt2 ] . in spite of these uncertainties , we believe that the relative stability of the various systems and our general understanding and conclusions will not be affected . typically , vibrational contributions to _ differences _ in the gibbs free energies of extended systems exhibit some cancellation . @xcite however , when there are additional atomic or molecular species which are not present in the reference system , the situation may be different as there will be no effective cancellation of the contributions of such species between surface and reference systems ( as is the case using the _ clean _ ag(111 ) as reference ) . we investigate this aspect in more detail for the present system ; in particular , we consider low coverages of on - surface oxygen and thin oxide - like structures . we calculate the difference in the vibrational and entropic contributions to the gibbs free energy , @xmath71 , as , @xmath72 \ } \quad , \label{eq : vibr}\end{aligned}\ ] ] where @xmath73 ( see e.g. ref . ) . the vibrations which enter eq . [ eq : vibr ] are those of the adsorbed o atoms . the calculated perpendicular vibrational frequency for 0.25 ml of oxygen in fcc sites on the surface is 50 mev ( 400 @xmath74 ) , and we assume that the lateral vibrations are the same ( this is a somewhat rough approximation but suffices for the present discussion , see below ) . thus , in calculating @xmath75 we multiply by a factor of three corresponding to these three modes . for the thin commensurate oxide - like structure containing one o atom on the surface in fcc sites and one sub - surface o in tetrahedral sites ( i.e. , a total coverage of 0.5 ml ) , the calculated perpendicular vibrational frequencies are 56 mev ( 455 @xmath74 ) and 65 mev ( 526 @xmath74 ) , respectively , for these two species . again , assuming each o atom has three modes equal to the perpendicular ones , we obtain the results presented in fig . [ fig : vibrations ] . it can be seen from fig . [ fig : vibrations ] that for on - surface o , for temperatures up to about 800 k , the contribution is a maximum of about @xmath762 mev / @xmath77 . for lower coverages of on - surface o such as the 1/16 and 1/9 ml structures , the contribution will be smaller ( assuming the frequency is similar ) . for the oxide - like structure at coverage 0.5 ml , the effect is larger , namely 5 - 6 mev / @xmath77 for temperatures up to about 400 k , thereafter decreasing . for the @xmath62 thin oxide structure with coverage 0.375 ml , we expect the vibrational frequencies of the o species will be similar to those of the structure at 0.5 ml since the local coordination of the o atoms is similar , as well as the electronic structure . in addition , the vibrational frequency of o in bulk ag@xmath3o is reported to be 66 mev ( 530 @xmath74 ) , @xcite which is very close to that of the o atom in the sub - surface tetrahedral site of the 0.5 ml coverage structure ( 65 mev ) . because of the lower o coverage of the @xmath62 structure , the contribution to the free energy will be smaller than the 0.5 ml structure by about 0.375/0.5 = 0.75 , giving a maximum contribution at low temperatures of 4.5 mev / @xmath77 ( dot - dashed line in fig . [ fig : vibrations ] ) . for the @xmath78 structure , in the temperature range 400 k to 730 k , it gives rise to a destabilization of less than 3.5 mev / @xmath77 ( the value at 400 k , which decreases with temperature ) , which on inspecting the energy scale of fig . [ fig : pt ] can be seen to be non - negligible . the relative energy _ difference _ , however between the on - surface species and @xmath78 structure , which determines the _ sequence _ of the stable phases , can be seen from fig . [ fig : vibrations ] to be affected by less than 1.9 mev / @xmath77 ( @xmath79 mev / @xmath77 ) for temperatures greater than 400 k , which will therefore have a small effect . we note that if the lateral vibrations would be 30 % greater than the perpendicular ones for the on - surface and oxide - like structures considered , this would result only in a small increase in the contribution to the free energy , i.e. of 0.5 and 1.2 mev / @xmath77 , respectively , at @xmath19 close to zero . the above analysis may be regarded as qualitative , but it gives an estimate of the magnitude of the changes in the free energy due to consideration of vibrational and entropic effects of the extended systems for the energetically most important o / ag structures . clearly , for other systems with adsorbed species that have high frequencies such as oh ( of the order 400 mev ) , @xcite the contribution to the free energy from vibrational and entropic effects will be significantly greater and will have to be taken into account in determination of the phase diagram . summarizing the results contained in fig . [ fig : pt2 ] , we obtain the following predictions : under uhv conditions ( @xmath80 atm . ) , chemisorption of oxygen on ag(111 ) with low coverage ( less than 0.0625 ml ) is stable up to about 360 k ( incl . corrections : 470 k ) . this is in - line with experimental results @xcite where it has been found that for temperatures less than 490 k , 0.03 ml of oxygen can be adsorbed on the ag(111 ) surface . under uhv conditions , the calculations also predict the stability of the @xmath62 phase at the lower temperatures of 160 k ( incl . corrections : 270 k ) to 260 k ( incl . corrections : 370 k ) . this is also consistent with scanning tunneling microscopy ( stm ) results obtained under uhv conditions at low ( 4 k ) temperatures @xcite where atomic resolution images of the @xmath62 phase have been achieved . the @xmath62 phase , however , can not be directly prepared under these conditions as it is known from experiment that temperatures greater than 400 k are required in order for the necessary atomic rearrangement take place . it is therefore clear that it is the kinetics that prevent the reconstruction and formation of this phase at low pressures and temperatures . from fig . [ fig : pt2 ] , it can be seen that the @xmath62 phase should develop into the oxide if the oxygen pressure is raised or if the temperature is lower than 160 k ( incl . corrections : 270 k ) . such transitions , however , may be prevented or hindered by kinetics ( e.g. diffusion barriers ) since extensive mass transport is required to form the oxide . at atmospheric pressure ( @xmath81 atm . ) , typical of that used in industry , the bulk oxide is calculated to be only stable up to around 350 k ( incl . corrections : 460 k ) . from this point of view , it can be concluded that for industrial reactions which take place at considerably higher temperatures , di - silver oxide can not be the active phase . the @xmath62 surface - oxide phase , however , is stable in the temperature range of 350 - 530 k ( incl . corrections : 460 - 640 k ) which coincides with that at which epoxidation of ethylene occurs . we therefore propose that such atomic o species are actuating the reaction . with regard to the catalytic reactions that take place at even higher temperatures , it can be seen that only very low coverages of chemisorbed o are stable up to 720 k ( incl . corrections : 830 k ) ; o atoms adsorbed at under - coordinated surface ag atoms are up to 865 k ( incl . corrections : 975 k ) ( see below ) . therefore these species could possibly play a role in these reactions . it is interesting to note that regardless of the pressure , fig . [ fig : pt2 ] shows that the _ ordering _ of the stable phases is the same , so that the same phases should be observable at different pressures if the temperature is adjusted accordingly . this indicates a way to bridge the pressure gap , i.e. , by appropriate choice of the pressure and temperatures to stay within the same phase . however , it is clear that sometimes this will not be possible within reasonable times because of kinetic hindering for dissociation and sticking of o@xmath3 ( i.e. , delivery of sufficient concentrations of atomic oxygen to the surface ) , as well as to kinetic limitations for adequate atomic rearrangement . experimentally , it has been reported that a `` strongly bound '' oxygen species ( named o@xmath82 ) forms after high temperature ( @xmath12900 k ) and atmospheric pressure exposure of polycrystalline silver , as well as ag(111 ) , to an oxygen environment . @xcite temperature programmed desorption spectra show that o@xmath82 desorbs at around 900 k and it is proposed by the authors to be the active species for the dehydrogenation of methanol @xcite where it diffuses to the surface from bulk sites . at the high temperature of @xmath12900 k in an oxygen atmosphere , silver reconstructs to form many ( 111 ) facets @xcite at the surface and it is speculated that this means that oxygen diffusion takes place by a so - called `` substitutional diffusion mechanism '' which the authors argue has higher barriers than usual interstitial diffusion , and to give rise to the o@xmath3 desorption peak at the high temperature of 900 k. from our calculations , this picture is difficult to understand , in particular , how sufficient amounts of bulk - dissolved oxygen can be formed . from our theoretical study , we know that at lower coverage , oxygen will prefer to stay on the surface , instead of in the sub - surface or bulk region , and for high o concentrations , oxide - like structures are preferred . @xcite furthermore , our results showed that substitutional oxygen adsorption in the bulk ( i.e. , o on ag sites ) is energetically unfavorable . @xcite the importance of steps and vacancies as active sites has been well documented , from both experimental and theoretical points of view . @xcite in our previous studies we investigated the binding of oxygen atoms on surfaces with pre - existing vacancies . @xcite we found the binding energy is stronger than on the ( 111 ) terraces , e.g. , for 0.25 ml adsorbed in fcc sites on a @xmath83/ag(111 ) vacancy array the binding energy is 3.83 ev while it is 3.52 ev on the perfect terrace , at the same coverage . a similar behavior can be expected for adsorption at under - coordinated sites at step edges , as has been shown theoretically for oxygen at other transition metals , see e.g. refs . and . to investigate the affect of pressure and temperature on the stability of such o species , we calculated the gibbs free energy of formation . we find that under atmospheric pressure , the o atoms are stable to around 770 k ( incl . corrections : 880 k ) on a @xmath84/ag(111 ) vacancy array , and even to 865 k ( incl . corrections : 975 k ) on a @xmath83/ag(111 ) vacancy array . the existence and concentration of these species , however , clearly depends on the number of defect sites at the surface . in view of the above , it could be speculated that the tpd signal of `` o@xmath82 '' is due not to substitutional diffusion of oxygen atoms from the bulk to the surface as proposed , but due to o atoms adsorbed at under - coordinated surface silver sites . as mentioned in the introduction , in the literature , the existence of surface ag vacancies has been proposed to stabilize a molecular - like ozone ( o@xmath7 ) species , which has been suggested to be active for the partial oxidation of ethylene . @xcite as reported in ref . , we calculated the adsorption energy of such a species on ( @xmath85 and @xmath64 vacancy arrays , corresponding to oxygen coverages of 1/3 and 1 ml , respectively . we found that these species were energetically _ unfavorable _ compared to the commensurate oxide - like structures described above . if , however , we assume that ag vacancies exist , i.e. , neglecting the energy cost to create the vacancy , then the _ binding energy _ is the most favorable of all for coverage 1 ml . we can estimate the concentration of such ag vacancies using our calculated values of the vacancy formation energy , together with an arrhenius type equation : for a @xmath64 vacancy array with a vacancy concentration of 1/3 ml , the formation energy is 0.53 ev ; it is 0.44 ev for a concentration of 0.25 ml , and 0.55 ev for a concentration of 0.11 ml . considering a temperature of 550 k , corresponding to that around which epoxidation of ethylene takes place , this implies a concentration of thermally induced surface vacancies of less than about @xmath86 ml , which is very small . furthermore , calculation of the _ free energy _ shows that this species is less stable than the others . our results thus do not support the molecular ozone - like species as playing an important role , but point to oxygen atoms of the thin @xmath62 surface - oxide as being the main species actuating epoxidation of ethylene . as seen from fig . [ fig : pt ] , our calculations predict that thick oxide - like films can form at ag(111 ) , but only for low temperatures [ e.g. , @xmath87350 k ( 460 k incl . corrections ) at atmospheric pressure ] . the extremely low sticking coefficient of o@xmath3 at ag(111 ) , @xcite as well as the found low thermal stability of these structures , may hinder their formation . for example , at high temperatures at which diffusion of o ( or ag ) atoms can effectively occur , the material may become unstable . furthermore , when using no@xmath3 to deliver o atoms to the surface , the substrate has to be held at temperatures _ higher _ than the temperature at which di - silver oxide decomposes , in order to desorb the no molecules . using , however , low temperatures and _ atomic oxygen _ or ozone ( which readily dissociates at the surface ) instead of o@xmath3 ( or no@xmath3 ) , we predict that thick oxide structures could be observed experimentally providing that diffusion barriers can be overcome . interestingly , it was found that di - silver oxide ( ag@xmath3o ) can be prepared using a mixture of ozone and o@xmath13 or microwave generated atomic oxygen , at room temperature on _ polycrystalline _ silver foil . @xcite the oxygen generated in these processes is quite active , for example , small amounts of ozone ( 5 mol.@xmath66 o@xmath88 in o@xmath13 ) can generate the o partial pressure equivalent to that of a pure o@xmath3 atmosphere at 10@xmath89 atm . at a temperature of 465 k. @xcite actually , in these mixtures of reaction gas , ag@xmath13o can be stable to temperatures in excess of 773 k. @xcite in the studies of waterhouse _ et . al . _ , @xcite it has been found that at room temperature , oxidation of polycrystalline silver foil is improved significantly after the sample has been `` scratched '' , which generates a high concentration of defects which act as the sites for the oxide film nucleation , highlighting the importance of defects . also diffusion of oxygen and silver atoms is seemingly improved significantly due to the non - uniformity of the growth . we note that subsequent to completion of our studies , very recent experimental work reported thick oxide formation at ag(111 ) by using a flux of atomic oxygen , @xcite thus supporting our findings . through density - functional theory calculations , and incorporating the effect of the atmospheric environment , we obtained the pressure - temperature phase diagram for the oxygen / ag(111 ) system , which reveals important insights into this system and its function as an oxidation catalyst : the calculations predict that a thin surface - oxide structure is most stable in the temperature and pressure range of ethylene epoxidation and we propose it ( and possibly other similar structures ) contains the main o species actuating the catalysis . low coverages of chemisorbed o become more stable for higher temperatures and could also play a role in other oxidation reactions . the only species stable for temperatures in excess of 720 k ( incl . corrections : 830 k ) , are o atoms adsorbed at under - coordinated surface ag atoms . thick bulk - like di - silver oxide can safely be ruled out as playing an important role in the technical catalytic reactions due to the low thermal stability , causing decomposition at temperatures lower than those used in the reactors . they are , however , the most stable structures for strongly o - rich conditions and low temperatures ( e.g. , up to 330 - 460 k at atmospheric pressure ) . interestingly , the electronic structure of thick oxide surfaces are very similar to that of the thin @xmath62 surface oxide structure , which raises the question of whether di - silver oxide might be used as a _ low - temperature _ oxidation catalyst . a molecular ozone - like species adsorbed at a surface vacancy , as had been proposed in the literature as being the active species for the epoxidation of ethylene , is energetically unfavorable , as is bulk - dissolved oxygen , which strongly questions the latters hitherto thought crucial role in , e.g. , the partial oxidation of methanol to formaldehyde . we also studied the bulk and ( 111 ) surface properties of ag@xmath13o , the most stable silver oxide . the interaction of oxygen and silver is via a weak hybridization between ag-4@xmath57 - 5@xmath90 and o-2@xmath18 orbitals , with substantial occupation of antibonding orbitals , which explains its low stability . various terminations of the ag@xmath13o(111 ) surface were investigated as a function of the oxygen chemical potential and it is found that the surface with oxygen in the hollow site ( i.e. , the stoichiometric surface ) is most favorable for all of the range of the o chemical potential considered , while the ag - terminated surface and surfaces with additional oxygen adsorbed on top of exposed `` cus '' surface ag atoms are unfavorable . on investigating the energetics of a transition from commensurate oxide - like structures to the `` real '' ( unstrained ) oxide , we find that the transition becomes favorable only at nine o - ag - o trilayers , which corresponds to @xmath124.5 ml oxygen . the approach of `` _ ab initio _ atomistic thermodynamics '' , i.e. , of calculating free energies for many possible atomic geometries and predicting the lowest energy structures in contact with species in a gas - phase environment , is a promising approach to investigate a system under different pressure and temperature conditions , in particular , from those of typical theoretical surface science , right up to those of technical heterogeneous catalysis . the obtained phase diagram also shows that if one stays within the region of one phase or along a phase boundary , then a bridging of the pressure gap should be possible . furthermore , it may be expected that the regions which lie close to the boundaries in the computed surface phase diagram may exhibit enhanced thermal fluctuations , i.e. , here the dynamics of atomistic processes may be particularly strong , and this may be particularly relevant for catalysis under realistic conditions . furthermore , we note however , that an understanding of a _ full catalytic cycle _ , requires a kinetic modeling that includes all reactant species and intermediates , which for complex systems such as those discussed in the present paper , is not yet possible . + a. nagy , g. mestl , d. herein , g. weinberg , e. kitzelmann , r. schlgl , j. catal . * 182 * , 417(1999 ) ; a. nagy , g. mestl , and r. schlgl , j. catal . * 188 * , 58(1999 ) ; a. nagy , and g. mestl , appl . catal . a * 188 * , 337 ( 1999 ) . c. rehren , m. muhler , x. bao , r. schlgl , and g. ertl , z. phys . chem . * 174 * , 11 ( 1991 ) ; x. bao , m. muhler , b. pettinger , r. schlgl , and g. ertl , catal . * 22 * , 215 ( 1993 ) ; h. schubert , u. tegtmeyer , and r. schlgl , catal . lett . * 28 * , 383 ( 1994 ) . t. zambelli , j. wintterlin , j. trost and g. ertl , science * 273 * , 1688 ( 1996 ) ; s. dahl , a. logadottir , r. c. egeberg , j. h. larsen , i. chorkendorf , e. trnqvist , and j. k. nrskov , phys . lett . * 83 * , 1814 ( 1999 ) ; b. hammer , phys . lett . * 83 * , 3681 ( 1999 ) .
to help understand the high activity of silver as an oxidation catalyst , e.g. , for the oxidation of ethylene to epoxide and the dehydrogenation of methanol to formaldehyde , the interaction and stability of many different oxygen species at the ag(111 ) surface has been studied for a wide range of coverages . through calculation of the free energy , as obtained from density - functional theory and taking into account the temperature and pressure via the oxygen chemical potential , we obtain the phase diagram of o / ag(111 ) . our results reveal that a _ thin _ surface - oxide structure is most stable for the temperature and pressure range of ethylene epoxidation and we propose it ( and possibly other similar structures ) contains the species actuating the catalysis . for higher temperatures , low coverages of chemisorbed oxygen are most stable , which could also play a role in oxidation reactions . for temperatures greater than about 775 k there are _ no _ stable oxygen species , except for the possibility of o atoms adsorbed at under - coordinated surface sites ( i.e. , imperfections , defects ) . at low temperatures ( @xmath0 400 k at atmospheric pressure ) , provided kinetic limitations can be overcome , thicker oxide - like structures are predicted . due to their low thermal stability , however , they can be ruled out as playing an important role in the heterogeneous reactions under technical conditions . bulk dissolved oxygen and a molecular ozone - like species adsorbed at a surface vacancy , as have been proposed in the literature , are found to be energetically unfavorable . the employed theoretical approach for calculating _ free energies _ and predicting the lowest energy structures in contact with species in the environment ( `` _ ab initio _ , atomistic thermodynamics '' ) , affords investigation of a system that seemlessly connects standard ( @xmath1 k ) dft results , characteristic of `` typical '' theoretical surface science studies , through to those valid for the conditions of catalysis . though the error bar of the noted theoretical temperatures is noticeable ( @xmath255 k ) , the identified trends and physical descriptions are useful .
let @xmath0 be the moduli space of reduced curves of degree @xmath1 in @xmath2 with a prescribed configuration of singularities @xmath3 . of the space @xmath4 of reduced plane curves with degree @xmath1 and set of singularities @xmath3 by the ` standard ' group action of @xmath5 . ] a pair of curves @xmath6 in @xmath0 is said to be a _ zariski pair _ if it satisfies the following two conditions ( cf . @xcite ) : 1 . [ c1 ] @xmath7 and @xmath8 have the same _ combinatoric _ , that is , there exist regular neighbourhoods @xmath9 and @xmath10 of @xmath7 and @xmath8 respectively such that the pairs @xmath11 and @xmath12 are homeomorphic ; 2 . [ c2 ] the pairs @xmath13 and @xmath14 are not homeomorphic . it is easy to check that if both @xmath7 and @xmath8 are _ irreducible _ then the first condition is always satisfied . the first zariski pair appears in the works by o. zariski @xcite ( see also @xcite and @xcite ) . the members of the pair are irreducible curves of degree @xmath15 , which is the smallest degree for which zariski pairs exist . the existence of a zariski pair in a moduli space gives an information about its connected components . indeed , if @xmath0 has a zariski pair @xmath6 , then @xmath7 and @xmath8 necessarily belong to different connected components ( cf . @xcite ) in particular , if @xmath0 has a zariski pair , then it is not connected . ( the converse statement is not clear . two curves coming from two different connected components may have the same embedded topology . ) now to check whether a given a pair of curves @xmath16 with the same combinatoric is a zariski pair , one can first try to calculate the generic alexander polynomials of the curves . if these polynomials are different , then the curves do not have the same embedded topology , and @xmath6 is a zariski pair . however , it may happen that these polynomials are the same although @xmath6 is a zariski pair . in this case the pair is said to be _ alexander equivalent_. the first example of such a pair was given by e. artal bartolo and j. carmona ruber @xcite for _ reducible _ curves of degree 7 . the first examples dealing with _ irreducible _ curves are due to the second author @xcite ( curves of degree @xmath17 ) and @xcite ( curves of degree @xmath18 ) . in @xcite , a. degtyarev proved that alexander equivalent zariski pairs also appear on irreducible curves of degree @xmath15 . however he did not give any explicit example ( only the existence is proved ) . the aim of the present paper is to construct a concrete example of such a pair . let @xmath19 be homogeneous coordinates on @xmath2 and @xmath20 the affine coordinates defined by @xmath21 and @xmath22 on @xmath23 , as usual . we consider the projective curves @xmath7 and @xmath8 in @xmath2 defined by the affine equations @xmath24 and @xmath25 respectively , where @xmath26 both @xmath7 and @xmath8 are _ irreducible sextics _ with the set of singularities @xmath27 . ( we recall that a point @xmath28 in a curve @xmath29 is said to be an @xmath30singularity ( @xmath31 ) if the germ @xmath32 is topologically equivalent to the germ at the origin of the curve defined by @xmath33 . ) for the curve @xmath7 , the @xmath34singularities are located at @xmath35 and @xmath36 while the @xmath37singularity is at @xmath38 . for @xmath8 , the @xmath34singularities are at @xmath39 and the @xmath37singularity is at @xmath38 . the _ real _ plane sections of @xmath7 and @xmath8 are shown in figure [ figa92a4rps ] and [ figa92a4narps ] respectively . ( in the figures we do not respect the numerical scale . ) the curve @xmath8 is symmetric with respect to the @xmath40axis . the curve @xmath7 has no particular symmetry . notice that after the analytic change of coordinates @xmath41 the equation of @xmath8 near the origin takes the form @xmath42 as the leading term @xmath43 is positive on @xmath44 , the origin is an isolated point of the real plane section of @xmath8 ( cf . figure [ figa92a4narps ] ) . finally , since @xmath7 and @xmath8 are not of torus type ( cf . @xcite and @xcite ) , their generic alexander polynomials are trivial ( cf . @xcite ) . @xmath45,width=302,height=188 ] @xmath46,width=302,height=188 ] [ maintheorem ] the fundamental group @xmath47 is isomorphic to @xmath48 while @xmath49 is isomorphic to @xmath50 , where @xmath51 is the dihedral group of order @xmath52 . in particular @xmath6 is an alexander equivalent zariski pair and the moduli space of irreducible sextics with the set of singularities @xmath53 has at least two connected components . theorem [ maintheorem ] is proved in sections [ abelian ] and [ nonabelian ] below . the curve @xmath8 is an example of so called _ @xmath51sextics_. ( a @xmath51sextic is a non torus irreducible sextic with simple singularities and whose fundamental group factors to the dihedral group @xmath51 . ) the existence of @xmath51sextics was first proved , purely arithmetically , by a. degtyarev @xcite who showed that there exist exactly 8 equisingular deformation families of such curves , one family for each of the following sets of singularities : @xmath54 @xmath55 first explicit examples and fundamental groups of @xmath51sextics were given in @xcite ( see also @xcite for the sets of singularities @xmath56 and @xmath57 ) . furthermore , degtyarev also observed in @xcite that the configurations @xmath58 can be realized not only by @xmath51sextics but also by irreducible non torus sextics @xmath29 for which the group @xmath59 does _ not _ factors to @xmath51 . ( in particular , since the alexander polynomial of an irreducible non torus sextic with simple singularities is always trivial , these 5 sets of singularities give rise to alexander equivalent zariski pairs of irreducible sextics . ) however , degtyarev did not give any explicit equation of such a curve @xmath29 ( only the existence is proved ) and did not calculate the group @xmath59 . the curve @xmath7 in theorem [ maintheorem ] is the first explicit example of such a curve i.e. , a curve whose fundamental group does not admit a dihedral quotient although its set of singularities can be realized by a @xmath51sextic as well . ( in particular , the pair @xmath6 in theorem [ maintheorem ] is also the first concrete example of an alexander equivalent zariski pair dealing with irreducible sextics . ) in this section , we show that @xmath61 . in fact , it suffices to prove that @xmath47 is abelian . indeed , by hurewicz s theorem , if @xmath47 is abelian , then it is isomorphic to first integral homology group @xmath62 and it is well known that @xmath63 . to show that @xmath47 is abelian , we use zariski van kampen s theorem with the pencil given by the vertical lines @xmath64 , @xmath65 ( cf . @xcite and @xcite ) . we always take the point @xmath66 as base point for our fundamental groups . this point is nothing but the axis of the pencil , which is also the point at infinity of the lines @xmath67 . note that it does not belong to the curve . the discriminant @xmath68 of @xmath69 as a polynomial in @xmath70 is the polynomial in @xmath40 given by @xmath71 where @xmath72 . this polynomial has exactly 10 distinct complex roots : @xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 . the singular lines of the pencil are the lines @xmath83 ( @xmath84 ) corresponding to these 10 roots . the lines @xmath85 and @xmath86 intersect the curve at its singular points . all the other singular lines are tangent to @xmath7 . see figure [ figa92a4rps ] . generators at @xmath87,width=302,height=188 ] we consider the generic line @xmath88 and choose generators @xmath89 of the fundamental group @xmath90 as in figure [ figa92a4eta4 ] , where @xmath91 is small enough . the @xmath92 s ( @xmath93 ) are _ lassos oriented counter clockwise _ around the 6 intersection points of the line @xmath88 with the curve i.e. , the 6 complex roots of the equation @xmath94 . ( in the figures , a lasso is represented by a path ending with a bullet . ) the zariski van kampen theorem says that @xmath95 where @xmath96 is the normal subgroup of @xmath97 generated by the monodromy relations associated with the singular lines of the pencil . to find these relations , we fix a ` standard ' system of generators @xmath98 for the fundamental group @xmath99 as follows . each @xmath100 is a lasso ( oriented counter clockwise ) around @xmath101 with base point @xmath102 . for @xmath103 , the tail of @xmath100 is a union of real segments and half circles around the exceptional parameters @xmath104 ( @xmath105 ) located in the real axis between the base point @xmath102 and @xmath101 . its head is the circle @xmath106 with centre @xmath101 and radius @xmath107 . the lasso @xmath108 corresponding to the non real root @xmath109 is given by @xmath110 , where @xmath111 is the loop obtained by moving @xmath40 once on the circle @xmath112 , starting at @xmath113 , while @xmath114 is the path obtained when @xmath40 moves on the real axis from @xmath102 to @xmath115 , makes half turn on the circle @xmath116 , from @xmath115 to @xmath117 , moves on the real axis from @xmath117 to @xmath118 , makes half turn on the circle @xmath119 , from @xmath118 to @xmath120 , moves on the real axis from @xmath120 to @xmath121 , makes half turn on the circle @xmath122 , from @xmath121 to @xmath123 , moves on the real axis from @xmath123 to @xmath124 , and finally moves in a straight line from @xmath124 to @xmath113 . ( here @xmath124 and @xmath125 denote the real and the imaginary parts of @xmath109 respectively . ) the lasso @xmath126 is defined similarly from a loop @xmath111 and a path @xmath114 meeting at @xmath127 . the monodromy relations around the singular line @xmath83 are obtained by moving the generic ` fibre ' @xmath128 isotopically ` above ' the loop @xmath100 , and by identifying each @xmath92 ( @xmath93 ) with its image by the terminal homeomorphism of this isotopy . ( for details see @xcite and @xcite . ) the remaining of the proof is to determine these relations . generators at @xmath129,width=302,height=188 ] when @xmath40 moves on the real axis from @xmath102 to @xmath115 , the 6 complex roots of the equation ( in @xmath70 ) @xmath24 ( and , consequently , the 6 generators @xmath131 ) are deformed as in figure [ figa92a4eta5 ] . the singular line @xmath132 is tangent to the curve at the simple point @xmath133 , and the intersection multiplicity @xmath134 of this line with the curve at this point is 2 . therefore , by the implicit functions theorem , the germ @xmath135 is given by @xmath136 where @xmath137 . it follows that when @xmath40 runs once counter clockwise on the circle @xmath116 , starting at @xmath115 , the variable @xmath70 makes half turn counter clockwise on the dotted circle around @xmath138 ( cf . figure [ figa92a4eta5 ] ) . the monodromy relation at @xmath130 is then given by @xmath139 at @xmath129 , the generators are as in figure [ figa92a4eta5 ] . in figure [ figa92a4eta6 ] , we show how the @xmath92 s are deformed when @xmath40 first makes half turn counter clockwise on the circle @xmath116 , from @xmath115 to @xmath117 , and then moves on the real axis from @xmath117 to @xmath118 . the singular line @xmath141 is also tangent to @xmath7 at a simple point @xmath142 and @xmath143 . therefore , as above , the monodromy relation we are looking for is simply given by @xmath144 generators at @xmath145,width=302,height=188 ] the singular line @xmath85 passes through the singular points @xmath38 and @xmath36 which are singularities of type @xmath37 and @xmath34 repectively . at @xmath38 , the curve has two branches @xmath147 and @xmath148 given by @xmath149 it follows that when @xmath40 runs once counter clockwise on the circle @xmath150 , starting at @xmath102 , the generators @xmath151 and @xmath152 make @xmath153 turns counter clockwise on the corresponding dotted circle ( cf . figure [ figa92a4eta4 ] ) . the monodromy relation around @xmath85 that comes from the singular point @xmath38 is then given by @xmath154 at @xmath36 , a puiseux parametrization of @xmath7 is given by @xmath155 hence , when @xmath40 goes once counter clockwise on the circle @xmath150 , starting at @xmath102 , the generators @xmath156 and @xmath157 make @xmath158turn counter clockwise on the corresponding dotted circle ( cf . figure [ figa92a4eta4 ] ) . the monodromy relation around @xmath85 that comes from the singular point @xmath36 is then given by @xmath159 at @xmath87 , the generators are as in figure [ figa92a4eta4 ] . now , when @xmath40 makes half turn counter clockwise on the circle @xmath150 , from @xmath102 to from @xmath161 , and then moves on the real axis from @xmath161 to @xmath162 , the @xmath92 s are deformed as in figure [ figa92a4eta3 ] , where @xmath163 the singular line @xmath164 is tangent to the curve at a simple point , with intersection multiplicity 2 , and the monodromy relation we are looking for is given by @xmath165 generators at @xmath166,width=302,height=188 ] generators at @xmath167,width=302,height=188 ] when @xmath40 makes half turn counter clockwise on the circle @xmath169 , from @xmath162 to @xmath170 , then moves on the real axis from @xmath170 to @xmath171 , the @xmath92 s are deformed as in figure [ figa92a4eta2 ] , where @xmath172 the monodromy relation at @xmath168 is also an usual multiplicity 2 tangent relation : @xmath173 generators at @xmath174,width=302,height=188 ] in figure [ figa92a4eta1 ] , we show how the @xmath92 s are deformed when @xmath40 makes half turn counter clockwise on the circle @xmath176 , from @xmath171 to @xmath177 , then moves on the real axis from @xmath177 to @xmath178 . the monodromy relation around @xmath179 is a multiplicity 2 tangent relation given by @xmath180 equivalently @xmath181 . since @xmath182 ( by ( [ rela92a4eta3 ] ) ) , it follows that @xmath183 combined with ( [ rela92a4eta6 ] ) , this gives @xmath184 combined with ( [ rela92a4eta2 ] ) , this in turn implies @xmath185 we recall that , at @xmath145 , the generators are as in figure [ figa92a4eta6 ] . when @xmath40 makes half turn counter clockwise on the circle @xmath119 , from @xmath118 to @xmath120 , the @xmath92 s are deformed as shown in figure [ figa92a4eta6bis ] , where @xmath187 generators at @xmath188,width=302,height=188 ] [ lemmaa92a4 ] when @xmath40 moves on the real axis from @xmath120 to @xmath121 , the @xmath92 s are deformed as figure [ figa92a4eta7 ] . lemma [ lemmaa92a4 ] is not obvious and will be proved at the end of this section . before , let us complete the calculation of @xmath47 . generators at @xmath189,width=302,height=188 ] the monodromy relation at @xmath186 is a multiplicity @xmath190 tangent relation given by @xmath191 notice that , by ( [ rela92a4s1 ] ) , this relation can also be written as @xmath192 moreover , still using ( [ rela92a4s1 ] ) , we can write the relation ( [ rela92a4eta400 ] ) as @xmath193 the latter , combined with ( [ rela92a4eta7s ] ) , implies @xmath194 this already shows that the fundamental group @xmath47 is abelian . ( we do not need to consider the monodromy relations around @xmath195 , @xmath196 and @xmath86 . ) to complete the calculation it remains to prove lemma [ lemmaa92a4 ] . we consider the polynomial @xmath197 for @xmath40 , @xmath198 and @xmath199 real . we denote by @xmath200 and @xmath201 the real and the imaginary part of @xmath202 respectively . they have degree @xmath15 and @xmath153 respectively in @xmath199 . suppose there exists a number @xmath203 $ ] such that @xmath204 complex solutions of the equation @xmath205 are on the same vertical line @xmath206 in the complex plane @xmath207 . this implies that the equations @xmath208 have @xmath204 common real solutions @xmath209 , @xmath210 , @xmath211 and @xmath212 . these solutions are non zero since the equation @xmath213 does not have any solution in @xmath214 $ ] . therefore , the equations @xmath215 where @xmath216 ( notice that @xmath199 divides @xmath201 , and @xmath217 is then a polynomial ) , also have @xmath209 , @xmath210 , @xmath211 and @xmath212 as common solutions . as @xmath218 has degree @xmath204 in @xmath199 , it follows that @xmath219 divides @xmath220 . therefore the remainder @xmath221 of @xmath222 by @xmath218 , as a polynomial in @xmath199 , must be identically zero for @xmath206 and @xmath223 ( of course , @xmath224 is written as @xmath225 where @xmath226 is a polynomial in @xmath40 , @xmath198 and @xmath199 while @xmath227 is a polynomial depending just on @xmath40 and @xmath198 ) . by an easy computation , we see that @xmath228 , where @xmath229 , @xmath230 , @xmath231 and @xmath232 are polynomials in @xmath40 and @xmath198 . thus , @xmath233 is a common real solution of the equations @xmath234 this implies that @xmath235 is a root of the resultant @xmath236 of @xmath229 and @xmath231 as polynomials in @xmath198 . note that the condition @xmath237 is necessary to have a real partner @xmath238 such that @xmath239 but it is not sufficient since the possible partner @xmath238 might be non real . there are @xmath153 real solutions @xmath240 of the equation @xmath241 in the interval @xmath214 $ ] . each of them gives a real number , say @xmath242 ( @xmath243 ) , such that @xmath244 is a solution of ( [ r ] ) . now , we have to check if these @xmath153 solutions give @xmath204 real roots of the polynomial @xmath245 . only the solution @xmath246 satisfies this requirement . therefore we can have one ( and only one ) overcrossing of @xmath204 complex roots . to check if it is the case , we look at the solutions @xmath70 of the equation , in @xmath70 , @xmath24 for values of @xmath40 near @xmath235 . ` maple ` actually gives an overcrossing . in this section , we prove that @xmath248 . as above , we use zariski van kampen s theorem with the pencil given by the vertical lines @xmath64 , @xmath65 . the singular lines of the pencil are the lines @xmath83 ( @xmath258 ) corresponding to these 8 roots . the lines @xmath85 and @xmath141 intersect the curve at its singular points , while all the other singular lines are tangent to the curve . see figure [ figa92a4narps ] . here we start with the generic line @xmath259 and we choose generators @xmath89 of @xmath260 as in figure [ figa92a4naeta4 ] , where @xmath91 is small enough . ( to determine the position of the complex roots of the equation @xmath261 one may use ( [ b0 ] ) and ( [ pp0 ] ) below . ) as above , to find the monodromy relations around the singular lines @xmath83 ( @xmath258 ) of the pencil , we fix a ` standard ' system of generators @xmath263 of the fundamental group @xmath264 , where each @xmath100 is a lasso ( oriented counter clockwise ) around @xmath101 and with base point @xmath161 . for @xmath265 , the tail of @xmath100 is a union of real segments and half circles around the exceptional parameters @xmath104 ( @xmath105 ) located in the real axis between the base point @xmath161 and @xmath101 , while its head is the circle @xmath106 . the lasso @xmath266 corresponding to the non real root @xmath267 has the form @xmath110 , where @xmath111 is the loop obtained by moving @xmath40 once on the circle @xmath122 , starting at @xmath268 , and @xmath114 the path obtained when @xmath40 makes half turn on the circle @xmath269 , from @xmath161 to @xmath102 , moves on the real axis from @xmath102 to @xmath115 , makes half turn on the circle @xmath116 , from @xmath115 to @xmath117 , moves on the real axis from @xmath117 to @xmath118 , makes half turn on the circle @xmath119 , from @xmath118 to @xmath120 , moves on the real axis from @xmath120 to @xmath270 , and finally moves in a straight line from @xmath270 to @xmath268 . the lasso @xmath108 is defined similarly from a loop @xmath111 and a path @xmath114 meeting at @xmath271 . when @xmath40 moves on the real axis from @xmath161 to @xmath162 , the @xmath92 s are deformed as in figure [ figa92a4naeta3 ] . the singular line @xmath164 is tangent to @xmath8 at one simple point , with intersection multiplicity 2 , so the monodromy relation around this line is simply given by @xmath272 now , when @xmath40 makes half turn counter clockwise on the circle @xmath169 , from @xmath162 to @xmath170 , and then moves on the real axis from @xmath170 to @xmath171 , the @xmath92 s are deformed as in figure [ figa92a4naeta2 ] . the singular line @xmath273 is tangent to the curve at two simple points , in both case with intersection multiplicity 2 . the monodromy relations around this line are given by @xmath274 in figure [ figa92a4naeta1 ] , we show how the @xmath92 s are deformed when @xmath40 makes half turn counter clockwise on the circle @xmath176 , from @xmath171 to @xmath177 , and then moves on the real axis from @xmath177 to @xmath178 . the monodromy relation around @xmath179 is a multiplicity 2 tangent relation : @xmath275 at @xmath262 , the generators are shown in figure [ figa92a4naeta4 ] . by ( [ rela92a4naeta3 ] ) , ( [ rela92a4naeta21 ] ) and ( [ rela92a4naeta22 ] ) , figure [ figa92a4naeta4 ] is the same as figure [ figa92a4naeta4bis ] . the singular line @xmath85 passes through the origin which a type @xmath37 singular point of the curve . at this point the curve has two branches @xmath148 and @xmath147 given by @xmath278 an easy computation shows that puiseux parametrizations of these branches are : @xmath279 the monodromy relations at @xmath146 are then given by @xmath280 note that , by ( [ rela92a4naeta1 ] ) , all of them are equivalent to @xmath281 at @xmath262 , the generators are as in figure [ figa92a4naeta4bis ] . now , when @xmath40 makes half turn counter clockwise on the circle @xmath269 , from @xmath161 to @xmath102 , then moves on the real axis from @xmath102 to @xmath115 , the @xmath92 s are deformed as in figure [ figa92a4naeta5 ] , where @xmath282 the singular line @xmath179 is tangent to the curve at two simple points , in both cases with intersection multiplicity 2 . the monodromy relations at @xmath130 are then given as follows : @xmath283 by ( [ rela92a4naeta1 ] ) and ( [ rela92a4nas1 ] ) , both of them are equivalent to @xmath284 in figure [ figa92a4naeta6 ] , we show how the generators are deformed when @xmath40 makes half turn counter clockwise on the circle @xmath116 , from @xmath115 to @xmath117 , and then moves on the real axis from @xmath117 to @xmath118 . the singular line @xmath141 passes through the points @xmath285 and @xmath286 which are both singularities of type @xmath34 . puiseux parametrizations of the curve at these points are given by : @xmath287 the monodromy relations at @xmath140 are then given by @xmath288 notice that ( [ rela92a4naeta61 ] ) is automatically satisfied , while ( [ rela92a4naeta62 ] ) can be written as @xmath289 or , equivalently , as @xmath290 indeed , using ( [ rela92a4nas2 ] ) under the form @xmath291 , the relation ( [ rela92a4naeta61 ] ) can be written as @xmath292 which is nothing but @xmath293 by ( [ rela92a4nas1 ] ) . by ( [ rela92a4nainfini ] ) this equality is always satisfied . the relation ( [ rela92a4naeta62 ] ) is written as @xmath294 which is the same as @xmath295 by ( [ rela92a4naeta52 ] ) . equivalently @xmath296 . by ( [ rela92a4naeta1 ] ) , @xmath151 can be eliminated so @xmath297 by ( [ rela92a4nainfini ] ) this is the same as @xmath298 , while ( [ rela92a4nainfini ] ) and ( [ rela92a4nas1 ] ) show that ( [ rela92a4nas5 ] ) is also the same as @xmath299 . figure [ figa92a4naeta6 ] shows the generators at @xmath145 . in figure [ figa92a4naeta7 ] ( respectively figure [ figa92a4naeta8 ] ) , we show how the @xmath92 s are deformed when @xmath40 makes half turn counter clockwise on the circle @xmath119 , from @xmath118 to @xmath120 , then moves on the real axis from @xmath120 to @xmath270 , and finally moves straight along the line @xmath303 from @xmath270 to @xmath268 ( respectively along the line @xmath304 from @xmath124 to @xmath271 ) , where @xmath305 ( in these figures we concentrate only on the generators which may give _ a priori _ some relations . ) the monodromy relations at @xmath186 and @xmath302 are multiplicity 2 tangent relations given by @xmath306 and @xmath307 respectively . in fact , these relations are automatically satisfied . indeed , the relation ( [ rela92a4naeta7 ] ) is written as @xmath308 but , by ( [ rela92a4naeta52 ] ) , we know that @xmath309 , so the relation above is always satisfied . now , using ( [ rela92a4naeta7 ] ) , the relation ( [ rela92a4naeta8 ] ) is written as @xmath310 which is equivalent to @xmath311 by ( [ rela92a4naeta52 ] ) . the latter is always satisfied , by ( [ rela92a4naeta1 ] ) . altogether , we have proved that the fundamental group @xmath49 is presented by the generators @xmath156 and @xmath157 and the relations ( [ rela92a4nainfini ] ) , ( [ rela92a4nas1 ] ) , ( [ rela92a4nas2 ] ) and ( [ rela92a4nas3 ] ) . the relation ( [ rela92a4nainfini ] ) can be written as @xmath314 this shows that ( [ rela92a4nas1 ] ) is equivalent to ( [ rela92a4nas3 ] ) . the relation ( [ rela92a4nas2 ] ) is automatically satisfied . indeed , by ( [ rela92a4nas3 ] ) , it is equivalent to @xmath315 but both sides are equal to @xmath277 , by ( [ rela92a4nas3 ] ) under the form ( [ rela92a4nas33 ] ) . hence , @xmath49 is simply presented by the generators @xmath156 and @xmath157 and the relations ( [ rela92a4nas3 ] ) and ( [ rela92a4nas4 ] ) . after the change @xmath316 and @xmath317 , the presentation is also given by @xmath318
the existence of alexander equivalent zariski pairs dealing with irreducible curves of degree 6 was proved by a. degtyarev . however , up to now , no explicit example of such a pair was available ( only the existence was known ) . in this paper , we construct the first concrete example . [ multiblock footnote omitted ]
neutrino physics can anticipate an era of higher precision measurements with the upcoming generation of neutrino experiments . in the past , measurements have shown that the mixing pattern of lepton sector is quite different from that of quark sector . in the pontecorvo - maki - nakagawa - sakata ( pmns ) parameterization @xcite for lepton mixing there are two large mixing angles . the atmospheric mixing angle @xmath1 is almost maximal while the solar mixing angle @xmath2 is also large and the reactor mixing angle @xmath3 nearly vanishes . the recent results are summarized in table[tab : data ] . we can see that the uncertainties in mixing angles are not particularly small . in most measurements there is roughly a @xmath4 deviation at @xmath5 confidence level . c|c|c|c|c|c & & & & & + & @xmath6 & @xmath7 & @xmath8 & @xmath9 & @xmath10 + & & & & & + & & & & & + central value & @xmath11 & @xmath12 & @xmath13 ( @xmath14 ) & @xmath15 ( @xmath16 ) & @xmath17 ( @xmath18 ) + & & & & & + @xmath19 range & @xmath20 & @xmath21 & @xmath22 & @xmath23 & @xmath24 + & & & ( @xmath25 ) & ( @xmath26 ) & ( @xmath27 ) + the mixing matrix which incorporates these angles and diagonalizes the mass matrix @xmath28 via @xmath29 is given by @xmath30 with @xmath31 for @xmath32 . @xmath33 is the dirac phase and we have neglected majorana phases . from table i we see that a good first approximation is to take @xmath34 which gives @xmath35 using this as a starting point , we wish to investigate whether there is an underlying symmetry @xmath36 of the neutrino mass matrix . this matrix must satisfy @xmath37=0 $ ] , or @xmath38 . given a @xmath36 , the transformation @xmath39 also diagonalizes @xmath40 . but @xmath41 is unique except for phases . this can be seen by supposing that @xmath42 is a unitary matrix such that @xmath43 also diagonalizes @xmath40 . for this to be true , @xmath42 must satisfy @xmath44 and this implies that @xmath45 . since @xmath39 diagonalizes @xmath40 , it must have the form @xmath46 , where @xmath42 has @xmath47 or @xmath48 diagonal elements . thus @xmath49 there are only eight possible combinations for the elements of @xmath42 . two of these are the unit matrix and its negative , both of which define @xmath36 as a multiple of the identity . of the remaining six , three have two entries of @xmath50 and one of @xmath48 , while the other three have two entries of @xmath48 and one of @xmath50 . these diagonal matrices differ by an overall minus sign , so only one of the two types is independent . if we choose the three with one entry of @xmath47 and two entries of @xmath48 ( @xmath51 ) , then it is easy to see that multiplying any pair of these diagonal matrices will result in the remaining matrix . hence , in reality , only two of these matrices are independent and both represent a @xmath52 symmetry . since the independent @xmath36 s commute , the horizontal symmetry of lepton mixing is @xmath53 if neutrinos are majorana fermions @xcite . a representation of @xmath36 can be obtained using @xmath54 since the eigenvalue @xmath47 can occur in three places , there are three symmetry matrices @xmath36 @xmath55 and @xmath56 where the subscript @xmath57 on @xmath58 denotes the component of @xmath59 that is @xmath50 . @xmath60 gives @xmath61 symmetry @xcite while @xmath62 is symmetric and commutes with @xmath60 . for simplicity we can parameterize the solar mixing angle @xmath63 as @xmath64 then the mixing matrix ( [ eq : unus ] ) takes the form @xcite @xmath65 consequently , the symmetry transformation matrix @xmath66 can be reexpressed in terms of @xmath67 @xmath68 although we can derive " a generalized form of @xmath62 symmetry transformation matrix ( [ g1 ] ) given the mixing matrix ( [ eq : uk ] ) , this relationship can not be reversed . the mixing matrix @xmath41 can not be uniquely determined solely by @xmath62 due to the fact that @xmath62 has degenerate eigenvalues . invariance under @xmath60 requires @xmath69 , @xmath70 , but invariance under @xmath62 does not , so below we assume that the neutrino mass matrix is invariant under @xmath62 , not only in the approximation @xmath71 , but for general values of all the mixing angles . in the next section we use this assumption in the form of eq.([g1 ] ) with general values of @xmath67 to derive relations among the mixing angles . in sec.3 we compare our results with the experimental values and in sec.4 we summarize . in this section we show explicitly the consequences of generalized @xmath62 symmetry . only two mass square differences have been measured and the neutrino s mass scale has not been determined by experiments . it is possible that one of the mass eigenvalues vanishes . this is also theoretically motivated by minimal seesaw model @xcite . we will explore the joint consequences of one vanishing mass eigenvalue and @xmath62 invariance . for simplicity we will postpone discussion of @xmath72 phases to a later article @xcite . if the neutrinos are majorana fermions , their mass matrix must be symmetric . we will consider the case that there are three generations of light neutrinos . then , the most general form of the neutrino mass matrix can be parameterized as @xmath73 which has six independent matrix elements . we assume @xmath28 is invariant under the @xmath62 symmetry transformation , @xmath74 with the help of ( [ g1 ] ) and ( [ m ] ) , eq.([gmg ] ) gives two conditions on the neutrino mass matrix elements of ( [ m ] ) @xcite , @xmath75 [ bcdak ] if there is a vanishing mass eigenvalue @xmath76 the corresponding mass eigenstate , which can be denoted as @xmath77 , must satisfy @xmath78 if we assume @xmath79 then we get three equations @xmath80 [ eq:0mass ] where @xmath81 . thus we have two sets of conditions , ( [ bcdak ] ) from @xmath62 invariance , and ( [ eq:0mass ] ) from the vanishing mass eigenvalue . from the relations ( [ eq:0mass ] ) we can express the matrix element @xmath82 in terms of @xmath83 , @xmath84 and @xmath85 @xmath86 now let us use these in the @xmath0 relations . eq.([bcdak1 ] ) and eq.([bcdak2 ] ) give @xmath87 d=0\ , . \label{c1c2d}\end{aligned}\ ] ] the above two relations can be reexpressed in terms of only two matrix elements , @xmath85 and @xmath84 or @xmath83 respectively @xmath88\,&=&\,0\,,\label{c2d } \\ ( \rho+\sigma - k)[(\rho\,k+1)c_1+(\sigma\,k+1)d]\,&=&\,0\ , . \label{c1d}\end{aligned}\ ] ] the mass eigenvalues that are nonzero are given by @xmath89\,,\ ] ] where we have used ( [ bcd ] ) and ( [ bdc ] ) and ( [ aa ] ) . from ( [ c2d ] ) and ( [ c1d ] ) it is obvious that one possible solution to the equations for @xmath90 is @xmath91 this makes @xmath92 and consequently @xmath93 given above would also be zero . since the experimental data shows that two mass square differences between the three neutrino mass eigenvalues are nonzero , we need at least two masses nonzero in order to have two oscillation lengths . a second solution of ( [ c2d ] ) , ( [ c1d ] ) is @xmath94 but then the three relations in ( [ eq:0mass ] ) simply reproduce the conditions ( [ bcdak ] ) . so the conclusion is that we must have @xmath95 and the conditions ( [ c2c1d ] ) or ( [ c1c2d ] ) reduce to an equation for @xmath96 @xmath97 this relation represents the constraint from @xmath62 invariance which was originally expressed as ( [ bcdak ] ) where there were two independent relations . using ( [ rsk ] ) , these two relations are satisfied simultaneously and reduce to a single constraint ( [ sccd ] ) . the condition ( [ rsk ] ) can also be substituted into ( [ abb ] ) , ( [ bcd ] ) , and ( [ bdc ] ) to give @xmath98 respectively . these three relations are a manifestation of vanishing mass eigenvalue . we can set these equations for @xmath96 equal to get relations among the matrix elements @xmath99 in terms of the parameter @xmath67 . not all of these equations are independent but two different relations are possible : @xmath100 [ eq : krel ] in the next subsection we will write @xmath99 in terms of the mixing angles and thereby get two relations among the mixing angles , again involving @xmath67 . using @xmath101 from eq.([utot ] ) in @xmath102 and comparing with ( [ m ] ) we get @xcite @xmath103 m_1 - c_x [ s_s c_s c_a + s_x s_a s_s^2 ] m_2 + c_x s_x s_a m_3 \label{vb1}\\ b_2 & = & c_x [ s_s c_s s_a + s_x c_a c_s^2 ] m_1 - c_x [ s_s c_s s_a - s_x c_a s_s^2 ] m_2 - s_x c_x c_a m_3 \label{vb2}\\ c_1 & = & ( s_s c_a - s_x c_s s_a)^2 m_1 + ( c_s c_a + s_x s_s s_a)^2 m_2 + c_x^2 s_a^2 m_3 \label{vc1 } \\ c_2 & = & ( s_s s_a + s_x c_s c_a)^2 m_1 + ( c_s s_a - s_x s_s c_a)^2 m_2 + c_x^2 c_a^2 m_3 \label{vc2 } \\ d & = & ( s_s s_a + s_x c_s c_a)(s_s c_a - s_x c_s s_a ) m_1 + ( c_s s_a - s_x s_s c_a)(c_s c_a + s_x s_s s_a ) m_2 - c_x^2 s_a c_a m_3 \label{vd}\end{aligned}\ ] ] [ eq : reconstruction ] where , as mentioned above , we have deferred consideration of @xmath72 violation to a later article . the mass eigenvalues can be further parameterized in terms of experimentally measured mass square differences : @xmath104 , @xmath105 and @xmath106 for inverted mass hierarchy and @xmath107 , @xmath108 and @xmath109 for normal mass hierarchy where @xmath110 and @xmath111 which is positive . to get relations between mixing angles we can substitute ( [ eq : reconstruction ] ) into ( [ eq : krel ] ) which gives @xmath112 m_1 m_2 & = & 0\,,\\ s_a c_x \left [ c_x ( c_a - s_a ) + k s_x \right ] m_1 m_2 & = & 0\,.\end{aligned}\ ] ] where we have assumed the mass hierarchy is inverted with vanishing @xmath113 and nonzero @xmath114 and @xmath115 , while @xmath116 , @xmath117 and @xmath118 are also nonzero . the only possible solution is @xmath119 where the last factor comes from expanding the two mixing angles @xmath120 and @xmath121 around the approximations @xmath122 and @xmath123 , @xmath124 with this solution for @xmath67 and the reconstructed mass matrix elements ( [ eq : reconstruction ] ) substituted back into ( [ abbk ] ) we find @xmath125 since @xmath126 is quite small , @xmath96 should be very large according to ( [ r2 ] ) . we still have the condition from @xmath0 , eq.([sccd ] ) . together with ( [ r2 ] ) and ( [ eq : reconstruction ] ) as well as ( [ r1 ] ) it gives @xmath127 there are two possible solutions @xmath128 or @xmath129 these relations between mixing angles can be used to predict the not well measured @xmath121 in terms of the solar and atmospheric mixing angles . for example ( [ r3a ] ) gives @xmath130 since @xmath121 is the focus of next generation of neutrino experiments , we use ( [ eq : sx - b ] ) to estimate its value . the scatter plot is shown in fig . [ fig : thetax ] . a scatter plot based on ( [ r3p ] ) would look similar with a steeper slope for the points . in terms of @xmath120 and @xmath63 at the 90% c.l . the vertical solid line denotes the experimentally measured central value of the atmospheric mixing angle @xmath120.,width=288 ] another way of expressing the results is to write all of the mixing angles in terms of the parameters @xmath96 and @xmath67 . using @xmath131 , this gives @xmath132 [ eq : s2asx - ih ] note that these equations are all unchanged under @xmath133 so only the absolute value of @xmath67 can be determined . the solutions ( [ r3p ] ) or ( [ r3a ] ) , which give ( [ s2s2 ] ) or ( [ s2s ] ) , are identical in the following sense - oscillation experiments measure @xmath134 and thus ca nt distinguish between @xmath135 and @xmath136 . further , @xmath137 , so a fit with ( [ r3p ] ) and @xmath63 assumed greater than @xmath138 is identical to one with ( [ r3a ] ) and @xmath63 assumed less than @xmath138 . having noted this we will proceed to fit both ( [ s2s2 ] ) and ( [ s2s ] ) with @xmath139 . using eqs.([s2a ] ) , ( [ s2x ] ) , and ( [ s2s2 ] ) , the fit to the data from ref.@xcite gives @xmath140 , @xmath141 and @xmath142 . at the minimum values of @xmath143 and @xmath144 , @xmath145 , @xmath146 and @xmath147 . the @xmath148 and @xmath149 confidence contours are shown in fig.[abc ] . ) , ( [ s2x ] ) , and ( [ s2s2 ] ) are shown in red and orange , respectively . the ( black ) dot indicates the @xmath150 minimum.,width=264 ] the distributions of this set of mixing angles are obtained from the likelihood distribution @xmath151 where @xmath82 is a normalization constant , using @xmath152 where @xmath153 is one of the functions on the righthand side of eqs.([eq : s2asx - ih ] ) . the results are shown in figs.([distna ] ) . obtained using eqs.([s2a ] ) , ( [ s2x ] ) , and ( [ s2s2 ] ) are shown.,title="fig:",width=220 ] obtained using eqs.([s2a ] ) , ( [ s2x ] ) , and ( [ s2s2 ] ) are shown.,title="fig:",width=220 ] obtained using eqs.([s2a ] ) , ( [ s2x ] ) , and ( [ s2s2 ] ) are shown.,title="fig:",width=220 ] as would be expected , none of the distributions is exactly gaussian . the largest contribution to the minimum @xmath150 is associated with @xmath154 and the influence of terms beyond the quadratic expansion of @xmath155 can be seen in the shape of this distribution . if we use eqs.([s2a ] ) , ( [ s2x ] ) and ( [ s2s ] ) , the fit to the data has two local minima . the lowest of these gives @xmath156 , @xmath157 and @xmath158 . at the minimum values of @xmath143 and @xmath144 , @xmath159 , @xmath160 and @xmath161 . at the other minimum , where @xmath162 , @xmath163 and @xmath164 are slightly different , but @xmath165 this is reflected in the individual mixing angle distributions . the confidence contours for this case are shown in fig.[abd ] . ) , ( [ s2x ] ) and ( [ s2s ] ) are shown in red and orange , respectively . the ( black ) dot indicates the @xmath150 minimum.,width=264 ] the distributions of this set of mixing angles are shown in figs.([distnb ] ) . here , too , the largest contribution to the minimum @xmath150 is associated with @xmath154 and the effect of the second local minimum this is reflected in the distortion on the high side of the probability distribution . obtained using eqs.([s2a ] ) , ( [ s2x ] ) and ( [ s2s ] ) are shown.,title="fig:",width=220 ] obtained using eqs.([s2a ] ) , ( [ s2x ] ) and ( [ s2s ] ) are shown.,title="fig:",width=220 ] obtained using eqs.([s2a ] ) , ( [ s2x ] ) and ( [ s2s ] ) are shown.,title="fig:",width=220 ] alternately we can fit for @xmath67 using the values of @xmath166 and @xmath163 from ref.@xcite but replace @xmath164 with the value for @xmath167 published by the minos collaboration @xcite . they report @xmath168 for inverted hierarchy and , for normal hierarchy , @xmath169 . from the inverted hierarchy result we get @xmath170 with a @xmath150 of @xmath171 or @xmath172 with a @xmath150 of @xmath173 . this was all for inverted hierarchy . normal hierarchy , with @xmath174 equal to zero , gives , after a lot of work , exactly the results of inverted hierarchy , ( [ r1 ] ) , ( [ r3p ] ) , ( [ r3a ] ) . the parameter @xmath96 is a different function than ( [ r2 ] ) , @xmath175 but this just amounts to a reparameterization of ( [ eq : s2asx - ih ] ) with no physical consequence . using the minos number for normal hierarchy we find the same values , including the errors , for @xmath143 as for the inverted hierarchy minos number . the @xmath150 values are smaller at @xmath176 or @xmath177 . with either minos value and for either value of @xmath143 the fitted value of @xmath163 is stable at @xmath13 , the fitted value of @xmath166 varies only slightly from @xmath178 for the larger @xmath143 to @xmath179 for the smaller value , but @xmath164 is less than @xmath180 for the larger @xmath143 but equal to @xmath181 for the smaller . a hidden @xmath0 symmetry , as given by eq.([gmg ] ) , results in only two possible sets of conditions on the neutrino mixing angles . assuming @xmath139 then either @xmath182 with confidence contours shown in fig.[abc ] , or @xmath183 with the confidence contours shown in fig.[abd ] . sfg was supported by the china scholarship council ( csc ) . dad and sfg were supported in part by the u. s. department of energy under grant no . de - fg03 - 93er40757 . wwr was supported in part by the national science foundation under grant phy-0555544 . dad is a member of the center for particles and fields and the texas cosmology center . it is always a pleasure to thank sacha kopp and karol lang for helpful discussions regarding the minos data . wwr thanks jim linnemann for an informative conversation about the procedures used in sec.(3 ) . 99 b. pontecorvo , sov . phys . jetp * 6 * , 429 ( 1957 ) [ zh . eksp . * 33 * , 549 ( 1957 ) ] ; z. maki , m. nakagawa and s. sakata , prog . * 28 * , 870 ( 1962 ) . g. l. fogli , e. lisi , a. marrone , a. palazzo and a. m. rotunno , phys . * 101 * , 141801 ( 2008 ) [ arxiv:0806.2649 [ hep - ph ] ] and g. l. fogli , e. lisi , a. marrone , a. palazzo and a. m. rotunno , arxiv:0809.2936 [ hep - ph ] ; g. l. fogli _ et al . _ , phys . d * 78 * , 033010 ( 2008 ) [ arxiv:0805.2517 [ hep - ph ] ] and references therein . see , for instance , the experimental reports at xxiii international conferences on `` neutrino physics and astrophysics '' ( neutrino 2008 ) , christchurch , new zealand , may 25 - 31 , 2008 . web link : http://www2.phys.canterbury.ac.nz/~jaa53 c. s. lam , phys . rev . * d74 * , 113004 ( 2006 ) . c. s. lam , phys . d * 78 * , 073015 ( 2008 ) [ arxiv:0809.1185 [ hep - ph ] ] . w. grimus , l. lavoura , p. o. ludl , j. phys . g * g36 * , 115007 ( 2009 ) . s.- f. ge , h. j. he and f. r. yin , jcap * 1005 * , 017 ( 2010 ) [ arxiv:1001.0940 [ hep - ph ] ] . p. f. harrison and w. g. scott , phys . b * 547 * , 219 ( 2002 ) [ arxiv : hep - ph/0210197 ] , r. n. mohapatra , jhep * 0410 * , 027 ( 2004 ) [ arxiv : hep - ph/0408187 ] , r. n. mohapatra , s. nasri and h. b. yu , phys . b * 615 * , 231 ( 2005 ) [ arxiv : hep - ph/0502026 ] , t. kitabayashi and m. yasue , phys . b * 621 * , 133 ( 2005 ) [ arxiv : hep - ph/0504212 ] , r. n. mohapatra and w. rodejohann , phys . d * 72 * , 053001 ( 2005 ) [ arxiv : hep - ph/0507312 ] , z. z. xing , h. zhang and s. zhou , phys . b * 641 * , 189 ( 2006 ) [ arxiv : hep - ph/0607091 ] . p. h. frampton , s. l. glashow and t. yanagida , phys . b * 548 * , 119 ( 2002 ) [ arxiv : hep - ph/0208157 ] ; m. raidal and a. strumia , phys . b * 553 * , 72 ( 2003 ) [ arxiv : hep - ph/0210021 ] . s.- f. ge , d. a. dicus , and w. w. repko , `` a @xmath184 symmetry prediction for the dirac @xmath72 phase of neutrino mixing , '' in preparation . these conditions for one vanishing neutrino mass and @xmath185 were derived in : d. a. dicus , s.- f. ge and w. w. repko , phys . d*82 * , 033005 ( 2010 ) arxiv:1004.3266 [ hep - ph ] . v. barger , d. a. dicus , h. j. he and t. j. li , phys . b * 583 * , 173 ( 2004 ) [ arxiv : hep - ph/0310278 ] . p. adamson _ et . lett . * 103 * , 261802 - 1 ( 2009 ) .
we explore the consequences of the neutrino mass matrix having a hidden @xmath0 symmetry and one zero eigenvalue . when implemented , these two conditions give relations among the mixing angles . in addition , fitting these relations to the existing oscillation data allows limits to be placed on the parameter of the symmetry .
the practical significance of the discrete logarithm problem on ordinary binary elliptic curves ( see , e.g. , @xcite ) renders these groups a natural target for implementing shor s algorithm @xcite . to implement an actual discrete logarithm computation , efficient quantum circuits to implement the pertinent curve arithmetic are needed , and a number of authors have explored circuits to implement the relevant elliptic curve operations @xcite . when considering a complete implementation of shor s algorithm in such a group , maslov et al.s proposal in @xcite shows that a quadratic depth circuit is sufficient . the reason for the quadratic depth is essentially two - fold : a double - and - add computation to compute the relevant scalar multiplications in shor s algorithm and a finite field inversion are the dominating operations . as shown in @xcite , inversion in @xmath2 can be implemented in depth @xmath5 , and so one may hope that the quadratic depth bound can indeed be overcome . [ [ our - contribution . ] ] our contribution . + + + + + + + + + + + + + + + + + below we show that an appropriate organization of the scalar multiplication(s ) in shor s algorithm in combination with an improved @xmath2-arithmetic enables a solution to the discrete logarithm problem on ordinary binary elliptic curves in depth @xmath1 . to implement the necessary group operations we use complete binary edwards curves as described in @xcite . [ [ structure - of - the - paper . ] ] structure of the paper . + + + + + + + + + + + + + + + + + + + + + + + in the next section we briefly review some background on elliptic curves and shor s algorithm . in particular we recall the the definition of binary edwards curves as needed for the main part of the paper . section [ sec : parallel ] details how with this curve representation the addition of any two curve points can be implemented in logarithmic depth . thereafter we discuss different options to organize the scalar multiplications in shor s algorithm , including a tree - based approach with polylogarithmic depth . after addressing the technical point of deriving a unique representation of group elements , in section [ sec : theend ] we establish our main result . this section reviews some known results on ordinary binary elliptic curves and on computing discrete logarithms with shor s algorithm . for @xmath6 a positive integer , we denote by @xmath7 a finite field of size @xmath8for a cryptographic application , e.g. , a digital signature scheme , a typical choice would be @xmath9 or @xmath10 @xcite . to represent elements in @xmath7 we use a polynomial basis representation . in other words we fix an irreducible polynomial @xmath11 $ ] with coefficients in the integers modulo @xmath12 and identify @xmath7 with the quotient @xmath13/(p)$ ] , so that each @xmath14 has a unique expression of the form @xmath15 with @xmath16 . using a _ short weierstrass form _ , each ordinary binary elliptic curve can be represented by a polynomial equation @xmath17 where @xmath18 with @xmath19 ( * ? ? ? * chapters 13.1.4 and 13.1.5 ) . more precisely , the elliptic curve represented by equation consists of the points @xmath20 , where @xmath21 is the unique projective point that is obtained when homogenizing equation . hasse s bound implies that the size of @xmath22 differs from @xmath8 by no more than @xmath23 , and the subgroups of @xmath22 considered in cryptographic applications typically have a very small cofactor . hence @xmath24 is a natural parameter to measure the complexity of a quantum circuit to solve the discrete logarithm problem in a @xmath22 . the set @xmath22 has a group structure , but implementing this group law directly comes at a certain inconvenience : case distinctions have to be made , which require the implementation of a ( nested ) if - then - else statement ( cf . the discussion in @xcite ) . to avoid this issue , subsequently we use _ complete binary edwards curves _ as introduced by bernstein et al . for @xmath25 each ordinary elliptic curve is birationally equivalent to a complete binary edwards curve , and we can represent such a curve by an equation @xmath26 with @xmath27 being non - zero , @xmath28 and @xmath29 . can equivalently be expressed as @xmath30 . ] by @xmath31 we denote the points in @xmath32 satisfying equation . the group law on @xmath31 is given by the formula @xmath33 where @xmath34 can be arbitrary curve points including the identity element @xmath35 . to derive efficient addition formula one can ( similarly as for a weierstrass form ) pass to projective coordinates . bernstein et al . @xcite show that from projective representations @xmath36 , @xmath37 of two points one can derive a projective representation @xmath38 of their sum by means of 21 multiplications in @xmath2 , four multiplications by one of the constant @xmath39 , @xmath40 , one squaring and 15 additions in @xmath2 . @xmath41 from an asymptotic point of view , it suffices to observe that the number of field operations is constant . it is not necessary , however , to perform these field operations sequentially , and @xcite suggest some parallelization , establishing the following upper bound for the depth of a point addition circuit , where @xmath42 stands for the depth of an @xmath2-multiplier @xmath43 . [ prop : edadd ] let @xmath44 and @xmath45 be projective representations of two ( not necessarily different ) points @xmath46 . then the addition map @xmath47 where @xmath48 is a projective representation of @xmath49 , can be implemented in depth @xmath50 . from @xcite it follows that we can choose @xmath51 , and in section [ sec : fieldmult ] we will show that through a suitable use of trees @xmath42 can be chosen to be of logarithmic depth . as the number of field operations to add to curve points is constant , this establishes immediately the existence of a logarithmic depth circuit for point addition . to optimize the circuit depth , we can exploit the bound from proposition [ prop : edadd ] : looking into the proof of ( * ? ? ? * proposition 3.3 ) , one recognizes that the term @xmath52 occurring as argument of the maximum in proposition [ prop : edadd ] describes the multiplication of a binary @xmath53-matrix with a binary vector . in the next section we will see that such a multiplication can be realized in logarithmic depth as well . a technical issue that we address in section [ sec : lowdepthdivision ] is the derivation of the unique ( affine ) representation from a projective representation of a curve point : the natural way to realize this is by means of an inversion in @xmath2 , but for none of the division circuits described in @xcite a polylogarithmic depth bound is available . we modify the construction in @xcite to achieve polylogarithmic depth . for our discussion we assume that a generator @xmath54 of a cyclic subgroup of @xmath31 is fixed and the order @xmath55 of this group generator is known . moreover , we assume that a group element @xmath56 in the subgroup generated by @xmath57 is fixed ; our goal is to find the unique integer @xmath58 such that @xmath59 . the algorithm proceeds as follows . first , two registers of length @xmath60 qubits can be represented with @xmath60 bits . ] are created and each qubit is initialized in the @xmath61 state . then a hadamard transform @xmath62 is applied to each qubit , resulting in the state @xmath63 . next , conditioned on the content of the register holding the label @xmath64 or @xmath65 , we add the corresponding multiple of @xmath57 and @xmath56 , respectively , i.e. , we implement the map @xmath66 hereafter , the third register is discarded and a quantum fourier transform @xmath67 on @xmath68 qubits is computed . finally , the state of the first two registers which hold a total of @xmath68 qubits is measured . as shown in @xcite , the factor @xmath69 can be computed from this measurement data via classical post - processing . the corresponding quantum circuit is shown in figure [ fig : shorcircuit ] . in the following sections , we will be concerned with parallelizing the parts of the circuit in this figure . in section [ sec : lowdepthdivision ] we will address the problem of having a non - unique representation of curve points , as the above description of shor s algorithm implicitly assumes group elements to have a unique representation . shor s algorithm to compute the discrete logarithm in the subgroup of an elliptic curve generated by a point @xmath57 . the input to the problem is a point @xmath56 , and the task is to find @xmath58 such that @xmath70 . the circuit naturally decomposes into three parts , namely ( i ) the hadamard layer on the left , ( ii ) a double scalar multiplication ( in this figure implemented as a cascade of point additions ) , and ( iii ) the quantum fourier transform @xmath71 at the end . we show that each of these parts can be implemented in a circuit depth of @xmath1 to obtain the main result of this paper.,scaledwidth=90.0% ] to reduce the circuit depth , we parallelize shor s algorithm on two different levels : ( i ) the computation of @xmath2-multiplications is parallelized and ( ii ) the computation of the scalar products @xmath72 is parallelized . a simple observation that will be useful is that we can implement the map @xmath73 in depth @xmath74 by arranging @xmath75 cnot gates as a tree . figure [ fig : multifanout ] , which derives from ( * ? ? ? * figure 2 ) , shows such a ` multi - fan - out cnot with @xmath61-input ' for the case @xmath76 . we note that in general such a tree is not functionally equivalent to a cnot with fan - out greater than @xmath77 , but for the case of a @xmath61-input this equivalence holds , and for our purposes this is the only case needed . as starting point to implement multiplication in @xmath2 we use the circuit proposed by maslov et al . in @xcite which builds on a classical mastrovito multiplier @xcite . this construction reduces the task of multiplying two elements in @xmath13/(p)$ ] to implementing a quantum circuit that evaluates two matrix - vector multiplications with a toeplitz matrix , one matrix - vector multiplication with a matrix that depends only on the polynomial @xmath78 , and an addition in @xmath79 . more specifically , the coefficients @xmath80 of the product @xmath81 are obtained as follows , where @xmath82 is independent of the specific field elements to be multiplied ; the matrix @xmath83 depends only on the irreducible polynomial @xmath78 defining the underlying finite field @xmath13/(p)$ ] : @xmath84 to implement the multiplications of @xmath87 and @xmath88 with @xmath89 respectively , we first observe that considering both the computation of @xmath85 and @xmath86 combined each coefficient @xmath90 ( @xmath91 ) occurs in exactly @xmath24 products of the form @xmath92 . similarly , each coefficient @xmath93 ( @xmath94 ) occurs in a total of exactly @xmath24 products . we want to compute _ all _ of these @xmath95-products in parallel . so we ensure that @xmath24 ` copies ' of each of @xmath96 are available , using a ` multi - fan - out cnot with @xmath61-input ' for each @xmath97 . as the cnot trees for @xmath97 and @xmath93 with @xmath98 operate on disjoint wires , they can be executed in parallel and implemented in depth @xmath3 . analogously , using a ` multi - fan - out cnot with @xmath61-input ' for each of @xmath99 we can in depth @xmath3 and in parallel to the trees for copying the @xmath97-values provide @xmath24 ` copies ' of each of @xmath99 . having , at the cost of @xmath0 qubits , all these copies at our disposal , we can now , in depth @xmath77 , compute all products @xmath100 that are necessary to find @xmath85 and @xmath86 in parallel , using @xmath101 toffoli gates . having evaluated all these products we can simply compute each entry of @xmath85 and @xmath86 by using a a depth @xmath3 addition tree for each entry of these two vectors . only cnot gates ( and no further ancillae ) are needed for this . figure [ fig : ubandlb ] shows an example for the case @xmath102 , i.e. , we have @xmath103 in this case all occurring multiplications can be evaluated in depth @xmath104 , and the final addition trees reduce to a single cnot gate to compute the ` last ' entry @xmath105 of @xmath85 . from equation , we see that the vector @xmath106 needs to be multiplied from the left with the fixed @xmath107-matrix @xmath83 . writing @xmath108 for the hamming weight of the @xmath109 column of @xmath83 and denoting by @xmath110 the @xmath109 entry of @xmath111 , we first create @xmath112 copies of @xmath110 ( @xmath113 ) , requiring @xmath0 qubits . for this we use again ` multi - fan - out cnot gates with @xmath61-input ' that operate in parallel and can be realized in depth @xmath3 . this allows us to compute all @xmath24 entries of @xmath114 in parallel : for each entry of the result we can use an @xmath3-depth addition tree that computes the scalar product of the corresponding row of @xmath83 with @xmath111 . as the matrix @xmath83 is fixed , this can be done by means of cnot gates . figure [ fig : mmult ] shows a ` worst case tree ' of depth @xmath115 for the case @xmath116 : multiplying a matrix row consisting entirely of @xmath77s with @xmath117 . finally , to complete the evaluation of equation , we add the binary vector @xmath118 to @xmath119 by means of @xmath24 cnot gates that operate in parallel . with all involved steps computing @xmath85 and @xmath86 , finding @xmath120 , and determining @xmath121being realizable in depth @xmath3 we obtain the following result . [ prop : gf2nlowdepthmult ] there is a polynomial - size quantum circuit of depth @xmath3 which on input polynomial basis representations of @xmath122 computes a polynomial basis representation of the product @xmath123 . the above - described technique to multiply a vector with the fixed matrix @xmath83 can also be used to implement other matrix - vector multiplications : given a binary @xmath53 matrix @xmath124 and a vector @xmath125 , we first use @xmath24 ` multi - fan - out cnot with @xmath61-input ' to create @xmath24 copies of @xmath126 , investing @xmath0 qubits . handling all entries of @xmath126 in parallel , this can be done in depth @xmath3 . hereafter we can use an addition tree to compute the necessary @xmath24 scalar products in parallel , just as in the discussion of the matrix @xmath83 . we can apply this observation to the matrix multiplications occurring in the proof of proposition [ prop : edadd ] given in @xcite , which replaces the argument @xmath52 of @xmath127 with a function in @xmath3 . in combination with theorem [ prop : gf2nlowdepthmult ] , we obtain the following . [ cor : lowdepthadd ] using projective coordinates , a projective representation of the sum of two points on a complete binary edwards curve @xmath31 can be computed in depth @xmath3 . an essential part of an implementation of shor s algorithm is a circuit which on input ( binary representations of ) @xmath128 and @xmath129 computes a ( unique ) representation of @xmath72because of hasse s bound , we can assume that @xmath64 and @xmath65 are represented with ( at most ) @xmath60 qubits each . the approach taken in @xcite can be seen as implementation of a right - to - left version of the double - and add - algorithm and then continue with the necessary additions of the points @xmath130 ; this change of addition order does not affect the circuit depth . ] : r i=0n 1 rr+2^ip rr+2^iq |return| r applying corollary [ cor : lowdepthadd ] , each point addition requires circuit depth @xmath3 , and so we immediately obtain an @xmath5-depth circuit to compute @xmath131 . a feature of this particular strategy is that all points @xmath132 and @xmath133 can be precomputed classically , and in all adders involved one argument is constant . this can be exploited to simplify the addition circuits , but for a realistic value of @xmath24 several hundred different addition circuits are involved . a left - to - right formulation of the double - and - add algorithm offers an alternative that involves only three types of circuits , but this uniformity comes at the cost of a general doubling circuit : r rr+p rr+q i = n-10 - 1 r2r rr+p rr+q |return| r this algorithm processes @xmath64 and @xmath65 simultaneously , so that the total number of doublings is @xmath24 ( rather than @xmath52)the latter technique goes back to straus @xcite and is also known as _ shamir s trick _ ( cf . @xcite ) . realizing point doubling and addition of @xmath57 respectively @xmath56 in depth @xmath3 , this yields again an @xmath5-depth circuit to find @xmath131 . to find @xmath131 with a quantum circuit of smaller depth , we parallelize the computation of @xmath131 . to simplify the description we assume that the length @xmath60 of the binary representation of @xmath64 and of @xmath65 is a power of @xmath12if this is not the case , the binary expansions can be padded with @xmath134s accordingly . ( r_0^(_2(n+1)), ,r_n^(_2(n+1)))(k_02 ^ 0p, ,k_n2^np ) ( s_0^(_2(n+1)), ,s_n^(_2(n+1)))(_02 ^ 0s, ,_n2^ns ) i=_2(n+1)-1 0 -1 ( r^(i)_0, ,r^(i)_j, ,r^(i)_2^i-1)=(r^(i+1)_0+r_1^(i+1), ,r^(i+1)_2j+r^(i+1)_2j+1 , , r^(i+1)_2^i+1 - 2+r^(i+1)_2^i+1 - 1 ) ( s^(i)_0, ,s^(i)_j, ,s^(i)_2^i-1)=(s^(i+1)_0+s_1^(i+1), ,s^(i+1)_2j+s^(i+1)_2j+1 , , s^(i+1)_2^i+1 - 2+s^(i+1)_2^i+1 - 1 ) |return| r_0^(0)+s_0^(0 ) the summation computed here is essentially the same as in the first sequential version of the double - and - add algorithm we discussed , therewith following the approach in @xcite . like maslov et al . we parallelize the execution of gates that operate on disjoint wires , but we process more than one point @xmath132 respectively @xmath135 at a time . the ( affine ) points @xmath136 and @xmath137 can be precomputed classically , and the leaves can be initialized with a sequence of cnot gates , conditioned on the individual bits of the binary representation of @xmath64 and @xmath65 . using ` multi - fan - out cnots with @xmath61-input ' , we can create @xmath52 copies of the binary representation of @xmath64 and @xmath52 copies of the binary representation of @xmath65 in depth @xmath3 . each leaf corresponds to a separate quantum register of linear length , initialized with @xmath138 , and hence the cnot gates for different points operate on disjoint wires . having @xmath52 copies of each coefficient @xmath139 and @xmath140 available , we can copy all @xmath52 qubits representing a point @xmath132 or @xmath135 on @xmath31 in constant depth . and @xmath135 is always @xmath77 . ] in particular , the complete initialization can be realized in depth @xmath3 . child node @xmath141 child node @xmath142 child node @xmath143 child node @xmath144 child node @xmath145 child node @xmath143 child node @xmath146 child node @xmath143 child node @xmath143 child node @xmath147 child node @xmath148 child node @xmath143 child node @xmath143 child node @xmath149 child node @xmath143 child node @xmath143 child node @xmath150 child node @xmath151 ; to proceed from one tree level to the next , a layer of addition circuits is used that operate in parallel , each circuit adding two curve points . realizing each such addition circuit in logarithmic depth as in corollary [ cor : lowdepthadd ] results in a quantum circuit of depth @xmath4 to compute @xmath131 . figure [ fig : treeaddition ] shows the resulting tree structure to compute @xmath131 . we obtain the following lemma . using projective coordinates , for a complete binary edwards curve the map @xmath152 where @xmath153 , can be realized in depth @xmath4 . in our discussion of the elliptic curve arithmetic we focused on the use of projective coordinates , as this avoids the use of a ( costly ) @xmath2-inversion . we pay for this by a non - unique point representation , however . to pass from projective coordinates @xmath154 for a point on @xmath31 to the unique affine coordinates @xmath155 , it is sufficient to compute the inverse of @xmath156 followed by two multiplications in @xmath2 . from theorem [ prop : gf2nlowdepthmult ] we know that these two multiplications can be implemented in depth @xmath3 , and we are left with the task of computing @xmath157 . in @xcite a depth @xmath5 circuit is proposed for this task which builds on a classical algorithm by itoh ans tsuji @xcite . this algorithm consists of two main parts followed by a final squaring in @xmath2 . the two main parts require a total of @xmath3 multiplications in @xmath2 and @xmath3 exponentiations with fixed powers of @xmath12 . the latter maps are bijective and @xmath95-linear and can be implemented as matrix - vector multiplications with a fixed matrix . consequently all necessary computations in these two parts can be implemented in depth @xmath4 . the final squaring operation can again be realized as a matrix - vector multiplication with a fixed matrix , and so the complete inversion can be implemented in depth @xmath4 . there is a polynomial - size quantum circuit of depth @xmath4 which on input a polynomial basis representation of @xmath158 computes a polynomial basis representation of the inverse @xmath159 . implementing an @xmath2-inverter in this way and combining it with two multiplication circuits as in theorem [ prop : gf2nlowdepthmult ] , we can derive the unique affine representation of a curve point in polylogarithmic depth . more specifically , we have the following . there is a polynomial - size quantum circuit of depth @xmath4 which on input a projective representation @xmath44 of a point on @xmath31 , returns the affine representation @xmath160 of this point . we now have all the pieces in place to establish our main result . shor s algorithm to compute the discrete logarithm on a complete binary edwards curve can be implemented using a quantum circuit of polynomial size and depth @xmath1 . referring to the three stages as in figure [ fig : shorcircuit ] , we first note that the hadamard gates on the left can be implemented in depth @xmath77 . next , as shown in subsection [ sec : doubleadd ] , we can implement the operation that computes @xmath161 in depth @xmath1 with a circuit of size polynomial in @xmath24 . in order to complete the proof , we have to bound the depth of the fourier transform @xmath162 . for this , we use a result shown in @xcite that the fourier transform on @xmath75 qubits with target error @xmath163 can be implemented with a circuit of depth @xmath164 and polynomial size . choosing @xmath165 and @xmath166 is sufficient for shor s algorithm to find the discrete logarithm with constant probability of success @xcite , i.e. , we can upper bound the depth of the @xmath71 by @xmath167 and hence the overall depth of the circuit by @xmath1 . @xmath168 the above discussion shows that the depth @xmath0 bound for discrete logarithm computations on ordinary binary elliptic curves can be improved exponentially : using parallelization at multiple levels , we can implement shor s algorithm on complete binary edwards curves in depth @xmath1 . to show our result we introduced the first sublinear - depth circuits for @xmath2-multiplication and @xmath2-inversion , which may be of independent interest . the price for the exponential reduction in depth is the introduction of ancillae a parallelized @xmath2-multiplier as discussed in section [ sec : fieldmult ] adds @xmath0 additional qubits . depending on the number of qubits available , for cryptographically significant parameters , say @xmath169 , trade - offs that put less emphasis on depth - optimization could be interesting . for instance , one could combine the parallelized double - and - add computation with a linear - depth field arithmetic . even though not being depth - optimal , thousands of qubits could be saved in this way . one could also try to avoid the general point addition circuits in the parallelized double - and - add procedure ( which involve many field multiplications ) and instead rely on a sequential depth @xmath5 solution with specialized point addition circuits as discussed in section [ sec : seq2a ] . it is not clear at this point how the best trade - off for a large - scale discrete logarithm computation looks like . it is interesting to compare our findings for bounding the depth of shor s algorithm over an additive group on an ordinary binary elliptic curve to the case of factoring . in both cases the high - level structure namely , phase estimation over an abelian group of the algorithm is the same , however , the details on how to implement the arithmetic differ significantly . in the case of the elliptic curve arithmetic , the bottle neck are the addition formulae for points on the curve which involve finite field divisions . our polylogarithmic depth implementation of shor s algorithm may be compared with the findings of @xcite in which an @xmath1 upper bound on the depth for shor s algorithm for integer factorization was given for a 2d nearest neighbor architecture and to @xcite , where a circuit of depth @xmath0 was derived , albeit on a 1d nearest neighbor architecture . the circuits presented in this paper assume an arbitrary coupling model and we leave it as an open problem whether they can be adapted to a 2d nearest neighbor architecture while maintaining the same upper bound on the circuit depth . to enable parallelization , we make intensive use of ` multi - fan - out cnots with @xmath61-input ' , and one may hope for further depth improvements if multi - fan - out gates are provided . at the moment we do not know how to derive an asymptotic benefit of such gates that goes beyond a constant factor improvement . the remaining bottleneck for an asymptotic improvement in the @xmath2-arithmetic seems the logarithmic depth for parity computations . finally , amy et al.s @xcite observation that @xmath2-multiplication can be realized in @xmath170-depth 2 suggests another natural direction for follow - up research : minimizing @xmath170-depth and the number of @xmath170-gates necessary to compute discrete logarithms on binary elliptic curves . we thank the reviewers for their constructive feedback and we thank schloss dagstuhl , germany , for providing an excellent environment for part of this research through a _ quantum cryptanalysis _ seminar . rs was supported by the spanish _ ministerio de economa y competitividad _ through the project grant mtm-2012 - 15167 and by nato s public diplomacy division in the framework of `` science for peace '' , project md.sfpp 984520 . this work was carried out while mr was with nec laboratories america , princeton , nj 08540 , usa . daniel j. bernstein , tanja lange , and reza rezaeian farashahi . . in elisabeth oswald and pankaj rohatgi , editors , _ cryptographic hardware and embedded systems ches 2008 _ , volume 5154 of _ lecture notes in computer science _ , pages 244265 . international association for cryptologic research , springer , 2008 . edoardo d. mastrovito . . in teo mora , editor , _ proceedings of the sixth symposium on applied algebra , algebraic algorithms and error correcting codes _ , volume 357 of _ lecture notes in computer science _ , pages 297309 . springer , 1988 .
improving over an earlier construction by kaye and zalka @xcite , in @xcite maslov et al . describe an implementation of shor s algorithm , which can solve the discrete logarithm problem on ordinary binary elliptic curves in quadratic depth @xmath0 . in this paper we show that discrete logarithms on such curves can be found with a quantum circuit of depth @xmath1 . as technical tools we introduce quantum circuits for @xmath2-multiplication in depth @xmath3 and for @xmath2-inversion in depth @xmath4 .
much of the stellar mass in the universe forms in starbursts ( e.g. blain et al . 1999 ; genzel et al . 2001 ; smail et al . 2002 ) , spectacular events during which the star formation rate ( sfr ) of a galaxy rises from few solar masses per year , typical of spirals such as the milky - way ( mckee & williams 1997 ) , to several hundred solar masses per year as in the local ultra luminous infrared galaxies ( ulirgs ) ( e.g. sanders & mirabel 1996 ; genzel et al . 1998 ; sanders & ishida 2004 ) where merger - driven starbursts take place in very compact ( d@xmath6100300pc ) dense gas disks ( e.g. downes & solomon 1998 ; sakamoto et al . 2008 ) . in the so called sub - mm galaxies ( smgs ) , similarly merger - driven starbursts in the distant universe , sfrs can even reach thousands of solar masses per year ( smail et al . 1997 ; hughes et al . 1998 ; eales et al . 1999 ) . while stars form invariably out of molecular gas throughout the universe ( e.g. solomon et al . 1997 ; frayer et al . 1998 , 1999 ; greve et al . 2005 , tacconi et al . 2006 ) , it is in its densest , uv - shielded , and cosmic - ray - dominated phase where the star formation initial conditions , and the characteristic mass of the emergent young stars @xmath19 are truly set ( bergin & tafalla 2007 ; elmegreen et al . 2008 ) . indeed , while photon - dominated - regions ( pdrs ) contain the bulk of the molecular gas mass in galaxies ( hollenbach & tielens 1999 ; ossenkopf et al . 2007 ) , their widely varying physical conditions remain always far removed from those prevailing in uv - shielded cr - dominated dense gas cores . the location of the latter deep inside molecular clouds , a strong temperature regulation via the onset of gas - dust thermal equillibrium at densities @xmath17@xmath2010@xmath4@xmath14 , and the nearly complete dissipation of supersonic turbulence ( larson 2005 ; bergin & tafalla 2007 ; jappsen et al . 2005 ) , all help to keep the thermal state of typical crdrs a near invariant over a wide range of ism environments . cosmic rays ( crs ) , accelerated in shocks around supernovae remnants ( snrs ) or winds associated with o , b star clusters ( e.g. binns et al . 2008 ) , are responsible for the heating , ionization level , and chemical state of such dense uv - shielded regions in molecular clouds ( e.g. goldmith & langer 1978 ; goldsmith 2001 ) . moreover , while their heating contribution relative to that from the ( far - uv)-induced photoelectric effect in pdrs , or turbulence , can vary within a galaxy and individual molecular clouds , crs remain responsible for setting the _ minimum _ temperature attainable in the gaseous ism ( e.g. goldmith & langer 1978 ; bergin & tafalla 2007 ; elmegreen et al . for the galaxy @xmath21(min)@xmath610k , reached in uv - shielded cores with @xmath17@xmath6(@xmath22@xmath23)@xmath14 . a recent study has shown that the average cr energy densities ( @xmath0 ) in compact starbursts typical in ulirgs ( recently found also in an smg at [email protected] , swinbank et al . 2010 ) , will be boosted by the tremendeous factors of @xmath24@xmath25(few)@xmath26(10@xmath310@xmath4)@xmath5 , potentially transforming their ism into extreme cosmic - ray - dominated regions ( crdrs ) . this will dramatically alter the thermal and ionization state of uv - shielded cores , and thus the initial conditions of star - formation , throughout their considerable molecular gas reservoirs ( @xmath2710@xmath28 - 10@xmath29m@xmath30 ) . higher characteristic mass @xmath31 for the young stars formed in such ism environments , and longer ambipolar diffusion timescales for the cr - heated uv - shieled cores could thus be a new paradigm for high - density star formation in the universe ( see papadopoulos 2010 for details ) . it is worth mentioning that recently such large @xmath32 boosts ( @xmath33@xmath34u@xmath35 ) have been found in the oriona bar , and while not representing the average sf initial conditions in the galaxy , they do allow local studies of the effects of large cr energy densities on the ism ( pellegrini et al 2009 ) . finally crs have been shown to effect the thermal state of primordial gas , catalyzing h@xmath36 formation in the absence of dust ( jasche , ciardi , & ensslin 2007 ) . in this work we solve the coupled chemical and thermal state equations for uv - shielded gas and its concomitant dust in extreme crdrs without the approximations employed by papadopoulos ( 2010 ) . we then use the thermal states computed for such a gas phase to examine the effects on @xmath37 and the so - called inflection point of the effective equation of state ( eos ) , set at gas densities above which @xmath38@xmath1@xmath16 ( e.g. spaans & silk 2000 ; larson 2005 and references therein ) . both of these play a crucial role in determining the stellar initial mass function ( imf ) in galaxies , which could thus turn out to be very different in the crdrs of extreme starbursts . throughout this work we make the standard assumption of @xmath39@xmath40@xmath41 ( sfr _ density _ ) as in papadopoulos 2010 , though we now discuss possible caveats ( section 4.4 ) . finally , the different imfs expected in extreme crdrs will modify the total sfr and @xmath42 values computed in galaxies ( from e.g. cm non - thermal or ir thermal continuum ) . in this new context and throughout this work sfr and @xmath43 will now denote these quantities only for the massive stars ( which define @xmath39 and power the ir / cm continuum of the ism ) , with their total values necessarily remaining uncertain , depending on the exact imf , which will no longer be an invariant . in this section we compute the equillibrium gas temperatures for the uv - shielded regions of molecular clouds immersed in the large cosmic ray energy densities of the extreme crdrs expected in starbursts with high sfr densities . we first consider the simple case where the chemical state of the uv - shielded dense gas cores remains unaltered and the only effect of large @xmath39 values is to heat them much more than in quiescent ism environments . following standard practice ( e.g. goldmith 2001 ) the temperature of such regions can be estimated from the thermal balance equation @xmath45 where @xmath46 is the cr heating with @xmath47@xmath48@xmath32 being the cosmic ray ionization rate . the cooling term @xmath49 is dominated by co and other molecular lines , while @xmath50 accounts for gas cooling due to the gas - dust interaction . a significant point of departure from the previous analysis by papadopoulos ( 2010 ) is the inclusion of the atomic fine structure lines oi ( at 63@xmath51 m ) and cii ( at 158@xmath51 m ) in @xmath52 . indeed for the cr - boosted temperatures expected in the uv - shielded dense gas of extreme crdrs , and unlike the corresponding gas phase in the quiescent ism of the galaxy , the cooling from the fine structure atomic line of oi ( 63@xmath51 m ) can also become important and along with the gas - dust interaction overtake the co lines for typical core densities of @xmath17@xmath5310@xmath12@xmath14 when @xmath15@xmath2050k . finally ism - cr interactions can generate internal uv radiation which destroys co , producing traces of cii , thus cooling via its fine structure line at 158@xmath54 m must also be included . the thermal balance equation encompassing all physical processes that can be important in the uv - shielded ism of extreme crdrs thus is : @xmath55 ( for the expressions used see the appendix ) . the solutions are shown in figure 1 from where it becomes obvious that the minimum temperatures attainable in dense gas cores in the ism of such starbursts are much higher than in the galaxy or galaxies with moderate @xmath41 values . turbulent gas heating was not considered since we want to compute the minimum temperature values for uv - shielded gas and since supersonic turbulence ( and thus shock - heating ) has usually dissipated in dense pre - stellar gas cores where the initial conditions of sf are set . from figure 1 it can be seen that the minimum gas temperatures possible in extreme crdrs are lower than those computed by papadopoulos ( 2010 ) , a result of the extra cooling power emanating from the now included atomic transitions that become excited at the high temperatures now possible in an ism permeated by large cr energy densities . nevertheless these minimum possible gas temperatures remain @xmath6510 times higher in extreme crdrs than those in the milky way or galaxies with moderate @xmath42 values . any ir radiation `` leaking '' deep in the cloud and warming the concomitant dust above @xmath1(810)k , or any remnant turbulent gas heating ( pan & padoan 2010 ) can only raise these minimum temperatures further . the thermal balance equation for uv - shielded , cr - heated gas ( equation 2 ) and its solutions ( figure 1 ) , while physically transparent , they do not account for the fact that ism chemistry ( and thus the abundance of coolants ) is also strongly cr - regulated in such regions . this makes the problems of thermal and chemical balance of cr - dominated gas strongly coupled that must be solved together , especially given the very large cr energy densities possible in the crdrs of compact starbursts . we do so by utilizing the advanced time - dependant ucl - pdr code ( the latest version described in bell et al . 2007 ) to compute gas temperatures of cr - heated molecular cloud interiors self - consistently with their cr - regulated chemistry for a wide range of cr energy densities . this model has already been successfully used several times to model pdrs in the galaxy ( e.g. thi et al . 2009 ) . the heating mechanisms included are : ( i ) photoelectric effect from silicate grains and polycyclic aromatic hydrocarbons ( pahs ) , ( ii ) h@xmath57 formation on gain surfaces , ( iii ) h@xmath57 photodissociation , ( iv ) h@xmath57 uv fluorescence , ( v ) cii recombination , ( vi ) interaction of low - energy cosmic rays with the gas . the latter dominate gas heating deep inside h@xmath36 clouds ( @xmath58@xmath595 ) where dense gas cores reside and the initial conditions for star formation are set . at cloud surfaces hot gas emits ly@xmath60 and oi63@xmath54 m lines while deeper in ( @[email protected] ) , gas cooling occurs via fine - structure line emission of oxygen and neutral and singly ionized carbon . vibrational and ro - vibrational lines of h@xmath57 and co also cool the gas while co rotational lines are excellent coolants for the cold inner regions as the first rotationally - excited level reaches down to 5.5 k. the interaction between the gas and the dust grains acts as cooling or heating agent for the gas depending on the difference between the gas and dust temperature . h@xmath57 and co self - shielding are taken into account . the code calculates the gas chemical abundances , emergent cooling line fluxes , and temperatures self - consistently at each depth in the cloud , and it has been benchmarked against other state - of - the art photodissociation region codes in rllig , m. et al . ( 2007 ) . we set a nominal incident far - uv field of @xmath62=1 in habing units ( expected to be even lower for the dense pre - stellar cores deep inside h@xmath36 clouds ) and a residual micro - turbulent line width of @xmath63=5kms@xmath64 ( which enters the expressions of line optical depths ) . in all cases chemical and thermal equillibrium is reached fast , within t@xmath65@xmath66yrs , and is always shorter than the local dynamical timescale which drives the ambient star formation events and thus the temporal evolution of @xmath0 . the results , shown in figure 2 , corroborate the high gas temperatures obtained using the simpler method . this reveals the cosmic rays as _ the most important feedback aspect of star formation _ capable of significantly altering its initial conditions during events of high star formation density via the strong heating of the deeply embedded and uv - shielded dense gas regions of molecular clouds . it can thus operate unhindered in the high dust extinction environments of extreme starbursts such as local ulirgs or high redshift smgs . we also ran the same grid of models with the incident far - uv radiation field scaling exactly as the cr energy densities which corresponds to the maximum incident radiation field that can be expected for dense gas cores ( in practice pre - stellar cores are always uv - shielded deep inside molecular clouds ) . the dust temperature is now more important than in the previous case of @xmath67=1 , and is computed as part of the code . its value inside the cloud can influence : the cooling by emitting ir photons which interact with line radiative transfer , the h@xmath36 formation by changing the sticking probability , the evaporation / condensation of molecules , and most importantly the gas - grain heating / cooling . the ucl - pdr code uses a modified version of the formula of hollenbach , takahashi , & tielens ( 1991 ) ( hereafter htt ) corrected for the various mean grain sizes . the htt formula has been further modified to include attenuation of the far - ir radiation . the incident fuv radiation is absorbed and re - emitted in the infrared by dust at the surface of the cloud ( up to @xmath68@xmath111mag ) . in the htt derivation , this fir radiation then serves as a second heat source for dust deeper into the cloud . however in their model this secondary radiation component is not attenuated with distance into the cloud , remaining undiminished with depth and leading to higher dust temperatures deep into the cloud , which in turn heats the gas to unrealistic temperatures . attenuation of the fir radiation has been introduced by using an approximation for the infrared - only dust temperature from rowan - robinson ( 1980 ) , his equation 30b : @xmath69=@xmath70 where @xmath71 is the cloud depth corresponding to @xmath68=1mag outer slab that processes the incident fuv radiation and then re - emits it in the fir . from figure 3 it can be readily seen that the gas temperatures remain invariant beyond @xmath68@xmath65 even with with incident far - uv fields as enhanced as the cr energy densities . this essentially leaves crs _ as the only strong feedback factor of star formation operating efficiently deep inside molecular clouds . _ moreover , intense cr heating of uv - shielded gas strongly decouples gas and dust temperature even at high densities ( figures 1 , 2 , 3 ) with important implications on the gas fragmentation properties and thus on the emergent stellar imf ( section 3.2 ) . the new detailed thermo - chemical calculations presented here support the results from the previous study of crdrs and the effects of high @xmath32 values on the initial conditions of star formation ( papadopoulos 2010 ) . the thermal state of uv - shielded dense gas is indeed dramatically altered in the crdrs of extreme starbursts , and this will affect its jeans mass , gas fragmentation properties , and the charecteristic mass @xmath72 of young stars . this contradicts the results by elemgreen et al . ( 2008 ) which , while correctly considering the uv - shielded gas phase as the one where the sf initial conditions are set , it omitted crs from its thermal balance . it must be pointed out that while several other studies investigated the effects of starburst environments on the imf , all ( besides the one by elmegreen et al . 2008 ) failed to consider the ism phase where the initial conditions of sf are actually set ( the crdrs ) . instead they used physical properties typical of pdrs ( e.g. klessen et al . 2007 ) which , while dominating the observed dust seds and molecular line spectral energy distributions ( sleds ) of galaxies , they are hardly representative of the physical conditions expected in their crdrs . even for mildly sf ir - luminous galaxies the dust temperature typical of their global seds is @xmath1(3040)k , which is well above @xmath16@xmath6(810)k expected in their crdrs . using inappropriate pdr - type initial conditions of star formation in numerical models : a ) would make them hard - pressed to account even for the near - invariant imf found in the galaxy ( as the wide range of physical conditions in its massive pdr gas phase would yield a wide range of imfs ) , b ) omits the significant , cr - induced , thermal decoupling between gas and dust ( figures 2 , 3 ) . the latter fundamentally impacts the so - called inflection point of effective equation of state ( eos ) used in all realistic ( i.e non isothermal ) numerical simulations of molecular cloud fragmentation ( e.g. jappsen et al . 2005 ; bonnell et al . 2006 ) , and this is further discussed in section 3.2 . the thermal state of these dense pre - stellar gas cores is no longer a near - invariant but is instead strongly altered by the high cr energy density backgrounds of extreme starbursts , inevitably affecting the fragmentation of those gas regions towards protostars . indeed , the jeans mass for such cr - heated cores @xmath73^{-1/2}\ , m_{\odot},\end{aligned}\ ] ] rises from @[email protected]@xmath75 when @xmath76(h@xmath36)=10@xmath12@xmath14 and @xmath15=10k ( typical for the galaxy ) , to @xmath77@xmath1(310)m@xmath30 for @xmath78@xmath6(50110)k expected for same density h@xmath36 gas immersed in the very large cr energy densities of extreme starbursts . the near - invariance of @xmath79 in galaxies ( elmegreen et al . 2008 ) is thus upended in starbursts with high sfr densities . in figure 4 the @xmath80 values computed for molecular gas immersed in the crdrs of compact extreme starbursts are shown for a gas density range typical for star - forming molecular clouds . for @xmath81@xmath5910@xmath3 , @xmath80 increases by a factor of @xmath610 across the entire range , which will invariably lead to a higher @xmath82 and thus a top - heavy imf ( larson et al . 2005 ; elmegreen et al . 2008 ) . the wide range of densities over which this occurs makes this conclusion independent of the specific details of molecular cloud fragmentation as long as @xmath83 $ ] drives the fragmentation process outcome at each spatial and temporal point @xmath84 . the latter is the current consensus on the role of the jeans mass in molecular cloud fragmentation towards a stellar imf , though the views regarding the particular gas phase whose @xmath80 sets the @xmath82 vary ( e.g. klessen 2004 ; bonnell et al . 2006 ) . if @xmath85@xmath48@xmath86 at the _ onset _ of molecular cloud fragmentation ( klessen 2004 ) , the cr - boost of @xmath82 in extreme crdrs is obvious from the large vertical displacement of the @xmath80 values in figure 4 . in a non - isothermal gravoturbulent molecular cloud fragmentation scheme ( e.g. jappsen et al . 2005 ) it is @xmath87@xmath11@xmath88(@xmath89 ) at a characteristic gas density @xmath89 above which efficient gas - dust thermal coupling lowers the gas temperature in crdrs towards a minimum value of @xmath90(min)@xmath6@xmath16 ( larson 2005 ) . for the galaxy this occurs at @xmath91@xmath110@xmath12@xmath14 , a density that marks also the transition to near - isothermal cloud regions where thermal motions dominate over supersonic turbulence ( larson 2005 ; bergin & tafalla 2007 ) . from figures 2 , 3 it is obvious that the high cr energy densities in extreme crdrs _ will keep the gas and dust thermally decoupled , with @xmath92(min)@xmath93@xmath9 _ , and thus a characteristic density can no longer be marked by the onset of thermal equillibrium between gas and dust . such cr - induced effects have been long suspected for the dense molecular gas in the galactic center ( yusef - zadeh et al . 2007 ) . in the cr - innudated molecular clouds of extreme starbursts a characteristic density in the ism can now remain only as the one where the supersonic turbulence has fully dissipated and overtaken by thermal motions in pre - stellar cores , a prerequisite for their efficient gravitational collapse ( e.g. krumholz & mckee 2005 ) . the average density of such transition regions can be estimated from two well - known scaling relations in molecular clouds : @xmath94=@xmath95 ( linewidth - size ) and @xmath96=@xmath97 ( density - size ) after setting @xmath98(min)=@xmath99 as the minimum linewidth possible , and solving for the corresponding mean volume density , @xmath100 for h=1/2 ( expected for gas clouds virialiazed under external pressure , elemgreen 1989 ) , @xmath101=1.1kms@xmath64 , and for @xmath102=few@xmath10310@xmath3@xmath14 , it is @xmath104@xmath25few@xmath10510@xmath4@xmath14 . for the cr - boosted gas temperatures in extreme starbursts where typically @xmath106(min)@xmath6100k ( figures 1 , 2 , 3 ) : @xmath107@xmath108few@xmath109@xmath14 , and the ( turbulent gas)@xmath110(thermal core ) transition will occur at lower densities _ as well as _ higher ( cr - boosted ) gas temperatures than in quiescent crdrs this will then shift the expected @xmath111 both vertically ( higher temperatures ) , as well as to the left ( lower densities ) of figure 4 , yielding even higher @xmath112 values than those resulting from only a vertical shift of @xmath113 because of cr - boosted gas temperatures . thus irrespective of molecular cloud fragmentation details , _ a much larger characteristic mass of young stars is expected in the cr - innudated molecular clouds of extreme starbursts . _ however the exact shape and mass scale of the new stellar imf can only be determined with dedicated molecular cloud fragmentation simulations that make use of the new sf initial conditions in extreme crdrs , along with an eos appropriate for the very different thermal and ionization state of their molecular gas . the importance of the effective equation of state ( eos ) , parametrized as a polytrope p=@xmath114 , in the fragmentation properties and emergent mass spectrum of turbulent self - gravitating gas has been well - documented in a few seminal papers ( li et al . 2003 ; jappsen et al . 2005 ; bonnell et al . 2006 ; klessen et al . 2007 ) revealing a crucial role for @xmath115 in defining the shape of the collapsed gas core mass distribution ( and hence the stellar imf ) . the eos , its polytropic index @xmath115 as a function of density , and their dependance on ism properties such as metallicity , and physical conditions such as the elevated cmb at high redshifts have also been well - documented ( spaans & silk 2000 , 2005 ) , with the so - called eos inflection point emerging as a general characteristic . in an eos that seems well - approximated by a piecewise polytrope with two distinct @xmath115 values , the latter is simply the characteristic density @xmath116 for which @xmath115@xmath1171 when n@xmath117@xmath118 , and @xmath115@xmath1191 for n@xmath120@xmath118 . its crucial role in determining the imf mass scale , discussed briefly in 3.1 , is documented throughout the literature ( larson 2005 ; jappsen et al . 2005 ; elmegreen 2008 ) . in starburst environments a `` sharpening '' of that inflection point is thought to occur because of warm dust . there @xmath115 reaches up to @xmath61.4 and stays above unity for densities past the inflection point @xmath121@xmath1(10@xmath410@xmath12)@xmath14 ( spaans & silk 2005 ) , which surpresses molecular cloud fragmentation while `` tilting '' it towards top heavy imfs ( e.g. li et al . 2003 ; klessen et al . 2007 ) . unfortunately pdr - like ism has been used to set the sf initial conditions for starbursts in all of the aforementioned models , even though the dust in crdrs ( where these conditions are really set ) remains cool , much lower than the gas , especially in extreme crdrs . this insensitivity of the crdr dust temperatures to the ambient far - uv radiation field can be demonstrated analytically by simply considering the location of crdrs deep inside far - uv illuminated molecular clouds at @xmath122 ( e.g. mckee 1989 ) . then @xmath123@xmath11@xmath124^{1/(4+a)}$ ] will be the radiation induced boost of the crdr dust temperature in starbursts @xmath125 with respect to that in the quiescent galactic environments @xmath126@xmath278k . for a dust emissivity law of @xmath60=2 , a large enhancement of the far - uv radiation field @xmath127=10@xmath4 , and @xmath128=5 it is @xmath129=2.15 , yielding @xmath16@xmath117k , in good agreement with the detailed dust temperature profiles past @xmath58@xmath15 shown in figure 3 . it is actually this robustness of dust temperatures inside crdrs , along with a strong gas - dust coupling at high densities ( when @xmath39 values are modest ) that helps retain : the thermal states of the pre - stellar gas cores , the @xmath72 values , and the imf , as near invariants over a wide range of ambient pdr conditions ( elmegreen et al . 2008 ) thus pdr conditions , while always dominating the observed dust seds and molecular sleds of even mildly sf galaxies ( as pdrs are warm _ and _ contain the bulk of the ism mass ) _ they are not representative of the sf initial conditions in galaxies , _ which are set in crdrs . this leaves the pre - stellar gas phase in them , amounting to a few% of the mass per giant molecular cloud ( @xmath6sf efficiency per typical gmc ) , yet to be explored in terms of its eos and fragmentation properties . a detailed exploration of those in crdrs will be the subject of a future paper , but figures 2 and 3 already show that crs , by inducing a strong decoupling of the gas and dust temperatures with @xmath21@xmath8@xmath16 , they `` erase '' any putative eos inflection point for gas densities up to @xmath130=10@xmath13@xmath14 . this will maintain efficient molecular gas cooling , with cooling times always much shorter than dynamical times ( figure 5 ) , and @xmath131@xmath1171 in crdrs over the entire gas density range explored here . in such regions gas gravoturbulent fragmentation will then occur equally efficiently from n@xmath25@xmath132@xmath14 up to @xmath110@xmath13@xmath14 yielding a pure power law imf ( spaans & silk 2000 ; li et al . 2003 ) . the latter result , along with our discussion in 3.1 makes clear that the stellar imf will be very different in extreme crdrs , though only detailed gravoturbulent simulations with the appropriate initial conditions , and performed with a new suitable eos can decide its exact shape and mass scale . it must be also noted that the very high ionization fractions in extreme crdrs can now `` anchor '' the magnetic field for much longer periods , and at much higher densities , as ambipolar diffusion in such regions is now much prolonged ( papadopoulos 2010 ) . magnetic fields may thus remain important during much later stages of gas fragmentation , even in a highly turbulent medium , and may thus have to be included in the simulations of this process in extreme crdrs . the current study supports the recently suggested role of crs as a decisive sf feedback factor , capable of altering the sf initial conditions and the imf in extreme crdrs . in this section we discuss some of the consequences beyond the immediate issues of the ism and the stellar imf in such enviornments , and outline current and future observations that can shed light on these new sf initial conditions that will be typical in high sfr density events in the universe . the dependance of the thermo - chemical and ionization state of pre - stellar dense gas in crdrs on the average @xmath39 permeating the ism of a galaxy naturally links the sf initial conditions and the resulting imf to the average sfr density @xmath134 , provided @xmath135 . the details of cr transport / escape mechanisms in quiescent sf disks and starbursts will define the proportionality factor ( see suckhov et al . 1993 for detailes ) . thus , once simulations of molecular gas fragmentation in crdrs produce libraries of the corresponding imfs , these will be naturally parametrized by @xmath134 , and hence actual observables ( especially in the upcoming era of high - resolution imaging at submm wavelengths with alma ) . in hierarchical galaxy formation models a bimodal imf , top - heavy for mergers / starbursts and regular for disk / regular - sf systems , has been postulated to explain the populations of the starbursting / merger - driven submillimeter galaxies ( smgs ) and regular sf disks like the ly - break galaxies ( lbgs ) at high redshifts in @xmath136cdm cosmology ( baugh et al . the cr - controlled sf initial conditions in galaxies , unhindered by dust extinction , can now set this on a firm physical basis . from figures 2 and 4 is obvious that for typical sf disks ( @xmath24@xmath137(0.210)@xmath103@xmath138 ) no substantial effects on the gas temperatures and jeans masses in crdrs are expected while in the extreme cdrds of compact starbursts ( @xmath24@xmath137(10@xmath310@xmath4)@xmath139@xmath138 ) very warm gas and large jeans masses will lead to an effective surpression of low - mass star formation and hence a top - heavy imf . for merger - driven star formation the latter may characterize entire star - forming episodes ( i.e. until the in - situ molecular gas reservoir is nearly exausted ) as dissipation can continously `` funnel '' the large molecular gas masses of the progenitors towards very compact ( @xmath1(100500)pc ) regions where extreme , eddington - limited sf , takes place with tremendeous ir brightness of @xmath140@xmath11(few)@xmath13910@xmath141l@xmath30kpc@xmath142 ( e.g. thompson 2009 ) . such high ir brightness , and thus high sfr densities , have been recently deduced also in a distant smg ( swinbank et al . 2010 ) . thus cr - controlled sf initial conditions naturally yield a bimodal sf process in galaxies , with high @xmath134 values providing the branching point , as follows : + ( merger - driven sf)@xmath143(high @xmath41)@xmath110(high @xmath32)@xmath143(top heavy imf ) , + ( regular disk sf)@xmath143(low @xmath41)@xmath110(low @xmath32)@xmath143(galactic imf ) . + this proposed @xmath41-imf link can then parametrize such an sf bimodality in hierarchical galaxy formation , once imf = f(@xmath144 ) `` libraries '' are obtained from e.g. gravoturbulent simulations of molecular cloud fragmentation for a range of @xmath144 values . a top - heavy imf in high-@xmath145 systems will have very serious implications on the actual sfrs deduced from the ir luminosities of these typically very dust - enshrouded systems ( where all the far - uv light from massive stars is reprocessed into ir ) . their cr - induced top - heavy imfs will now correspond to several times higher energy outputs per stellar mass , and the sfrs in such systems will have to be revised downwards by identical factors . a @xmath43-dependent imf as the natural outcome of cr - controlled sf initial conditions in galaxies can also naturaly yield integrated galactic imfs ( igimfs ) that depend on the star formation history ( sfh ) of galaxies . this is because @xmath146 can be a strong function of the evolution of a galaxy , especially during its early and presumably gas - rich state even when no mergers are involved . such a sfh - dependance of the igimfs has been considered before as the cause of the well - known mass - metallicity relation of galaxies ( kppen , weidner , & kroupa 2007 ) . setting @xmath39@xmath147@xmath134 in crdrs assumes that crs can freely penetrate and warm the entire volume of a gmc down to its densest uv - shielded regions . recent gamma - ray observations have directly measured cr energy densities in the nucleus of a nearby galaxy on par with those computed for extreme crdrs and expected from its sfr density , while indicating only mildly calorimetric conditions for the cr / ism p - p interactions ( acero et el . 2009 ) . the latter supports an unhintered cr penetration through the molecular clouds . nevertheless alfvn waves generated _ outside _ such clouds by a net flux of crs in their environments could be effective in keeping low energy crs below a few hundred mev ( those most effective in heating the gas ) outside their high density regions ( skilling & strong 1976 ) , which in turn would affect their thermal balance , their fragmentation properties , and the emergent stellar imf . thus the coupled problems of cr transport , magnetic field dynamics , and gravoturbulent evolution of non - isothermal dense molecular gas with high degrees of ionization like those expected in compact extreme starbursts such as arp220 ( e.g. greve et al . 2009 ) must be investigated with dedicated mhd simulations . the detections of @xmath148-ray emission from two nearby galaxies , namely m82 and the starburst nucleus of ngc253 ( acciari et al . 2009 , acero et al . 2009 ) , are two very important recent developments in @xmath115-ray astronomy , given that @xmath115-ray emission results from cr proton interactions with the molecular ism . this allows such observations to directly measure the average cr energy densities in sf environments , and thus sensitive such measurements in ulirgs such as the nearby ulirg arp220 will be very important in revealing extreme crdrs in such systems . on the other hand , direct measurements of the expected top - heavy imfs in the heavily dust - enshrouded compact starbursts where extreme crdrs would reside are impossible . however any cases where the `` end '' products , i.e. the stellar populations , of high-@xmath145 sf events can be observed and their imf determined such as the galactic center ( gc ) and local ultra compact dwarfs ( ucds ) allow such direct observational tests . there is indeed strong evidence of cr heating of the dense molecular gas in the gc , setting @xmath149@xmath20@xmath16 , and expected strong effects on the local jeans mass and ambipolar diffusion timescales ( yusef - zadeh et al . 2007 ) . unfortunately , initial assertions about a top - heavy imf in the gc ( e.g. nayakshin & sunyaev 2005 ) and the massive arches cluster near it ( figer et al . 1999 ; stolte et al . 2002 , 2005 ) have not been corroborated by later studies ( kim et al . 2006 ; espinoza et al . 2009 ; lckmann et al . 2010 ) . on the other hand some evidence exists for a top - heavy imf in the nuclear starburst of m82 ( e.g. rieke et al . 1980 ; 1993 ) , and a significantly enhanced @xmath32 is certainly there ( @xmath6500@xmath2@xmath138 , directly measured via @xmath131-ray observations ; acciari et al . 2009 ) to achieve this via a cr - boosted @xmath19 imf mass scale . indeed for such @xmath24 values , setting @xmath150@xmath25@xmath151 yields @xmath152(m82)@xmath202m@xmath30 ( fig.4 ) . however the galactic center does offer an example of regions with locally boosted @xmath0 values that do not necessarily correspond to top - heavy imfs in their vicinity , possibly because such boosts are too short and transient to systematically bias the local emergent imf . stronger evidence for top - heavy imfs exists in ucds , small stellar systems with r@xmath153@xmath1(1030)pc ( half - light radius ) where @xmath27(10@xmath1310@xmath154)m@xmath155 of gas have been converted into stars at @xmath156@xmath6(10100)m@xmath30yr@xmath64 ( see dabringhausen et al . 2009 and references therein ) . unlike the gc where imf studies are hindered by high extinction and tidal disruption of clusters , and where any high-@xmath145 sf episodes may have been short - lived ( and thus could not significantly bias its average imf towards a top - heavy one ) , the imfs in ucds are much more amenable to systematic studies , while their sf episodes have had @xmath157@xmath120@xmath158 ! they are thus ideal targets for studying the imf resulting from completed high-@xmath159 sf events . the next best thing besides studies of the imf itself is studies of the ism in crdrs from which the imf emerges , especially those in compact extreme starbursts where @xmath160 is very high ( e.g. ulirgs ) versus systems with more distributed sf and much lower @xmath145 values . the unique observational signatures of the very warm dense gas with high ionization fractions expected in extreme crdrs have already been discussed in some detail by papadopoulos ( 2010 ) , and the interested reader is refered to this work . in summary , in the era of herschel and the upcoming era of alma the sole main obstacle is the degeneracy of extreme crdr diagnostic with that of x - ray dominated regions ( xdrs ) induced in the ism by an x - ray luminous agn . indeed , unless an x - ray luminous agn can be excluded by other independent means ( e.g. deep x - ray observations ) , it can be difficult to tell extreme crdrs and xrds apart in a simple fashion as even exotic high excitation species such as oh@xmath161 and h@xmath36o@xmath161 , recently detected by herschel in the archetypal ulirg / qso mrk231 and attributed to xdrs ( van der werf et al . 2010 ) , could be partly due to crdrs . in recent work meijerink 2010 provide a detailed discussion of diagnostics lines in extreme crdrs versus xdrs and the key observations that can separate them . finally it is worth noting that x - rays and xdrs will have similar effects as crs on the jeans mass and the imf of uv - shielded dense gas regions as both can volumetrically warm such regions to high temperatures . this was pointed out recently by bradford et al . ( 2009 ) for the ism of a distant qso , and constitutes an omitted aspect of agn feedback on star formation . this has been studied recently and shown to lead to a top - heavy imf ( hocuk & spaans 2010 ) . xdrs besides elevating @xmath162(min ) to much higher values than in the galaxy they also yield @xmath38@xmath8@xmath9 ( and thus they can also `` erase '' the eos inflection point ) , though unlike crdrs their influence on the ism is necessarily limited by the @xmath6@xmath163 geometric factor ( @xmath164 the distance of the irradiated gas from the agn ) . we have conducted new , detailed calculations on the thermal balance of uv - shielded dense gas in cosmic - ray - dominated - regions ( crdrs ) , the ism phase where the initial conditions of star formation are set in galaxies , in order to examine in detail the effects of extreme crdrs on the mass scale of young stars and the imf recently suggested by papadopoulos ( 2010 ) . our results can be summarized as follows : \1 . we confirm the high temperatures of dense gas cores in the extreme crdrs expected in ulirgs and all systems with large star formation rate densities ( @xmath145 ) , albeit at lower levels than the original study . these cr - induced higher gas temperatures will lead to @xmath610 times larger jeans masses across the entire density range typical in such regions , and thus higher charecteristic mass of young stars in the ism of galaxies with high @xmath145 . the crs , by decoupling the thermal state of the gas from that of the dust in crdrs keep @xmath15@xmath8@xmath9 for all densities typical of molecular clouds . this effectively `` erases '' the so called inflection point of the effective equation of state , especially in extreme crdrs , inevitably altering the fragmentation properties of the gas . this effect , along with the expected much higher gas ionization fractions , necessitates a search for a new eos , appropriate for crdrs . these new cr - induced ism conditions provide a new paradigm for all high - density star formation in the universe , with crs as the key sf feedback mechanism , operating unhindered by extinction even in the most dust - enshrouded starbursts . the resulting @xmath43-dependent imf , yields : a ) a natural bimodal behaviour for merger - driven / starburst versus disk / regular - sf galaxies , and b ) integrated galactic imfs ( igimfs ) that depend on their star formation history , both with important consequences in galaxy formation and evolution . imf studies of stellar systems formed under extremely high @xmath165 values such as ultra compact dwarfs , and molecular line observations of the extreme crdrs in nearby ulirgs can provide the full picture of crdr properties and their effects on the imf . regarding the latter , the now spaceborn herschel space observatory , and the upcoming alma hold particular promise . we would like to thank pavel kroupa for numerous useful conversations , especially regarding the imf of ucds , and joerg dabringhausen for pointing out his relevant work . we also thank andrew strong for discussing uncertainties regarding cr transport in dense molecular clouds , and the referee for comments regarding the cr energy densities and their effect on the imf of m82 . * padelis p. papadopoulos would like to dedicate this work to the memory of his dear friend yannis bakopoulos , an excellent mathematician , fellow scientist , and committed social activist who died on 6th of december 2010 . * acero f. et al . ( h.e.s.s . collaboration ) , 2009 , science , vol . 326 , 1080 acciari v. a. et al . ( veritas collaboration ) , 2009 , nature , vol . 462 , 770 baugh c. m. , lacey c. g. frenk c. s. , granato g. l. , silva l. , & bressan a. 2005 , mnras , 356 , 1191 bayet e. , hartquist t. w. , viti s. , et al . 2010 , a&a , 521 , 16 bell t. a. , viti s. , & williams d. a. 2007 , mnras , 378 , 983 bergin e. a. , & tafalla m. 2007 , ara&a , vol . 45 issue 1 , pg.339 binns w. r. , wiedenbeck m. e. , arnould m. , et al . 2008 , new astronomy reviews , vol . 52 , issue 7 - 10 , p. 427 blain a. , smail i. , ivison r. j. , kneib , j .- 1999 , mnras , 302 , 632 bonnell i. a. , clarke c. j. , & bate r. 2006 , mnras , 368 , 1296 bradford c. m. , et al . 2009 , apj , 705 , 112 burke & hollenbach 1983 , apj 265 , 223 dabringhausen j. , kroupa p. , & baumgardt h. 2009 , mnras , 394 , 1529 downes d. , & solomon p. m. 1998 , apj , 507 , 615 eales s. , lilly s. , gear w. et al . 1999 , apj , 515 , 518 elmegreen b. g. 1989 , apj , 338 , 178 elmegreen b. g. , klessen r. s. , & wilson c. d. 2008 , apj , 681 , 365 espinoza p. , selman f. j. , & melnick j. 2009 , a&a , 501 , 563 frayer d. t. , ivison r. j. , scoville n. z. , yun m. , evans a. s. , smail ian , blain a. w. , & kneib j .- p 1998 , apj , 506 , l7 frayer d. t. et al . 1999 , apj , 514 , l13 genzel r. , lutz d. , sturm e. , et al . 1998 , apj , 498 , 579 genzel r. , tacconi l. j. , rigopoulou d. , lutz d. , & tecza m. 2001 , apj , 563 , 527 goldsmith p. f. , & langer w. d. 1978 , apj , 222 , 881 goldsmith p. f. 2001 , apj , 557 , 736 greve t. r. , bertoldi f. , smail i. et al . 2005 , mnras , 359 , 1165 greve t. r. , papadopoulos p. p. , gao y. , & radford s. j. e. 2009 , apj , 692 , 1432 groenewegen m. a. t. 1994 , a&a , 290 , 531 hocuk s. , & spaans m. 2010 , a&a , ( in press , arxiv:1008.5284 ) hollenbach d. j. , takahashi t. , & tielens a. g. g. m. 1991 , apj , 377 , 192 hollenbach d. j. , & tielens a. g. g. m. 1999 , reviews of modern physics , vol . 1 , january 1999 jasche j. ciardi b. , & ensslin t. a. 2007 , mnras , 380 , 417 jappsen a .- k . , klessen r. s. , larson r. b. , et al . 2005 , a&a , 435 , 611 kim s. s. , figer d. f. , kudritzki r. p. , & najarro f. 2006 , apj , 653 , l113 klessen r. s. 2004 , astrophysics & space science 292 , 215 klessen r. s. , spaans m. , & jappsen a .- k . 2007 , mnras , 374 , l29 kppen j. , weidner c. , & kroupa p. 2007 , mnras , 375 , 673 krumholz m. r. , & mckee c. f. 2005 , 630 , 250 larson r. b. 2005 , mnras , 359 , 211 lehnert m. d. , & heckman t. m. 1996 , apj , 472 , 546 li y. , klessen r. s. , mclow m .- m . 2003 , apj , 592 , 975 liseau r. et al . 1999 , a&a 344 , 342 lckmann u. , baumgardt h. , & kroupa p. 2010 , mnras , 402 , 519 mckee , c. f. 1989 , apj , 345 , 782 nayakshin s. , sunayev r. 2005 , mnras , 364 , l23 papadopoulos , p. p. 2010 , apj , 720 , 226 pellegrini e. w. , baldwin j. a. ferland g. j. , et al . 2009 , apj , 693 , 285 ossenkopf v. , rollig m. , cubick m. , & stutzki j. 2007 , in molecules in space and laboratory , meeting held in paris , france , editors : j.l . lemaire , f. combes , pg . 95 pan l. , & padoan p. 2009 , apj , 692 , 594 rieke g. h. , lebofsky m. j. , thompson r. i. , low f. j. , & tokunaga a. t. 1980 , apj , 239 , 24 rieke g. h. , loken k. , rieke m. j. , & tamblyn p. 1993 , 412 , 99 rllig m. , et al . 2007 , a&a , 467 , 187 rowan - robinson m. 1980 , apjs , 44 , 403 sakamoto k. , et al . 2008 , apj , 684 , 957 sanders d. b. & mirabel f. 1996 , ara&a , vol . 34 pg . 749 sanders . d. b. , & ishida c. 2004 , asp conference series , vol . 230 skilling j. , & strong a. w. 1976 , a&a , 53 , 253 smail i. 2002 , royal society of london philosophical transcations a , vol . 360 , issue 1801 , pg . 2697 solomon p. m. , downes d. , radford s.j.e . , & barrett j. w. 1997 , apj , 478 , 144 spaans m. , & silk j. 2000 , apj , 538 , 115 spaans m. , & silk j. 2005 , apj , 626 , 644 stolte a. , grebel e. k. , brandner w. , figer d. f. 2002 , a&a , 394 , 459 stolte a. , brandner w. , grebel e. k. , et al . 2005 , apj , 628 , l113 suchkov a. , allen r. j. , & heckman t. m. 1993 , apj , 413 , 542 swinbank a. m. , et al . 2010 , nature , 464 , 733 tacconi l. j. , neri r. , chapman s. c. , et al . 2006 , apj , 640 , 228 thompson t. a. 2009 , _ the starburst - agn connection _ asp conference series , vol . 408 , pg . 128 thi w .- f . , van dishoeck e. f. , bell t. , et al . 2009 , mnras , 400 , 622 van der werf p. p. , and the hercules team , 2010 , a&a , 518 , l42 wolfire m. g. , et al . 1995 , apj , 443 , 152 yusef - zadeh f. , wardle m. , & roy s. 2007 , apj , 665 , l123 the cr heating rate used in this work is ( wolfire et al . 1995 ) where @xmath167@xmath40@xmath39 being the cosmic ray ionization rate ( in s@xmath64 ) , with a adopted galactic value of @xmath168=5@xmath210@xmath169 s@xmath64 ( corresponding to @xmath24=@xmath170 ) , and @xmath171=2n(h@xmath36 ) for fully molecular gas cores . for an optically thin oi line the cooling is @xmath172 , which becomes computed in a similar fashion as the oi(63@xmath51 m ) line cooling , and for a fully transparent medium . gas - grain accommodation can heat or cool the gas depending on the difference between their temperatures ( burke & hollenbach 1983 ) . it can be expressed as where @xmath184 is the number density of dust grains and @xmath185 is the grain cross - section ( cm@xmath142 ) . if the gas - to - dust ratio is 100 , the dust mass density is @xmath187=3.5g@xmath14 , and a dust radius of @xmath188 0.1 @xmath189 m then which we set to the maximum value @xmath60=0.35 ( i.e. maximum gas cooling from gas - dust interaction ) . thus the mainly cooling term ( for the ism conditions explored here ) due to the gas - dust interaction in equation a6 becomes where we set @xmath195}}$]=(0.97,0.98,0.99,1.0 ) for densities of 5@xmath210@xmath196 , 10@xmath197 , 10@xmath181 , 10@xmath198 @xmath14 respectively ( with the value of = 1 corresponding to all carbon locked onto co ) . solutions of equation 2 in the main text for densities spanning the range present in star - forming h@xmath36 clouds in galaxies , and @xmath0=(0.2 , 1.0 , 10 , 10@xmath199 , 10@xmath3 , 10@xmath4)@xmath105@xmath200 are then obtained numerically and are shown in figure 1 .
we examine in detail the recent proposal that extreme cosmic - ray - dominated - regions ( crdrs ) characterize the ism of galaxies during events of high - density star formation , fundamentally altering its initial conditions ( papadopoulos 2010 ) . solving the coupled chemical and thermal state equations for dense uv - shielded gas reveals that the large cosmic ray energy densities in such systems ( @xmath0@xmath1 few@xmath2(10@xmath310@xmath4)@xmath5 ) will indeed raise the minimum temperature of this phase ( where the initial conditions of star formation are set ) from @xmath610k ( as in the milky way ) to @xmath650100k . moreover in such extreme crdrs the gas temperature remains fully decoupled from that of the dust , with @xmath7@xmath8@xmath9 , even at high densities ( @xmath10@xmath1110@xmath1210@xmath13@xmath14 ) , quite unlike crdrs in the milky way where @xmath15@xmath6@xmath16 when @xmath17@xmath1810@xmath12@xmath14 . these dramatically different star formation initial conditions will : a ) boost the jeans mass of uv - shielded gas regions by factors of @xmath1110100 with respect to those in quiescent or less extreme star forming systems , and b ) `` erase '' the so - called inflection point of the effective equation of state ( eos ) of molecular gas . both these effects occur across the entire density range of typical molecular clouds , and may represent _ a new paradigm for all high - density star formation in the universe , with cosmic rays as the key driving mechanism _ , operating efficiently even in the high dust extinction environments of compact extreme starbursts . the characteristic mass of young stars will be boosted as a result , naturally yielding a top - heavy stellar initial mass function ( imf ) and a bimodal star formation mode ( with the occurance of extreme crdrs setting the branching point ) . such crdrs will be present in local ultra luminous infrared galaxies ( ulirgs ) and merger - driven gas - rich starbursts across the universe where large amounts of molecular gas rapidly dissipate towards compact disk configurations where they fuel intense starbursts . in hierarchical galaxy formation models cr - controlled sf initial conditions lend a physical basis for the currently postulated bimodal imf in merger / starburst versus quiescent / disk star forming environments , while naturally making the integrated galactic imfs ( igimfs ) a function of the sf history of galaxies . [ firstpage ] cosmic rays dust , extinction galaxies : starburst galaxies : star formation galaxies : interaction ism : molecules ism : supernovae remants .
the dynamics of a large diversity of physicochemcal systems can be mathematically modeled as reaction - diffusion systems in which it is described how the composition of multiple chemical species distributed in space change under the influence of competitive chemical reactions between the species ( giving origin to a new species ) and the diffusion which causes the species to spread out in the space . it is well known that depending on the relative importance of the kinetics and the diffusion these systems can provide a large diversity of behaviors , including the formation of complex structures and patterns see @xcite . such a structure formation occurs for example during the solid phase formation and evolution in intercalation and conversion reactions in rechargeable lithium batteries @xcite , during the self - organisation of materials occuring with the fabrication process of composite electrodes for electrochemical devices applications @xcite , during the microstructural evolution of composite elecrodes upon their degradation @xcite and in other competitive chemically reactive systems like in the belousov - zhabotinsky reaction @xcite . + designing appropriate controllers of these reaction - diffusion systems can reveal of great relevance within a reverse engineering approach for example towards the optimization of discharge - charge of lithium batteries ( by for example enhancing the formation of solid phases during discharge more reversible upon charge ) and the optimization of the structure of the fabricated electrodes as function of the fabrication parameters ( e.g. temperature dynamics , reactant flow , etc . ) . + in this paper , we consider the one - dimensional allen - cahn equation @xmath00,1 [ , t > 0,\\ u_x(0,t)=\alpha(t ) , u_x(1,t)=0 & \forall t > 0,\\ u(x,0)=u_0(x ) & x\in ] 0,1[.\end{aligned}\ ] ] + this reaction - diffusion equation describes the process of phase separation in many situations . it was originally introduced in @xcite by allen and cahn to model the motion of anti - phase boundaries in crystalline solids . in equation ( 1 ) , @xmath1 represents the concentration of one of the possible phases , @xmath2 represents the interfacial width , supposed to be small as compared to the characteristic length of the laboratory scale . the homogenous neumann boundary condition ( when @xmath3 ) traduces that there is no loss of mass across the boundary walls however , the allen - cahn equation is invoqued in a large number of complicated moving interface problems in materials science through a phase - field approach , therefore a large litterature in mathematical analysis and in numerical analysis is devoted to the study of the mathematical properties of this equation and of its simulation ( see @xcite and the references therein ) . + in equation ( 1 ) , @xmath4 represents the potential energy and @xmath5 represents the control flux at one of the boundaries ; @xmath4 is assumed having stable roots @xmath6 , @xmath7 such that @xmath8 and @xmath9 . it is observed in many cases that when @xmath10 and as @xmath11 goes to @xmath12 , the solutions tend to steady states @xmath13 which consist in ( almost ) piecewise constant functions whose the different values are equal to the stable roots of @xmath14 which represent the different phase stripes . hence @xmath13 exhibits large gradient near @xmath6 , as illustrated in figure ( [ fig1 ] ) . ( left ) and for @xmath15 ( right).,title="fig:",width=321,height=264 ] ( left ) and for @xmath15 ( right).,title="fig:",width=321,height=264 ] + + an important issue in the conception of rechargeable lithium and post - lithium batteries , is the design of active materials providing upon the battery discharge a number of interphases as low as possible . the morphological simplicity of such discharged materials is expected to enhance the rechargeability of these type batteries and thus to increase their efficiency @xcite . in this paper we propose a first numerical strategy to calculate the boundary flux function @xmath5 on a given time interval @xmath16 $ ] , with @xmath17 large enough , in such a way the number of interphase of the steady state @xmath13 is minimized . to this end , we consider as control function @xmath5 , @xmath2 being constant . for the sake of simplicity , we first restrict ourselves to the case @xmath18 which possesses 3 roots : @xmath19 which are stable and @xmath20 which is unstable . + the article is organized as follows : first , in section 2 , we present first the global numerical strategy by deriving the estimation of the number of interphases , which will be the merit function to minimize . then , we present the finite differences discretization of the system in space and we describe the numerical solver , that includes the optimization process as well as the full discretized problem to be solved at each iteration . in section 3 , we present some numerical results demonstrating the numerical controllability of the problem : we calculate optimal@xmath21 for different values of @xmath22 , @xmath17 and @xmath2 . finally , in section 4 we conclude and indicate further perspectives of development of our work . we consider the finite differences discretization in space of the allen - cahn equation which leads to a differential system . the grid points @xmath23 , @xmath24 are regularly spaced for simplicity , @xmath25 is the corresponding stepsize . we assume that @xmath25 is small enough in order the discrete solution captures the strong gradients near the interphases . the steady solution @xmath13 is considered to be almost piecewise constant , so its approximations at grid points @xmath26 , @xmath24 take the values @xmath27 . hence @xmath28 therefore , the number of interphases is @xmath29 this quantity can be related to the @xmath30-norm of @xmath31 , indeed @xmath32 in figure ( [ fig1 ] ) ( left ) , we count 10 changes , the result given by formula ( [ formula_changes ] ) is 9.9968 and in figure ( [ fig1 ] ) ( right ) 48 changes are counted while ( [ formula_changes ] ) estimation is 47.7475 . + we remark that an interesting numerical issue could be to plug an adaptive grid strategy since the steady solution needs only few points to be represented . our approach applies when more than 2 interphases are present . indeed , consider for simplicity the case of @xmath33 stable phases . to obtain the number of interphases , it is sufficient to split the final signal profile into 4 parts , each of them reprensenting the state of one phase stirp ( see figure below in the case of @xmath34 ) . once done , we can apply formula ( [ formula_changes ] ) separately . this procedure allows using a weighted merit function @xmath35 where @xmath36 is the number of connex components for phase @xmath37 and @xmath38 the associated weight : a large value of @xmath39 enforces the optimal state to provide small number of phases of type @xmath37 . so it is possible to select a given profile . it has to be pointed out that @xmath40 , the total number of interphases , in fact @xmath41 , however the simplified summation formula allows selecting easily given phase stripes ( figure ( [ fig3 ] ) ) . 0.2 cm + 0.2 cm when the different phases @xmath42 are known _ a priori _ , the merit function can be defined more precisely as @xmath43 for instance , with @xmath44 and @xmath18 , the two stable phases are @xmath45 and the merit function associated to @xmath46 is @xmath47 which can be considered directly in the continuous case with , e.g. , @xmath48 norm , giving rise to the merit function @xmath49 . we denote by @xmath50 the approximation of @xmath51 generated by the semi discrete scheme @xmath52 in which we have implemented neumann boundary condition to calculate the values @xmath53 and @xmath54 . as a time marching scheme , we will use a semi implicit one , in order to have a good stability : it is important since , as we will see hereafter , the calculation of optimal @xmath5 requires a great number of numerical solutions of this system and a not too small time step @xmath55 must be used . we fix a value for the final time @xmath17 , @xmath17 being large enough to obtain a steady state : in practice the solution converges toward equilibrium relatively fastly for small values of @xmath2 , which is the case here . the time interval @xmath16 $ ] is subdivided into @xmath56 subintervals of length @xmath55 , @xmath57 $ ] , @xmath58 , with @xmath59 and @xmath60 . we note @xmath61 the numerical approximation to @xmath13 . a first idea is to compute @xmath5 as a piecewise constant function in time , say @xmath62}}.\end{aligned}\ ] ] + we remark here that calculating @xmath5 as a piecewise constant function , is a simple and stable approach : other techniques allowing orthogonal polynomial ( such as fourier or laguerre ) could be used but the strong decreasing of the fourier coefficients make the problem ill - conditioned . also , an heuristic method can be used for accelerating the numerical convergence : once computed @xmath63 for a given @xmath56 and @xmath55 , one can repeat the computation for @xmath64 time steps ( with @xmath65 ) , in order to get a more precise result @xmath66 , starting from a mid - point interpolated value . + the problem we want to solve is then expressed as + @xmath67 + we can now describe the global approach under the following algorithm initialization : start from an initial guess @xmath68}}$ ] compute @xmath61 by time integration of ( [ eq1])-([eq4 ] ) with @xmath69}}\ ] ] compute @xmath70 . update @xmath71 from @xmath72 by a derivative free optimization process ( nonlinear search ) also , we will have to establish numerical convergence by varying @xmath56 , @xmath55 . we subdivide @xmath16 $ ] into @xmath56 subintervals of length @xmath55 and note @xmath73 the approximation of @xmath74 at time @xmath75 . let @xmath76 be the discretization matrix of the negative seconde derivative en @xmath77 with homogeneous neumann boundary conditions and @xmath78 . we consider the following linearized implicit euler scheme which reads , after usual simplifications as @xmath79 where @xmath80 . we have the assume @xmath81 . if @xmath82 and if @xmath83 then the sequence @xmath84 defined by the scheme ( [ scheme1 ] ) satisfies @xmath85 . if @xmath86 , we set @xmath87 and @xmath88 . we ll show by induction that @xmath89 and @xmath90 . let us fix @xmath91 and assume that @xmath92 says @xmath89 and @xmath93 . we have , after simplifications @xmath94 if @xmath89 , then @xmath95 , so the matrix @xmath96 has the discrete maximum property ( @xmath97 is a m - matrix ) and since @xmath89 and @xmath98 , we have @xmath99 so that @xmath100 we proceed in a similar way for @xmath101 . as , we will see hereafter in the numerical simulations , this is observed for moderate values of @xmath21 . + it is important to underline that it is crucial that the numerical scheme heritates of the intrinsic properties of the equation ( maximum principle , asymptotic behavior ) for calculating a numerical control : if not , non - physical control can be determined@xcite . as pointed out before , the principal key of this approach is the choice of the optimization method to compute @xmath5 . this is not an easy task since the merit function is not differentiable in the @xmath30 case ; generally the gradient is not available as in the @xmath48 case . hence , gradient methods can not be used . we have then to address to derivative free optimization algorithms @xcite . + in order to illutrate the numerical cacutation of optimal @xmath21 , we use matlab build - in house solver , such as x = fminsearch(fun , x0 ) or x = fminunc(fun , x0 ) where fun is the objective function which in our case is built numerically by solving the allen cahn equation up to @xmath102 with @xmath21 as an input . the minimization procedure used is fminunc available in from the optimization toolbox of matlab computational software , @xcite . we fix @xmath2 and we start from @xmath103 $ ] . we hereafter display results for different values of @xmath2 and @xmath17 . the merit function is @xmath70 . starting from a regular initial data allows to observe numerical convergence of the optimal control function @xmath5 as the number of discretisation points @xmath104 increases , and as , for a fixed @xmath17 , the time step @xmath105 decreases . we can see , in figures ( [ fig4])-([fig5])-([fig6])-([fig7 ] ) the good coherence of the results for fixed values of @xmath106 and @xmath17 when varying @xmath55 and the number @xmath104 of grid points . in all the cases , the global procedure allows minimizing the number of interphases or stripes . , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] here , in figures ( [ fig8 ] ) and ( [ fig9 ] ) the discrete components of @xmath22 are randomly calculated following a uniform law on @xmath107 $ ] and we start from @xmath103 $ ] . as we see , here again , in all cases , the global procedure , illustrating the effective numerical controllability by the boundaries and the robustness of the approach , since the data are very oscillating . , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] finally , we give hereafter a numerical illustration of the optimization process when considering a weighted merit function as in ( [ weighted ] ) : we adopt here @xmath108 . as we see in figure ( [ fig10 ] ) , the merit function favors the formation of the phase @xmath109 , the initial datum is , as above , randomly generated on @xmath107 $ ] by an uniform law . , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] + , title="fig:",width=302,height=264 ] , title="fig:",width=302,height=264 ] in this paper , we have presented a simple approach to calculate numerically a boundary control that allows obtaining an optimal steady - sate configuration , i.e. , with a minimal number of interphases . we have also demonstrated that we can also favor the formation of a given phase by following the same procedure . the results we obtained are encouraging and show the numerical faisability of the proposed method .. of course , we have here considered first a relatively simple case , namely the one dimensional case before extending the approach to 2d or 3d models which correspond to more realistic situations found , for example , in electrochemistry . furthermore , the monitoring of the number of interphases by @xmath2 ( problem 2 ) is an important feature that we will study in a near future . finally the integration of such optimization algorithms in an in - house multiscale simulator of electrochemical power generators will be also considered @xcite . 99 s. m. allen and j. w. cahn . a microscopic theory for antiphase boundary motion and its application to antiphase domain coarsening . acta metall . ( 27 ) ( 1979 ) , pp 1085-1095 . a. r. conn , k. scheinberg , and l. n. vicente , _ introduction to derivative - free optimization _ , mps - siam series on optimization , siam , philadelphia , ( 2009 ) . a.a . franco , k.h . xue , carbon - based electrodes for lithium air batteries : scientific and technological challenges from a modeling perspective , ecs journal of solid state science and technology , 2 ( 10 ) ( 2013 ) , pp m3084-m3100 . a.a . franco , multiscale modeling and numerical simulation of rechargeable lithium ion batteries : concepts , methods and challenges , rsc advances , 3 ( 32 ) ( 2013 ) , pp 1302713058 . franco , ms liber t computational software , http://www.modeling-electrochemistry.com/ms-liber-t/ ignat l. i. , pozo a. , zuazua e. large - time asymptotics , vanishing viscosity and numerics for 1-d scalar conservation laws , submitted ( 2013 ) + http://www.bcamath.org/documentospublic/archivos/publicaciones/asymptoticsinnumerics.pdf matlab s optimization toolbox , http://www.mathworks.fr/fr/products/optimization/index.html k. malek , m. eikerling , q. wang , t. navessin , z. liu , self organization in catalyst layers of pem fuel cells , j. phys . c , 111 ( 36 ) , ( 2007 ) , pp 13627-13634 . k. malek , a.a . franco , microstructural resolved modeling of aging mechanisms in pemfc , j. phys . b , 115 ( 25 ) ( 2011 ) , pp 8088-8101 . m. pierre , _ etude numrique et mathmatique de quelques modles de transition de phase , de sparation de phases et de cristaux liquides _ , habilitation diriger les recherches , ( in french ) , universit de poitiers ( oct . d. kondepudi , i. prigogine , modern thermodynamics : from heat engines to dissipative structures , john & wiley sons ltd . , new york ( 1998 ) . c. sachs , m. hildebrand , s. vlkening , j. wintterlin , g. ertl , science ( 293 ) no . 5535 ( 2001 ) , pp . sirimungkala a. , frsterling h - d . , dlask v. , bromination reactions important in the mechanism of the belousov-zhabotinsky system " . a 103 ( 8) , ( 1999 ) , 1038-1043 . j. shen , x. yang , numerical approximations of allen - cahn and cahn - hilliard equations . dcds , series a , ( 28 ) , ( 2010 ) , pp 16691691 .
the calculation of optimal structures in reaction - diffusion models is of great importance in many physicochemical systems . we propose here a simple method to monitor the number of interphases for long times by using a boundary flux condition as a control . we consider as an illustration a 1-d allen - cahn equation with neumann boundary conditions . numerical examples are given and perspectives for the application of this approach to electrochemical systems are discussed .
many lower - dimensional antiferromagnets show a singlet ground state and an energy gap in their excitation spectrum originating from quantum many body effects . the first indication of the quantum energy gap came out in 1983 . haldane@xcite conjectured that the excitation spectrum of one - dimensional ( 1d ) heisenberg antiferromagnets ( hafs ) with integer spin quantum number ( @xmath0 ) has an energy gap between the ground and first excited states , while the corresponding system with half - odd integer @xmath0 has no energy gap . this conjecture has been tested both theoretically@xcite and experimentally@xcite . it is now generally accepted that the quantum energy gap ( haldane gap ) does exist in 1d hafs with integer @xmath0 . in real materials , there always exists an interaction between spin chains . we call this class of materials a quasi - one - dimensional ( q1d ) magnet . usually q1d antiferromagnets exhibit a long - range ordering ( lro ) at finite temperature due to the inter - chain coupling ( @xmath11)@xcite . the compound ni(c@xmath3h@xmath12n@xmath3)@xmath3no@xmath3(clo@xmath13 ) , abbreviated nenp , is a typical example of @xmath0=1 q1d hafs@xcite and does not show any indication of lro down to 300 @xmath14k@xcite , whereas the @xmath0=1 q1d haf csnicl@xmath4 shows lro at @xmath154.7 k@xcite . the different behavior between the two compounds comes from the difference in the strength of @xmath11 . how robust is the haldane phase in a q1d haf against perturbation ? sakai and takahashi@xcite studied theoretically the ground state properties of an @xmath0=1 q1d haf with a single - ion anisotropy of the form @xmath16 . they showed that the haldane disordered phase exists for @[email protected] in a rather wide range of @xmath19 values , where @xmath20 is the number of neighboring chains and @xmath21 is the intra - chain exchange interaction . as was demonstrated experimentally@xcite , strong magnetic fields destroy the haldane gap and the system recovers magnetism . then , we expect a magnetic ordering to occur in an @xmath0=1 q1d haf under high fields and at low temperatures . the heat capacity measurement made on nenp in magnetic fields did not reveal any indication of lro@xcite . one of the reasons for this is that an external magnetic field applied along the chain axis of nenp induces a staggered field on the ni@xmath22 sites because of the presence of two crystallographically inequivalent sites for ni@xmath22@xcite . this staggered field causes a small energy gap near the transition field from the disordered to the magnetized states@xcite and thus prevents the occurrence of lro at low temperatures . we have reported the observation of a field - induced lro in the @xmath0=1 q1d haf ni(c@xmath1h@xmath2n@xmath3)@xmath3n@xmath4(clo@xmath13 ) , abbreviated ndmaz@xcite , and in ni(c@xmath1h@xmath2n@xmath3)@xmath3n@xmath4(pf@xmath5 ) , abbreviated ndmap@xcite . since there is only one site for ni@xmath22 in ndmaz and mamap , these compounds are ideal for studying the field - induced lro . from a heat capacity ( @xmath23 ) measurement on a single crystal sample of ndmaz , we have observed an anomaly at about 0.6 k and 12 t@xcite which indicated that a magnetic ordering occurred there . because of limitations in our calorimeter , we were unable to follow how the position of the anomaly in @xmath23 changes with temperature ( @xmath24 ) and magnetic field ( @xmath6 ) . we then tried to synthesize the @xmath0=1 q1d haf , ndmap , with a weaker @xmath21 than ndmaz so that lro is expected to be induced at a lower field . we have measured @xmath24 and @xmath6 dependence of @xmath23 in a single crystal sample of ndmap and constructed the @xmath6-@xmath24 phase diagram@xcite . we found an anisotropy in the phase boundary curve separating the paramagnetic and lro phases , and we presented a qualitative interpretation for this anisotropy@xcite . electron spin resonance ( esr ) measurements made on a single crystal of ndmap gave further evidence for the existence of the field - induced lro phase and of the anisotropy in the phase boundary curve.@xcite in this paper we report detailed results of heat capacity and magnetization measurements made on ndmap . we also report a theoretical analysis of the @xmath6-@xmath24 phase diagram . the format of this paper is as follows . in sec.ii we present the relevant background information and details . the experimental results are given in sec.iii . in sec.iv , a theoretical consideration is given on the @xmath6-@xmath24 phase diagram . the last section ( sec.v ) is devoted to discussion and conclusions . the compound ni(c@xmath1h@xmath2n@xmath3)@xmath3n@xmath4(pf@xmath5 ) ( ndmap ) has the orthorhombic structure@xcite with the space group @xmath25 shown in fig . 1 . the lattice parameters are , @xmath8= 18.046 @xmath26 , @xmath9= 8.7050 @xmath26 and @xmath10= 6.139 @xmath26@xcite . the structure consists of ni(c@xmath1h@xmath2n@xmath3)@xmath3n@xmath4 chains along the @xmath10 axis . these chains are well separated from each other by pf@xmath5 molecules . all the ni@xmath22 sites in a chain are equivalent , @xmath27 , only one site exists for ni@xmath22 . from the analysis of the magnetic susceptibility data , the following values are obtained@xcite ; @xmath28= 30.0 k , @xmath29= 0.3 , @xmath30= 2.10 and @xmath31= 2.17 , where @xmath30 and @xmath31 are the @xmath32 values parallel and perpendicular to the chain @xmath10 axis , respectively . neutron inelastic scattering measurements were done on single crystals of deuterated ndmap at @xmath24= 1.4 k @xcite . from the analysis of the data , the values of exchange and anisotropy parameters are determined to be , @xmath21= 2.28 mev (= 26.5 k ) , @xmath33= 3.5@xmath3410@xmath35 mev (= 4.1@xmath3410@xmath36 k ) , @xmath37= 1.8@xmath3410@xmath36 mev (= 2.1@xmath3410@xmath38 k ) and @xmath19= 0.70 mev (= 8.1 k ) , where @xmath33 and @xmath37 are the inter - chain exchange interactions along the @xmath8 and @xmath9 axes , respectively . the single crystals of ndmap used in this study were grown from an aqueous solution of nan@xmath4 , ni(no@xmath4)@xmath3@xmath396h@xmath3o and 1,3-diamino-2,2-dimethylpropane . after filtration , kpf@xmath5 was added to the solution . well shaped blue single crystals up to 5mm@xmath345mm@xmath3420 mm were obtained after several weeks . fully deuterated single crystals of ndmap ( ndmap - d@xmath40 ) were grown from a d@xmath3o solution of the same ingredients except that a deuterated 1,3-diamino-2,2-dimethylpropane was used . the single crystals thus obtained were checked by a four circle x - ray diffractometer . we confirmed that the lattice parameters of our crystal are almost identical with those reported before@xcite . we measured @xmath23 of ndmap and ndmap - d@xmath40 and found that the @xmath6-@xmath24 phase diagrams of the two systems are essentially the same . this means that the magnetic parameters in ndmap and ndmap - d@xmath40 are not different . heat capacity measurements were performed with a maglab@xmath41 microcalorimeter ( oxford instruments , uk ) . the temperature and magnetic field ranges accessible with this calorimeter are , 0.45 k@xmath18@xmath24@xmath18 200 k and 0@xmath18@xmath6@xmath1812 t. magnetization measurements were done with a maglab@xmath42 vibrating - sample - magnetometer ( oxford instruments , uk ) . the temperature and magnetic field ranges available with this magnetometer are , 1.5 k@xmath18@xmath24@xmath18300 k and 0@xmath18@xmath6@xmath1812 t. figure 2 shows the temperature dependence of the heat capacity of ndmap , including the contribution of the lattice measured in zero field . the data are well expressed by the following equation , @xmath43 in the temperature range between 2 and 5 k with @xmath8= 0.109 and @xmath9= 0.00653 . we use hereafter this @xmath44 term to subtract the contribution of lattice heat capacity as has been done by many authors . we show in figs . 3(a)-(c ) the temperature dependence of magnetic heat capacity ( @xmath45 ) of ndmap , after subtracting the lattice heat capacity , in magnetic fields applied parallel to the @xmath8 , @xmath9 , and @xmath10 axes , respectively . in all field directions , we see an anomaly in @xmath45 at finite fields above a critical value . this anomaly signals that a magnetic ordering occurs there . one of the advantages of our calorimeter is that a field dependent @xmath23 can be measured under a constant temperature@xcite . strictly speaking , we need to change temperature to measure @xmath23 . however , the temperature increment necessary for the measurement is 0.5 - 1 % of the temperature we set so that temperature change during the measurement may be considered as small . figures 4(a)-(c ) show such `` field scan '' data measured at several temperatures for the magnetic field directions parallel to the @xmath8 , @xmath9 , and @xmath10 axes , respectively . in addition to the sharp peak at the field denoted by @xmath46 , a broad feature is seen around the field named as @xmath47 . here , @xmath46 means the position of @xmath6 at which the field - induced lro occurs for a given @xmath24 . combining all the information obtained from the heat capacity measurements , both of the `` temperature scan '' and `` field scan '' procedures , we present the @xmath6-@xmath24 phase diagram of ndmap in fig . in addition to the anisotropic phase boundary ( curve ( a ) ) separating the disordered and lro phases@xcite , we have another boundary ( curve ( b ) ) separating the haldane and the disordered phases which is also anisotropic . the two curves ( a ) and ( b ) seem to merge at a finite @xmath6 when extrapolated to @xmath24= 0 k , for respective field directions . figures 6(a)-(c ) show the temperature dependence of susceptibility ( magnetization divided by applied magnetic field , @xmath48 ) in ndmap measured in magnetic fields applied parallel to the @xmath8 , @xmath9 , and @xmath10 axes , respectively . the behavior of the susceptibility at @xmath6= 1 t is reminiscent of that taken at a much lower field ( @xmath6= 0.01 t)@xcite ; a broad peak around 35 k and a steep decrease in susceptibility with decreasing temperature below about 20 k. on increasing @xmath6 , @xmath48 does not extrapolate to zero with @xmath24@xmath490 . this behavior of @xmath48 is similar to the one observed in an @xmath0=@xmath50 1d haf@xcite and indicates that a transition from the gapped to a gapless phase occurs at a higher field . on increasing @xmath6 further , @xmath48 shows a minimum and an up turn at low temperatures . the insets of figs . 6(a)-(c ) show the low temperature part of the data . we see in the inset of figs . 6(a ) and ( b ) that @xmath48 becomes almost temperature independent below a temperature whose value is field dependent . the temperature independent susceptibility reminds us of the perpendicular susceptibility ( @xmath51 ) of an anisotropic antiferromagnet below the nel temperature ( @xmath7 ) . we show below that we are actually observing @xmath51 in this compound . because the sign of the single - ion anisotropy term ( @xmath16 ) is positive in this compound , spins in the ordered phase are expected to lie in a plane perpendicular to the @xmath10 axis ( the quantization axis of the @xmath19 term is taken parallel to the @xmath10 axis ) . the anisotropy in the @xmath10 plane of the form @xmath52 is very small@xcite . therefore , when @xmath6 is applied along the @xmath8 or @xmath9 axes , spins point perpendicularly to @xmath6 in the @xmath10 plane keeping an antiferromagnetic arrangement , thus giving @xmath51 . we plot in fig . 5 the transition points obtained from the @xmath48 data shown in fig . these points are defined as the temperatures where @xmath53 shows a minimum for a given @xmath6 . we see that the transition points determined from the heat capacity and magnetization measurements agree well with each other . in this section , we try to reproduce the @xmath6-@xmath24 phase diagram observed in ndmap , exhibiting an interesting behavior of @xmath7 as a function of @xmath6 : at low temperatures , there occurs lro only above a certain critical field and @xmath7 shows an increase with increasing fields , the rate of which depends on the direction of @xmath6 . we focus our attention especially on the physics behind the phase diagram . to this end , we adopt the mean - field approximation for the interchain interaction@xcite , which is known to work quite well except for fields in the vicinity of the critical field . in the following , we use the energy unit @xmath21=1 , so that @xmath54 , @xmath55 , @xmath56 and @xmath57 are renormalized quantities of @xmath24 , @xmath58 , @xmath11 and @xmath19 , respectively . according to the mean - field theory , the renormalized nel temperature @xmath59 as a function of the renormalized field @xmath55 is given by the solution satisfying the following equation : @xmath60 where @xmath61 is the staggered susceptibility for the _ one - dimensional _ magnetic system . then , we calculate the staggered susceptibility of the magnetic chain by means of the quantum transfer matrix method combined with the finite - temperature density matrix renormalization group . as was mentioned in the previous section , ni@xmath22 spins in ndmap has an easy - plane anisotropy , and thus the magnetic chain is well described by the following hamiltonian : @xmath62,\ ] ] where @xmath63 represents the spin-@xmath64 operator at the @xmath65th site . neglecting the small anisotropy in the @xmath10 plane , we consider the following two cases for the field @xmath55 applied ( i ) along the @xmath20(@xmath10)-axis ( perpendicular to the easy plane ) and ( ii ) along the @xmath66-axis ( in the easy plane ) . it is noted here that the situation is quite different between the two cases , since the former field reserves the axial symmetry around the @xmath10-axis while the latter breaks it . the nature of quantum fluctuations and hence the staggered susceptibility depends crucially on the symmetry of the system . in the following , we consider the two cases separately . \(i ) _ the field applied perpendicularly to the easy plane _ ( @xmath67 ) + as was mentioned , below the critical field @xmath68 the nonmagnetic haldane phase called a quantum disordered phase is the ground state of the system , while above it the so called tomonaga - luttinger liquid state becomes the ground state . although the former has an excitation gap to the triplet state , the latter has a gapless excitation spectrum and hence is critical . this criticality of the tomonaga - luttinger liquid is characterized by the critical exponent @xmath69 defined by the divergence of the staggered susceptibility at zero temperature@xcite : @xmath70 where @xmath71 is a constant , which scarcely depends on @xmath55 in our calculation . note that in the classical system @xmath72 . we may safely use eq . ( 4 ) at low temperatures well below the temperature at which the susceptibility is maximum ( @xmath1535 k ) . in fig . 7 , @xmath69 s , estimated from the numerical calculations for the temperature region @xmath73 , are shown by the solid circles , each of which has an error bar of @xmath74 . the data have been shifted in fields so that the theoretical critical field coincides with the one observed . the solid curve represents the phenomenological relation between @xmath69 and @xmath75 : @xmath76 with @xmath77 and @xmath78 . now , we reproduce the phase diagram , using eqs . ( 2 ) and ( 4 ) with @xmath79k@xcite . in fig . 8 we show the theoretical result by the solid curve and the experimental data by the solid circles . here , we adjusted the theoretical curve to reproduce the experimental point @xmath80 k at 11 t. from figs . 7 and 8 , we see that the increase of @xmath7 with @xmath6 is a consequence of the decrease in @xmath69 with @xmath6 . remembering that @xmath69 measures the degree of quantum fluctuations , we can say that the quantum fluctuation out of the easy plane is reduced by @xmath6 so that the nel state becomes more stable . in contrast to this quantum system , the field dependence of the constant @xmath71 is an only source of the field dependence of @xmath7 in the classical system , being very mild . we show in fig . 7 @xmath69 estimated from the phase diagram ( fig . 5 ) by the open circles , which follows also the phenomenological relation but @xmath81 . considering the large error bars in @xmath69 of our estimations , we do not think the discrepancy in @xmath82 so seriously . observation of the field dependence of @xmath69 by other methods is desired . in stronger fields , spins ca nt in the field direction and hence @xmath71 decreases seriously . thus , the phase boundary curve closes at the upper critical field , where the magnetic moment saturates . from the phase diagram , we estimate the interchain coupling @xmath83 to be @xmath84 , using the value @xmath85 . this value is favorably compared with that obtained from the neutron inelastic scattering experiments ( 2@xmath86 + 2@xmath87=@xmath88@xcite . \(ii ) _ the field applied in the easy plane _ ( @xmath89 ) + in this case , above @xmath68 , the ground state has the nel order and a gap opens again in the excitation spectrum because of the symmetry breaking field . this situation is quite different from the former case . the field dependence of @xmath7 reminds us of the _ soliton scenario _ in the classical system@xcite . remembering the form of the staggered susceptibility in the classical system , we postulate the following in this case : @xmath90 where @xmath91 is a constant . this form is confirmed by our numerical calculations , in the temperature region @xmath73 , with the coefficient @xmath92 , a little less than unity . although @xmath92 may represent quantum effects in the formation energy of soliton , we assume , for simplicity , the classical value 1 for it . now , we again reproduce the phase diagram for this case using eqs . ( 2 ) and ( 5 ) . the theoretical curves are adjusted as before using the experimental point @xmath93 k at 12 t and the critical field value @xmath94 t for @xmath55 parallel to the @xmath8-axis , and @xmath95 k at 12 t and @xmath96 t for @xmath55 parallel to the @xmath9-axis , respectively . the different values of @xmath97 are due to an in - plane anisotropy , being neglected in this paper . the phase boundary curves are shown by the solid curve for @xmath55 parallel to the @xmath8-axis and by the dotted curve for @xmath55 parallel to the @xmath9-axis with the corresponding experimental points , respectively , by the solid and the open circles in fig . the agreement between theory and experiment is satisfactory . we mention that the soliton scenario still works in our quantum system : the symmetry breaking field yields the uniaxial symmetry and hence the soliton is a dominant source for the fluctuations in this system . since the soliton formation energy , i.e. the gap , increases with @xmath98 and hence the staggered susceptibility increases exponentially at low temperatures , @xmath7 shows a rapid increase with @xmath6 , the rate of which is marked contrast with the former case . although the soliton scenario is effective in our case , more sophisticated study is required to establish further a quantum analogue of the soliton in classical spin chains . the phase boundary curve closes also at the upper critical field as in the former case . we estimate the interchain coupling @xmath83 to be @xmath99 , using the value @xmath100 . the value @xmath56 estimated again agrees with the value observed@xcite . iii , we have presented detailed results of heat capacity and magnetization measurements on ndmap from which we have constructed the @xmath6-@xmath24 phase diagram shown in fig . we have been successful in explaining theoretically the phase boundary curve separating the paramagnetic and lro phases in sec . iv using the values of intra - chain exchange interaction and anisotropy constants determined from the neutron inelastic scattering measurements@xcite . the inter - chain exchange interactions estimated theoretically are close to those obtained from the neutron experiment@xcite . we discuss the lower field boundary separating the haldane and paramagnetic phases ( curve ( b ) in fig . we argue below that the anomaly in @xmath45 observed along this curve is due to the field dependence of the first excited triplet@xcite . we analyzed the low temperature part of the magnetic heat capacity data using a two - level system model , with a singlet ground state and the lowest state of the excited triplet with an energy difference ( @xmath101 ) , which gives a schottky type anomaly . we show in fig . 10 , @xmath102 , @xmath103 and @xmath104 thus obtained for the field directions parallel to the @xmath105 , @xmath106 and @xmath107 axes , respectively . the solid curves in fig . 10 represent the theoretical energy level as a function of @xmath6@xcite . here , we used the value determined from the neutron inelastic scattering experiment@xcite for the energy gap in respective field directions at @xmath6@xmath490 . we see in this figure that the agreement between theory and experiment is satisfactory . finally , we discuss the temperature dependence of @xmath48 ( figs . 6(a)-(c ) ) . we calculated @xmath48 as a function of @xmath24 using the quantum transfer matrix method with a density - matrix renormalization group technique . we compare theory and experiment in fig . 11(a ) for selected values of @xmath6 parallel to the @xmath8 axis . here , we used @xmath108=26.5 k obtained from the neutron inelastic scattering study@xcite and @xmath32=2.14 determined from the esr measurement@xcite . we see a good agreement between theory and experiment without any adjustable parameters . we have obtained the value @xmath108=30.0 k from a fitting of the theory with the susceptibility data at high temperature range above about 40 k@xcite . because the lattice parameters change with temperature@xcite , it is not surprizing if the exchange interaction constant determined at low temperatures is different from that at high temperatures . figure 11(b ) shows the case when @xmath6 is applied along the @xmath10 axis . since no esr data are available along this direction , we assumed the value 2.05 for @xmath32 . the agreement between theory and experiment is not as good as in fig . further study is necessary to clarify this point . in conclusion , we have reported detailed results of heat capacity and magnetization measurements on a single crystal sample of the @xmath0=1 q1d haf , ndmap . from these results , we constructed the @xmath6-@xmath24 phase diagram which exhibits the quantum disordered haldane , field - induced lro and thermally disordered paramagnetic phases . the phase boundary curve separating the paramagnetic and lro phases is anisotropic ; the increase of @xmath7 with @xmath6 along the @xmath8 and @xmath9 axes is more rapid than that along the @xmath10 axis . we calculated @xmath7 as a function of @xmath6 by taking into account the inter chain coupling as the form of a mean - field . we first evaluated numerically the staggered susceptibility of the @xmath0=1 1d antiferromagnet for @xmath6 applied perpendicularly to the easy plane , which shows , at low temperatures , a typical divergence of the tomonaga - luttinger liquid with the critical exponent @xmath69 . then , we got a satisfactory agreement with the experimental results using the exchange and anisotropy constants obtained from the neutron scattering experiment@xcite . it is interesting to note that the transition temperature @xmath7 is governed by the critical exponent @xmath69 of the tomonaga - luttinger liquid . on the other hand , for @xmath6 applied in the easy plane , we invoked the soliton scenario and got again a satisfactory agreement with experiment . this work was partially supported by a grant - in - aid for scientific research from the japanese ministry of education , science , sports and culture . z. h. was supported by the research fellowships of the japan society for the promotion of science for young scientists . the computation in this work has been done using the facilities of the supercomputer center , issp , university of tokyo . also at graduate school of science and engineering , saitama university , urawa , saitama 338 - 8570 , japan f. d. m. haldane , phys . lett . * 93*a , 464 ( 1983 ) ; phys . lett . * 50 * , 1153 ( 1983 ) . for a review see , i. affleck , j. phys . : condens . matter * 1 * , 3047 ( 1989 ) . for a review see , k. katsumata , j. magn . magn . mater . * 140 - 144 * , 1595 ( 1995 ) . for a review see , l. j. de jongh and a. r. miedema , @xmath109 @xmath110 @xmath111 @xmath112 @xmath113 @xmath114 ( taylor & francis ltd . , london , 1974 ) . a. meyer , a. gleizes , j. girerd , m. verdaguer and o. kahn , inorg . * 21 * , 1729 ( 1982 ) . j. p. renard , m. verdaguer , l. p. regnault , w. a. c. erkelens , j. rossat - mignod and w. g. stirling , europhys . lett . * 3 * , 945 ( 1987 ) . o. avenel , j. xu , j. s. xia , m - f . xu , b. andraka , t. lang , p. l. moyland , w. ni , p. j. c. signore , c. m. c. m. van woerkens , e. d. adams , g. g. ihas , m. w. meisel , s. e. nagler , n. s. sullivan and y. takano , phys . rev . b*46 * , 8655 ( 1992 ) . d. moses , h. shechter , e. ehrenfreund and j. makovsky , j. phys . c : solid state phys . * 10 * , 433 ( 1977 ) . t. sakai and m. takahashi , phys . rev . b*42 * , 4537 ( 1990 ) . k. katsumata , h. hori , t. takeuchi , m. date , a. yamagishi and j. p. renard , phys . lett . * 63 * , 86 ( 1989 ) . t. kobayashi , y. tabuchi , k. amaya , y. ajiro , t. yosida and m. date , j. phys . . jpn . * 61 * , 1772 ( 1992 ) . m. chiba , y. ajiro , h. kikuchi , t. kubo and t. morimoto , phys . rev . b*44 * , 2838 ( 1991 ) . p. p. mitra and b. i. halperin , phys . lett . * 72 * , 912 ( 1994 ) . t. sakai and h. shiba , j. phys . 63 * , 867 ( 1994 ) . z. honda , k. katsumata , h. aruga katori , k. yamada , t. ohishi , t. manabe and m. yamashita , j. phys . : condens . matter * 9 * , l83 ( 1997 ) ; 3487 ( 1997 ) . z. honda , h. asakawa and k. katsumata , phys . lett . * 81 * , 2566 ( 1998 ) . z. honda , k. katsumata , m. hagiwara and m. tokunaga , phys . rev . b * 60 * , 9272 ( 1999 ) . m. monfort , j. ribas , x. solans and m. f. bardia , inorg . chem . * 35 * , 7633 ( 1996 ) . a. zheludev , y. chen , c. l. broholm , z. honda and k. katsumata , cond - mat/0003223 . we thank g. shirane for suggesting this possibility . see for example , j. p. a. m. hijmans , k. kopinga , f. boersma and w. j. m. de jonge , phys . rev . lett . * 40 * , 1108 ( 1978 ) . t. sakai and m. takahashi , j. phys . * 60 * , 3615 ( 1991 ) . i. harada , k. sasaki and h. shiba , solid state commun . * 40 * , 29 ( 1981 ) . o. golinelli , th . jolicoeur and r. lacaze , j. phys . : condens . matter * 5 * , 7847 ( 1993 ) . z. honda , k. katsumata , and m. hagiwara , unpublished .
detailed results of heat capacity and magnetization measurements are reported on a single crystal sample of the spin @xmath0=1 quasi - one - dimensional heisenberg antiferromagnet ni(c@xmath1h@xmath2n@xmath3)@xmath3n@xmath4(pf@xmath5 ) , abbreviated ndmap . from these results , we constructed the magnetic field ( @xmath6 ) versus temperature phase diagram of this compound which exhibits the quantum disordered haldane , field - induced long - range - ordered ( lro ) and thermally disordered paramagnetic phases . the phase boundary curve separating the paramagnetic and lro phases is anisotropic ; the increase of the nel temperature ( @xmath7 ) with @xmath6 along the @xmath8 and @xmath9 axes is more rapid than that along the @xmath10 axis . we calculated @xmath7 as a function of @xmath6 by taking into account the inter chain coupling as the form of a mean - field . the calculation of the staggered susceptibility of the @xmath0=1 one - dimensional antiferromagnet with an easy - plane anisotropy shows quite different behavior in the different field directions , resulting in the anisotropic phase boundary curve . a good agreement is obtained between theory and experiment using the exchange and anisotropy constants obtained from the neutron scattering experiment .
gamma ray emission from classical novae is dominated , during the first hours , by positron annihilation following the beta decay of @xmath0f@xcite . however , even though it has been the object of many recent experiments@xcite the rate of its main mode of destruction , through the @xmath0f(p,@xmath1o reaction , remains highly uncertain . this was mainly due to the unknown proton widths of the first three @xmath3ne levels above the proton emission threshold ( @xmath4 , @xmath5 = 6.419 mev , 3/2@xmath6 ; 6.437 mev , 1/2@xmath7 and 6.449 mev , 3/2@xmath6 ) . the tails of the corresponding resonances ( at respectively , @xmath8 = 8 , 26 and 38 kev ) can dominate the astrophysical s - factor in the relevant energy range@xcite . as a consequence of these nuclear uncertainties , the @xmath0f production in nova and the early gamma ray emission was uncertain by a factor of @xmath9300@xcite . since a direct measurement of the relevant resonance strengths is impossible due to the very low coulomb barrier penetrability , we used an indirect method aiming at determining the one nucleon spectroscopic factors in the analog levels of the mirror nucleus ( @xmath3f ) by the neutron transfer reaction d(@xmath0f , p)@xmath3f@xcite . recently , the same reaction has been studied at higher energy@xcite . we present here additional information on the extraction of spectroscopic factors as well as new @xmath0f(p,@xmath1o reaction rates . f excitation energy calibration spectrum with a global fit ( see text ) . the @xmath10 mev level is the most populated.,width=566 ] we refer to de srville et al.@xcite for experimental details and present in figure [ f : spect ] the reconstructed excitation energy spectrum . the well populated @xmath3f levels at @xmath11 and 5.106 mev observed in single events as well as the @xmath12 mev group observed in coincidence events were used to make an internal calibration . this procedure resulted in an uncertainty on the excitation energy calibration of @xmath13 kev in the region of interest . during this analysis a careful study of the systematic errors has been done . to extract the relative contribution of the two 3/2@xmath6 levels , a simultaneous fit of the 6.5 mev group ( @xmath14 = 6.497 + 6.528 mev ) , the 6.25 mev group ( @xmath14 = 6.255 + 6.282 + 6.330 mev ) and the 6.9 mev group ( @xmath14 = 6.787 + 6.838 + 6.891 mev ) was performed . the background is described by a lorentzian of width @xmath15 = 280 kev corresponding to the @xmath16 @xmath17 6.429 mev . the result favors the dominant contribution of one single level ( @xmath10 mev ) while another more recent study @xcite favors the other one . in any case , the nuclear structure of these two 3/2@xmath6 levels seem to be very different according to the results of an inelastic electron scattering measurement on @xmath3f @xcite . from the angular information of the 6.5 mev peak , we have obtained the angular distribution that we have analyzed performing a finite range dwba analysis ( fresco@xcite code ) , including a compound nucleus component . the extracted spectroscopic factor is @xmath18 ( 0.21 when neglecting the compound nucleus contribution@xcite ) and is weakly dependent of the optical potential parameters . it is to be noted that our experiment has been recently repeated at higher energy@xcite and that a slightly lower spectroscopic factor is found ( @xmath19 ) . before deriving a new @xmath0f(p,@xmath1o reaction rate and because the previous analysis showed that the 3/2@xmath6 levels of interest _ can not _ be neglected , it seems important to focus on the remaining uncertainties . first , the spectroscopic factors obtained correspond to @xmath3f levels and should be transposed to the analog @xmath3ne levels in order to deduced the proton widths used to calculate the reaction rate . such a common practice in nuclear astrophysics leads to an uncertainty of a factor of about two on the proton widths ( based on a statistical study of analog states in the same mass region ) . moreover , due to the small separation energy of this doublet ( only 30 kev ) , the assignation of analog levels is not very clear and in the following we will always deal with two cases : _ no inversion _ where the low energy level in @xmath3f is the analog of the low energy one in @xmath3ne and _ inversion _ where the low energy level in @xmath3f is the analog of the high energy one in @xmath3ne . second , the @xmath20widths of the @xmath3ne levels are unknown and calculated from the reduced widths of the corresponding analog levels in @xmath3f . unlike the one nucleon case ( see above ) , the associated uncertainty could be as large as a factor of 10 @xcite . third , the two @xmath21 levels of astrophysical interest can interfere with another @xmath21 level at @xmath22 mev and no information about the sign of interferences is available at present day . in case of destructive interference , the astrophysical s - factor decreases drastically in the gamow region at nova temperature . ne levels @xmath23 and 7.076 mev ( dashed lines ) . cases for different @xmath20width are reported . ( the 6.449 mev spectroscopic factor is from our experiment while data points are from bardayan et al.@xcite . ) . solid lines represent the astrophysical s - factor used for the lower and upper @xmath0f(p,@xmath1o reaction rate ( see text ) . gamow windows for temperatures relevant to nova are shown.,width=566 ] figure [ f : taux ] is an example of the influence on the astrophysical s - factor of the @xmath20width variation when constructive and destructive interferences are considered between the two levels @xmath23 and 7.076 mev ( no inversion ) . experimental data from bardayan et al.@xcite are also displayed . these data points are used to constrain the astrophysical s - factor with the help of r - matrix fits ( anarki @xcite code ) . the free parameters are the @xmath20width of the @xmath24 = 8 or 38 kev resonances , the proton width of the @xmath24 = 330 kev , the proton and @xmath20width and position of the @xmath24 = 665 kev resonance and the sign of the interferences . the best fit is obtained for constructive interferences for the @xmath23 mev level and is then used for determining the new @xmath0f(p,@xmath1o nominal rate . the upper reaction rate is given for a constructive interference for the @xmath25 mev ( inversion ) and an @xmath20width such as the astrophysical s - factor correspond to the upper limit of the error bar of the low energy data point from @xcite ( @xmath26 kev ) . in the same way , the lower rate is given for a destructive interference for the @xmath23 mev ( no inversion ) and an @xmath20width such as the astrophysical s - factor correspond to the lower limit of the error bar at @xmath26 kev . since we consider that the inversion of the analog levels is a possibility , the conclusions are the following . the new nominal rate is within a factor of two of the former one @xcite . furthermore the upper rate is reduced and the global uncertainty is reduced but remains large . this work is based on experimental results obtained in collaboration with c. angulo , m. assuno , d. beaumel , b. bouzid , s. cherubini , m. couder , p. demaret , f. de oliveira santos , p. figuera , s. fortier , m. gaelens , f. hammache , j. kiener , d. labar , a. lefebvre schuhl , p. leleux , m. loiselet , a. ninane , s. ouichaoui , g. ryckewaert , n. smirnova , v. tatischeff and j .- p . thibaud .
the @xmath0f(p,@xmath1o reaction is recognized as one of the most important reaction for nova gamma ray astronomy as it governs the early @xmath2 511 kev emission . however , its rate remains largely uncertain at nova temperatures due to unknown low energy resonance strengths . we report here on our last results concerning the study of the d(@xmath0f , p@xmath1n reaction , as well as on the determination of the @xmath0f(p,@xmath1o reaction rate using the r - matrix theory . remaining uncertainties are discussed .
most of the stars of all masses in the galaxy form in rich clusters . despite this , the details of the clustered star formation process are still poorly understood . studies of _ low - mass protoclusters _ have started to unveil similarities and differences between isolated and clustered dense cores ( e.g. andr et al . @xcite , foster et al . @xcite , friesen et al . globally these studies suggest that cluster environment has a relatively smaller influence on the properties of the cores ( temperature , mass , velocity dispersion , chemical abundances of early phase molecules ) than is typically assumed ( foster et al . however , the conclusions described above do not include observations of _ high - mass star forming regions_. because the phenomena associated with massive star formation have a stronger impact on the environment ( massive outflows , uv radiation , expanding regions ) , it is plausible that these energetic phenomena have major effects on the surrounding dense material . the study of such interaction is especially important to quantify the effect of protostellar feedback on the environment and test recent models of high - mass star formation including outflows and radiation from the newly born stars ( e.g. krumholz et al . 2011 , hennebelle et al . @xcite ) . interferometric observations of dense gas and dust tracers ( , mm continuum , ) have revealed the presence of pre - stellar core candidates surrounding ultracompact regions ( ) and other massive young stellar objects ( ysos ) that do show evidences of such an interaction . for example , fontani et al . ( 2009 ) found that in the protocluster associated with iras 05345 + 3157 the kinematics of two pre - stellar core candidates is influenced by the passage of a massive outflow . uv radiation and powerful outflows affect the chemistry of starless cores in iras 20293 + 3952 ( palau et al . 2007a ) . on the other hand , a crucial chemical process in pre stellar cores , i.e. the deuteration of species like and , seems to remain as high as in pre - stellar cores isolated and associated with low - mass star forming regions ( fontani et al . @xcite , busquet et al . 2010 , pillai et al . therefore , to date it is not clear if and how the presence of massive objects affects the properties and evolution of the other ( pre-)protocluster members . the protocluster associated with ( hereafter ) represents an excellent laboratory to study this issue . the iras source is located to the northeastern side of the cygnus ob2 association , at 1.4 kpc of distance from the sun ( sridharan et al . 2002 , rygl et al . @xcite ) , and two bright nebulous stars , irs 1 ( north ) and irs 3 ( south ) , are found inside the iras error ellipse when observed at high angular resolution ( kumar et al . the bright infrared stars are embedded in a cometary - like cloud whose head , facing the cygnus ob2 association , is bright at centimeter wavelengths and whose tail , bright in the mid - infrared , is extending for about @xmath0 ( @xmath1 pc ) towards the north - east ( fig . [ fig_largefov ] ) . this kind of clouds are also known as bright rimmed clouds . thanks to interferometric observations of and 1.3 mm continuum , palau et al . ( @xcite ) concluded that irs 1 is an intermediate - mass class i yso driving a molecular outflow in the east - west direction , while irs 3 is likely a more evolved intermediate / high - mass star . this is further confirmed through mid - infrared photometric and spectroscopic observations , from which campbell et al . ( @xcite ) also estimated a bolometric luminosity of the order of 1000 for both irs 1 and irs 3 . furthermore , irs 3 is at the centre of an uc -region detected through vla centimeter continuum emission ( carral et al . @xcite ) , and of a fan - shaped emission in the 2.12 @xmath2 m rovibrational line of molecular hydrogen ( kumar et al . @xcite ) . east and west of this fan - shaped feature , palau et al . ( @xcite ) detected molecular gas and dust resolved into several millimeter continuum compact sources . palau et al . ( @xcite ) interpreted these starless condensations as being accumulated on the walls of the expanding shock front , but could not derive firm conclusions on their origin and nature . this work aims at better understading the nature of the dense cores in , and its relation with the neighbouring more evolved objects . to achieve the goal we performed observations of molecular species obtained at high angular resolution with the combined array for research in millimeter astronomy ( carma ) at 3 mm and the very large array ( vla ) at 1.3 cm . all selected molecular transitions are commonly used to characterise dense gas : ( i ) and are excellent tracers of dense and cold cores because either species do not suffer from depletion up to @xmath3 , and is extensively used as thermometer in both low- and high - mass star forming regions ; ( ii ) provides an estimate of the degree of deuteration ( with ) . this combination of diagnostic lines was successfully used by busquet et al . ( 2010 ) to identifying pre - protostellar cores in the proto - cluster associated with iras 20293 + 3952 ; ( iii ) c@xmath4h and are both high - density pdr tracers useful to shed light on the interaction among the cold gas and the uv radiation field coming from irs 1 and irs 3 . is also a tracer of cold gas ( e.g. beuther et al . 2008 , padovani et al . @xcite ) . in this paper we concentrate on the gas morphology , temperature and kinematics of the region adjacent to irs 1 and irs 3 , and confirm the hypothesis proposed by palau et al . ( @xcite ) that irs 3 is opening a cavity in the surrounding dense gas and starless material is being accumulated on the cavity walls . in sect . [ obs ] we describe the observations . the observational results are presented in sect . [ res ] , and discussed in sect . [ discu ] . in sect . [ conc ] we summarise the main findings of the work and give a general conclusion . 3 mm carma observations of were obtained on 29 mar 2010 in c- and 01 may 2010 in d - configuration under good weather conditions for observations at 3 mm , characterized by about 5 mm of precipitable water and atmospheric noise rms of about 300 @xmath2 m as measured on a baseline of 100 m at the frequency of 225 ghz . the phase centre was the same as in palau et al . ( @xcite ) , namely : r.a . ( j2000 ) = @xmath5 and dec . ( j2000 ) = @xmath6.-2pt@xmath72 . the local standard of rest velocity of the cloud is assumed to be 11.5 , as determined from single - dish ammonia observations ( sridharan et al . 2002 ) . the primary beam of the 10 m and 6 m dishes at about 85 ghz is @xmath8 and 121 , respectively . the single - side - band system temperature during the observations was below 150 k. during c - configuration observations , the correlator provided 4 bands which were configured to observe the continuum , the , the and the lines simultaneously . d - array observations were obtained with the new carma correlator , which provides more bands . two 500 mhz bands were used to observe the continuum and 5 bands set up to observe the , , , ccs and line emission . the pass - band was calibrated by observing 1733 - 130 ; flux calibration was set by observing mwc349 . the estimated uncertainty of the absolute flux calibration is 10% , and it is determined from periodic observations of mwc349 . atmospheric and instrumental effects were corrected by observing the nearby quasar 2007 + 404 every 15 minutes . the tracers observed and the main observational parameters ( frequency , synthesised beam , linear resolution , spectral resolution , 1@xmath9 rms channel noise , largest detectable angular scale ) are reported in table [ obs_par ] . the ccs line is the only undetected transition and will be not discussed in the following . the continuum was derived by averaging the 500 mhz bands . visibility data were edited and calibrated with the miriad package . a minor flagging of the data was performed using the uvflag task , mainly to remove the intervals characterized by the bad atmospheric phase coherence . the channel spacing and the corresponding 1 @xmath9 rms noise are shown in table [ obs_par ] . merging the visibilities obtained in c and d configuration , imaging , deconvolution , and analysis of channel maps and continuum were performed using the standard tasks of the gildas package ( e.g. uvmerge , uvmap , clean ) . images were created applying natural weighting to the visibilites . [ cols="<,^,^,^,^,^,^ " , ] in table [ tab_coldensratio ] we report the column densities of , and , and the column density ratios -to- and -to- . the -to- ratio is of order 0.1 ( from @xmath10 to 0.15 ) , and does not change greatly from core to core . such values are much larger than the cosmic d / h ratio , estimated to be @xmath11 ( linsky et al . @xcite ) , and comparable to those measured towards low - mass pre stellar cores ( roueff et al . @xcite ) and infrared dark clouds ( pillai et al . this implies that the deuteration in the cores of is as high as in colder and more quiescent environments , despite the relatively higher gas temperature and turbulence , and confirms previous findings in other dense cores associated with intermediate- to high - mass star - forming regions ( e.g. busquet et al . @xcite , pillai et al . interestingly , relatively high values are found also close to the uc region associated with irs 3 , in mmb and irs3-sw . because is efficiently formed on dust grains , a strong uv radiation can heat the dust and cause evaporation thus increasing its abundance . on the other hand , a strong uv radiation could ( at least partly ) decrease the abundance of , due to its interaction with h@xmath12 to form nh@xmath13 ( e.g. fuente et al . @xcite ) . concerning the -to- ratio , we find the largest value in mme ( @xmath14 ) . the enhancement of the -to- ratio can be understood when freeze - out of species heavier than he becomes important ( see e.g. flower et al . @xcite ) , so that it is expected to increase when the starless core gets closer to the onset of star formation . in this scenario , the fact that mme has the largest -to- ratio suggests that this core could be close to the onset of gravitational collapse , i.e. mme could be a candidate massive pre stellar core . however , palau et al . ( @xcite ) measured with the sma a mass of only 0.7 from the 1.3 mm continuum , while we find 23 . this discrepancy likely comes from extended flux filtered out by the sma , which means that the core is quite flat and not centrally - peaked as expected for a pre stellar core . based on the results of this work , we propose our final interpretation for the nature of each one of the 3 mm condensations : * mma is probably a protostar candidate . although it does not show any embedded infrared source , its relatively high , large line broadening , and -to- lower than in other cores suggest that this condensation is evolved . * mme is likely a pre stellar core , because it shows high -to- and -to- ratios , is more quiescent than mma and it does not appear fragmented into smaller condensations when observed at higher angular resolution ( palau et al . @xcite ) . assuming a typical star formation efficiency of @xmath15% , the core , the mass of which is 23 , has the potential to form an intermediate- to high - mass object . * the nature of mmb , mmc and mmd is less clear . due to the low -to- and -to- ratios , mmd could be a protostellar object , consistent with clear hints of contraction motions seen in the ( 1,1 ) spectrum , while for mmb we found hints of expansion due to asymmetric emission in the two inner satellites ( see sect . [ res_linesvla ] ) . certainly , all condensations are perturbed ( mmb by the ionisation front from irs 3 , mmc by irs 1 and the outflow associated with it , mmd perhaps by a combination of both ) . only higher sensitivity and angular resolution observations will allow to better understand the nature of these cores . the most striking result of this work is the clear confirmation of a cavity opened by irs 3 in the molecular surrounding gas , and a tight interaction between this cavity and the surrounding dense gas . we have found several evidences of this : ( i ) the morphology of all the molecular tracers , especially in and , delineates a cavity around irs 3 and the 1.3 cm continuum map resolving the ionised gas perfectly matches the profile of the cavity ; ( ii ) the integrated intensity ( 2,2)/(1,1 ) ratios are large near irs 3 ; ( iii ) the line widths are also large near irs 3 , specially in and ; ( iv ) the position - velocity plot of shows a u - structure typical of an expanding shell ; ( v ) in the mmb core we found hints of expansion in the ( 1,1 ) spectrum due to different intensity of the two inner satellites . these evidences of such an interaction are shown for the first time in this work . if we put together all the results obtained , we speculate about a possible scenario that describes the star formation history in : irs 1 and irs 3 , both having bolometric luminosities of about 1000 , seem to come from the same natal cloud while being clearly in different evolutionary stages , which points towards different generations of ( intermediate- to high - mass ) star formation in . in this context , irs 3 could have induced the formation of irs 1 , as could be inducing star formation on the west ( in mma ) . on the other hand , in this bright - rimmed cloud the star formation probably has not been triggered by the uv radiation from the cygnus ob2 association stars , because irs 3 , the massive star that formed first , is relatively distant from the bright rim , and the dense gas where we find the bulk of the current star formation activity is around irs 3 and away from the bright rim . therefore , the star formation seems to be dominated by irs 3 , which has been caught in the act of pushing away and disrupting its natal cloud . the protocluster associated with the centre of is an excellent location where the interaction between evolved intermediate- and high - mass protostellar objects and dense ( starless ) cores can be studied . we have derived the physical and chemical properties of the dense gas by means of carma and vla observations of the millimeter and centimeter continuum , and of several molecular tracers ( , , , , ) . below , we summarise the main results . * morphologically , the dense gas is distributed in a filament oriented east - west that passes in between irs 1 and irs 3 , the two most massive and evolved objects . we resolve the dense gas into five millimeter continuum condensations . all of them show column densities consistent with potentially being the birthplace of intermediate- to high - mass objects , although the masses derived from continuum suggest that they can form intermediate - mass objects more likely . * we confirm the presence of an expanding cavity driven by irs 3 demonstrated mainly by the shape of the emission in the two pdr tracers and , as well as by hints of expanding motions from both the position - velocity diagrams and the asymmetric intensity of the two inner satellites of the ( 1,1 ) line of the millimetre core closest to irs 3 ( mmb ) . * the non - thermal line widths across the filament indicate that the gas kinematics is dominated by turbulence , similarly to other intermediate- and high - mass star - forming regions and different from low - mass dense starless cores . * the rotation and kinetic temperatures derived from ammonia are on average larger than those typically found in cores associated with low - mass star - forming regions , especially around the cavity walls . the most massive and extended millimeter core , mme , shows physical and chemical signatures of an intermediate- to high - mass pre stellar core candidate . * we have better constrained the spectral index of the radio - continuum emission associated with irs 3 , which turns out to be flat , and thus the ionised gas comes from a region photoionised by the b2 zams star at the centre of irs 3 . * the column density ratio / is of the order of 200 - 400 across the source and is higher where the dense gas is getting disrupted . * the deuterated fraction -to- is of the order of 0.1 in all cores , as large as in low - mass pre stellar cores and infrared dark clouds . we find high levels of deuteration also close to the cavity driven by irs 3 . we speculate that evaporation of and destruction caused by the uv radiation field can influence this high deuteration . these findings undoubtedly confirm a tight interaction in between the most massive and evolved objects and the dense surrounding starless cores in several respects ( kinematics , temperature , chemical gradients ) , and suggest that irs 3 could have induced the formation of irs 1 and of the other gaseous condensations accumulated on the cavity walls . however , the large - scale morphology of the molecular tracers suggests that we are likely seeing only a limited portion of a very extended gaseous filament . only a large pc - scale mosaic will allow us to fully trace the distribution of the dense gas in the region and delineate a complete view of the core population . support for carma construction was derived from the gordon and betty moore foundation , the kenneth t. and eileen l. norris foundation , the james s. mcdonnell foundation , the associates of the california institute of technology , the university of chicago , the states of california , illinois , and maryland , and the national science foundation . ongoing carma development and operations are supported by the national science foundation under a cooperative agreement , and by the carma partner universities . we acknowledge support from the owens valley radio observatory , which is supported by the national science foundation through grant ast 05 - 40399 . ap is grateful to inma seplveda for insightful discussions . ap is supported by the spanish micinn grant aya2008 - 06189-c03 ( co - funded with feder funds ) and by a jae - doc csic fellowship co - funded with the european social fund under the program ` junta para la ampliacin de estudios ' . gb is funded by an italian space agency ( asi ) fellowship under contract number i/005/07/01 . we are grateful to the anonymous referee for his / her valuable comments and suggestions . andr , ph . , belloche , a. , motte , f. , peretto , n. 2007 , a&a , 472 , 519 arce , h.g . , borkin , m.a . , goodman , a.a . , pineda , j.e . , beaumont , c.n . 2011 , apj , 742 , 105 beuther , h. , semenov , d. , henning , th . , linz , h. 2008 , apj , 675 , l33 beuther , h. , schilke , p. , menten , k.m . 2002 , apj , 566 , 945 bourke , t.l . , myers , p.c . , caselli , p. et al . 2011 , arxiv:1111.4424 briggs , d.s . 1995 , 187th american astronomical society meeting , vol . 27 , p.1444 busquet , g. , palau , aina , estalella , r. , et al . 2009 , a&a , 506 , 1183 busquet , g. , palau , a. , estalella , r. et al . 2010 , a&a , 517 , l6 campbell , m. f. , sridharan , t.k . , beuther , h. et al . 2008 , apj , 673 , 954 carral , p. , kurtz , s. , rodriguez , l.f . et al . 1999 , rmxaa , 35 , 97 caselli , p. , walmsley c.m . , zucconi , a. et al . 2002 , apj , 565 , 344 flower , d. r. , pineau des forts , g. , & walmsley , c. m. 2006 , 456 , 215 fontani , f. , caselli , p. , zhang , q. et al . 2012 , arxiv:1203.3412 fontani , f. , zhang , q. , caselli , p. , bourke , t.l . 2009 , a&a , 499 , 233 fontani , f. , caselli , p. , bourke , t.l . , cesaroni , r. , brand , j. 2008 , a&a , 477 , l45 foster , j.b . , rosolowsky , e.w . , kauffmann , j. , et al . 2009 , apj , 696 , 298 friesen , r.k . , di francesco , j. , shirley , y.l . , myers , p.c . 2009 , apj , 697 , 1457 fuente , a. , martin - pintado , j. , cernicharo , j. , bachiller , r. 1993 , a&a , 276 , 473 gerin , m. , kamierczak , m. , jastrzebska , m. et al . 2011 , a&a , 525 , a116 gerin , m. , goicoechea , j.r . , pety , j. , hily - blant , p. 2009 , a&a , 494 , 977 hennebelle , p. ; commeron , b. ; joos , m. et al . 2011 , a&a , 528 , 72 hill , t. , thompson , m.a . , burton , m.g . 2006 , mnras , 368 , 1223 ho , p.t.p . & townes , c.h . 1983 , ara&a , 21 , 239 kirk , h. , johnstone , d. , & tafalla , m. 2007 , apj , 668 , 1042 kramer , c. , alves , j. , lada , c. , et al . 1998 , a&a , 329 , l33 krumholz , m.r . , klein , r.i . , mckee , c.f . 2011 , apj , 740 , 74 krumholz , m. & mckee , c.f . 2008 , nature , 451 , 1082 kumar , m.s.n . , bachiller , r. , & davis , c. j. 2002 , apj , 576 , 313 lee , h .- g . , koo , b .- c . , park , y .- s . , ho , p. t. p. 2002 , jkas , 35 , 105 linsky , j. l. , et al . 2006 , apj , 647 , 1106 longmore , s. n. , burton , m. g. , barnes , p. j. et al . 2007 , mnras , 379 , 535 lucas , r. & liszt , h.s . 2000 , a&a , 358 , 1069 maclaren , i. , richardson , k.m . & wolfendale , a.w . 1988 , apj , 333 , 821 miralles , m.p . , rodriguez , l.f . , tapia , m. , et al . 1994 , a&a , 282 , 547 molinari , s. , brand , j. , cesaroni , r. , & palla , f. 2000 , a&a , 355 , 617 mul , p.m. & mcgowan , j.w . 1980 , apj , 237 , 749 mller , h. s. p. , thorwirth , s. , roth , d. a. , & winnewisser , g. 2001 , a&a , 370 , l49 padovani , m. , walmsley , c.m . , tafalla , m. , galli , d. , mller , h.s.p . 2009 , a&a , 505 , 1199 palau , a. , estalella , r. , girart , j.m . , et al . 2007a , a&a , 465 , 219 palau , aina , estalella , r. , ho , p.t.p . , beuther , h. , beltrn , m.t . 2007b , a&a , 474 , 911 palau , aina , snchez - monge , . , busquet , g. , et al . 2010 , a&a , 510 , a5 panagia , n. 1973 , apj , 78 , 929 park , y .- s . 2001 , a&a , 376 , 348 peeters , e. , hony , s. , van kerckhoven , c. et al . 2002 , a&a , 390 , 1089 pety , j. , teyssier , d. , foss , d. et al . 2005 , a&a , 435 , 885 pillai , t. , kauffmann , j. , wyrowski , f. et al . 2011 , a&a , 530 , 118 pillai , t. , wyrowski , f. , hatchell , gibb , j.a.g . , thompson , m.a . 2007 , a&a , 467 , 207 purcell , c. r. , minier , v. , longmore , s. n. et al . 2009 , a&a , 504 , 139 roueff , e. , lis , d.c . , van der tak , f.f.s . , gerin , m. , goldsmith , p.f . 2005 , a&a , 438 , 585 rygl , k.l.j . , brunthaler , a. , sanna , a. et al . 2011 , arxiv1111.7023 sakai , n. , saruwatari , o. , sakai , t. , takano , s. , yamamoto , s. 2010 , a&a , 512 , 31 sridharan , t. k. , beuther , h. , schilke , p. , menten , k. m. , wyrowski , f. 2002 , apj , 566 , 931 stutzki , j. & winnewisser , g. 1985 , a&a , 144 , 13 tafalla , m. , myers , p.c . , caselli , p. , & walmsley , c.m . 2004 , a&a , 416 , 191 torrelles , j.m . , verdes - montenegro , l. , ho , p.t.p . , rodriguez , l.f . , canto , j. 1993 , apj , 410 , 202 walsh , a. j. , myers , p. c. , di francesco , j. , et al . , 2007 , apj , 655 , 958 zhang , q. , hunter , t.r . , brand , j. et al . 2005 , apj , 625 , 864 zhang , q. , hunter , t.r . , sridharan , t.k . , ho , p.t.p . 2002 , apj , 566 , 982 let us assume that the linewidth of the magnetic hyperfine components of the nh@xmath17 @xmath18 inversion transition is large enough to make them unresolved , so that only the five electric hyperfine lines , i.e. one main line ( `` m '' ) , two inner satellites ( `` is '' ) , and two outer satellites ( `` os '' ) , are resolved . assuming that the excitation temperature is @xmath19 , the ratio of intensities of the main line to the inner satellite line is @xmath20 the usual assumption is that we are observing a homogeneous isothermal region , so that both excitation temperatures , @xmath21 and @xmath22 , are equal , and that @xmath23 . in this case , @xmath24 the lower limit corresponds to the optically thin case , while the upper limit is the optically thick case . in sect . 3.2.2 we have shown nh@xmath17 @xmath18 spectra for each of the 3 mm continuum clumps . for the case of mma , the intensity ratio of the inner satellites and the main line is @xmath25 , lower than the optically thin limit , @xmath26 . the assumption that the observed ratio is close to 0.28 , and that the emission in mma is optically thin , leads to inconsistent results . for an optical depth of the main line of @xmath27 , we obtain that @xmath28 k , which is much higher than the kinetic temperature estimated from the intensities ratio @xmath29 ( see text ) , @xmath30 k. this result is improbable , since we expect the excitation temperature to be , in general , lower than the kinetic temperature . the intensity of the @xmath31 is close to the kinetic temperature , indicating that the optical depth of the main line is probably @xmath32 . the optical depth of the satellite , however , can be lower , so that both lines are tracing the emission of different layers of the region observed : the main line , the outer layer facing the observer ; and the satellite line , a deeper layer of material ( see fig . [ figappendix ] ) . the easiest explanation of the anomalous ratio @xmath33 is to assume that the region is not isothermal , and that the two layers at different physical depths , have different temperatures . so , the two excitation temperatures , @xmath21 and @xmath22 , are not equal . thus , @xmath34 assuming that @xmath22 is lower than @xmath35 , @xmath36 giving that the optical depth of the main line must be @xmath37 . thus , taking into account that @xmath23 , we obtain @xmath38 resulting in @xmath39 the result is that the deeper layer traced by the satellite line is colder than the outer layer traced by the main line . if we assume that the kinetic temperature of mma , obtained from the ratio @xmath29 , is tracing the outer layer , the outer layer temperature is 22 k , while the inner layer temperature is between 5 k and 8 k. the higher temperature of the outer layer is indicative of external heating , as discussed in the text .
the intermediate- to high - mass star - forming region is an excellent laboratory to study the influence of high- and intermediate - mass young stellar objects on nearby starless dense cores , and investigate for possible implications in the clustered star formation process . we present 3 mm observations of continuum and rotational transitions of several molecular species ( , , , ) obtained with the combined array for research in millimetre - wave astronomy , as well as 1.3 cm continuum and observations carried out with the very large array , to reveal the properties of the dense gas . we confirm undoubtedly previous claims of an expanding cavity created by an ultracompact region associated with a young b2 zero - age main sequence ( zams ) star . the dense gas surrounding the cavity is distributed in a filament that seems squeezed in between the cavity and a collimated outflow associated with an intermediate - mass protostar . we have identified 5 millimeter continuum condensations in the filament . all of them show column densities consistent with potentially being the birthplace of intermediate- to high - mass objects . these cores appear different from those observed in low - mass clustered environments in sereval observational aspects ( kinematics , temperature , chemical gradients ) , indicating a strong influence of the most massive and evolved members of the protocluster . we suggest a possible scenario in which the b2 zams star driving the cavity has compressed the surrounding gas , perturbed its properties and induced the star formation in its immediate surroundings . [ firstpage ] stars : formation ism : individual objects : iras 20343 + 4129 ism : molecules
with the successful launch of the _ swift _ gamma - ray burst explorer , an era of systematic , multi - wavelength observations of grb early afterglows has been ushered in . very early optical / ir detections have been made with ground - based telescopes ( akerlof et al . 1999 ; fox et al . 2003 ; li et al . 2003 ; blake et al . 2005 ; vestrand et al . 2005 ) before and during the initial operation of _ swift_. in the x - ray band , some evidence of the early afterglows has been collected earlier ( e.g. piro et al . 1998 ; giblin et al . 1998 ; burenin et al . 1999 ; piro et al . however , it is the operation of the _ swift _ x - ray telescope ( xrt ) that offers the possibility to unveil the final gap between the prompt emission and the late afterglow stage . there has been widespread expectation that the early x - ray observations could answer a series of core questions in grb studies . what is the connection between the grb prompt emission and the afterglow ? are prompt emission and afterglow both from the external shock ( mszros & rees 1993 ; dermer & mitman 1999 ) or do they come from different locations [ i.e. prompt emission from the internal shocks ( rees & mszros 1994 ; paczynski & xu 1994 ) , while the afterglow comes from the external shock ( mszros & rees 1997a ; sari et al . does the central engine become dormant after the burst is over ? what is the immediate environment of the burst , an ism or a wind ? are there density clumps surrounding the grb progenitor ? what is the role of the reverse shock ? what is the initial lorentz factor of the fireball ? all these questions can be at least partially answered with the early x - ray afterglow data , sometimes in combination with the prompt gamma - ray data and the early optical / ir afterglow data . although early afterglow lightcurves have been extensively modeled in the optical band ( mainly driven by the observations and by the theoretical argument that the reverse shock emission component plays an important role in the optical band , e.g. mszros & rees 1997a ; sari & piran 1999 ; mszros & rees 1999 ; kobayashi 2000 ; kobayashi & zhang 2003a , b ; zhang et al . 2003 ; wei 2003 ; wu et al . 2003 ; li et al . 2003b ; fan et al . 2004a ; zhang & kobayashi 2005 ; fan et al . 2005a , b ; nakar & piran 2004 ; mcmahon et al . 2004 ) , possible early x - ray afterglow signatures have been only sparsely studied ( e.g. kumar & panaitescu 2000a ; kobayashi et al . 2005a ; fan & wei 2005 ) . in its first six months of operations , the _ swift _ x - ray telescope ( xrt ) has already accumulated a rich collection of early afterglow features in the x - ray band . the xrt is a sensitive broad - band ( 0.2 - 10 kev ) imager , which can be promptly slewed to grb targets triggered by the burst alert telescope ( bat ) within 1 - 2 minutes ( burrows et al . it is therefore an ideal instrument to study the transition between the grb prompt emission and the very early x - ray afterglow . the following features are all detected by xrt in a good sample of bursts ( see , e.g. chincarini et al . 2005 ; nousek et al . 2005 for a collection of data ) , reflecting some common underlying physics of grbs . \1 . in most cases ( e.g. grb 050126 and grb 050219a ) , a steep decay is observed up to several hundred seconds after the burst trigger , which is followed by a more conventional shallower decay ( tagliaferri et al . 2005 ; goad et al . this conclusion is drawn by choosing the grb trigger time as the zero time point ( @xmath0 ) . at later times ( e.g. @xmath1 , where @xmath2 is the duration of the grb ) , the afterglow decay slope @xmath3 is essentially independent on the adopted @xmath0 . however , at early times ( e.g. @xmath4 not much larger than @xmath2 , the measured decay slope could be very sensitive to the assumed @xmath0 . tagliaferri et al . ( 2005 ) explored the @xmath0 effect and concluded that the two distinct lightcurve segments are likely intrinsic rather than due to a poor choice of @xmath0 . furthermore , in some cases , the steep decay segment also has a different spectral index ( e.g. for grb 050319 , cusumano et al . usually it also connects to the spectral extrapolation of the bat prompt emission lightcurve smoothly ( barthelmy et al . all these facts indicate that the steep decay component is a distinct new component that may be unrelated to the conventional afterglow component . \2 . in a good fraction of grbs ( e.g. grb 050128 , campana et al . 2005 ; grb 050315 , vaughan et al . 2005 ; grb 050319 , cusumano et al . 2005 ; and grb 050401 , de pasquale et al . 2005 ) , the early x - ray afterglow is characterized by a shallow to `` normal '' transition . during the transition the spectral index is not changed . the decay slope after the break ( e.g. @xmath5 ) is consistent with the standard ism afterglow model , while the decay slope before the break is usually much shallower ( e.g. @xmath6 ) . \3 . in some cases ( e.g. grb 050315 , vaughan et al . 2005 ) , a further steepening is detected after the `` normal '' decay phase , which is consistent with a jet break . bright x - ray flares have been detected in the early x - ray lightcurves of nearly one half of the burst population ( e.g. grb 050406 , grb 050202b , burrows et al . 2005 ; romano et al . 2005 ; falcone et al . 2005 ) . in particular , the x - ray afterglow of the short - hard burst grb 050724 also shows at least three flares ( barthelmy et al . the flares typically happen hundreds of seconds after the trigger or earlier , but in some cases they occur around a day ( e.g. grb 050502b , falcone et al . 2005 ; grb 050724 , barthelmy et al . the amplitudes of the flares are usually larger than the underlying afterglow component by a factor of several ( e.g. a factor of 6 in grb 050406 , burrows et al . 2005 ; romano et al . 2005 ) , but can be much larger ( e.g. @xmath7 in the case of grb 050202b , burrows et al . 2005 ; falcone et al . a similar feature was evident for grb 011121 detected by bepposax ( piro et al . 2005 ) . in summarizing the current x - ray afterglow data , one can tentatively draw a synthetic cartoon lightcurve in the x - ray band , which consists of 5 components ( see figure [ xrtlc ] ) : i. an initial steep decay ( with a typical slope @xmath8 or steeper ) ; ii . a shallower - than - normal decay ( with a typical slope @xmath6 ) ; iii . a normal decay ( with a typical slope @xmath5 ) ; iv . a late steeper decay ( with a typical slope @xmath9 ) ; and v. one or more x - ray flares . we note that nousek et al ( 2005 ) also arrived at the similar schematic diagram that includes the segments i , ii and iii in our cartoon picture ( see their fig.3 ) . limited by the quality of the data , the current analyses indicate that the spectral indices remain unchanged in the segments ii , iii and iv , with a typical value of @xmath10 ( @xmath11 , nousek et al . 2005 ) . in some bursts , the segments i and ii have different spectral indices ( e.g. grb 050319 , cusumano et al . 2005 ) . in some cases , a time - evolution of the spectral index has been detected ( e.g. in the giant flare of grb 050502b , falcone et al . 2005 ) . in fig.1 , we have indicated the typical temporal index for each segment . throughout the paper the transition times between the adjacent segments for the four lightcurve segments are denoted as @xmath12 , @xmath13 , and @xmath14 , respectively . in this paper , we systematically study the physical processes that shape an early x - ray lightcurve , and discuss possible theoretical interpretations to the above phenomena . in [ sec : curvature ] , we discuss the grb tail emission arising from high angular latitude relative to the viewing direction , which takes longer to reach the observer due to the extra distance it travels , as a conical ( or spherical ) shell suddenly stops shining . this is the so - called curvature effect " ( fenimore et al . 1996 ; kumar & panaitescu 2000a ; dermer 2004 ; dyks et al . 2005 ) . in [ sec : fs ] , we review the main emission properties from the external forward shock region in the x - ray band , summarizing the temporal and spectral indices expected in the x - ray band for both the ism and the wind models . furthermore , we also discuss the case of a continuously refreshed shock as well as its three possible physical mechanisms . several case studies are investigated to reveal an intriguing refreshed shock phase commonly present in many bursts . in [ sec : rs ] , we briefly discuss whether and how the reverse shock emission would influence the x - ray band emission . in [ sec : flare ] , we explore various mechanisms that might give rise to the x - ray flares observed in many bursts ( e.g. grb 050406 and grb 050202b ) , and conclude that the phenomenon is best interpreted as due to a late central engine activity . our conclusions are summarized in [ sec : conclusions ] . the temporal bridge between the grb prompt emission and the afterglow emission is essential for revealing whether the prompt emission and the afterglow originate from the same component . the earliest grb relativistic shock model invoked the external shock as the site for prompt @xmath15-ray emission ( rees & mszros 1992 ; mszros & rees 1993 ) . the rapid variability observed in many grbs is in great contrast with the intuitive expectations in the external shock model , which generally predicts a smooth burst temporal profile , and it has been argued that the radiative efficiency is too low for the model so that a much larger energy budget is required ( sari & piran 1997 ) . dermer & mitman ( 1999 , 2003 ) argued that if the grb ambient medium is sufficiently clumpy , an external shock grb model could reproduce the observed variability with a high energy efficiency . within such a picture , the prompt emission and the afterglow originate from the same component ( i.e. the external shock ) , and it is expected that the two emission components are likely smoothly connected in the early phase . on the other hand , it is now commonly believed that grb prompt emission originates from some `` internal processes '' , i.e. the @xmath15-rays are emitted before the fireball is decelerated by the ambient medium . the most widely discussed model is the internal shock model ( rees & mszros 1994 ; paczynski & xu 1994 ; kobayashi et al . 1997 ; daigne & mochkovitch 1998 ; fan et al . 2004b ) . alternatively , the internal emission may be caused by dissipation of strong magnetic fields ( e.g. drenkhahn & spruit 2002 ) , or comptonization of the photospheric emission ( rees & mszros 2005 ) . within such scenarios , there exist two distinct temporal episodes dominated by the prompt emission and the afterglow , respectively , since the latter is emitted at a much larger distance from the central engine when the fireball is decelerated . generally one should expect a flux contrast between these two episodes . before the _ swift _ era , no solid observation was available to finally differentiate both scenarios , and evidence in favor of each scenario had been collected ( see e.g. zhang & mszros 2004 for a review ) . it is one of the major tasks of _ swift _ to pin down the emission site of the grb prompt emission . if the prompt emission and the afterglow arise from different emission sites , as is expected in the internal shock ( or similar ) scenario , and if the prompt emission flux level is much higher than the afterglow emission flux level , one expects to see a steeply decaying lightcurve during the transition from the prompt emission phase to the afterglow phase . such a steep decay is due to the so - called `` curvature effect '' ( e.g. kumar & panaitescu 2000a ; dermer 2004 ; dyks et al . 2005 ; panaitescu et al . 2005 ) . in principle , such an effect also applies to the tail emission of the x - ray flares ( burrows et al . hereafter we generally define such an emission component as `` grb tail emission '' . for a conical jet with an opening angle @xmath16 , emission from the same radius @xmath17 but from different viewing latitudes @xmath18 ( @xmath19 ) would reach the observer at different times . even if the emission is turned off instantaneously , due to the propagation effect the observer would receive the emitted photons at the angle @xmath18 at @xmath20 . such a tail emission thus lasts for a duration of t_tail = ( 1+z ) ( r_cr / c ) ( _ j^2/2 ) 330 s ( ) ( ) . if the line of sight is not too close to the jet edge . we consider the simplest case of a jet moving with a constant bulk lorentz factor @xmath21 ( or a constant velocity @xmath22 ) . the electrons are shock - heated up to a radius @xmath17 , beyond which no fresh shocked electrons are injected , and the already heated electrons cool rapidly . the comoving emission frequency @xmath23 is boosted to @xmath24 in the observer s frame , where @xmath25^{-1}$ ] is the doppler factor , which is @xmath26 for @xmath27 , and @xmath28 for @xmath29 . since @xmath30 , one gets @xmath31 for @xmath29 . the observed flux @xmath32 is related to the comoving surface brightness @xmath33 by f_l_ d^2 ( )^- d^2 ^- d^2 + ^- t^-2- , where @xmath34 is the observed spectral index around the observed frequency @xmath35 , and the last proportionality is valid for @xmath36 . with the standard convention @xmath37 , one has the well - known result for the curvature effect ( e.g. kumar & panaitescu 2000a ; dermer 2004 ; fan & wel 2005 ; dyks et al . 2005 ; panaitescu et al . 2005 ) = 2 + . in reality , the lorentz factor of the shell could be decreasing right before the sudden cessation of the emission . this is valid for the external shock case in the deceleration phase , or even in the internal shock case . we perform numerical calculations to investigate such an effect , normalized by a local injection rate ( moderski et al . the electrons cool down through synchrotron radiation and adiabatic expansion . all kinematic effects which affect the observed flux ( e.g. doppler boost , propagation time effects ) have been taken into account rigorously , following salmonson ( 2003 ) . the second code was developed by y. z. fan . this is an afterglow code developed from the dynamical model of huang et al . ( 2000 ) , and has been used in several previous studies ( e.g. fan et al . 2004a ; fan et al . the latest addition is to also include the kinetic evolution of the electron distribution ( moderski et al . we have used both codes in various calculations in this paper , and the results are consistent with each other . the figures presented in this paper are all generated from the second code . ] . the curvature effect for the sudden switching - off of radiation in a decelerating outflow is presented figure [ fig : cur - ism ] , in which we plot the x - ray lightcurve from an expanding jet blastwave with @xmath38ergs , @xmath39 , @xmath40 , @xmath41 , and @xmath42 . we manually turn off the radiation at radii ( @xmath43 ) of @xmath44 cm , @xmath45 cm , and @xmath46 cm , respectively , and investigate the subsequent curvature effect . the radiation from fluid elements at @xmath47 has been rigorously integrated . the result indicates that the @xmath48 conclusion is essentially unchanged . . [ fig : cur - ism ] another interesting issue is the jet structure . in principle , grb jets could be structured ( zhang & mszros 2002a ; rossi et al . 2002 ; zhang et al . 2004a ; kumar & granot 2003 ) . since the curvature effect allows one to see the high - latitude emission directly , an interesting question is whether the decay slope associated with the curvature effect depends on the unknown jet structure . we have investigated this effect with the first code , and find that for a relativistic outflow the temporal slope of the curvature effect is largely insensitive to the jet structure as long as the viewing angle is not far off the bright beam . the main reason is that the decrease of flux because of the curvature effect occurs on a much shorter timescale than that for the jet structure to take effect . for a spectral index @xmath49 ( @xmath50 ) , the flux decreases by @xmath51 orders of magnitude after a time of @xmath52 , where @xmath53 is the observer time at which the curvature effect began . for a typical @xmath54 , the flux drops by one order of magnitude after a short time @xmath55 . a drop of three orders of magnitude occurs in no more than a decade in time . on the other hand , the observer can perceive the switch - off of emissivity at an angle @xmath18 measured from the line of sight at a time @xmath56t_{cr}$ ] . one can see that the structure of the outflow must have a typical angular scale smaller than @xmath57 in order to affect the observed flux before @xmath58 . for @xmath59 , the parameters of the outflow would have to vary strongly on a scale smaller than one degree . nonetheless , the effect of the jet struture would start to play a noticeable role if the line of sight is outside the bright beam . detailed calculations are presented elsewhere ( dyks et al . 2005 ) . almost all the _ swift _ xrt early afterglow lightcurves are categorized by a steep decay component followed by a more `` normal '' decaying afterglow lightcurve ( tagliaferri et al . 2005 ; nousek et al . 2005 ) - see segment 1 in fig.[xrtlc ] . in most of these cases , the measured @xmath61 and @xmath34 values in this rapidly decaying component are close to the @xmath60 relation , but in most cases , they do not match completely . this does not invalidate the curvature effect interpretation , however , since in principle the following factors would lead to deviations from the simple @xmath60 law . \1 . the time zero point ( @xmath0 ) effect . in grb studies , the afterglow lightcurves are plotted in the log - log scale , with @xmath62 defined as the trigger time of the burst . when discussing the late afterglows , shifting @xmath0 by the order of the burst duration @xmath2 does not make much difference . when discussing the early afterglow and its connection to the prompt emission , however , the decay power law index ( @xmath63 ) is very sensitive to the @xmath0 value one chooses . correctly choosing @xmath0 is therefore essential to derive the correct temporal decay index @xmath61 ( see e.g. kobayashi et al . 2005b for a detailed discussion about the @xmath0 issue ) . in the case of the internal shock model , the case is straightforward . the observed pulses essentially track the behavior of the central engine ( kobayashi et al . each pulse marks another re - start of the central engine , so that @xmath0 should be re - defined for each pulse when the curvature effect of that pulse is considered . keeping the same @xmath0 as the beginning of the first pulse ( i.e. the grb trigger ) would inevitably lead to a false , very steep power law decay for later pulses . figure [ fig : internal ] gives an example to show this point . when a series of shells are successively ejected , each pulse will be followed by its tail emission due to the curvature effect , but most of these tails are buried under the main emission of the next pulse . the observed curvature effect is only the tail emission of the very last pulse . as a result , properly shifting @xmath0 is essential to interpret the steep decay component observed in xrt bursts . \2 . the superposition effect . the observed steep - to - shallow transition in the early phase of xrt bursts ( tagliaferri et al . 2005 ) suggests that by the end of the tail emission , the fireball is already decelerated , and the forward shock emission also contributes to the x - ray band . as a result , the observed steep decay should also include the contribution from the forward shock . assuming the later has a temporal decay index @xmath64 , the x - ray flux at the early phase should read @xmath65 where @xmath66 and @xmath67 are constants , and @xmath68 , @xmath69 are the time zero points for the steep decay component ( presumably of the internal origin ) and for the shallow decay component ( presumably of the external origin ) , respectively . in the intermediate regime between the two power - law segments , both components are important , and the observed @xmath61 should be shallower than @xmath70 during the steep decaying phase . this effect flattens the decay instead of steepening it . if with the above two adjustments , the observed @xmath61 is still steeper than @xmath70 , one can draw the conclusion that the solid angle of the emitting region is comparable to or smaller than @xmath71 . this would correspond to a patchy shell ( kumar & piran 2000a ) or a mini - jet ( yamazaki et al . 2004 ) . a caveat on such an interpretation is that the probability for the line of sight sitting right on top of such a very narrow patch / mini - jet is very small . as a result , this model can not interpret an effect that seems to be a general property of x - ray afterglows . if with the first two adjustments , the observed @xmath61 is flatter than @xmath70 but is still much steeper than that expected from a forward shock model , there could be two possibilities . one is that the emission is still from the internal dissipation of energy but the emission in the observational band does not cease abruptly . this is relevant when the observational band is below the cooling frequency . the adiabatic cooling therefore gives a decay slope of @xmath72 rather than @xmath73 ( e.g. sari & piran 1999 ; zhang et al . the second possibility is that one is looking at a structured jet ( zhang & mszros 2002a ; rossi et al . 2002 ) , with the line of sight significantly off - axis . the curvature effect in such a configuration typically gives a flatter decay slope than @xmath70 ( dyks et al . 2005 ) . this is particularly relevant for x - ray rich grbs or x - ray flashes for which a large viewing angle is usually expected ( zhang et al . 2004 ; yamazaki et al . further analyses of xrt data suggest that at least in some grbs , the decay slope is shallower than @xmath70 ( obrien et al . the above two possibilities are in particular relevant for these bursts . we suggest that most of the rapid - decay lightcurves observed by the _ swift _ xrt may be interpreted as grb ( or x - ray flare ) tail emission through the curvature effect , with the first two adjustments discussed above . in order to test this hypothesis , after the submission of this paper we have performed more detailed data analyses on a large sample of xrt bursts ( liang et al . 2005 ) . by assuming that the decay slope should be @xmath70 , we search the appropriate @xmath0 that allows such an assumption to be satisfied . it is found that @xmath0 is usually at the beginning of the last pulse ( for the steep decay following the prompt emission ) or at the beginning of the x - ray flare ( for steep decay following flares ) . this fact strongly suggests that the curvature effect is likely to be the correct interpretation , and at the same time lends strong support to the internal origin of the prompt emission and x - ray flares ( see 2.3 & 5.7 for the arguments in favor of the internal models for both components ) . another potential test of the curvature effect is to search for a correlation between the spectral peak energy ( @xmath74 ) and the flux at the peak ( @xmath75 ) ( dermer 2004 ) . this requires a well measured @xmath74 in the xrt band . in most cases , the xrt spectrum is consistent with a single power law . more detailed analyses on future bright x - ray flares are desirable to perform such a test . the current _ swift _ xrt observations of the early rapid - to - shallow decay transition of the x - ray lightcurves ( tagliaferri et al . 2005 ) , when interpreted as the curvature effect , have profound implications for the understanding of the grb phenomenon . \1 . it gives a direct observational proof that the grb prompt emission very likely comes from a different site than the afterglow emission . this suggests that the emission either comes from the internal shocks due to the collisions among many engine - ejected shells ( rees & mszros 1994 ; paczynski & xu 1994 ) , or is due to magnetic or other dissipation processes at a radius smaller than the fireball deceleration radius ( e.g. drenkhahn & spruit 2002 ; rees & mszros 2005 ) . in both scenarios , the energy dissipation region is well inside the region where the deceleration of the whole fireball occurs . an interesting fact is that in most cases , after the prompt emission , the x - ray emission level ( that spectrally extrapolated from the bat data ) drops by several orders of magnitude ( through the curvature effect , in our interpretation ) before `` landing '' on the afterglow emission level . one could roughly estimate the expected `` drop - off '' . the flux level in the xrt band during the prompt phase could be roughly estimated as @xmath76 , where @xmath77 is the isotropic energy of the emitted gamma - rays , @xmath2 is the duration of the burst , @xmath78 kev is the typical energy in the xrt band , @xmath79 kev is the typical peak energy in the grb spectrum , and @xmath80 is the low - energy spectral index for a band - spectrum ( band et al . 1993 ) . assuming that the x - ray band for the afterglow emission is above both the typical synchrotron frequency @xmath81 and the cooling frequency @xmath82 ( which is usually the case for the ism model , see eqs.[[tm]],[[tc ] ] ) , the x - ray afterglow flux level can be estimated as ( e.g. freedman & waxman 2001 ) @xmath83 , where @xmath84 is the isotropic energy of the afterglow kinetic energy , and @xmath85 is the electron equipartition parameter in the shock . the flux contrast can be estimated as ~ ( ) ( ) . for typical parameters , one has @xmath86 and @xmath87 , so that the term in the bracket @xmath88 . although @xmath89 would generally suggest that @xmath90 should be higher than @xmath91 , the large contrast between the two components observed in many bursts is usually not accounted for unless @xmath77 is ( much ) larger than @xmath84 . this refers to a very high apparent grb radiation efficiency - even higher than the one estimated using the late x - ray afterglow data ( lloyd - ronning & zhang 2004 ) as discussed in 3.2 . because of the injection , the effective @xmath84 in the early epochs is smaller than that in the later epochs . as a result , a larger @xmath92 contrast is expected . ] . the commonly - invoked internal shock model predicts a low emission efficiency ( e.g. panaitescu et al . 1999 ; kumar 1999 ) . understanding such a high apparent radiation efficiency is therefore desirable ( see e.g. beloborodov 2000 ; kobayashi & sari 2001 ) . the common steep - to - shallow transition feature indicates that the fireball has already been decelerated at the time when the grb tail emission fades . otherwise , one would see an initially rising lightcurve peaking at the fireball deceleration time . this fact alone sets a lower limit to the intial lorentz factor of the fireball , since the deceleration time @xmath93 must be earlier than the transition time @xmath12 . the numerical expression is _ 0 125 ( ) ^1/8 t_b1,2 ^ -3/8 ( ) ^3/8 , [ gam0 ] where @xmath77 is the isotropic gamma - ray energy ( which is an observable if the redshift @xmath94 is known ) , @xmath95 is a conversion factor between the isotropic afterglow energy @xmath84 and @xmath77 . throughout the paper , the convention @xmath96 is adopted in cgs units . applying the method to the bursts with measured @xmath94 ( chincarini et al . 2005 ) , we get the lower limits og @xmath97 for several bursts ( table [ tab : gamma ] ) . given the weak dependence on the unknown parameters ( i.e. @xmath98 ) , we conclude that the data suggest that grbs are highly relativistic events with typical lorentz factors higher than 100 . this is an independent method , as compared with previous ones using the high energy spectrum ( e.g. baring & harding 1997 ; lithwick & sari 2001 ) , the reverse shock data ( sari & piran 1999 ; wang et al . 2000 ; zhang et al . 2003 ) , and the superluminal expansion of the radio afterglow source image ( waxman et al . 1998 ) . . constraints on the initial lorentz factors of several grbs . [ cols="^,^,^,^,^",options="header " , ] [ tab : alpha - beta ] \3 . the energy injection is also brief , but the outflow has a significant fraction of poynting flux ( e.g. usov 1992 ; thompson 1994 ; mszros & rees 1997b ; lyutikov & blandford 2005 ) . assigning a parameter @xmath99 for the outflow , which is the ratio between the poynting flux and baryonic kinetic energy flux , zhang & kobayashi ( 2005 ) modeled the reverse shock emission from ejecta with an arbitrary @xmath99 value . they found that during the crossing of the reverse shock , the poynting energy is not transferred to the ambient medium . the poynting energy ( roughly by a factor of @xmath99 ) is expected to be transferred to the medium ( and hence , to the afterglow emission ) after the reverse shock disappears . zhang & kobayashi ( 2005 ) suggest that the transfer is delayed with respect to the traditional case of @xmath100 . the energy transfer process , however , is poorly studied so that one does not have a handy conversion relation with the @xmath101 value derived in the first scenario . in this subsection , we discuss several _ swift _ grbs with well - monitored early afterglow data detected by xrt . the notations for the break times and the temporal slopes are per those marked on fig . [ xrtlc ] . * grb 050128 * ( campana et al . 2005 ) : the lightcurve can be fitted by a broken power law with the break time at @xmath102s . the temporal decay indices before and after the break are @xmath103 and @xmath104 , respectively . the spectral indices before and after the break are essentially unchanged , i.e. @xmath105 , and @xmath106 . campana et al . ( 2005 ) discussed two interpretations . a jet model requires a very flat electron spectral index , i.e. @xmath107 , as well as a change of the spectral domain before and after the jet break . alternatively , the data may be accommodated in a wind model , but one has to assume three switches of the spectral regimes during the observational gap from 400s to about 4000s . so neither explanation is completely satisfactory . by comparing the predicted indices in table [ tab : alpha - beta ] , the observation may be well - interpreted within the ism model with an initial continuous energy injection episode . the segment after the break is consistent with a standard ism model for @xmath108 , with @xmath109 . the lightcurve before the break , on the other hand , is consistent with an injection model with @xmath110 and @xmath111 in the same spectral regime . the break time is naturally related to the cessation of the injection process , and a slight change of electron spectral index ( from 2.2 to 2.6 ) is required . from the beginning of the observation ( 100s ) to @xmath13 , the total energy is increased by a factor of @xmath112 . * grb 050315 * ( vaughan et al . 2005 ) : after a steep decay ( @xmath113 ) up to @xmath114s , the lightcurve shows a flat `` plateau '' with a temporal index of @xmath115 . it then turns to @xmath116 at @xmath117s . finally there is a third break at @xmath118s , after which the temporal decay index is @xmath119 . so this burst displays all four segments presented in fig.[xrtlc ] . the spectral indices in segments ii , iii and iv are essentially constant , i.e. @xmath120 , @xmath121 and @xmath122 , respectively . segment iii is consistent with an ism model with @xmath123 and @xmath124 , since in this model @xmath125 , @xmath126 , in perfect agreement with the data . the third temporal break @xmath14 is consistent with a jet break . according to dai & cheng ( 2001 ) , the post - break temporal index for @xmath127 is @xmath128 , which is also consistent with the observed @xmath129 . the plateau between @xmath12 and @xmath13 is then due to an energy injection in the same ism model ( @xmath123 ) , with @xmath130 and @xmath131 . the total injected energy is increased by a factor of @xmath132 . * grb 050319 * ( cusumano et al . 2005 ) : after a steep decay ( @xmath133 ) up to @xmath134s , the lightcurve shows a shallow decay with a temporal index of @xmath135 . it steepens to @xmath136 at @xmath137s . the spectral indices in segment ii and iii are @xmath138 and @xmath139 , respectively . again segment iii is well consistent with an ism model for @xmath140 with @xmath141 , which gives @xmath142 and @xmath143 , in excellent agreement with the data . interpreting the segment ii ( the shallow decay phase ) as the energy injection phase , for the same ism model ( @xmath144 ) , one gets @xmath145 and @xmath146 . the total injected energy is increased by a factor of @xmath147 . the uvot observations are also consistent with such a picture ( mason et al . 2005 ) . * grb 050401 * ( de pasquale et al . 2005 ) : the early x - ray lightcurve is consistent with a broken power law , with @xmath148 , @xmath149 , and @xmath150s . the spectral indices before and after the break are all consistent with @xmath151 . the @xmath152 relation does not fit into a simple @xmath127 jet model . on the other hand , the energy injection model gives a natural interpretation . after the break , the lightcurve is consistent with an ism model for @xmath153 with @xmath154 . before the break , it is consistent with the same model with @xmath155 . the total injected energy is increased by a factor of @xmath156 . the injection signature is also inferred in other bursts such as grb 050117 ( hill et al . 2005 ) and xrf 050416 ( sakamoto et al . 2005 ) , where similar conclusions could be drawn . the injection model is supported by an independent study of panaitescu et al . ( 2005 ) . the following conclusions could be drawn from the above case studies . a common feature of the early x - ray afterglow lightcurves is a well defined temporal steepening break . a crucial observational fact is that there is essentially no spectral variation before and after the break . this suggests that the break is of hydrodynamic origin rather than due to the crossing of some typical frequencies of the synchrotron spectrum in the band . it is worth mentioning that a lightcurve transition similar to the transition between segments ii and iii is expected in a radiative fireball ( e.g. bttcher & dermer 2000 ) , see e.g. figs . 1 & 2 of wu et al . however , that transition is due to the crossing of @xmath81 in the observational band . one therefore expects a large spectral variation before and after the break , which is inconsistent with the data . another straightforward interpretation would be a jet break , but there are three reasons against such an interpretation . first , in all the cases , @xmath127 has to be assumed . this is in stark contrast to the late jet breaks observed in the optical band , which typically have @xmath157 . furthermore , the @xmath152 relation predicted in the jet model is usually not satisfied . second , the post - break @xmath152 relation is usually satisified in a standard slow cooling ism model , with the x - ray band either below or above the cooling frequency . in such a sense , this segment is quite `` normal '' . third , in some cases ( e.g. grb 050315 ) , another steepening break is observed after this normal segment , which is consistent with the jet break interpretation . since only one break could be attributed to a jet break , the `` shallow - to - normal '' break must be due to something else . . a natural interpretation of the shallow decay phase is to attribute it to a continuous energy injection , so that the forward shock is `` refreshed '' . three possibilities exist to account for the refreshed shock effect ( [ sec : fs2 ] ) : a long - lived central engine with progressively reduced activities , an instantaneous injection with a steep power - law distribution of the shell lorentz factors , and the deceleration of an instantaneously - injected highly magnetized ( high-@xmath99 ) flow . in terms of afterglow properties , these possibilities are degenerate ( e.g. the connection between @xmath101 and @xmath158 ) and can not be differentiated . in principle , the first scenario may give rise to additional observational signatures ( e.g. rees & mszros 2000 ; gao & wei 2004 , 2005 ) , which may be used to differentiate the model from the others . two interesting characteristics during the injection phase are that the injection process is rather smooth , and that the effective @xmath101 value is around 0.5 . this gives interesting constraints on the possible physical mechanisms . ( 1 ) for the scenario of a continuously injecting central engine ( zhang & mszros 2001 ) , the central engine luminosity must vary with time smoothly . this is in contrast to the conventional grb central engine which injects energy erratically to allow the observed rapid variability in the lightcurves . this usually requires two different energy components , i.e. one `` hot '' fireball component that leads to the prompt emission and a `` cold '' poynting flux component that gives to the smooth injection . a natural poynting flux component is due to the spin - down of a new - born millisecond pulsar ( dai & lu 1998a ; zhang & mszros 2001 ) . however , a straightforward prediction from such a model is @xmath159 , not consistent with @xmath111 inferred from the observations . modifications to the simplest model are needed . alternatively , the system may be a long - lived black hole torus system with a reducing accretion rate . however , at later times the long - term central engine power corresponds to @xmath160 ( macfadyen et al . 2001 ; janiuk et al . 2004 ) , too steep to give an interesting injection signature . it is worth mentioning that in the collapsar simulations ( macfadyen et al . 2001 ) , an extended flat injection episode sometimes lasts for @xmath161s , which could potentially interpret the short injection phase of some bursts , but is difficult to account for some other bursts whose injection phase is much longer . ( 2 ) for the scenario of a power - law distribution of lorentz factors ( rees & mszros 1998 ) , one should require that a smooth distribution of lorentz factors is produced after the internal shock phase . in the internal shock model , slow shells are indeed expected to follow the fast shells , but they tend to be discrete and give rise to bumpy lightcurves ( e.g. kumar & piran 2000a ) especially when the contribution from the reverse shock is taken into account ( zhang & mszros 2002b ) . it is also unclear how an effective @xmath111 is expected . ( 3 ) deceleration of a promptly - ejected poynting - flux dominated flow ( e.g. zhang & kobayashi 2005 ) naturally gives a smooth injection signature observed . above case studies indicate that the injected energy is by a factor of several to 10 . within such a picture , the unknown @xmath99 value is about several to 10 . however , it is unclear how long the delay would be and whether one can account for the shallow decay with @xmath111 extending for @xmath162 seconds . further more detailed theoretical modeling is needed to test this hypothesis . any model needs to interpret the sudden cessation of the injection at @xmath163 . this time has different meanings within the three scenarios discussed above . ( 1 ) within the long - lived central engine model , this is simply the epoch when the injection process ceases . in the pulsar - injection model , there is a well - defined time for injection to become insignificant ( dai & lu 1998a ; zhang & mszros 2001 ) , but within a black - hole - torus injection model , such a time is not straightforwardly defined . ( 2 ) in the varying lorentz factor scenario , this time corresponds to a cut - off of the lorentz factor distribution at the low end below which the distribution index @xmath158 is flatter than 1 so that they are energetically unimportant . this lowest lorentz factor is defined by _ m = 23 ( ) ^1/8 t_b2,4 ^ -3/8 ( ) ^3/8 , a successful model must be able to address a well - defined @xmath164 in this model . ( 3 ) within the poynting flux injection model , a well - defined time cut - off is expected , which corresponds to the epoch when all the poynting energy is transferred to the blastwave . if the shallow decay is indeed due to poynting energy transfer , the cut - off time ( @xmath13 ) could be roughly defined by the @xmath99 parameter through @xmath165 , where @xmath93 is the conventional deceleration time defined by @xmath166 , when only a fraction of @xmath167 energy is transfered to the ism ( zhang & kobayashi 2005 ) . \5 . although we have not tried hard to rule out a wind - model interpretation , the case studies discussed above suggest that the early afterglow data are consistent with an ism model for essentially all the bursts . this conclusion also applies to other well - studied _ swift _ bursts ( e.g. grb 050525a , blustin et al . this result is intriguing given that long grbs are associated with the death of massive stars , from which a strong wind is expected . previous analyses using late time afterglow data ( e.g. panaitescu & kumar 2002 ; yost et al . 2003 ) have also suggested that most afterglow data are consistent with an ism model rather than a wind model . in order to accommodate the data , it has been suggested that the wind parameter may be small so that at a late enough time the blastwave is already propagating in an ism ( e.g. chevalier et al . the _ swift _ results push the ism model to even earlier epochs ( essentially right after the deceleration ) , and indicate the need for a re - investigation of the problem . the epoch shortly before the deaths of massive stars is not well studied ( woosley et al . one possibility is that the stellar wind ceases some time before the star collapses . careful analyses of early afterglows of a large sample of long grbs may shed light on the final stage of massive star evolution . it is generally believed that a short - lived reverse shock exists during the intial deceleration of the fireball and gives interesting emission signatures in the early afterglow phase . given the same internal energy in both the forward - shocked and the reverse - shocked regions , the typical synchrotron frequency for the reverse shock emission is typically much lower than that in the forward shock region , since the ejecta is much denser than the medium . while the early forward shock synchrotron emission peaks in x - rays at early times , the reverse shock synchrotron emission usually peaks in the optical / ir band or even lower ( e.g. mszros & rees 1997a ; sari & piran 1999 ; kobayashi 2000 ; zhang et al . 2003 ; zhang & kobayashi 2005 ) . this model has been succussful in interpreting the early optical emission from grb 990123 ( akerlof et al . 1999 ; sari & piran 1999 ; mszros & rees 1999 ) , grb 021211 ( fox et al . 2003 ; li et al . 2003 ; wei 2003 ) , and grb 041219a ( blake et al . 2005 ; vestrand et al . 2005 ; fan et al . 2005b ) . as a result , it is expected that the reverse shock component has a negligible contribution in the x - ray band . in the above argument , it has been assumed that the shock parameters ( @xmath85 , @xmath168 and @xmath169 ) are the same in both shocks . in reality this might not be the case . in particular , the grb outflow may itself carry a dynamically important magnetic field component ( or poynting flux ) . this magnetic field would be shock - compressed and amplified , giving a larger effective @xmath168 ( fan et al . 2004a , b ; zhang & kobayashi 2005 ) . since the medium is generally not magnetized , it is natural to expect different @xmath168 values in both regions , and a parameter @xmath170 have been used in the reverse shock analysis . it has been found that @xmath171 is indeed larger than unity for grb 990123 and grb 021211 ( zhang et al . 2003 ; fan et al . 2002 ; kumar & panaitescu 2003 ; panaitescu & kumar 2004 ; mamahon et al . hereafter the subscript / superscript `` f '' and `` r '' represent the forward shock and the reverse shock , respectively . according to zhang & kobayashi ( 2005 ) , the case of grb 990123 corresponds to the most optimized case with @xmath172 , so that @xmath171 is the largest . more generally , @xmath85 and @xmath169 may also vary in both shocks . fan et al . ( 2002 ) performed a detailed fit to the grb 990123 data and obtained @xmath173 and @xmath174 . a general treatment therefore requires that we introduce one more parameter , i.e. @xmath175/[(p^{\rm f}-2)/(p^{\rm f}-1 ) ] ( \epsilon_{\rm e}^{\rm r}/\epsilon_{\rm e}^{\rm f})$ ] . following the treatment of zhang et al . ( 2003 ) , we have the following relations in the thin - shell regime ( see also fan & wei 2005 ) , the treatment becomes more complicated . the treatment presented here is generally valid for @xmath176 . for @xmath177 , the reverse shock emission starts to be suppressed ( zhang & kobayashi 2005 ) . since we are investigating the most optimistic condition for the reverse shock contribution , in this paper we adopt the standard hydrodynamic treatment which is valid for @xmath178 . ] & ~&r_b r_e^2 ( ) ^2,[rnum ] + & ~ & r_b^-3 ( ) ^2 , [ rnuc ] + & ~ & r_b ~r_b _ 0 [ rfnum ] , where @xmath179 and @xmath180 are the compton parameters for the forward and the reverse shock emission components , respectively ; @xmath181 is the reverse shock crossing time , which is essentially @xmath93 ( eq.[[tdec ] ] ) for the thin shell case ; @xmath182 is the bulk lorentz factor of the outflow at @xmath181 ; @xmath183 is the lorentz factor of the shocked ejecta relative to the unshocked one . the synchrotron photons in the reverse shock region will be scattered by the same electrons that produce these photons . the characteristic energy of this component is typically in the @xmath15-ray range . however , under some conditions , this synchrotron self - compton ( ssc ) component would also stand out in the x - ray band , giving rise to an x - ray bump in the lightcurve . a detailed discussion has been presented in kobayashi et al . ( 2005 ) , which we do not repeat here . the general conclusion is that the ssc component could account for an early x - ray flare bump by a factor of several under certain optimized conditions . an advantage of this model over the reverse shock synchrotron model is that a bright uv - optical flash is avoided . however , this model can not account for a flare with a very large contrast ( e.g. by a factor of 500 , as seen in grb 050502b , burrows et al . xrt observations indicate that x - ray flares are common features in the early phase of x - ray afterglows ( the component v in figure [ xrtlc ] ) . after the report of the first two cases of flares in grb 050406 and grb 050502b ( burrows et al . 2005 ) , later observations indicate that nearly half of long grbs harbor early x - ray flares ( e.g. obrien et al . more intriguingly , the early x - ray afterglow of the latest localized short grb 050724 ( barthelmy et al . 2005b ) also revealed flares similar to those in the long grbs ( e.g. 050502b ) . the common feature of these flares is that the decay indices after the flares are typically very steep , with a @xmath205 much smaller than unity . in some cases ( e.g. grb050724 , barthelmy et al . 2005 ) , the post - flare decay slopes are as steep as @xmath206 . in this section we discuss various possible models to interpret the flares and conclude that the data require that the central engine is active and ejecting these episodic flares at later times . as discussed in [ sec : rs ] , synchrotron or ssc emission from the reverse shock region could dominate that from the forward shock emission in the x - ray band under certain conditions . because of the lack of strong uv - optical flares in the uvot observations , we tentatively rule out the reverse shock synchrotron emission model . the prediction of the lightcurve in the ssc model could potentially interpret the x - ray flare seen in grb 050406 ( burrows et al . 2005 ) , but the predicted amplitude is too low to interpret the case of grb 050502b ( burrows et al . 2005 ) . in some bursts ( e.g. grb 050607 ) , more than one flare are seen in a burst . although one of these flares may be still interpreted as the reverse shock ssc emission , an elegant model should interpret these flares by a model with the same underlying physics . we tentatively conclude that the reverse shock model can not account for most of the x - ray flares detected by the _ swift _ xrt in a unified manner . long - duration grbs are believed to be associated with the deaths of massive stars , such as wolf - rayet stars . according to woosley et al . ( 2003 ) , there are no observations to constrain the mass loss rate of a wolf - rayet star during the post - helium phase ( 100 - 1000 years before the explosion ) , no stability analyses have been carried out to assess whether such stars are stable . at the high end of a reasonable range of the mass loss rate , dense clouds surrounding a grb progenitor may exist . for a wind velocity @xmath207 , the density clumps could occur at a radius @xmath208 cm . these density clumps have been invoked by dermer & mitman ( 1999 , 2003 ) to interpret the grb prompt emission variabilities . one immediate question is whether these density clumps , if they exist , could give rise to the x - ray flares detected by xrt . in order to check this possibility , we investigate the following toy model . for simplicity , we assume a dense clump extends from @xmath209 to @xmath210 with a number density @xmath211 . the background ism density is taken as @xmath212 . other parameters in the calculation include : @xmath213ergs , @xmath42 , @xmath39 , @xmath214 , @xmath215 , @xmath198 and @xmath216 . no sideways expansion of the jet is included , which is consistent with previous simulations ( kumar & granot 2003 ; cannizzo et al . the x - ray lightcurve is shown in fig.[fig : cloud ] , the general feature of which is reproduced in both codes used in this paper . although the rising phase could be very sharp , the decaying slope is rather flat . this is because of two effects . first , after entering the dense cloud , the blastwave lorentz factor falls off rapidly . the observed time @xmath217 is hence significantly stretched . the dashed line in fig.[fig : cloud ] represents the observed emission before the fireball exits the cloud . second , after the fireball exits the cloud ( dotted line in [ fig : cloud ] ) , the fireball does not decelerate immediately since the lorentz factor is already too low to be further decelerated in a medium with a low density . it is decelerated again when enough medium is swept up at a larger radius . the interaction of a fireball with density bumps has been studied previously by many authors ( e.g. lazzati et al . 2002 ; dai & wu 2003 ) . our detailed calculations suggest that the variation caused by density inhomogeneities is generally not very significant ( see also ramirez - ruiz et al . 2005 ) . the lightcurves for a blastwave surfing on a density wave should be generally quite smooth ( cf . lazzati et al . 2002 ) . the one - dimensional model presented here effectively calculates the feature when the fireball encounters a high density shell . a more realistic model for density clumps should include the size of the clump ( @xmath218 ) . the lightcurve decay slope after the peak therefore depends on the comparison between @xmath219 and the angle the clump extends from the central engine , @xmath220 . if @xmath221 , the above calculation is still valid since the observer would not notice the edge of the clump . the resulting decay slope is rather shallow after the peak , which is in distinct contrast to the x - ray flare data . this is the case for the parameters we use in fig . [ fig : cloud ] . for smaller clumps , the lightcurve will steepen as @xmath219 becomes comparable to @xmath222 . our numerical calculations indeed show such a feature . however , the flux does not fall rapidly right after the peak . the peak time is typically shortly after the blastwave enters the density clump when the lorentz factor is still very high . the epoch when @xmath223 is satisfied happens much later . the lightcurve therefore still show a shallow decay segment before the steep decay component . this is in contrast with the x - ray flare data that show rapid decays right after the flare peak ( e.g. burrows et al . 2005 ; falcone et al . 2005 ; romano et al . 2005 ; obrien et al . 2005 ) . one could make the clouds small enough , so that @xmath224 is satisfied from the very beginning . however , in such a case it is very hard to achieve a large contrast between the flares and the underlying afterglow level ( ioka et al . 2005 ) . the 500 time contrast observed in grb 050502b is in any case very difficult to achieve within all the density clump models we have explored . we therefore tentatively conclude that the x - ray flares commonly detected in the early x - ray afterglow lightcurves are likely not caused by the putative density clouds surrounding the grb progenitors . the lack of these well - modeled features in the data also suggest that the lumpiness of circumburst medium is , if any , rather mild . we note , however , dermer ( 2005 , in preparation ) suggests that more detailed 3-d modeling of the density clump problem could potentially reproduce the observed x - ray flares . in the collapsar progenitor models , it is expected that the ejecta generally has two components , one ultra - relativistic component powering the grb , and another moderately relativistic cocoon component ( e.g. zhang . et al . 2004b ; mszros & rees 2001 ; ramirez - ruiz et al . this gives a physical motivation to the phenomenological two - component jet model ( e.g. lipunov et al . 2001 ; berger et al . 2003 ; huang et al . an interesting possibility would be whether the x - ray flare following the prompt gamma - ray emission ( which arises from the central relativistic component ) is caused by the deceleration of the wider mildly - relativistic cocoon component as it interacts with the ambient medium . a straightforward conclusion is that the wide , off - beam component must contain more energy than the narrow component in order to give noticeable features in the lightcurve . also the decay after the lightcurve peak of the second component should follow the standard afterglow model , and the variability time scale satisfies @xmath225 . the optical lightcurves of the two component jet model have been calculated by huang et al . ( 2004 ) and granot ( 2005 ) . fig.[fig : twocomp ] shows a sample x - ray lightcurve in the two - component jet model . it is obvious that this model can not interpret the rapid fall - off observed in the xrt x - ray flares , and is therefore ruled out . a related model considers a jet with large energy fluctuations in the angular direction , so that its energy distribution is patchy ( kumar & piran 2000b ) . this is a variation of the two - component jet , and could be approximated as a multi - component jet . when the @xmath71 cone of different patches enter the field of view , the observed lightcurve may present interesting signatures . however , the general feature of the two - component jet still applies : only when a patch has a substantial energy compared with the on - beam jet would it give a bump feature on the lightcurve . after each bump , the afterglow level is boosted and would not resume the previous level . the variability time scale is also typically @xmath225 . figure [ fig : patchy ] gives an example . apparently this model can not account for the observed x - ray flares . in the internal shock model , it is expected that after the collisions are finished , the shells are distributed such that a shell with a higher lorentz factor always leads a shell with a lower lorentz factor . as the fast moving shell ( blastwave ) is decelerated , the trailing slow shell will catch up with it and inject energy into the blastwave ( kumar & piran 2000a ) . such an injection also happens if the central engine further ejects high-@xmath21 shells that catch up with the decelerating blastwave ( zhang & mszros 2002b ) . such a collision would give rise to a bump signature on the lightcurve . a detailed treatment suggests that the overall lightcurve should include emission from three components : a forward shock propagating into the medium , a second forward shock propagating into the blastwave , and a reverse shock propagating into the injected shell . in the x - ray band , the contribution from the reverse shock is negligible . the lightcurve generally shows a step - like signature , due to the increase of the total energy in the blastwave ( zhang & mszros 2002b ) . after the peak , the flux level does not resume the previous level since more energy has been injected into the fireball . the decay slope after the injection peak follows the standard afterglow model , and @xmath225 is expected . these are also inconsistent with the data of x - ray flares . another interesting possibility is whether x - ray flashes are the signature of the existence of free neutrons in the fireball . derishev et al . ( 1999 ) , beloborodov ( 2003 ) and rossi et al . ( 2004 ) suggested that a baryonic fireball contains free neutrons , the decay of which would leave important imprints on the lightcurve . fan et al . ( 2005a ) modeled the process carefully and calculated the lightcurves . according to fan et al . ( 2005 ) , the neutron - feature is rather smooth in a wind model , and it is hard to detect . in the ism case , on the other hand , a bump does exist ( in all bands ) . the physical reason is that the trailing proton shell catches up with the decelerated neutron - decay products . the physical process is analogous to the post energy injection effect discussed in [ subsec : model6 ] . the amplitude of the flare is modest , at most a factor of several . since the injection model is not favored , this possibility is also disfavored . after ruling out various `` external - origin '' mechanisms , we are left only with the possibility that involves the re - activation of the central engine . in this interpretation , the x - ray flares share essentially the same origin as the prompt gamma - ray emission , i.e. they are caused by some `` internal '' energy dissipation processes which occur before the ejecta is decelerated by the ambient medium . the leading scenario is the `` late '' internal shock model , which suggests that the central engine ejects more energy in the form of an unsteady wind with varying lorentz factors at a late time . these discrete shells collide with each other and produce the observed emission . alternatively , the late injection could be mainly in the form of magnetic fields / poynting flux , and the x - ray flares are due to the intermittent dissipation of the magnetic fields , likely through magnetic reconnection events . fan , zhang & proga ( 2005c ) argued that at least for the flares following short grbs , the process that powers the flares has to be magnetic - origin . there are two advantages of the `` internal '' models over the `` external '' models . first , re - starting the central engine equivalently re - set the time zero point . in this interpretation , the observed flare component and the underlying decaying component observed at the same observer time @xmath226 originate from different physical sites at different central engine times . let us assume that the initial burst lasts @xmath2 , that the central engine re - activates after a time interval @xmath227 , and that it ejects an unsteady wind with a typical variability time scale of @xmath228 . at the observer time @xmath229 s , the underlying decaying component ( external afterglow ) happens at a distance @xmath230 ^ 2 c t \sim 5\times 10^{16}$ ] cm , where @xmath231 is the lorentz factor of the blastwave at the time @xmath226 . the flare , on the other hand , happens at a distance of @xmath232 cm , where @xmath97 is the initial lorentz factor of the late - time ejecta . according to the clocks attached to the central engine , the photons from the external afterglow component are emitted at @xmath233s , while the photons from late central engine activity are emitted at @xmath234s . because of the relativistic effect , these photons reach the observer at exactly the same time @xmath226 , and superpose onto the lightcurve detected by the observer . when plotted as a single @xmath235 lightcurve with the origin at the burst trigger , a very steep apparent decay slope can be produced for a large time shift @xmath227 ( fig.[fig : internal ] ) . this naturally overcomes the @xmath236 constraint encountered by the external models ( ioka et al . 2005 and references therein ) . second , invoking a late central engine activity greatly eases the required energy budget . in most of the external models , in order to give rise to a significant bump on the lightcurve , the total newly - added energy ( either from the radial direction - late injection case , or from the horizontal direction - patchy jets or multiple - component jets ) must be at least comparable to the energy that defines the afterglow emission level . this model therefore demands a very large energy budget . for the internal model , on the other hand , since the lightcurve is a superposition of two independent physical components , the energy budget is greatly reduced , especially if the bump happens at later times when the background afterglow level is much lower . for example , for an x - ray lightcurve with the decay index of -1 following a burst with duration 10 s , a significant flare at @xmath237s only requires a luminosity slightly larger than @xmath238 times of that of the prompt emission . this model is therefore very `` economical '' in interpreting the very late ( @xmath239s ) flares detected in some bursts ( e.g. grb 050502b , falcone et al . 2005 ; and grb 050724 , barthelmy et al . 2005b ) . can late central engine activity give rise to softer bursts ( e.g. x - ray flares as compared with the prompt gamma - ray or hard x - ray emission ) ? according to the internal shock model , the peak energy @xmath240 of the synchrotron spectrum satisfies ( zhang & mszros 2002c ) e_p b l^1/2 r^-1 l^1/2 ^-2 t^-1 , where @xmath241 is the luminosity , @xmath242 is the comoving magnetic field strength , @xmath21 is the typical lorentz factor of the wind ( for internal shock collisions , this @xmath21 is for slow shells ) , and @xmath243 is the variability time scale . we can see that a smaller @xmath241 , a higher @xmath21 and a larger @xmath243 would be favorable for a softer burst . the observed xrt x - ray flares generally have a smaller luminosity @xmath241 . the lightcurve is smoother with a larger @xmath243 ( burrows et al . 2005 ; barthelmy et al . also , at later times , the environment tends to be cleaner so that @xmath21 could be larger . one therefore naturally expects softer flares at later times . can the central engine re - start after being quiescent for some time ( e.g. @xmath244s , but sometimes as late as @xmath245s ) ? this is an interesting theoretical problem . the collapsar model predicts a central engine time scale of minutes to hours ( macfadyen et al . the star may also fragment into several pieces during the collapse ( king et al . 2005 ) . the largest piece collapses first onto the black hole , powering a prompt gamma - ray burst . other fragments are intially ejected into elliptical orbits , but would eventually fall into the black hole after some time , powering the late x - ray flares . fragmentation could also happen within the accretion disk itself due to gravitational instabilities ( perna et al . . a magnetic - dominated accretion flow could also give rise to intermittent accretion flows due to interplay between the gravity and the magnetic barrier ( proga & begelman 2003 ; proga & zhang 2005 ) . flares could also occur in other central engine scenarios ( dai et al . the rich information collected by the _ swift _ xrt suggests that we are getting closer to unraveling of the details of the bursting mechanisms of grbs . more detailed studies are called for to unveil the mystery of these explosions . an interesting fact ( see next subsection for case studies ) is that the duration of the flares are positively correlated with the time at which the flare occurs . the later the flare , the longer the flare duration . a successful central engine model must be able to address such a peculiar behavior . perna et al . ( 2005 ) suggest that if the accretion disk is fragmented into blobs or otherwise has large amplitude density fluctuation at large radii from the central engine , the viscous disk evolution would cause more spread for blobs further out from the central engine . this gives a natural mechanism for the observed correlation . perna et al . ( 2005 ) suggest that gravitational instability in the outer part of the disk is likely the origin of the density inhomogeneity within the disk . in this subsection , we briefly discuss the x - ray flares discovered in several grbs . * grb 050406 & grb 050502b * ( burrows et al . 2005 ) : these were the first two bursts with flares detected by xrt . for the case of grb 050406 whose @xmath2 is @xmath246s , an x - ray flare starts at @xmath247s and reaches a peak at @xmath248s . the flux rebrightening at the flare peak is by a factor of 6 . the rising and the decaying indices are @xmath249 and @xmath250 , respectively . the total energy emitted during the flare is about 10% of that emitted in the prompt emission . this suggests that the central engine becomes active again after @xmath247s , but with a reduced power . for the case of grb 050502b whose @xmath2 is @xmath251s , a giant flare starts at @xmath252s and reaches a peak at @xmath253s . the flux rebrightening is by a factor of @xmath7 . the decay index after the peak is @xmath254 . the total energy emitted during the flare is comparable to that emitted in the prompt emission . this suggests that the central engine re - starts and ejects a substantial amount of energy . the duration of the x - ray flare is much longer that the duration of the prompt emission ( @xmath2 ) , so that the luminosity of the flare is much lower . in grb 050502b , there is yet another late flare - like event that peaks at @xmath255s . the post - peak decay index is @xmath8 . this is also consistent with the case of a late - time central engine activity , and the decaying slope is consistent with that expected from the curvature effect . the total energy emitted in this flare is @xmath256 of that emitted in the giant flare at @xmath253s . * grb 050724 * ( barthelmy et al . 2005b ) : this is a short , hard burst with @xmath257s whose host galaxy is an elliptical galaxy , similar to the case of the first _ swift_-localized short burst grb 050509b ( gehrels et al . 2005 ) . the xrt lightcurve reveals rich features which are quite similar to the case of grb 050502b . the xrt observation starts at @xmath258s after the trigger , and the early xrt lightcurve initially shows a steep decay with a slope @xmath9 . this component is connected to the extrapolated bat lightcurve that shows a flare - like event around @xmath259s . this extended flare - like epoch ( including the -2 decay component ) stops at @xmath260s after which the lightcurve decays even more rapidly ( with an index @xmath261 ) . a second , less energetic flare peaks at @xmath252s , which is followed by another steep ( with index @xmath261 ) decay . a third , significant flare starts at @xmath262s , and the decay index after the peak is @xmath263 . this quite similar to those the late flare seen in grb 050502b . the energy emitted during the third flare is comparable to that of the second one , and both are several times smaller than the energy of the first flare . all the rapid - decay components following the flares could be potentially interpreted as the high - latitude emission , given a proper shift of the zero time point . since the duration of the third flare is very long , the @xmath0 effect does not affect the decay index too much . this is why the decay index @xmath264 is quite `` normal '' , i.e. consistent with the @xmath265 prediction . the earlier steeper decays ( e.g. @xmath261 ) are all preceded by flares with a sharp increasing phase . shifting @xmath0 in these cases would lead to significant flattening of the decaying index , which could be still consistent with the high - latitude emission . all these discussions also apply to the case of grb 050502b . * grb 011121 * ( piro et al . 2005 ) . this _ bepposax _ burst also indicated a flare - like event around @xmath266s . piro et al . ( 2005 ) interpreted the x - ray bump as the onset of the afterglow phase . in view of the fact that x - ray flares are commonly detected in _ bursts , it is natural to spectulate that the event is also an x - ray flare caused by late central engine activity . fan & wei ( 2005 ) have suggested the late central engine activity and performed a detailed case study on this event . during the past several months , the _ swift _ xrt has collected a rich sample of early x - ray afterglow data . this , for the first time , allows us to peer at the final temporal gap left by the previous observations and to explore many interesting questions of grb physics . in this paper , we have systematically investigated various possible physical processes that could give interesting contributions to the early x - ray afterglow observations . this includes the tail emission of the prompt gamma - ray emission , both the forward shock and the reverse shock emission components , refreshed shocks , post energy injection , medium density clumps near the burst , angular inhomogeneities of the fireball , emission component due to the presence of free neutrons , as well as emission from late central engine activity . we discuss how the above processes might leave interesting signatures on the early x - ray afterglow lightcurves . based on the xrt data collected so far , we summarize the salient features and suggest a tentative synthetic lightcurve for the x - ray afterglow lightcurves . as shown in fig . [ xrtlc ] , the synthetic lightcurve includes five components : i. an intial steep decay component ; ii . a shallow - than - normal decay component ; iii . a `` normal '' decay component ; iv . a post - jet break component ; and v. x - ray flares . the components i and iii appear in almost all the bursts . other three components also commonly appear in some bursts . flares have been detected in nearly half of the xrt early lightcurves , and the shallow decay segment has also been discovered in a good fraction of _ swift _ grbs . we therefore believe that they represent some common underlying physics for grbs . after comparing data with various physical models , we tentatively draw the following conclusions . \1 . the rapid decay component ( tagliaferri et al . 2005 ) commonly observed in the very early afterglow phase ( which usually has a different spectral slope than the late shallow decay components ) is very likely the tail emission of the prompt gamma - ray bursts or of the early x - ray flares . allowing proper shifting of the time zero point and considering the contribution of the underlying forward shock emission , we speculate that essentially all the steep decay cases could be understood in terms of the `` curvature effect '' of the high - latitude emission as the emission ceases abruptly . more detailed data analyses ( liang et al . 2005 ) support such a speculation . the transition between the prompt emission and the afterglow emission appears to be universally represented by a rapid decay followed by a shallower decay , indicating that the grb emission site is very likely different from the afterglow site , and that the apparent gamma - ray efficiency is very high . \3 . in a good fraction of grbs ( e.g. grbs 050128 , 050315 , 050319 , 050401 , campana et al . 2005 ; vaughan et al . 2005 ; cusumano et al . 2005 ; de pasquale et al . 2005 ) , a clear temporal break exists in the early x - ray lightcurves . there is no obvious spectral index change across the break . the temporal decay index before the break is very flat , while that after the break is quite `` normal '' , i.e. is consistent with the standard afterglow model for a fireball with constant energy expanding into an ism . we suggest that these breaks are likely not `` jet breaks '' . rather they mark the cessation of an early continuous energy injection phase during which the external shock is refreshed . we suggest three possible physical mechanisms for the refreshed shocks , i.e. a long - lived central engine with a decaying luminosity , a power law distribution of the shell lorentz factors before deceleration begins , and the deceleration of a high-@xmath99 flow . further studies are needed to better understand this phase . \4 . in the `` normal '' phase , the data for many bursts are consistent with an ism medium rather than a wind medium . this has important implications for understanding the massive star progenitors of long grbs , including their late evolution stage shortly before explosion . given that most of the shallow - to - normal transitions are due to the cessation of the refreshed shock phase , the cases with a well - identified jet break are not very common . nonetheless , jet breaks are likely identified in some bursts , e.g. grb 050315 ( vaughan et al . 2005 ) , grb 050525a ( blustin et al . 2005 ) , xrf 050416 ( sakamoto et al . 2005 ) , and some others . the x - ray flares detected in nearly half of the _ swift _ bursts are most likely due to late central engine activity , which results in internal shocks ( or similar energy dissipation events ) at later times . it seems to us that there is no evidence for the existence of density clumps in the grb neighborhood , and there is no support for strong angular inhomogeneities ( e.g. two - component jet , patchy jets ) for the grb fireball . however , their existence is not ruled out . the similar lightcurves for some long grbs ( e.g. grb050502b , burrows et al . 2005 ) and some short grbs ( e.g. grb050724 , barthelmy et al . 2005 ) indicate that different progenitor systems may share some similarities in the central engine . this might be caused by the same underlying physics that controls the hyper - accreting accretion disk - a common agent in charge both types of systems ( perna et al . 2005 ) . the _ swift _ xrt is still rapidly accumulating data on the early x - ray afterglows . more careful statistical analyses of these x - ray data , as well as more detailed case studies , also including low frequency data collected by uvot and other ground - based telescopes , would greatly imporve our knowledge about grb prompt emission and afterglows , advancing the quest for the final answers to the core questions in the study of grbs . we thank the referee for helpful comments . b.z . acknowledges useful discussions with t. abel , p. armitage , m. begelman , z. g. dai , c. d. dermer , b. dingus , c. fryer , l. j. gou , j. granot , a. heger , d. lazzati , a. panaitescu , r. perna , d. proga , s. woosley , x. y. wang , x. f. wu , and w. zhang on various topics covered in this paper . this work is supported by nasa nng05gb67 g , nng05gh91 g ( for bz ) , nng05gh92 g ( for bz , sk and pm ) , eberly research funds of penn state and by the center for gravitational wave physics under grants phy-01 - 14375 ( for sk ) , nsf ast 0307376 and nasa nag5 - 13286 ( for pm ) , and nasa nas5 - 00136 ( db & jn ) . 99 akerlof , c. et al . 1999 , nature , 398 , 400 band , d. et al . 1993 , apj , 413 , 281 baring , m. g. & harding , a. k. 1997 , apj , 491 , 663 barthelmy , s. et al . 2005a , apj , 635 , l133 . 2005b , nature , 438 , 994 beloborodov , a. m. 2000 , apj , 539 , l25 . 2003 , apj , 585 , l19 berger , e. et al . 2003 , nature , 426 , 154 berger , e. , cenko , s. b. & kulkarni , s. 2005 , gcn circ . # 3088 blake , c. h. et al . 2005 , nature , 435 , 181 blustin , a. j. et al . 2005 , apj , in press ( astro - ph/0507515 ) bttcher , m. & dermer , c. d. 2000 , apj , 532 , 281 burenin , r. a. et al . 1999 , apj , 344 , l53 burrows , d. n. et al . 2005a , astro - ph/0508071 burrows , d. n. et al . 2005b , science , 309 , 1833 campana , s. et al . 2005 , apj , 625 , l23 chevalier , r. a. , & li , z .- y . 2000 , apj , 536 , 195 chevalier , r. a. , li , z .- y . & fransson , c. 2004 , apj , 606 , 369 chincarini , g. et al . 2005 , ( astro - ph/0506453 ) cannizzo , j. k. , gehrels , n. & e. t. vishniac , 2004 , apj , 601 , 380 cusumano , g. , et al . 2005 , apj , in press ( astro - ph/0509689 ) dai , z. g. 2004 , apj , 606 , 1000 dai , z. g. & cheng , k. s. 2001 , apj , 558 , l109 dai , z. g. & lu , t. 1998a , a&a , 333 , l87 . 1998b , mnras , 298 , 87 dai , z. g. , wang , x. y. , wu , x. f. & zhang , b. 2005 , submitted dai , z. g. & wu , x. f. 2003 , apj , 591 , l21 daigne , f. & mochkovitch , r. 1998 , mnras , 296 , 275 de pasquale , m. et al . 2005 , mnras , in press ( astro - ph/0510566 ) derishev , e. v. , kocharovsky , v. v. & kocharovsky , vl . v. 1999 , apj , 521 , 640 dermer , c. 2004 , apj , 614 , 284 dermer , c. , & mitman , k. e. 1999 , apj , 513 , l5 . 2003 , astro - ph/0301340 drenkhahn , g. & spruit , h. c. 2002 , a&a , 391 , 1141 dyks , j. , zhang , b. & fan , y. z. 2005 , apj , submitted falcone , a. et al . 2005 , apj , submitted fan , y. z. , dai , z. g. , huang , y. f. , & lu , t. 2002 , chinese j. astron . astrophys . 2 , 449 fan , y. z. , & wei , d. m. 2005 , mnras , 364 , l42 fan , y. z. , wei , d. m. , & wang , c. f. 2004a , a&a , 424 , 477 fan , y. z. , wei , d. m. , & zhang , b. 2004b , mnras , 354 , 1031 fan , y. z. , zhang , b. & proga , d. 2005c , apj , 635 , l129 fan , y. z. , zhang , b. , & wei , d. m. 2005a , apj , 628 , 298 . 2005b , apj , 628 , l25 fenimore , e. e. , madras , c. d. & nayakshin , s. 1996 , apj , 473 , 998 fynbo , p. u. et al . 2005a , gcn circ . # 3136 . 2005b , gcn circ . # 3176 fox , d. et al . 2003 , apj , 586 , l5 freedman , d. l. & waxman , e. 2001 , apj , 547 , 922 gao , w. h. , & wei , d. m. 2004 , apj , 604 , 312 . 2005 , apj , 628 , 853 gehrels , n. et al . 2005 , nature , 437 , 851 giblin , t. w. et al . 1999 , apj , 524 , l47 goad , m. et al . 2005 , a&a , in press ( astro - ph/0511751 ) gou , l. j. , dai , z. g. , huang , y. f. & lu , t. 2001 , a&a , 368 , 464 granot , j. 2005 , apj , 631 , 1022 granot , j. & kumar , p. 2003 , apj , 591 , 1086 hill , j. et al . 2005 , apj , in press ( astro - ph/0510008 ) huang , y. f. , gou , l. j. , dai , z. g. & lu , t. 2000 , apj , 543 , 90 huang , y. f. , wu , x. f. , dai , z. g. , ma , h. t. & lu , t. 2004 , apj , 605 , 300 ioka , k. , kobayashi , s. & zhang , b. 2005 , apj , 631 , 429 janiuk , a. , perna , r. , di matteo , t. & czerny , b. 2004 , mnras , 355 , 950 kelson , d. & berger , e. 2005 , gcn circ . # 3100 king , a. , obrien , p. t. , goad , m. r. , osborne , j. , olsson , e. & page , k. 2005 , apj , 630 , l113 kobayashi , s. 2000 , apj , 545 , 807 kobayashi , s. , piran , t. , & sari , r. 1997 , apj , 490 , 92 kobayashi , s. & sari , r. 2001 , apj , 551 , 934 kobayashi , s. , & zhang , b. 2003a , apj , 582 , l75 .2003b , apj , 597 , 455 kobayashi , s. , zhang , b. , mszros , p. , & burrow , w. 2005a , apj , submitted ( astro - ph/0506157 ) kobayashi , s. et al . 2005b , in preparation kumar , p. 1999 , apj , 532 , l113 kumar , p. & granot , j. 2003 , apj , 591 , 1075 kumar , p. & panaitescu , a. 2000a , apj , 541 , l51 . 2000b , apj , 541 , l9 . 2003 , mnras , 346 , 905 kumar , p. , & piran , t. 2000a , apj , 535 , 152 . 2000b , apj , 532 , 286 lazzati , d. et al . 2002 , a&a , 396 , l5 li , w. d. , filippenko , a. v. , chornock , r. , & jha , s. 2003a , apj , 586 , l9 li , z. , dai , z. g. , lu , t. & song , l. m. 2003b , apj , 599 , 380 liang , e. w. et al . 2005 , apj , to be submitted lithwick , y. , & sari , r. 2001 , apj , 555 , 540 lipunov , v. m. , postnov , k. a. , & prokhorov , m. e. 2001 , astron . 45 , 236 lloyd - ronning , n. m. & zhang , b. 2004 , apj , 613 , 477 lyutikov , m. & blandford , r. d. 2005 , mnras , submitted macfadyen , a. i. , woosley , s. e. & heger , a. 2001 , apj , 550 , 410 mason , k. o. et al . 2005 , apj , in press ( astro - ph/0511132 ) mcmahon , e. , kumar , p. , & panaitescu , a. 2004 , mnras , 354 , 915 mszros , p. , & rees , m. j. 1993 , apj , 405 , 278 . 1997a , apj , 476 , 232 . 1997b , apj , 482 , l29 . 1999 , mnras , 306 , l39 . 2001 , apj , 556 , l37 moderski , r. , sikora , m. , & bulik , t. 2000 , apj , 529 , 151 nakar , e. , & piran , t. 2004 , mnras , 353 , 647 nousek , j. et al . 2005 , apj , submitted ( astro - ph/0508332 ) obrien , p. et al . 2005 , apj , to be submitted paczynski , b. , & xu , g. 1994 , apj , 427 , 708 panaitescu , a. , & kumar , p. 2002 , apj , 571 , 779 . 2004 , mnras , 353 , 511 panaitescu , a. & mszros , p. 1999 , apj , 526 , 707 panaitescu , a. , mszros , p. , & rees , m. j. 1998 , apj , 503 , 314 panaitescu , a. , mszros , p. , gehrels , n. , burrows , d. & nousek , j. 2005 , mnras , in press ( astro - ph/0508340 ) panaitescu , a. , spada , m. & mszros , p. 1999 , apj , 522 , l105 perna , r. , armitage , p. j. & zhang , b. 2006 , apj , 636 , l29 piro , l. et al . , 1998 , a&a 331 , l41 piro , l. et al . 2005 , apj , 623 , 314 preece , r. d. , briggs , m. s. , mallozzi , r. s. , pendleton , g. n. , paciesas , w. s. , & band , d. l. 2000 , apjs , 126 , 19 proga , d. & begelman , m. 2003 , apj , 592 , 767 proga , d. & zhang , b. 2005 , apj , submitted ramirez - ruiz , e. , celotti , a. , & rees , m. j. 2002 , mnras , 337 , 1349 ramirez - ruiz , e. , garcia - segura , g. , salmonson , j. d. & perez - rendon , b. 2005 , apj , 631 , 435 rees , m. j. , & mszros , p. 1992 , mnras , 258 , 41 . 1994 , apj , 430 , l93 . 1998 , apj , 496 , l1 . 2000 , apj , 545 , l73 . 2005 , apj , 628 , 847 rhoads , j. e. 1999 , apj , 525 , 737 romano , p. et al . 2005 , a&a , submitted rossi , e. , beloborodov , a. m. & rees , m. j. 2004 , aipc , 727 , 198 rossi , e. , lazzati , d. & rees , m. j. 2002 , mnras , 332 , 945 sakamoto , t. et al . 2005 , apj , submitted salmonson , j. d. 2003 , apj , 592 , 1002 sari , r. , & mszros , p. 2000 , apj , 535 , l33 sari , r. , & piran , t. 1997 , apj , 485 , 207 . 1999 , apj , 517 , l109 sari , r. , piran , t. & halpern , j. p. 1999 , apj , 519 , l17 sari , p. , piran , t. & narayan , r. 1998 , apj , 497 , l17 tagliaferri , g. et al . 2005 , nature , 436 , 985 thompson , c. 1994 , mnras , 270,480 usov , v. v. 1992 , nature , 357 , 472 vaughan , s. et al . 2005 , apj , in press ( astro - ph/0510677 ) vestrand , w. t. et al . 2005 , nature , 435 , 178 wang , x. y. , dai , z. g. & lu , t. 2000 , mnras , 319 , 1159 waxman , e. , kulkarni , s. r. & frail , d. 1998 , apj , 497 , 288 wei , d. m. 2003 , a&a , 402 , l9 woosley , s. e. , zhang , w. , & heger , a. 2003 , aip conf . proc . 662 , 185 wu , x. f. , dai , z. g. , huang , y. f. , & lu , t. 2003 , mnras , 342 , 1131 . 2005 , apj , 619 , 968 yamazaki , r. , ioka , k. & nakamura , t. 2004 , apj , 607 , l103 yost , s. , harrison , f. a. , sari , r. , & frail , d. a. 2003 , apj , 597 , 459 zhang , b. , dai , x. , lloyd - ronning , n. m. & mszros , p. 2004a , apj , 601 , l119 zhang , b. , & kobayshi , s. 2005 , apj , 628 , 315 zhang , b. , kobayshi , s. , & mszros , p. 2003 , apj , 595 , 950 zhang , b. , & mszros , p. 2001 , apj , 552 , l35 . 2002a , apj , 571 , 876 . 2002b , apj , 566 , 712 . 2002c , apj , 581 , 1236 . 2004 , ijmpa , 19 , 2385 zhang , w. , woosley , s. e. , & heger , a. 2004b , apj , 608 , 365
with the successful launch of the _ swift _ gamma - ray burst explorer , a rich trove of early x - ray afterglow data has been collected by its on - board x - ray telescope ( xrt ) . some interesting features are emerging , including a distinct rapidly decaying component preceding the conventional afterglow component in many sources , a shallow decay component before the more `` normal '' decay component observed in a good fraction of grbs ( e.g. grb 050128 , grb 050315 , grb 050319 , and grb 050401 ) , and x - ray flares in nearly half of the afterglows ( e.g. grb 050406 , grb 050502b , grb 050607 , and grb 050724 ) . these interesting early afterglow signatures reveal valuable and unprecedented information about grbs , including the prompt emission - afterglow transition , grb emission site , central engine activity , forward - reverse shock physics , and the grb immediate environment . in this paper , we systematically analyze the possible physical processes that shape the properties of the early x - ray afterglow lightcurves , and use the data to constrain various models . we suggest that the steep decay component is consistent with the tail emission of the prompt gamma - ray bursts and/or of the x - ray flares . this provides strong evidence that the prompt emission and afterglow emission are likely two distinct components , supporting the internal origin of the grb prompt emission . the shallow decay segment observed in a group of grbs suggests that very likely the forward shock keeps being refreshed for some time . this might be caused either by a long - lived central engine , or by a wide distribution of the shell lorentz factors , or else possibly by the deceleration of a poynting flux dominated flow . x - ray flares suggest that the grb central engine is very likely still active after the prompt gamma - ray emission is over , but with a reduced activity at later times . in some cases , the central engine activity even extends to days after the burst triggers . analyses of early x - ray afterglow data reveal that grbs are indeed highly relativistic events , and that early afterglow data of many bursts , starting from the beginning of the xrt observations , are consistent with the afterglow emission from an interstellar medium ( ism ) environment .
we report the study of the vector to pseudoscalar conversion decay @xmath0 with @xmath13 . in conversion decays , @xmath14 , the radiated photon is virtual and the squared dilepton invariant mass , @xmath15 , corresponds to the photon 4-momentum transferred , @xmath16 . the probability of having a lepton pair of given invariant mass is determined by the electromagnetic dynamical structure of the transition @xmath17 . the differential decay rate , normalized to the radiative width , is @xcite : @xmath18^{3/2},\,\ \ \ \ \ \ \ \ \ \ \ \ \ \ \nonumber\end{aligned}\ ] ] where @xmath19 is the mass of the electron and @xmath20 , @xmath21 are the masses of the @xmath22 and @xmath23 mesons , respectively . @xmath24 is the transition form factor , tff , that describes the coupling of the mesons to virtual photons and provides information on its nature and underlying structure . the slope of the transition form factor , @xmath25 , is defined as : @xmath26 in the vector meson dominance model , vmd , the transition form factor is parametrized as : @xmath27 the vmd successfully describes some transitions , such as @xmath28 , while is failing for others , as in the case of @xmath29 @xcite . recently , new models have been developed to overcome such a kind of discrepancies @xcite and they should be validated with the experimental data from other channels . the only existing data on @xmath0 come from the snd @xcite and cmd-2 @xcite experiments . their measurements of the branching ratio , br@xmath30 , are ( [email protected]@xmath310.07)@xmath32 and ( [email protected]@xmath310.06)@xmath32 , respectively . the vmd expectation is br@xmath33 = 1.1@xmath32 @xcite . the snd experiment has also measured the slope of the transition form factor from the @xmath34 invariant mass distribution , on the basis of 213 events : @xmath35= ( [email protected] ) gev@xmath36 @xcite . the vmd expectation is @xmath35=1 gev@xmath36 @xcite . + due to the large data sample , we have performed three different measurements : 1 . the determination of the branching fraction of the @xmath0 decay ; 2 . the direct measurement of the transition form factor slope b@xmath37 with a fit to the dilepton invariant mass spectrum ; 3 . the extraction of the @xmath38 as a function of the dilepton invariant mass . da@xmath2ne , the frascati @xmath22-factory , is an @xmath4 collider running at center of mass energy of @xmath39 mev . positron and electron beams collide at an angle of @xmath40 - 25 mrad , producing @xmath22 mesons nearly at rest . the kloe experiment operated at this collider from 2000 to 2006 , collecting 2.5 fb@xmath3 . the kloe apparatus consists of a large cylindrical drift chamber surrounded by a lead - scintillating fiber electromagnetic calorimeter both inserted inside a superconducting coil , providing a 0.52 t axial field . the beam pipe at the interaction region is a sphere with 10 cm radius , made of a 0.5 mm thick beryllium - aluminum alloy . the drift chamber @xcite , 4 m in diameter and 3.3 m long , has 12,582 all - stereo tungsten sense wires and 37,746 aluminum field wires , with a shell made of carbon fiber - epoxy composite with an internal wall of @xmath41 mm thickness . the gas used is a 90% helium , 10% isobutane mixture . the momentum resolution is @xmath42 . vertices are reconstructed with a spatial resolution of @xmath43 3 mm . the calorimeter @xcite , with a readout granularity of @xmath43(4.4 @xmath44 4.4 ) @xmath45 , for a total of 2440 cells arranged in five layers , covers 98% of the solid angle . each cell is read out at both ends by photomultipliers , both in amplitude and time . the energy deposits are obtained from the signal amplitude while the arrival times and the particles positions are obtained from the time differences . cells close in time and space are grouped into energy clusters . energy and time resolutions are @xmath46 and @xmath47 , respectively . the trigger @xcite uses both calorimeter and chamber information . in this analysis the events are selected by the calorimeter trigger , requiring two energy deposits with @xmath48 mev for the barrel and @xmath49 mev for the endcaps . + machine parameters are measured online by means of large angle bhabha scattering events . the average value of the center of mass energy is evaluated with a precision of about 30 kev each 200 nb@xmath3 of integrated luminosity . collected data are processed by an event classification algorithm @xcite , which streams various categories of events in different output files . the analysis of the decay chain @xmath0 , @xmath50 , has been performed on a data sample of about 1.7 fb@xmath3 . the monte carlo ( mc ) simulation for the signal has been produced with @xmath51 according to vmd model . the signal production corresponds to an integrated luminosity one hundred times larger than collected data . final state radiation has been included using photos monte carlo generator @xcite . for the background , all @xmath22 decays and the not resonant @xmath52 process have been simulated with a statistics two times larger than data . + all mc productions take into account changes in da@xmath2ne operation and background conditions on a run - by - run basis . data - mc corrections for cluster energies and tracking efficiencies are evaluated with radiative bhabha and @xmath53 samples , respectively . the main steps of the analysis are : 1 . a preselection requiring two tracks of opposite sign extrapolated to a cylinder around the interaction point and 6 prompt photon candidates ; 2 . a loose cut on the six photon invariant mass : @xmath54 mev ; 3 . a 3@xmath55 cut on the recoil mass against the @xmath4 pair , @xmath56 , shown in fig . [ fig : mmiss ] : @xmath57 mev mass ( @xmath43 547.85 mev ) . the shift is due to the treatment of the energy loss for the electrons in the tracking reconstruction , that assumes the energy loss for pions . ] ; 4 . a cut on the invariant mass and the distance between the two tracks extrapolated to the beam pipe and at the drift chamber wall surfaces , to reject photon conversion ; 5 . a cut based on the time of flight ( tof ) of the tracks to the calorimeter to reject events with charged pions in the final state . these cuts are described in details in ref . @xcite , which reports the results for a search of a light vector boson using the same data sample . the @xmath34 and @xmath58 variable is defined as the angle between the @xmath23 and the @xmath59 in the rest frame . ] distributions , after the @xmath56 cut and at the end of the analysis chain , are shown in fig . [ fig : compneubefore ] , compared to mc expectations . the residual background contamination is concentrated at high masses and is dominated by @xmath60 events with an early @xmath61 decay . cc + the analysis efficiency for signal events as a function of the @xmath4 invariant mass is shown in fig . [ fig : effi ] for 5 mev mass bins . it is about 10@xmath62 at low masses and increases to @xmath63 at 460 mev , due to the larger acceptance for higher momentum tracks . at the end of the analysis chain , 30,577 events are selected , with @xmath64 background contamination . after bin to bin background subtraction , 29,625@xmath31178 @xmath65 , @xmath50 , candidates are present in the dataset . the branching ratio has been calculated using bin - by - bin efficiency correction : @xmath66 the luminosity measurement is obtained using very large angle bhabha scattering events @xcite , giving an integrated luminosity of @xmath67 . the effective @xmath22 production cross section takes into account the center of mass energy variations ( at 1% level ) @xcite : @xmath68 . the value of the br(@xmath50)=([email protected])% is taken from @xcite . our result is : @xmath69 where the error includes the uncertainties on luminosity and @xmath22 production cross section . the systematic error has been evaluated moving by @xmath70 the analysis cuts on the recoil mass and tof , and by @xmath31 20% those related to conversion cuts ( table [ tab : syst_br ] ) . in order to evaluate the systematic due to the variation of the analysis efficiency for low @xmath34 values , the br has been measured for m@xmath71 mev , where the efficiency has a smoother behaviour . these systematics are negligible with respect to the normalization error . .systematics on the branching ratio . relative variation of each contribution with respect to the @xmath56 , tof , photon conversion , event classification cuts are reported . [ cols="^,^,^",options="header " , ] [ tab : mee_point ] the @xmath72 distribution has been fitted as a function of the invariant mass with two free parameters , one corresponding to the normalization and the other to @xmath73 , as shown in fig . [ fig : data_5mev ] , together with the predictions form the vmd and from ref . @xcite . from this fit , the value of the slope @xmath35 is : @xmath74 in agreement within the uncertainties with the value obtained from the fit to the invariant mass spectrum ( eq . [ eq : fit1 ] ) . analysing the @xmath0 decay channel , an precise measurements of both , the br(@xmath0 ) , and the transition form factor slope @xmath25 are obtained . we measured a value of @xmath7@xmath8@xmath9 and a value of the slope of @xmath10@xmath11 . + the br(@xmath65 ) is in agreement with vmd predictions @xcite and with the snd and cmd-2 results @xcite . the transition form factor slope is in agreement with vmd predictions @xcite , with a precision that is a factor of five better than previous snd measurement . + the transition form factor has been used @xcite to derive the upper limit for the production of a light dark boson u in @xmath75 decay . present measurement confirms the exclusion plot obtained by kloe in the mass range ( 5 @xmath76 m@xmath77 @xmath76 470 ) mev , where b@xmath37 = 1 gev@xmath36 was assumed @xcite . we warmly thank our former kloe colleagues for the access to the data collected during the kloe data taking campaign . we thank the da@xmath2ne team for their efforts in maintaining low background running conditions and their collaboration during all data taking . we want to thank our technical staff : g.f . fortugno and f. sborzacchi for their dedication in ensuring efficient operation of the kloe computing facilities ; m. anelli for his continuous attention to the gas system and detector safety ; a. balla , m. gatta , g. corradi and g. papalino for electronics maintenance ; m. santoni , g. paoluzzi and r. rosellini for general detector support ; c. piscitelli for his help during major maintenance periods . this work was supported in part by the eu integrated infrastructure initiative hadron physics project under contract number rii3-ct- 2004 - 506078 ; by the european commission under the 7th framework programme through the ` research infrastructures ' action of the ` capacities ' programme , call : fp7-infrastructures-2008 - 1 , grant agreement no . 227431 ; by the polish national science centre through the grants no . dec-2011/03/n / st2/02641 , 2011/01/d / st2/00748 , 2011/03/n / st2/02652 , + 2013/08/m / st2/00323 , and by the foundation for polish science through the mpd programme and the project homing plus bis/2011 - 4/3 .
we have studied the vector to pseudoscalar conversion decay _ @xmath0 _ , with @xmath1 , with the kloe detector at da@xmath2ne . the data set of 1.7 fb@xmath3 of @xmath4 collisions at @xmath5 contains a clear conversion decay signal of @xmath6 events from which we measured a value of @xmath7@xmath8@xmath9 . the same sample is used to determine the transition form factor by a fit to the @xmath4 invariant mass spectrum , obtaining @xmath10@xmath11 , that improves by a factor of five the precision of the previous measurement and is in good agreement with vmd expectations . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , @xmath12 collisions , conversion decay , transition form factor 13.66.bc , 13.40.gp
the anomalous suppression of @xmath2 production in relativistic heavy - ion collisions has been considered as a signal of the existence of deconfined matter , i.e. , the quark - gluon plasma " ( qgp ) @xcite . in nucleus - nucleus collisions , the production of charmonia is affected by interactions within the incoming nuclei ( cold nuclear matter , cnm ) and the subsequently formed medium ( hot nuclear matter , hnm ) . the former include a shadowing of parton distribution functions , cronin effect and nuclear absorption . they occur prior to the formation of qgp and , at high energies , usually do not involve charmonium eigenstates , but rather pre - resonance states which therefore result in a similar modification of the eventually formed eigenstates . the hnm effects include inelastic collisions with partons , color screening and the recombination of charm and anti - charm quarks in qgp . differences between the measured nuclear modification factors ( @xmath3 ) of @xmath2 and @xmath4 should mainly be caused by the hot medium . the sequential suppression idea , based on the hierarchy of binding energies , suggests that @xmath4 should suffer much stronger suppression than @xmath2 in qgp , due to its smaller binding energy @xcite , with estimated dissociation temperatures of @xmath5@xmath6 1.5 - 2@xmath7 and @xmath8@xmath6@xmath7 @xcite . this would mean that the @xmath4 can hardly survive in qgp and produce a small @xmath9 . indeed , applying an instantaneous melting picture with the above quoted temperature , transport model calculations result in a double ratio of @[email protected] in central pb - pb(@xmath10=2.76tev ) collisions for 6.5gev / c@xmath11@xmath12@xmath1130gev / c @xcite . this underestimates the experimental data significantly @xcite . one idea to improve on this situation is to include regeneration processes for the @xmath13 , which , because of its smaller dissociation temperature , would likely be operative near the hadronization transition and in the subsequent hadronic phase @xcite . on the other hand , both qgp and the initially formed @xmath14 pairs in heavy - ion collisions have time evolution scales which can interfere and cause significant deviations from an equilibrium evolution @xcite this situation is reinforced if the binding potential is weakened due to color screening , and for high - momentum charmonia due to time dilation effects @xcite . instead of being an eigenstate , the @xmath0 dipole evolves as a superposition of eigenstates through the medium . most currently used quarkonium transport models @xcite do not account for transitions between ground and excited states . while this is probably a good approximation for the predominantly produced ground states , a dynamical feeddown " may have significant effects on the yields of excited states @xcite . the composition will not only change with time but also as a consequence of medium effects on the potential , which specifically may enhance the overlap with the larger - size excited states . since the internal motion of @xmath15 and @xmath16 inside charmonium bound states is approximately non - relativistic , we employ the time - dependent schrdinger equation to describe its wavefunction evolution . we explore the in - medium heavy - quark ( hq ) potential , @xmath17 , within two limits , the free ( @xmath18 ) and internal @xmath19 energies , for which we can employ results from lattice qcd @xcite . after separating off the angular parts , the radial time - dependent schrdinger equation for the quarkonium wavefunction can be written as @xmath20\psi(r , t)\end{aligned}\ ] ] where @xmath21 is the relative distance between @xmath15 and @xmath16 , @xmath22 is the proper time in the center of mass frame , and @xmath23 the reduced mass . to numerically solve eq . ( [ seq ] ) , we employ the crank - nicholson method @xcite , which utilizes a discretization of the time evolution as @xmath24 with @xmath25 the non - zero entries are given by @xmath26 with @xmath27 and @xmath28 . from the @xmath0 dipole wavefunction , @xmath29 , one can obtain the coefficient , @xmath30 ( @xmath31=0,1 , ... ) , of a @xmath32-wave charmonium eigenstate in the @xmath0 dipole at each time step as @xmath33 the pertinent nuclear modification factor of an eigenstate evolvig in a qgp fireball can be defined as @xmath34 where @xmath35 and @xmath36 are the initial position and momentum of the center of @xmath0 dipole , with @xmath37 , @xmath38 and @xmath39 since we neglect in / elastic interactions of the @xmath0 dipole with the medium . for the initial @xmath0 phase space distribution , @xmath40 , one usually assumes a nuclear collision profile , includes a shadowing factor , and samples it using monte - carlo techniques . however , we here constrain ourselves to evolve a @xmath0 dipole in a static medium , to focus on the color screening effects on the internal evolution of the wavepackage . the direct nuclear modification factor then simply becomes @xmath41 . dipole evolution in heavy - ion collisions . after a hard collision in pb - pb collisions , the @xmath0 dipole is produced within @xmath42 fm / c . the vacuum formation time of charmonium eigenstates , @xmath43 , is usually comparable to the qgp thermalization time @xmath44 , but can become significantly longer at high momenta ( lorentz time dilation ) and/or for color - screened potentials . , scaledwidth=49.0% ] [ fig - diagram ] finally , we need to specify our input for the initial @xmath0 dipole wavefunction . it can be a sharp gaussian function during the production time , @xmath45fm / c . however , in @xmath46 collisions , one should recover the observed fractions of charmonium states . this can be achieved by making a gaussian ansatz , @xmath47 and adjust the the width , @xmath48 accordingly . we find @xmath48=0.23fm and 0.62fm to satisfy the experimental value of @xmath49 in @xmath46 , representing 2 different values of the relative velocity between @xmath15 and @xmath16 inside the dipole . the coefficient @xmath50 can be fixed to the normalization of the radial wavefunction , @xmath51 . we neglect medium effects in the hq potential in pre - equilibrium stage prior to qgp formation , which amounts to conserving the initial fraction of eigenstates in the wavepackage . therefore , we start the @xmath0 evolution from @xmath44 where qgp reaches local equilibrium , cf . [ fig - diagram ] . we start with a stationary @xmath0 in a static qgp medium at @xmath52=1.5@xmath7 , initialized as a pure @xmath2 state , cf . fig . [ fig - jpsi-15tc ] . with the hq potential as the free energy , @xmath18 , the @xmath2 is a loosely bound dipole under these conditions . its wavefunction expands outside , decreasing the overlap between the @xmath0 and the @xmath2 eigen - function , while increasing the overlap with the @xmath4 eigen - function for times @xmath22@xmath112fm/@xmath15 . subsequently , the @xmath0 dipole continues to expand decreasing the overlap with both @xmath2 and @xmath4 components , although the latter stays larger than the former . dipole in static medium with temperature @xmath53 and the free energy as the in - medium hq potential . the initial wavefunction of the @xmath0 dipole is initialized as the @xmath2 eigenstate . solid and dashed lines are the fractions of @xmath2 and @xmath4 in the @xmath0 wavefunction , respectively . , scaledwidth=40.0% ] [ fig - jpsi-15tc ] next , we initialize the @xmath0 dipole wavefunction @xmath54 at @xmath44 with the ratio of direct @xmath2 and @xmath4 yields in @xmath46 collisions using eq . ( [ gauss ] ) with @xmath55fm , and evolve the @xmath0 dipole in the static medium with @xmath56 . for the hq potential we consider both the free energy , @xmath18 , and internal energy , @xmath19 . similar to the previous case , the @xmath4 fraction increases at first and then decreases with time , cf . upper panel of fig . [ fig - sigma023 ] . the pertinent double ratio of direct @xmath2 and @xmath4 nuclear modification factors ( without electromagnetic decay feeddowns from excited states ) , @xmath57 , markedly increases with time , see lower panel of fig . [ fig - sigma023 ] . this effect is significantly less pronounced when using the internal energy as hq potential , since the much stronger attraction tends to keep the @xmath0 wavepackage more compact thus increasing ( reducing ) the overlap with the ground ( excited ) state , see the red line ( with circles ) in the lower panel of fig . [ fig - sigma023 ] . a similar effect occurs at lower temperatures , e.g. , at @xmath52=@xmath7 . even for the free energy , a large overall reduction in the time dependence of the double ratio is found , dropping below one for times @xmath22@xmath582fm / c , cf . the line with triangles in the lower panel of fig . [ fig - sigma023 ] . and @xmath4 nuclear modification factors , @xmath3 , as a function of time in a static medium with constant temperature , @xmath52=1.5@xmath7 , and @xmath59=@xmath60 . the initial @xmath0 wavefunction at @xmath44 is taken as a gaussian function of width @xmath48=0.23fm . lower panel : double ratio , @xmath57 , of direct @xmath13 over @xmath2 corresponding to the curves in the upper panle ( squares ) , and for @xmath59=@xmath61 ( circles ) , as well as for a lower @xmath52=@xmath7 with @xmath59=@xmath60 ( triangles ) . , scaledwidth=45.0% ] [ fig - sigma023 ] we repeat the same calculations as described in the previous paragraph for larger width of the initial gaussian wavepackage , @xmath62fm ( which also recovers the empirical @xmath13-over-@xmath2 ratio in @xmath46 collisions ) , cf . fig.[fig - sigma062 ] . the enhancement in the @xmath13 @xmath3 is significantly delayed , as is the suppression of the @xmath2 due to a smaller relative velocity of @xmath15 and @xmath16 . these delays also manifest themselves in the slower growth of the @xmath57 double ratio , starting out at values below one for all 3 scenarios considered , while still reaching well above one at later times for @xmath52=1.5@xmath7 and both potential types ( less for stronger attraction ) . however , for temperatures near @xmath7 , the double ratio drops dramatically within the first 1fm/@xmath15 of the evolution and stays small thereafter . , but for a larger gaussian width of @xmath48=0.62fm for the initial @xmath0 wavepackage at @xmath44 . , scaledwidth=45.0% ] [ fig - sigma062 ] the inclusion of inelastic charmonium break - up reactions will introduce an imaginary part into our calculation which will lead to a suppression of both @xmath2 and @xmath4 , which is most likely stronger for the excited states . therefore , we expect the fractions in the lower panels of figs . [ fig - sigma023 ] and [ fig - sigma062 ] to be generally reduced . however , the transition mechanisms , @xmath63 , still exist within the @xmath0 dipole and may well play an important role in the finally observed @xmath57 , including a nontrivial momentum dependence . we have also employed this approach to bottomonia . due to the large binding energy of @xmath64 , the screening of the hq potential at the relevant distances , @xmath65 , is less pronounced causing fewer transitions to occur . however , for the excited @xmath66 and @xmath67 states , we expect transitions to be active , and thus provide an independent handle from the experimental side . this will be discussed in an upcoming manuscript . in summary , we have employed the time - dependent schrdinger equation to simulate the internal evolution of hq dipoles in a hot medium . an in - medium potential causes primordially produced correlated @xmath0 dipoles to undergo transitions between different asymptotic ( vacuum ) eigenstates over timescales which are significantly longer than the vacuum formation times of these states . in particular , for weak binding the dipole expands rather easily shifting wavefunction overlap from the ground to excited states . however , we have also found that whether these mechanisms result in an enhancement or suppression of the final double ratio , @xmath57 , of the charmonium nuclear modification factors , rather sensitively depends on the strength of the binding potential . a stronger attraction , i.e. , less screening , due to either a stronger potential type ( e.g. , @xmath61 rather than @xmath60 ) or due to lower temperatures ( e.g. , near @xmath7 ) , generally leads to smaller double ratios . related effects are expected to occur in the @xmath66-@xmath67 sector . + _ acknowledgment_. we thank p. zhuang and y. liu for valuable discussions and input . this work has been supported by the us nsf under grant phy-1614484 , the nsfc and most grant nos . 11335005 , 11575093 , 2013cb922000 , 2014cb845400 and 11547043 . t. matsui and h. satz , phys . b * 178 * , 416 ( 1986 ) h. satz , j. phys . g * 32 * , r25 ( 2006 ) b. chen , y. liu , k. zhou and p. zhuang , phys . b * 726 * , 725 ( 2013 ) v. khachatryan _ et al . _ [ cms collaboration ] , phys . * 113 * , 262301 ( 2014 ) x. du and r. rapp , nucl . a * 943 * , 147 ( 2015 ) b. z. kopeliovich _ et al . _ , d * 92 * , 034023 ( 2015 ) r. katz and p. b. gossiaux , annals phys . * 368 * , 267 ( 2016 ) l. gerland _ et al . _ , c * 69 * , 014904 ( 2004 ) x. zhao and r. rapp , proc . of 4@xmath68 winter workshop on nuclear dynamics ( wwnd 2008 ) ; arxiv:0806.1239 [ nucl - th ] . t. song , c.m . ko and s.h . lee , phys . c * 91 * , 044909 ( 2015 ) x. zhao and r. rapp , phys . b * 664 * , 253 ( 2008 ) x. l. zhu , p. f. zhuang and n. xu , phys . b * 607 * , 107 ( 2005 ) m. strickland and d. bazow , nucl . a * 879 * , 25 ( 2012 ) e. g. ferreiro , phys . b * 749 * , 98 ( 2015 ) b. chen , phys . c * 93 * , 054905 ( 2016 ) h. sorge , e. v. shuryak and i. zahed , phys . lett . * 79 * , 2775 ( 1997 ) s. digal _ et al . _ , eur . j. c * 43 * , 71 ( 2005 ) j. crank and p. nicolson , adv . * 6 * , 207 ( 1996 )
we employ a time - dependent schrdinger equation to study the evolution of a @xmath0 dipole in a quark - gluon plasma ( qgp ) . medium effects on the heavy - quark potential in the qgp are found to significantly affect the timescales of the internal evolution of the dipole . color - screening can enhance the overlap of the expanding wavepackage with excited states at high temperature , while it is reduced at lower temperatures where the dipole favors the formation of the charmonium ground state . we investigate the consequences of this mechanism on the double ratio of charmonium nuclear modification factors , @xmath1 , in heavy - ion collisions . the impact of the transition mechanisms on this ratio turns out to be rather sensitive to the attractive strength of the potential , and to its temperature dependence .
supernovae ( sn ) with a massive progenitor star leave behind a stellar remnant . a massive star with a mass ( @xmath3 ) of 10 @xmath4 leaves a neutron star while that with a m @xmath5 40 leaves a blackhole . the cygnus loop is a well evolved supernova remnant ( snr ) and has been widely observed from radio to x - ray wavelengths . miyata et al . ( 1998b ) investigated the central portion of the cygnus loop and concluded that there were si , s , and fe rich plasmas in the center of the loop . comparing their abundance ratios with the theoretical calculations of nucleosynthesis by a sn explosion , they estimated the progenitor mass of @xmath625 . furthermore , miyata et al . ( 1999 ) reported a progenitor mass of @xmath615 based on the radial distribution of heavy elements from the north - eastern region toward the central portion . in both cases , we expect that a neutron star must have been left behind the sn which produced the cygnus loop . as mentioned in miyata et al . ( 1998a ; hereafter paper i ) , there have been several observations to search for a neutron star associated with the cygnus loop . so far , no compact source has been established . the asca observatory is a powerful tool to discover embedded neutron stars because of its high detection efficiency at hard x - ray region and has discovered more than a dozen neutron stars . we have reported the discovery of a compact x - ray source , ax j2049.6 + 2939 , during the asca survey project of the cygnus loop ( paper i ) . its x - ray properties suggest that it is a neutron star or an active galactic nucleus ( agn ) . if it is a neutron star , it is definite that the cygnus loop must have originated from a massive progenitor star . in this paper , we performed optical photometric and spectroscopic observations to investigate whether or not it is an agn . we also performed the timing analysis for follow - up asca observations and rxte observations to investigate the possibility of a neutron star which might associate with the cygnus loop . because of the possibility that the x - ray source is an agn , we examined it via optical observations . we performed optical imaging observations of the field containing ax j2049.6 + 2939 in the @xmath2-band and the @xmath7-band with the wiyn 3.5 m at kpno in 1997 november . the exposure time was 600 s for each band . the pixel scale was @xmath10.@xmath82 . the data were reduced and analyzed in the usual manner with _ photometric calibration was made using landolt s standard stars ( sa 112 and sa 92 ) ( landolt , 1992 ) and a photometric error was estimated to be 0.03 mag . the magnitudes of the objects are @xmath9 mag and @xmath10 mag with an accuracy of @xmath11 mag . astrometry was made using the _ usno - a1 _ catalog and the positional error was found to be less than 1@xmath12 . the @xmath7-band image centered on the x - ray position is shown in figure [ fig : optical ] with an error circle of a radius of 10@xmath12 obtained by the rosat hri ( paper i ) . if the x - ray source is an agn , the optical magnitude estimated from the x - ray flux obtained in the first asca observation is @xmath13 mag by a consideration of the galactic hi column density ( paper i ) . thus we selected possible optical counterparts as the objects which have a @xmath2-band magnitude of @xmath14 mag located within the error circle ; we extended the possible optical magnitude range to be more secure in the identification . the number of candidates thus selected is four , of which coordinates and magnitudes are listed in table 1 . we also added one more object ( no . 5 ) as a candidate for the spectroscopic observation , because it is located very close to the error circle as shown in figure [ fig : optical ] . although we searched for an object behind the bright source in the error circle , no significant object was found . it should be noted that the brightest optical object ( @xmath15 mag ) in the error circle was identified with a g - type star based on an optical spectroscopy reported in paper i and was excluded from the candidates in this observation . optical spectroscopy of the candidates was conducted in 1998 october with the gold camera spectrograph attached to the kpno 2.1 m telescope . a grating of 158 lines mm@xmath16 and an order cut filter of gg400 were used , which gave a spectral coverage of @xmath17 with a spectral resolution of @xmath18 . one pixel covered 4.6 and 0.@xmath878 . the slit width used was 2@xmath12 and seeing during the observing run was @xmath19 . an exposure time was 20 to 30 min and typically two or three exposures were taken for each object . we used an internal quartz lamp taken just after each exposure as a flat . data were reduced and analyzed in the usual manner using _ iraf_. the obtained spectrum of no . 1 shows a tio band feature at around 6160 as well as mg b absorption line and is identified with an m - type star . the spectra of nos . 2 , 3 , 4 , and 5 show an mg b absorption and a na d absorption at @xmath20 , and these objects are identified with stars ( presumably g or k stars ) . the spectral types of identified stars are also consistent with obtained @xmath21 colors . we therefore conclude that there are no agns in the expected magnitude range within the error circle . we performed two follow - up observations with asca . ( for a detailed description of asca , see tanaka , inoue , & holt , 1994 ) . in the following asca analysis , we focused on the gis data since the gis possesses higher detection efficiency in the hard x - ray band and a high timing resolution below 10 kev ( ohashi et al . 1996 ; makishima et al . the second observation with asca was performed on 1998 may 7 . we excluded all the data taken at elevation angle below 5@xmath22 from the night earth rim and 25@xmath22 from the day earth rim , a geomagnetic cutoff rigidity lower than 6 gv , and the region of the south atlantic anomaly . we also applied a `` flare - cut '' to maximize the signal - to - noise ratio , as described in ishisaki et al . ( ishisaki y. et al , 1997 , in the asca news letter , no.5 ) . the exposure time is @xmath1 25.4 ksec after the data screening . we constructed three kinds of images for the gis data set : a total count image , a non - x - ray background image , and a cosmic x - ray background image . the non - x - ray background image was produced with the h02-sorting method in _ display45 _ ( the detailed description of this method is in ishisaki ( 1996 ) ) . the image of the cosmic x - ray background was extracted from the large sky survey ( lss ) data ( ueda et al . 1999 ) acquired during the asca pv phase . we also extracted the non - x - ray background image from the lss data to produce an image of the mean cosmic x - ray background only . in the energy band below 2 kev , the thermal emission from the cygnus loop is dominant . thus , we extracted background - subtracted gis images in the energy band of @xmath23 kev shown in figure [ fig : second ] . the x - ray flux of ax j2049.6 + 2939 is much lower than that of the first observation ( paper i ) . the count rate of ax j2049.6 + 2939 is @xmath24 c s@xmath16/gis and is a factor of @xmath1 50 lower than that of the first observation . as clearly seen in figure [ fig : second ] , a bright compact source can be found at the north - eastern direction of ax j2049.6 + 2939 . this source was detected in the first observation but not detected with the rosat hri ( see figure 1c in paper i ) . the location ( j2000 ) of this source is @xmath25 and @xmath26 with an error radius of 70@xmath12 . we tentatively call this source as ax j2050.0 + 2941 . unfortunately , this source is outside of the sis fov . we observed the region in the vicinity of ax j2049.6 + 2939 again for the third observation . it was performed in 1998 november 10 . we screened the data using the same method as for the second observation , giving a total observation time of 7.7 ksec . the image obtained with the gis is shown in figure [ fig : third ] . the gis count rate of ax j2049.6 + 2939 is @xmath27 c s@xmath16/gis and is consistent with that of the second observation . + we should note that ax j2049.6 + 2939 was detected at only a two sigma significance level in the sis data . we extracted the x - ray spectrum of ax j2049.6 + 2939 and the background spectrum from the same region in the same manner as for the first observation ( described in paper i ) . we made spectra both from the gis2 and gis3 data and added them together . the response matrix was created from a count - weighted sum of gis2 and gis3 responses . the cross - calibration uncertainties are less than 5 % as may been found at the asca gof web page . we applied several simple models to our data set and the results are described in table 3 . @xmath28 is fixed to @xmath29 which is the same value as that obtained in the first observation ( paper i ) . we should note that the fitting results were not affected when we chose the other values that we obtained in table 1 of paper i. in all models listed in table 3 , we obtained consistent results with those of the first observation within statistical uncertainties . therefore , the spectral shape seems to be unchanged whereas the flux was decreased by a factor of @xmath1 50 . we next extracted the x - ray spectrum of ax j2050.0 + 2941 for all three data sets . as shown in fig 1 ( b ) in paper i , ax j2050.0 + 2941 is severely contaminated by ax j2049.6 + 2939 . the point spread function ( psf ) of the asca xrt has a sharp core with a broad wing structure ( serlemitsos et al . although there are complex structures in the psf , the on - axis psf shows a well - defined axially symmetric shape . the 7@xmath30 off - axis psf shows a small elongation along the tangential direction but is still axially symmetric , as shown in figure 5b of serlemitsos et al . we observed ax j2049.6 + 2939 and ax j2050.0 + 2941 within 7@xmath30 from the on - axis for all three observations . therefore , we select the region which is on the opposite side of ax j2050.0 + 2941 across the location of ax j2049.6 + 2939 to accumulate the background spectrum . the count rates of ax j2050.0 + 2941 are @xmath31 c s@xmath16 /gis , @xmath32 c s@xmath16 /gis , and @xmath33 c s@xmath16 /gis for the first , second , and third observations , respectively . we applied the same models as those for ax j2049.6 + 2939 and the results are given in table 4 . in the third observation , we could not obtain meaningful results due to low x - ray intensity and relatively short exposure time . only bremsstrahlung and power - law models are acceptable for both data sets . we found no difference in their spectral shapes within the statistical uncertainties . in the case of the power - law model , the x - ray flux in the energy range of @xmath23 kev is @xmath34 and @xmath35 for the first and the second data sets , respectively . in the third observation , we calculated the x - ray flux assuming the power - law model with a photon index of @xmath36 and @xmath37 of @xmath38 to be @xmath39 . ax j2050.0 + 2941 shows the x - ray intensity variation by a factor of 4 . we obtained the einstein ipc data set in the vicinity of ax j2050.0 + 2941 from the heasarc / gsfc online service using the observation i d of 3782 . unfortunately , ax j2050.0 + 2941 is located just on the support grid of ipc and it is difficult to estimate the x - ray flux with a high degree of accuracy . the rosat hri also observed the region of ax j2050.0 + 2941 as the sequence number of 500462 . however , it was not detected above the 5 @xmath40 level . this might be consistent with the fact that the obtained @xmath28 value was high . thus , we could not investigate the long term x - ray intensity variation for ax j2050.0 + 2941 . we searched for short term intensity variations of the two compact sources in the second observation using the gis data . since we assigned the timing bit of the gis to be 10 for the asca second and third observations , the timing resolutions for high and medium bit rate are 61 and 488 @xmath41s ( ohashi et al . 1996 ) . we used the data obtained both with the high and the medium bit rate for the temporal analysis . after applying the barycentric correction on photon arrival times , we performed the fft analysis with a nyquist frequency of 1024 hz . we found no coherent pulsations for neither ax j2049.6 + 2939 nor ax j2050.0 + 2941 . the 99 % upper limits on the pulsed fraction for the sinusoidal pulse shape are 42 % and 36 % for @xmath42 kev band and 62 % and 55 % for @xmath23 kev band , respectively , in the second observation . because the fourier transform method is not sensitive enough for non - sinusoidal pulse shapes , we also tried the epoch folding search method . arrival times were folded into 10 phase bins for assumed trial periods . the epoch folding method is extremely time consuming on the computer and we restricted our search to the period range of @xmath43 . no significant pulsation was detected . upper limits of pulse amplitude ( 99 % confidence ) were obtained as 23 % ( 0.7@xmath4410 kev ) and 44 % ( 2@xmath4410 kev ) for ax j2049.6 + 2939 and 21 % ( 0.7@xmath4410 kev ) and 41 % ( 2@xmath4410 kev ) for ax j2050.0 + 2941 , respectively . to further examine the aperiodic variations , we extracted light curves of 0.7@xmath4410 kev with the bin sizes of 512 , 1024 , 2048 , and 4096 s for the two sources . no significant aperiodic variations were found at the 99 % confidence level . in these timescales , the rms noise was found to be 26 % and 20 % for ax j2049.6 + 2939 and ax j2050.0 + 2941 , respectively . these results show that there is no excess variation above the poisson error . due to poor statistics , we could not obtain a meaningful result for the third data set . ax j2049.6 + 2939 was observed with rxte on 1997 december 22 - 23 . rxte carries three scientific instruments : the proportional counter array ( pca ) , the high - energy x - ray timing experiment ( hexte ) , and the all - sky monitor ( asm ) . detailed description of rxte can be found in bradt , rothschild , and swank ( 1993 ) . here we analyze the data from the pca which consists of five collimated ( 1@xmath22 fwhm ) proportional counter units ( pcus ) that contain three multianode detector layers with a mixture of xenon and methane gas ( jahoda et al . 1996 ) . to maximize the signal - to - noise ratio between 2 and 10 kev ( @xmath45 ph channel ) , we accumulated photon events from the top xenon / methane layer of the pca . we excluded all the data taken at elevation angle below 10@xmath22 from the earth rim , the region from the south atlantic anomaly ( saa ) to within 30 min after the end of the saa passage , and the period when there was severe electron contamination ( electron0 , electron1 , electron2 @xmath46 0.1 ) . since pcu-3 was switched off for some part of our observation , we ignored pcu-3 data to maximize the observation time . this yields a total integration time of @xmath1 26 ks . we constructed light curves for the @xmath23 kev band with 64 , 128 , and 512 s binning and estimated the background for each time bin using pcabackest with a background model of pca_bkgd_faintl7_e3v19990824.mdl and pca_bkgd_faint240_e3v19990909.mdl retrieved from the gsfc ftp site . there is no variability in the x - ray flux during the observation . the rms noise is 69 , 64 , and 64 % for the 64 , 128 , and 512s binned data . these results show that there is no excess variation above the poisson error . the mean count rate is @xmath47 c s@xmath16 /4pcu . we estimated the contamination from the shell emission of the cygnus loop . we extracted the asca 2@xmath445 kev image with the same method as miyata et al . ( 2000 ) , removed the region with the radius of 4@xmath30 centered on ax j2049.6 + 2939 , and estimated the intensity of shell emission there based on the region shown in fig 1-a in paper i. we obtained the collimator response file for the rxte from the calibration database ( caldb ) supported by gsfc ( p0coll_96jun05.fits ) and convolved it with the processed asca image . thus , we estimated the rxte countrate of the shell emission of the cygnus loop to be @xmath48 c s@xmath16 / 4pcu assuming the plasma temperature ranges from [email protected] kev ( miyata 1996 ) . therefore , the contamination of thermal emission can be negligibly small . as mentioned in section [ sec : asca ] , we found a hard x - ray compact source , ax j2050.0 + 2941 , near ax j2049.6 + 2939 . ax j2050.0 + 2941 shows x - ray variability by a factor of 4 . since the angular separation is less than 7@xmath30 , both sources were simultaneously observed with rxte . assuming the x - ray flux of ax j2050.0 + 2941 to decrease linearly from the first to the second observation , the x - ray flux of ax j2050.0 + 2941 at the rxte observation phase can be estimated to be @xmath49 . we assumed the x - ray spectrum of ax j2049.6 + 2939 to be a power - law spectrum with a photon index of @xmath50 and @xmath28 to be @xmath51 ( paper i ) , resulting in the x - ray flux of ax j2049.6 + 2939 to be @xmath52 erg @xmath53 . we should note that we take into account the 20 % cross calibration uncertainties between asca and rxte . this value is an order of magnitude lower than that of the asca first observation and is similar to that of the second asca observation . we have performed a fft analysis for the frequency range of @xmath54500 hz on the barycenter corrected time series data of 2@xmath4410 kev detected in the top xenon / methane layer . no significant coherent pulsation was found . the poor signal to noise statistics due to the relatively low x - ray flux and high background rate ( @xmath55 95 % of the total counts ) did not allow us to constrain the better upper limit of the coherent pulsations than those with the first and second asca observations . we also tried the epoch folding search method for the rxte data in the same way as that for asca . although we searched for the period range @xmath56 , the @xmath57 values were found to become large for the period range @xmath58 s due to the time variable background counts . we therefore restricted the search range to @xmath59 . an upper limit of pulse amplitude was found to be @xmath6 70 % of the estimated source count rate of ax j2049.6 + 2939 . we constructed the long term light curve of ax j2049.6 + 2939 , using the einstein ipc , the rosat hri , rxte pca , and asca gis data , which span a period of 19 years . for the ipc and hri observations , we extracted the count rate of ax j2049.6 + 2939 while the background count rate was estimated from regions where the intensity of thermal emission from the cygnus loop was similar to that of ax j2049.6 + 2939 . we assumed a power - law function and took into account the error regions obtained in paper i to calculate the x - ray flux . figure [ fig : long - term ] shows the obtained long term light curve of ax j2049.6 + 2939 . we find that the x - ray flux of ax j2049.6 + 2939 generally decreases during 19 years while a `` flare event '' occurred during the first asca observation . we should note that the x - ray variability shown in figure [ fig : long - term ] contains a lot of uncertainties : insufficient ( or no ) cross calibration , different energy bands , assumed model and @xmath28 . however , we can still confirm a large x - ray intensity variation with the asca data , solely . ax j2050.0 + 2941 shows an x - ray intensity variation of a factor of 4 during the asca observations . it shows no short term variability . the x - ray spectrum of ax j2050.0 + 2941 can be fitted by a power - law function with a photon index of @xmath60 or a thermal bremsstrahlung with @xmath61 of 3 kev . as mentioned in paper i , radio emission was detected in the vicinity of ax j2050.0 + 2941 using _ skyview _ supported by heasarc / gsfc . williams et al . ( 1992 ) analyzed both radio - loud and radio - quiet quasars with ginga in the energy range of @xmath62 kev and found a correlation between the radio - loudness and the x - ray spectral index . the x - ray spectra of radio - loud and radio - quiet quasars were well represented with a power - law function with an index of @xmath63 and @xmath64 , respectively . since we have found a radio counterpart of ax j2050.0 + 2941 , it is likely to be a radio - loud quasar . if ax j2050.0 + 2941 is an agn , the reddening - corrected optical magnitude estimated from the x - ray flux is @xmath65 mag ( zamorani et al . 1981 ) . we then checked for a counterpart in optical wavelengths using the _ usno - a1 _ catalog . however , ax j2050.0 + 2941 is detected only in the asca observation and the error circle is relatively large ( 70@xmath12 ) . more than forty objects were found within the error circle . we therefore need further x - ray observations of ax j2050.0 + 2941 with chandra or xmm - newton to determine the position of ax j2050.0 + 2941 with high accuracy for its identification . we have found a long - term x - ray variability of over 19 years for ax j2049.6 + 2939 . only in the sequence of asca observations , ax j2049.6 + 2939 shows a long - term x - ray variability with a factor of @xmath1 50 but no short - term x - ray variability . the x - ray spectrum can be fitted by a power - law function with an index of @xmath66 or a thermal bremsstrahlung with @xmath61 of 4 kev for three observations . such a large x - ray variability suggests that it is not likely to be an ordinary rotation - powered pulsar but it could be an agn . we have performed optical photometric and spectroscopic observations in the vicinity of ax j2049.6 + 2939 and found no agns within its error circle . thus , we conclude that ax j2049.6 + 2939 is not likely to be an agn . we should note , however , that we performed the optical photometric observations in 1997 november where the expected x - ray flux might be lower than that of the asca first observation . we can estimate the x - ray flux of ax j2049.6 + 2939 at 1997 november to be @xmath67 erg s@xmath16 @xmath68 from figure [ fig : long - term ] assuming a smooth decrease of the x - ray flux from the asca first observation to the asca second observation . the optical magnitude expected from this value is @xmath69 mag ( zamorani et al . we have investigated five objects within the error circle whose @xmath2-band magnitude range @xmath14 mag based on wiyn observations . so , even if the x - ray flux is lower than our estimate , we can conclude that no agns are found within the error circle . we investigated the possibility of the g - type star considering our new x - ray observations . the x - ray spectrum of the g - type star is well represented by an isothermal , two - temperature , or dem ( differential emission measure ) thermal model . the @xmath61 values range from @xmath1 80 ev to @xmath12.5 kev and abundances of heavy elements are roughly solar values ( gdel et al 1997 ) . the @xmath61 value and abundances of ax j2049.6 + 2939 are a little higher and much lower , respectively , than those of g - type stars if we apply a thin thermal model ( paper i ) . for these stars , flares with much larger luminosities have been observed where @xmath61 and emission measure becomes larger than those of quiescent state . however , the timescales of their flare are roughly several tens of minutes or hours which is much shorter than ` flare event' seen in figure [ fig : long - term ] . therefore , we again conclude that ax j2049.6 + 2939 is not likely to be a g - type star . it is well known that the x - ray pulsar 1e 2259 + 586 in the supernova remnant [email protected] shows a flux variation of a factor of a few on timescales of a few years ( corbet et al . this pulsar belongs to a member of anomalous x - ray pulsars ( axps ) . members of this group have several common properties including a relatively long pulse period of @xmath70 s and a steep x - ray spectrum characterized by a power - law function with a photon index of @xmath71 . comparing these characteristics , long - term x - ray variability seems to be too large for axps , there is no coherent pulsation , and the spectrum is flatter than those of axps . thus , ax j2049.6 + 2939 is not likely to be an axp . a remarkably similar situation is found for the compact source , 1e 161348 - 5055 , in the supernova remnant rcw 103 . gotthelf , petre & vasisht ( 1999 ) reported the x - ray variability of 1e 161348 - 5055 . they found that its x - ray flux decreased by an order of magnitude over a period of 18 years . no coherent pulsation has been detected from this source ( gotthelf et al 1999 and references therein ) . these characteristics are similar to those of ax j2049.6 + 2939 whereas the x - ray spectrum of 1e 161348 - 5055 is much softer than that of ax j2049.6 + 2939 . they propose that 1e 161348 - 5055 is either a cooling neutron star ( but less plausible ) , an advection dominated accretion flow ( adaf ) around a blackhole , or an axp . if we apply the same analogy for ax j2049.6 + 2939 , this source might well be accreting through an adaf . since there are co molecular clouds at the western edge of the cygnus loop ( scoville et al . 1977 ) , such a model might be applicable to ax j2049.6 + 2939 . the x - ray spectra of ax j2049.6 + 2939 are consistent with that expected from the adaf model ( narayan , barret , & mcclintock 1997 ; manmoto , mineshige , & kusunose 1997 ) . the adaf model also predicts that the shape of the x - ray spectrum is independent of the x - ray flux . thus , there is no definite evidence for adaf but apparent x - ray features are consistent with those expected from the adaf model . we performed wide - band spectroscopic observations in the vicinity of ax j2049.6 + 2939 . based on the optical photometric and spectroscopic observations , we can rule out the possibility of an agn . ax j2049.6 + 2939 shows a large x - ray variability over 19 years , which excludes the possibility of an ordinary rotation - powered pulsar . thus , the adaf model would be a plausible one , although no definite evidence is obtained . as calculated by narayan , barret , & mcclintock ( 1997 ) and manmoto , mineshige , & kusunose ( 1997 ) , the spectrum generated by adaf shows the synchrotron peak which comprises the rayleigh - jeans slope and the optically thin synchrotron emission in the radio regime . the comptonization of the synchrotron photons leads to one or two peaks from ir to uv regimes . in the x - ray and @xmath72 ray regimes , the bremsstrahlung emission plus photons suffering multiple compton scattering and emission from saturated comptonized photons are dominant . therefore , further simultaneous observations over a wide energy band are essential to investigate its nature . we have found another hard x - ray emitting compact source , ax j2050.0 + 2941 , which is in a direction north - eastern from ax j2049.6 + 2939 . we also found radio ( 4850mhz ) emission in the vicinity of ax j2050.0 + 2941 . due to its large error circle , we found more than forty objects for the optical candidate of ax j2050.0 + 2941 . we therefore need further x - ray observations in the vicinity of ax j2050.0 + 2941 with chandra or xmm - newton in order to determine its position more accurately . it would be essential to investigate the relationship with the cygnus loop . we thank profs . s. kitamoto , f. nagase , and s. mineshige for invaluable comments and suggestions . we appreciate mr . c. baluta for careful reading of manuscript . asca_anl _ and _ display 45 _ developing team members , especially dr . part of this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center . this research is partially supported by act - jst program , japan science and technology corporation . asai k. dotani t. , mitsuda k. , hoshi r. , vaughan b. , tanaka y. , inoue h. 1996 , , 48 , 257 bradt h. v. , rothschild r. e. , swank j.h . 1993 , a&as , 97 , 355 bradt h.d.v , mcclintock j.e . 1983 , . 21 , 13 gotthelf e.v . , petre r. , vasisht g. 1999 , . 514 , l107 gdel m , guinan e.f . , skinner s.l . 1997 , , 483 , 947 ishisaki y. 1996 , ph.d . thesis of univ . of tokyo , isas rn 613 jahoda k. , swank j.h . , giles a.b . , stark m.j . , strohmayer t. , zhang w. , morgan e.h . 1996 , , 2808 , 59 landolt , a.u . 1992 , , 104 , 340 lasker b.m . , sturch c.r . , mclean b.j . , russell j.l . , jenkner h. , shara m.m . 1990 , , 99 , 2019 lochner j.c . , roussel - dupr d. 1994 , , 435 , 840 narayan r. , barret d. , mcclintock j.e . 1997 , , 482 , 448 makishima k. , tashiro m. , ebisawa k. , ezawa h. , fukazawa y. , gunji s. , hirayama m. , idesawa e. et al . 1996 , , 48 , 171 manmoto t. , mineshige s. , kusunose m. 1997 , , 489 , 791 mitsuda k. , inoue h , nakamura n. , tanaka y. 1989 , , 41 , 97 miyata e. , tsunemi h. , torii k. , hashimotodani k. , tsuru t. , koyama k. , ayani k. , ohta k. , yoshida m. 1998a , , 50 , 475 ( paper i ) miyata e. 1996 , phd thesis , osaka university ( isas rn 591 ) miyata e. , tsunemi h. , kohmura t. , suzuki s. , kumagai s. 1998b , , 50 , 257 miyata e. , tsunemi h. 1999 , , 525 , 305 miyata e. , tsunemi h. , koyama k. , ishisaki y. 2000 , adv . space . 25 , 555 ohashi t. , ebisawa k. , fukazawa y. , hiyoshi k. , horii m. , ikebe y. , ikeda h. , inoue h. et al . 1996 , , 48 , 157 scoville n.z . , 216 , 320 serlemitsos p.j . 1995 , , 47 , 105 tanaka y. , inoue h. , holt s.s . 1994 , , 46 , l37 turner t.j . , pounds k.a . 1989 , , 240 . 833 ueda y. , takahashi t. , inoue h. , tsuru t. , sakano m. , ishisaki y. , ogasaka y. , makishima k. et al . 1999 , , 518 , 656 williams o.r . et al . 1992 , , 389 , 157 zamorani g. , et al . 1981 , , 245 , 357 ccccc no.1 & 20@xmath73 49@xmath74 34.@xmath7584 & 29@xmath22 38@xmath76 53.@xmath778 & 22.0 & 19.6 + no.2 & 20@xmath73 49@xmath74 35.@xmath7501 & 29@xmath22 38@xmath76 48.@xmath777 & 21.0 & 19.2 + no.3 & 20@xmath73 49@xmath74 35.@xmath7524 & 29@xmath22 38@xmath76 41.@xmath778 & 20.9 & 19.6 + no.4 & 20@xmath73 49@xmath74 36.@xmath7502 & 29@xmath22 38@xmath76 49.@xmath775 & 20.9 & 19.7 + no.5 & 20@xmath73 49@xmath74 36.@xmath7505 & 29@xmath22 38@xmath76 43.@xmath779 & 18.9 & 17.0 + cccc blackbody & @xmath78 kev & @xmath79 erg s@xmath80 & 22.1 / 33 + bremsstrahlung & @xmath81 kev & @xmath82 @xmath83 pc & 19.6 / 33 + power law & @xmath84 = @xmath85 & @xmath86 photons kev@xmath16 @xmath87 & 20.3 / 33 + ccccccccc blackbody & @xmath88 kev & @xmath894 & @xmath90 & 38.0 / 27 & @xmath91 kev & @xmath893 & @xmath92 & 54.6 / 40 + bremsstrahlung & @xmath93 kev & @xmath94 & @xmath95 & 35.2 / 27 & @xmath96 kev & @xmath898 & @xmath97 & 47.6 / 40 + power law & @xmath84 = @xmath98 & @xmath99 & @xmath100 & 35.6 / 27 & @xmath84 = @xmath101 & @xmath8911 & @xmath102 & 47.9 / 40 +
ax j2049.6 + 2939 is a compact x - ray source in the vicinity of the southern blow - up region of the cygnus loop supernova remnant ( miyata et al . 1998a ) . this source was the brightest x - ray source inside the cygnus loop observed during the asca survey project . the x - ray spectrum was well fitted by a power - law function with a photon index of @xmath0 . short - term timing analysis was performed and no coherent pulsation was found . follow - up observations with asca have revealed a large variation in x - ray intensity by a factor of @xmath1 50 , whereas the spectral shape did not change within the statistical uncertainties . in the second asca observation , we found another x - ray source , ax j2050.0 + 2941 , at the north east of ax j2049.6 + 2939 . during the three asca observations , the x - ray intensity of ax j2050.0 + 2941 varied by a factor of @xmath14 . no coherent pulsations could be found for ax j2050.0 + 2941 . we have performed optical photometric and spectroscopic observations in the vicinity of ax j2049.6 + 2939 at the kitt peak national observatory ( kpno ) . as a result , all objects brighter than @xmath2-band magnitude of 22 in the error box can be identified with normal stars . combined with the x - ray results and the fact that there are no radio counterparts , ax j2049.6 + 2939 is not likely to be either an ordinary rotation - powered pulsar or an agn . the nature of ax j2049.6 + 2939 is still unclear and further observations over a wide energy band are strongly required . as to ax j2050.0 + 2941 , the long - term x - ray variability and the radio counterpart suggests that it is an agn .
following the first experimental demonstration of graphene@xcite , two - dimensional ( 2d ) materials have attracted increasing attention both experimentally and theoretically . especially transition metal dichalcogenides ( tmds)@xcite with chemical formula mx@xmath2 ( where m is a transition metal atom and x is a chalcogen atom ) have been a favored subject . there are also other stoichiometric forms of transition metal dichalcogenides such as titanium trisulfide ( tis@xmath3 ) that can form monolayer crystals.@xcite tmds have a special 2d layered structure . their mono- and few - layered forms offer many opportunities for fundamental and technological research@xcite because of their exceptional electronic , mechanical and optical properties@xcite . furthermore , it is well known that various kinds of tmds such as mos@xmath0 , ws@xmath0 , mose@xmath0 , wse@xmath0 , res@xmath0 , nbs@xmath2 , tis@xmath2 , and tise@xmath2 have been synthesized@xcite and studies have revealed that tmds exhibit metallic , semimetallic , semiconducting , and even superconducting behavior with different phases such as @xmath4h , @xmath4 t and their distorted forms . the presence of exotic properties in @xmath5d materials , that stemmed from increasing quantum confinement effects , has also motivated researchers to further reduce their dimension and to investigate one - dimensional ( @xmath4d ) nanoribbons . in early studies it was shown that armchair and zigzag graphene nanoribbons ( nrs ) were semiconductors with an energy gap decreasing with increasing ribbon width.@xcite in addition , zigzag graphene nanoribbons ( zgnrs ) have ferromagnetically ordered edge states and can display half - metallic behavior when an external electric field is applied.@xcite furthermore , motivated by the potential use of single layer mos@xmath2 in nanoscale optoelectronic devices , its nanoribbons have been studied intensively.@xcite armchair mos@xmath2 nrs are direct band gap semiconductors with a nonmagnetic ground state . unlike gnrs their band gaps do not vary significantly with the ribbon width.@xcite however , zigzag mos@xmath2 nrs are ferromagnetic metals regardless of their width and thickness.@xcite despite the comprehensive research on graphene and single layer tmds , studies on the electronic properties of the group ivb tmds in the t phase , namely the two - dimensional @xmath4t - mx@xmath2 structures , are sparse . nevertheless , @xmath4t - tise@xmath2@xcite is an extensively studied quasi-@xmath5d tmd , which has a charge density wave ( cdw ) state and in condensed matter physics transitions from superconductivity to charge density wave phases has been shown to be very important@xcite . however , whether @xmath4t - tise@xmath2 is a semimetal or a semiconductor is still an open question@xcite . since tite@xmath2 is a semimetal with overlapping valence and conduction bands@xcite and tis@xmath2 is a semiconductor with an indirect gap@xcite , it can be expected that the band gap of tise@xmath2 is smaller or even nonexistent . note that in the periodic table selenium is in between sulfur and tellurium , and also selenium is less electronegative than sulfur . therefore , both experimental and theoretical techniques have been used to identify the semiconducting or semimetallic nature of 1t - tise@xmath2@xcite . very recently , peng et al.@xcite grew tise@xmath2 ultrathin films on a graphitized sic(0001 ) substrate by using molecular beam epitaxy ( mbe ) . their findings offer important insights into the nature of the charge density wave in tise@xmath2 , and paved the way for potential applications based on its collective electronic states@xcite . the successful mbe growth of tise@xmath2 ultrathin films down to monolayer thickness motivated us to investigate one - dimensional tise@xmath2 nanoribbons because of its interesting electronic and physical properties that are essentially related with its low dimensionality and effects due to quantum confinement . the main goal of this study is to find the characteristics of zigzag- and armchair - edged @xmath4t - tise@xmath2 nanoribbons . the paper is organized as follows . details of the computational methodology are given in sec . the calculated structural and electronic properties of single layer 1t - tise@xmath2 are described in sec . then we analyze 1t - tise@xmath2 nanoribbons and present results from spin - unpolarized and spin - polarized calculations in detail in sec . iv . the last section , sec . v , is devoted to the conclusion . ( color online ) ( a ) atomic structure of monolayer 1t - tise@xmath2 with top and side views where the dashed yellow area denotes the unitcell of the monolayer , and ( b ) the band structure calculated with pbe and hse06 , ( c ) partial density of states as calculated with pbe . labels m@xmath4 and m@xmath5 are discussed in fig . [ zigzagband - dec].,width=321 ] the optimized structures and electronic properties of @xmath4t - tise@xmath2 nanoribbons with desired edges ( zigzag or armchair ) reported here are based on first - principle calculations within the density functional theory ( dft ) using the plane - wave projector - augmented wave ( paw ) method@xcite implemented in the vienna _ ab - initio _ simulation package ( vasp).@xcite the perdew - burke - ernzerhof ( pbe)@xcite form of the generalized gradient approximation ( gga ) were adopted to describe the electron exchange and correlation for both spin - polarized and spin - unpolarized cases . in order to correct the pbe band structure for a monolayer of tise@xmath2 , we also used the heyd - scuseria - ernzerhof 06 ( hse06 ) functional@xcite which is known to give better electronic structure description that is close to experiments and produce accurate band gaps . since it improves the accuracy of standard band gaps , we determined hse06 functional parameters as an enhanced fraction of the hartree - fock exchange @xmath6 = 0.25 and screening @xmath7 @xmath8 . the kinetic energy cutoff for the plane - wave expansion was set to @xmath9 ev where the brillouin zone ( bz ) was sampled with monkhorst pack ( mp ) by 7@xmath101@xmath101 k - point grids . for all band structure calculations , we used a 75@xmath101@xmath101 @xmath11-centered k - point mesh . to avoid the interaction between periodic images , we ensured a sufficient large supercell which is @xmath1 long perpendicular to the nanoribbon plane and with an edge - to - edge distance of at least @xmath12 . at the same time , all the atoms in the supercell were fully relaxed during the geometry optimization . the convergence threshold for energy was chosen as 10@xmath13 ev and 10@xmath14 ev / for the force . the charge distribution on the atoms were calculated by using the bader analysis.@xcite moreover , we investigated hydrogen saturated nanoribbons in order to study the edge stability . the hydrogen saturation was realized by adding one hydrogen atom to the edge of ti and se atoms for the zigzag nanoribbons , however for the armchair nanoribbons one hydrogen atom was added to the edge of se atoms and two hydrogen atoms are added to the ti atom . for the determination of the most favorable structure which means the structure after hydrogenation , the binding energies were estimated from : e@xmath15=e@xmath16[nr]+ne@xmath16[h]-e@xmath16[nr+nh ] where e@xmath16[nr ] is the total energy of the tise@xmath2 nanoribbon , e@xmath16[h ] is the energy of the free hydrogen atom , e@xmath16[nr+nh ] is the total energy of the tise@xmath2 nanoribbon saturated by hydrogen atoms , and n is the total number of saturated hydrogen atoms . before a comprehensive investigation of tise@xmath2 nanoribbons , we first present the atomic , electronic and magnetic properties of the tise@xmath2 monolayer . principally , layered structures of tmds can form several different phases , @xmath17 h and t , that result in diverse electronic properties . monolayer tise@xmath2 has a hexagonal crystal structure composed of three atom layers with a metal atom ti layer sandwiched between two chalcogen se layers . here octahedral coordination of the metal atoms results in the @xmath4 t structure as shown in fig . [ bandmonotise2](a ) . similar to graphite and graphene , in bulk tise@xmath2 the monolayers are bound together through the interlayer van der waals ( vdw ) interaction . the bond lengths are uniformly d@xmath18=@xmath19 , d@xmath20=@xmath21 , where the angle between the ti - se bonds is @xmath22=@xmath23 and the optimized lattice constant is @xmath24 from pbe calculation . the pbe electronic band dispersion , shown in fig . [ bandmonotise2](b ) , shows that single layer tise@xmath2 is a metal with a nonmagnetic ground state . in addition , the partial density of states ( pdos ) reveals that while there is negligible contribution from the se orbitals around the fermi level ( e@xmath25 ) , those bands are mainly composed of ti-@xmath26d orbitals ( d@xmath27 , d@xmath28 , d@xmath29 ) . at the same time , a bader analysis indicates that each ti atom gives 1.4 electrons to the se atoms which means that 0.7 electrons are taken by one se atom , hence this situation shows that the character of the bonding is ionic . in contrast , the band structure of 1 t tis@xmath2 is semiconducting . usually the difference in chalcogen atoms affects the structural properties , but has little influence on the electronic properties . for instance single layers of mose@xmath2 and mos@xmath2 are both direct band gap semiconductors . however , a tise@xmath2 sheet exhibits a metallic behavior with a low band crossing of the fermi level , which is different from tis@xmath2 . to further examine the electronic properties of 1t - tise@xmath2 , we also calculated the band structure with the hse06 method which is shown in fig . [ bandmonotise2](b ) . as can be seen the calculated bands below the fermi level are shifted upward while above the fermi level they are slightly shifted downward . at the same time , below the fermi level the bands are decomposed but the bands above the fermi level almost overlap with those of the pbe result . in general , relative to the experimental values , band gaps of semiconducting materials are underestimated by pbe . however , pbe+hse06 provides better aggrement with the experimental values . applying hse06 corrections to metallic systems is not very common due to its computational cost , and no expected qualitative change in the band structures . its effect is to introduce some shifts to the bands but the metallic character is preserved . for instance , single - layered vs@xmath2 and t - mos@xmath2 are still found metallic with hse06 correction.@xcite consequently , from both the pbe and hse06 methods we may conclude that tise@xmath2 is metallic . the tise@xmath2 nanoribbons ( tise@xmath2-nrs ) are obtained by cutting the @xmath5d - tise@xmath2 monolayer . according to the different directions of termination , there are two kinds of nanoribbons : zigzag ( tise@xmath2-znr ) , and armchair ( tise@xmath2-anr ) . apart from the termination , tise@xmath2-nrs are defined by their widths . the width of the zigzag nanoribbon is denoted as @xmath30 ( tise@xmath2-@xmath30znr ) and for armchair nanoribbon , the width is denoted by @xmath31 ( tise@xmath2-@xmath31anr ) . in fig . [ zigzagarm ] the lattice structure of tise@xmath2 - 8znr and tise@xmath2 - 5anr are presented . in our calculations , we consider width @xmath30 from 2 to 10 and @xmath31 from 2 to 8 . ( color online ) top view of ( a ) zigzag and ( b ) armchair tise@xmath2 nanoribbons . the unitcell is indicated by the dashed box.,width=321 ] the fully optimized nrs exhibit structural deviation at the edges . for example tise@xmath2-anrs are strongly distorted after relaxation , compared to tise@xmath2-znrs . in the triple layer networks , the edge selenium atoms shift their position from the se layers to the ti layer for both zigzag and armchair nanoribbons whereas the ti atoms at the edges shift their position from the ti layer to the se layers for only zigzag nanoribbons . at one of the edges the ti atom is closer to the lower se layer , and the ti atom at the other edge is closer to the upper se layer . as seen in fig . [ zigzagarm](b ) for armchair nanoribbons reconstruction takes place , as the ti atoms at the edges moved towards the ribbon s center and the se atoms tend to shift slightly outward . for tise@xmath2 - 8znr , shown in fig . [ zigzagarm](a ) , the ti atoms moved slightly out of the plane , leading to a change of the ti - se bond length along the ribbon - axis . nevertheless , the triple - layer networks are well kept intact for both ribbons . for instance , the average ti - se bond lengths for tise@xmath2 - 7znr are @xmath19 in the inner site , and @xmath32 at the two edges . the angle between se - ti - se bond is @xmath33 between the center and edge of the @xmath30=7 zigzag nanoribbon . for the tise@xmath2 - 8anr , coordination of atoms are different so that the ti - se bond length is different with values of @xmath34 , @xmath35 , and @xmath36 in the inner site , at the edges it decreases to @xmath37 . all of the nanoribbons display the same structural property , and the only difference is that the bond lengths between the edge ti - se atoms are longer in znrs than those in anrs . similar to the case of mos@xmath2 nanoribbons@xcite , at the edges the ti - se bond lengths decrease because of the irregular force on the edge atoms . also , a bader charge analysis tells us that charges on both ti and se atoms are equally distributed along the ribbon axis , since all of the ti atoms lose the same amount of electron charge which is taken by the se atoms . likewise in the 2d - tise@xmath2 layer , every ti atom loses 1.4 electrons to the se atoms which gain 0.7 electrons along the ribbon axis . ( color online ) energy gap of zigzag ( @xmath38 ) and armchair ( @xmath39 ) 1t - tise@xmath2 nanoribbons as function of the ribbon width . dashed curves are exponential fits.,width=321 ] during the geometry optimization , we first carried out both spin - polarized and spin - unpolarized total energy calculations in order to determine the ground state of the different tise@xmath2-@xmath30znr ( tise@xmath2-@xmath31anr ) . there is no energy difference between spin - polarized and spin - unpolarized calculations which indicates that zigzag and armchair tise@xmath2 nanoribbons have a nonmagnetic ground state . to be more confident about the magnetization of the edges , we also performed calculations for four different magnetic orderings for tise@xmath2 - 4znr and also tise@xmath2 - 5znr by taking a double unitcell , such as antiferromagnetic ( afm ) , ferromagnetic ( fm ) ( where , the atoms are located at different edges are afm coupled , and at the same edge are fm coupled ) ( see fig . [ mag](a ) ) . we take the case of a tise@xmath2 - 5znr as an example . calculations starting from the four magnetic states , namely afm - afm , afm - fm , fm - afm , and fm - fm , and results in the same total energy . the same magnetic test is also applied to armchair nanoribbons ( see fig . [ mag](b ) ) . all the test results gave the same total energy and zero net magnetic moment . as a result , tise@xmath2 armchair nanoribbons have a nonmagnetic ground state like mos@xmath2-anrs@xcite . thus , our calculation demonstrates that tise@xmath2-znrs and tise@xmath2-anrs are not magnetic and the edge states do not effect the magnetization of the structures . ( color online ) different magnetic interaction cases for ( a ) tise@xmath2 - 5znr and ( b ) tise@xmath2 - 5anr.,width=321 ] after analyzing the structural and magnetic properties , we investigated the band dispersion of the tise@xmath2-nrs . electronic structures of tise@xmath2-nrs show similar behavior like the single - layer 1t - tise@xmath2 . in fact , we found that reducing the dimensionality from 2d to 1d , at a certain ribbon width a metal to semiconductor transition is found for both zigzag and armchair nanoribbons as seen in fig . [ gapvswidth ] . the band gap decays monotonically with the ribbon width for armchair nanoribbons , however for zigzag nanoribbons the rapid band gap decrease is superposed with an even - odd oscillation with increasing @xmath30 and finally both structures switches to the zero energy gap of monolayer tise@xmath2 ( for @xmath30 @xmath40 , and @xmath31 @xmath41 ) . similar oscillatory behavior is also observed in the equilibrium lattice constant for tise@xmath2-@xmath30znrs , when we increase the ribbon width @xmath30 , the lattice constant approached slowly the value @xmath24 which is the same as that calculated for the @xmath5d - tise@xmath2 . the edge reconstructions are more effective in changing the equilibrium lattice constant of ultra narrow ribbons . as illustrated in fig . [ gapvswidth ] , the band gaps as a function of ribbon width for both zigzag and armchair - edged nanoribbons decay very rapidly , except for a small superposed oscillation observed in ultranarrow zigzag nanoribbons . similar band gap oscillations as a function of ribbon width were also predicted for other semiconducting nanoribbons@xcite . nevertheless , due to the rapid decay in both types of nanoribbons , to provide a quantitative measure for these decays the band gap variations are fitted to the exponential functions , e@xmath42(n)= @xmath6 exp(-n@xmath43 ) , where n is the width of the nanoribbon ( for znrs n=@xmath30 and for anrs n=@xmath31 ) , and @xmath6 and @xmath43 are fitting parameters . for armchair and zigzag nanoribbons , the values of the fitting parameters are found to be @xmath6=@xmath44 , @xmath43=@xmath45 ev and @xmath6=@xmath46 , @xmath43=@xmath47 ev , respectively . for @xmath48 , both types of nanoribbons show metallic behavior . ( color online ) electronic band structure of a series of zigzag and armchair nanoribbons of 1t - tise@xmath2 by using the pbe method.,width=321 ] spin - unpolarized band structures of tise@xmath2-@xmath30znrs are presented in fig . [ bands ] . notice that the band structures show similar property at the x - point for odd and even numbers of ribbon width . for the ribbon width of @xmath30=2 a large gap of about @xmath49 ev is found . among the four znrs in fig . [ bands ] , tise@xmath2 - 4znr has the largest band gap of @xmath50 ev , tise@xmath2 - 3znr has a medium band gap of @xmath51 ev , tise@xmath2 - 5znr and tise@xmath2 - 6znr have the smallest band gaps of @xmath52 mev and @xmath53 mev , respectively . both the direct band gap of zigzag - edged and the indirect band gap of armchair - edged ultranarrow nanoribbons decrease with increasing ribbon width and eventually vanish for @xmath30 @xmath40 , and @xmath31 @xmath41 . the conduction band minimum ( cbm ) and the valance band maximum ( vbm ) cross resulting in a semimetallic band structure with overlapping bands . in order to investigate this width - dependent transition in the band structure , as well as the odd - even variations observed in the narrowest znrs , we have considered partial charge density ( pcd ) profiles corresponding to vbm and cbm , or for some specific pair of points in the band structures . these pair of points are m1 and m2 for 2d - tise2 ( fig . [ bandmonotise2 ] ) , z1 and z2 for znrs , and a1 and a2 for anrs ( fig . [ bands ] ) . the pcd plots of the vbm and the cbm as shown in fig . [ zigzagband - dec ] indicate the electronic states around the fermi level . for tise@xmath2 - 3znr ( tise@xmath2 - 4znr ) , the vbm and the cbm originate from a hybridized mixture of 3d electrons of ti and 4p electrons of se atoms with the hybridization being stronger in the vbm than that in the cbm . a comparison of the vbm states of tise@xmath2 - 3znr and tise@xmath2 - 4znr indicate that they are localized more at the edges for odd @xmath30 , whereas they are more uniform distributed for even @xmath30 ribbons . for wider ribbons ( @xmath30 @xmath54 @xmath55 ) , both the vbm and cbm states tend to delocalize and the metallic character is attained ( this is evident for @xmath30=7 and @xmath30=8 in fig . [ zigzagband - dec ] ) . with increasing @xmath30 , the pcd plots at the z1 and z2 points tend to converge to those at the m1 and m2 pair for 2d - tise@xmath2 , where the corresponding states are localized on the se and ti atoms , respectively . the opening of a band gap in very narrow ribbons can be attributed to quantum size effects . ( color online ) band decomposed charge density plots of monolayer and @xmath30=3,4,7,8 nanoribbons of tise@xmath2 where z1 and z2 are shown in the band - structures ( see fig . [ bands ] ) . inset shows the @xmath11-point charge densities of m1 and m2 band edges ( shown in fig . [ bandmonotise2 ] ) of 2d tise@xmath2.,width=321 ] typical band structures for a series of armchair tise@xmath2 nanoribbons are also shown in fig . [ bands ] . unlike zigzag nanoribbons , the electronic structure of the armchair ribbons exhibit an indirect band gap for @xmath31 @xmath566 . the gap decreases exponentially with the ribbon width . the band gap is almost halved when the ribbon width is increased from @xmath31=2 to @xmath31=4 . tise@xmath2-@xmath53anr still has a band gap of about @xmath57 mev . starting with @xmath31=6 , the cbm dips into the fermi level , so that the armchair nanoribbons become metallic for wider widths . some partial charge density plots for tise@xmath2-@xmath31anrs are also illustrated in fig . [ armband - dec ] . similarly , the vbm and cbm states are composed of a hybridized mixture of ti-3d and se-4p orbitals for small nanoribbons , however for the ribbon width larger than four , the hybridization becomes lost . ( color online ) tise@xmath2 - 3anr band structure and band decomposed charge densities of @xmath31=3,4,7,8 nanoribbons of tise@xmath2 where the a1 and a2 refer to the states indicated in fig . [ bands].,width=321 ] in order to investigate the effect of dangling states present at the edges of the nanoribbons , we have passivated the edge atoms by hydrogen atoms . these unsaturated bonds influence the electronic properties of the ribbons . naturally these states do not exist in the infinite tise@xmath2 single layer , therefore reducing dimensionality from 2d to 1d it will be of importance control the dangling bonds . earlier , it was shown for graphene nanoribbons that when the dangling bonds at the edges are passivated with hydrogen atoms the electronic and magnetic properties of the ribbons are modified.@xcite unlike graphene , the tise@xmath2-nrs have two types of atoms at the edges so that both ti and se atoms have to be passivated by hydrogen atoms to compensate the edge states . ( color online ) passivation of the edge states with hydrogen atoms ( blue colored ) for the zigzag and armchair nanoribbons.,width=321 ] among possible configurations for the edge termination with hydrogen atoms , the most energetically favorable structure is shown for the tise@xmath2 - 4znr in fig . [ hpas - types ] . as seen in the figure where the edge atoms are passivated by hydrogen atoms symmetrically , hydrogenation of the nanoribbons also enhances the stability of the structures . after hydrogenation the ground state energies is lowered , and the binding energy is found to be @xmath58 ev for the case of tise@xmath2 - 4znr . the band structures for several hydrogenated znrs are shown in fig . [ band - hpas ] . the tise@xmath2-@xmath30znrs are all metallic except for @xmath30=4 . ( color online ) band structures for zigzag and armchair nanoribbons where the edge atoms are passivated by hydrogen atoms.,width=321 ] we performed a analysis for the armchair nanoribbons . in tise@xmath2 - 3anr , as an example shown in fig . [ hpas - types ] , the edge se and ti atoms are passivated by one and two hydrogen atoms , respectively . the binding energy of the tise@xmath2 - 3anr is 23.4 ev . after the hydogenation , tise@xmath2 - 3anr and tise@xmath2 - 4anr are semiconductors with an increasing band gap . also , the vbm state moves a little away from the @xmath11-point in case of @xmath31=3 . tise@xmath2 - 7anr and tise@xmath2 - 8anr are still metallic after hydrogenation , however the overlap of the conduction and valance bands is reduced . in this work , we have investigated the electronic and magnetic properties of zigzag and armchair - edged tise@xmath2 nanoribbons by means of first - principles calculations . overall , our results demonstrate that these tmd nanoribbons which are in 1 t phase have quite different characteristics from nanoribbons of other widely studied materials such as graphene or mos@xmath2 . our calculations revealed that only ultranarrow zigzag and armchair nanoribbons exhibit semiconducting behavior and their band gap rapidly decreases to zero with increasing ribbon width . @xmath31 @xmath59 6 and @xmath30 @xmath59 7 nanoribbons exhibit metallic behavior like two - dimensional tise@xmath2 . the width dependency of the band gap can be fairly represented by an exponential decay function . both zigzag and armchair ribbons have nonmagnetic ground states . in addition , the robust metallic behavior of both zigzag and armchair tise@xmath2 nanoribbons remains unaltered even after passivation of the edges by hydrogen atoms . the metallic character of the wider ribbons of tise@xmath2 regardless of their edge symmetry is an advantageous property for utilizing them as one - dimensional interconnects of nanoscale circuits . this work was supported by the flemish science foundation ( fwo - vl ) and the methusalem foundation of the flemish government . computational resources were provided by tubitak ulakbim , high performance and grid computing center ( tr - grid e - infrastructure ) . hs is supported by a fwo pegasus long marie curie fellowship . jk is supported by a fwo pegasus short marie curie fellowship . hdo , hs and rts acknowledge the support from tubitak through project 114f397 . j. n. coleman , m. lotya , a. oneill , s. d. bergin , p. j. king , u. khan , k. young , a. gaucher , s. de , r. j. smith , i. v. shvets , s. k. arora , j. j. boland , j. j. wang , j. f. donegan , j. c. grunlan , g. moriarty , a. shmeliov , r. j. nicholls , j. m. perkins , e. m. grieveson , k. theuwissen , d. w. mc comb , p. d. nellist , and v. nicolosi , science * 331 * , 568 ( 2011 ) . f. iyikanat , h. sahin , r. t. senger , and f. m. peeters , vacancy formation and oxidation characteristics of single layer tis@xmath3 [ just accepted ] , doi : 10.1021/acs.jpcc.5b01562 . published online : april , 28 , 2015 .
motivated by the recent synthesis of single layer tise@xmath0 , we used state - of - the - art density functional theory calculations , to investigate the structural and electronic properties of zigzag and armchair - edged nanoribbons of this material . our analysis reveals that , differing from ribbons of other ultra - thin materials such as graphene , tise@xmath0 nanoribbons have some distinctive properties . the electronic band gap of the nanoribbons decreases exponentially with the width and vanishes for ribbons wider than @xmath1 angstroms . for ultranarrow zigzag - edged nanoribbons we find odd - even oscillations in the band gap width , although their band structures show similar features . moreover , our detailed magnetic - ground - state analysis reveals that zigzag and armchair edged ribbons have nonmagnetic ground states . passivating the dangling bonds with hydrogen at the edges of the structures influences the band dispersion . our results shed light on the characteristic properties of t phase nanoribbons of similar crystal structures .
many graph optimisation problems do not admit an exact solution by a fast distributed algorithm . this is true not only for most np - hard optimisation problems , but also for problems that can be solved using sequential polynomial - time algorithms . this work is a contribution to the _ distributed approximability _ of such a problem : the minimum vertex cover problem on bipartite graphs we call it @xmath5 , for short . our focus is on _ negative _ results : we prove an optimal ( up to constants ) time lower bound @xmath6 for a randomised distributed algorithm to find a close - to - optimal vertex cover on bipartite 2-coloured graphs of maximum degree @xmath7 . in particular , this rules out the existence of a sublogarithmic - time approximation scheme for @xmath5 on sparse graphs . our lower bound result exhibits the following features : * the proof is relatively simple as compared to the strength of the result ; this is achieved through an application of _ expander graphs _ in the lower - bound construction . * to explain the source of hardness for @xmath5 we introduce a certain _ distributed cut minimisation problem _ , which might have applications elsewhere . * many previous distributed inapproximability results are based on the hardness of _ local symmetry breaking_. this is not the case here : the difficulty we pinpoint for @xmath5 is in the semi - global task of _ gluing together _ two different types of local solutions . * our result states that _ knig s theorem is non - local_see section [ sec : konig ] . we work in the standard @xmath8 model of distributed computing @xcite . as input we are given an undirected graph @xmath9 . we interpret @xmath10 as defining a communication network : the nodes @xmath11 host processors , and two processors can communicate directly if they are connected by an edge . all nodes run the same distributed algorithm @xmath12 . the computation of @xmath12 on @xmath10 starts out with every node @xmath13 knowing an upper bound on @xmath14 and possessing a globally unique @xmath15-bit identifier @xmath16 ; for simplicity , we assume that @xmath17 and @xmath18 . also , we assume that the processors have access to independent ( and unlimited ) sources of randomness . the computation proceeds in synchronous communication rounds . in each round , all nodes first perform some local computations and then exchange ( unbounded ) messages with their neighbours . after some @xmath19 communication rounds the nodes stop and produce local outputs . here @xmath19 is the _ running time _ of @xmath12 and the output of @xmath20 is denoted @xmath21 . the fundamental limitation of a distributed algorithm with running time @xmath19 is that the output @xmath21 can only depend on the information available in the subgraph @xmath22\subseteq g$ ] induced on the vertices in the radius-@xmath19 neighbourhood ball @xmath23 conversely , it is well known that an algorithm @xmath12 can essentially discover the structure of @xmath22 $ ] in time @xmath19 . thus , @xmath12 can be thought of as a function mapping @xmath19-neighbourhoods @xmath22 $ ] ( together with the additional input labels and random bits on @xmath24 ) to outputs . while the @xmath8 model abstracts away issues of network congestion and asynchrony , this only makes our _ lower - bound _ result stronger . below , we concentrate on bipartite 2-coloured graphs @xmath10 . that is , @xmath10 is not only bipartite ( which is a global property ) , but every node @xmath20 is informed of the bipartition by an additional input label @xmath25 , where @xmath26 is a proper 2-colouring of @xmath10 . in the @xmath5 problem we are given a 2-coloured graph @xmath27 and the objective is to output a minimum - size vertex cover of @xmath10 . a distributed algorithm @xmath12 computes a vertex cover by outputting a single bit @xmath28 on a node @xmath20 indicating whether @xmath20 is included in the solution . this way , @xmath12 computes the set @xmath29 . moreover , we say that @xmath12 computes an _ @xmath30-approximation _ of @xmath5 if @xmath31 is a vertex cover of @xmath10 and @xmath32 where @xmath33 denotes the size of a minimum vertex cover of @xmath10 . our main result is the following . [ thm : main ] there exists a @xmath2 such that no randomised distributed algorithm with running time @xmath3 can find a @xmath4-approximation of @xmath5 on graphs of maximum degree @xmath7 . a matching time upper bound follows directly from the well - known network decomposition algorithm due to linial and saks @xcite . for every @xmath0 a @xmath1-approximation of @xmath5 can be computed with high probability in time @xmath34 on graphs of maximum degree @xmath35 . the subroutine _ construct_block _ in the algorithm of linial and saks @xcite computes , in time @xmath36 , a set @xmath37 with the following properties . each component in the subgraph @xmath38 $ ] induced by @xmath39 has _ weak diameter _ at most @xmath19 , i.e. , @xmath40 for each pair @xmath41 belonging to the same component of @xmath38 $ ] . moreover , they prove that , w.h.p . , @xmath42 let @xmath43 be a component of @xmath38 $ ] . every node of @xmath43 can discover the structure of @xmath43 in time @xmath44 by exploiting its weak diameter . thus , every node of @xmath43 can internally compute _ the same _ optimal solution of @xmath5 on @xmath43 . we can then output as a vertex cover for @xmath10 the union of the optimal solutions at the components together with the vertices @xmath45 . this results in a solution of size at most @xmath46 } + \epsilon n.\ ] ] but since @xmath47 for connected @xmath10 , this is a @xmath48-approximation of @xmath5 . the classic theorem of knig ( see , e.g. , diestel @xcite ) states that , on bipartite graphs , the size of a maximum matching equals the size of a minimum vertex cover . a modern perspective is to view this result through the lens of linear programming ( lp ) duality . the lp relaxations of these problems are _ the fractional matching problem _ ( primal ) and _ the fractional vertex cover problem _ ( dual ) : @xmath49 in particular , on bipartite graphs , the above lps do not have an integrality gap ( see , e.g. , papadimitriou and steiglitz @xcite ) : among the optimal feasible solutions are integral vectors @xmath50 and @xmath51 that correspond to maximum matchings and minimum vertex covers . in the context of distributed algorithms , the following is known on ( bipartite ) _ bounded degree graphs _ : 1 . _ primal lp and dual lp admit local approximation schemes . _ as part of their general result , kuhn et al . @xcite give a strictly local @xmath1-approximation scheme for the above lps . their algorithms run in constant time independent of the number of nodes . integral primal admits a local approximation scheme . _ strand et al . @xcite describe a strictly local @xmath1-approximation scheme for the maximum matching problem on 2-coloured graphs . again , the running time is a constant independent of the number of nodes . integral dual does not admit a local approximation scheme . _ the present work shows in contrast to the above positive results that there is no local approximation scheme for @xmath5 even when @xmath7 . there are relatively few independent methods for obtaining negative results for distributed approximation in the @xmath8 model . we list three main sources . [ [ local - algorithms . ] ] local algorithms . + + + + + + + + + + + + + + + + + linial s @xcite lower bound @xmath52 for @xmath53-colouring a cycle together with the ramsey technique of naor and stockmeyer @xcite establish basic limitations on finding _ exact solutions _ strictly locally in constant time . these impossibility results were later extended to finding _ approximate solutions _ on cycle - like graphs by lenzen and wattenhofer @xcite and czygrinow et al . a recent work @xcite generalises these techniques even further and proves that deterministic local algorithms in the @xmath8 model are often no more powerful than algorithms running on anonymous port numbered networks . for more information on this line of research , see the survey of local algorithms @xcite . here , the inapproximability results typically exploit the inability of a local algorithm to break local symmetries . by contrast , in this work , we consider the case where the local symmetry is already broken by a @xmath54-colouring . [ [ kmw - bounds . ] ] kmw - bounds . + + + + + + + + + + + kuhn , moscibroda and wattenhofer @xcite prove that any randomised algorithm for computing a constant - factor approximation of minimum vertex cover on general graphs requires time @xmath55 and @xmath56 . roughly speaking , their technique consists of showing that a fast algorithm can not tell apart two adjacent nodes @xmath20 and @xmath57 , even though it is globally more profitable to include @xmath20 in the vertex cover and exclude @xmath57 than conversely . the lower - bound graphs of kuhn et al . are necessarily of unbounded degree : on bounded degree graphs the set of all non - isolated nodes is a constant factor approximation of a minimum vertex cover . by contrast , our lower - bound graphs are of bounded degree and they forbid close - to - optimal approximation of @xmath5 . [ [ sublinear - time - centralised - algorithms . ] ] sublinear - time centralised algorithms . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + parnas and ron @xcite discuss how a fast distributed algorithm can be used as _ solution oracle _ to a centralised algorithm that approximates parameters of a sparse graph @xmath10 in sublinear time given a randomised query access to @xmath10 . thus , lower bounds in this model of computation also imply lower bounds for distributed algorithms . in particular , an argument of trevisan ( presented in @xcite ) implies that computing a @xmath58-approximation of a minimum vertex cover requires @xmath6 time on @xmath59-regular graphs , where @xmath60 is sufficiently large . we note that @xmath5 is easy to approximate in this model : nguyen and onak @xcite give a centralised constant - time algorithm to approximate the size of a maximum matching on a graph @xmath10 . if we are promised that @xmath10 is bipartite , the same algorithm approximates the size of @xmath5 by knig duality . to best explain the basic idea of our lower bound result , we first prove theorem [ thm : main ] for a toy model that we define in section [ ssec : toy ] ; in this model , we only consider a certain class of deterministic distributed algorithms in anonymous networks . later in section [ sec : randomised ] we will show how to implement the same proof technique in a much more general setting : randomised distributed algorithms in networks with unique identifiers . in the present section , we find a source of hardness for @xmath5 as follows . first , we argue that any approximation algorithm for the @xmath5 problem also solves a certain cut minimisation problem called @xmath61 . we then show that @xmath61 is hard to approximate locally , which implies that @xmath5 must also be hard to approximate locally . throughout this section we consider deterministic algorithms @xmath12 running in time @xmath62 that operate on _ input - labelled anonymous networks _ @xmath63 , where @xmath9 and @xmath64 is a labelling of @xmath11 . more precisely , we impose the following additional restrictions in the @xmath8 model : * the nodes of @xmath10 are not given random bits as input . * the output of @xmath12 is invariant under reassigning node identifiers . that is , if @xmath65 is isomorphic to @xmath10 via a mapping @xmath66 , then for @xmath13 , @xmath67 put otherwise , the only symmetry breaking information we supply @xmath12 is the radius-@xmath19 neighbourhood topology together with the input labelling the nodes are anonymous and do not have unique identifiers . we will also consider graphs @xmath10 that are _ directed_. in this case , the directions of the edges are merely additional symmetry - breaking information ; they do not restrict communication . problem . in this example , we have used a simple distributed algorithm @xmath12 to find a recut @xmath68 with a small boundary @xmath69 : a node outputs red iff there is a red node within distance @xmath70 in the input . while the solution is not optimal , in a grid graph the boundary will be relatively small . however , our lower bound shows that any fast distributed algorithm including algorithm @xmath12fails to produce a small boundary in some graph . ] in the following , we consider partitions of @xmath11 into red and blue colour classes as determined by a labelling @xmath71 . we write @xmath72 for the fraction of edges crossing the red / blue cut , i.e. , @xmath73 in _ the recut problem _ we are given a labelled graph @xmath63 as input and the objective is to compute an output labelling ( _ a recut _ ) @xmath68 that minimises @xmath74 subject to the following constraints : ( a ) if @xmath75 , then @xmath76 . ( b ) if @xmath77 , then @xmath78 . in words , if we have an all - red input , we have to produce an all - red output , and if we have an all - blue input , we have to produce an all - blue output . otherwise the output can be arbitrary . see figure [ fig : recut ] for an illustration . needless to say , the global optimum for an algorithm @xmath12 would be to produce a constant output labelling @xmath79 ( either all red or all blue ) having @xmath80 . however , a distributed algorithm @xmath12 can only access the values of the input labelling @xmath64 in its local radius-@xmath19 neighbourhood : when encountering a neighbourhood @xmath81 with @xmath82 , the algorithm is forced to output red at @xmath20 to guarantee satisfying the global constraint ( a ) , and when encountering a neighbourhood @xmath81 with @xmath83 , the algorithm is forced to output blue at @xmath20 to satisfy ( b ) . thus , if a connected graph @xmath10 has two disjoint @xmath19-neighbourhoods @xmath84 with @xmath82 and @xmath85 the algorithm @xmath12 can not avoid producing at least some red / blue edge boundary . indeed , the best we can hope @xmath12 to achieve is a recut @xmath79 of size @xmath86 for some small constant @xmath87 . [ [ discussion . ] ] discussion . + + + + + + + + + + + the @xmath61 problem models the following abstract high - level challenge in designing distributed algorithms : each node in a local neighbourhood @xmath88 can , in principle , internally compute a completely _ locally optimal _ solution for ( the subgraph induced by ) @xmath89 , but difficulties arise when deciding which of these proposed solution are to be used in the final distributed output . in particular , when the _ type _ of the produced solution changes from one ( e.g. , red ) to another ( e.g. , blue ) across a graph @xmath10 one might have to introduce suboptimalities to the solution at the ( red / blue ) boundary in order to glue together the different types of local solutions . in fact , the @xmath61 problem captures the first non - trivial case of this phenomenon with only _ two _ solution types present . one can think of the input labelling @xmath64 as recording the _ initial preferences _ of the nodes whereas the output labelling @xmath79 records how an algorithm @xmath12 decides to combine these preferences into the final unified output . in the end , our lower - bound strategy will be to argue that any @xmath12 can be forced into producing too large an edge boundary @xmath90 resulting in too many suboptimalities in the produced output . next , we show how the above discussion is made concrete in the case of the @xmath5 problem . we call a graph @xmath10 _ tree - like _ if all the @xmath19-neighbourhoods in @xmath10 are trees , i.e. , @xmath10 has girth larger than @xmath91 . furthermore , if @xmath10 is directed , we say it is _ balanced _ if @xmath92 for all vertices @xmath20 . we note that a deterministic algorithm @xmath12 produces the same output on every node of a balanced regular tree - like digraph @xmath10 , because such a graph is _ locally homogeneous _ : all the @xmath19-neighbourhoods of @xmath10 are pairwise isomorphic . using this terminology we give the following reduction . [ thm : reduction ] suppose @xmath12 computes a @xmath1-approximation of @xmath5 on graphs of maximum degree @xmath7 . then , there is an algorithm ( with running time @xmath19 ) that finds a recut @xmath79 of size @xmath93 on balanced @xmath94-regular tree - like digraphs . the proof of theorem [ thm : reduction ] follows the usual route : we give a local reduction ( i.e. , one that can be computed by a local algorithm ) that transforms an instance @xmath63 of @xmath61 into an instance @xmath95 of @xmath5 . then we simulate @xmath12 on the resulting instance and map the output of @xmath12 back to a solution of the @xmath61 instance @xmath63 . let @xmath9 be a balanced @xmath94-regular tree - like digraph and let @xmath71 be a labelling of @xmath10 . the instance @xmath95 is obtained by replacing each vertex @xmath13 by one of two local gadgets depending on the label @xmath96 . we first describe and analyse simple gadgets yielding instances of @xmath5 with @xmath97 ; the gadgets yielding instances with @xmath98 are described later . the red gadget replaces a vertex @xmath13 by two new vertices @xmath99 ( white ) and @xmath100 ( black ) that share a new edge @xmath101 . the incoming edges of @xmath20 are reconnected to @xmath99 , whereas the outgoing edges of @xmath20 are reconnected to @xmath100 . note that the @xmath5 instance @xmath102 ( where we denote by @xmath103 the constant labelling @xmath104 ) contains @xmath105 as a perfect matching . since @xmath106 is locally homogeneous , the solutions output by @xmath12 on the endpoints of @xmath107 are isomorphic across all @xmath20 . assuming @xmath108 it follows that the algorithm @xmath12 must output either the set of all white nodes or the set of all black nodes on @xmath102 . our reduction branches at this point : we choose the structure of the blue gadget to counteract this white / black decision made by @xmath12 on the red gadgets . we describe the case that @xmath12 outputs all white nodes on @xmath102 ; the case of black nodes is symmetric . the blue gadget replacing @xmath13 is identical to the red gadget with the exception that a third new vertex @xmath109 ( white ) is added and connected to @xmath100 . similarly as above , we can argue that @xmath12 outputs exactly the set of all black nodes on the instance @xmath110 . this completes the description of @xmath111 . next , we simulate @xmath12 on @xmath95 . the output of @xmath12 is then transformed back to a labelling @xmath112 by setting @xmath113 note that @xmath79 satisfies both feasibility constraints ( a ) and ( b ) of @xmath61 . it remains to bound the size @xmath90 of this recut . [ [ recut - analysis . ] ] recut analysis . + + + + + + + + + + + + + + + call a red vertex @xmath20 in @xmath114 _ bad _ if @xmath20 has a blue out - neighbour @xmath57 . by the definition of `` @xmath115 '' , the vertex cover produced by algorithm @xmath12 does not contain the white node @xmath116 . thus to cover the edge @xmath117 , the vertex cover has to contain the black node @xmath100 . but by the definition of `` @xmath118 '' , we must have @xmath99 or @xmath109 in the solution as well . hence , at least two nodes are used to cover the gadget at @xmath20 , which is suboptimal as compared to the minimum vertex cover @xmath119 , which uses only one node per gadget . this implies that we must have at most @xmath120 bad vertices as @xmath12 produces a @xmath1-approximation of @xmath5 on @xmath95 . on the other hand , exactly half of the edges crossing the cut @xmath79 are oriented from red to blue since @xmath10 is balanced . each bad vertex gives rise to at most two of these edges , so we have that @xmath121 which gives @xmath122 , as required . this proves theorem [ thm : reduction ] for @xmath97 . [ [ gadgets - for - delta-3 . ] ] gadgets for @xmath7 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the maximum degree used in the gadgets can be reduced to 3 by the following modification . the red gadget replaces a vertex @xmath13 by a path of length 3 . again , to achieve a @xmath123-approximation of @xmath5 on @xmath102 the algorithm @xmath12 has to make a choice : either leave out the middle black vertex or the middle white vertex from the vertex cover . supposing @xmath12 leaves out the middle black , the blue gadget is defined to be identical to the red gadget with an additional white vertex connected to the middle black one . after simulating @xmath12 on an instance @xmath95 we define @xmath124 iff @xmath12 outputs only black nodes at the gadget at @xmath20 . the recut analysis will then give @xmath125 . intuitively , the difficulty in computing a small red / blue cut in the @xmath61 problem stems from the inability of an algorithm @xmath12 to overcome the neighbourhood expansion of an input graph in @xmath62 steps an algorithm can not hide the red / blue boundary as the radius-@xmath19 neighbourhoods themselves might have large boundaries . to formalise this intuition , we use as a basis for our lower - bound construction an infinite family @xmath126 of @xmath94-regular _ @xmath127-expander graphs _ , where each @xmath128 satisfies the edge expansion condition @xmath129 here , @xmath130 is the number of edges leaving @xmath39 and @xmath2 is an absolute constant independent of @xmath14 . on such graphs it is enough for us to force an algorithm @xmath12 to output a _ nearly balanced recut _ @xmath79 having both colour classes close to @xmath131 in size . this is because if the number of , say , the red nodes is @xmath132 then the expansion property implies that @xmath133 that is , @xmath12 computes a recut of size @xmath134 . indeed , the following simple fooling trick makes up the very core of our argument . [ lem : fooling ] suppose @xmath12 produces a feasible solution for the @xmath61 problem in time @xmath62 . then for each @xmath94-regular graph @xmath10 there exists an input labelling for which @xmath12 computes a nearly balanced recut . fix an arbitrary ordering @xmath135 for the vertices of @xmath10 and define a sequence of labellings @xmath136 by setting @xmath137 iff @xmath138 . that is , in @xmath139 all nodes are red , in @xmath140 all nodes are blue , and @xmath141 is obtained from @xmath142 by changing the colour of @xmath143 from red to blue . when we switch from the instance @xmath144 to @xmath145 the change of @xmath143 s colour is only registered by nodes in the radius-@xmath19 neighbourhood of @xmath143 . this neighbourhood has size @xmath146 , and so the number of red nodes in the outputs @xmath147 and @xmath148 of @xmath12 can only differ by @xmath149 . as , by assumption , we have that @xmath12 computes the labelling @xmath150 on @xmath151 and the labelling @xmath152 on @xmath153 , it follows by continuity that some labelling in our sequence must force @xmath12 to output @xmath154 red nodes . we now have all the ingredients for the lower - bound proof : we can take @xmath155 if we choose @xmath126 to be the family of @xmath94-regular ramanujan graphs due to morgenstern @xcite . these graphs are tree - like , as they have girth @xmath156 . they can be made into balanced digraphs since a suitable orientation can always be derived from an euler tour . thus , @xmath126 consists of balanced @xmath94-regular tree - like digraphs . lemma [ lem : fooling ] together with the discussion above imply that every algorithm for @xmath61 produces a recut of size @xmath134 on some labelled graph in @xmath126 . hence , the contrapositive of theorem [ thm : reduction ] proves theorem [ thm : main ] for our deterministic toy algorithms . even though our model of deterministic algorithms in section [ sec : deterministic ] is an unusually weak one , we can quickly recover the standard @xmath8 model from it by equipping the nodes with independent sources of randomness . in particular , as is well known , each node can choose an identifier uniformly at random from , e.g. , the set @xmath157 , and this results in the identifiers being globally unique with probability at least @xmath158 . when discussing randomised algorithms many of the simplifying assumptions made in section [ sec : deterministic ] no longer apply . for example , a randomised algorithm need not produce the same output on every node of a locally homogeneous graph . consequently , the homogeneous feasibility constraints in the @xmath61 problem do not strictly make sense for randomised algorithms . however , we can still emulate the same proof strategy as in section [ sec : deterministic ] : we force the randomised algorithm to output a nearly balanced recut with high probability . below , we describe this strategy in case of the easy - to - analyse `` @xmath97 '' gadgets with the understanding that the same analysis can be repeated for the `` @xmath7 '' gadgets with little difficulty . fix a randomised algorithm @xmath12 with running time @xmath62 and let @xmath9 , @xmath14 , be a large @xmath94-regular expander . again , we start out with the all - red instance . we denote by @xmath159 and @xmath160 the number of black and white nodes output by @xmath12 on @xmath102 . as each of the edges @xmath107 must be covered , we have that @xmath161 hence , by linearity of expectation , at least one of @xmath162 \geq n/2 $ ] or @xmath163 \geq n/2 $ ] holds . we assume that @xmath162 \geq n/2 $ ] ; the other case is symmetric . in reaction to @xmath12 preferring white nodes , the blue gadgets are now defined exactly as in section [ sec : deterministic ] with an additional white vertex . furthermore , for any input @xmath71 we interpret the output of @xmath12 on @xmath95 as defining an output labelling @xmath79 of @xmath11 , where , again , @xmath124 iff @xmath12 outputs only the black node at the gadget at @xmath20 . this definition translates our assumption of @xmath162 \geq n/2 $ ] into @xmath164 \geq n/2,\ ] ] where @xmath165 counts the number of gadgets ( i.e. , vertices of @xmath10 ) relabelled red by @xmath12 on @xmath95 . if @xmath12 relabels a blue gadget red , it must output at least two nodes at the gadget . this means that the size of the solution output by @xmath12 on @xmath110 is at least @xmath166 . thus , if @xmath12 is to produce a @xmath167-approximation on @xmath110 in expectation , we must have that @xmath168 \leq n/2.\ ] ] the inequalities and provide the necessary boundary conditions ( replacing the feasibility constraints of @xmath61 ) for the argument of lemma [ lem : fooling ] : by continuously changing the instance @xmath106 into @xmath169 we may find an input labelling @xmath170 achieving @xmath171 = n/2 - o(n).\ ] ] it remains to argue that @xmath12 outputs a nearly balanced recut not only `` in expectation '' but also with high probability . focusing on the instance @xmath172 we write @xmath173 and @xmath174 where @xmath175 indicates whether @xmath12 relabels the gadget at @xmath20 red . the variables @xmath176 are not _ too _ dependent : the @xmath177th power of @xmath10 , denoted @xmath178 , where @xmath179 are joined by an edge iff @xmath180 , is _ a dependency graph _ for the variables @xmath176 . every independent set @xmath181 in @xmath178 corresponds to a set @xmath182 of mutually independent random variables . since the maximum degree of @xmath178 is at most @xmath183 , this graph can always be partitioned into @xmath184 independent sets . indeed , janson @xcite presents large deviation bounds for sums of type by applying chernoff hoeffding bounds for each colour class in a @xmath185-colouring of @xmath178 . for any @xmath0 , theorem 2.1 in janson @xcite , as applied to our setting , gives @xmath186 + \epsilon n ) \leq \exp\left(-2\frac{(\epsilon n)^2}{\chi(g^{2r})\cdot n}\right ) \to 0,\quad\text{as}\ n\to \infty,\ ] ] and the same bound holds for @xmath187 - \epsilon n\right)$ ] . that is , @xmath188 is concentrated around its expectation . in conclusion , the combination of and implies that , for large @xmath189 , the algorithm @xmath12 outputs a nearly balanced recut on @xmath172 with high probability . by the discussion in section [ sec : deterministic ] , this proves theorem [ thm : main ] . many thanks to valentin polishchuk for discussions . this work was supported in part by the academy of finland , grants 132380 and 252018 . http://arxiv.org/abs/1002.0125[matti strand , valentin polishchuk , joel rybicki , jukka suomela , and jara uitto . local algorithms in ( weakly ) coloured graphs , 2010 . manuscript , arxiv:1002.0125 [ cs.dc ] . ] http://dx.doi.org/10.1007/978-3-540-87779-0_6[andrzej czygrinow , micha hakowiak , and wojciech wawrzyniak . fast distributed approximations in planar graphs . in _ proc . 22nd symposium on distributed computing ( disc 2008 ) _ , volume 5218 of _ lncs _ , pages 7892 . springer , berlin , 2008 . ] http://dx.doi.org/10.1145/1011767.1011811[fabian kuhn , thomas moscibroda , and roger wattenhofer . what can not be computed locally ! in _ 23rd symposium on principles of distributed computing ( podc 2004 ) _ , pages 300309 . acm press , new york , 2004 . ] http://dx.doi.org/10.1145/1109557.1109666[fabian kuhn , thomas moscibroda , and roger wattenhofer . the price of being near - sighted . in _ proc . 17th symposium on discrete algorithms ( soda 2006 ) _ , pages 980989 . acm press , new york , 2006 . ] http://dx.doi.org/10.1007/978-3-540-87779-0_27[christoph lenzen and roger wattenhofer . leveraging linial s locality limit . in _ proc . 22nd symposium on distributed computing ( disc 2008 ) _ , volume 5218 of _ lncs _ , pages 394407 . springer , berlin , 2008 . ] http://dx.doi.org/10.1109/focs.2008.81[huy n. nguyen and krzysztof onak . constant - time approximation algorithms via local improvements . in _ proc . 49th symposium on foundations of computer science ( focs 2008 ) _ , pages 327336 . ieee computer society press , los alamitos , 2008 . ]
knig s theorem states that on bipartite graphs the size of a maximum matching equals the size of a minimum vertex cover . it is known from prior work that for every @xmath0 there exists a _ constant - time _ distributed algorithm that finds a @xmath1-approximation of a maximum matching on 2-coloured graphs of bounded degree . in this work , we show somewhat surprisingly that no _ sublogarithmic - time _ approximation scheme exists for the dual problem : there is a constant @xmath2 so that no randomised distributed algorithm with running time @xmath3 can find a @xmath4-approximation of a minimum vertex cover on 2-coloured graphs of maximum degree 3 . in fact , a simple application of the linial saks ( 1993 ) decomposition demonstrates that this lower bound is tight . our lower - bound construction is simple and , to some extent , independent of previous techniques . along the way we prove that a certain cut minimisation problem , which might be of independent interest , is hard to approximate locally on expander graphs .
one of the key features of strongly correlated electron systems is their zero temperature behavior where quantum fluctuations have the dominant role to classify different novel phases which are separated by quantum phase transitions @xcite . in this case , ground state ( gs ) is the sole candidate which receives drastic changes at qcp driven by some external parameters . identification of qcps is always a challenging task specially when quantum correlations diminish a single - particle picture . within last couple of years different measures of quantum entanglement have been proposed to be a new toolkit to detect and categorize qcps @xcite . recently , ground state fidelity the overlap of ground state at two slightly different values of coupling constants has attracted intensive attention as a proper quantity to signal qcp without the need to know the structure of phases close to phase transition @xcite . an abrupt drop of fidelity in the vicinity of qcp is a consequence of an essential change in the structure of gs which is usually accompanied with a divergence of fidelity susceptibility . meanwhile , it is rather difficult to obtain the gs of a strongly correlated many body system . however , recent successful attempts to calculate some measures of entanglement @xcite by quantum renormalization group ( qrg ) approach @xcite lead to a unified formalism @xcite to obtain fidelity in the thermodynamic limit ( @xmath1 ) without knowing the gs of a large system . it suggests to implement qrg for obtaining the renormalization of fidelity for the bond alternating ising model with dzyaloshinskii - moriya ( dm ) interaction . dm interaction @xcite which roots to the spin - orbit coupling is the antisymmetric super - exchange which leads to helical magnetic structures as the most likely candidates to host ferroelectricity @xcite . in this article we study the quantum critical behavior of the one dimensional bond alternating ising model with dm interaction in terms of renormalization of fidelity . the model represents an antiferromagnetic ( nel ) long range order for small dm interaction while it undergoes via a continuous phase transition to a helical ordered phase . the ground state in the nel phase is a product state which has essentially zero quantum entanglement and nonzero staggered magnetization while the helical phase poses a correlated quantum ground state with zero staggered magnetization . within a classical picture the helical phase could be assumed as slightly rotating spins along the direction of chain . in terms of landau theory of critical phenomena the staggered magnetization could be chosen as the proper order parameter . we have shown that the divergence in fidelity susceptibility ( fs ) at the qcp is an appropriate signature to find qcp which is more pronounced than the second derivative of ground state energy @xcite . the presented scheme makes us to find the critical points and its corresponding exponents more accurately . the hamiltonian of bond alternating ising chain with dm interaction on a periodic chain of n sites is defined @xmath2,\ ] ] where @xmath3 is the spin-1/2 operator at site @xmath4 , @xmath5 describes the relative strength of the alternating coupling and @xmath6 shows the nearest - neighboring antiferromagnetic coupling . @xmath7 is the vector of dm interaction which is considered in @xmath8-direction , i.e. @xmath9 . to apply qrg @xcite , the spin chain is decomposed to three - sites blocks ( see fig.[blocks ] ) where the intra - block hamiltonian is @xmath10 and the inter - block one is @xmath11 and their sum defines the whole hamiltonian , @xmath12 ( see the bottom part of fig.[blocks ] ) . in this respect we have @xmath13 , where @xmath14s_{l , i}^zs_{l+1,i}^z \nonumber \\ + d(s_{l , i}^xs_{l+1,i}^y - s_{l , i}^ys_{l+1,i}^x)\big),\end{aligned}\ ] ] and similarly @xmath15 , @xmath16s_{3,i}^zs_{1,i+1}^z \nonumber \\ + d(s_{3,i}^xs_{1,i+1}^y - s_{3,i}^ys_{1,i+1}^x)\big).\end{aligned}\ ] ] the block hamiltonian ( @xmath17 ) is diagonalized exactly and the two lowest energy eigenvectors ( @xmath18 ) are kept to construct the embedding operator ( @xmath19 ) to the renormalized hilbert space , @xmath20 . here , @xmath21 , @xmath22 represent the renamed states @xmath23 and @xmath24 , respectively in the renormalized space to be considered as the new base kets . @xmath25 and @xmath26 have the following expression ( for odd blocks ) in the original spin hilbert space where @xmath27 and @xmath28 represent the eigenvectors of @xmath29 operator at each site , @xmath30 where @xmath31 and @xmath32 is the ground state energy of the block . the presence of bond alternation imposes to consider two types of blocks as depicted in fig.[blocks ] , namely even and odd which are the mirror image of each other . for even blocks we find similar eigenstates by replacing @xmath33 , @xmath34 and @xmath35 . ) and inter - block one ( @xmath11 ) are represented schematically in the bottom part.,height=102 ] the global embedding operator is the direct product of the embedding operator of each block , @xmath36 which gives the renormalized hamiltonian by @xmath37 @xcite . the renormalized hamiltonian is similar to the original one , eq.[eq1 ] , while the coupling constants are replaced with the renormalized one ( denoted with @xmath38 ) as given in the following equations , @xmath39 where @xmath40 . these relations , eq.[rgflow ] , define the qrg - flow of our model which will be used in next sections and their features will be discussed in sec.[summary ] . we implement the formalism introduced in @xcite to calculate the ground state fidelity in terms of quantum renormalization group . according to the qrg - flow ( eq.[rgflow ] ) , @xmath0 does not run within the qrg iterations which hints to study the quantum phase transition by variation of @xmath41 . the fidelity ( f ) associated to the ground states @xmath42 for a system of size @xmath43 is defined by @xmath44 where @xmath45 and @xmath46 is a small deviation around @xmath41 . according to the renormalization group approach @xmath47 in which @xmath48 is the global embedding operator and @xmath49 is the ground state of the renormalized hamiltonian . thus , fidelity can be written in terms of renormalized ground state , @xmath50 . a straightforward calculation shows that @xmath51 , \end{aligned}\ ] ] for both even and odd types of blocks where @xmath52 is the identity operator . therefore , the first qrg iteration leads to @xmath53 , where fidelity of the original model is expressed in terms of fidelity of the renormalized one , i.e. @xmath54 . it defines the renormalization of fidelity in terms of qrg . the qrg procedure is iterated @xmath55-times to reach the renormalized system of @xmath56 and the ground state fidelity is expressed by @xmath57 where @xmath58 has the same expression as given in eq.[r0 ] for @xmath59 in which @xmath60 , @xmath61 and @xmath62 are calculated at the @xmath4-th qrg iteration and @xmath63 is the fidelity of a single block with three sites and @xmath55-times renormalized couplings . we have plotted fidelity ( eq.[frg ] ) versus @xmath41 in fig.[fs]-(left ) for different chain length ( @xmath43 ) , @xmath64 and at @xmath65 . by definition , fidelity is bounded by @xmath66 and an abrupt drop is a signature of quantum phase transition . we observe a sharp drop in fig.[fs]-(left ) for @xmath67 which manifests that the ground state has encountered an essential change . the deep of drop is enhanced as the system size is increased which justifies an unfailing drop in the thermodynamic limit . this signature of quantum phase transition is more pronounced in the fidelity susceptibility ( fs ) which is the leading nonzero term in the expansion of fidelity and shows the change rate of fidelity , i.e. @xmath68 . thus , fs is obtained by the following relation @xmath69 fig.[fs]-(right ) presents @xmath70 versus @xmath41 for various system sizes , @xmath64 and at @xmath65 . a maximum appears in @xmath71 which is increased by the size of system representing a divergence in the thermodynamic limit . the position of @xmath72 is exactly at the point where fidelity receives a drop . it has been shown @xcite that the fidelity susceptibility at the quantum critical point obeys a scaling relation . the scaling analysis for finite system size ( @xmath43 ) states @xmath73 is the position of maximum in fs and @xmath74 is the critical exponent governing the divergence of correlation length . the analysis of data of fig.[fs]-(right ) is presented as an inset to this figure . it clearly verifies that the scaling relation eq.[scaling ] is satisfied with @xmath75 and @xmath76 for @xmath65 . moreover , we have obtained similar behavior for different values of @xmath0 ( not presented here ) where the critical point is found to be a function of bond alternating parameter , @xmath77 while @xmath78 is the same for all values of @xmath0 . in contrast to what stated in ref.@xcite the critical exponent which we got does not depend on @xmath0 . it means that the whole phase boundary for @xmath79 belongs to the same universality class of a second order phase transition . we have implemented the quantum renormalization group approach to study the zero temperature phase diagram of bond alternating ising chain in the presence of dm interaction . to get a self similar hamiltonian two types of blocks with 3-sites have been considered which leads to the qrg - flow equations of eq.[rgflow ] . the qrg - flow tells that @xmath0 does not vary within qrg procedure while @xmath41 is renormalized . to get the phase diagram of the model we have calculated the renormalization of ground state fidelity which has been developed recently @xcite . fidelity as a geometric quantity @xcite shows how much the ground state encounters an essential change by slightly moving in the parameter space . therefore , a sharp drop of fidelity versus a control parameter is a signature of quantum phase transition . the renormalization of fidelity obtained in eq.[frg ] in addition to qrg - flow , eq.[rgflow ] give the fidelity of our model for very large system sizes without the need to get the ground state . a clear drop of fidelity versus @xmath41 in fig.[fs]-(left ) and consequently a maximum in the fidelity susceptibility , fig.[fs]-(right ) , verifies the existence of a quantum phase transition at @xmath80 . we have applied the finite size scaling on the susceptibility data presented in the inset of fig.[fs]-(right ) for @xmath65 and generally got the critical phase boundary @xmath81 . the phase boundary which separates the antiferromagnetic ( af ) nel order from a helical order is shown by the blue line in fig.[phase - diagram]-(right ) . however , our scaling analysis gives a single value for the correlation length exponent @xmath82 ( independent of @xmath0 ) for the whole phase boundary which is in contrast to @xcite . although the presence of bond alternation breaks the translation invariance of the hamiltonian it does not change the symmetry of the ground state which has already been spontaneously broken due to antiferromagnetic long range order . a comparison of our results with @xcite concludes that the bond alternation does not change the universality class of the model as far as @xmath83 . we calculate the staggered magnetization ( sm ) and helical order parameter which is presented in fig.[phase - diagram]-(left ) . staggered magnetization is defined by @xmath84 which can be expressed in terms of the renormalized ground state by replacing @xmath47 . we use the projection of spin operators into the renormalized hilbert space which finally leads to @xmath85 where @xmath86 is the staggered magnetization of the renormalized chain . a large number of iterations of eq.[smrg ] give the staggered magnetization in the thermodynamic limit . similarly , the helical order parameter ( @xmath87 ) is defined by the following relation @xcite @xmath88 which can be calculated using the qrg - flow . in this respect , @xmath87 is expressed in terms of the helical order parameter in the renormalized model ( @xmath89 ) by the following relation @xmath90 the above equation is iterated by replacing the couplings with the renormalized ones to reach the stable thermodynamic limit . we have plotted both staggered magnetization and helical order parameter versus @xmath41 in fig.[phase - diagram]-(left ) for three values of @xmath91 and @xmath1 . for @xmath92 , @xmath93 is at its saturated value @xmath94 and @xmath95 . the onset of nonzero @xmath41 induces a helical order on the spins in the presence of antiferromagnetic order . the increase of @xmath41 reduces the antiferromagnetic order and enhances @xmath87 ( helical order ) . exactly at the quantum critical point @xmath80 , @xmath93 vanishes and remains zero for @xmath96 while @xmath87 saturates to a finite amount which is less than its maximum attainable value . similar behavior has been observed for all @xmath79 while the saturated value of @xmath87 is slightly decreased by increase of @xmath0 . to complete the phase diagram let us concentrate on the @xmath92 axis ( fig.[phase - diagram]-(right ) ) . at @xmath97 the model is simply an antiferromagnetic ising chain with nel order @xmath98 and the ground state energy is @xmath99 . for @xmath100 , there are two types of couplings @xmath101 and @xmath102 which are positive and induce the previous nel order and ground state energy . exactly at @xmath103 , the weak interaction @xmath101 becomes zero and the spin model decomposes to @xmath104 independent pairs of antiferromagnetically coupled spins . the ground state energy is still @xmath99 while the ground state is exponentially degenerate , namely @xmath105 . the degeneracy arises from the pairs which are decoupled . for instance , @xmath98 , @xmath106 and @xmath107 are examples of different configurations which is possible for the ground state . the entropy ( s ) is proportional to @xmath108 which leads to @xmath109 . this high amount of entropy at @xmath103 is a signature of a phase transition which is actually of first order . meanwhile , for @xmath110 one of the interactions becomes ferromagnetic , @xmath111 and the other @xmath102 is still antiferromagnetic which totally leads to @xmath112 . thus , the derivative of @xmath113 with respect to @xmath0 receives a discontinuity at @xmath103 . the ground state for @xmath110 is either @xmath114 or @xmath115 . the first order tri - critical point is represented by the filled black circle in fig.[phase - diagram]-(right ) and the ferro - antiferromagnetic ground state is denoted by the red line for @xmath110 . * acknowledgments * we would like to thank a. t. rezakhani for fruitful discussions . this work was supported in part by sharif university of technology s center of excellence in complex systems and condensed matter and the office of vice - president for research . a. l. acknowledges partial support from the alexander von humboldt foundation and max - planck - institut fr physik komplexer systeme ( dresden - germany ) . p. zanardi and n. paunkovi , phys . e * 74 * , 031123 ( 2006 ) ; h .- q . zhou and j. p. barjaktarevi , j. phys . a : math . theor . * 41 * , 412001 ( 2008 ) . s .- b * 24 * , 4371 ( 2010 ) , and references therein . m. a. martin - delgado and g. sierra , int . j. mod , phys . a * 11 * , 3145 ( 1996 ) ; a. langari , phys . b * 58 * , 14467 ( 1998 ) ; * 69 * , 100402(r ) ( 2004 ) . r. jafari , a. langari , phys . b * 76 * , 014412 ( 2007 ) ; physica a * 364 * , 213 ( 2006 ) . p. zanardi , p. giorda , and m. cozzini , phys . lett . * 99 * , 100603 ( 2007 ) ; a. t. rezakhani _ * 103 * , 080502 ( 2009 ) ; a. t. rezakhani _ et al . _ , a * 82 * , 012321 ( 2010 ) .
we present the zero temperature phase diagram of the bond alternating ising chain in the presence of dzyaloshinskii - moriya interaction . an abrupt change in ground state fidelity is a signature of quantum phase transition . we obtain the renormalization of fidelity in terms of quantum renormalization group without the need to know the ground state . we calculate the fidelity susceptibility and its scaling behavior close to quantum critical point ( qcp ) to find the critical exponent which governs the divergence of correlation length . the model consists of a long range antiferromagnetic order with nonzero staggered magnetization which is separated from a helical ordered phase at qcp . our results state that the critical exponent is independent of the bond alternation parameter ( @xmath0 ) while the maximum attainable helical order depends on @xmath0 .
technological advances within the past twenty or so years have led to a greater understanding of the cluster environment . sensitive radio telescopes such as the _ very large array _ ( vla ) have mapped radio emission originating from individual galaxies , with such emission frequently extending well beyond the physical size of the host galaxy ( @xmath3kpc ) . in addition , x - ray satellites such as _ einstein _ and _ rosat _ have probed the hot x - ray emitting gas associated with galaxy clusters . in fact , x - ray observations are now frequently used to identify clusters and compact groups ( see , e.g. , crawford & fabian 1996 , mulchaey _ et al . _ 1996 , ostriker _ _ 1995 , pildis _ et al . _ 1995 , doe _ et al . _ 1995 , burns _ _ 1996 ) . radio studies have not shown a correlation between the probability that a galaxy is a radio source and the degree of local clustering ( see , e.g. , ledlow & owen 1995 ) . specifically , the radio luminosity function does not differ for cluster vs. field radio galaxies ( fanti 1984 , ledlow & owen 1996 ) , nor does the probability for a galaxy to exhibit radio emission depend on global cluster galaxy density ( e.g. , zhao _ et al . _ 1989 , ledlow & owen 1995 ) . these results seem counterintuitive , as all current radio models require an intracluster medium ( icm ) to confine the radio plasma and associated b - fields . furthermore , cluster galaxy radio jets are frequently bent into u - shaped head - tail morphologies which are believed to result from dynamic bending caused by transonic movement through the icm . thus , the existence of very extended radio sources associated with galaxies apparently outside of an icm is difficult to understand . analysis of _ einstein _ data showed that the x - ray emitting gas in clusters is commonly distributed in clumps ( e.g. , mohr _ et al . _ 1993 ) within larger regions of extended emission . radio galaxies are often correlated with the positions of these x - ray clumps ( burns _ et al . _ 1994 ) , although there is some debate over the cause of this correlation ( edge & rttgering 1995 ) . while the radio emission is produced by synchrotron radiation , there are several possible origins for the x - ray emission . if it originates from free - free radiation of hot icm gas , it would be extended and could provide pressure confinement for the radio plasma . extended x - ray emission could also arise from inverse compton scattering of microwave background photons by the relativistic electrons generating the synchrotron emission ( daly 1992 ) , though such emission would be quite weak ( burns _ et al . another viable option for compact x - ray emission is non - thermal emission from active galactic nuclei ( agn ) . this x - ray emission would be unresolved and coincident with the core of radio emission . the question then becomes : are radio galaxies associated with real clumps in the icm or do the peaks in x - ray emission arise from the galaxies themselves ? arguments for each progenitor abound in the literature ( for extended emission from cluster gas , see e.g. feigelson & berg 1983 ; for point emission from agn , see e.g. fabbiano _ et al . some sources may even exhibit both point - like and extended emission , possibly resulting from cooling inflows of the hot cluster gas ( e.g. , see birkinshaw & worrall 1993 for a discussion of ngc6251 ) . it seems likely that different specific circumstances require different explanations ( rhee _ et al . _ 1994 , burns _ 1994 ) . in this paper , a complete nearby sample ( @xmath0 ) of 25 radio galaxies are examined in an attempt to characterize their environments . x - ray emission in the vicinity of the galaxies is investigated through use of the _ rosat all - sky survey _ ( rass ) and _ rosat _ pointed observations . this analysis has two components : first , the determination of whether the radio galaxies are also significant x - ray sources ; and second , whether this emission is extended or unresolved . luminosities are also determined for the purpose of investigating the known correlation of cluster richness and x - ray luminosity . optical data from the digitized sky survey ( dss ) are used to determine spatial two - point correlation functions about each radio source to quantify the galaxy clustering . a list of potential poor clusters is thereby identified as those sources with significant extended x - ray emission and high optical spatial correlation coefficients . furthermore , the specific environments of fanaroff - riley class i and ii ( fr i and fr ii , respectively ; fanaroff & riley 1974 ) radio sources are compared to help elucidate the nature of the origins of these two types of sources . throughout this paper , values of @xmath4 km / sec / mpc and @xmath5 will be adopted . in section 2 , the data analysis and procedure are summarized . this summary includes the selection of the sample in addition to specific x - ray data reduction techniques . a brief overview of spatial two - point correlation functions is provided with a description of the assumptions we have used in our calculations . in section 3 the results are summarized , and section 4 provides a discussion of these results which focuses on correlations between radio , x - ray , and clustering properties . the results and conclusions are summarized in section 5 . the sample was chosen from the 3crr catalog ( laing , riley , & longair 1983 , hereafter lrl ) and the wall & peacock all - sky survey ( wall & peacock 1985 , wp ) . the lrl sample is complete to a flux density limit @xmath6 for sources smaller than 10 in extent . the wp survey was performed at a higher frequency than the lrl survey to avoid bias against flat - spectrum sources , and includes southern sky sources and sources of larger extent . it is complete to @xmath7 over the entire sky for which @xmath8 , except for regions near the magellenic clouds . from this combined survey , all radio sources with @xmath0 and which were not cataloged members of abell clusters ( abell , corwin , and olowin 1989 , hereafter aco ) were selected.mpc from the center of this cluster , and its recessional velocity does not rule out its being a cluster member ( z=0.022 vs. z=0.019 for the cluster ) . as will be shown later , 3c66b resides in its own separate region of x - ray emission and its projected distance from the core of aco 347 will not greatly contaminate its @xmath1 value . ] the resulting 25 radio galaxies are listed in table 1 along with their fanaroff - riley classification from owen & laing ( 1989 ) , redshift , and optical position in j2000 coordinates . of the sample , 16 galaxies correspond to fr i sources and 6 correspond to fr ii sources . the remaining three - da240 , 3c310 , and ngc6251 - are transitional fr i / ii sources , with the first two being classified as fat doubles and the third a twin jet . based on the conclusions of owen & white ( 1991 ) , which noted that the fr i / fr ii break depends on optical as well as radio luminosity , 3c310 and ngc6251 are grouped with the fr i sources and da240 with the fr ii sources . the 1.4ghz radio powers are listed in table 5 . two - degree square fields centered on each radio source were taken from the rass . the rass was performed between august 1990 and february 1991 , covering an effective energy range of 0.1 to 2.4 kev with a resolution of @xmath9 at 1 kev . exposure times vary for different locations on the sky as a result of rosat s elliptical orbit which provided longer exposures at the ecliptic poles , but in general are on the order of hundreds of seconds . for a more thorough description of the rass , see voges ( 1993 ) . the images correspond to the energy range 0.5 - 2.0 kev . in addition , a search of the _ rosat _ archives found eleven of the sources ( 3c31 , 3c66b , 3c296 , 3c305 , ngc6109 , ngc6251 , 2152 - 69 , 3c442a , 3c449 , 3c88 , and pictor a ) were either the target of or were within the field of publicly - available ( including some sources which lay outside the central ring of the pspc - b rib structure ) pointed pspc - b observations . for both rass and pointed images , photometry was performed to determine x - ray luminosities and characterize the emission as extended or unresolved . photometry of the rass images was performed with the task iring found within the nrao s astronomical image processing system ( aips ) . iring sums counts within annuli about the given central position of the source , which was determined by a best - fit gaussian to the smoothed x - ray image . the chosen aperture corresponded to a linear radius of 250 @xmath10kpc , and sixteen equally - spaced annuli were used within this aperture . the background was estimated using border pixels around the image , rejecting values greater than @xmath11 above the mean . in converting net count rates to luminosities , energy conversion factors ( ecfs ) which assume a 1 kev raymond - smith plasma ( raymond & smith 1977 ) of 0.3 solar metallicity were used . corrections for photoelectric absorption based on hydrogen column densities from stark _ ( 1992 ) were also applied . photometry of available pointed observations follows the same procedure , with two notable differences . due to the position - dependent sensitivity of the pspc - b , background counts were determined using three equivalent - sized apertures located in similar regions of the detector . care was taken to avoid including rib structures in these background estimates . the ecfs include the same assumptions as those used for rass data , with the exception of the different response function for the pspc - b detector . in all cases , the luminosities calculated from these pointed observations were consistent with those derived from the rass . to differentiate between point and extended sources , a point source profile was created for the rass images using point sources located within the fields of the 25 sources . this profile is depicted in figure 1 , and is well - described by a gaussian , but with a long tail . comparison of the profile of each source with the generated point response function was performed to differentiate between point and extended sources . two x - ray sources lacked adequate s / n for characterization ( ngc1692 and 3c88 ) , although the latter of these was also present in pointed observations so the differentiation was achieved . for pointed observations , the task pcrpsf within the ftools software package was used to generate theoretical point response curves for comparison with radial profiles . this task accounts for the variation in the point response function of the pspc - b with off - axis angle and energy . optical data from the digitized sky survey ( dss ) were used to determine a quantitative measure of the degree of clustering . abell characterized clusters by simply counting the number of galaxies within a magnitude range @xmath12 to @xmath13 , where @xmath12 is the apparent magnitude of the third - brightest galaxy . these counts were performed after correcting for the background and corresponded to galaxies within a radius defined by a set angular distance divided by the assumed redshift of the cluster ( a linear radius of about @xmath14 for nearby clusters ) . using this procedure he established richness classes of clusters based on their number of members . a more statistical measure of clustering is provided by the spatial two - point correlation function , which measures the probability in excess of random of finding two objects in a given volume of space . it is frequently characterized as a coefficient multiplied by a power - law in the separation . this power - law holds over many scales , meaning the coefficient represents a measure of the degree of clustering ( and is frequently referred to as the amplitude of the correlation function ) . andersen & owen ( 1994 , ao94 ) and yee & lopez - cruz ( 1999 , ylc99 ) demonstrated that there is a correlation between this coefficient and abell s richness class . hence , we rely on the spatial two - point correlation coefficient as our means to assess clustering . a brief background is provided ( for a more thorough review , see peebles 1980 ) , followed by a more specific procedure which parallels that presented by ao94 . given the presence of one object , the two - point correlation function , @xmath15 , is defined in terms of the probability of finding another object within a volume element @xmath16 : @xmath17 where @xmath18 is the mean volume density of objects and the volume is described by a radius vector , @xmath19 . integrating this expression simply gives the number of galaxies counted within the volume . note that for @xmath20 the distribution is random and the expected number of objects is simply the mean density times the volume . positive values of @xmath15 therefore denote clustering in excess of random . groth and peebles ( 1977 ) determined that the distribution of galaxies was well - fit by a power law such that @xmath21 where @xmath22 . @xmath1 is therefore called the spatial two - point correlation coefficient and represents a measure of the degree of clustering of a system . groth and peebles found that , on average , galaxies exhibit clustering on scales up to @xmath23 with @xmath24 @xmath25 . lacking three - dimensional data on the distribution of galaxies , @xmath1 can be estimated from the angular distribution . analagous to the spatial two - point correlation function is the angular two - point correlation function , @xmath26 defined such that @xmath27 substituting this into the angular equivalent of equation 1 and integrating yields : @xmath28 where @xmath29 is the mean surface density of galaxies and @xmath30 is the counted number of galaxies brighter than some limiting magnitude , @xmath31 . assuming little evolution of the correlation function and the galaxy luminosity function , the angular correlation coefficient is related to the spatial one via : @xmath32 with @xmath33 representing the luminosity distance ; @xmath34 the galaxy luminosity function ( lf ) ; and @xmath35 is a constant resulting from the integration ( equal to 3.78 for @xmath36 ) . the above discussion shows that there are several parameters which must be specified in order to determine values of @xmath1 for the radio galaxies in our sample . the primary parameters include a lf , a limiting magnitude , and an angular counting radius . further issues include the choice of a value for @xmath37 and determination of the mean surface density of galaxies , @xmath29 . we have relied upon signal - to - noise ( s / n ) criteria in our determination of these parameters and hence @xmath1 . an abell lf ( abell 1962 ; the lf is approximated as two power laws meeting at @xmath38 ) is assumed and we count down to the corresponding value of @xmath39 , corrected for extinction . our angular counting radius varies slightly , but is typically around @xmath40 . a summary of our chosen parameters is included in table 2 , and a detailed description of the procedure may be found in ao94 . galaxies were selected by eye from the dss and we performed our own photometry . to eliminate mis - classifications of stars as galaxies we performed gaussian fits to the objects and rejected objects with stellar profiles . several other studies have used @xmath1 to quantify clustering . of these , prestage & peacock ( 1988 ; pp88 ) also investigated a few of the radio sources in this sample . more recently , ylc99 calculated @xmath1 values for abell clusters and argued for their use as a more robust measure of cluster richness . table 3 presents the derived @xmath1 values by richness class for each of these two studies . for comparison , the results of ao94 , who used essentially the same procedure as we have in this paper , are also presented . as there are clear discrepancies across these studies , some discussion is warranted . the three studies may be normalized by applying scaling relations based on their chosen values of @xmath41 and lf . for example , as @xmath1 @xmath42 @xmath43 , @xmath1 values computed with @xmath41 = 50 km / sec / mpc ( pp88 , ylc99 ) will be about a factor of two greater than those computed with @xmath41 = 75 km / sec / mpc ( ao94 ) . the remaining large - scale differences can be explained by noting the differences in assumed lfs . since @xmath1 is inversely proportional to the lf , its shape and normalization have dramatic effects on the resulting values . the size of this effect depends on the difference in the lfs at the limiting magnitude used for galaxy counting . for example , the lf and associated parameters used in ylc99 result in higher values of @xmath1 by an additional factor of two over those in ao94 . applying this net factor of four to the ao94 values brings them into line with those of ylc99 . as alluded to in section 2 , a number of additional parameters lead to smaller discrepancies in calculated @xmath1 values . a discussion of these and their relative effects may be found in ylc99 . as the radio galaxies analyzed in this study are not members of rich clusters , the optimal choices of limiting magnitude and counting radius may not be the same as those for evaluating @xmath1 for abell clusters . the basic arguments are that by counting to too faint a magnitude or too large an angular extent , the background counts will begin to dominate over the cluster counts and thereby reduce the s / n of the result . to investigate these parameters we created a computer model . the ` clusters ' were represented by king profiles ( king 1962 ; @xmath44 models ) of galaxy density , with luminosities obeying the schechter lf ( schechter 1976 ) used in ylc99 . models with different core radii and richnesses were run to explore a range of cluster redshifts and background counts . the rich clusters were modeled by a king function with core radius equal to @xmath45 and @xmath44 equal to one , values suggested in adami _ the environments around the radio galaxies were modeled with a smaller core radius of @xmath46 , and the richness was varied to find a model which approximately matched the representative @xmath1 values derived for the actual radio galaxies . evaluation of the best choice of limiting magnitude and counting radius was made by the combination which produced the highest s / n for that model . with core radius and richness fixed , isoclines of constant s / n for varying limiting magnitude and counting radius result . figure 2 depicts the results for a both a model radio galaxy environment and a model aco cluster of richness class 1 . contours in the figure depict isoclines of constant s / n from 50@xmath47 to 100@xmath47 of the peak s / n , in increments of 10@xmath47 . the model results clearly show that for richer environments , the s / n improves by counting out to greater radii and fainter limiting magnitudes . the figure also shows that the s / n rises fairly rapidly and reaches a broad plateau . this means that a reasonably wide range of counting radii and limiting magnitudes produce s / n which are better than 90@xmath47 of the peak value , and confirms the results of ylc99 which demonstrate @xmath1 is reasonably insensitive to these parameters . the optimal values of counting radius and limiting magnitude are presented in table 4 for average richness class 0 , 1 , and 2 clusters and the average radio galaxies of this study . these models confirm that the procedure of ao94 applied in this study is a good choice for the poorer environments of the sample radio galaxies . compared to abell clusters , higher s / n for the radio galaxies @xmath1 values is obtained by counting to brighter magnitudes and smaller counting radii . while the ao94 parameters are still not quite optimal for the radio galaxy environments in this sample , they usually result in only a small reduction in s / n of 10@xmath47 or less . the ao94 parameters lead to greater reduction in s / n as one moves toward richer environments . for the richest radio galaxy environment in our sample ( ngc1692 ) , the reduction amounts to about 30@xmath47 and for an abell cluster of richness class 2 it is on the order of 50@xmath47 . for comparison , the parameters adopted by ylc99 would reduce the s / n in the @xmath1 for a typical radio galaxy in this sample by around 30@xmath47 . consequently , we argue that the use of the ao94 parameters is the better choice for this study . the parameters of ylc99 are , however , a good choice for the richer environments of abell clusters . while our model results suggest that ylc99 would improve the s / n in their estimates by using a slightly brighter limiting magnitude and larger counting radius , the net improvement is less than 10@xmath47 . furthermore , our models confirm that for the fixed counting radius of about 700 @xmath10kpc used by ylc99 , they achieve the best results by counting down to their value of @xmath39 + 2 . most of the radio galaxies were associated with significant detections in the x - ray . of the 18 fr i sources , only 3 were not detected at the @xmath2 confidence level ( 3c76.1 , 3c293 , and 3c386 ) . five of the 7 fr ii sources were significant detections , with only da240 and 3c98 not being detected . table 5 lists the x - ray luminosities for each field , derived using a 250 @xmath10kpc radius aperature . the x - ray luminosities for all of the detected sources lie at the high end or above those expected for normal e / s0 galaxies ( fabbiano 1989 ) . the reported x - ray luminosities and associated s / n for several of the detected sources appear low in table 5 as a result of their being fairly weak , compact sources whose signal is overwhelmed by the background in the 250 @xmath10kpc aperature . the x - ray extents of the sources are clearly essential to understand their environment . consequently , radial profiles for the significant x - ray detections were compared with the point response function determined for the rass . figure 3 illustrates rass profiles for 3c31 , 3c66b , ngc6251 , and pictor a. the first two are clearly extended sources while pictor a is unresolved . ngc6251 looks like a point source at small radii , but shows evidence for weak , extended emission at larger radii . this is consistent with the findings of birkinshaw & worrall ( 1993 ) , who suggest ngc6251 is best represented by a point source with a faint x - ray halo . in all , eleven of the detected sources were characterized as extended with three others being unresolved ( 3c78 , 2152 - 69 , and pictor a ) . the remainder of the detected sources lacked adequate signal for characterization ( 3c278 , 3c296 , 3c305 ; and 0131 - 36 , 3c88 , 1637 - 77 ) . these results were confirmed with profiles derived from pointed observations , which also showed 3c88 to be unresolved . a simple estimation of the radial extents was performed by noting the radius at which the source signal was less than @xmath11 above the background . to eliminate the possibility of chance fluctuations leading to incorrect characterization of source size , this criterion had to be met in three successive annuli . if this occurred , the radial extent was assigned as the linear distance corresponding to the last annulus significantly above the background . as it is inaccurate to assume that the x - ray distributions are symmetric and centered on the radio source , and our linear resolution is dependent on the redshift of the source , these extents should be considered as rough approximations . to illustrate this and gain a better understanding of the nature of the x - ray emission , overlays of x - ray contours onto dss optical images are provided in figure 4 . in addition to the eleven extended sources , an overlay for 2152 - 69 ( an unresolved source with high @xmath1 ) is included . figure 4b shows that 3c66b is associated with its own region of extended x - ray emission , with the x - ray emission associated with the core of aco 347 lying well to the south of the image . for this reason , we justify its inclusion in our sample of non - cluster radio galaxies . 3c353 was removed from the list of extended sources as it is not within an extended x - ray source but lies next to one ( see figure 4k ; also siebert _ et al . consequently , averaging over the fluxes contained in successive annuli gives the false appearance of extended x - ray emission about the source . overlays of x - ray contours onto radio images for five of the larger radio sources ( 3c31 , 3c66b , 3c310 , 3c442a , and 3c449 ) are depicted in figure 5 . in each of these sources , the extent of the radio emission is approximately the same as the extent of the x - ray emission . in optical measures of clustering , 11 of 18 ( @xmath48 ) fr i sources exhibit @xmath1 values @xmath49 in excess of that expected for a typical galaxy ( @xmath50 ; groth & peebles 1977 , determined as the average for all galaxies present in the lick survey ) . only 2 of the 7 ( @xmath51 ) fr ii sources were significant at this level . one of these two sources was 3c353 , which as noted previously in the x - ray discussion is the result of a superposition of the radio galaxy with an offset x - ray and optical group . results for three sources ( ic2082 , ic4296 , and 2104 - 25 ) located within aco clusters are also included in table 5 . these sources were included initially in the sample to provide a consistency check on the results . the evidence for clustering is even stronger than suggested by the @xmath1 values . for poor clusters such as those of this study , low values of @xmath1 do not necessarily imply a lack of physical association for galaxies in the vicinity of the radio galaxy . this is because the @xmath1 calculation for a close collection of galaxies may be dominated by the background . the presence of clustering was also investigated with poisson statistics . given our assumption of a uniform distribution of field galaxies @xmath52 , we can easily determine the expected number of field galaxies within each radio galaxy s survey area . this may be compared to the actual number of galaxies counted to test the null hypothesis that there is no cluster present and the galaxy counts represent only chance fluctuations in the field galaxy distribution . performing this simple test , 18 of the 25 galaxies represented clustering in excess of the field galaxy distribution at the @xmath53 significance level or greater . this number breaks down into 15 of 18 fr i galaxies and only 3 of 7 fr ii galaxies ( including 3c353 ) . the counted number of members and expected background counts per field are also listed in table 5 . the primary result of this investigation is that many of these nearby radio galaxies do , in fact , reside in cluster environments . as one would expect based on their not being cataloged members of clusters , these cluster environments are poorer than those prescribed by abell . the data backing this claim are the high incidence of significant x - ray emission associated with the radio galaxies and their significant optical clustering above that expected for the field . to demonstrate this point , table 3 includes average values of x - ray luminosity and @xmath1 for abell clusters of richness class 0 , 1 , and 2 . these data are also included in figure 8 , where it can be seen that the radio galaxies form a natural extension of the classical abell clusters in regards to x - ray luminosity and optical clustering . the difference between the fr i and fr ii sources is striking . all ten of the extended x - ray sources are fr i sources , whereas none of the seven fr ii sources are resolved . furthermore , most of the fr i sources represent significant associations of galaxies above the background ( 15 of 18 ) , including all ten of the extended x - ray sources . conversely , only three of the seven fr ii sources are indicative of significant clustering above an expected field distribution . the significance of such clustering merely strengthens this argument , as only two fr ii sources possess @xmath1 values in excess of that expected for a typical galaxy . this is even including 3c353 , which can be seen to lie at the edge of an optical group of galaxies coincident with x - ray emission . consequently , it is more correct to say that the fr i radio galaxies represent a low richness extension of the abell clusters . they are frequently associated with extended x - ray emission suggestive of an icm , and possess a significant excess of nearby galaxies above that expected for the field . the strong radio - emitting fr ii galaxies appear to reside in environments more consistent with the field . they are either non - detections in the x - ray or possess unresolved x - ray emission . this suggests a more internal mechanism for their x - ray emission , namely agn activity . thus , fr i s are more likely to be associated with some amount of clustering than fr ii s . these results confirm those of several prior studies , including longair & seldner s ( 1979 ) pioneering work which noted that fr i s were more likely to be associated with regions of greater clustering . pp88 strengthened this result and added that fr ii sources avoid high richness environments at low redshift whereas at higher redshifts fr i sources and fr ii sources are found in similar environments . it is also interesting to note the overlap in the x - ray and optical results : of the five x - ray non - detections , only one ( da240 ) represents a marginally significant excess of galaxies above the field . the overlap is exact for the fr i radio galaxies : of the 18 fr i sources , fifteen possess significant x - ray emission and an excess of galaxies above the field , whereas the remaining three possess neither ( 3c76.1 , 3c293 , and 3c386 ) . prior studies have noted that there appears to be no correlation between radio power and richness for abell clusters ( e.g. , zhao _ et al . _ 1989 , ledlow & owen 1995 ) . correlation analyses between radio power at 1.4 ghz ( @xmath54 ) and optical clustering ( @xmath1 ) were performed to test this result for our sample . kendall s @xmath55 for this correlation was determined for the entire sample as well as subsets defined by radio morphology ( see table 6 and figure 6 ) . in performing these tests , censoring of data was used when @xmath1 values were considered to be upper limits . these upper limits corresponded to the @xmath1 value at which any underlying galaxy population would be significantly in excess of the expected field counts at the @xmath53 confidence level . for reference , these upper limits are all roughly equal to the groth & peebles ( 1977 ) average galaxy @xmath1 value of @xmath56 . we can not make any statements about the correlation between radio power and richness for our sample as a whole . however , breaking the sample into different classes provided interesting insight , though caution must be exercised in evaluating the correlations as the sample sizes are small . performing the correlation analysis for only the extended fr i sources indicated a potential correlation between @xmath54 and @xmath1 , at a probability of @xmath57 . this result suggests that the presence of an icm as indicated by extended x - ray emission does have a significant effect on fr i radio sources , in accordance with present theories of such sources . the importance of x - ray gas was further investigated by analyzing potential correlations between @xmath54 and @xmath58 . the results of the kendall s @xmath55 tests ( again with censoring ) are presented in table 6 , and figure 7 graphically depicts the data . as expected from the @xmath59 correlations , little can be said about potential correlations for the entire sample or the fr i sources in particular . correlations between @xmath54 and @xmath58 appear to exist for both the fr ii and extended fr i sources ( at @xmath60 and @xmath61 , respectively ) , though this is most likely the result of selection bias . as our sample is derived from flux - limited all - sky surveys , the radio powers and redshifts are very strongly correlated . since @xmath58 is also redshift dependent , it is not surprising that it and @xmath54 appear correlated . given this caveat , figure 7 shows different trends for the fr ii and extended fr i sources , which is a further suggestion of different emission mechanisms . a qualitative understanding of the relationship between the radio and x - ray for the fr i sources may be derived from figure 5 . in this figure , radio images of five of the extended fr i sources are shown with x - ray contour overlays . in each case , the extents of the radio and x - ray emission are consistent with one another . correlation between @xmath58 and @xmath1 is less straight forward . for the entire sample , the two variables are likely to be correlated ( greater than a @xmath62 chance that a correlation is present ; see table 6 ) . this can be interpreted as richer systems being more likely to have associated x - ray gas . broken down by radio morphology , this correlation appears to hold for the entire fr i portion of the sample ( again a greater than @xmath62 probability ) while there is no significant correlation for the fr ii sources . while one would expect @xmath58 and @xmath1 to be strongly correlated for the extended fr i sources , the results of the kendall s @xmath55 test do not support this . figure 8 plots @xmath1 vs. @xmath58 for each type of source . a thorough study of the environments of a statistical sample of nearby radio galaxies has been presented . by focusing on both x - ray emission and optical clustering , stronger conclusions may be reached than for studies which address only one of these wavelengths . several important conclusions are summarized as follows : * radio galaxies not cataloged to be in rich clusters are frequently associated with significant x - ray emission ( 20 of 25 , or @xmath63 of the total sample ) and galaxy clustering in excess of that expected for the field ( 18 of 25 , or @xmath64 ) . this suggests that powerful radio galaxies are associated with cluster environments , even if these clusters are too poor to be classified as such in well - known catalogs . * of the radio galaxies with associated x - ray emission , many ( 10 of 20 ) are resolved by _ rosat_. this suggests the presence of an icm around these sources , which is consistent with current models of extended radio sources . * though the small sample size limits strong statistical conclusions , fr i sources are more likely to lie in cluster environments . ten of 18 ( @xmath65 ) of the fr i sources exhibit extended x - ray emission vs. none of the fr ii sources . this same trend is reflected in the @xmath1 values , with 11 of the 18 ( @xmath66 ) fr i sources exhibiting @xmath1 s significantly in excess of that expected for a typical galaxy . all of the fr i sources with significant @xmath1 are also extended x - ray sources . * fr ii sources are more likely associated with compact x - ray sources , and hence are consistent with agn . five of the seven fr ii sources are significant x - ray detections yet none are resolved . these results , coupled with the optical results showing fr ii galaxies to generally lie in field - like environments , suggest fr ii galaxies are governed more by influences within the galaxies and less by the presence of a cluster environment . * while fr ii sources may lie in regions of modest clustering ( 2 of 7 have significant values of @xmath1 ) , they do not appear to be centrally located in such clusters . this is evidenced by the fact that none of the fr ii sources lie within regions of extended x - ray emission , and at least one example ( 3c353 ) can be seen to lie on the outskirts of such a region . hence , we have demonstrated that for this nearby sample fr i radio galaxies are associated with cluster environments , even if these environments are too poor to be classified as clusters by classical measures . moreover , the environments of these nearby fr i and fr ii sources differ , with fr i sources more likely to be associated with x - ray gas and regions of higher optical clustering . thus , we can conclude that a more significant icm is present for fr i radio sources . other studies aimed at cluster radio galaxies have found that higher redshift fr i and fr ii galaxies reside in similar environments , and suggested that this similarity appears to break down at lower redshifts . this study has strengthened this latter claim , thereby adding importance to the issue of evolution of fr i and fr ii environments . why are fr i and fr ii galaxies found in similar cluster environments at high redshift while the fr ii galaxies appear to avoid cluster environments at low redshift ? that is , what evolutionary factors cause this change in the environment of fr ii radio galaxies ? such questions need to be the topic of future studies . even within this study s sample of nearby fr i galaxies there are additional unanswered questions . if a cluster environment seems to be so important for fr i galaxies , what is the explanation for the three fr i sources of the sample without x - ray emission or significant clustering ? are their radio properties and morphologies somehow different than the other fr i galaxies of the study ? answering these questions requires more evidence of the conclusions reported in this paper . the measures of clustering we have used are statistical in nature . more direct evidence of clustering can be obtained through collection of spectra of the galaxies in these environments . from this information , velocity dispersions and subsequently mass estimates can be made . we have undertaken such a project , and thereby hope to better determine whether these systems do in fact represent poor clusters of galaxies . this information will demonstrate more conclusively if the clustering environments of nearby fr i and fr ii galaxies differ , and should further elucidate the differences in the environments of the sample fr i galaxies for which no evidence of clustering was found . in addition to the radio galaxies of this study , we hope to increase the size of our sample of nearby radio galaxies in order to strengthen our statistical conclusions . the authors would like to thank mpe and its director , j. trmper , for making the rass data available for this study . clay smith provided invaluable assistance by writing routines to assist in the reduction of the x - ray data . we also thank victor andersen , mark bliton , percy gmez , chris loken , glenn morrison , and elizabeth rizza for conversations regarding this research , and heinz andernach and an anonymous referee for valuable comments on the manuscript . we acknowledge nasa grant nagw-3152 to f.n.o . and j.o.b . and nsf grants ast-9317596 and ast-98 to j.o.b . n.a.m . would like to thank the nrao pre - doctoral program for support of this research . shane , c.d . , and wirtanen , c.a . 1967 , _ publ . _ , 22 , pt 1 siebert , j. , brinkmann , w. , morganti , r. , tadhunter , c.n . , danziger , i.j . , fosbury , r.a.e . , and di serego alighieri , s. 1996 , , 279 , 1331 c c c c c + & & & & & & & & 3c31 & fr i & 0.017 & 01 07 25 & 32 24 45 + 3c66b & fr i & 0.022 & 02 23 11 & 42 59 32 + 3c76.1 & fr i & 0.0324 & 03 03 15 & 16 26 19 + 3c78 & fr i & 0.029 & 03 08 26 & 04 06 39 + ngc1692 & fr i & 0.035 & 04 55 24 & -20 34 13 + 3c278 & fr i & 0.015 & 12 54 37 & -12 33 23 + ngc5090 & fr i & 0.011 & 13 21 09 & -43 42 39 + 3c293 & fr i & 0.045 & 13 49 41 & 31 02 31 + 3c296 & fr i & 0.024 & 14 16 53 & 10 48 27 + 3c305 & fr i & 0.0417 & 14 49 21 & 63 16 14 + 3c310 & fr i / ii & 0.054 & 15 04 57 & 26 00 59 + ngc6109 & fr i & 0.0296 & 16 17 41 & 35 00 16 + ngc6251 & fr i / ii & 0.024 & 16 32 32 & 82 32 16 + 3c386 & fr i & 0.0177 & 18 38 26 & 17 11 49 + 2152 - 69 & fr i & 0.027 & 21 57 07 & -69 41 33 + 3c442a & fr i & 0.0171 & 22 14 47 & 13 50 27 + 3c449 & fr i & 0.017 & 22 31 21 & 39 21 30 + ngc7385 & fr i & 0.0243 & 22 49 55 & 11 36 33 + 0131 - 36 & fr ii & 0.030 & 01 33 58 & -36 29 36 + 3c88 & fr ii & 0.030 & 03 27 55 & 02 33 44 + 3c98 & fr ii & 0.031 & 03 58 55 & 10 25 47 + pictor a & fr ii & 0.035 & 05 19 49 & -45 46 46 + da240 & fr i / ii & 0.036 & 07 48 37 & 55 48 59 + 1637 - 77 & fr ii & 0.043 & 16 44 16 & -77 15 35 + 3c353 & fr ii & 0.030 & 17 20 32 & -00 58 46 + \(1 ) source name ; ( 2 ) fanaroff - riley classification from owen & laing ( 1989 ) ; ( 3 ) redshift ; ( 4 ) right ascension of optical center of galaxy , in j2000 coordinates ; ( 5 ) declination of optical center of galaxy , in j2000 coordinates . c c c + & & & & luminosity function & @xmath67 & abell 1962 + & @xmath68 @xmath69 & peebles 1971 + & @xmath70 & + background galaxy counts & @xmath71 & + & @xmath72 @xmath73 & peebles 1973 + limiting magnitude & @xmath31 _ corresponds to @xmath74 at the _ & andersen & owen 1994 + & _ distance of the source _ & + angular counting radius & @xmath75 _ determined such that a cluster with _ & andersen & owen 1994 + & @xmath76 _ be detected with s / n=3 _ & + power of correlation function & @xmath36 & groth & peebles 1977 + c c c c c + & & & & & & & & & & & & 0 & 162 @xmath77 41 & 146 @xmath77 15 & 293 @xmath77 98 & 8.2 + 1 & 272 @xmath77 52 & 350 @xmath77 37 & 488 @xmath77 98 & 18.0 + 2 & 515 @xmath77 74 & 500 @xmath77 61 & 683 @xmath77 98 & 39.4 + \(1 ) abell s richness class : where 0 indicates 30 - 49 members , 1 indicates 50 - 79 members , and 2 indicates 80 - 109 members ; ( 2 ) average @xmath1 per richness class ( in units of @xmath25 ) , from andersen & owen ( 1994 ) who use a similar procedure to that adopted in this study ; ( 3 ) average @xmath1 as presented by prestage & peacock ( 1988 , 1989 ) , in units of @xmath25 ; ( 4 ) suggested @xmath1 per richness class ( in units of @xmath25 ) , from yee & lopez - cruz ( 1999 ) . note that these represent suggested values of @xmath1 by richness class , based on a reasonable linear fit to their results . their median values of @xmath1 by richness class are roughly 465 , 465 , and 830 ; ( 5 ) x - ray luminosity , based on the fit of voges _ et al . _ this is derived from the rass using a 500kpc aperature , and the values presented here assume 40 members for an r=0 cluster , 65 for r=1 , and 105 for r=2 ( i.e. , the midpoints for each richness as defined by abell ) . c c c c c c + & & & & & & & & & & rg & 73 & 50 & 0.03 & @xmath78 & 0.9 + aco 0 & 289 & 100 & 0.07 & @xmath78 & 1.3 + aco 1 & 487 & 100 & 0.07 & @xmath79 & 1.9 + aco 2 & 694 & 100 & 0.07 & @xmath80 & 2.1 + \(1 ) model representation , where ` rg ' represents a typical radio galaxy from this study and ` aco ' represent abell clusters of richness class 0 , 1 , and 2 ; ( 2 ) @xmath1 value , corresponding to parameters of yee & lopez - cruz but with @xmath4 km / sec / mpc ; ( 3 ) core radius , in kpc , used in the king profile of the model ; ( 4 ) redshift of the simulated cluster ; ( 5 ) limiting magnitude to count down to in order to maximize s / n for each cluster model ; ( 6 ) linear counting radius , in mpc , in order to maximize sn for each cluster model . c c c c c c c c + & & & & & & & & & & & & & & & & & all fr i and fr ii & 25 & 0.280 & 0.700 & -0.067 & 0.193 & 0.621 & 0.982 + fr i & 18 & 0.366 & 0.722 & 0.235 & 0.519 & 0.797 & 0.985 + fr ii & 7 & 1.048 & 0.964 & -0.095 & 0.133 & -0.095 & 0.149 + fr i , extended & 10 & 1.244 & 0.988 & 1.200 & 0.984 & 0.444 & 0.631 + \(1 ) types of sources included in test ; ( 2 ) number of sources included in test ; ( 3 ) kendall s @xmath55 for correlation of radio power at 1.4 ghz with rass - derived x - ray luminosity ; ( 4 ) corresponding probability that radio power and x - ray luminosity are correlated ; ( 5 ) kendall s @xmath55 for correlation of radio power at 1.4 ghz with @xmath1 ; ( 6 ) corresponding probability that radio power and @xmath1 are correlated ; ( 7 ) kendall s @xmath55 for correlation of rass - derived x - ray luminosity with @xmath1 ; ( 8) corresponding probability that x - ray luminosity and @xmath1 are correlated .
investigations of the cluster environment of radio sources have not shown a correlation between radio power and degree of clustering . however , it has been demonstrated that extended x - ray luminosity and galaxy clustering do exhibit a positive correlation . this study investigates a complete sample of 25 nearby ( @xmath0 ) radio galaxies which are not cataloged members of abell clusters . the environment of these radio galaxies is studied in both the x - ray and the optical by means of the _ rosat all - sky survey _ ( rass ) , _ rosat _ pointed observations , and the palomar optical digitized sky survey ( dss ) . x - ray luminosities and extents are determined from the rass , and the dss is used to quantify the degree of clustering via the spatial two - point correlation coefficient , @xmath1 . of the 25 sources , 20 are @xmath2 detections in the x - ray and 11 possessed @xmath1 s significantly in excess of that expected for an isolated galaxy . adding the criterion that the x - ray emission be resolved , 10 of the radio galaxies do appear to reside in poor clusters with extended x - ray emission suggestive of the presence of an intracluster medium . eight of these galaxies also possess high spatial correlation coefficients . taken together , these data suggest that the radio galaxies reside in a low richness extension of the abell clusters . the unresolved x - ray emission from the other galaxies is most likely associated with agn phenomena . furthermore , although the sample size is small , it appears that the environments of fr i and fr ii sources differ . fr i s tend to be more frequently associated with extended x - ray emission ( 10 of 18 ) , whereas fr ii s are typically point sources or non - detections in the x - ray ( none of the 7 sources exhibit extended x - ray emission ) . 0.5truein
in models of open inflation , which lead to a density parameter @xmath4 , the `` horizon '' and `` flatness '' problems are solved by two very different mechanisms . although open inflation can be realized with a single scalar , realistic models look more natural when the task of solving each one of these two problems is entrusted to a different scalar field . nevertheless , models with two fields introduce a host of new effects which should be carefully investigated . in particular , as we shall see in this paper , most of the two - field models that have been recently proposed do not give rise to an infinite open universe but to a large inflating island of finite size : a quasi - open universe . the picture of open inflation is the following . the universe starts in a de sitter phase driven by the potential energy of a scalar field @xmath5 which is trapped in a false vaccuum . this false vacuum decays through quantum tunneling , and spherical bubbles of true vacuum nucleate in the smooth de sitter background . after nucleation , the bubbles expand with constant acceleration , following a `` trajectory '' which is invariant under lorentz transformations @xmath6 , see refs . @xcite and fig . since this is also the symmetry of an open friedmann - robertson - walker ( frw ) universe , the @xmath7 surfaces in the interior of the bubble can be identified with the @xmath8 sections of an open universe @xcite . in this way the symmetry of the bubble takes care of the `` homogeneity '' problem . a second period of ` slow - roll ' inflation inside the bubble , lasting for approximately 60 @xmath9-foldings , would solve the `` flatness '' problem @xcite . in open ( and in quasi - open ) inflation , the dynamics of bubble nucleation and subsequent expansion turns out to be very important in determining the spectrum of gravity waves and density perturbations . the reason is that , unlike the case of stardard inflation , the amount of slow - roll inflation is minimal and the `` initial conditions '' right after the bubble nucleates are not washed out completely . thus , for instance , quantum fluctuations of the slow - roll field generated outside the bubble can penetrate to the interior @xcite , causing perturbations whose wavelength is larger than the curvature scale . these are the so - called supercurvature modes @xcite . also , the scattering of tensor modes off the bubble wall determines the spectrum of very long wavelength gravitational waves @xcite . in the limit of a weakly gravitating wall , this effect can be alternatively described as a fluctuation of the bubble wall itself , which induces supercurvature anisotropies inside the bubble @xcite . in principle , tunneling and slow - roll can be done by the same scalar field @xcite , but this requires a very special form of the inflaton potential @xmath10 . denoting by @xmath11 the hubble rate during inflation , a sharp barrier where @xmath12 is necessary the phase transition can proceed via the hawking - moss instanton @xcite . however , this channel represents tunneling to the top of the barrier of a region of size @xmath13 , and does not lead to an open universe . ] for bubble nucleation @xcite . but this barrier must be right next to a flatter region where @xmath14 , which is needed to make slow - roll possible . moreover , the duration of slow - roll , which depends on the length of the plateau in the inflaton potential , has to be fine - tuned to some extent , because a few @xmath9-foldings more or less can make the difference between an almost flat universe and an almost empty one . as mentioned above , models with two fields were introduced in order to overcome these difficulties ; one doing the tunneling and the other doing the slow - roll @xcite . in this way , the coexistence of two different mass scales seems more natural . also , it was argued that in some models the value of the slow - roll field after bubble nucleation can be different in each nucleated bubble , and hence the duration of open inflation would be different in each one . as a result , for a given temperature of the cmb , one would obtain a different value of the density parameter in each universe , and there would always be some open universes with a density parameter in the interesting range @xcite . = 5.5 cm + the purpose of this paper is to show that in models of this sort , with variable @xmath15 , the picture is actually more complicated . indeed , instead of an infinite open universe inside of each bubble , what we find is an infinite number of inflating islands of finite size inside each bubble . quasi - open universes are not entirely new . the simplest two - field model of open inflation , where the tunneling field @xmath5 and the slow - roll field @xmath16 are decoupled , is actually a quasi - open one , as emphasized by the authors of ref . quasi - openness is in principle not a desirable feature , since to a typical observer , the universe looks anisotropic @xcite . in the simple `` decoupled '' model , this `` classical '' anisotropy is large and , combined with the effect of quantum `` supercurvature '' fluctuations mentioned above @xcite , it basically rules out the model @xcite . to circumvent this problem , linde and mezhlumian introduced a class of two - field models where the slow - roll field is coupled to the tunneling field . as we shall see , these models are also quasi - open . this does not mean that they are not good cosmological models . if the co - moving size of the inflating islands is sufficiently large , then the resulting classical anisotropy may be unobservable . even so , the fact that these islands are finite leads to a dramatically different picture of the large scale structure of the universe in open models . the plan of the paper is the following . in section ii we briefly review the `` supernatural '' model of inflation , introduced in ref . @xcite . in this model , an approximate @xmath3 symmetry is broken by a tunneling field in a first order phase transition , and slow - roll inflation inside the nucleated bubbles is driven by the pseudo - goldstone boson . in section iii , we study the quantum fluctuations of the pseudo - goldstone in the bubble background . section iv is the core of the paper , where we argue that after tunneling , we do not obtain an infinite open universe , but an infinite ensemble of quasi - open universes inside a single bubble . section v is devoted to more general models , like the `` coupled '' and `` uncoupled '' two - field models . in section vi we briefly describe the observational implications of our results and in section vii we summarize our conclusions . an attractive scenario for open inflation is the model of a complex scalar field with a slightly tilted mexican hat potential , see fig . [ fig2 ] , where the radial component of the field does the tunneling and the pseudo - goldstone does the slow - roll . this model was called `` supernatural '' inflation in ref . @xcite , because the hierarchy between tunneling and slow - roll mass scales is protected by the approximate global @xmath3 symmetry . the action is given by @xmath17,\ ] ] where we use the metric signature @xmath18 . expanding the field in the form @xmath19 , where @xmath20 is the expectation value of @xmath5 in the broken phase , we consider a potential of the form @xmath21 where @xmath22 is @xmath3 invariant and @xmath23 is a small perturbation that breaks this invariance . it is asumed that @xmath10 has a local minimum at @xmath24 which makes the symmetric phase metastable . we shall consider a ` tilt ' in the potential of the form @xmath25 where @xmath26 is a slowly varying function of @xmath5 which vanishes at @xmath27 . for definiteness we can take @xmath28 . = 5.5 cm + the idea is that @xmath5 tunnels from the symmetric phase @xmath27 to the broken phase , landing at a certain value of @xmath16 away from the minimum of the tilted bottom . once in the broken phase , the potential @xmath23 can not be neglected , and the field @xmath16 slowly rolls down to its minimum , driving a second period of inflation inside the bubble . another attractive feature of this model is that depending on the value of @xmath16 on which we land after tunneling , the number of @xmath9-foldings of inflation will be different . hence it appears that in principle we can get a different value of the density parameter in each nucleated bubble . as we shall see , however , this picture is somewhat oversimplified . we should point out that the supernatural model is not free from certain restrictions . indeed , in order for the pseudo - goldstone to realize inflation as in the simple free field `` chaotic '' scenario , we would need @xmath29 , where @xmath30 is the planck mass . on the other hand , if @xmath22 is a typical quartic potential , the bubble walls would undergo topological inflation @xcite for @xmath31 , and this would spoil the open scenario . topological inflation occurs when the thickness of the walls is larger than the hubble rate at the top of the potential barrier separating two local minima ( degenerate or not ) . this is the same condition under which the coleman - de luccia instantons @xcite cease to exist , and we have a hawking - moss @xcite transition instead . hence the condition @xmath32 also represents the regime where the transition is not of the coleman - de luccia type , as it would be necessary for a successful open universe . as emphasized in ref . @xcite in the case of single - field open inflation , a transition of the hawking - moss type would leave unacceptable anisotropies in the cmb . these constraints can be made less severe by choosing a suitable form for @xmath22 , with higher curvature at the top of the potential between the two minima , or perhaps a special form for @xmath33 . in any case , we shall take @xmath34 in what follows . in this section , we compute the amplitude of quantum fluctuations of the pseudo - goldstone field in the supernatural inflation model . for this , we need to review the formalism for quantizing fields in the background of a bubble . the field equation for @xmath35 is @xmath36 where @xmath37 . in terms of the modulus @xmath5 and phase @xmath38 , we have @xmath39 e^{i \phi / v}=0 . \label{eom}\ ] ] it should be noted that classical solutions with @xmath40 exist only if @xmath41 is an extremum of @xmath33 , so that @xmath42 this is true also for the euclidean solutions ( instantons ) describing the tunneling , so strictly speaking these instantons can only take us from the false vacuum to the extrema of @xmath23 in the broken phase . even if we relax the condition that @xmath16 should be constant , there may not be any instantons which can take us to non extremum values . of course , this nonexistence of the corresponding instanton does not mean that the field can not tunnel to a non - extremum value of @xmath16 , it simply means that tunneling away from the extremum will be somewhat suppressed . this question will be addressed in section iv . since the effect we are studying is not due to gravity , we shall start with the case of a bubble in flat spacetime . including gravity is quite straightforward and will be done below . according to the theory of vacuum decay @xcite , the tunneling rate is dominated by the @xmath43 symmetric solution of the euclideanized equations of motion ( [ eom ] ) with appropriate boundary conditions , which is called the instanton or bounce , and which we shall write as @xmath44 here , we have introduced the euclidean radial coordinate @xmath45 , where @xmath46 are cartesian coordinates in euclidean space . as we move from spatial infinity to the origin , the bounce interpolates between false vacuum @xmath47 and a certain value of the field in the basin of the true vacuum , see fig . [ fig2 ] , @xmath48 . in addition , the bounce has to satisfy the boundary condition @xmath49 . the solution describing the bubble after nucleation is given by the analytic continuation of the instanton to minkowski time @xmath50 through the substitution @xmath51 . then , the bubble solution depends only on the lorentz invariant ` distance ' to the origin @xmath52 , where @xmath53 are the usual minkowski coordinates . it is useful to change to the new coordinates @xmath54 in terms of which the line element reads @xmath55 here @xmath56 is the line element of a 2 + 1 dimensional de sitter space of unit ` radius ' , and in flat space @xmath57 . including gravity , @xmath58 has to satisfy the ( euclideanized ) friedmann equation , as described below . this 2 + 1 dimensional de sitter space can be thought as the hyperboloid swept by the bubble wall during its time evolution ( suitably rescaled ) . in spite of its name , the coordinate @xmath59 is timelike , whereas @xmath60 is a ` radial ' spacelike coordinate . the above coordinates cover only the exterior of the light - cone from the origin . in order to cover the interior , which is where the open universe sits , we use the coordinates @xmath61 in terms of these the metric reads @xmath62 where @xmath63 is the metric on the unit 3-dimensional hyperboloid and ( in flat space ) the scale factor is given by @xmath64 . notice that ( [ metric3 ] ) is the metric of an open frw . when gravity is included , the scale factor @xmath65 will no longer be proportional to the cosmological time @xmath66 but it will be given by the solution of the friedmann equation @xmath67 in the general case , the metrics ( [ metric1 ] ) and ( [ metric3 ] ) are related by the analytic continuation of coordinates and scale factor in the following way @xmath68 these relations can be used to analytically continue solutions from the outside to the inside of the light - cone from the origin . in what follows , we shall assume that tunneling occurs along the real direction for @xmath35 , i.e. @xmath69 in ( [ bounce ] ) , and we shall consider perturbations around the classical solution of the form @xmath70 note that @xmath71 substituting into the action ( [ actionsuper ] ) we obtain the second order action for linearized perturbations @xmath72 + \int\,d^4x \sqrt{-g}[\box \sigma_0-v_0'(\sigma_0)]\varphi_1 + s_1[\varphi_1]+s_2[\varphi_2],\ ] ] where @xmath73 @xmath74 \varphi_2 ^ 2\right ) . \label{s2}\ ] ] here @xmath75 is a small @xmath60-dependent ` squared mass ' due to the potential @xmath23 . in the last equality we have used @xmath33 of the form @xmath76 . when the field @xmath5 is in the broken phase , then @xmath77 is the mass of the pseudo - goldstone . of course , using the unperturbed equations of motion the linear term in ( [ pertaction ] ) drops out . the action ( [ s1 ] ) has been studied in some detail in the past @xcite because it is the same as the one for a one - field model . in particular , it describes the fluctuations of the bubble wall itself . here , we shall concentrate on @xmath78 $ ] , which describes fluctuations in the direction transverse to tunneling . in order to study quantum fluctuations , the field @xmath79 is expanded as a sum over modes times the corresponding creation and anihilation operators @xmath80 the equation of motion satisfied by the modes is @xmath81 \varphi_{plm } = 0.\ ] ] following @xcite , we take the ansatz @xmath82 where @xmath83 are coordinates on the 2 + 1 de sitter space spanned by the motion of the bubble wall . introducing the conformal coordinate @xmath84 defined through the relation @xmath85 , the equation of motion separates into a schrdinger equation @xmath86 f_p = p^2 f_p , \label{schrodinger}\ ] ] where the separation constant @xmath87 plays the role of an energy eigenvalue , and a klein - gordon equation for the modes of a scalar field of mass @xmath88 living in the 2 + 1 dimensional de sitter space @xmath89 here @xmath90 is the covariant dalembertian in this lower dimensional space , and @xmath91 is the four - dimensional ricci scalar , which from the unperturbed einstein s equations can be written as @xmath92 $ ] . the modes @xmath93 should be klein - gordon normalized on a cauchy surface such as @xmath94 . this amounts to klein - gordon normalizing the lower dimensional modes @xmath95 in the 2 + 1 dimensional sense , and then normalizing @xmath96 as in the schrdinger problem @xcite , @xmath97 where the delta function will be discrete or continuous depending on whether we are considering discrete or continuous eigenstates of the schrdinger equation . in flat space , @xmath98 , where @xmath99 is an arbitrary constant which can be conveniently taken to be of order the radius of the bubble at the time of nucleation . therefore , the effective potential in the schrdinger equation @xmath100 \label{effpot}\ ] ] tends to zero at @xmath101 ( center of the bubble ) and to infinity at @xmath102 ( false vacuum ) , see fig . [ fig3 ] . in curved space , it can be shown that @xcite @xmath103 , so @xmath104 vanishes at both ends . therefore , in both cases , the spectrum will be continuous for @xmath105 , and there may be a discrete spectrum for @xmath106 . = 5.5 cm + in a time dependent background , the choice of a vacuum state is always somewhat ambiguous @xcite . here , this ambiguity corresponds to the freedom of choosing the ` positive frequency ' modes @xmath107 on the hyperboloid . in principle , the ambiguity can be resolved dynamically if the initial quantum state before the bubble nucleates is given . the quantum state of a nucleating bubble has been extensively studied both in flat and in curved space @xcite . in our model , @xmath79 is treated as a free field which couples to the bubble via a @xmath5-dependent mass term . for this type of models , it has been shown that if the initial quantum state is de sitter invariant before the bubble nucleates , then right after bubble nucleation the field @xmath79 will be in an @xmath6 symmetric state . this is perhaps not too surprising : the appearance of the bubble breaks the @xmath108 de sitter symmetry by selecting a `` nucleation point '' in spacetime , but otherwise the bubble solution respects an @xmath6 subgroup of isometries . what this means is that the positive frequency modes @xmath95 must be taken as the bunch - davies modes , which guarantee the desired symmetry . these are given by @xmath109^{1/2 } \frac{p^{-l-1/2}_{ip-1/2}(i \sinh\rho)}{\sqrt{i \cosh\rho } } y_{lm}(\theta,\phi).\ ] ] where @xmath110 are the legendre functions and @xmath111 are the usual spherical harmonics . when analytically continued to the inside of the light - cone , through the relations ( [ analytical ] ) , they become @xmath112^{1/2 } \frac{p^{-l-1/2}_{ip-1/2}(\cosh r)}{\sqrt{\sinh r } } y_{lm}.\ ] ] these are proportional to the often used harmonics @xmath113 which are normalized on the hyperboloid @xmath114 @xcite , @xmath115 these analytically continued modes are normalizable on @xmath114 only for @xmath116 . since the mode with @xmath117 has wavelength comparable to the curvature scale , the non - normalizable modes with @xmath106 have been dubbed `` supercurvature '' modes @xcite . writing @xmath118 , the supercurvature modes are given by @xmath119^{1/2 } { p_{\lambda-1/2}^{-l-{1/2}}(\cosh r ) \over \sqrt{\sinh r } } \ , . \label{harmonics}\ ] ] ( we have added a comma after the subindex @xmath26 to indicate that it is the value of @xmath120 rather than @xmath121 ) . we repeat , however that if the corresponding bound state exists in the schrdinger equation ( [ schrodinger ] ) , then these modes are perfectly normalizable on the cauchy surface , and hence they must be included in the expansion of the field operator ( [ expansion ] ) . to begin with , let us neglect the mass term @xmath122 ( [ m2 ] ) which comes from the tilt in the potential , @xmath23 . then , using the equation of motion satisfied by @xmath123 , it is straightforward to show that @xmath124 is a solution of ( [ schrodinger ] ) with eigenvalue @xmath125 , or @xmath126 . moreover , this solution is normalizable and it belongs to the discrete spectrum . the normalization constant @xmath127 is found from ( [ norm ] ) @xmath128 here , we have changed back to the physical coordinate @xmath60 , which measures the physical distance to the center of the bubble . in flat space @xmath129 is actually infinite , but the integral is finite because @xmath123 vanishes exponentially fast outside the bubble . including gravity @xmath129 becomes a finite value , so the integral is also finite . the explicit value of @xmath127 can be calculated numerically for any given model . if @xmath99 is the size of the bubble and @xmath130 is the value of the field at the center of the bubble , see fig . [ fig1 ] , at the time of nucleation , then we can estimate @xmath131 the above estimate is for bubbles such that @xmath99 is small compared with the hubble radius , so that @xmath132 . also , we have substituted @xmath123 by @xmath130 in the integrand , and we have integrated @xmath60 from zero up to @xmath99 . because @xmath133 our estimate is actually a lower bound on @xmath127 , and in some thick wall models the effect can be somewhat higher . if we asume a large energy difference between false and true vacua , then we are in the thick wall regime and @xmath99 is of the order of the thickness of the bubble wall @xmath134 here @xmath135 is the mass of the field in the false vacuum . if , on the contrary , the vacua are sufficiently degenerate , then we are in the thin wall regime , and the expression for @xmath99 can be found e.g. in appendix c and ref . @xcite . the normalized mode has an amplitude @xmath136 physically , what happens is that the field does not simply tunnel to a sharply defined value of @xmath16 , but a distribution of values . taking into account that the phase of our complex scalar field @xmath35 is given by eq . ( [ phi2 ] ) , the @xmath137 dependence in ( [ fluctuation ] ) shows that , after nucleation , different points on the bubble have different values of the angle @xmath16 . when analytically continued to the interior of the light - cone from the origin , through the relations @xmath138 , @xmath60 is replaced with the cosmological time and @xmath139 quickly follows its evolution towards its expectation value @xmath140 . hence , inside the light - cone , the normalized supercurvature mode will take the form @xmath141 which is time independent . in order for the bubble solution to exist , the mass of the field in the false vacuum should be @xmath142 , where @xmath143 is the hubble rate in the false vacuum . in our model , @xmath143 is considered to be much greater than the hubble rate in the true vacuum @xmath144 . ) the two minima become degenerate , @xmath145 . this weakens the constraints and makes the model viable in certain range of parameters . ] therefore , the amplitude of supercurvature perturbations is of order @xmath146 . this exceeds by far the amplitude of the usual `` subcurvature '' fluctuations , which is only of order @xmath144 . the corresponding effect in the cmb places severe constraints on the model @xcite . we shall come back to this question in section vi . note that for @xmath126 the amplitude of the homogeneous mode @xmath147 diverges because one of the gamma functions in ( [ harmonics ] ) has vanishing argument . this is related to the fact that , strictly speaking , the @xmath147 mode should have been quantized as a collective coordinate instead of as a harmonic oscillator @xcite . the divergence simply means that all values of @xmath16 are equally probable after nucleation . when we include the effect of @xmath23 , the degeneracy will be broken and the @xmath147 mode will have a finite amplitude . when the tilt @xmath23 in the potential is included , @xmath125 is no longer an eigenvalue of the schrodinger operator ( [ schrodinger ] ) . however , it is clear that for small @xmath122 there will still be a discrete eigenmode whose eigenvalue we can calculate in perturbation theory . denoting by @xmath148 the unperturbed bound state ( [ -1 ] ) , the perturbation to the eigenvalue will be given by @xmath149 again , @xmath1 can be computed numerically for any particular model , but we can estimate it as being of order @xmath150 where @xmath151 , see eq . ( [ m2 ] ) . the above estimate uses the same approximations as the estimate for @xmath127 after eq . ( [ intn ] ) . the normalized @xmath147 mode will now be given by @xmath152\over \sinh r}\ , \ ] ] where , from now on , we shall use the notation @xmath153 to denote the modes with @xmath154 . in this expression , and the ones that will follow , we give only the unperturbed time dependence , without including the first order correction in @xmath1 . the field @xmath139 quickly rolls down from @xmath130 and settles down to its minimum at @xmath155 . it is understood that , after that , the time derivative of the mode will be dominated by the corrections of order @xmath1 , which slowly drive the goldstone modes to the minimum of the tilted potential . the amplitude of the higher-@xmath156 excitations will also suffer corrections of order @xmath1 , but since these amplitudes were already finite , the effect on those is not so dramatic . to leading order , the amplitudes are the ones calculated in the previous subsection . it is interesting to look at the distribution of the field on a hyperboloid @xmath157 . note that the amplitude of the @xmath147 mode near the origin @xmath158 is of order @xmath159 which is a factor @xmath160 larger than the amplitude of the individual @xmath161 modes found in eq . ( [ rmsdeg ] ) . but the amplitude of the @xmath147 mode decays exponentially for @xmath162 , which means that at large distances it will become negligible . however , the quantum state that we have chosen is @xmath6 symmetric , which means that the r.m.s . fluctuation of the field can not depend on @xmath163 . therefore , the loss in amplitude of the @xmath147 mode as we move away from the origin has to be made up for by the joint contribution of the @xmath164 modes , smeared over a suitable length scale . this is analogous to what happens for a massive field in de sitter space , except that here we are considering a spacelike manifold @xmath8 rather than a spacetime . in the appendix b we show that for @xmath162 we have @xmath165 where @xmath166 is a certain cut - off . if we smear over a fixed comoving length @xmath167 , as we move away from the origin we have to include more and more modes in the sum . since the @xmath156-th multipole has wavelength proportional to @xmath168 , we take @xmath169 . with this , we find @xmath170 . notice that the result is rather insensitive to the choice of @xmath167 . as long as @xmath171 , the added contribution of all relevant modes at large @xmath163 is the same as the contribution of the @xmath147 mode near the origin , given by ( [ spread2 ] ) . we also show in appendix a that the two - point correlations on a @xmath157 surface die off with comoving distance @xmath172 as @xmath173 . hence , around the time @xmath174 when the field @xmath123 settles down to its minimum @xmath175 the supercurvature fluctuations of the field are of order @xmath176 here , @xmath177 is the mass of the pseudo - goldstone in the true vacuum . in the last step we have asumed that @xmath178 is a slowly varying function of @xmath5 , and set @xmath179 . note that the fluctuation in @xmath79 can easily reach planckian values . it suffices to take @xmath180 gev and @xmath181 gev . with these values , the field is displaced enough from its minimum that it can drive inflation ( recall that we are asuming @xmath182 ) . tunneling to a large value of the field is usually understood in an `` adiabatic '' sense . the idea is that since the motion of the phase is dictaded by the explicit symmetry breaking potential @xmath23 , it will be much slower than the motion of the radial component of the field dictated by the large u(1 ) symmetric part of the potential . hence , one can estimate the rate for tunneling at any @xmath16 by solving the euclidean equations of motion for @xmath5 while @xmath16 is kept as a frozen parameter . this frozen parameter is then used as the initial value of the slow - roll field on the @xmath183 hypersurface . here , as above , @xmath174 is the time at which @xmath123 reaches its expectation value @xmath20 . however , it is clear that when the spread of the pseudo - goldstone on the @xmath183 hyperboloid is comparable to its range , @xmath184 , the picture that each bubble nucleates with a different value of @xmath79 is not adequate . instead , all of the ` vacuum ' manifold is pretty much sampled _ inside a single bubble_. in this case , inflation does not take place coherently on the @xmath8 hyperboloids . as mentioned in the previous section , the comoving correlation length for quantum fluctuations ( in units of the curvature scale ) is @xmath185 . thus , patches of comoving size @xmath186 where @xmath79 is large and positive would be right next to patches where @xmath79 is large and negative . these patches would be separated by regions where @xmath79 is small and the universe does not inflate . patches with positive and negative values can also be separated by `` walls '' where @xmath79 is close to @xmath187 . in the case @xmath182 these domain walls of the pseudo - goldstone potential would be topologically inflating @xcite . hence , instead of a smooth inflating hyperboloid , we have a patchy mosaic of inflating regions , as depicted in fig . [ fig4 ] . in principle , each patch can give rise to a successful cosmology , but this cosmology will not be an open universe in the traditional sense . at best it will be a ` quasi - open ' one , i.e. one which locally resembles an open frw . = 5.5 cm + when @xmath188 the spread in the distribution of the pseudo - goldstone in a @xmath183 section is small . what this means is that if the nucleated bubble is described by the @xmath6 symmetric quantum state , then most of the surface @xmath183 has a non - inflating value of the field . however , in an infinite hypersurface , there will be a certain density of occasional large fluctuations which will lead to inflating islands of comoving size @xmath185 . each one of these islands will be a quasi - open universe . this is in clear opposition with the conventional adiabatic picture described in the first paragraph . it is interesting to pursue the adiabatic picture for a moment , in order to see in which sense it is adequate or not . to keep the discussion simple , let us consider tunneling to a range of values of @xmath16 for which the linearized expressions ( [ pertaction ] ) are still valid , but sufficiently large that it can be distinguished from tunneling to the bottom @xmath189 . this will be the case if @xmath79 , see eq . ( [ phi2 ] ) , is in the range @xmath190 where @xmath191 was computed in the previous section . since @xmath192 and @xmath79 decouple , we may take an _ approximate _ euclidean solution of the form @xmath193 , where @xmath16 is taken as constant in the adiabatic approximation . substituting this configuration in the euclideanized version of ( [ pertaction ] ) we find , after straightforward algebra , @xmath194= s_e[\sigma_0]+{1\over 2}\int d^4x \sqrt{-g}\,m^2\varphi_2 ^ 2 = s_e[\sigma_0]+{\gamma\pi^2 \over { \cal n}^2 v^2 } \phi^2.\ ] ] if the decay rate is proportional to @xmath195 , the relative probability of having a bubble with a certain value of @xmath79 at nucleation will be @xmath196,\ ] ] where we have used ( [ phi2 ] ) and , perhaps not too surprisingly , found @xmath197 , the same expression obtained in the previous section from considerations of quantum fluctuations in the @xmath6 invariant state . thus , two approaches which in principle are aimed at answering different questions end up giving the same answer . here , we were asking how likely is it for a bubble to nucleate at a large value of @xmath16 , whereas in the previous section we were computing the amplitude of fluctuations inside a given bubble . undoubtedly , both questions are related , since at least locally we can not distinguish a large value of the field induced by the nucleation of the bubble from a fluctuation of the field inside the bubble . however , even though the adiabatic approximation may give the right answer for the probability of tunneling to a large value of @xmath16 , it suggests the wrong picture for bubble nucleation . since @xmath198 is used in the above estimate , one might imagine that an infinite open universe with homogeneous @xmath79 in the range ( [ range ] ) can be created . we shall argue that the probability for this to happen is actually zero . first of all , the formula @xmath199 $ ] is only justified when the euclidean action is evaluated on a solution of the equations of motion . but @xmath198 is only a solution in the case when @xmath23 is neglected . we can try and correct this configuration so that it will be a solution , while still keeping @xmath43 symmetry . in the linear regime , what this means is that we want a solution of eq . ( [ schrodinger ] ) with @xmath125 , so that eq . ( [ laplacian ] ) is satisfied by @xmath200 and we have a homogeneous solution inside the bubble . however , for @xmath201 the lowest eigenvalue is @xmath202 , which means that the solution with @xmath125 is not normalizable . as a result , the corresponding euclidean action is badly divergent . if the action is regularized with a cut - off and if we take the @xmath125 solution to be well behaved at the center of the bubble @xmath203 , then it is easy to show that the action starts growing exponentially , @xmath204 $ ] , as the cut - off @xmath205 in the radial direction @xmath60 becomes larger than the size of the bubble . here @xmath135 is the mass of @xmath35 outside the bubble . hence it is not justified to say that the estimate ( [ probability ] ) gives the nucleation rate for a homogeneous bubble . rather , using a homogeneous solution we would get a divergent action and hence a vanishing probablility . to compare , we can now ask what is the amplitude for tunneling from false vacuum to a spherically symmetric but inhomogeneous configuration with a large value of @xmath79 inside the bubble . this is what we call an `` inflating island '' or quasi - open universe . again , this amplitude will depend on the action of a semiclassical euclidean trajectory . in order to make the metric ( [ metric3 ] ) into one of euclidean signature , we must consider the analytic continuation of the coordinates @xmath206 and @xmath207 . with this we have @xmath208 where @xmath209 is the metric on the three - sphere , and the range of @xmath210 is from @xmath211 to @xmath212 . the semiclassical trajectory we shall consider is simply the analytic continuation of the @xmath147 supercurvature mode where we match the euclidean solution to the lorentzian one . hence it is a complex trajectory which has a small imaginary part ( of order @xmath1 ) in the classically forbidden region . the imaginary part decreases exponentially fast in the lorentzian section on a timescale of order @xmath213 ( [ l=0 ] ) @xmath214 where @xmath215 and @xmath216\over\sin r_e.}\ ] ] note that this solution is not regular at one of the poles of the three - sphere , @xmath217 . this is of course expected , since @xmath218 is an eigenfunction of the laplacian with eigenvalue @xmath1 , which is not in the spectrum . hence , ( [ inhomogeneous ] ) is not a regular `` bounce '' around which we should expand in order to find the false vacuum persistence amplitude @xcite . this is not a problem for us here , because we are not estimating the total decay rate of the false vacuum but a particular transition amplitude . for this , we only need `` half '' of the euclidean solution , which interpolates between false vacuum at @xmath219 , @xmath220 and an inhomogeneous configuration at @xmath221 . at @xmath222 this solution is matched to a lorentzian solution at @xmath94 ( the spacelike surface where the bubble nucleation takes place ) simply by the analytic continuation @xmath223 and @xmath224 [ see eq . ( [ analytical ] ) ] . the solution is then propagated to the interior of the bubble ( the `` open '' universe ) through eq . ( [ analytical ] ) . the transition amplitude @xmath225 is wkb suppressed only in the euclidean regime ; the lorentzian evolution contributing an oscillatory phase . therefore @xmath226 where @xmath227.\ ] ] it is easy to show that since @xmath79 is a solution of the equations of motion , after integration by parts the only contribution to the integral comes from the boundary term at @xmath222 . a straightforward calculation gives @xmath228 where @xmath229 is de constant introduced in ( [ inhomogeneous ] ) . after analytic continuation to the interior of the bubble , we have @xmath230 . hence , the relative probability for nucleating through the inhomogeneous trajectory ( [ inhomogeneous ] ) is again given by @xmath231 . \label{creation}\ ] ] although this is in perfect agreement with ( [ probability ] ) , it is now clear that it does nt mean the whole open universe will have the value @xmath79 , but only a patch of comoving size @xmath232 will have this value . thus , tunneling to a value of the field which is far from the one indicated by the @xmath43 symmetric instanton is perfectly possible , with a somewhat suppressed probability . however , the resulting universe is not an infinite open universe but just a quasi - open one . the arguments used in the previos subsection leading to eq . ( [ creation ] ) are somewhat heuristic . in particular , we have not attempted to justify why the semiclassical trajectories of the form ( [ inhomogeneous ] ) should be the only relevant ones . note , however , that the probability distribution ( [ creation ] ) for nucleating at a high value of the field @xmath79 near @xmath158 , is the same as the gaussian distribution for the amplitude of the @xmath233 mode in the @xmath6 invariant state . in fact , the possibility of nucleating at different values of the field is already accounted for by this quantum state and need not be considered separately . the analysis of refs . @xcite , whose result we described in section iii.a , takes into account all paths in @xmath16-field space , not just the semiclassical one used above . that analysis should be regarded as a more rigorous derivation of the result ( [ creation ] ) . because of the invariance of the quantum state , it is clear that there is nothing particular about the point @xmath158 , and an inflating ` blob ' is equally likely to develop around any point on the @xmath183 hyperboloid . therefore we are led to the picture where in each comoving volume of size comparable to the correlation lenght @xmath0 , the probability distribution for @xmath79 is given by ( [ creation ] ) . clearly , since the volume of the hyperboloid is infinite , inflating islands with all possible values of the field at their center will be realized inside of a single bubble . we may happen to live in one of those patches of comoving size @xmath234 , where the universe _ appears _ to be open . also , in the supernatural model , it is possible to modify the shape of the potential near the false vacuum so that there is a misalignment between the prefered direction for tunneling @xcite and the direction of the minimum of the pseudo - goldstone potential in the broken phase . in this case , we expect that the @xmath183 surfaces inside the bubble will have a mean value of @xmath235 , determined by the most probable escape path ( i.e. , the instanton , which in this case will not land on @xmath189 ) . this value of @xmath236 will determine the number of @xmath9-foldings of inflation and hence the mean value of the density parameter @xmath237 on the hyperboloid . let us call this value @xmath238 . if the tunneling path is not too narrow , there will still be a supercurvature mode which will cause fluctuations in the density parameter on co - moving scales of order @xmath0 , which are of course much larger than the hubble radius . the picture is then that we have an ensemble of large patches with different values of the density parameter . this is an interesting situation which deserves further study . however , since this model involves more parameters , for the remainder of this paper we shall concentrate on the simplest case discussed above , where the preferred tunneling direction and the minimum of the pseudo - goldstone potential are aligned . in this section we shall consider the class of two - field models with a potential of the form @xmath239 models of this type were introduced in ref . @xcite . here @xmath22 is a non - degenerate double well potential , with a false vacuum at @xmath27 and a true vacuum at @xmath240 . when @xmath5 is in the false vacuum , @xmath22 dominates the energy density and we have an initial de sitter phase with expansion rate given by @xmath241 . once a bubble of true vacuum @xmath240 forms , the energy density of the slow - roll field @xmath16 may drive a second period of inflation . as pointed out in refs . @xcite , the simplest two - field model of open inflation , given by ( [ general ] ) with @xmath242 and @xmath243 , is actually a quasi - open one . since there is no coupling between the two fields except for the gravitational one , we shall call this the `` decoupled '' two - field model . in this model , inflation starts chaotically at large values of @xmath244 . in some regions of the universe , the field @xmath5 will be trapped in the false vacuum , while @xmath16 rolls down from large values . if a bubble nucleates at a point where @xmath245 , the value of the slow - roll field will be large enough to drive a short second period of inflation inside the bubble . one problem with this model , is that the slow - roll field moves also outside the bubble , so the synchronization of the @xmath246 and @xmath7 surfaces inside the bubble is not perfect , as pointed out in ref . @xcite . in ref . @xcite , the classical evolution of the slow - roll field from the outside to the inside of the bubble was studied , and it was found that on the hypersurface @xmath247 the high value @xmath248 decays exponentially with the distance to the origin as @xmath249 \label{classical}\ ] ] where @xmath250 . hence , the size of the inflating region in this model is finite . we use the subindex @xmath251 to stress that this result follows from purely classical evolution . the larger @xmath143 , the larger will be the inflating region . the reason is that the cosmological friction term in the equation of motion for @xmath16 is proportional to @xmath143 , so the larger is @xmath143 , the slower the field @xmath16 will roll down the potential outside the bubble , and the better is the synchronization between @xmath246 and @xmath7 surfaces . however , @xmath143 can not be taken to be too large because otherwise the quantum fluctuations generated outside the bubble produce too large an amplitude for the supercurvature mode inside the bubble @xcite . the combination of these two effects severely constrains this model @xcite . in order to construct a truly open model , linde and mezhlumian suggested taking @xmath252 and @xmath253 . we shall call this the `` coupled '' two - field model . in this way , the mass of the slow - roll field vanishes in the false vacuum , and it would appear that the problem of classical evolution outside the bubble is circumvented . however , this is not exactly so , and the whole class of models ( [ general ] ) leads to quasi - open universes . the basic reason is that , as we shall see below , the ( linear ) equation of motion for @xmath16 in the presence of the bubble , does not admit @xmath6 invariant solutions which are regular at the origin , except for the trivial one , @xmath189 . thus , we are back to a situation analogous to the supernatural inflation model . = 5.5 cm + even if the mass of the field in the false vacuum vanishes , one must not expect that @xmath16 will not evolve at all outside the bubble . [ fig5 ] shows the result of a numerical evolution of the field @xmath16 in the coupled model . the figure represents a conformal diagram of a bubble expanding in de sitter space ( for simplicity , the gravitational field of the bubble has been neglected ) . the bubble wall is indicated by the thick timelike hyperbola . as initial conditions , we have taken @xmath246 and @xmath254 . surfaces of constant @xmath16 are indicated by different shadings . even though the field is massless outside the bubble , we find that it does not stay exactly constant there . due to the finite size of the bubble at the time of nucleation , the field @xmath16 feels the presence of the bubble everywhere inside the light - cone from the `` point '' p. as a result , inside the bubble , the hypersurfaces @xmath246 are not perfectly synchronized with the @xmath7 hypersurfaces . thus , we are back to a situation where the inflating region inside the bubble has a finite size , as in the decoupled model ( note that the @xmath246 lines cross the bubble wall trajectory ) . note that this effect is due to the finite size of the bubble . we shall see below that the effect is of order @xmath255 . in fig . [ fig5 ] , the parameters have been chosen so that the effect is very dramatic and the size of the inflating islands is comparable to the curvature scale , but one can choose parameters so that the inflating islands are as large as desired . however , except in the case where gravity is neglected so that @xmath256 , their size is always finite and the large value of the field at the time of nucleation ends up decaying at large distances @xmath163 from the origin . hence , just as in the case of the supernatural model , the infinite @xmath8 surfaces would be almost empty at large distances , if it was nt for the occasional quantum fluctuations which may ignite inflating islands here and there . as in the case of the supernatural model , we expand the field operator @xmath16 in terms of creation and anihilation operators , @xmath257 in the present case , the equation of motion for the modes is given by @xmath258 \varphi_{plm } = 0.\ ] ] here @xmath123 is the `` background '' bubble solution . using the ansatz ( [ ansatz ] ) , @xmath259 we have the following schrdinger equation for @xmath96 , @xmath260 f_p = p^2 f_p . \label{schrodinger2}\ ] ] this equation determines the spectrum of allowed eigenvalues @xmath87 , which correspond to normalizable eigenfunctions @xmath96 . all of these eigenvalues have to be included in the expansion of the field operator ( [ expansion2 ] ) . as before , the harmonics @xmath95 must satisfy equation ( [ laplacian ] ) . in the case of supernatural inflation , we saw that there was a discrete eigenstate with @xmath106 , which actually dominated the r.m.s . fluctuations of the field on a @xmath8 hypersurface . we shall see that a similar situation happens in this case . since the hubble rate inside the bubble @xmath144 is smaller than the hubble rate outside , @xmath143 , we need @xmath261 in order to have slow - roll inflation inside the bubble . this condition suggests taking a perturbative approach . to lowest order , we can neglect the mass term for @xmath16 and the gravitational backreaction of the bubble , so that @xmath16 is just a massless field in de sitter space . in this case , ( [ schrodinger2 ] ) has the well known supercurvature mode with @xmath125 , which corresponds to @xcite @xmath262 the field configurations corresponding to this @xmath121 are @xmath263 . as mentioned before , for @xmath147 the harmonic @xmath264 is not klein - gordon normalizable in the 2 + 1 dimensional sense . this is because , for a massless field in de sitter space , the zero mode corresponding to the translations @xmath265 has to be treated as a collective coordinate and not as an oscillator . when the mass of @xmath16 is included , the mode becomes normalizable . indeed , the bound state will shift to a perturbed eigenvalue @xmath154 which , as before , can be calculated perturbatively : @xmath266 d\eta . \label{perturbation2}\ ] ] the normalized modes corresponding to the discrete eigenvalue take the form @xmath267 , and in particular , for the @xmath147 mode , @xmath268\over \sinh r}[1+o(\gamma)].\label{zero2}\ ] ] the uncertainty of order @xmath1 comes from the fact that we have not evaluated the correction to the `` wave function '' @xmath269 which gives the temporal dependence of the field . this can be done in principle , but it is not really necessary for our purposes . it is clear that the mass term will cause the field to have the temporal dependence corresponding to slow - roll inside the bubble . near the origin @xmath158 the r.m.s . fluctuation of the field @xmath270 will be dominated by the mode ( [ zero2 ] ) , and for @xmath271 it will be given by @xmath272 as mentioned in section iii , because of the @xmath6 invariance of the quantum state , this will also be the r.m.s . fluctuation of the field at any point on the @xmath183 hypersurface ( see appendix b ) . in the case of thin walls , the value of @xmath1 can be calculated explicitly ( see appendix c ) , @xmath273 here @xmath274 is the mass of the slow - roll field in the true vacuum , @xmath275 is the mass in the false vacuum , @xmath143 is the hubble rate in the false vacuum , and @xmath99 is the intrinsic radius of the bubble at the time of nucleation . the origin of the different terms in ( [ gamma ] ) is easy to understand . the first one is independent of the existence of the bubble , and comes from the fact that the slow - roll field has a mass @xmath276 in the false vacuum ( the @xmath277 in the denominator can be understood from simple dimensional considerations ) . the second term is due to the perturbations of the effective potential in the schrdinger equation ( [ schrodinger2 ] ) caused by the bubble solution . in the bubble , the scale factor @xmath58 is of order @xmath99 , so the factor @xmath278 in the integrand of ( [ perturbation2 ] ) will yield the factor @xmath279 in front of the second term of ( [ gamma ] ) . even though the decoupled model ( @xmath242 ) is not a very good candidate to an open cosmological model @xcite , it is instructive to consider it as a first step . in this model , inflation inside the bubble can be initiated because of the large `` classical '' value of @xmath16 at the time of nucleation , but at very large distances from the origin @xmath280 , the classical field dies off , and only the quantum fluctuations remain . let us consider the amplitude of quantum fluctuations . from ( [ gamma ] ) we obtain ( @xmath281 ) @xmath282 accordingly , from ( [ rms ] ) , the r.m.s . quantum fluctuations of the field on the @xmath183 surfaces will be of order @xmath283 and the comoving correlation length will be of order @xmath284 , the same as we found in the classical case ( [ classical ] ) from a completely different approach . in the decoupled model @xmath285 is the same as @xmath286 , and @xmath143 is not too much larger than @xmath144 @xcite . since @xmath287 , we have @xmath288 and quantum fluctuations on that surface would typically not reach `` inflating '' values of order @xmath30 . still , as in the supernatural model , an occasional large quantum fluctuation can initiate inflation on a patch of size @xmath0 . let us reflect upon the meaning of eq . ( [ great ] ) . this r.m.s . amplitude is actually the same as the bunch - davies one for fluctuations of the slow - roll field outside the bubble @xcite . hence , the correct interpretation of the result ( [ great ] ) seems to be the following . inflation inside and near the bubble wall may start because the field is large at the point where the bubble nucleates . however , even after bubble nucleation , the field will continue its random walk outside the bubble , and it may occasionally become large . if the bubble wall hits a patch where the field is large , then this will generate a local inflating patch inside the bubble , and we might inhabit one of those inflating patches . however , this model is not a very good open model , and it would only agree with observations if @xmath237 were very close to one . let us now consider the `` coupled '' model , where the fields are coupled but @xmath289 . in this case @xmath290 clearly , by choosing parameters such that the size of the bubble is much smaller than the hubble rate outside the bubble , or such that the mass of the field in the true vacuum is sufficiently small , @xmath1 can be made as small as desired . hence the size of inflating regions can be made as large as desired . in this case , the field is massless outside the bubble and quantum fluctuations of @xmath16 pile up to arbitrarily large values far from the bubble . however , from eq . ( [ rms ] ) , we find a finite answer for the fluctuations inside the bubble @xmath291 the first interesting thing to note about this result is that it does not depend explicitly on the hubble rate inside or outside the bubble ( the only dependence is through @xmath99 ) . the second observation is that it is very similar to the expression ( [ similar ] ) we had for the supernatural case , so a connection between the physics of both models can be anticipated . the finiteness of ( [ finite ] ) is not surprising , since the slow - roll field is coupled to the bubble , and piling of modes in the vicinity of the bubble is suppressed by the mass term . also , nucleation of bubbles at high values of @xmath16 is suppressed because the degeneracy between true and false vacuum is lower . as we discussed in section iv , the quantum state already encodes the information that tunneling to a large value of the field is suppressed . a difference with the supernatural inflation case is that now the amplitude of the supercurvature modes with @xmath292 is of order @xmath143 rather than @xmath293 , hence the constraints on this model from microwave background anisotropies will be easier to accomodate . our results from the previos sections have important consequences for two - field models of open inflation . first of all , our models are quasi - open , rather than open , which leads to classical anisotropies @xcite . second , we saw that in the supernatural model , the amplitude of supercurvature excitations is quite large . in this section , we shall give order of magnitude estimates for the expected cmb anisotropies from these effects . a detailed investigation of the power spectrum will be presented elsewhere @xcite . quasi - open universes are finite , and hence they look anisotropic to a typical observer . this effect was studied in ref . @xcite for the uncoupled model , and was called a `` classical anisotropy '' . the name was given because the finiteness of inflating islands was due to the classical motion of the slow - roll field outside of the bubble . clearly , the same effect arises in all quasi - open universes we have considered . in some cases the appearance of the inflating island is better described as a semiclassical effect , but the resulting inflating islands are just as classical here as they were in ref . hence , we shall use the same name for this type of anisotropies . to proceed , it will be important to distinguish between two different cases . the first case arises when the r.m.s . fluctuation of the slow - roll field @xmath16 on the spacelike surfaces @xmath271 is small compared with @xmath30 . in this case , `` high peaks '' where the field is comparable to @xmath30 and which will lead to inflating regions of co - moving size @xmath294 will be very `` rare '' on that hypersurface . here , @xmath1 is the correction to the supercurvature eigenvalue , calculated in section iii.c for the supernatural model and in section v.a for the more general models . high peaks of a homogeneous gaussian random field tend to be spherical , and so our inflating islands will have approximate spherical symmetry . in the opposite case , when the r.m.s . fluctuation of the slow - roll field is comparable to @xmath30 , we will have a patchy mosaic of `` overlapping '' inflating islands , as described in the second paragraph of section iv . it is easy to check that in the second case the `` classical '' effect is small compared with the effect of quantum fluctuations which we shall consider in the next subsection , so here we shall only consider the first case . let us begin with the supernatural model . the quantum state we are considering leads to a gaussian distribution for the random field @xmath16 that is @xmath6 symmetric . hence , to compute the probabilities for the field distribution around any point it suffices to study them around the origin @xmath158 . here we shall only be concerned with fluctuations due to the supercurvature modes , which have a long range . the effect of subcurvature modes can be incorporated in the usual way . note that the r.m.s . amplitude for the @xmath147 supercurvature mode is a factor @xmath160 larger than the amplitude for @xmath164 modes ( recall that @xmath295 ) . hence , even if the r.m.s . of @xmath16 is far below @xmath30 , there is a certain probability for @xmath16 to reach @xmath30 in a certain region near the origin . the spherically symmetric mode is the one that is most likely to contribute to this possibility . even though there is a small probability for this to happen , it is clear that only those rare regions with @xmath245 will undergo a second stage of inflation ; so they will be the only ones that matter . the value of the field on those inflating islands will have the radial dependence of the @xmath147 mode , which decays as @xmath296 at large distances , @xmath297 . let us now discuss the more general models where the slow - roll field has a small mass or it is massless outside the bubble . in this case , one may ask what happens when a bubble nucleates in a place where the slow - roll field already had a large ( classical ) value . this may occur , for instance , if the whole universe was created at a large value of @xmath16 , and at the time when the bubble nucleates @xmath16 is still rolling down from large values . this possibility would in principle be relevant for bubbles nucleated at early times , and is the one considered in @xcite . however , as time goes by , the initially large classical value of the slow - roll field in the false vacuum will decrease , and all that will remain are the quantum fluctuations which should be well described by the @xmath108 or de sitter - invariant quantum state . occasionally , fluctuations of the slow - roll field in the false vacuum may create a localized region with a higher value of the field . the nucleation of a bubble on top of one of these regions will not be very different from the case discussed in the previous paragraph . whether the bubble nucleates on one of these high peaks or not , the field outside the bubble will continue to fluctuate , and the bubble walls will from time to time bump into regions with a higher value of the field , as discussed in section v.b . hence , also in this case , there will be an ensemble of inflating regions with some distribution inside the bubble . from a formal point of view , notice that the appearance of the bubble has selected a point in spacetime , thus breaking the @xmath108 invariance , but otherwise respects a residual @xmath6 symmetry . therefore it seems reasonable to expect that , at least in a statistical sense , the field inside the bubble will be well described by the @xmath6 invariant gaussian distribution , corresponding to the quantum state we have studied . just as in the case of the supernatural model , here we also expect that the high peaks which lead to inflating islands will have spherical symmetry , and the value of the field on those islands will have the radial dependence of the @xmath147 mode , which decays as @xmath296 at large distances from the center of the island . the co - moving size of the inflating islands is @xmath298 . since the volume on the hyperboloid grows exponentially with the distance to the origin , as @xmath299 , a typical observer in a quasi - open universe is most likely to be at @xmath297 , where the scalar field behaves as @xmath300 . up to exponentially small corrections , this is the same radial dependence that was considered in @xcite . in that case , the fields were uncoupled and @xmath301 . the arguments used in @xcite to estimate the temperature anisotropies measured by a typical observer can be directly applied to the models discussed here . changing from the coordinates @xmath302 to a new set @xmath303 such that the point @xmath304 ( with @xmath305 ) is the new origin of coordinates , one finds that the perturbation of the field @xmath16 around @xmath306 can be described as @xcite @xmath307 , where @xmath308 and @xmath309 is the value of the field at the point @xmath306 . the corresponding gauge invariant potential at horizon crossing is @xmath310 the effect on the microwave background temperature fluctuations can be computed by integrating the sachs - wolfe effect along the line of sight @xcite . the dominant effect is in the quadrupole @xcite , and it is of order @xmath311 this is just a very rough order of magnitude estimate , which works well for @xmath312 . a more detailed study of the power spectrum of temperature anisotropies will be presented elsewhere @xcite . note that if the universe is sufficiently flat , the factor @xmath313 may completely erase the effect . otherwise , for a universe with appreciable curvature , we obtain a constraint on @xmath1 , @xmath314 where we have used @xmath315 , from the requirement that this effect does not dominate the temperature anisotropies on large scales , as seen by cobe @xcite . since @xmath144 is larger than @xmath286 , it is clear that @xmath1 has to be very small in order to avoid large temperature fluctuations . in the supernatural model , @xmath316 , this constraint implies @xmath317 which is not difficult to accomodate . note that the size of the bubble @xmath99 is necessarily less than the hubble radius in the false vacuum and hence can easily be much less than the hubble radius in the true vacuum . however , in the decoupled model discussed in @xcite the constraint ( [ semiclassical ] ) forces @xmath144 to be quite large , and this causes a problem of large quantum fluctuations in the supercurvature modes @xcite . in the class of models ( [ general ] ) , the size of @xmath1 is determined by eq . ( [ gamma ] ) . clearly , the effect can be made small by choosing parameters such that @xmath285 and the size of the bubble are sufficiently small . this is not always straightforward to implement . for instance , the `` hybrid '' open inflation model considered in ref . @xcite turns out to be quasi - open and suffers from too large semiclassical anisotropies . however , it is possible to write an open hybrid model , with a massless inflaton in the false vacuum , that satisfies the constraints @xcite . in the previous subsection we have considered the case where the @xmath147 mode was `` oversized '' , meaning that it took an amplitude much larger than its expected r.m.s . because of this , an observer far from the center of the inflating region would see the anisotropy ( [ anis ] ) . in this section we shall estimate the anisotropies caused by the @xmath164 supercurvature modes . here we are not thinking that these higher modes are `` oversized '' ; they simply take random values of the order of their r.m.s . for simplicity , we shall consider an observer located at @xmath158 , but the effect should not be much different for an observer located elsewhere . the size of cmb anisotropies caused by the @xmath164 supercurvature modes has been estimated in @xcite . for the class of models ( [ general ] ) , where the supercurvature mode is normalized as in ( [ usual ] ) , the quadrupole cmb anisotropies are of order @xmath318 here @xmath319 are the temperature anisotropies caused by the subcurvature modes ( with @xmath105 ) . the supercurvatre effect decreases very fast with multipole number , basically as @xmath320 . if the fluctuations we observe in the cmb are due to inflation , then we need @xmath321 , and from ( [ superanisg ] ) we have that @xmath143 can not be too much larger than @xmath144 , unless the universe is almost flat . for the supernatural model , the supercurvature mode ( [ rmsdeg ] ) has a normalization @xmath322 times larger than its counterpart ( [ usual ] ) [ we are ignoring the mild enhancement due to the factor @xmath323 . hence , the analog of ( [ superanisg ] ) is @xmath324 therefore we need @xmath325 . since @xmath99 has to be necessarily smaller than @xmath326 , we have a two - fold restriction . on one hand , @xmath327 , and on the other @xmath328 . thus , it seems fair to say that the model is not as natural as it was thought to be @xcite : the difference in energy density between the true and the false vacuum can not span many orders of magnitude . the reason is the following : in spite of the fact that the field is massive in the false vacuum , a supercurvature mode exists . its normalization is not proportional to @xmath143 as in the usual case ( [ general ] ) , but to @xmath329 , which is even larger . the effect can be thought as the excitation of the pseudo - goldstone modes due to the acceleration of the domain wall `` boundary '' . the model may still be viable in a certain range of parameters . determining this range requires detailed analysis , which is left for future research @xcite . open inflation is an appealing way of reconciling an infinite open universe with the inflationary paradigm . in this scenario , a symmetric bubble nucleates in de sitter space , and its interior undergoes a second stage of slow - roll inflation to almost flatness . single - field models of open inflation can in principle be constructed , but it does not seem possible to do so without a certain amount of fine tuning @xcite . the basic problem is that there is a hierarchy between the large mass needed for successful tunneling and the small mass required for successful slow - roll . for that reason , it seems natural to consider two - field models of open inflation @xcite where one field does the tunneling and the other drives slow - roll inflation inside the bubble . in this paper we show that a large class of two - field models of open inflation do not lead to infinite open universes , as it was previous thought , but to an ensemble of inflating islands of finite size . the reason is that the quantum tunneling does not occur simultaneously along both field directions , and the equal - time hypersurfaces in the open universe are not synchronized with equal - density or fixed - field hypersurfaces . technically , one finds that there are no @xmath43 invariant instantons for the two - field system which would describe the formation of a bubble with `` large '' values of the slow - roll field in its interior . large values of the inflaton field , needed for the second period of inflation inside the bubble , only arise as localized fluctuations . the interior of each nucleated bubble will contain an infinite number of such inflating regions , giving rise to a rather unexpected form of the large scale structure of the universe in these models . the picture is the following . right after the bubble has nucleated there will be , on the @xmath8 hypersurfaces inside the bubble , a certain density of occasional large fluctuations of the slow - roll field that lead to inflating islands . this fluctuations are caused by modes whose wavelength is larger than the curvature scale . denoting by @xmath1 the eigenvalue of the laplacian on the unit hyperboloid ( with @xmath330 ) , the co - moving size of the inflating islands is given by @xmath185 [ the parameter @xmath1 can be determined in terms of the parameters of the model , see eqs . ( [ gamma1 ] ) and ( [ gamma ] ) , and it is important in deriving observational constraints ] . each one of the inflating islands will be a quasi - open universe . since the volume of the hyperboloid is infinite , inflating islands with all possible values of the field at their center will be realized inside of a single bubble . we may happen to live in one of those patches where the universe appears to be open . the fact that the inflating regions are finite gives rise to classical anisotropies like those discussed in ref . @xcite . in particular , we have studied the supernatural model introduced by linde and mezhlumian @xcite . we have shown that in spite of the large mass of the inflaton field in the false vaccuum , there is a supercurvature mode . its amplitude is proportional to @xmath329 , rather than the usual @xmath143 . here @xmath99 is the radius of the bubble at the time of nucleation and @xmath143 is the hubble rate in the false vacuum . since @xmath331 , this effect is quite important . in order to make the model compatible with observations , it is required that the energy density in the false vacuum should not be much larger than in the true vacuum . this means that @xmath332 can not span many orders of magnitude , as it was previously believed @xcite . the supercurvature mode can be understood as the pseudo - goldstone mode associated with the choice of a tunneling direction in field space . combining the supercurvature anisotropies with the classical ones we find that the range of @xmath237 will also be restricted . detailed analysis is required in order to determine the range of parameters in which the model may still be viable @xcite . for the more general class of models ( [ general ] ) , the size of the inflating islands can be chosen to be comfortably large by an appropriate choice of parameters . in this way , the classical anisotropy will be unobservably small . by order of magnitude , the constraint is given by eq . ( [ semiclassical ] ) , where @xmath1 is given in eq . ( [ gamma ] ) . the constraint will be satisfied if the mass of the slow - roll field is sufficiently small in the false vacuum _ and _ @xmath99 is much smaller than @xmath326 . in a future publication @xcite we will give more precise constraints from the observed power spectrum of temperature anisotropies of the cmb . finally , there are some two - field models of open inflation , such as the one introduced by green and liddle @xcite in the context of induced gravity , which need not be affected in principle by the classical anisotropies mentioned above . in these models , the value of @xmath237 is not variable ; it is determined in terms of the parameters in the potential . it would be interesting to check whether @xmath43 symmetric instantons do indeed exist in this model . j.g.b . and j.g . thank andrei linde for very stimulating discussions . j.g . also thanks alex vilenkin for very useful conversations , and the theory division at cern for their hospitality during part of this work . j.g . and x.m . acknowledge financial support from cicyt under contract aen95 - 0882 and from european project ci1-ct94 - 0004 . we will show in this appendix that the two - point function on a @xmath157 surface for the state @xmath333 dies off as @xmath173 , where @xmath172 is the comoving distance between the points . we will compute the two - point function ouside the lightcone , and then continue it to the inside . to compute the two - point function for @xmath79 , hereafter @xmath334 , we will use the fact that the modes @xmath95 are properly normalized as klein - gordon modes of mass @xmath88 living in the @xmath335 ( 2 + 1 ) de sitter hypersurfaces of the outside lightcone metric . thus we can define the fields @xmath336 and the de sitter invariant vaccuum @xmath337 anihilated by @xmath338 . notice that @xmath339 in ( [ metric1 ] ) corresponds to the line element of a closed coordenization of a ( 2 + 1 ) de sitter space , for which the two - point functions can be found in ref . @xcite . we can now write the two - point function @xmath340 in terms of the two - point functions for @xmath341 , @xmath342 where the sum over p has to be understood like a sum over the discrete eigenvalues @xmath87 of the schrdinger equation ( [ schrodinger ] ) and like an integration over its continuum spectrum @xmath105 . on a given @xmath343 hypersurface , the @xmath59 dependence of @xmath334 will be given by @xmath344 , weighted for each @xmath121 by @xmath345 . the two - point function @xmath346 can be found in @xcite @xmath347 where @xmath348 is the hypergeometric function and @xmath349 is the scalar product of the position vectors at points @xmath137 and @xmath350 in the embedding ( 3 + 1 ) mikowski space , @xmath351 here @xmath352 is the angle on the 2-sphere between the two points . we recall that for the lowest discrete eigenmode to first order in the shift @xmath1 , @xmath353 , so we will denote by @xmath354 the two - point function for this eigenmode . taking @xmath306 and @xmath356 , so @xmath357 , and using eq . ( 9.131.1 ) in ref . @xcite , we find that inside the light - cone the two - point function between points separated a comoving distance @xmath172 can be written as @xmath358 as @xmath359 , the hypergeometric function in ( [ twopointinside ] ) tends to a constant , and the assymptotic behaviour of @xmath354 is given by @xmath360 which dies off exponentially with @xmath172 . to compute @xmath362 for large @xmath163 , we will need the asymptotic expressions for the hyperbolic harmonics @xmath363 for @xmath162 . the legendre functions are given by ( see e.g. ref . @xcite ) @xmath364 from equation ( [ hyper ] ) , the supercurvature mode @xmath365 is given by @xmath366^{1/2 } \frac{\sinh \lambda r}{\lambda\sinh r},\ ] ] where @xmath367 . for large @xmath163 , we have @xmath368.\ ] ] for @xmath164 , we can express @xmath369 in terms of @xmath370 using the recursion formula @xmath371 the legendre function @xmath372 acquires then the form @xmath373 where @xmath374 are some functions depending on @xmath375 . in fact , for large @xmath163 , we do not need to compute all @xmath374 . we have to take into account that for a supercurvature mode , @xmath376 behaves for large @xmath163 as @xmath377 ( as can be seen from ( [ legzero ] ) ) . thus , for the supercurvature mode @xmath378 , the term @xmath379 in ( [ sum ] ) grows exponentially faster than the rest of terms in the sum , so the main contribution for large @xmath163 will be given by this term . the coefficienty @xmath380 can be easily read from ( [ recursion ] ) : @xmath381 for large @xmath163 , using ( [ sum ] ) , ( [ important ] ) and ( [ hyper ] ) , we obtain @xmath382^{1/2}\gamma(1-\gamma/2 ) e^{-\gamma r/2 } y_{lm}(\omega).\ ] ] finally , using @xmath383 , we can write @xmath384 for large @xmath156 , to order @xmath1 , as @xmath385.\ ] ] using the results derived above , we can compute the amplitude of the @xmath147 mode near the origin , @xmath386,\ ] ] and for large @xmath163 , @xmath387 as we can see , the amplitude of the mode @xmath147 decays exponentially for @xmath162 . taking into account that we have choosen an @xmath6 symmetric vaccuum , this decrease in amplitude for large @xmath163 must be compensated by the joint contribution of the @xmath164 modes , smeared over a suitable length scale , in such a way that the r.m.s fluctuations of the field are independent of @xmath163 . let us check it . we need to compute @xmath388 where @xmath389 is a certain cutoff , which has to grow as we move away from the origin to include more and more modes in the sum . if we smear the field over a fixed comoving lenght @xmath167 , realizing that the wavelength of the @xmath156-th multipole is proportional to @xmath168 , we can take @xmath390 . finally , we obtain @xmath391 as we can see , as long as @xmath392 , the added contribution of the relevant modes is the same as the one given by the @xmath147 mode near the origin . in the thin wall approximation , neglecting gravitational backreaction , the background geometry is found @xcite to be described by two de sitter pieces with different hubble constant glued together at some @xmath393 . the scale factor is given by @xmath394 where @xmath395 and @xmath396 are the scale factors in the false and in the true vaccuum , @xmath397 continuity of @xmath58 at the wall implies @xmath398 where @xmath99 is the radius of the wall , and @xmath399 is given by @xmath400 to complete the description , we need to know the value of @xmath99 . it can be found in ref . @xcite : @xmath401 where @xmath402 and @xmath403 is the wall tension . we want to find the lowest eigenvalue of the schrdinger equation ( [ schrodinger ] ) in the background given above . the effective potential is given in this case by @xmath404 + ( { \cal h}_f-{\cal h}_t)\delta(\eta-\eta_w),\ ] ] where @xmath405 and similarly for @xmath406 . we will take a perturbative approach . we will divide the effective potential @xmath407 in an unperturbed one , @xmath408 , and in a small perturbation part , @xmath409 : @xmath410 the unperturbed @xmath408 corresponds to the effective potential of a massless scalar field in de sitter space , which has as a ground state a supercurvature mode with energy @xmath411 and wavefunction @xcite @xmath412 to first order in perturbation theory , the shift of the energy @xmath413 is given by @xmath414 where @xmath285 is the efective mass of the slow - roll field in the false vaccuum , and @xmath286 the effective mass in the true vaccuum . in this case , @xmath275 and @xmath415 .
we show that a large class of two - field models of single - bubble open inflation do not lead to infinite open universes , as it was previously thought , but to an ensemble of very large but finite inflating ` islands ' . the reason is that the quantum tunneling responsible for the nucleation of the bubble does not occur simultaneously along both field directions and equal - time hypersurfaces in the open universe are not synchronized with equal - density or fixed - field hypersurfaces . the most probable tunneling trajectory corresponds to a zero value of the inflaton field ; large values , necessary for the second period of inflation inside the bubble , only arise as localized fluctuations . the interior of each nucleated bubble will contain an infinite number of such inflating regions of comoving size of order @xmath0 , where @xmath1 depends on the parameters of the model . each one of these islands will be a quasi - open universe . since the volume of the hyperboloid is infinite , inflating islands with all possible values of the field at their center will be realized inside of a single bubble . we may happen to live in one of those patches of comoving size @xmath2 , where the universe appears to be open . in particular , we consider the `` supernatural '' model proposed by linde and mezhlumian . there , an approximate @xmath3 symmetry is broken by a tunneling field in a first order phase transition , and slow - roll inflation inside the nucleated bubble is driven by the pseudo - goldstone field . we find that the excitations of the pseudo - goldstone produced by the nucleation and subsequent expansion of the bubble place severe constraints on this model . we also discuss the coupled and uncoupled two - field models .
let @xmath3 be a one parameter family of riemannian metrics for @xmath7 $ ] . we say that the time - dependent @xmath8 is a generalized geometric flow if it is evolving by the following equation @xmath9\ ] ] with @xmath10 , where @xmath11 , @xmath12 is the maximal time of existence , i.e. , @xmath13 is the first time where the flow blows - up and @xmath14 is a general time - dependent symmetric @xmath15-tensor . here @xmath14 is smooth in both variables @xmath16 and @xmath17 . this is obvious since @xmath18 is smooth in both variables also . the scaling factor @xmath19 in ( [ eq11 ] ) is insignificant while the negative sign may be important in some specific applications for the purpose of keeping the flow either forward or backward in time . two popular examples of geometric flows are the ricci flow where @xmath20 ( the ricci curvature tensor ) and the mean curvature flow , in which case @xmath21 ( here @xmath22 is the mean curvature and @xmath23 is the second fundamental form on @xmath24 ) . one may impose boundedness condition on the tensor @xmath14 . in fact , such boundedness and sign assumptions on @xmath14 are preserved as long as the flow exists , so it follows that the metrics are uniformly equivalent . precisely , if @xmath25 , where @xmath26 $ ] is the flow , then @xmath27 to see the above bounds ( [ eq2.3 ] ) we consider the evolution of a vector form @xmath28 . by the equation ( [ eq11 ] ) and by the boundedness of the tensor @xmath29 we have @xmath30 which implies ( by integrating from @xmath31 ) @xmath32 taking the exponential of this estimate with @xmath33 and @xmath34 yields @xmath35 from which the uniform boundedness of the metric follows . thus , if there holds boundedness assumption @xmath25 , the metric @xmath3 are uniformly bounded below and above for all time @xmath36 under the geometric flow ( [ eq11 ] ) . then , it does not matter what metric we use in the argument that follows . in this paper , we discuss the evolution , monotonicity and differentiability of the first nonzero eigenvalue of @xmath4-laplace operator on the @xmath1-dimensional compact riemannian manifold whose metric evolves by ( [ eq11 ] ) . the @xmath4-laplace operator is defined as @xmath37 for @xmath38 , where @xmath39 is the divergence operator , the adjoint of gradient ( @xmath40 ) for the @xmath41-norm induced by @xmath42 on the space of differential forms . when @xmath6 , @xmath43 is the usual laplace - beltrami operator . the eigenvalues and the corresponding eigenfunctions of @xmath44 satisfy the following nonlinear eigenvalue problem @xmath45 it is easily verifiable that the principal symbol of ( [ eq12 ] ) is nonnegative everywhere and strictly positive at the neighbourhood of the point where @xmath46 . we know that ( [ eq12 ] ) has weak solutions with only partial regularity of class @xmath47 in general . interested readers can find the classical papers by l. evans @xcite and p. tolksdorff @xcite . notice that the least eigenvalue of @xmath48 on compact manifold without boundary is zero with the corresponding eigenfunction being a constant . hence , we refer to the infimum of the positive eigenvalues as the first nonzero eigenvalue or simply the first eigenvalue . the first eigenvalue of @xmath44 is characterised by the min - max principle @xmath49 satisfying the following constraint @xmath50 , where @xmath51 is the volume measure on @xmath0 . obviously , the infimum does not change when one replaces @xmath52 by @xmath53 . the corresponding eigenfunction is the energy minimizer of rayleigh quotient ( [ eq13 ] ) and satisfies the following euler - lagrange equation @xmath54d\mu_g= 0\ ] ] for @xmath55 in the sense of distribution . the problem of finding @xmath56 is related to the problem of finding the best constant @xmath57 in the @xmath58-sobolev inequality @xmath59 which is obtained by continuous embedding of @xmath60 under the sobolev norm @xmath61 it is well - known that @xmath4-laplacian has discrete eigenvalues but still remains unknown whether it only has discrete eigenvalues for bounded connected domains . another well - known results tell us that the first nonzero eigenvalue is simple and isolated @xcite . here the simplicity shows that any nontrivial eigenfunction corresponding to @xmath62 does not change sign and that any two first eigenfunctions are constant multiple of each other . in contrast to the spectrum of laplace - beltrami operator ( the case @xmath6 ) , the @xmath4-laplacian is nonlinear in general . for instance , the case @xmath63 gives @xmath64 , the negative of mean curvature operator . aside being of geometric interests , the @xmath4-laplacian appears naturally in the study of non - newtonian fluids , nonlinear elasticity , heat radiation , porous media flow , rheology , petroleum extraction , brownian motions , torsional creep problem among others . for details of physical applications of @xmath4-laplacian , the reader is referred to @xcite . moreover , it is not known if @xmath62 or its corresponding eigenfunction is @xmath65-differentiable ( or even locally lipschitz ) along any geometric flow of the form ( [ eq11 ] ) . however , it has been pointed out that the differentiability for the case @xmath6 is a consequence of eigenvalue perturbation theory , see for instance @xcite . for this reason , the method of l. ma @xcite , which assumes differentiability of eigenvalues and eigenfunctions under the ricci flow , can only be applied to find the monotonicity of first eigenvalue for the case @xmath6 along ( [ eq11 ] ) . now to avoid the differentiability assumption on the first eigenvalue and the corresponding eigenfunction in the case @xmath66 , we shall apply techniques of cao @xcite as used by wu @xcite and wu , wang and zheng @xcite under the ricci flow to study the evolution and monotonicity of @xmath67 , where @xmath68 and @xmath69 are assumed to be smooth . the evolution and the monotonicity formulas for the first eigenvalue ( in both cases @xmath6 and @xmath70 ) derived here do not depend on the evolution of the eigenfunction . the eigenfunction only needs to satisfy certain normalization condition . there are many results on the evolution and monotonicity of eigenvalues of the laplace operator on evolving manifolds with or without curvature assumptions . one can find @xcite under the ricci flow , @xcite under ricci - harmonic map flow and @xcite along abstract geometric flow with entropy methods . the study of the properties of eigenvalues of the @xmath4-laplacian on evolving manifold is still very young . the main aim of this paper is to investigate if those known properties of @xmath62 on static metric and for the case @xmath6 on evolving metric can be extended to various geometric flow . we however intend to develop a unified algorithm that can be used for this purpose on time - dependent metrics . many interesting results concerning the behaviour of @xmath62 can be found in @xcite for static metrics and @xcite for evolving metrics along various geometric flow . the rest of this paper is planned as follows . in section [ sec2 ] , we discuss the main results of this paper . firstly , we highlight some notations about differential geometry including metrics , gradient , divergence and integration by parts , which form the core of analytic tools used in the paper . we also prove some technical lemmas about evolutions of some geometric objects relating to the eigenvalues under the flow ( [ eq11 ] ) . in section [ sec3 ] , we study the first eigenvalue of laplace - beltrami operator ( @xmath6 ) under this geometric flow . here , we assume the first eigenvalue to be a function of time only and obtain its general evolution and monotonicity formula under certain condition . in section [ sec4 ] , we use a different approach to derive @xmath4-eigenvalue s evolution and monotonicity without differentiability assumption on the eigenfunction . in fact , the differentiability of @xmath4-eigenvalue is a consequence of the monotonicity formula . in the last section , we list some examples of geometric flows where the approaches used in this paper are applicable . in fact , this section reveals that our generalised geometric flow is not a trivial generalisation . throughout , @xmath24 will be taken to be a closed manifold ( i.e. , compact without boundary ) . most of our calculations are done in local coordinates , where @xmath71 is fixed in a neighbourhood of every point @xmath72 . we shall adopt einstein summation convention with repeated indices summed up . the riemannian metric @xmath73 at any point @xmath11 is a bilinear symmetric positive definite matrix denoted in local coordinates by @xmath74 the laplace - beltrami operator acting on a smooth function @xmath75 on @xmath24 is defined as divergence of gradient of @xmath76 , written as @xmath77 where @xmath78 and @xmath79 are determinant and the inverse metric matrix respectively . by the above we note that @xmath80 where @xmath81 is a smooth vector field . also we have the metric norm @xmath82 the riemann structure allows us to define riemannian volume measure @xmath51 on @xmath24 by @xmath83 by the divergence theorem , we have the following integration by parts formulas : let @xmath81 be a vector field , @xmath84 and @xmath85 be smooth function , then @xmath86 also for functions @xmath87 @xmath88 we write in local coordinates gradient , hessian and covariant derivative as @xmath89 respectively . also we write time derivative as @xmath90 . interestingly , all the geometric quantities associated with the underlying manifold evolve as the riemannian metric evolves along the geometric flow . this also serves as a motivation considering the behaviours of some other important geometric quantities such as eigenvalues of the manifold under the flow . the next two lemmas give us these evolutions . [ lem211 ] suppose a one - parameter family of smooth metrics @xmath3 solves the geometric flow ( [ eq11 ] ) , then , we have the following evolutions @xmath91 here @xmath92 , the metric trace of a symmetric @xmath19-tensor @xmath14 and @xmath76 is a smooth function defined on @xmath24 . recall that both @xmath18 and @xmath93 are symmetric tensors and @xmath94 . note also that levi - civita connection is not a tensor , but the time derivative of a connection is @xmath95 tensor . then @xmath96 holds as a tensor equation in any coordinates system and at any point . the proofs of @xmath97 and @xmath96 are the same as those of ( * ? ? ? * lemmas 3.1 and 3.2 ) . we give the computations in local coordinates which lead to the proofs of @xmath98 and @xmath99 . in fact , we have @xmath100 which is exactly ( 3 ) . we prove ( 4 ) by using ( 1 ) and ( 2 ) as follows : @xmath101 \\ & = \frac{\partial}{\partial t } ( g^{ij } ) ( \partial_i \partial_j - \gamma_{ij}^k \partial_k ) f + g^{ij } ( \partial_i \partial_j - \gamma_{ij}^k \partial_k ) \frac{\partial}{\partial t } f - g^{ij } \big ( \frac{\partial}{\partial t } \gamma_{ij}^k \big ) \partial_k f \\ & = 2 h^{ij } \nabla_i \nabla_j f + \delta f_t + g^{ij } g^{kl } \big ( \nabla_i h_{jl } + \nabla_j h_{il } - \nabla_l h_{ij}\big ) \nabla_k f \\ & = \delta f_t + 2 h_{ij } f_{ij } + 2 g^{kl } \big ( g^{ij } \nabla_i h_{jl } - \frac{1}{2 } g^{ij } \nabla_l h_{ij } \big ) \nabla_k f \\ & = \delta f_t + 2 h_{ij } f_{ij } + 2 \langle \ div \ h , \nabla f \rangle - \mathcal{h } _ i f_j . \end{aligned}\ ] ] in local coordinates , the volume form is written as @xmath102 , then , @xmath103 by chain rule of differentiation @xmath104 @xmath105 therefore @xmath106 this completes the proof of the lemma . [ lem212 ] suppose a one - parameter family of smooth metrics @xmath3 solves the geometric flow ( [ eq11 ] ) , then , we have the following evolutions @xmath107 where @xmath108 and @xmath76 is a smooth function defined on @xmath24 . when @xmath6 we have ( 2 ) and ( 3 ) of lemma [ lem211 ] . the proof follows standard computation as in ( * ? ? ? * lemma 2.2 ) . but we include it here for completeness . by ( 3 ) of lemma [ lem211 ] we can get @xmath109 which is ( 1 ) . ( 2 ) follows from the same calculation as the above . let @xmath108 , then @xmath110 by ( 1 ) and ( 3 ) of the lemma , we have @xmath111 recall that we already mentioned that sign assumption on the tensor @xmath14 is preserved throughout the flow . to prove the monotonicity of @xmath62 , we will need the condition @xmath112 for all @xmath113 $ ] . this condition is informed by the hamilton s maximum principle for tensors . for clarity we state the principle without proof . ( * null - eigenvector assumption * ) a quantity @xmath114 \to sym^2t^*m$ ] is said to satisfy the null eigenvector assumption if whenever @xmath115 is a nonnegative symetric @xmath116-tensor at a point @xmath117 and if @xmath118 is such that @xmath119 and then @xmath120 for any @xmath121 $ ] . the symetric tensor @xmath115 is defined to be nonnegative if and only if @xmath122 for all vectors @xmath123 ( i.e. , the quadratic form induced by @xmath115 is semi - positive definite ) and we write @xmath124 . ( ( * ? ? ? * theorem 9.1 ) , ( * ? ? ? * theorem 4.6 ) ) let @xmath3 be a smooth one parameter family of riemannian metrics satisfying ( [ eq11 ] ) . let @xmath125 be a symmetric @xmath116-tensor satisfying @xmath126 where @xmath81 is a time - dependent vector field and @xmath127 is a symmetric @xmath116-tensor which is locally lipschitz in @xmath17 , continuos in @xmath16 and satisfies the null - eigenvector assumption . if @xmath128 for all @xmath129 , then @xmath130 for all @xmath129 and @xmath131 by the above maximum principle for tensors we can prove the following conclusion . let @xmath3 be a smooth one parameter family of riemannian metrics satisfying ( [ eq11 ] ) . if @xmath132 then @xmath133 for some @xmath134 $ ] and all @xmath113.$ ] the first main results on evolution and monotonicity of @xmath62 for the case @xmath6 are the following and are proved in section [ sec3 ] . [ thm311 ] let @xmath135 be a closed manifold evolving by the geometric flow ( [ eq11 ] ) and @xmath136 be the first nonzero eigenvalue of the laplacian @xmath137 corresponding to the eigenfunction @xmath138 , then @xmath136 evolves by @xmath139 by setting @xmath140 , along the flow we have the following monotonicity formula from the last theorem @xmath141 by the definition of @xmath136 or from ( [ eq21 ] ) we know that @xmath142 and @xmath143 . suppose further that @xmath144 , then , @xmath145 the monotonicity of @xmath136 here depends on the sign of @xmath146 . note that in applications the sign of @xmath146 is usually preserved throughout the evolution . an interesting case is when the manifold is being evolved under the ricci flow @xcite , where the nonnegativity of scalar curvature is preserved . recall also that @xmath146 evolves by @xmath147 where @xmath148 , in particular , under ricci flow where @xmath20 and @xmath149 , we have @xmath150 . here we will assume that @xmath151 this is motivated by an error term appearing in a result of mller ( * ? ? ? * lemma 1.6 ) . for our case the error term reads ; for any time - dependent vector field @xmath81 on @xmath24 @xmath152 where @xmath153 is the ricci curvature tensor of @xmath24 . clearly the last three terms in ( [ eq224 ] ) above is the same as the quantity @xmath154 . it does make sense to assume ( [ eqb ] ) holds whenever @xmath155 is nonnegative . an application of this is that we are on a steady or shrinking soliton ( self - similar solution to the geometric flow ) if the equality in ( [ eqb ] ) holds . writing @xmath156 and using the condition that @xmath157 , we have a governing differential inequality for the evolution of @xmath146 as follows @xmath158 suppose @xmath159 we can apply the maximum principle by comparing the solution of ( [ eq318 ] ) with that of the following ordinary differential inequality @xmath160 solving to @xmath161 therefore @xmath162 for all @xmath163 as long as the flow exists . hence we write ( [ eq25 ] ) as follows @xmath164 by this we can prove the following let @xmath135 be a closed manifold evolving by the geometric flow ( [ eq11 ] ) . let @xmath136 be the first nonzero eigenvalue of the laplacian @xmath137 . suppose @xmath165 then @xmath166 \geq 0,\ ] ] where @xmath167 and @xmath168 for @xmath36 . in the next we state our results concerning the general @xmath4 . the proofs are discussed in section [ sec4 ] . [ thm411 ] let @xmath169 be an @xmath1-dimensional closed riemannian manifold evolving by the geometric flow ( [ eq11 ] ) . let @xmath170 be the first eigenvalue of the @xmath4-laplacian on @xmath24 corresponding to the eigenfunction @xmath171 at time @xmath113 $ ] . then @xmath172 for all time @xmath173 $ ] . moreover , if it holds that @xmath174 then @xmath175 is monotonically nondecreasing along the geometric flow and it is differentiable almost everywhere . precisely , @xmath176 provided @xmath146 is nonnegative . [ cor4 ] with the conditions of theorem [ thm411 ] @xmath177 where @xmath178 furthermore , if @xmath179 and satisfies the governing inequality ( [ eq318 ] ) , it then holds for all time @xmath180 that @xmath181 finally , we show that the following quantity @xmath182 is nondecreasing ( see theorem [ thm44 ] ) and that @xmath183 is differentiable almost everywhere along the geometric flow . the eigenvalue problem involving laplace - beltrami operator on a closed manifold consists in finding all possible real @xmath184 such that there exists non - trivial functions @xmath185 satisfyng @xmath186 it is well - known that @xmath187 has a discrete spectrum on a closed riemannian manifold . this set consists of an infinite sequence @xmath188 and can be found that only constant functions correspond to @xmath189 . the eigenfunctions are @xmath190 orthonormal basis @xmath191 of real @xmath192 function such that @xmath193 while the eigenvalues are @xmath190 orthogonal . by the min - max principle , the first non - zero eigenvalue @xmath194 can be characterised as follows @xmath195 satisfying @xmath196 , where @xmath197 is the completion of @xmath198 with respect to the sobolev norm @xmath199 note that the @xmath4-laplacian @xmath200 is a natural generalisation of the laplace - beltrami operator . in this section , we consider the eigenvalues of the laplace - beltrami operator under the geometric flow , assuming the least eigenvalue @xmath201 is a function of time only . next we discuss the proof of theorem [ thm311 ] . let @xmath24 be a closed riemannian manifold and @xmath3 evolve by the generalized geometric flow ( [ eq11 ] ) in the interval @xmath202 let @xmath203 be the corresponding eigenfunction to the first nonzero eigenfunction @xmath204 of @xmath205 , i.e , @xmath206 taking derivative with respect to time , we have @xmath207 multiplying the above by @xmath208 and integrate with respect to the volume measure on @xmath24 , we have @xmath209 notice that by the application of integration by parts and ( [ eq21 ] ) @xmath210 then , we arrive at @xmath211 use the evolution of the laplacian under the geometric flow , ( i.e. , ( 4 ) of lemma ( [ lem211 ] ) ) , so that we have @xmath212 we express the first and the last terms of the above as follows @xmath213 and @xmath214 putting these together we have @xmath215 hence , we have proved theorem [ thm311 ] on the evolution of @xmath216 . in some applications @xmath146 may be required to be a constant or bounded by a constant . thus when one takes @xmath14 to be ricci curvature tensor one is talking about a manifold with constant scalar curvature . in this situation we have the following as a corollary . let @xmath135 be a closed manifold evolving by the geometric flow ( [ eq11 ] ) . let @xmath136 be the first nonzero eigenvalue of the laplacian @xmath137 . then if @xmath217 in @xmath218 $ ] for some uniform constant @xmath219 along the flow we have @xmath220 and @xmath221 in this section , we consider the nonlinear eigenvalue problem @xmath222\ ] ] with the normalization condition @xmath223 . we want to derive general evolution for the @xmath4-eigenvalues ( eigenvalues of @xmath48 ) and show that @xmath62 is monotone on metrics evolving by the geometric flow . in order to do these we need to calculate time evolution for @xmath62 and its corresponding eigenfunction . but unfortunately , we do not know whether @xmath62 or its corresponding eigenfunction @xmath224 is @xmath65-differentiable or not along the flow . so a similar approach to the one in @xcite ( see also wu @xcite , wu , wang and zheng @xcite and zhao @xcite ) will be used to avoid this difficulty . precisely , let @xmath225 $ ] be a smooth compact riemannian manifold evolving by the flow ( [ eq11 ] ) . define a genral smooth function as follows @xmath226 where @xmath227 is a smooth function satisfying the normalisation condition @xmath228 by this we claim that there exists a smooth function @xmath229 at time @xmath230 $ ] satisfying ( [ eqdef ] ) . to see this claim , we first assume that at time @xmath231 , @xmath229 is the eigenfunction corresponding to @xmath232 of @xmath233 which implies @xmath234 then we consider the following smooth function @xmath235 under the flow @xmath3 . we normalize this smooth function @xmath236 under the flow @xmath3 . by ( [ eqdef2 ] ) we can easily check that @xmath227 satisfies ( [ eqdef ] ) . note that in general @xmath237 is not equal to @xmath170 . but at time @xmath231 , if @xmath229 is the eigenfunction of the first eigenvalue @xmath232 , then we conclude that @xmath238 notice that the normalisation condition implies @xmath239 which by direct computation ( at @xmath231 ) yields the following @xmath240 by this it holds that @xmath241 we now present a proposition let @xmath3 be a smooth solution of the flow ( [ eq11 ] ) on @xmath24 . let @xmath170 be the first eigenvalue of the @xmath4-laplacian under ( [ eq11 ] ) . assume @xmath229 is the corresponding eigenfunction of @xmath170 at time @xmath242 $ ] , that is , @xmath243 let @xmath237 be a smooth function defined by @xmath244 then we have @xmath245 where @xmath227 is any smooth function satisfying ( [ eqdef ] ) such that at any time @xmath246 $ ] , @xmath229 is the eigenfunctin for @xmath232 . here @xmath247 denotes covariant derivative of @xmath185 with respect to the levi - civita connection of @xmath3 at time @xmath231 . since the evolution formula ( [ eq5.9 ] ) does not depend on the time derivative of the eigenfunction @xmath185 , we have that @xmath248 at some time @xmath231 . we are now set to prove a theorem about the evolution , monotonicity and differentiability ( theorem [ thm411 ] ) of the first eigenvalue of the @xmath4-laplacian under the geometric flow . clearly , we can now set @xmath249 the evolution of @xmath62 then follows @xmath250 the proof also follows by direct computation using evolution of quantities in lemmas [ lem211 ] and [ lem212 ] . denote @xmath251 , then working in local orthonormal coordinates we have @xmath252 u d\mu \\ \displaystyle & = \frac{\partial } { \partial t } \int_m \big ( g^{ij } \nabla_i z \nabla_j u + z \delta u \big ) u d\mu \\ \displaystyle & = \int_m \frac{\partial } { \partial t } \big ( g^{ij } \nabla_i z \nabla_j u + z \delta u \big ) u d\mu + \int_m \delta_p u \frac{\partial } { \partial t } ( u d\mu)\\ \displaystyle & = : i + ii.\end{aligned}\ ] ] by the evolution of @xmath48 in lemma [ lem212 ] we have @xmath253 using integration by parts on the second and third terms of the last integral we have @xmath254 therefore we have after using the evolution @xmath255 from lemma [ lem212 ] @xmath256 computing the first and the third terms on the right hand side ( rhs for short ) of ( [ eq411])as follows @xmath257 @xmath258 putting these back into ( [ eq411 ] ) we have @xmath259 using the integrability condition ( [ eq47 ] ) and the definition of @xmath260 in ( [ eq41 ] ) , the last two terms on the rhs of the above equation vanish and we then arrive at @xmath261 the next is to compute the second term on the rhs of the last equality using integration by parts as follows @xmath262 putting this into ( [ eq412 ] ) we obtain ( [ eq48 ] ) at once . hence , the first part of the theorem is proved . using the condition ( [ eq49 ] ) in ( [ eq48 ] ) we have the monotonicity formula ( [ eq49 ] ) with the condition @xmath263 . clearly , when @xmath6 we have @xmath264 , the usual laplace - beltrami operator . also @xmath265 , the first eigenvalue of @xmath266 and the corresponding eigenfunction are smoothly differentiable . then theorem [ thm311 ] reduces to a corollary . this further explains that the @xmath4-laplacian is a nonlinear generalisation of laplace - beltrami operator . integrating both sides of ( [ eq410 ] ) from @xmath267 to @xmath268 on a sufficiently small time interval @xmath269 , @xmath7 $ ] , we then obtain [ eq514 ] . of corollary [ cor4 ] . by the definition of @xmath270 in ( [ eq13 ] ) and the normalization condition @xmath271 we know that @xmath272 then ( [ eq410 ] ) reduces to @xmath273 from which ( [ eq414 ] ) follows by integrating on the interval @xmath274 $ ] with @xmath275.$ ] note that both @xmath276 and @xmath277 are functions of time only . setting @xmath278 we can evaluate @xmath279 therefore integrating both sides of ( [ eq415 ] ) from @xmath267 to @xmath268 and together with the above equality , yields @xmath280 for any time @xmath180 sufficiently close to @xmath268 . by this we have @xmath281 then @xmath282 is nondecreasing along the geometric flow . hence we conclude this section with the following . [ thm44 ] with the assumption of theorem [ thm411 ] . the following quantity @xmath182 is nondecreasing and @xmath183 is differentiable almost everywhere along the geometric flow ( [ eq11 ] ) . the last theorem has been proved using a different method under the ricci flow in @xcite . since @xmath170 is nondecreasing on the time interval @xmath283 ( under curvature assuption of the theorem ) , by the classical lebesgue s theorem [ a. mukherjea and k. pothoven : real and functional analyisis ( chap 4 ) ] , it is easy to see that @xmath170 is differentiable almost everywhere , our proofs of the first eigenvalue evolution and monotonicity do not use any differentiability of the first eigenvalue @xmath170 or its corresponding eigenfunction @xmath284 of the @xmath4-laplacian under @xmath285-flow . in fact , it is not known whether they are differentiable in advance . it would be interesting to find out whether the corresponding first eigenfunction of the @xmath4-laplacian is a @xmath65-differentiable function with respect to @xmath16-variable along @xmath285-flow . in this section , we give some examples of geometric flows where our results are valid . we remark that in these cases the error term @xmath286 and the quantity @xmath287 are nonnegative . more examples can be found in ( * ? ? ? * section 2 ) . let @xmath135 be a solution to the hamilton s ricci flow @xmath288 this is the case where @xmath289 is the ricci tensor and @xmath290 is the scalar curvature on @xmath24 . here , the scalar curvature evolves by @xmath291 by twice contracted second bianchi identity @xmath292 , which implies @xmath293 the quantity @xmath286 vanishes identically and @xmath294 . note that the positivity of curvature is preserved along the ricci flow @xcite . the evolution equation and monotonicity formula for the first eigenvalue follow easily ( see @xcite-@xcite and @xcite-@xcite ) . a fundamental result here is perelman s paper @xcite , where he defines his @xmath295-energy @xmath296 and proves that it is monotonically nondecreasing . he also defines @xmath297 with @xmath298 ( being the least eigenvalue of the geometric operator @xmath299 ) and its corresponding eigenfunction @xmath300 satisfying the eigenvalue problem @xmath301 he shows that monotonicity of @xmath298 follows from that of @xmath295 . let @xmath169 and @xmath302 be compact ( without boundary ) riemannian manifolds of dimensions @xmath303 and @xmath1 respectively . let a smooth map @xmath304 be a critical point of the dirichlet energy integral @xmath305 , where @xmath306 is isometrically embedded in @xmath307 by the nash embedding theorem . the configuration @xmath308 of a one parameter family of riemannian metrics @xmath309 and a family of smooth maps @xmath310 is defined to be ricci - harmonic map flow if it satisfies the coupled system of nonlinear parabolic equations @xmath311 where @xmath312 is the ricci curvature tensor for the metric @xmath42 , @xmath313 is a time - dependent coupling constant , @xmath314 is the intrinsic laplacian of @xmath315 , which denotes the tension field of map @xmath315 and @xmath316 is the pullback of the metric @xmath317 on @xmath306 via the map @xmath318 . see list @xcite when the target manifold is one dimensional . here @xmath319 , @xmath320 and @xmath321 using the twice contracted second bianchi identity , we have @xmath322 a straightforward computation gives @xmath323 then , @xmath324 and @xmath325 for all @xmath81 on @xmath24 . thus both @xmath326 and @xmath327 are nonnegative as long as @xmath328 is nonincreasing in time . the first author has considered this in @xcite . see also @xcite for the monotonicity of the first eigenvalue of laplace - betrami operator and versions of perelman s entropy under the ricci - harmonic map flow . let @xmath329 be a family of space - like hypersurfaces in ambient lorentzian manifold evolving by lorentzian mean curvature flow @xmath330 for @xmath331 \times m$ ] , where @xmath332 is the position of @xmath333 in @xmath334 satisfying @xmath335 . here @xmath336 and @xmath337 are respectively the outer normal vector and mean curvature at the point @xmath338 then , the induced metric evolves by @xmath339 where @xmath23 denotes the components of the second fundamental form @xmath340 on @xmath24 and @xmath341 denotes the mean curvature of @xmath24 . in this case @xmath342 and @xmath343 . letting @xmath344 and @xmath345 denote the ricci and riemman curvature tensor of @xmath334 respectively , we have the gauss equation @xmath346 the codazzzi equation @xmath347 the evolution equation @xmath348 and @xmath349 see the explicit forms of the gauss and the codazzi equations for @xmath350 in @xcite . combining the above equation we obtain the quantity @xmath351 where @xmath352 denotes future - oriented timelike unit normal vector on @xmath24 . obviously both @xmath286 and @xmath353 are nonnegative when assuming nonnegativity on sectional curvature of @xmath334 . see @xcite for the evolution and monotonicity of the first eigenvalue of @xmath4-laplace operator under the @xmath354 powers of the mean curvature flow . see also huisken @xcite , the second author s paper @xcite and @xcite for related results . this is the case when @xmath355 , where @xmath356 is the scalar curvature of the metric . yamabe flow is then the following evolution equation @xmath357 \\ \ \\ g_{ij}(x , 0 ) = g_0(x ) \end{array } \right.\ ] ] as introduced by r. hamilton who first establishes the existence of its unique solution for all time and shows that the metric @xmath3 approaches constant as @xmath358 . his proof is done for volume preserving flow @xmath359 with @xmath360 is the average of scalar curvature for the metric in a conformal class . for more on the global existence and convergence of ( [ eq518 ] ) see chow @xcite and ye @xcite . note that under the yamabe flow the volume measure evolves as @xmath361 and the normalization condition , @xmath362 , implies @xmath363 here , the evolution of scalar curvature is given @xcite as @xmath364 and by the strong maximum principle @xmath365 for all t. we can also compute @xmath366 and @xmath367 these imply that the quantitty @xmath368 is nonnegative on the einstein tensor @xmath369 hence our results hold . see ( * ? ? ? * section 7 ) , where with assumption that @xmath370 and @xmath371 , they prove that @xmath372 is strictly increasing and differentiable almost everywhere along the yamabe flow . yamabe flow coincides with the ricci flow on riemann surfaces . the first author was partially supported by nigeria tetfund research grant through oscotech , esa - oke . the second author was partially supported by the nsf of china ( grant no . 11401131 ) . the first version of the paper was completed during the first author s research visit to african institute for mathematical sciences ( aims ) , senegal in february 2016 . he therefore thanks aims - senegal for finacial suport during the visit . he also wishes to thank the research chair , prof . moustapha fall for the hospitality and useful discussions which helped in improving this paper .
let @xmath0 be an @xmath1-dimensional compact riemannian manifold ( @xmath2 ) whose metric @xmath3 evolves by the generalized abstract geometric flow . this paper discusses the evolution , monotonicity and differentiability for the first eigenvalue of the @xmath4-laplacian on @xmath5 with respect to time evolution . we prove that the first nonzero @xmath4-eigenvalue is monotone nondecreasing along the flow under certain geometric condition and that the first eigenvalue is differentiable almost everywhere . when @xmath6 , we recover the corresponding results for the usual laplace - beltrami operator . our results provide a unified approach to the study of @xmath4-eigenvalue under various geometric flows .
nuclear reactions are the ubiquitous experimental tool to study atomic nuclei . while many astrophysically relevant reactions proceed at relatively low energies @xmath2 mev @xcite , in the laboratory , these reactions are often studied indirectly with beams at higher energy ( @xmath35 mev / u ) . one of the most important open questions currently being explored today in our field concerns the astrophysical site for the r - process , the process that gave rise to about half of the heavy elements in our planet . in order to perform simulations of neutron star mergers or supernovae explosions ( the two possible sites under consideration ) , neutron capture rates are needed on rare isotopes of nuclei as heavy as uranium @xcite . despite all the effort with _ ab initio _ approaches to nuclear reactions , which include the study of elastic scattering @xcite , transfer @xcite , photo reactions @xcite , and capture reactions @xcite , only selected nuclei and specific reaction channels can be addresssed with the various ab - initio methods in the market ( refs . @xcite for recent reviews ) . a more general approach to reactions involving heavier nuclei is based on a reduction of the many - body picture to a few - body one , where only the most relevant degrees of freedom are retained @xcite . in such approaches one introduces effective interactions ( the so - called optical potentials ) between the clusters considered . traditionally these interactions have been constrained by data , particularly using data on @xmath4-stable isotopes @xcite . clearly , the application of these global parameterizations to exotic regions of the nuclear chart is unreliable and has uncontrolled uncertainties . it is critical for progress in the field of reactions that these effective interactions be connected to the underlying microscopic theory so that extrapolations to exotic regions can be better understood . in most cases , phenomenological optical potentials are made local for simplicity . we know based on the feshbach projection formalism that , in its most general form , the microscopic optical potential should be complex , non - local and energy dependent @xcite . recently , a series of studies has shown that nonlocality can affect transfer reaction observables ( e.g. @xcite ) and it is expected that it can equally affect other reaction channels . so far we have not been able to identify an experimental method to constrain nonlocality . it is essential that microscopic theories provide guidance on this aspect of the optical potential . the goal of this work is to provide a proof - of - principle for a new method to compute nuclear optical potentials from ab - initio many - body coupled cluster calculations . it is the first of a series of studies that aims at constructing an optical potential rooted in the underlying microscopic formulation of the problem , potentials which can then be incorporated , consistently with other ingredients , into the general few - body formalism . in an approach based on feshbach projection operators , the optical potential is the self - energy term in the dyson equation @xcite . thus , the solution of the dyson equation by self - consistent green s function methods can be used to compute optical potentials @xcite . in this paper , we compute the green s function directly from the coupled cluster method @xcite , thus circumventing the usual self - consistency approach . the self - energy can then be determined by inverting the dyson equation . the key elements in our approach to compute the green s function are : i ) an analytical continuation in the complex energy plane based on a berggren basis consisting of bound , resonant , and non - resonant scattering states @xcite , and ii ) a generalized non - symmetric lanczos method @xcite that allows us to write the green s function as a continued fraction @xcite . the first of these two elements is essential because it allow us to properly deal with the poles of the green s function along the real energy axis , and obtain numerically stable green s functions and optical potentials for arbitrary energies . the second element is essential to make the problem computationally feasible . in this work we demonstrate that optical potentials , converged with respect to the models space , can indeed be determined from the green s functions generated from coupled cluster many - body calculations . we note that the computation of green s functions with the coupled - cluster method is well established in quantum chemistry @xcite , and that very recently this approach has also been used to extract the optical potential @xcite . our approach is similar to that effort , but applied to nuclear many - body problem . this paper is organized as follows . in sec . ( [ sec_formalism ] ) we introduce the formalism of the green s function and the coupled - cluster method along with the berggren basis and discuss the application of the lanczos method for the numerical calculations of the green s function . in sec . [ secres ] , we show an application for the elastic scattering on @xmath5 and discuss the results . finally , we will conclude and discuss future possible applications in sec . [ conclusion ] . the single - particle green s function of an @xmath6-nucleon system has matrix elements @xmath7 here , @xmath8 and @xmath4 denote single - particle states , and @xmath9 is the ground state of the @xmath6-body system with energy @xmath10 . as usual , the parameter @xmath11 is such that @xmath12 at the end of the calculation . the operators @xmath13 and @xmath14 create and annihilate a fermion in the single - particle state @xmath8 and @xmath4 , respectively , and are shorthands for the quantum numbers @xmath15 . here , @xmath16 label the radial quantum number , the orbital angular momentum , the total orbital momentum , its projection on the @xmath17 axis , and the isospin projection , respectively . the intrinsic hamiltonian @xmath18 is @xmath19 here , @xmath20 is the momentum of the nucleon @xmath21 of mass @xmath22 and @xmath23 is the momentum of the center of mass . we limit ourselves to a two - body interactions @xmath24 and neglect contributions from three - nucleon forces . it is useful to rewrite the hamiltonian as @xmath25 separating one - body and two - body contributions . in what follows , we take the single - particle states from the hartree - fock ( hf ) basis . we recall that the hf basis is an excellent starting point for coupled - cluster calculations and that the hf green s function @xmath26 is a first order approximation to the green s function ( [ gf1 ] ) . in eq . ( [ hfgf ] ) @xmath27 is the hf reference state of the @xmath6-nucleon system and @xmath28 the corresponding energy . as the single - particle states @xmath29 are given by the hf basis eq . ( [ hfgf ] ) can be written as , @xmath30.\end{aligned}\ ] ] here , @xmath31 is the single - particle energy associated with @xmath32 and @xmath33 the unit step function . for a single - particle state @xmath8 above the occupied shells at the hf approximation , @xmath34 , whereas @xmath35 for @xmath36 below the fermi level . the green s function fulfills the dyson equation @xmath37 here , @xmath38 is the self energy , which can be obtained from an inversion of eq . ( [ dys ] ) @xmath39^{-1}-g^{-1}(e ) .\end{aligned}\ ] ] to obtain the optical potential we introduce the quantity @xmath40 with the hf potential @xmath41 . for @xmath42 , @xmath43 in eq . ( [ optdef ] ) corresponds to the optical potential for the elastic scattering from the @xmath6-nucleon ground state @xcite . we are interested in the scattering amplitude @xmath44 where @xmath45 is the elastic scattering state of a nucleon on the target with the energy @xmath46 . the scattering amplitude @xmath47 is the solution of the schrdinger equation containing the optical potential @xmath48 here , @xmath49 is the reduced mass of the nucleus - nucleon system , and @xmath50 the corresponding radial coordinate . for simplicity , we supressed any spin and isospin labels . the optical potential is non - local , energy - dependent and complex @xcite and , for @xmath51 , its imaginary component describes the loss of flux due to absorption . similarly , the overlap @xmath52 for a bound state @xmath53 of energy @xmath54 in the @xmath55 system , fulfills the schrdinger equation with the optical potential at the discrete energy @xmath56 . in this paper , we construct the optical potential by an inversion of the dyson equation ( [ dys ] ) after a direct computation of the green s function ( [ gf1 ] ) following the coupled - cluster method @xcite . in the following section , we present the main steps involved in the computation of the green s function in our approach . the hf reference state for the nucleus consisting of @xmath6 nucleons is @xmath57 denotes the cluster operator @xmath58 we note that @xmath59 and @xmath60 induce @xmath61-@xmath62 and @xmath63-@xmath64 excitations of the hf reference , respectively . here and in what follows , the single - particle states @xmath65 refer to hole states occupied in the reference state @xmath27 while @xmath66 denote valence states above the reference state . in practice , the expansion ( [ t_cluster ] ) is truncated . in the coupled cluster with singles and doubles ( ccsd ) all operators @xmath67 with @xmath68 are neglected . in that case , the ground - state energy and the amplitudes @xmath69 are obtained by projecting the state ( [ cc1 ] ) on the reference state and on all @xmath61-@xmath62 and @xmath63-@xmath64 configurations for which @xmath70 here , @xmath71 + { 1\over 2!}\left[\left[h , t\right],t\right ] + \ldots\end{aligned}\ ] ] denotes the similarity transformed hamiltonian and it can be computed systematiccaly y via the baker - campbell - hausdorff expansion . for two - body forces and in the ccsd approximation , this expansion actually terminates at fourfold nested commutators . the ccsd equations ( [ ccsd ] ) show that the ccsd ground state is an eigenstate of the similarity - transformed hamiltonian in the space of @xmath72-@xmath73 , @xmath61-@xmath62 , @xmath63-@xmath64 configurations . the transformed hamiltonian is not hermitian because the operator @xmath74 is not unitary . as a consequence @xmath75 has left- and right - eigenvectors which constitute a bi - orthogonal basis with the corresponding completeness relation @xmath76 the right ground state is the reference state @xmath77 , while the left ground - state is given by @xmath78 where @xmath79 is a linear combination of particle - hole de - excitation operators . using the ground state of the similarity - transformed hamiltonian , we now can write the coupled cluster green s function @xmath80 as @xmath81 here , @xmath82 and @xmath83 are the similarity - transformed annihilation and creation operators , respectively , and the baker - campbell - hausdorff expansion yields the relations @xmath84 , \\ \overline{a^{\dagger}_{\beta}}&=&a^{\dagger}_{\beta}+[a^{\dagger}_{\beta},t ] .\end{aligned}\ ] ] we note that the truncation of the cluster operator @xmath85 is reflected in the expression of the coupled - cluster green s function ( [ gfcc ] ) , and if all excitations up to @xmath86-@xmath87 were taken into account in the expansion ( [ t_cluster ] ) , the green s function ( [ gfcc ] ) would be identical to ( [ gf1 ] ) . one might be tempted to use the completeness relations for the @xmath88 systems to obtain the lehmann representation of the green s function @xmath89 here , @xmath90 ( @xmath91 ) is an eigenstate of @xmath75 for the @xmath55 ( @xmath92 ) system with energy @xmath93 ( @xmath94 ) . to simplify the notation , the completeness relations are written in ( [ gfcc2 ] ) as discrete summations over the states in the @xmath95 systems . in principle , the green s function ( [ gfcc2 ] ) could be obtained by calculating the spectrum of the @xmath95 systems using the particle - attached equation - of - motion ( pa - eom ) and particle - removed equation - of - motion ( pr - eom ) coupled - cluster methods @xcite . however in practice , this approach is difficult to pursue as the sum over all states also involves eigenstates in the continuum . to avoid this problem , we return to the expression in eq . ( [ gfcc ] ) and use the lanczos technique @xcite for its computation . in this section , we compute the coupled - cluster green s function ( [ gfcc ] ) using the lanczos method @xcite . to simplify the notation we introduce the following shorthands @xmath96 and write the green s function as @xmath97 for a truncation of @xmath85 at the @xmath98-@xmath99 level , the states @xmath100 and @xmath101 belong to the vector space @xmath102 spanned by the states built from @xmath61-@xmath73, ... ,@xmath98-@xmath103 excitations of the reference state @xmath77 . similarly , the states @xmath104 and @xmath105 belong to the vector space @xmath106 spanned by @xmath72-@xmath62, .. ,@xmath107-@xmath99 excitations of the reference state . introducing @xmath108 and @xmath109 defined as @xmath110|x_{\beta}\rangle&\equiv&|v^{a+1}_{\beta}\rangle , \label{eqx } \\ \left [ z_-+\overline{h}\right]|y_{\alpha}\rangle & \equiv&|v^{a-1}_{\alpha}\rangle , \label{eqy}\end{aligned}\ ] ] with @xmath111 and @xmath112 , we can write : @xmath113 this matrix element of the green s function is calculated by solving the systems of linear equations ( [ eqx ] ) and ( [ eqy ] ) in the lanczos basis . the advantage of working in the lanczos basis is twofold . first , the actual dimensions of the linear systems ( defined by the number of lanczos vectors @xmath114 ) needed to reach convergence , are much smaller than the dimension of the full space @xmath102 and @xmath106 . second , the resolution has to be done only once for all energies @xmath115 . let us now focus on the first term on the right hand side of ( [ gfcc4 ] ) , i.e. the term associated with the particle part of the green s function . starting with the normalized states @xmath116 and @xmath117 ( where the norm is @xmath118 ) as right and left lanczos pivots , we construct iteratively a set of @xmath114 pairs of lanczos vectors . by construction , @xmath75 is conveniently represented in the lanczos basis as a tridiagonal matrix : @xmath119 using the cramer s rule for the resolution of linear systems , one can then show that @xmath120 is given by the continued fraction @xmath121 as it is clear from the expression above , one just needs to solve the linear system ( [ eqx ] ) only once in order to calculate @xmath122 for any value of the energy @xmath115 . the convergence as a function of @xmath114 is quickly reached as we will show in sec . ( [ secres ] ) . the calculation of the second term in ( [ gfcc4 ] ) , i.e. the hole part of the green s function , proceeds in a similar manner . ultimately we want to compute the optical potential describing scattering processes at arbitrary energies . however , as @xmath12 , the coupled - cluster greens function in eq . ( [ gfcc2 ] ) has poles at energies @xmath123 which make the numerical calculation unstable . there have been various proposed solutions to this problem , such as using a complex scaling technique @xcite , or carrying calculations at finite values of @xmath124 and extrapolate to @xmath12 @xcite . another ( phenomenological ) approach to this problem is to employ a finite energy dependent width which account for damping and decay processes that are not included in the employed theoretical approach @xcite . in this work , we suggest a different approach based on an analytic continuation of the green s function in the complex energy plane using a berggren basis @xcite , that includes bound- , resonant , and discretized non - resonant continuum states . as we will demonstrate below , by employing the berggren basis it is possible to obtain stable numerical results as @xmath12 . thus , the set of hf states includes bound , resonant ( when they exist ) and complex - continuum states single - particle states . accordingly , the many - body spectrum for the @xmath55 ( @xmath92 ) systems obtained with the pa - eom ccsd ( pr - eom ccsd ) is composed of bound , resonant and complex - continuum states . in other words , the poles of the green s function [ cf eq . ( [ gfcc2 ] ) ] have either a negative real or complex energy . as a consequence , as @xmath12 , the values of the green s function matrix elements for @xmath125 smoothly converge to a finite value . in the case of a real hf basis consisting of bound , and discretized real energy continuum states , the calculation would become unstable for small @xmath124 since the green s function poles would then be located at real values of @xmath115 . in order to fulfill the berggren completeness @xcite , the complex - continuum single - particle states must be located along a contour @xmath126 in the fourth quadrant of the complex momentum plane below the resonant single - particle states . according to the cauchy theorem , the precise form of the contour @xmath126 is not important , provided all resonant states lie between the contour and the real momentum axis . the berggren completeness then reads @xmath127 where the discrete states @xmath128 correspond to bound and resonant solutions of the single - particle potential , and @xmath129 are complex - energy scattering states along the complex - contour @xmath126 . in practise , the integral along the complex continuum is discretized yielding a finite discrete basis set . we now present results for the elastic scattering of a neutron on @xmath0 . the choice of this problem is motivated by the fact that @xmath130 is a doubly magic nucleus and as such can be computed relatively precisely using the coupled - cluster method . we will work at in the ccsd approximation and use the @xmath131 @xcite nucleon - nucleon interaction . we also want to point out that we introduced a simplification for the solutions of the pa - eom ccsd and the pr - eom ccsd equations . instead of solving these problems with the mass @xmath55 ( @xmath92 ) for the @xmath55 ( @xmath92 ) sytems @xcite , we have used in all calculations the mass @xmath132 . this introduces a small error ( of the order @xmath133 ) that is not relevant in this proof - of - principle calculation . table [ tabres ] shows the pa - eom ccsd energies for the low - lying states in @xmath134 . the first two states ( @xmath135 ) are bound whereas the second excited state ( @xmath136 ) is resonant . in the computation of these states , we start the hf calculations in a single - particle basis that employs a mixed representation of harmonic oscillator states and berggren states . we include all harmonic oscillator shells such that @xmath137 and for a given @xmath138 state in @xmath134 , we only use berggren states for the partial wave @xmath139 that couples with the @xmath140 gs state in @xmath5 to the total angular momentum @xmath138 . for instance , for the @xmath141 ground state in @xmath142o we use harmonic oscillator states for all partial waves excepted for the neutron @xmath143 orbital . we have checked that the results remain unchanged when the berggren basis is used for multiple orbitals . the harmonic oscillator frequency is kept fixed at @xmath144 mev . energies are practically converged for @xmath145 at a precision of few kev for the ground state in @xmath146 and few tens of kev for the excited states . we note that due to the non - hermitian character of both the cc and the representation of the hamiltonian in the berggren basis , the dependence of the energy with the size of the model space is not necessarily monotonic . this can be seen , for instance , in the result for the ( complex ) energies of the @xmath136 resonance in @xmath147 . table [ tabres ] also shows the ccsd ground - state energy in @xmath148 . lcccc @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 + + 8 & -4.35 & -2.62 & 2.68-i0.32 & -121.68 + 10 & -4.49 & -2.73 & 2.24-i0.25 & -123.24 + 12 & -4.56 & -2.76 & 2.34-i0.21 & -123.49 + 14 & -4.57 & -2.80 & 2.26-i0.12 & -123.52 + the calculated ground state of @xmath154o at the ccsd level is underbound by about 4 mev compared to the experimental value at -127.62 mev , while ccsd with a perturbative triples correction gives a ground - state energy of @xmath155 mev @xcite . the ground - state of @xmath146 is found to be overbound by about 0.4 mev ( @xmath156 mev ) . the first excited state is underbound by about 0.5 mev ( @xmath157 mev ) , and the real part of the energy of the resonant @xmath136 state is about 1.3 mev above the experimental value @xmath158 mev . one can speculate whether higher order correlations such as @xmath159-@xmath64 excitations in the pa - eom approach , and the neglected three - nucleon forces will impact these low - lying states in @xmath142o . we also remind the reader that we have used @xmath132 in the pa - eom ccsd calculations of @xmath142o which introduce a small error in the computation of total binding energies . a more significant effect is seen if one looks at the energies of @xmath142o with respect to the ground - state of @xmath154o , shown in tab . [ tabres ] . in the pa - eom ccsd computations of @xmath142o , the energies are given with respect to the ground - state of @xmath154o , i.e. @xmath160 . using @xmath161 for @xmath142o the ground - state energy @xmath162 of @xmath154o is computed with the same mass @xmath163 , so in order to get the correct threshold one needs to add the energy shift @xmath164 where @xmath165 is the ground - state energy of @xmath154o with @xmath132 , this shift is about @xmath166 mev for the states show in in table [ tabres ] ( see e.g. @xcite for more details ) . we now illustrate the efficiency of the lanczos method to calculate the green s function matrix elements ( cf . [ seclanc ] ) by figure [ fig : pot_lanc ] shows the convergence of the real part of the radial ( diagonal @xmath167 ) @xmath168-wave optical potential as a function of the number of lanczos iterations @xmath114 . here , the single - particle basis is based on a model space with harmonic - oscillator shells up to @xmath169 and 50 discretized berggren @xmath170 shells . we show results in fig . [ fig : pot_lanc ] for @xmath171 mev . after about 10 lanczos iterations , the ( diagonal ) potential quickly converges except in the vicinity of the origin @xmath172 where the convergence is slower . however , close to the origin the @xmath168-wave scattering wavefunction @xmath173 , and the small dependence on @xmath114 will have a negligible impact on observables . -wave at @xmath171 mev as a function of @xmath174 the number of lanczos iteration . calculations were performed with @xmath175 and 50 discretized berggren shells for the neutron @xmath170 partial wave . ] results should be independent on the choice of the contour @xmath126 in the complex momentum plane as long as its discretization is adequate for the infrared scales under consideration @xcite . figure [ fig : pot_real_dif_cont ] shows the real part of the radial ( diagonal ) neutron @xmath168-wave potential at @xmath176 and @xmath171 mev using two contours @xmath177 and @xmath178 . both contours are defined by two segments @xmath179 $ ] and @xmath180 $ ] located on the fourth quadrant of the complex momentum plane where @xmath181 is taken as the origin . for the contour @xmath182 , the segment @xmath179 $ ] has a norm of @xmath183 , with an argument equal to @xmath184 and @xmath185 $ ] is a horizontal segment with @xmath186 . for the contour @xmath187 , the segment @xmath179 $ ] has a norm of @xmath188 and an angle equal to @xmath189 and @xmath185 $ ] is a horizontal segment with @xmath186 . we take 10 and 50 points on each segments for @xmath182 , whereas we take 5 and 45 points for the discretization of @xmath187 , respectively . figure [ fig : pot_real_dif_cont ] shows that the results are practically independent of the choice of the contour . -wave at @xmath190 and @xmath171 mev using two different contours @xmath191 for the single - particle neutron @xmath170 shells , in a model space with @xmath175 . ] + in fig . [ fig : gf ] , we illustrate the numerical stability of our approach as @xmath12 . we show the imaginary part of the ( diagonal ) @xmath168-wave green s function @xmath192 using ( i ) a complex and ( ii ) real set of hf orbitals for the @xmath170 neutron shells . while the shown results are for @xmath193 fm , we note that the qualitative behavior is independent of the value of @xmath194 . as expected ( see sec . [ berg ] ) , for @xmath124 values significantly larger than zero , both bases give the same results . let us first consider the real hf basis , corresponding to the dashed lines in fig . [ fig : gf ] . for @xmath195 mev the results are smooth but , as @xmath124 decreases , the considerable oscillations appear , and for @xmath196 peaks with widths proportional to @xmath124 start to appear near the green s function poles , at real energies . if instead we use a complex single - particle basis ( solid lines in fig . [ fig : gf ] ) no such instability occurs as @xmath12 . at @xmath193 fm ( see text for details ) . the dashed lines correspond to calculations with a real hf basis whereas the full lines were obtained using a complex contour . calculations were performed with @xmath197 and 50 discretized shells in the @xmath198 partial wave . ] next , we show in fig . [ fig : pot_conv ] , the convergence of the real part of the ( diagonal ) the @xmath168-wave optical potential as the size of the model space increases from @xmath199 to @xmath200 . results are shown for @xmath171 mev and , in all cases , 50 discretized shells are used for the berggren basis in the @xmath168-wave , and @xmath201 . for @xmath202 and @xmath171 mev , the results agree with those shown in fig . [ fig : pot_real_dif_cont ] . convergence is achieved for @xmath145 for @xmath203 fm . for small values of @xmath194 , the optical potential depends on @xmath204 . this is understandable because short - range physics gets more resolved as the model space increases . , and thus convergence becomes harder . again we note that in this region the scattering wave function @xmath205 and the dependence of the potential on @xmath149 does not impact observables . to demonstrate this point , fig . [ fig : pot_equi ] shows the integrated quantity @xmath206 -wave at @xmath171 mev . results are shown for at @xmath207 and 50 discretized shells in the @xmath170 partial wave ] in the neutron s - wave at e=10 mev . results are shown for at @xmath207 . for illustration purpose , we also show the results obtained with the phenomenological potential from ref . @xcite . ] the potential @xmath208 can be viewed as the local equivalent potential @xmath209 multiplied by the scattering wave function , and corresponds to the source term in the one - body optical - model type schodinger equation . the variations of the optical potential with the model space for small values of @xmath194 do not impact the behavior of @xmath208 . for illustration , [ fig : pot_equi ] also shows a result for @xmath210 obtained using a phenomenological potential based on a woods - saxon form factor @xcite . so far , we have only presented results for the diagonal part of the optical potential . figure [ fig : s_contour ] shows a contour plot for the nonlocal neutron @xmath168-wave optical potential . introducing the relative coordinate @xmath211 and the center - of - mass coordinate @xmath212 we plot the optical potential as a function of @xmath213 at fixed @xmath214 fm in fig . [ fig : counter_diag_s_10_mev ] . we can see that the full width at half maximum is about 2.2 fm . clearly , this potential is very different from a model of a dirac @xmath215 function in @xmath213 and examplifies the degree of nonlocality which is predicted microscopically . we note that due to the non - hermitian nature of the coupled - cluster method , the potential @xmath216 is slightly non - symmetric in @xmath194 and @xmath217 , and as a consequence @xmath218 is not quite an even function of @xmath213 . in figs . [ fig : s_contour ] , and [ fig : counter_diag_s_10_mev ] the energy is @xmath171 mev and results were obtained for @xmath219 and 50 discretized shells for the @xmath168-wave along a contour in the complex plane . for @xmath219 and 50 discretized @xmath168-wave shells at @xmath171 mev . ] -wave optical potential at e=10 mev plotted as @xmath220 at fixed @xmath214 fm . here @xmath221 and 50 discretized @xmath168-wave shells are included in the single - particle basis . ] calculations of the optical potential in other partial waves follow along the same lines . for illustration , we show a contour plot of the @xmath222-wave potential in fig . [ fig : d_contour ] . results are shown for @xmath171 mev at @xmath219 and 50 discretized shells for the @xmath223-wave along a complex contour . as in other cases , we take the limiting value @xmath224 . -wave potential for @xmath225 at @xmath171 mev . ] + -wave at @xmath171 mev . calculations were performed at @xmath226 with 50 @xmath170 discretized shells . results are shown for several values of @xmath124 . ] we finally turn to the imaginary part of the optical potential . the imaginary part describes the loss of flux due to inelastic processes . for most nuclei , and particularly for heavier systems , there are many compound - nucleus resonances above the particle threshold , and absorption is known to be significant . our results for the imaginary part of the potential , along the diagonal @xmath167 are shown in fig . [ fig : pot_imag ] for the neutron @xmath227 wave at @xmath171 mev . the model space consists of @xmath202 and 50 discretized shells for the s - wave . we consider various values of @xmath124 . in the limit @xmath201 , the imaginary part of the potential is very small , and this is true for the whole range of energies up to @xmath171 mev . as one can see in fig . [ fig : pot_imag ] , as @xmath124 decreases to zero , the imaginary part also decreases and becomes very small for @xmath201 . we observed the same qualitative behavior for all other considered partial wave , up to @xmath143 , a result that does not change when the model space increases . to further illustrate our difficulties with the imaginary part , we plot in fig . [ fig : jw ] the imaginary volume integral @xmath228 @xmath229 for the optical potential in the @xmath168-wave , taking a model space with @xmath145 and 50 discretized s - waves . -wave imaginary volume integral @xmath230 for several values of @xmath124 . calculations were performed at @xmath145 with 50 discretized @xmath170 shells . ] in order to understand these results , we recall that the compound states that contribute to the flux removal from the elastic channel consist of a high number of particle - hole excitations and are usually described by stochastic approaches @xcite . however , the coupled - cluster approach to the optical potential presented in this paper employs only @xmath61-@xmath62 and @xmath63-@xmath64 excitations and is thus limited to absorption on resonant states that are dominated by @xmath61-@xmath62 excitations . in our example of scattering off @xmath154o , its @xmath231 state ( at about 6 mev of excitation energy ) is thought to be of @xmath61-@xmath62 structure . with the nnlo@xmath232 interaction , we computed this state using eom - ccsd and found it at about 10 mev of excitation . another relevant excited state in @xmath154o is the first excited @xmath233 state also at @xmath234 mev , which is known to have a strong @xmath235 configuration . in our coupled cluster calculations this state is above @xmath236 mev . in fact , there are no other excited states below 10 mev . in general , positive parity states of @xmath154o are dominated by @xmath63-@xmath64 excitations , and are therefore not well described in eom - ccsd . thus , from this analysis , we conclude that it is not possible to produce significant absorption at low - energies for neutron scattering on @xmath154o due to the employed low - order cluster truncations in our eom - ccsd and pa / pr - eom ccsd approximations . one path forward is to introduce a phenomenological and energy dependent width in the green s function , to account for higher - order correlations such as @xmath159-@xmath64 and @xmath63-@xmath237 not included in pa / pr - eom ccsd @xcite . as shown in fig.[fig : pot_imag ] , this will increase the absorption at lower energies . finally we show , in fig . [ fig : phase_shift ] , the neutron elastic scattering phase shift obtained with the optical potential in the @xmath168 and @xmath238 partial waves , as a function of the model space . in the relative coordinate , with the reduced mass @xmath239 of the @xmath240o system . however , with the optical potential being calculated in the laboratory frame ( the hamiltonian @xmath18 ( [ hami ] ) is defined in the laboratory ) a correction to the reduced mass is needed . this correction is such that the reduced mass @xmath241 used to solve the schrdinger eq . ( [ schro ] ) is @xmath242 ( cf eq . [ hami ] ) . doing so , the bound states of the optical potential in the @xmath143 and @xmath170 partial waves correspond to respectively , the gs and first excited state in @xmath142o obtained with the pa - eom ccsd method . ] we find that , for @xmath145 all calculated phase shifts have converged ( all calculations here are done with 50 discretized shells ) . the sharp rise of the phase shift in the @xmath223 partial wave is the standard signature of the resonance @xmath136 in @xmath142o , which is numerically predicted to be at @xmath243 mev from our pa - eom ccsd calculations ( see table [ tabres ] ) . and @xmath238 wave as a function of @xmath204 . in all cases 50 discretized berggren shells are included . ] we constructed microscopic nuclear optical potentials combining the green s function approach with the coupled - cluster method . we calculated the green s function with the coupled - cluster method in an affordable approximation , and inverted the dyson equation to obtain the optical potential . to compute the green s function at arbitrary real energies we used an analytical continuation in the complex energy plane , based on a berggren basis . using the lanczos method , we wrote the green s function as a continued fraction , and reached convergence as the number of lanczos iterations increases . in the coupled - cluster singles and doubles approximation , the optical potential and the neutron elastic scattering phase shifts on @xmath0 converge well with respect to the size of the single - particle basis , for the low partial waves . the predicted optical potential has a strong nonlocality , that is not gaussian . in addition , we found an almost vanishing imaginary part of the potential for scattering energies below 10 mev . this lack of an absorptive component was attributed to neglected higher - order correlations in the employed coupled - cluster methods . in the future , we plan to update the nn force currently used , to one that is able to reproduce charge radii of heavier systems . we also plan to include three - nucleon forces in the coupled - cluster calculations of the green s functions , as well as higher - order correlations in the employed coupled - cluster methods . we expect this will produce an increase in the imaginary part of the derived optical potential . once these improvements are in place , this work can be extended to other systems , and to other reaction channel such as transfer , capture , breakup and charge - exchange . systematic studies involving heavier nuclei and consistent calculations along isotopic chains will provide critical information on how to extrapolate the optical potential to unknown regions of the nuclear chart . we acknowledge beneficial discussions with carlo barbieri , willem dickhoff , charlotte elster and grgory potel . this work was supported by the office of nuclear physics , u.s . department of energy under contracts de - fg02 - 96er40963 , de - fg52- 08na28552 ( ribss center ) and de - sc0008499 ( nuclei scidac collaboration ) , and the field work proposal erkbp57 at oak ridge national laboratory ( ornl ) . we also acknowledge the support of the national science foundation under grants no . phy-1520929 and phy-1403906 . computer time was provided by the institute for cyber- enabled research at michigan state university and the innovative and novel computational impact on theory and experiment ( incite ) program . this research used resources of the oak ridge leadership computing facility located at ornl , which is supported by the office of science of the department of energy under contract no . de - ac05 - 00or22725 . 53ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ ( , ) link:\doibase 10.1103/physrevlett.99.022502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.092501 [ * * , ( ) ] link:\doibase 10.1103/physrevc.86.021602 [ * * , ( ) ] link:\doibase 10.1103/physrevc.88.054622 [ * * , ( ) ] link:\doibase 10.1038/nature16067 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1103/physrevlett.108.042503 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.112301 [ * * , ( ) ] http://stacks.iop.org/0954-3899/34/i=12/a=r02 [ * * , ( ) ] link:\doibase 10.1103/physrevc.90.064619 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.232502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.192503 [ * * , ( ) ] http://stacks.iop.org/0954-3899/41/i=12/a=123002 [ * * , ( ) ] link:\doibase 10.1088/0031 - 8949/91/5/053002 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevc.89.034609 [ * * , ( ) ] link:\doibase 10.1103/physrevc.92.044607 [ * * , ( ) ] link:\doibase 10.1103/physrevc.93.014604 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/j.ppnp.2004.02.038 [ * * , ( ) ] link:\doibase 10.1103/physrevc.72.014613 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/0370 - 1573(78)90081 - 9 [ * * , ( ) ] link:\doibase 10.1088/0034 - 4885/77/9/096302 [ * * , ( ) ] link:\doibase 10.1016/0375 - 9474(68)90593 - 9 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.89.042502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.89.042501 [ * * , ( ) ] http://stacks.iop.org/0305-4470/37/i=38/a=006 [ * * , ( ) ] link:\doibase 10.1103/physrevc.73.034321 [ * * , ( ) ] http://hdl.handle.net/1903/786 [ * * ( ) ] link:\doibase 10.1103/revmodphys.66.763 [ * * , ( ) ] link:\doibase 10.1103/physrevb.52.r9827 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.4944960 [ * * , ( ) , 10.1063/1.4944960 ] @noop _ _ ( , ) link:\doibase 10.1103/revmodphys.79.291 [ * * , ( ) ] link:\doibase 10.1103/physrevc.74.024310 [ * * , ( ) ] link:\doibase 10.1143/ptp.113.1273 [ * * , ( ) ] , link:\doibase 10.1103/physrevc.75.044602 [ * * , ( ) ] link:\doibase 10.1016/j.ppnp.2013.10.003 [ * * , ( ) ] link:\doibase 10.1103/physrevc.91.021001 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.165112 [ * * , ( ) ] http://adsabs.harvard.edu/abs/2016arxiv160608822d [ ( ) ] , link:\doibase 10.1103/physrevlett.110.192502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.182501 [ * * , ( ) ] link:\doibase 10.1103/physrevc.86.031301 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.2845 [ * * , ( ) ]
we develop a method to construct a microscopic optical potential from chiral interactions for nucleon - nucleus scattering at arbitrary energies . the optical potential is constructed by combining the green s function approach with the coupled - cluster method . to deal with the poles of the green s function along the real energy axis , and to obtain the optical potential at arbitrary energies , we employ a berggren basis in the complex energy plane combined with the lanczos method . using this approach , we perform a proof - of - principle calculation of the optical potential for the elastic neutron scattering on @xmath0 . for the computation of the ground - state of @xmath0 , we use the coupled - cluster method in the singles - and - doubles approximation , while for the @xmath1 nuclei we use particle - attached / removed equation - of - motion methods truncated at two - particle - one - hole and one - particle - two - hole excitations , respectively . we verify the convergence of the optical potential and scattering phase shifts with respect to the model - space size and the number of discretized complex continuum states . we also investigate the absorptive component of the optical potential ( which reflects the opening of inelastic channels ) by computing its imaginary volume integral and find an almost negligible absorptive component at low - energies . to shed light on this result , we computed excited states of @xmath0 using equation - of - motion coupled - cluster with singles - and - doubles excitations and we found no low - lying excited states below 10 mev . furthermore , most excited states have a dominant two - particle - two - hole component , making higher order particle - hole excitations necessary to achieve a precise description of these core - excited states . we conclude that the reduced absorption at low - energies can be attributed to a lack of correlations coming from the low - order cluster truncation in the employed coupled - cluster method .
the main propose of ultrarelativistic heavy ion collisions is to probe the thermal information of the dense thermalized matter which is named quark - gluon plasma ( qgp).the electromagnetic radiation emitted from the center of the collision is a kind of very clean information due to the real and virtual photons which do not interact strongly , therefore photons and dileptons can escape from the dense medium without strong interactions , and may carry the information of the center dynamics@xcite . however , so far no evident experiments show that some information is exactly produced from the thermalized medium@xcite . the @xmath0 meson couples to the @xmath1 channel strongly , and the lifetime of @xmath0 meson is much shorter than the one of the expected hot hadronic gas . the yield of the @xmath0 mess spectrum may be modified in the hot medium due to the chiral symmetry restoration . the scenario of mass dropping or melting in a hot medium successfully interprets the dilepton yield enhancement in the low mass region at super proton synchrotron ( sps ) for pb - au 158 gev / a collisions@xcite . recently , the measurement of the dilepton continuum at relativistic heavy ion collider ( rhic ) energies has been performed by the phenix experiments for au - au 200 gev / a collisions@xcite . the dilepton yield in the low mass range between 0.2 and 0.8 gev is enhanced by a factor of 2@xmath23 compared with the expectation from hadron decays . however , such a modifying scenario can not well explain the enhancement in the au - au collisions at rhic . the imperfect modifying models of hadron decays present other probable mechanisms to explain the enhancement of the dilepton yield at low mass . moreover , the dilepton enhancement at rhic is implied that such phenomena are related strongly to the hot plasma scenario . the contribution of thermal dileptons at the low mass is covered by the cocktail of hadron decays because the vector meson peaks is more pronounced than the thermal spectrum . the thermal information is dominant in the intermediate mass region between the @xmath3 and @xmath4 vector meson for the phase transition theory , but the contribution of dileptons in this region also can be explained by the decays of charmed mesons . the na60 collaboration has also observed an enhancement at intermediate mass . the data suggest that such an enhancement may include a thermal information and not just charm decays@xcite . except for the above problems , a new puzzle was performed by phenix collaboration @xcite . the @xmath5 spectra of dileptons in @xmath6 collisions for different mass bins are in agreement with the expectation of the cocktail , charm decays and direct contributions . the agreement also exists in the au - au collisions at high @xmath5 ( @xmath7gev ) . the excess contribution above the cocktail and charm decays is consistent with the contribution from direct dileptons . in the soft region of @xmath8gev , pqcd is out of use , and the spectra are still higher than the expectation of cocktail and charm decays . if fitted with exponential in this soft window , one has a slope @xmath9mev , which is almost twice smaller than the typical slope of the standard thermal component . another candidate in the soft window is the new `` cold '' component which is still valid in the low @xmath5 region . in ref @xcite the authors suggest a new cold dilepton production mechanism which is not discussed in the standard theory list before . the semi - coherent two - photon processes are the so - called dilepton source for small mass bins and low @xmath5 . shuryak e. et . al . use the equivalent photon approximation to determine the differential cross section for the @xmath10 processes in au - au collisions , and have concluded that the semi - coherent production of dileptons does not contribute significantly to the phenix data@xcite . however , we derive the yield of dileptons for @xmath10 processes with the restriction @xmath11 , and finally find that the production of double photon interaction has a positive contribution to the cocktail and charm decays in the soft region . one notes that the impact parameter @xmath12 is smaller than the transverse dimension of the system @xmath13 , and that the dense hot medium can not be created in the heavy ion collisions if @xmath14 because the qgp is centrality dependent . the equivalent photon spectrum corresponding to a point charge @xmath15 , moving with a velocity @xmath16 is given by @xmath17,\end{aligned}\ ] ] where the argument of the modified bessel functions is @xmath18 , and @xmath19 corresponds to the radius of the radiation system with the maximum energy of the photons . after the macdonald approximation the above formulation can be written into a compact form as@xcite @xmath20 where @xmath21 and @xmath22 for au - au 200gev / a collisions at rhic . in this article we use the natural units , namely @xmath23 . the differential cross section of two - photon interaction for au - au collisions is given by @xmath24 where the total cross section of @xmath25 is@xcite @xmath26,\end{aligned}\ ] ] and @xmath27^{1/2}.\end{aligned}\ ] ] in the semi - coherent case , @xmath28 , the total transverse momentum of dileptons is @xmath29 , where @xmath30 is the transverse momentum of a photon . therefore the differential cross section can be written in the terms of dilepton transverse momentum as the following @xmath31 & & \times\int_{p_{2tmin}}^{\gamma / r_{min}}\mathrm{ln}\frac{\gamma}{q_{2t}r_{min}}\frac{1}{q_{2t}}dq_{2t},\end{aligned}\ ] ] where the minimum transverse momentum of photon @xmath32 is @xmath33gev due to the single track acceptance condition@xcite . if both the photon transverse momenta ( @xmath34 and @xmath35 ) are the same large ( non - coherent ) or small ( coherent ) , the total transverse momentum would have a very small value ( @xmath36 ) , then a dilepton could not obtain large transverse momentum in the @xmath37 interaction . the single track condition @xmath38gev allows us to discuss large transverse momentum dilepton production in the two - photon reaction , the contribution of non - coherent and coherent photon - photon processes is weak compared with the semi - coherent processes . this is the reason why the semi - coherent approach is essential in the two - photon interactions . for different mass bins . ( panel a)the solid line : the sum of cocktail and charm decay ; the dot line : the sum of cocktail and charm decay plus the contribution of direct dileptons [ 32 ] ; ( panel b)the contribution of @xmath39 , the dot line : 100mev@xmath40200mev ; the solid line : 200mev@xmath40300mev ; the dash line : 300mev@xmath40500mev ; the dash dot line : 500mev@xmath40750mev ; the dash dot dot line : 810mev@xmath40990mev . , width=9 ] in the semi - coherent case ( @xmath41 and @xmath42 ) , a non - coherent photon with large transverse momentum is radiated from a nucleus and a coherent photon with small transverse momentum is radiated from a proton of another nucleus in the relativistic heavy ion collisions . the large transverse momentum yield for different mass bins is given as follows @xmath43 where the yield relates to the differential cross section @xmath44 for nucleus - nucleus collisions with the total cross section @xmath45 in the form @xmath46 . the authors in ref@xcite use @xmath47 for the au - au 200gev / a collisions at rhic . from fig . 1 one can see that the contribution of two - photon processes is only valid in the region of @xmath483gev due to the radiation limit condition @xmath49 . it also implies that the spectra of @xmath10 interaction depend on the nucleus radiation energies . the energy of photons radiated from nuclei is localized in the range of relatively small @xmath30 . shuryak e. et . al . have discussed the two - photon processes by using a charge distribution form factor@xcite . a charge distribution is necessary to the form factor , but the woods - saxon approximation they used has depressed the value of the form factor , therefore the spectra of dileptons under @xmath39 interaction are also depressed quickly with the rising dilepton transverse momentum . in this article we derive the yield of dileptons for @xmath10 processes with the restriction @xmath50 , where the quark - gluon plasma may be created , and it is found that the spectra for different mass bins are essential in the soft region . at low mass , @xmath51200mev , the contribution of cocktail and charm decays is in agreement with the phenix data . however , in the soft region ( @xmath481gev ) the data are still higher than the expectation of cocktail and charm decays at @xmath52200mev . the yield enhancement is more evident with the increasing of dilepton mass . in order to avoid the influence of @xmath53 decays of narrow resonance vector mesons , the mass regions around the @xmath54 meson and @xmath3 meson are excluded by phenix collaboration . in fig . 2 we plot the large transverse momentum dilepton yield under @xmath39 interactions for different mass bins . as we discussed above the yield enhancement is not evident for the mass bins as 100mev@xmath40200mev and 200mev@xmath40300mev , so the correlation of cold @xmath10 component is weak ; in the mass region of 300mev@xmath40500mev , 500mev@xmath40750mev and 810mev@xmath40990mev the data are higher almost one order than the expectation of cocktail and charm decays , and the modification of @xmath10 reaction is essential now . therefore the two - photon interaction plays an important role in the large transverse momentum dilepton production in the soft @xmath5 region . the dot line is the contribution of the cocktail and charm decay , the dash line is the photon - photon production , and the solid line is the total contribution.,width=566 ] we derived the two - photon processes from the passage of @xmath39 . the large transverse momentum dilepton can be created in the semi - coherent two - photon interaction . the @xmath10 processes may be an essential complementarity to the previous standard cold dilepton production . in the macdonald approximation the energy of the radiated photon is limited in the region of @xmath55 , therefore the cold dilepton component of two - photon processes is still valid in the soft @xmath5 region where the pqcd is out of use . the numerical results are plotted by comparing with the phenix experimental data . the modification of two - photon processes is more evident with the rising dilepton mass , especially in the mass bins of 300mev@xmath40500mev , 500mev@xmath40750mev and 810mev@xmath40990mev .
the cold component of large transverse momentum dilepton production via semi - coherent two - photon interaction is calculated . the cold contribution is essential to the dilepton spectra in the soft region for different mass bins . the results are compared with the phenix experimental data at rhic , and we find that the modification of semi - coherent two - photon processes is more evident with the rising dilepton mass bins . pacs numbers:12.38.mh , 25.75.nq , 21.65.qr
the main goal of this paper is to study the representation theory of infinitesimal cherednik algebras @xmath2 , a deformation of the representation theory of @xmath1 with infinitely many deformation parameters @xmath4 . namely , @xmath1 can be represented as @xmath5 , where @xmath6 are the natural representations of @xmath0 on vectors and covectors . in this representation of @xmath1 , the elements of @xmath7 commute with each other , as do the elements of @xmath8 . the commutation relations of @xmath0 with @xmath6 are given by the usual action of matrices on vectors and covectors , while commutators of @xmath7 with @xmath8 produce elements of @xmath0 . to pass to the deformation @xmath2 , one needs to change only the last relation : commutators of @xmath7 and @xmath8 will now be not just elements of @xmath0 but rather some polynomial @xmath9 of them , where @xmath10 are the deformation parameters mentioned above and @xmath11 are basis polynomials introduced in @xcite . this deformation turns out to be very interesting , since it unifies the representation theory of @xmath1 with that of degenerate affine hecke algebras ( @xcite,@xcite ) and of symplectic reflection algebras ( @xcite ) . the main results of this paper are the following . in section [ section : shapform ] , we generalize a classical result from the representation theory of kac - moody algebras by computing the determinant of the contravariant ( or shapovalov ) form , thus determining when the verma module over @xmath2 is irreducible . this proof requires knowledge of the quadratic central element and its action on the verma module . in section [ section : gln ] , we find the quadratic central element of @xmath2 ; this extends the work of tikaradze @xcite , who proved using methods of homological algebra that the center of @xmath2 is a polynomial algebra in @xmath12 generators , but did not get any explicit formulas for these generators . in section [ section : finite ] , we provide a complete classification and character formulas for finite dimensional representations of @xmath2 , generalizing chmutova s unpublished work . in sections [ section : poisson ] to [ section : sp2n ] , we introduce poisson analogues of the infinitesimal cherednik algebras , compute their poisson center , and use them to give a second proof of the formula for the quadratic central element of @xmath2 . we also provide an analogous formula for the center of the poisson analogue of @xmath3 . finally , in section [ section : kostant ] , we investigate an analogue of kostant s theorem for @xmath3 . @xmath13 let us formally define the infinitesimal cherednik algebras of @xmath0 , which we denote by @xmath2 . let @xmath14 be the basic @xmath12-dimensional representation of @xmath0 and @xmath15 the dual representation . for any @xmath0 invariant pairing @xmath16 , define an algebra @xmath2 as the quotient of the semi - direct product algebra @xmath17 by the relations @xmath18={\zeta}(y , x)$ ] and @xmath19=[y , y']=0 $ ] for all @xmath20 and @xmath21 . let us introduce an algebra filtration on @xmath2 by setting @xmath22 for @xmath23 , @xmath24 , and @xmath25 for @xmath26 . we say that @xmath2 satisfies the pbw property if the natural surjective map @xmath27 is an isomorphism , where @xmath28 denotes the symmetric algebra ; we call these @xmath2 the _ infinitesimal cherednik algebras _ of @xmath0 . in @xcite , theorem 4.2 , it was shown that the pairings @xmath29 such that @xmath2 satisfy the pbw property are given by @xmath30 where @xmath31 and @xmath32 is the symmetrization of the coefficient of @xmath33 in the expansion of @xmath34 . note that for @xmath35 with @xmath36 , there is an isomorphism @xmath37 given by @xmath38 for @xmath39 , @xmath40 , @xmath41 , and @xmath42 this isomorphism allows us to view @xmath2 for general @xmath29 as an interesting deformation of @xmath43 , even though any formal deformation of @xmath43 is trivial . [ example : gl1 ] the infinitesimal cherednik algebras of @xmath44 are generated by elements @xmath45 , @xmath46 , and @xmath47 , satisfying the relations @xmath48=e$ ] , @xmath49=-f$ ] , and @xmath50=\phi(h)$ ] for some polynomial @xmath51 . in literature , these algebras are known as generalized weyl algebras ( @xcite ) . similarly to the representation theory of @xmath1 , we define the verma module of @xmath2 as @xmath52 where the set of positive root elements @xmath53 is spanned by the positive root elements of @xmath0 ( i.e. , matrix units @xmath54 with @xmath55 ) and elements of @xmath7 ; the set of negative root elements @xmath56 is spanned by the negative root elements of @xmath0 ( i.e. , matrix units @xmath54 with @xmath57 ) and elements of @xmath8 ; and the cartan subalgebra @xmath58 is spanned by diagonal matrices . the highest weight , @xmath59 , is an element of @xmath60 , and @xmath61 is the corresponding highest - weight vector . let us denote the set of positive roots by @xmath62 , so that @xmath63 for @xmath64 , @xmath65 . to denote the positive roots of @xmath0 , we use @xmath66 , and to denote the weights of @xmath67 , we use @xmath68 . we define @xmath69 , a _ quasiroot _ to be an integral multiple of an element in @xmath70 , and @xmath71 to be the set of linear combinations of positive roots with nonnegative integer coefficients . finally , @xmath72 denotes the @xmath73 weight - space of @xmath74 , where @xmath75 . as in the classical representation theory of lie algebras , the shapovalov form can be used to investigate the basic structure of verma modules . similarly to the classical case , @xmath76 possesses a maximal proper submodule @xmath77 and has a unique irreducible quotient @xmath78 . define the harish - chandra projection @xmath79 with respect to the decomposition @xmath80 , and let @xmath81 be the anti - involution that takes @xmath67 to @xmath82 and @xmath54 to @xmath83 . the _ shapovalov form _ $ ] is a bilinear form given by @xmath85 . the bilinear form @xmath86 on the verma module @xmath76 is defined by @xmath87 , for @xmath88 . this definition is motivated by the following two properties ( compare with @xcite ) : [ proposition : shap ] @xmath89 for @xmath90 , + @xmath91 . statement 1 of proposition [ proposition : shap ] reduces @xmath28 to its restriction to @xmath92 , which we will denote as @xmath93 . statement 2 of proposition [ proposition : shap ] gives a necessary and sufficient condition for the verma module @xmath76 to be irreducible , namely that for any @xmath75 , the bilinear form @xmath94 is nondegenerate , or equivalently , that @xmath95 , where the determinant is computed in any basis ; note that this condition is independent of basis . for convenience , we choose the basis @xmath96 , where @xmath97 runs over all partitions of @xmath98 into a sum of positive roots and @xmath99 with @xmath100 of weight @xmath101 . we will use the notation @xmath102 to mean that @xmath103 is a partition of @xmath104 into a sum of @xmath12 nonnegative integers when @xmath105 , and @xmath106 to mean that @xmath97 is a partition of @xmath98 into a sum of elements of @xmath70 when @xmath75 . then , the basis we will work with is @xmath107 . now , we present a formula for the determinant of the shapovalov form for @xmath2 generalizing the classical result presented in @xcite . this formula uses the following result proven in section [ subsection : actiongln ] : for a deformation @xmath108 , the central element @xmath109 ( introduced in section [ section : gln ] ) acts on the verma module @xmath76 by a constant @xmath110 , where @xmath111 are the complete symmetric functions ( we take @xmath112 ) and @xmath113 are linearly independent linear functions on @xmath114 . define the kostant partition function @xmath115 as @xmath116 . then : [ theorem : shapovalov ] up to a nonzero constant factor , the shapovalov determinant computed in the basis @xmath107 is given by @xmath117 in the case @xmath35 with @xmath36 , we get the classical formula from @xcite . the proof of this theorem is quite similar to the classical case with a few technical details and differences that will be explained below . we begin with the following lemma , which shows that irreducible factors of @xmath118 must divide @xmath119 for some @xmath120 . [ lemma : shap1 ] suppose @xmath121 . then , there exists @xmath122 such that @xmath123 . note that @xmath124 implies that the verma module @xmath76 has a critical vector ( a vector on which all elements of @xmath53 act by 0 ) of weight @xmath125 for some @xmath120 satisfying @xmath126 . thus , @xmath127 is embedded in @xmath76 . since @xmath109 acts by constants on both @xmath76 and @xmath127 , which can be considered as a submodule of @xmath76 , we get @xmath128 . the top term of the shapovalov determinant @xmath129 in the basis @xmath107 comes from the product of diagonal elements , that is , @xmath130^{\mathbf{m}_{\alpha}}(\lambda)$ ] . the top term of @xmath131(\lambda)$ ] for @xmath55 is @xmath132 where @xmath133 is the weight of @xmath54 . the following lemma gives the top term of @xmath134(\lambda)$ ] : the highest term of @xmath134(\lambda)$ ] for @xmath135 is @xmath136 , where the sum is over all partitions @xmath137 of @xmath138 into @xmath12 summands . from @xcite , theorem 4.2 , we know that the top term of @xmath134 $ ] for @xmath139 is given by the coefficient of @xmath140 in @xmath141 . because the set of diagonalizable matrices is dense in @xmath0 , we can assume @xmath142 is a diagonal matrix @xmath143 so that + @xmath144 and @xmath145 . multiplying these series gives the statement in the lemma . thus , we see that the top term of the determinant computed in the basis @xmath107 , up to a scalar multiple , is of the form @xmath146 since @xmath147 is the number of partitions of a weight @xmath148 , the sum @xmath149 over all partitions @xmath97 of @xmath98 with @xmath133 fixed must equal @xmath150 , so the expression above simplifies to @xmath151 this highest term comes from the product of the highest terms of factors of @xmath119 for various @xmath120 . [ lemma : irreducible ] for all @xmath152 , @xmath153 , @xmath119 is irreducible as a polynomial in @xmath59 . + for @xmath154 , @xmath153 , @xmath155 is irreducible . if lemma [ lemma : irreducible ] is true , then all @xmath148 contributing to the above product must be quasiroots : if @xmath152 for some @xmath156 , the highest term of the irreducible polynomial @xmath119 , @xmath157 , does not match any factor in the highest term of the shapovalov determinant unless @xmath148 is a @xmath7-quasiroot . moreover , if @xmath158 for @xmath156 , since @xmath155 is irreducible for @xmath153 , comparison with the highest term of the determinant shows that only the linear factor @xmath159 of @xmath160 appears in the shapovalov determinant . we will prove that @xmath119 is irreducible for @xmath152 ( @xmath153 ) ; similar arguments will show that @xmath155 is irreducible for any @xmath153 , @xmath161 . consider the parameters @xmath162 as formal variables . then , we have @xmath163 . we can absorb the @xmath164 vector into the @xmath59 vector . for this polynomial to be reducible in @xmath162 and @xmath165 , the coefficient of @xmath166 should be zero : @xmath167 . also , since the coefficient of @xmath168 is linear in @xmath165 , it must divide the coefficients of every other @xmath162 . in particular , the highest term of @xmath169 must divide that of @xmath170 . the highest term of @xmath169 is @xmath171 and the highest term of @xmath170 is given by @xmath172 , the evaluation of the gradient @xmath173 at @xmath148 . since this term is quadratic and is divisible by @xmath174 , we can write @xmath175 for some @xmath176 . now , let us match coefficients of @xmath177 for @xmath178 and of @xmath179 on both sides of the equation . by doing so ( and using the fact that @xmath180 ) , we obtain @xmath181 and @xmath182 . since @xmath183 and @xmath184 , at least two of @xmath185 are nonzero , say @xmath186 and @xmath187 . from the two equations , we obtain @xmath188 . if @xmath189 , then by similar arguments , @xmath190 , which is impossible since @xmath191 . thus , @xmath119 is reducible only if exactly two of the @xmath185 are nonzero and opposite to each other ; that is , @xmath158 for @xmath153 . to prove that the power of each factor in the determinant formula of theorem [ theorem : shapovalov ] is correct , we use an argument involving the jantzen filtration , which we define as in @xcite , page 101 ( for our purposes , we switch @xmath192 to @xmath2 ) . the jantzen filtration is a technique to track the order of zero of a bilinear form s determinant . instead of working over the complex numbers , we consider the ring of localized polynomials @xmath193 , q(0 ) \neq 0\}$ ] . a word - to - word generalization of @xcite , lemma 3.3 , proves that the power of @xmath160 for @xmath194 and of @xmath159 for @xmath153 is given by @xmath195 , completing the proof of theorem [ theorem : shapovalov ] . @xmath13 let @xmath196 ( which can be identified as elements of @xmath197 under the trace - map ) be defined by the power series @xmath198 , and let @xmath199 be the image of @xmath200 under the symmetrization map from @xmath197 to @xmath201 . the center of @xmath201 is a polynomial algebra generated by these @xmath199 . define @xmath202 $ ] . according to @xcite , theorems 2.1 and 1.1 , the center of @xmath203 is a polynomial algebra in @xmath204 , and there exist unique ( up to a constant ) @xmath205 such that the center of @xmath2 is a polynomial algebra in @xmath206 , @xmath207 . the _ casimir element _ of @xmath2 is defined ( up to a constant ) as @xmath109 . we will construct the casimir element of @xmath2 and prove that its action on the verma module @xmath76 is given by @xmath208 , where @xmath209 are linear functions in @xmath10 . let us switch to the approach elaborated in @xcite , section 4 , where all deformations satisfying the pbw property were determined . define @xmath210 with @xmath211 being a standard delta function at 0 , i.e. , @xmath212 . let @xmath213 be a polynomial satisfying @xmath214 , where @xmath215 is the generating series of the deformation parameters : @xmath216 . since @xmath213 is defined up to a constant , we can specify @xmath217 . recall from @xcite , section 4.2 , that for @xmath218 , @xmath219=\frac{1}{2 \pi^n}\int_{v\in \mathbb{c}^n : |v|=1 } ( x,(v\otimes \bar{v})y ) \int_{-\pi}^{\pi } \left(1-e^{-i \theta}\right)\hat{f}(\theta)e^{i\theta(v\otimes \bar{v } ) } \ , d\theta \ , dv.\ ] ] [ theorem : firstcentralelementtheorem ] let @xmath220 . the casimir element of @xmath2 is given by @xmath221 . define @xmath222 . let us compute @xmath223= \sum_j [ y , x_j ] y_j + [ y , c']$ ] . the first summand is : @xmath224 y_j & = & \frac{1}{2 \pi^n}\sum_j \int_{v \in \mathbb{c}^n : |v|=1 } \int_{-\pi}^{\pi } ( 1-e^{-i\theta})\hat{f}(\theta ) e^{i\theta ( v\otimes \bar{v } ) } ( x_j,(v\otimes \bar{v})y)y_j \ , d\theta \ , dv \\ & = & \frac{1}{2 \pi^n } \int_{|v|=1 } \int_{-\pi}^{\pi } ( 1-e^{-i\theta})\hat{f}(\theta ) e^{i\theta ( v\otimes \bar{v } ) } \otimes ( v\otimes \bar{v})y \ , d\theta \ , dv . \end{aligned}\ ] ] following @xcite , section 4.2 , we define @xmath225 . there , it was proven that @xmath226 thus , we can write @xmath227 which implies that @xmath228=\frac{1}{2 \pi^n } \int_{|v|=1 } \int_{-\pi}^{\pi } \hat{f}(\theta ) [ y , e^{i\theta ( v\otimes \bar{v } ) } ] \ , d\theta \ , dv $ ] . since @xmath229 = e^{-i\theta ( v\otimes \bar{v } ) } ye^{i\theta ( v\otimes \bar{v } ) } -y = e^{-i\theta \mathrm{ad}(v\otimes \bar{v } ) } y - y = ( e^{-i\theta}-1)(v\otimes \bar{v})y $ ] , we get @xmath228=\frac{1}{2 \pi^n } \int_{|v|=1 } \int_{-\pi}^{\pi } \hat{f}(\theta ) e^{i\theta ( v\otimes \bar{v})}(e^{-i\theta}-1)(v\otimes \bar{v})y \ , d\theta \ , dv $ ] , and so @xmath230 y_i + [ y , c ' ] = 0 $ ] as desired . by using the anti - involution @xmath231 defined in the beginning of section [ section : shapform ] , this implies @xmath232=0 $ ] for any @xmath23 , while @xmath233=0 $ ] by @xcite , and hence , @xmath234 . this proof resembles calculations in @xcite , section 4 . in particular , proposition 5.3 of @xcite provides a formula for the casimir element of continuous cherednik algebras . however , adopting this formula for the specific case of infinitesimal cherednik algebras is nontrivial and requires the above computations . in this section , we justify our claim that the action of the casimir element @xmath109 is given by @xmath110 . obviously , @xmath109 acts by a scalar on @xmath235 , which we will denote by @xmath236 . since @xmath237 , @xmath238 , we see that @xmath239 where @xmath240 denotes the constant by which @xmath241 acts on @xmath235 . [ theorem : casimirgln ] let @xmath242 be the unique degree @xmath243 polynomial satisfying @xmath244 . then @xmath245 because @xmath240 is a polynomial in @xmath59 , we can consider a finite - dimensional representation of @xmath201 instead of the verma module @xmath235 of @xmath2 . for a dominant weight @xmath246 ( so that the highest weight @xmath0-module @xmath247 is finite dimensional ) we define the normalized trace @xmath248 for any @xmath249 satisfying @xmath250 ( note that @xmath251 does not depend on @xmath249 ) . to compute @xmath251 , we will use the weyl character formula ( see @xcite ) : @xmath252 , where @xmath253 denotes the weyl group ( which is @xmath254 for @xmath0 ) . however , direct substitution of @xmath255 into this formula gives zero in the denominator , so instead we compute @xmath256 for a general diagonal matrix @xmath148 . without loss of generality , we may suppose @xmath257 , so that @xmath258 then @xmath259 partition @xmath260 into @xmath261 , where @xmath262 . for @xmath263 , @xmath264 so @xmath265 . next , we compute the numerator . we can divide @xmath266 , where @xmath267 . note that @xmath268 , where @xmath269 and @xmath270 denotes the subgroup of @xmath254 corresponding to permutations of @xmath271 . we can then write @xmath272 where @xmath273 and @xmath274 . combining the results of the last two paragraphs , we get @xmath275 using the weyl character formula again , we see that @xmath276 where @xmath277 is half the sum of all positive roots of @xmath278 . thus , @xmath279 we substitute to obtain @xmath280 our original goal was to calculate @xmath281 . we obtain @xmath282 using the dimension formula ( @xcite , equation 15.17 ) : @xmath283 we get @xmath284 . since @xmath285 , we have @xmath286 . thus , we get @xmath287 where @xmath288 . let @xmath289 . we verify that @xmath290 and it is easy to see that the polynomial solution to @xmath244 is unique . @xmath13 in this section , we investigate when the irreducible @xmath2 representation @xmath291 is finite dimensional . as in the case of classical lie algebras , any finite dimensional irreducible representation is isomorphic to @xmath291 for a unique weight @xmath59 . theorem [ theorem : classification ] provides a necessary and sufficient condition for @xmath291 to be finite dimensional . in particular , all such representations have a _ rectangular form_. in section [ subsection : finiteexistence ] , we prove that for any allowed _ rectangular form _ there exists a deformation @xmath29 such that the representation @xmath291 of @xmath2 has exactly that shape . [ theorem : classification ] the representation @xmath291 is finite dimensional if and only if @xmath59 is a dominant @xmath0 weight and there exists @xmath292 such that @xmath293 . for every @xmath294 let @xmath295 be the smallest nonnegative integer such that @xmath296 ( we set @xmath297 if no such nonnegative integer exists ) . we define parameters @xmath298 . if @xmath291 is finite dimensional , then as a @xmath0 module it decomposes into @xmath299 where @xmath300 are the parameters defined above ( depending on @xmath29 and @xmath59 ) . in order for @xmath291 to be finite dimensional , it is clearly necessary for @xmath59 to be a dominant @xmath0 weight . recalling the pbw property and the definition of the verma module @xmath76 , we see that as a @xmath0 module , @xmath76 decomposes as @xmath301 , where @xmath302 . we can further decompose each @xmath303 into irreducible modules of @xmath0 ; once we do so , we find that @xmath76 has a simple @xmath0 spectrum . note that @xmath304 can be decomposed as @xmath305 ( taking @xmath306 if @xmath307 is not dominant ) . we can thus associate each @xmath308 for @xmath309 in the decomposition of @xmath76 with a lattice point @xmath310 . we draw a directed edge from @xmath311 to @xmath312 if @xmath313 is in the decomposition of @xmath304 , and we say @xmath312 is _ smaller _ than @xmath311 . a key property of this graph is that any @xmath2-submodule of @xmath76 intersecting the module @xmath314 must necessarily contain @xmath314 and all @xmath313 such that @xmath312 is reachable from @xmath311 by a walk along directed edges . recall that @xmath315 , where @xmath316 is the maximal proper @xmath2-submodule of @xmath76 . the aforementioned property guarantees that as a @xmath0 module , @xmath317 for some set @xmath28 of vertices closed under walks , so that @xmath291 is finite dimensional if and only if @xmath318 ( the complement of @xmath28 ) is a finite set . we now prove part ( a ) . first , suppose that @xmath291 is finite dimensional . the finiteness of @xmath318 implies the existence of some @xmath319 such that @xmath320 ( note that @xmath321 ) . let @xmath322 be the minimal such @xmath319 . we define @xmath323 as the set of vertices that can be reached by walking from @xmath324 . because @xmath325 , the verma module @xmath76 must possess a submodule @xmath326 . by considering the action of the casimir element on @xmath76 and @xmath326 , we get @xmath327 . next , suppose that there exists @xmath292 such that @xmath293 . the determinant formula of theorem [ theorem : shapovalov ] implies that the verma module @xmath76 contains the submodule @xmath328 for some @xmath329 . define @xmath323 to be the set of vertices that can be reached by walking from @xmath330 . its complement @xmath331 is finite , since for any vertex @xmath332 of our graph , we have @xmath333 . because @xmath334 , @xmath318 is finite , finishing the proof of ( a ) . we note that explicitly , @xmath335 and the corresponding finite dimensional quotient is @xmath336 . part ( b ) requires an additional argument . namely , if @xmath291 is finite dimensional , then it can also be considered as a lowest weight representation . let @xmath337 be the vertex corresponding to the lowest weight of @xmath291 . if the statement of ( b ) was wrong , there would be a vertex @xmath338 with two nonzero coordinates , such that @xmath339 for any @xmath340 . without loss of generality , suppose @xmath341 . as we can walk along reverse edges from @xmath342 to both points @xmath343 and @xmath344 , we can also walk along reverse edges to @xmath345 , which is a contradiction . this proves part ( b ) and explains our terminology `` _ _ rectangular form _ _ '' . of @xmath346 , with the highest weight of each @xmath347 module indicated . , scaledwidth=58.0% ] the decomposition of @xmath291 as a @xmath0 module provides the character formula for @xmath291 as the sum of the characters of @xmath0 modules : @xmath348 as in the classical theory , this character allows us to calculate the decomposition of finite dimensional representations into irreducible ones . let us illustrate the decomposition of @xmath291 from the proof of theorem [ theorem : classification ] ; for clarity , we will work with @xmath349 representations instead of @xmath350 representations . using the notation of the proof , @xmath351 , the irreducible @xmath349 representation of dimension @xmath352 . by the clebsch - gordon formula , @xmath353 we can use the above formula to draw the graph , as in figure 2 , representing the decomposition of @xmath354 , with @xmath355 , into @xmath349 modules . this representation is the quotient of @xmath356 by the submodules represented by the shaded areas of the diagram , and @xmath357 as @xmath349 modules . , with @xmath355 , into @xmath349 modules.,scaledwidth=90.0% ] for @xmath358 , the irreducible finite dimensional representation @xmath291 , for @xmath359 , has character @xmath360 , where @xmath98 is some nonnegative integer . if we describe @xmath358 as in example [ example : gl1 ] , we can easily calculate the casimir element to be @xmath361 , where @xmath362 satisfies the equation @xmath363 . then , @xmath98 is the smallest nonnegative integer such that @xmath364 . for @xmath365 , the irreducible finite dimensional representations are necessarily of the form @xmath291 with @xmath366 , where @xmath367 . the character of @xmath291 equals @xmath368 let @xmath369 and @xmath370 . again , @xmath371 is defined as the minimal nonnegative integer satisfying @xmath372 , while @xmath373 is either @xmath138 or the minimal nonnegative integer satisfying @xmath374 . for instance , if @xmath375 with @xmath376 , then @xmath377 is a multiple of @xmath148 , and so the only solution to the equation @xmath372 is @xmath378 , which is negative . thus , @xmath379 has no finite dimensional irreducible representations . if @xmath380 with @xmath36 , @xmath381 , so @xmath382 . thus , @xmath291 is finite dimensional if and only if @xmath383 is a positive integer . this agrees with the description of finite dimensional representations of @xmath384 . for any @xmath0 dominant weight @xmath59 and @xmath385 such that @xmath386 for all @xmath294 , there exists a deformation @xmath29 , such that the irreducible representation @xmath291 of @xmath2 is finite dimensional and its character is given by . let @xmath387 . we can write @xmath388 for @xmath389 ( we have strict inequalities because of the shift by @xmath164 ) . recall that @xmath390 for @xmath162 defined as in theorem [ theorem : casimirgln ] . let @xmath391 . we will find @xmath162 such that @xmath392 , while for all @xmath393 , @xmath394 . this implies that there are embeddings of @xmath395 into @xmath396 with an irreducible quotient @xmath397 , due to theorem [ theorem : classification ] . define @xmath398 for @xmath399 with the @xmath243 at the @xmath400-th location . we must prove that there exist @xmath401 such that @xmath402 and @xmath403 for all @xmath404 . we can write @xmath405 , where @xmath406 note that the condition @xmath407 determines a hyperplane @xmath408 in the space @xmath409 ( @xmath408 might in fact be the entire space , but the following argument would be unaffected ) . hence , the intersection @xmath410 belongs to the union @xmath411 if and only if it belongs to some @xmath412 . thus , it suffices to show that @xmath413 are linearly independent as functions of @xmath162 for all @xmath414 and @xmath415 . this condition of linear independence is satisfied if @xmath416 now we shall prove that using column transformations , we can reduce the above matrix to its evaluation at @xmath417 . we proceed by induction on the column number . the elements of the first column , @xmath418 , are of degree zero with respect to @xmath419 , so @xmath420 . suppose that using column transformations , all columns before column @xmath421 are reduced to their constant terms . now , we note that @xmath422 thus , we see that @xmath423 is a linear combination of @xmath424 , the entries of the other columns : @xmath425 by selecting pivots of @xmath426 , we can eliminate every term except @xmath427 . by repeating this step , we reduce the matrix to its evaluation at @xmath417 : @xmath428 let us now rewrite @xmath429 : @xmath430 where @xmath431 and @xmath432 . the third equality is because @xmath433 . it is easy to see that the above determinant can be reduced further to @xmath434 where @xmath435 and the determinant is @xmath436 by the vandermonde determinant formula . now , recalling the conditions @xmath437 we get @xmath438 for any @xmath55 and so @xmath439 is nonzero . similarly , we get @xmath440 . hence , the determinant is nonzero , and so @xmath441 are linearly independent as desired . now we will study infinitesimal cherednik algebras by using their poisson analogues . the poisson infinitesimal cherednik algebras are as natural as @xmath2 , and their theory goes along the same lines with some simplifications . although these algebras have not been defined before in the literature , the authors of @xcite were aware of them , and technical calculations with these algebras are similar to those made in @xcite . this approach provides another proof of theorem [ theorem : firstcentralelementtheorem ] . let @xmath29 be a deformation parameter , @xmath442 . the poisson infinitesimal cherednik algebra @xmath443 is defined to be the algebra @xmath444 with a bracket defined on the generators by : @xmath445 \text { for } a , b \in { \mathfrak{gl}}_n,\\ \{g , v\ } & = g(v ) \text { for } g\in { \mathfrak{gl}}_n , v\in v\oplus v^*,\\ \{y , y'\}&=\{x , x'\ } = 0 \text { for } y , y ' \in v , x , x ' \in v^*,\\ \{y , x\ } & = { \zeta}(y , x ) \text { for } y\in v , x \in v^*.\end{aligned}\ ] ] this bracket extends to a poisson bracket on @xmath443 if and only if the jacobi identity @xmath446 holds for any @xmath447 . as can be verified by computations analogous to @xcite , theorem 4.2 , the jacobi identiy holds iff @xmath448 where @xmath31 and @xmath449 is the coefficient of @xmath33 in the expansion of @xmath450 . actually , we can consider the infinitesimal cherednik algebras of @xmath0 as quantizations of @xmath443 . [ remark : glnpoisson ] note that @xmath451 this follows from @xmath452 when @xmath453 . in fact , if @xmath454 , the jacobi identity implies that @xmath455 for some @xmath456 invariant function @xmath457 , and that @xmath458 , where @xmath459 is the matrix with @xmath460 . one can then show that the only @xmath456 invariant functions @xmath457 satisfying this partial differential equation are linear combinations of @xmath461 . our main goal is to compute explicitly the poisson center of the algebra @xmath443 . as before , we set @xmath462 to be the coefficient of @xmath463 in the expansion of @xmath464 , @xmath465 , and @xmath466 . [ theorem : poissoncentergln ] the poisson center @xmath467 $ ] , where @xmath468 is the coefficient of @xmath469 in the series @xmath470 first , we claim that @xmath471 $ ] . the inclusion @xmath472 \subseteq \operatorname{\mathfrak{z}_{\mathrm{pois}}}(h_0'({\mathfrak{gl}}_n))$ ] is straightforward , while the reverse inclusion follows from the structure of the coadjoint action of the lie group corresponding to @xmath473 ( as in the proof of @xcite , theorem 2 ) . we prove that the poisson center of @xmath474 can be lifted to the poisson center of @xmath443 by verifying that @xmath475 are indeed poisson central . since @xmath476 and @xmath477 , @xmath478 poisson - commutes with elements of @xmath197 . we can define an anti - involution on @xmath443 that acts on basis elements by taking @xmath54 to @xmath83 and @xmath67 to @xmath82 . by using the arguments explained in the proof of theorem 2 in @xcite , we can show that @xmath479 is fixed by this anti - involution , while @xmath480 is also fixed since it lies in @xmath481 . applying this anti - involution , we see that it suffices to show that @xmath480 satisfies @xmath482 for basis elements @xmath483 . first , notice that if @xmath484 , then @xmath485 , and together with the equation @xmath486 ( see the proof of lemma 2.1 in @xcite ) , we get @xmath487 thus , we have @xmath488 hence , @xmath482 is equivalent to the system of partial differential equations : @xmath489 multiplying both sides by @xmath463 and summing over @xmath490 we obtain an equivalent single equation @xmath491 since all terms above are @xmath456 invariant and diagonalizable matrices are dense in @xmath0 , we can set @xmath492 : @xmath493 and it is easy to see that @xmath494 satisfies the above equation . [ example : l1 ] in particular , @xmath495 . another way of writing the formula for @xmath480 is @xmath496 where @xmath497 and @xmath498 , the character of an irreducible @xmath0 module corresponding to a hook young diagram representations , @xmath499-[\bigwedge^{k+1 } v \otimes s^{m-1 } v]+\cdots+(-1)^m[\bigwedge^{k+m } v]=[v_{(m+1,1,\ldots,1)}]$ ] due to pieri s formula . ] . this provides a better insight for the quantization construction . we expect that for any @xmath500 , the induced symplectic structure on @xmath501 has only finitely many symplectic leaves . note that @xmath502 for @xmath484 and @xmath24 ; we can thus identify @xmath503 with the element @xmath504 . [ lemma : com2noncom ] @xmath505 = \left\{\sum_{j=0}^{k}\frac{(-1)^j}{k+n+1 } { k+n+1 \choose j+1}\operatorname{tr}s^{k+1-j } a , y \right\}.\ ] ] it is enough to consider the case @xmath506 . recall that @xmath507 can be written as a sum of degree @xmath352 monomials of form @xmath508 where @xmath509 and the sequence @xmath510 is a permutation of the sequence of @xmath511 ones , @xmath512 twos , and so forth ; for conciseness , we will denote the above monomial by @xmath513 . the only terms of @xmath514 that contribute to @xmath515 $ ] and to @xmath516 have @xmath517 . since to compute @xmath515 $ ] we first symmetrize @xmath518 , we will compute @xmath519-\{\mathrm{sym}(e_{1,i_1 } \cdots e_{n , i_{k+1}}),y_1\}$ ] . for both the lie bracket and the poisson bracket , we use leibniz s rule to compute the bracket , but whereas in the poisson case we can transfer the resulting elements of @xmath7 to the right since the poisson algebra is commutative , in the lie case when we do so extra terms appear . consider a typical term that may appear after we use leibniz s rule to compute @xmath515 $ ] : @xmath520 when we move @xmath521 to the right , we get , besides @xmath522 , additional residual terms like @xmath523 and @xmath524 , up to @xmath525 . without loss of generality , we can consider only the last expression , since the others will appear in the smaller chains @xmath526 and @xmath527 and so forth , with the same coefficients . for notational convenience , we let @xmath528 denote the coefficient of @xmath529 in the residual term , i.e. , the term represented by the ellipsis : @xmath530 . then , @xmath531 is a term in the expression @xmath532 , which appears in @xmath533 . thus , we can write @xmath534=\left\{\sum_{n=0}^{k } ( -1)^n c_n \operatorname{tr}s^{k+1-n}a , y_1 \right\}\ ] ] for some coefficients @xmath535 . next , we compute @xmath535 . we first count how many times @xmath531 appears in @xmath536 . notice that since @xmath528 is the product of @xmath537 @xmath538 s , we can insert @xmath539 in @xmath540 places to obtain @xmath541 such that @xmath542 contains @xmath531 . now we compute the coefficient of @xmath541 in @xmath543 . as noted before , @xmath544 can be written as a sum of degree @xmath545 monomials of form @xmath546 . any term that is a permutation of those @xmath545 unit matrices will appear in the symmetrization of @xmath543 . we count the number of sequences @xmath547 such that @xmath541 is the product of the elements @xmath548 ( in some order ) ; this tells us the multiplicity of @xmath541 in the symmetrization of @xmath543 . suppose @xmath549 for a certain sequence @xmath547 . then , @xmath550 if and only if @xmath551 is a permutation of @xmath552 for all @xmath400 . thus , @xmath541 appears @xmath553 times in @xmath543 . since each term has coefficient @xmath554 in the symmetrization , @xmath541 appears with coefficient @xmath555 in the symmetrization of @xmath543 . in conjunction with the previous paragraph , we see that @xmath531 appears @xmath556 times in @xmath536 . it remains to calculate how many times @xmath531 appears in @xmath557 $ ] . recall that @xmath528 is obtained from a term like : @xmath558 where the ordered union of the ellipsis equals @xmath528 . thus , @xmath528 comes from terms of the following form : we choose arbitrary numbers @xmath559 , and insert @xmath560 into @xmath528 . there are @xmath561 ways for this choice for any fixed @xmath559 . any such term @xmath562 appears in @xmath514 with coefficient @xmath563 where @xmath564 is the total number of @xmath565 s ( for some @xmath340 ) in @xmath562 , i.e. , @xmath566 , @xmath567 . combining the results of the last two paragraphs , we see that @xmath536 must appear with coefficient @xmath568 where the summation is over all length-@xmath569 sequences @xmath570 of integers from 1 to @xmath12 . we claim that @xmath571 to see this , notice that @xmath572 is the coefficient of @xmath573 in the expression @xmath574 the above generating function equals @xmath575 , and the coefficient of @xmath573 in this expression is @xmath576 . finally , we arrive at the simplified coefficient of @xmath536 : @xmath577 as desired . now we will give an alternative proof of theorem [ theorem : firstcentralelementtheorem ] . let @xmath213 be the polynomial satisfying @xmath578 and @xmath579 ( in the expression for @xmath580 , we discard any negative powers of @xmath581 ) . note that if @xmath582 , then @xmath583 lemma [ lemma : com2noncom ] allows us to write @xmath584&=\left\{\sum_{j=1}^{k+1 } \sum_{i=0}^{j-1 } \frac{1}{j+n}{j+n \choose i+1 } ( -1)^i g_j\operatorname{tr}s^{j - i } a , y\right\}\\ & = \left\ { \sum_{j=1}^{k+1 } \sum_{i=0}^{k - j+1 } \frac{1}{j+i+n}{j+i+n \choose i+1 } ( -1)^i g_{j+i}\operatorname{tr}s^{j } a , y\right\ } = \left\{\sum_{j=1}^{k+1 } { \zeta}_{j-1}\operatorname{tr}s^{j}a , y\right\}.\end{aligned}\ ] ] hence , @xmath585=\sum_{i=1}^n [ x_i , y ] y_i=\sum_{i=1}^n \{x_i , y\ } y_i=-\left\{\sum_{j=1}^{k+1 } { \zeta}_{j-1}\operatorname{tr}s^{j}a , y\right\}=-\left[\sum_{j=1}^{k+1 } g_{j } \operatorname{tr}s^{j}a , y\right],\end{aligned}\ ] ] where the third equality follows from the fact that @xmath586 is poisson - central in @xmath443 ( see example [ example : l1 ] ) . thus , we get @xmath234 , where @xmath587 comparing the formula for @xmath588 in example [ example : l1 ] to the one from theorem [ theorem : firstcentralelementtheorem ] , we see that they differ only by a change @xmath589 . we expect that a similar _ twist _ of the formula for @xmath494 given in theorem [ theorem : poissoncentergln ] will provide the formulas for the actions of the generators of @xmath590 on the verma module @xmath235 . let @xmath7 be the standard @xmath592-dimensional representation of @xmath593 with symplectic form @xmath594 , and let @xmath595 be an @xmath593 invariant bilinear form . the _ infinitesimal cherednik algebra _ @xmath3 is defined as the quotient of @xmath596 by the relation @xmath597={\zeta}(x , y)$ ] for all @xmath598 , such that @xmath3 satisfies the pbw property . in @xcite , theorem 4.2 , it was shown that @xmath3 satisfies the pbw property if and only if @xmath599 where @xmath32 is the symmetrization of the coefficient of @xmath600 in the expansion of @xmath601 note that for @xmath602 , the expansion of @xmath603 contains only even powers of @xmath581 . for @xmath376 , there is an isomorphism @xmath604 , where @xmath605 is the @xmath12-th weyl algebra ( see @xcite example 4.11 ) . thus , we can regard @xmath3 as a deformation of @xmath606 . choose a basis @xmath607 of @xmath7 , so that @xmath608 with @xmath609 as before , we study the noncommutative infinitesimal cherednik algebra @xmath3 by considering its poisson analogue @xmath591 . we define @xmath610 and @xmath611 where @xmath612 is dual to @xmath613 ( that is , @xmath614 ) . when viewed as an element of @xmath615 $ ] , @xmath616 so @xmath617 is @xmath593 invariant and independent of the choice of basis @xmath618 . the poisson center of @xmath619 is @xmath620 $ ] . we will follow a similar approach as in the proof of theorem 2.1 , @xcite . let @xmath621 be the lie algebra @xmath622 and @xmath28 be the lie group of @xmath621 . we need to verify that @xmath623 = \operatorname{\mathfrak{z}_{\mathrm{pois}}}(h_0'({\mathfrak{sp}}_{2n}))$ ] , the latter being identified with @xmath624^s$ ] . let @xmath625 be the @xmath592-dimensional subspace consisting of elements of the form @xmath626 where all the @xmath627 s belong to @xmath628 . in what follows , we identify @xmath629 and @xmath621 via the non - degenerate pairing , so that the coadjoint action of @xmath28 is on @xmath621 . we use the following two facts proved in @xcite : first , that the orbit of @xmath630 under the coadjoint action of @xmath28 on @xmath629 is dense in @xmath629 ; and second , that @xmath624^s \cong \mathbb{c}[f_1 , \ldots , f_n]$ ] , where @xmath631 and @xmath632 is the @xmath400-th elementary symmetric polynomial . it is straightforward to see that @xmath633 , and so @xmath624^s \cong \mathbb{c}[{\uptau}_1 , \ldots , { \uptau}_n]$ ] as desired . as before , let @xmath634 . [ theorem : sp2npoissoncenter ] the poisson center @xmath635 $ ] , where @xmath636 is the coefficient of @xmath637 in the series @xmath638 since @xmath639 , @xmath640 for any @xmath641 , and so it suffices to show that @xmath642 for all @xmath643 . by the jacobi rule , @xmath644 thus , @xmath645 in the case of @xmath443 , @xmath646 by straightforward application of properties of the determinant . however , for @xmath591 , @xmath647 . to calculate this sum , let @xmath648 be a basis of @xmath649 ( the basis elements are given in the appendix , but for the purposes of this section , the specific elements are not needed ) . write @xmath650 [ lemma : sp2nkey ] @xmath651 the proof of this lemma is quite technical and is provided in the appendix . using the fact that @xmath652 , we can restrict ( [ eqn : sp2npoisson ] ) to diagonal matrices , which are spanned by elements @xmath653 with 1 at the @xmath654-th coordinate . thus , the condition @xmath642 is equivalent to : @xmath655 multiplying the above equation by @xmath656 and summing over @xmath340 for @xmath657 , the required condition transforms into : @xmath658 it suffices to check this condition for basis vectors @xmath659 and @xmath660 . substituting , we get @xmath661 and @xmath662 these last two formulas both reduce to @xmath663 and it is straightforward to verify that @xmath494 satisfies the above equation . we now briefly consider the center of @xmath3 . let @xmath664 be the symmetrization of @xmath200 , and let @xmath665 v_j^*.\ ] ] clearly , @xmath666 is independent of the choice of basis @xmath667 and @xmath593 invariant . . ] [ conjecture : sp2ncenter ] the center of @xmath3 is @xmath668 $ ] for some @xmath669 . recall kostant s theorem in the classical case ( @xcite ) : let @xmath670 be a reductive lie algebra with an adjoint - type lie group @xmath671 , and let @xmath672 $ ] be the ideal generated by the homogeneous elements of @xmath673^g$ ] of positive degree . then : + @xmath192 is a free module over its center @xmath674 ; + the subscheme of @xmath670 defined by @xmath675 is a normal reduced irreducible subvariety that corresponds to the set of nilpotent elements in @xmath670 . in @xcite , kostant s theorem was generalized to @xmath2 . in this section , we provide a similar generalization for @xmath3 assuming conjecture [ conjecture : sp2ncenter ] : @xmath676 $ ] . as in section [ section : gln ] , we define @xmath677 . introduce a filtration on @xmath3 with @xmath678 for all @xmath679 and @xmath680 for all @xmath681 , where @xmath138 is half the degree of @xmath215 ; this choice of filtration is also clarified by @xcite . let @xmath682 where @xmath683 are the generators of @xmath684 given in theorem [ theorem : sp2npoissoncenter ] ; if conjecture [ conjecture : sp2ncenter ] is true , @xmath685 is also the highest term of @xmath686 . assuming that conjecture [ conjecture : sp2ncenter ] is true , @xmath3 is a free module over its center . + @xmath687 is a normal complete - intersection integral domain . \(1 ) introduce a filtration on @xmath687 with @xmath678 for @xmath679 and @xmath688 for @xmath643 . define @xmath689 by @xmath690 . the formula in theorem [ theorem : sp2npoissoncenter ] implies that @xmath691^{s_n}$ ] is a free and finite module over @xmath692 $ ] , so @xmath693^w$ ] is finite and free over @xmath694 $ ] . since @xmath695 is free over @xmath693^w$ ] by the classical kostant s theorem , @xmath695 , and hence @xmath696 , is free over @xmath694 $ ] . thus , @xmath697 is free over @xmath698 $ ] , implying the result . \(2 ) to show that @xmath687 is a normal integral domain , it suffices to show that the smooth locus of the zero set of @xmath699 has codimension 2 and is irreducible . let @xmath700 be a closed subscheme of @xmath701 defined by @xmath702 , and let @xmath703 where jac is the jacobi matrix of @xmath699 at @xmath704 with respect to some basis of @xmath7 and @xmath593 . it suffices to show that @xmath705 is a codimension 2 subvariety of @xmath706 and that the latter is irreducible . now , recall that @xmath707 by changing basis , we can rewrite @xmath708 as @xmath709 , where @xmath710 and @xmath711 are polynomial expressions in @xmath712 and @xmath713 ( in particular , there is no dependence on @xmath249 ) . we can and will use the jacobian of @xmath714 instead of @xmath715 to describe @xmath705 . let us calculate the derivatives of @xmath716 with respect to @xmath717 and @xmath718 : @xmath719 thus , if @xmath720 for some @xmath721 , then @xmath722 for all @xmath723 . equivalently , @xmath724 . now we will consider the situation in @xmath725 . we know that @xmath726 , where @xmath727 and @xmath728 is the nilpotent cone of @xmath593 . since @xmath7 and @xmath728 are irreducible , @xmath729 , and hence @xmath706 , is irreducible . recall that @xmath705 was defined as the locus of points @xmath730 such that @xmath731 , or in other words , all @xmath732 minors of the jacobian matrix have determinant 0 . since each of those determinants is homogeneous with respect to our second filtration , it is natural to define @xmath733 as a locus of points where @xmath731 . then , @xmath734 . note that @xmath735 , where @xmath736 and @xmath737 . the codimension of a regular nilpotent s orbit is 2 , so @xmath738 . it suffices to show that @xmath739 as well . we shall do this by showing that given a regular nilpotent @xmath142 , @xmath740 , where @xmath741 . let us switch to a basis of @xmath593 where the skew symmetric form is represented by the matrix @xmath742 if we define @xmath743 then @xmath744 , implying @xmath602 . now , suppose that @xmath745 at @xmath746 , for @xmath747 . by examining the @xmath748 components of @xmath749 , we get @xmath750 ; moreover , either @xmath751 , or @xmath752 . the conditions @xmath753 define a codimension two subspace as desired . we thus need to show that if @xmath750 and @xmath752 , then @xmath745 implies a nontrivial condition on @xmath249 . to find such a condition , note that @xmath754 and that @xmath755 does not depend on @xmath249 . now , let us take @xmath756 ; we can verify that @xmath757 , so @xmath758 . we note that @xmath759 , @xmath760 , @xmath761 and so forth . thus , @xmath762 . however , setting @xmath763 , we get @xmath764 , which is a nontrivial degree two polynomial in @xmath765 that should equal the number @xmath766 . this gives the other codimension 1 condition , and so @xmath767 is at least of codimension 2 in @xmath729 as desired . in this section , we will outline the proof of lemma [ lemma : sp2nkey ] , which states : @xmath768 we use the basis for @xmath7 defined in section [ section : sp2n ] , in which @xmath594 is represented by the matrix @xmath675 . let us multiply ( [ eqn : identity ] ) by @xmath637 and sum over @xmath340 to get the equivalent assertion that @xmath769 since the whole sum is @xmath649-invariant ( even though each term considered separately is not ) , we can look at the restriction of the sum to @xmath58 . thus , this sum equals zero if and only if @xmath770 we choose the following basis @xmath648 for @xmath593 : @xmath771 , @xmath772 , @xmath773 , for all @xmath774 , and for all @xmath775 , the elements @xmath776 , @xmath777 , @xmath778 , and @xmath779 . we observe that for any @xmath780 , there exists a unique basis vector in @xmath648 that takes @xmath781 to @xmath782 ; we shall denote this element by @xmath783 . these @xmath784 are not pairwise distinct since there are basis vectors with two nonzero entries . since @xmath785 acts transitively on @xmath7 , we can assume @xmath786 . using the above basis , we get @xmath787 where @xmath788 we now restrict to @xmath58 . we have @xmath789 only when the matrices for @xmath790 and @xmath791 have nonzero entries on the diagonal , or if @xmath790 and @xmath791 have nonzero entries at the @xmath340-th row @xmath400-th column and @xmath400-th row @xmath340-th column , respectively . this can only happen when @xmath792 for some @xmath793 . we can list all the ways this can happen for @xmath794 or @xmath795 with @xmath796 ( keeping in mind that @xmath797 and @xmath798 ) : to calculate the derivatives , let @xmath805 be the 4 by 4 matrix formed by the intersections of the first , second , @xmath806-th , and @xmath807-th rows and columns of @xmath142 , and let @xmath808 be the @xmath809 by @xmath809 matrix formed by the intersections of the remaining rows and columns . the space of all such @xmath808 is isomorphic to @xmath810 , and we denote the cartan subalgebra of diagonal matrices of this space by @xmath811 . all six of the above derivatives evaluate to the same polynomial in @xmath811 times the corresponding derivative in @xmath812 ; for instance , @xmath813 with @xmath814 and @xmath815 $ ] . thus , we can reduce our problem to @xmath812 , and straightforward computations verify ( [ eqn : identity ] ) for @xmath812 . similarly , when @xmath816 ( that is , when the term is of the form @xmath817 ) , all computations will reduce to analogous ones in @xmath818 . * acknowledgments : * the authors would like to thank pavel etingof for suggesting this research topic , for stimulating discussions , and for reviewing the rough draft of this paper . the authors would also like to thank the primes program at mit for sponsoring this research . finally , the authors would like to thank the referee for pointing out some inaccuracies in the original proof of theorem [ theorem : classification ] . j. bernstein and v. lunts , _ a simple proof of kostant s theorem that @xmath192 is free over its center _ , amer . j. math . 118 ( 1996 ) , no . 5 , 979987 . mr1408495 ( 97h:17012 ) v. drinfeld , _ degenerate affine hecke algebras and yangians _ , funcktsional . prilozhen . 20 ( 1986 ) , no . 1 , 6970 ( russian ) . mr0831053 ( 87m:22044 ) p. etingof and v. ginzburg , _ symplectic reflection algebras , calogero - moser space , and deformed harish - chandra homomorphism _ , invent . math . 147 ( 2002 ) , 243348 , doi 10.1007/s002220100171 . mr1881922 ( 2003b:16021 ) p. etingof , w.l . gan , and v. ginzburg , _ continuous hecke algebras _ , transform . groups 10 ( 2005 ) , no . 34 , 423447 , doi 10.1007/s00031 - 005 - 0404 - 2 , mr2183119 ( 2006h:20006 ) w. fulton and j. harris , _ representation theory : a first course _ , graduate texts in mathematics , vol . 129 , springer - verlag , new york , 1991 . mr1153249 ( 93a:20069 ) h. kaneta , _ the invariant polynomial algebras for the groups @xmath819 and @xmath820 _ , nagoya math . j. 94 ( 1984 ) , 6173 . mr748092 ( 86f:17012b ) v. g. kac and d. a. kazhdan , _ structure of representations with highest weight of infinite - dimensional lie algebras _ , adv . 34 ( 1979 ) , no . 1 , 97108 , doi 10.1016/0001 - 8708(79)90066 - 5 . mr547842 ( 81d:17004 ) a. tikaradze and a. khare , _ center and representations of infinitesimal hecke algebras of @xmath349 _ , comm . algebra 38 ( 2010 ) no . 2 , 405439 , doi 10.1080/00927870903448740 . mr2598890 ( 2011g:17024 ) g. lusztig , _ affine hecke algebras and their graded version _ , j. amer soc . 2 ( 1989 ) , no . 3 , 599635 , doi 10.2307/1990945 . mr0991016 ( 90e:16049 ) i. losev and a. tsymbaliuk , _ infinitesimal cherednik algebras as w - algebras _ , arxiv : 1305.6873 . a. tikaradze , _ center of infinitesimal cherednik algebras of @xmath0 _ , represent . theory 14 ( 2010 ) , 18 , doi 10.1090/s1088 - 4165 - 10 - 00363 - 8 . mr2577654 ( 2011b:16111 ) a. tikaradze , _ on maximal primitive quotients of infinitesimal cherednik algebras of @xmath0 _ , j. algebra 355 ( 2012 ) , 171175 , doi 10.1016/j.jalgebra.2012.01.013 . mr2889538
infinitesimal cherednik algebras are continuous analogues of rational cherednik algebras , and in the case of @xmath0 , are deformations of universal enveloping algebras of the lie algebras @xmath1 . in the first half of this paper , we compute the determinant of the shapovalov form , enabling us to classify all irreducible finite dimensional representations of @xmath2 . in the second half , we investigate poisson - analogues of the infinitesimal cherednik algebras and generalize various results to @xmath3 , including kostant s theorem .
the demonstration of metamaterial perfect absorbers @xcite represents one of the most important applications employing the astonishing properties found in metamaterials @xcite . the original idea @xcite is that , in metamaterials with simultaneous electrical and magnetic resonances , both of the effective permittivity @xmath0 and permeability @xmath1 are highly dispersive and can be tailored independently . at certain frequencies the effective impedance , which is defined as @xmath2 , matches to the free space impedance @xmath3 , and therefore the reflection is minimized . if at the same time the metamaterial is also of high loss which causes very low transmission , then near - unity absorption can occur within an ultra thin layer of material . under such considerations , in a typical metamaterial absorber , the magnetic resonance from the bi - layered metal structure @xcite is essential , and each layer of metal structure also provides the electrical response . as a direct evidence of the magnetic resonance in metamaterial absorbers , the surface currents excited in the two metal layers are found to be anti - parallel @xcite . while all of the above arguments seem to be plausible and accepted by a majority of researchers @xcite , fundamental questions remain . in fact , such a metamaterial absorber can be equivalent to a single layer of atoms ( or molecules ) , and it is strongly inhomogeneous in the wave propagating direction @xcite . so it can hardly be considered as an effective bulk medium where the constitutive parameters @xmath0 and @xmath1 apply . additionally , fabry - prot resonance @xcite , conventional transmission line model @xcite , and cavity resonance @xcite have been proposed to explain metamaterial absorbers or antireflection coatings , which also raise questions regarding the existence or involvement of the magnetic resonance in metamaterial absorbers . in this paper , we start with the multiple reflections interference model @xcite , and show in a typical metamaterial absorber there is negligible near - field interaction or magnetic response between the neighboring metal structures . the two layers of metal structures can be decoupled and the only link is multiple reflections between them . through combination of numerical simulations and analytical calculations , we further derive the surface currents and reveal that the anti - parallel directions are a result of interference and superposition , rather than excited by the magnetic component of the incident electromagnetic fields . and @xmath4 directions forming a square array with periodicity of @xmath5 . ( b ) unit cell used to obtain the reflection and transmission coefficients at air - spacer interface with the cross - resonator array . ( c ) the solid red curve is the reflection @xmath6 from the metamaterial absorber shown in unit cell ( a ) , and the dashed blue curve is the resonant transmission @xmath7 through the cross - resonator array shown in unit cell ( b ) . ( d ) surface currents of the metamaterial absorber at the peak absorption frequency . the current at the cross flows to left direction , while it flows to right direction at the ground plane . , without losing generality , we use a simple and typical metamaterial absorber structure operating at terahertz frequencies as an example . it consists of a cross - resonator array and a ground plane separated by @xmath8 m thick polyimide dielectric spacer @xcite . the unit cell is shown in fig . [ fig1](a ) , where the periodicity is @xmath9 @xmath10 m , the length and width of cross wires are @xmath11 m and @xmath12 m , respectively , and the thickness of metal is @xmath13 m . due to the presence of the ground plane , the substrate provides the mechanical support only and the choice of material does not affect the performance of the metamaterial absorber . we carry out numerical simulations using cst microwave studio 2009 @xcite . perfect electric conductor ( pec ) is used to simulate the metal , while the ohmic loss is effectively put in the polyimide spacer with dielectric constant @xmath14 . periodic boundary conditions are applied to the side walls parallel to @xmath15 direction , and the ports are at the front and back surfaces of the unit cell . the the simulated electric field reflection @xmath6 is shown by the solid red curve in fig . [ fig1](c ) , which reveals a nearly zero reflection at a frequency slightly below 1 thz . this also suggests that the incident light is almost completely absorbed by the metamaterial in this narrow frequency band , according to @xmath16 , where @xmath17 is the reflectance , and @xmath18 is the zero transmittance due the presence of the ground plane . the surface currents in the cross - resonator array and the ground plane exhibit anti - parallel directions , as shown in fig . [ fig1](d ) . the above model treats the metamaterial absorber as a _ coupled _ system , i.e. , the possible near - field interactions and magnetic resonance have been taken into account between the cross - resonator array and the ground plane . however , we may also consider the presence of the metal cross - resonator array resulting in an impedance - tuned air - spacer interface @xcite with dramatically modified complex reflection and transmission coefficients shown in fig . [ fig2 ] . the ground plane , on the other hand , functions as a perfect reflector with reflection coefficient @xmath19 . in the interference model we _ decouple _ the metamaterial absorber into two tuned interfaces , respectively with cross - resonator array and ground plane located at the two sides of the spacer . they are only linked by multiple reflections , as shown in the inset to fig . [ fig2](a ) , while any near - field interaction or magnetic resonance has been neglected . in this model , we need to know the reflection / transmission coefficients at the air - spacer interface with cross - resonator array , which are simulated using the unit cell shown in fig . [ fig1](b ) , where the ground plane and the substrate have been removed from the unit cell shown in fig . [ fig1](a ) . the cross - resonator array reveals a dipole resonance as indicated by the deep transmission dip at about 1.1 thz , shown in the dashed blue curve in fig . [ fig1](c ) . comparing the two curves , the feature of particular interest and worthy of mentioning is that near - zero reflection ( or unity absorption ) of the metamaterial absorber occurs at a frequency different from and lower than the resonance frequency of the cross - resonator array . inset : multiple reflections and interference model of the metamaterial absorber , where the cross - resonator array is indicated by the dashed line at the air - spacer interface . , width=288 ] as shown in the inset to fig . [ fig2](a ) , at the air - spacer interface with cross - resonator array , the incident light is partially reflected back to air with a reflection coefficient @xmath20 and partially transmitted into the spacer with a transmission coefficient @xmath21 . the latter continues to propagate until it reaches the ground plane , with a complex propagation phase @xmath22 , where @xmath23 is the free space wavenumber , @xmath24 is the propagation phase , and @xmath25 represents the absorption in the spacer . after the reflection at the ground plane and addition of another propagation phase @xmath26 , partial reflection and transmission occur again at the air - spacer interface with cross - resonator , with coefficients of @xmath27 and @xmath28 , respectively . similar to the light propagation in a stratified media @xcite , the overall reflection is then the superposition of the multiple reflections : @xmath29 where the first term is the reflection directly from the cross - resonator array , and the second term , including the `` @xmath30 '' sign , is the reflection resulting from superposition of the multiple reflections between the cross - resonator array and ground plane . the absorptance is then retrieved through @xmath31 since the transmission is zero . using the reflection and transmission coefficients shown in fig . [ fig2 ] , the absorptance is calculated in the decoupled metamaterial absorber system and the results are shown in fig . [ fig3 ] , where we have purposely varying the spacer thickness @xmath32 from 4 @xmath10 m to 16 @xmath10 m . the spacer thickness dependent absorptance in the coupled metamaterial absorber is also simulated directly using the unit cell shown in fig . [ fig1](a ) and the results are shown in the inset to fig . [ fig3 ] , which is in excellent agreement with the results calculated in the decoupled metamaterial absorber using the interference model . with all of these spacer thicknesses , we observe an absorption peak , which increases in the beginning and its frequency red - shifts as the spacer thickness increases . it reaches near - unity absorption at 1 thz when the spacer thickness is about 10 @xmath10 m , and then the peak value decreases and its frequency continuously red - shifts when the spacer thickness further increases . further numerical simulations and calculations reveal that , with any chosen spacer dielectric material , there is always an optimized spacer thickness where the reflection can reach zero , i.e. unity - absorption . the excellent agreement shown in fig . [ fig3 ] validates the interference model we have presented above . the physical explanation of a metamaterial absorber is then as follows . the multiple reflections in the metamaterial absorber , i.e. the second term in eq . ( [ reflection ] ) , constructively interfere as evidenced by the fact that near 1 thz the phase change of a round trip is @xmath33 . the superposition of the multiple reflections then destructively interferes with the direct reflection from the air - spacer interface with cross - resonators , i.e. the first term in eq . ( [ reflection ] ) . with the optimized spacer thickness , these two terms cancel each other out resulting in zero reflection ; while with other spacer thicknesses the amplitude and phase do not match and these two terms only partially cancel each other out , resulting in a reduced absorption peak and corresponding frequency shift . obviously , this explanation does not involve any near - field interaction or magnetic resonance between the two metal layers in the metamaterial absorber . ) and at the ground plane ( @xmath34 ) , as well as their difference ( @xmath35 ) , calculated through multiple reflections and superposition based on the interference model . the corresponding magnitude of the surface currents are shown in the inset . ( b ) backward radiation field ( reflection ) calculated through @xmath36 for various spacer thicknesses.,width=288 ] the remaining question is then why we observe the anti - parallel directions of the excited surface currents as shown in fig . [ fig1](d ) , which has been used as the evidence of a magnetic resonant response @xcite similar to fish - net metamaterials @xcite . in order to elucidate this mystic observation , by considering the boundary conditions at the interfaces we derive the excited surface currents based on interference and superposition . when the incident light with electric field @xmath37 is first reflected and transmitted at the air - spacer interface with cross - resonator array , according to the boundary condition requirements @xcite , it excites a surface current @xmath38 { e}_{\rm c}^{\rm i } \hat{\textbf{x}}$ ] , where @xmath3 and @xmath39 are impedances of the free space and spacer , respectively . the transmitted light continues to propagate and reflected by the ground plane , and the latter excites a surface current at the ground plane @xmath40 . the reflection from the ground plane propagates back to the air - spacer interface with cross - resonator array , where partial reflection and transmission occur which excites a surface current @xmath41 \tilde t_{12 } e^{i 2 \tilde\beta } ( -1 ) e_{\rm c}^{\rm i } \hat{\textbf{x}}$ ] . this procedure continues and the overall surface currents at the cross - resonator array and ground plane are then superpositions of multiple excitations : @xmath42 with a spacer thickness of @xmath8 m , the phase and magnitude of surface currents in eqs . ( [ eqjc ] ) and ( [ eqjg ] ) are shown in fig . [ fig4](a ) and its inset , respectively , which reveal a phase difference of @xmath43 and comparable surface current magnitude at 1 thz . that is , they are almost out - of - phase , directly resulted from the interference rather than the magnetic resonance . we also notice that , in the calculated results shown in fig . [ fig4](a ) , at frequencies below 1 thz the phase difference is @xmath44@xmath45 , while at frequencies above the resonance of the cross - resonator array ( about 1.1 thz ) the phase difference is nearly @xmath46 . this agrees well with the results in numerical simulations when treating the whole metamaterial absorber as a coupled system . finally , we use the obtained surface currents to calculate the overall backward radiation field ( i.e. reflection ) by @xmath47 for different spacer thicknesses . the results shown in fig . [ fig4](b ) reveal a reflection dip , in which the spacer thickness dependence of the reflection minimum and its frequency is again consistent with the absorption results shown in fig . we have shown that it is _ not _ required in a metamaterial absorber to have simultaneous electric and magnetic responses , which have been considered as the foundation of metamaterial absorbers . we explicitly demonstrate that the assumed magnetic resonance plays a negligible role in the impedance matching of metamaterial absorbers to free space . in contract , it is the destructive interference between the direct reflection and the following multiple reflections that effectively traps light in the metamaterial absorbers and eventually causes the high absorption . based on such an interference model , we derive the surface currents through multiple excitations and superposition , which clearly show the anti - parallel directions of surface currents in the resonator array and ground plane . this profound understanding of the underlying physics will undoubtedly provide valuable guidance in the future developments of more advanced metamaterial absorber based devices for microwave and photonic applications . the author acknowledges the fruitful discussions with j. zhou , y. zeng , l. huang , a. k. azad , j. f. ohara , w. j. padilla , r. d. averitt , and a. j. taylor . we acknowledge support in part from the los alamos national laboratory ldrd program . this work was performed , in part , at the center for integrated nanotechnologies , a us department of energy , office of basic energy sciences nanoscale science research center operated jointly by los alamos and sandia national laboratories . los alamos national laboratory , an affirmative action / equal opportunity employer , is operated by los alamos national security , llc , for the national nuclear security administration of the us department of energy under contract de - ac52 - 06na25396 .
the impedance matching in metamaterial perfect absorbers has been believed to involve and rely on magnetic resonant response , with a direct evidence from the anti - parallel directions of surface currents in the metal structures . here we present a different theoretical interpretation based on interferences , which shows that the two layers of metal structure in metamaterial absorbers are linked only by multiple reflections with negligible near - field interactions or magnetic resonances . this is further supported by the out - of - phase surface currents derived at the interfaces of resonator array and ground plane through multiple reflections and superpositions . the theory developed here explains all features observed in narrowband metamaterial absorbers and therefore provides a profound understanding of the underlying physics . 99 a. al , f. bilotti , n. engheta , and l. vegni , `` a thin absorbing screen using metamaterial complementary pairs , '' in _ proceedings of the international conference on electromagnetics in advanced applications _ , ( turin , italy , september 12 - 16 , 2005 ) , pp . 7578 . n. i. landy , s. sajuyigbe , j. j. mock , d. r. smith , and w. j. padilla , `` perfect metamaterial absorber , '' phys . rev . lett . * 100 * , 207402 ( 2008 ) . d. r. smith , w. j. padilla , d. c. vier , s. c. nemat - nasser , and s. schultz , `` composite medium with simultaneously negative permeability and permittivity , '' phys . rev . lett . * 84 * , 41844187 ( 2000 ) . h .- t . chen , j. f. ohara , a. k. azad , and a. j. taylor , `` manipulation of terahertz radiation using metamaterials , '' laser photon . rev . * 5 * , 513533 ( 2011 ) . w. cai and v. shalaev , _ optical metamaterials : fundamentals and applications _ ( springer , 2010 ) . s. zhang , w. fan , n. c. panoiu , k. j. malloy , r. m. osgood , and s. r. j. brueck , `` experimental demonstration of near - infrared negative - index metamaterials , '' phys . rev . lett . * 95 * , 137404 ( 2005 ) . v. m. shalaev , w. cai , u. k. chettiar , h .- k . yuan , a. k. sarychev , v. p. drachev , and a. v. kildishev , `` negative index of refraction in optical metamaterials , '' opt . lett . * 30 * , 33563358 ( 2005 ) . h. tao , n. i. landy , c. m. bingham , x. zhang , r. d. averitt , and w. j. padilla , `` a metamaterial absorber for the terahertz regime : design , fabrication and characterization , '' opt . express * 16 * , 71817188 ( 2008 ) . y. avitzour , y. a. urzhumov , and g. shvets , `` wide - angle infrared absorber based on a negative - index plasmonic metamaterial , '' phys . rev . b * * 7**9 , 045131 ( 2009 ) . n. liu , m. mesch , t. weiss , m. hentschel , and h. giessen , `` infrared perfect absorber and its application as plasmonic sensor , '' nano lett . * 10 * , 23422348 ( 2010 ) . j. hao , l. zhou , and m. qiu , `` nearly total absorption of light and heat generation by plasmonic metamaterials , '' phys . rev . b * 83 * , 165107 ( 2011 ) . j. zhou , h .- t . chen , t. koschny , a. k. azad , a. j. taylor , c. m. soukoulis , and j. f. ohara , `` application of metasurface description for multilayered metamaterials and an alternative theory for metamaterial perfect absorber , '' http://arxiv.org/abs/1111.0343 . h .- t . chen , j. zhou , j. f. ohara , f. chen , a. k. azad , and a. j. taylor , `` antireflection coating using metamaterials and identification of its mechanism , '' phys . rev . lett . * 105 * , 073901 ( 2010 ) . j. sun , l. liu , g. dong , and j. zhou , `` an extremely broad band metamaterial absorber based on destructive interference , '' opt . express * 19 * , 2115521162 ( 2011 ) . q. y. wen , y. s. xie , h. w. zhang , q. h. yang , y. x. li , and y. l. liu , `` transmission line model and fields analysis of metamaterial absorber in the terahertz band , '' opt . express * 17 * , 2025620265 ( 2009 ) . d. y. shchegolkov , a. k. azad , j. f. ohara , and e. i. simakov , `` perfect subwavelength fishnetlike metamaterial - based film terahertz absorbers , '' phys . rev . b * 82 * , 205117 ( 2010 ) . h. tao , c. m. bingham , a. c. strikwerda , d. pilon , d. shrekenhamer , n. i. landy , k. fan , x. zhang , w. j. padilla , and r. d. averitt , `` highly flexible wide angle of incidence terahertz metamaterial absorber : design , fabrication , and characterization , '' phys . rev . b * 78 * , 241103(r ) ( 2008 ) . m. diem , t. koschny , and c. m. soukoulis , `` wide - angle perfect absorber / thermal emitter in the terahertz regime , '' phys . rev . b * 79 * , 033101 ( 2009 ) . x. liu , t. starr , a. f. starr , and w. j. padilla , `` infrared spatial and frequency selective metamaterial with near - unity absorbance , '' phys . rev . lett . * 104 * , 207403 ( 2010 ) . cst microwave studio 2009 , http://www.cst.com/. c. l. holloway , a. dienstfrey , e. f. kuester , j. f. ohara , a. k. azad , and a. j. taylor , `` a discussion on the interpretation and characterization of metafilms / metasurfaces : the two - dimensional equivalent of metamaterials , '' metamaterials * 3 * , 100112 ( 2009 ) . m. born and e. wolf , _ principles of optics _ ( pergamon , 1980 ) , 6th ed . j. d. jackson , _ classical electrodynamics _ ( wiley , 1998 ) , 3rd ed .
since akimitsu and co - workers presented a remarkably high temperature ( @xmath8 39 k ) superconducting magnesium diboride mgb@xmath3 in the beginning of 2001 @xcite , much attention has been focused on this diboride material and some related boride compounds @xcite . the material mgb@xmath3 takes the layered structure of b - mg - b stacking and has a coupling between 2-dimensional ( 2d ) b - b layers @xcite . according to several band calculations @xcite , in the mgb@xmath3-type crystal structure there exist a strong covalent b - b bonding and an ionic b - mg bonding . the strong covalent b - b bonding and the presence of the ionic mg atoms lead to hole - like cylindrical fermi surfaces in mgb@xmath3 @xcite , and the holes are conducted in each 2d b - b layer . it is observed that the 2d b - b layers , like the cu - o layers in high-@xmath2 cuprates , play an important role in the electrical transport properties @xcite . the superconductor mgb@xmath3 can then be beautiful realization of the essential physics of superconductivity in cuprates , without the complications of cu @xmath9 orbitals @xcite . in a previous paper @xcite , on the basis of a theoretical model of the b - b layers analogous to the cu - o layers @xcite , we investigated the ionic diboride compounds ( m@xmath10)(b@xmath11 , @xmath12e.g . , ( mg@xmath10)(b@xmath11 in the case of mgb@xmath1@xmath13 , and discussed their superconducting critical temperature @xmath2 . we made investigation into @xmath2 assuming a simple linear relation between @xmath2 and the inter - layer charging energy @xmath4 @xcite . in the present paper , instead of assuming the linear relation , we will analyze @xmath2 and @xmath4 in more detail on the basis of an original formula for @xmath2 presened in refs . @xcite , and also mention general boride compounds mb@xmath0 in which the interstitial ionic atoms m between b - b layers are not restricted to m@xmath10 . the purpose of this paper is to grope for various possible boride compounds with @xmath2 over 39 k and to give a hint as to new material synthesis . in the viewpoint of our theoretical model , the recently discovered 117-kelvin superconductor c@xmath5chbr@xmath6 @xcite is also discussed . we model the boride compounds mb@xmath0 ( m@xmath14 mg , be , etc . ) as quantum - capacitive josephson junction arrays ( jja ) with weakly coupled superconducting grains on a 3d lattice , following kawabata - shenoy - bishop ( ksb ) theory @xcite , originally proposed to the high-@xmath2 cuprates with 2d cu - o layers and a coupling between the cu - o layers . in the present jja model , the josephson coupling energy @xmath15 via the cooper pair tunneling within and between the b - b layers enhances the critical temperature @xmath2 , while the charging energy @xmath4 between the b - b layers depresses @xmath2 . the origin of @xmath4 is that , when the cooper pairs tunnel between the layers , local inter - layer charge unbalance is induced between the superconducting grains in jja . this charge unbalance gives rise to the electric fields between the grains and it costs the electromagnetic energies and destroys the phase coherence of superconductivity @xcite . here , an important point for our analysis is that the interstitial ionic atoms , m , located between the b - b layers play a role of capacitors , reducing the charging energy @xmath4 by the electric polarization of the ionic atom . we estimate @xmath2 which is a function of @xmath4 . according to the ksb theory @xcite , in the present jja model @xmath2 is expressed , with an unknown function @xmath16 , as @xmath17 and for zero charging energy @xmath18 it must reduce to a known equation @xmath19 which is the dimensionless critical coupling @xcite in terms of the coupling anisotropy @xmath20 in the classical anisotropic 3d xy / jja system @xmath12@xmath21 , where @xmath22 is the inter - layer josephson coupling energy , @xmath23 the intra - layer one , and @xmath24@xmath13 . in the 3d limit @xmath25 , the present jja model may be applied to the superconductor c@xmath5chbr@xmath6 , where the coupled superconducting grains on a 3d lattice in the 3d jja are composed of c@xmath7 molecules . we discuss this superconductor in sec . [ c60 ] . expanding @xmath16 with respect to @xmath26 , we obtain from eq . ( [ eq : tc - ksb-1 ] ) the expression for @xmath2 as @xcite @xmath27 where @xmath28 . in this paper , we rewrite eq . ( [ eq : tc - ksb-3 ] ) as @xmath29 with the phenomenological parameters @xmath30 and @xmath31 , which should be determined empirically . we use eq . ( [ eq : tc - ksb-4 ] ) to fit data and estimate @xmath2 for some borides in the next section . referring to kittel s textbook and shockley s tables @xcite for the ionic radius @xmath32 and electronic polarizability @xmath33 of the ionic atoms m , we have obtained the charging energy @xmath34 between the b - b layers as in table [ table : ec ] , following the ksb theory @xcite . we discuss first the ionic compounds m@xmath10(b@xmath35)@xmath1 expected to have nearly the same electronic structure @xcite as mgb@xmath3 ( mg@xmath10(b@xmath35)@xmath1 ) and to have almost the same josephson coupling energy @xmath15 @xmath36i.e . , the same `` bare @xmath2 '' ( @xmath30 ) without any influence of the m - atom - dependent charging energy @xmath4 , and the same value of @xmath31 in eq . ( [ eq : tc - ksb-4])@xmath37 $ ] . by hypothesizing the same @xmath15 for any m@xmath10(b@xmath35)@xmath1 compounds , we estimate critical temperatures @xmath2 ( @xmath38 ) . in fig . [ fig : tc ] we show two points , as experimentally known data , for beb@xmath3 ( @xmath39 k ( @xmath40 k ) @xcite , @xmath41 ) and mgb@xmath3 ( @xmath4239 k @xcite , @xmath43 ) , where @xmath4 ( @xmath44 ) are of table [ table : ec ] and in arbitrary units . from eq . ( [ eq : tc - ksb-4 ] ) and the values of @xmath2 and @xmath4 for mgb@xmath3 , we obtain an equation @xmath45 . from this equation and eq . ( [ eq : tc - ksb-4 ] ) with @xmath46 k and @xmath41 for beb@xmath3 , we obtain @xmath47 , which is contrary to the condition that the charging energy @xmath4 decreases @xmath2 in eq . ( [ eq : tc - ksb-4 ] ) . it should be remedied by taking higher order terms of @xmath26 when obtaining eq . ( [ eq : tc - ksb-3 ] ) . instead , we plot the @xmath2 vs. @xmath4 curve in fig . [ fig : tc ] on the basis of eq . ( [ eq : tc - ksb-4 ] ) , assuming @xmath48 k as representative . in this case @xmath49beb@xmath50 k at @xmath41 . even if @xmath30 is increased further , @xmath2 at @xmath41 only decreases a little . on the other hand , when @xmath30 is decreased , @xmath2 at @xmath41 is increased and deviates more from the experimental result @xcite @xmath49beb@xmath51 k. we then believe that the @xmath2 vs. @xmath4 curve drawn in fig . [ fig : tc ] is not so unreasonable . it is expected that the diborides , which contain the ionic atoms m with the electric polarizability @xmath33 and have the corresponding @xmath4 in table [ table : ec ] , are distributed on the @xmath2 vs. @xmath4 graph along the curve drawn in fig . [ fig : tc ] . in ref . @xcite , @xmath52 k is observed in beb@xmath53 , which has the crystal structure different from mgb@xmath3 . it has also been suggested that nonstoichiometry may be an important point in the superconductivity of some boride compounds @xcite . it is then interesting to synthesize various borides mb@xmath0 and try to measure their @xmath2 . we give in table [ table : ec ] various atoms m for which the polarizability has been obtained @xcite . the present theory can be applied to the borides mb@xmath0 which have the b - m - b stacking structure . in this case the values of the josephson coupling energy @xmath15 and the coupling anisotropy @xmath20 @xmath12and thus , the corresponding @xmath30 and @xmath31 in eq . ( [ eq : tc - ksb-4])@xmath13 would be different in general for each compound . it is , however , expected that the compounds with smaller @xmath4 is adventageous to the superconductivity , which can be a guide in the synthesis of new materials . the thin films of mb@xmath0 with the b - m - b stacking structure could be synthesized by , for example , the epitaxial methods . quite recentry , schn _ et al . _ @xcite found a new superconducting system c@xmath5chbr@xmath6 which exhibits very high critical temperature @xmath54 k. it is a c@xmath7 superconductor intercalated with chbr@xmath6 molecules . this @xmath2 is about two times larger than 52 k @xcite of c@xmath7 without intercalations . the critical temperatures of c@xmath7 superconductors increase with the lattice constants @xmath55 ; c@xmath7 ( @xmath56 k , @xmath57 ) , c@xmath5chcl@xmath6 ( @xmath58 k , @xmath59 ) , and c@xmath5chbr@xmath6 ( @xmath54 k , @xmath60 ) @xcite . one of the promising course of attaining higher @xmath2 is certainly to search for new spacer molecules which expand the separation between c@xmath7 molecules @xcite . however , it seems to be open to further discussion whether the only 2 % expansion of the lattice constant is sufficient for the increase of the electronic density of states which corresponds to rather drastic rise of @xmath2 from 52 to 117 k. in the viewpoint of our theory , we would like to point out another possibility that the intercalation molecules would play a role of the capacitors in the c@xmath7 superconductors as discussed for the atoms m in the boride compounds mb@xmath0 . by the electric polarization of the neutral molecules , the intercalation molecules such as chbr@xmath6 may reduce the charging energy @xmath4 between the c@xmath7 molecules which constitute a 3d jja as superconducting grains on a 3d lattice . the smaller the charging energy @xmath4 is , the higher the critical tempereture @xmath2 is , as discussed in the preceding sections . according to experimental data listed in table 11 of ref . @xcite , the molecular polarizability of chbr@xmath6 is indeed larger than that of chcl@xmath6 , which is in accord with the present scenario that the intercalation molecules act as capacitors which reduce @xmath4 . we can estimate @xmath2 of other intercalated c@xmath7 superconductors as follows . the experimental values of the molecular polarizability are @xmath33(chcl@xmath61 au and @xmath33(chbr@xmath62 au @xcite . assuming an approximately constant value of the lattice constant @xmath55 , the charging energies @xmath63 in arbitrary units are obtaind as @xmath64c@xmath5chcl@xmath65 and @xmath64c@xmath5chbr@xmath66 . on the @xmath2 vs. @xmath4 graph , extrapolating the straight line which goes through two points @xmath67 for c@xmath5chcl@xmath6 and @xmath68 for c@xmath5chbr@xmath6 , namely , @xmath69 , we predict @xmath70 k ( c@xmath5chi@xmath6 ) , 127 k ( c@xmath5ch@xmath3i@xmath3 ) , 137 k ( c@xmath5c@xmath71h@xmath72cl@xmath3 ) , 127 k ( c@xmath5c@xmath71h@xmath73no@xmath3 ) , 121 k ( c@xmath5c@xmath71h@xmath73cl ) , and 175 k ( c@xmath5c@xmath74h@xmath75 ) . here , we use @xmath76 au ( chi@xmath6 ) , 87.05 au ( ch@xmath3i@xmath3 ) , 95.83 au ( c@xmath71h@xmath72cl@xmath3 ) , 87.19 au ( c@xmath71h@xmath73no@xmath3 ) , 82.67 au ( c@xmath5c@xmath71h@xmath73cl ) , and 153.86 au ( c@xmath74h@xmath75 ) @xcite . depending on actual lattice constants , other molecules with relatively large polarizability listed in ref . @xcite may also be interesting as candidates of the intercalation molecules . we indicated that , if the boride compounds are synthesized by using ionic atoms with larger polarizability @xmath33 or smaller diameter instead of the mg atoms , the critical temperatures @xmath2 will become higher than 39 k of mgb@xmath1 . we also pointed out that the recently descovered superconductor c@xmath5chbr@xmath6 will owe its very high critical temperature to the intercalation molecules chbr@xmath6 _ as electoric capacitors _ which reduce the charging energy @xmath4 @xmath77 among the superconducting grains of c@xmath7 molecules , rather than as simple spacer molecules which expand the separation @xmath55 between c@xmath7 molecules . we hope that the present proposals are a hint or an encouragement to future synthesis of new materials . c. kawabata , s. r. shenoy , and a. r. bishop , _ advances in science and technology 8 , superconductivity and superconducting materials technologies _ , edited by p. vincenzini ( techna srl . , faenza , 1995 ) p.13 . versus charging energy @xmath4 ( see text ) . points are for mgb@xmath1 ( @xmath80 k @xcite ) and beb@xmath1 ( @xmath39 k ( @xmath40 k ) @xcite ) . the values of @xmath4 are of table [ table : ec ] for mg@xmath10 and be@xmath10 . , width=529 ] . ionic radius @xmath32 @xcite ionic electric polarizability @xmath33 @xcite and charging energy @xmath81 . the data of @xmath33 in brackets are of ref.@xcite originally and the values of @xmath32 for cu are of ref.@xcite [ cols= " < , < , < , < , < " , ]
the boride compounds mb@xmath0 related to the magnesium - boron stacking layered material mgb@xmath1 are discussed in terms of the b - b layers in the borides analogous to the cu - o ones in the cuprates . we propose a possibility of superconducting materials which exhibit higher critical temperature @xmath2 than 39 k of mgb@xmath3 . we point out a role of interstitial ionic atoms m ( e.g. , mg in mgb@xmath3 ) as capacitors , which reduce the condensation - energy loss due to the charging energy @xmath4 between the b - b layers . in the viewpoint of the present model , the recently discovered 117-kelvin superconductor c@xmath5chbr@xmath6 is also discussed in terms of the intercalation molecules chbr@xmath6 as possible capacitors among the superconducting grains of c@xmath7 molecules . pcp-2 /iss 2001 , , and mgb@xmath3 , c@xmath5chbr@xmath6 , josephson junction array model , electric polarizability , charging energy 74.62.-c , 74.62.bf
over the last decade , deep pencil beam surveys of high - redshift galaxies have been the observational basis for studying the processes that drive galaxy formation and evolution . for a given amount of observing time , such surveys have to trade off imaging area and imaging depth : among hst imaging surveys , gems @xcite , aegis @xcite and cosmos @xcite are examples that aim at comparatively wide fields to observe a large sample of galaxies . alternatively , the hubble deep field ( * ? ? ? * hdf ) and the ultra deep field ( * ? ? ? * udf ) are examples of extremely deep surveys in small areas , aimed at detecting faint galaxies ( which can be either of low mass or at high redshift ) . one of the most fundamental properties of galaxy sub - populations at any epoch is their number density . however , observational estimates of galaxy number densities in finite volumes are subject to uncertainty due to _ cosmic variance _ , arising from underlying large - scale density fluctuations and leading to uncertainties in excess of nave poisson errors . note that this source of uncertainty is referred to as sample variance in other branches of cosmology . for sampling volumes much larger than the typical clustering scale of the observed objects , cosmic variance is not significant . however , many important existing surveys have a sampling volume that is small enough that cosmic variance may dominate the uncertainties . this may be particularly true at high redshift , where galaxies are expected to be much more strongly clustered than dark matter @xcite . still , many published quantities which are based on number density ( e.g. luminosity functions , stellar mass functions , etc . ) are quoted with error budgets that do not properly account for cosmic variance . as shown by @xcite , the normalization and the slope of high - redshift luminosity functions can be affected by cosmic variance errors . sound estimates of cosmic variance are helpful both for analyzing existing observational data and for designing future surveys . for a detailed and general treatment of the cosmic error , we refer to @xcite , @xcite and @xcite . our goal here is to provide a simple recipe to derive cosmic variance for galaxy samples selected by their stellar mass @xmath3 , for a field of specified angular dimensions @xmath14 and @xmath15 , at a mean redshift @xmath1 and for a redshift bin size of @xmath2 . several previous works have provided estimates of cosmic variance for specific sets of assumptions . for example , @xcite presented estimates of the cosmic variance as a function of field area and axis ratio for the redshift range @xmath16 , relevant to the deep2 redshift survey , but did not account for galaxy bias ( the clustering amplitude of galaxies relative to dark matter ; see eqn.[eqn : bias ] ) . @xcite provided predictions that could be used to estimate cosmic variance as a function of mean redshift and survey volume , using the number density of the population to estimate the bias , assuming one galaxy per halo . however , @xcite used the approximation of spherical ( rather than rectangular or pencil beam ) volumes . @xcite presented estimates of cosmic variance for pencil beam geometries , again using the number density of the population to estimate the host halo mass and hence the bias . many galaxy samples in observational studies , however , are selected by their stellar mass , and stellar mass is one of the fundamental properties of galaxies . galaxy clustering and bias are strong functions of stellar mass , making it extremely important to account for the mass dependence in cosmic variance estimates . moreover , one does not necessarily know the number density of a population of interest a priori . we therefore take a different approach here , providing estimates of cosmic variance explicitly as a function of stellar mass and redshift . to do this , we make use of the results previously presented in ( * ? ? ? * m09 ) . we used a halo occupation distribution ( hod ) model to empirically establish the relationship between stellar mass and dark matter halo ( or sub - halo ) mass at different redshifts , and then used a dissipationless n - body simulation to compute the galaxy bias as a function of stellar mass and redshift . we presented fitting functions for this quantity in m09 . we combine these with estimates of the cosmic variance for the dark matter for rectangular cells as described in @xcite . the paper is organized as follows : in section [ sec : method ] we describe how to define and how to compute cosmic variance for a galaxy population , based on the underlying cold dark matter ( cdm ) theory , a model for galaxy bias and for a combination of separate survey fields . in section [ sec : recipe ] we spell out a recipe to compute cosmic variance for a pencil beam geometry in four steps : ( 1 ) select survey geometry , ( 2 ) choose mean redshift and redshift bin size , ( 3 ) determine stellar mass interval and ( 4 ) compute cosmic variance . we also apply this recipe to four existing surveys at a redshift of @xmath17 and @xmath18 . in section [ sec : properties ] we compare cosmic variance for different galaxy samples at different redshifts and different survey geometries . finally , we summarize our methods and conclusions in section [ sec : conclusions ] . throughout , we assume cosmological parameters consistent with results from wmap-3 @xcite for a flat @xmath19 cosmological model : matter density @xmath20 , cosmological constant @xmath21 , hubble parameter @xmath22 , fluctuation amplitude @xmath23 and primordial power - spectrum @xmath24 . all stellar masses are calibrated to a @xcite initial mass function . the mean @xmath25 and the variance @xmath26-@xmath27 are given by the first and second moments of the probability distribution @xmath28 , which describes the probability of counting @xmath29 objects within a volume @xmath30 . the relative cosmic variance is defined as @xmath31 the last term in the numerator represents the correction for poisson shot noise . the second moment of the object counts is @xmath32 where @xmath33 is the two - point correlation function of the sample and @xmath30 is the sample volume ( see * ? ? ? 234 ) . combining this with equation , the cosmic variance can be written as @xmath34 there are two approaches to evaluating this equation . the first is applicable to populations with a known correlation function , when the integral can be solved either analytically or numerically as a function of cell radius or cell volume . however , in practice , there are a number of difficulties with this approach : while on small scales in the nonlinear regime ( @xmath35 ) the correlation function can typically be approximated by a power law , it deviates on larger scales ( in the linear regime ) , from this form . additionally , in many cases of interest the correlation function is not known a priori . the second approach to calculating the cosmic variance can be used if the correlation function can not be approximated by a power law or if it is unknown . it makes use of the galaxy bias , which can be predicted for a given galaxy population using halo occupation models . we can substitute the galaxy correlation function @xmath36 in equation for the correlation function for dark matter @xmath37 : @xmath38 where @xmath39 is the galaxy bias , which depends both on redshift @xmath40 , and the stellar mass of the galaxies @xmath3 . in general , the bias is also a function of @xmath41 , however , in the linear regime , we assume that it is independent of scale . this assumption will be made throughout the paper . employing the bias in this manner yields @xmath42 the integral leading to @xmath43 can then be solved using linear theory in cdm for any volume @xmath30 with a given geometry . the cosmic variance for galaxies can thus be determined by multiplying the dark matter cosmic variance at a given redshift with the galaxy bias for a given stellar mass at that redshift . for a pencil beam geometry the volume element @xmath30 is set by the angular size of the survey and the redshift bin size . we now pursue this second approach , using the galaxy bias as a function of redshift and stellar mass as computed by @xcite . in this paper the authors used a halo occupation model to obtain a parameterized stellar - to - halo mass relation by populating halos and subhalos in an @xmath29-body simulation with galaxies and requiring that the observed stellar mass function be reproduced . using the halo positions obtained from the simulation , they find that predictions for the galaxy correlation function are in excellent agreement with observed clustering properties at low redshift . the derived stellar - to - halo mass relation is finally used to predict the stellar - mass - dependent galaxy correlation function and bias at high redshift . we calculate the cosmic variance for dark matter in each redshift bin by solving the integral in eqn for dark matter using the code quickcv which is described in @xcite . the code computes the dark matter correlation function using the transfer function given in @xcite and a window function which is 1 inside the volume @xmath30 and 0 elsewhere . the volume elements are determined by the angular dimensions of the survey field and the redshift bin . finally , the cosmic variance for galaxies is computed using equation . once cosmic variance has been determined for a number of individual survey fields that are widely separated on the sky , compared to the correlation length , as is often the case ( e.g. for the various hst survey fields ) , one can easily derive the cosmic variance for the combined sample . for the combination of fields @xmath44 the total volume is @xmath45 implying a cosmic variance for the total sample of @xmath46 where we have used @xmath47 since the galaxies in separated fields are uncorrelated . for a pencil beam geometry ( angular extend @xmath48 radian ) this can be written as @xmath49 ^ 2}\;\ ] ] where @xmath14 and @xmath15 are the angular dimensions of each field . note that this volume - weighted average is identical to an inverse - variance - weighted average only in the range where the cosmic variance scales with volume . in order to compute the cosmic variance for galaxies in a given volume in the universe one has to fix five parameters . the size of the volume is determined by two dimensions tangential to the line of sight and one dimension parallel to the line of sight . the tangential dimensions we use are the angular size of the field @xmath14 and @xmath15 , while the vertical dimension is set by the size of the redshift bin under consideration @xmath2 . the fourth parameter is the mean redshift of the volume @xmath1 , since the cosmic variance depends on the clustering strength which is a function of redshift . moreover @xmath1 determines the length scale of the volume bins by setting the conversion factor from angular dimensions and @xmath2 into comoving mpc . in order to determine the cosmic variance for galaxies one finally has to fix the stellar mass @xmath3 of the sample under consideration , as the clustering strength and the bias depend strongly on stellar mass . thus the five parameters which have to be set are @xmath14 , @xmath15 , @xmath1 , @xmath2 and @xmath3 . the first step to calculate the cosmic variance is to fix the size of the volume element under consideration . for now we set the size of the redshift bin to a reference value of @xmath50 . this is a convenient bin size for low redshift while for higher redshift it is rather small . however , this choice can be changed at any redshift ( for values @xmath51 ) using the recipe presented in section [ sec : redshiftbins ] . for the tangential dimensions , we choose five popular fields : the hubble ultra deep field ( * ? ? ? * udf ) , one field of the great observatories origins deep survey ( * ? ? ? * goods ) , the extended chandra deep field south ( e - cdfs ) used in the galaxy evolution from morphology and seds survey ( * ? ? ? * gems ) , the extended groth strip ( egs ) used in the all - wavelength extended groth strip international survey ( * ? ? ? * aegis ) and the field used in the cosmic evolution survey ( * ? ? ? * cosmos ) . the angular dimensions of the surveys and the resulting observation areas are summarized in table [ t : surveyarea ] . we note that due to the small scales for the udf numerical uncertainties in the integration of the correlation function are likely to be high . as a result the values for cosmic variance in the udf computed here , especially for @xmath52 ( where size of the volume elements is very small ) have considerable uncertainties . at these scales , the assumption of linear biasing is also not accurate . as a consequence the cosmic variance for the udf presented here is to be treated with caution and is mainly included to demonstrate the large cosmic variance in small fields . l r r r r r udf & 1 & 3.3 & 3.3 & 11 & 11 + goods & 2 & 10 & 16 & 160 & 320 + gems & 1 & 28 & 28 & 784 & 784 + egs & 1 & 10 & 70 & 700 & 700 + cosmos & 1 & 84 & 84 & 7056 & 7056 + [ t : surveyarea ] c c c c c c c c c + & & & + @xmath1 & @xmath2 & @xmath43 & & & & & + & & & 8.75&9.25&9.75&10.25&10.75&11.25 + + 0.1 & 0.2 & 0.316 & 0.349 & 0.358 & 0.380 & 0.409 & 0.430 & 0.526 + 0.3 & 0.2 & 0.237 & 0.272 & 0.288 & 0.295 & 0.318 & 0.339 & 0.428 + 0.5 & 0.2 & 0.218 & 0.262 & 0.273 & 0.283 & 0.307 & 0.332 & 0.435 + 0.7 & 0.2 & 0.204 & 0.259 & 0.271 & 0.280 & 0.305 & 0.335 & 0.456 + 0.9 & 0.2 & 0.192 & 0.259 & 0.274 & 0.282 & 0.308 & 0.346 & 0.488 + 1.1 & 0.2 & 0.182 & 0.264 & 0.281 & 0.289 & 0.317 & 0.365 & 0.531 + 1.5 & 0.2 & 0.167 & 0.282 & 0.305 & 0.319 & 0.352 & 0.426 & 0.648 + 1.9 & 0.2 & 0.156 & 0.311 & 0.341 & 0.369 & 0.407 & 0.518 & 0.807 + 2.5 & 0.2 & 0.143 & 0.374 & 0.416 & 0.482 & 0.529 & 0.718 & 1.129 + 3.5 & 0.2 & 0.128 & 0.523 & 0.593 & 0.781 & 0.845 & 1.236 & 1.898 + + 0.1 & 0.2 & 0.244 & 0.270 & 0.277 & 0.294 & 0.317 & 0.333 & 0.407 + 0.3 & 0.2 & 0.205 & 0.236 & 0.243 & 0.255 & 0.276 & 0.293 & 0.371 + 0.5 & 0.2 & 0.179 & 0.215 & 0.224 & 0.232 & 0.252 & 0.272 & 0.357 + 0.7 & 0.2 & 0.160 & 0.203 & 0.213 & 0.220 & 0.239 & 0.263 & 0.358 + 0.9 & 0.2 & 0.146 & 0.197 & 0.208 & 0.214 & 0.234 & 0.263 & 0.371 + 1.1 & 0.2 & 0.135 & 0.195 & 0.207 & 0.214 & 0.235 & 0.270 & 0.392 + 1.5 & 0.2 & 0.119 & 0.200 & 0.217 & 0.227 & 0.250 & 0.303 & 0.460 + 1.9 & 0.2 & 0.107 & 0.215 & 0.235 & 0.255 & 0.281 & 0.357 & 0.557 + 2.5 & 0.2 & 0.096 & 0.250 & 0.278 & 0.322 & 0.354 & 0.480 & 0.754 + 3.5 & 0.2 & 0.083 & 0.336 & 0.382 & 0.503 & 0.544 & 0.795 & 1.221 + + 0.1 & 0.2 & 0.226 & 0.250 & 0.256 & 0.272 & 0.293 & 0.308 & 0.377 + 0.3 & 0.2 & 0.174 & 0.200 & 0.206 & 0.216 & 0.234 & 0.249 & 0.315 + 0.5 & 0.2 & 0.144 & 0.173 & 0.180 & 0.187 & 0.203 & 0.219 & 0.288 + 0.7 & 0.2 & 0.124 & 0.158 & 0.165 & 0.171 & 0.186 & 0.204 & 0.279 + 0.9 & 0.2 & 0.110 & 0.149 & 0.157 & 0.162 & 0.177 & 0.199 & 0.280 + 1.1 & 0.2 & 0.100 & 0.144 & 0.154 & 0.158 & 0.174 & 0.200 & 0.291 + 1.5 & 0.2 & 0.085 & 0.143 & 0.155 & 0.162 & 0.179 & 0.217 & 0.329 + 1.9 & 0.2 & 0.075 & 0.150 & 0.164 & 0.178 & 0.196 & 0.250 & 0.389 + 2.5 & 0.2 & 0.065 & 0.170 & 0.189 & 0.219 & 0.240 & 0.326 & 0.512 + 3.5 & 0.2 & 0.054 & 0.221 & 0.251 & 0.331 & 0.358 & 0.523 & 0.804 + + 0.1 & 0.2 & 0.215 & 0.238 & 0.244 & 0.259 & 0.279 & 0.293 & 0.358 + 0.3 & 0.2 & 0.157 & 0.180 & 0.186 & 0.195 & 0.211 & 0.224 & 0.284 + 0.5 & 0.2 & 0.128 & 0.154 & 0.160 & 0.167 & 0.181 & 0.195 & 0.256 + 0.7 & 0.2 & 0.120 & 0.140 & 0.146 & 0.151 & 0.164 & 0.181 & 0.246 + 0.9 & 0.2 & 0.097 & 0.131 & 0.139 & 0.143 & 0.156 & 0.175 & 0.247 + 1.1 & 0.2 & 0.088 & 0.127 & 0.135 & 0.139 & 0.153 & 0.176 & 0.256 + 1.5 & 0.2 & 0.075 & 0.126 & 0.136 & 0.143 & 0.157 & 0.191 & 0.290 + 1.9 & 0.2 & 0.066 & 0.132 & 0.145 & 0.157 & 0.173 & 0.220 & 0.343 + 2.5 & 0.2 & 0.057 & 0.150 & 0.167 & 0.194 & 0.213 & 0.288 & 0.453 + 3.5 & 0.2 & 0.048 & 0.197 & 0.224 & 0.295 & 0.319 & 0.466 & 0.716 + + 0.1 & 0.2 & 0.181 & 0.200 & 0.205 & 0.218 & 0.235 & 0.247 & 0.302 + 0.3 & 0.2 & 0.117 & 0.134 & 0.138 & 0.145 & 0.157 & 0.167 & 0.211 + 0.5 & 0.2 & 0.088 & 0.105 & 0.109 & 0.114 & 0.123 & 0.133 & 0.175 + 0.7 & 0.2 & 0.070 & 0.089 & 0.094 & 0.097 & 0.105 & 0.116 & 0.158 + 0.9 & 0.2 & 0.059 & 0.080 & 0.084 & 0.087 & 0.095 & 0.107 & 0.151 + 1.1 & 0.2 & 0.051 & 0.074 & 0.079 & 0.083 & 0.090 & 0.103 & 0.150 + 1.5 & 0.2 & 0.041 & 0.070 & 0.075 & 0.079 & 0.087 & 0.105 & 0.160 + 1.9 & 0.2 & 0.035 & 0.070 & 0.077 & 0.083 & 0.092 & 0.117 & 0.182 + 2.5 & 0.2 & 0.029 & 0.076 & 0.085 & 0.098 & 0.108 & 0.146 & 0.230 + 3.5 & 0.2 & 0.023 & 0.095 & 0.108 & 0.142 & 0.154 & 0.225 & 0.346 + [ t : cvsurveylin ] the next step is to calculate the cosmic variance for dark matter for a set of mean redshifts @xmath1 . we do this using the angular dimensions of the five surveys presented in section [ sec : survey ] and the reference redshift bin size of @xmath50 . the cosmic variance is then computed for a set of mean redshifts . the resulting values of the root cosmic variance for dark matter are given in the third column of table [ t : cvsurveylin ] and are plotted in figure [ fig : cv ] ( symbols ) . to make it possible to obtain the cosmic variance for mean redshifts other than these tabulated values , we introduce a fitting function : @xmath53 l c c c c c c udf & & 0.251 & & 0.364 & & 0.358 + goods & & 0.261 & & 0.854 & & 0.684 + gems & & 0.161 & & 0.520 & & 0.729 + egs & & 0.128 & & 0.383 & & 0.673 + cosmos & & 0.069 & & 0.234 & & 0.834 + [ tab : meanredshift ] the parameters @xmath54 , @xmath55 and @xmath56 depend on the angular dimensions of the field and are given in table [ tab : meanredshift ] . the lines in figure [ fig : cv ] show the fitting function for the five surveys and agree very well with the computed values for the redshift bins . using equation for any field we are thus able to calculate the dark matter cosmic variance as a function of mean redshift . up until now we have assumed a reference redshift bin size of @xmath57 in order to determine the size of the volume element under consideration . however , especially at higher redshift , larger bin sizes are desirable . we thus investigate how the cosmic variance depends on @xmath2 . for this we plot @xmath58 as a function of @xmath59 for different fields and mean redshifts in figure [ fig : deltaz ] . we find that independent of angular dimensions and mean redshift the cosmic variance has the same dependance on the redshift bin size : @xmath60 this means that for a reference redshift bin size of @xmath61 as assumed in equation the root cosmic variance can be calculated for a different @xmath2 as : @xmath62 there is a simple picture that explains the dependence on @xmath63 . if we divide a redshift bin of size @xmath2 into @xmath29 redshift bins with size @xmath64 then @xmath65 . assuming that all these redshift bins are uncorrelated , we can invoke equation which yields @xmath66 . using @xmath67 we finally get @xmath68 . it is interesting that the simplified assumption of uncorrelated volume elements yields a relation that fits so well to the results calculated for the total redshift bin , as seen in figure [ fig : deltaz ] . however , we note that this approximation is only valid for redshift bin sizes @xmath69 . for smaller bin sizes we refer to the more accurate model presented in section 3.5 of @xcite . the last step to derive the cosmic variance for a given sample of galaxies is to apply the galaxy bias . as we have shown in equation , in the linear - biasing limit , the cosmic variance for a sample of galaxies with a given stellar mass is the product of the squared bias and the dark matter cosmic variance . the stellar mass dependent galaxy bias has been derived in @xcite . the authors present the redshift dependence of the bias using parameterized functions of the form : @xmath70 with the parameters @xmath71 , @xmath72 , and @xmath73 which are given in table [ tab : biasfit ] for six stellar mass bins and six stellar mass thresholds . in order to compute the root cosmic variance for galaxies of that mass , we have to multiply the bias and the root cosmic variance of dark matter for the same redshift . this has been done for the five surveys presented in table [ t : cvsurveylin ] : the cosmic variance for galaxies of mass @xmath3 is given in columns 4 - 9 for the reference redshift bin size of @xmath74 . cosmic variance for massive galaxies is always larger than that of low mass galaxies , since the galaxy bias increases with increasing stellar mass . lccc 8.5 - 9.0 & 0.062 @xmath75 0.017 & 2.59 @xmath75 0.18 & 1.025 @xmath75 0.062 + 9.0 - 9.5 & 0.074 @xmath75 0.008 & 2.58 @xmath75 0.26 & 1.039 @xmath75 0.028 + 9.5 - 10.0 & 0.042 @xmath75 0.003 & 3.17 @xmath75 0.05 & 1.147 @xmath75 0.021 + 10.0 - 10.5 & 0.053 @xmath75 0.014 & 3.07 @xmath75 0.17 & 1.225 @xmath75 0.077 + 10.5 - 11.0 & 0.069 @xmath75 0.014 & 3.19 @xmath75 0.13 & 1.269 @xmath75 0.087 + 11.0 - 11.5 & 0.173 @xmath75 0.035 & 2.89 @xmath75 0.20 & 1.438 @xmath75 0.061 + @xmath76 & 0.063 @xmath75 0.008 & 2.62 @xmath75 0.08 & 1.104 @xmath75 0.028 + @xmath77 & 0.085 @xmath75 0.008 & 2.50 @xmath75 0.06 & 1.098 @xmath75 0.064 + @xmath78 & 0.058 @xmath75 0.005 & 2.96 @xmath75 0.06 & 1.192 @xmath75 0.021 + @xmath79 & 0.072 @xmath75 0.013 & 2.90 @xmath75 0.13 & 1.257 @xmath75 0.051 + @xmath80 & 0.093 @xmath75 0.015 & 3.02 @xmath75 0.11 & 1.332 @xmath75 0.054 + @xmath81 & 0.185 @xmath75 0.032 & 2.86 @xmath75 0.25 & 1.448 @xmath75 0.098 + [ tab : biasfit ] finally we can summarize our recipe to derive the cosmic variance for a particular survey : step 1 ) : : choose the survey ( field ) in table [ t : surveyarea ] that is closest to the field you are using . step 2 ) : : choose a mean redshift @xmath82 and a redshift bin size @xmath2 . step 3 ) : : choose the stellar mass range of your galaxy sample from table [ tab : biasfit ] . step 4a ) : : calculate the dark matter root cosmic variance @xmath83 using equation and the parameters for your survey as given in table [ tab : meanredshift ] . step 4b ) : : calculate the galaxy bias @xmath84 using equation and the parameters for your stellar mass bin or threshold as given in table [ tab : biasfit ] . step 4c ) : : compute the root cosmic variance for galaxies : @xmath85 step 4d ) : : combine widely separated fields using equation . this recipe allows for easy computation of cosmic variance for the representative surveys we have presented . to enable readers to compute cosmic variance for different survey areas or geometries , we provide a software tool as an online supplement . for specified angular dimensions and redshift bins the code computes the cosmic variance for dark matter and galaxies in six stellar mass bins . having derived the recipe to determine the cosmic variance , we can apply it to the five example surveys . we first give an overview of the surveys . the udf is an exposure of a square field in the southern sky covering a total area of 11@xmath86 with the advanced camera for surveys ( acs ) on the hubble space telescope ( hst ) . to date it is the deepest image of the universe with uniform limiting magnitudes @xmath87 for point sources and contains at least 10,000 objects . however , of the five surveys studied here , the udf is the smallest . the goods project covers two fields , the hubble deep field north ( hdfn ) and the chandra deep field south ( cdfs ) which both have similar dimensions of @xmath88 and are so widely separated as to be uncorrelated . since for widely separated fields the cosmic variance goes as @xmath89 , the total variance for both fields decreases by a factor of 2 . the gems survey has imaged the e - cdfs , a square area of @xmath90 , with the acs on the hst . the e - cdfs is centered on the cdfs and contains roughly 10,000 galaxies down to a depth of 24th magnitude in the r - band . aegis is a multi - wavelength , deep , wide - field , photometric and spectroscopic survey in the extended groth strip ( egs ) area . while the total area is similar to that of gems , the field is a long strip with dimensions of @xmath91 . the widest survey for which we compute cosmic variance is cosmos which covers an equatorial square field of @xmath92 with imaging by most of the major space - based telescopes and several large ground - based telescopes . the magnitude limit in the i - band is @xmath93 . if we apply this recipe to the udf , goods , gems , aegis and cosmos at @xmath4 and @xmath5 , the relative cosmic variance for massive galaxies with @xmath6 is @xmath94 for the udf , while it is @xmath95 for goods , @xmath8 for gems , @xmath96 for aegis and @xmath9 for cosmos . for galaxies of intermediate mass ( @xmath10 ) the relative cosmic variance is @xmath97 for the udf , @xmath11 for goods , @xmath12 for gems , @xmath98 for aegis and @xmath13 for cosmos . this implies that at @xmath4 cosmic variance is a significant source of uncertainty for small fields and massive galaxies . however , for larger fields and intermediate mass galaxies cosmic variance is less serious . at a higher redshift of @xmath99 and @xmath5 the relative cosmic variance is higher , since the galaxy bias is much larger . for massive galaxies with @xmath6 , we find a relative cosmic variance of @xmath100 for the udf , @xmath101 for goods , @xmath102 for gems , @xmath103 for aegis and @xmath104 for cosmos . for intermediate mass galaxies with @xmath10 the relative cosmic variance is @xmath105 for the udf , @xmath106 for goods , @xmath107 for gems , @xmath108 for aegis and @xmath109 for cosmos . this shows that at @xmath99 cosmic variance can be serious in small fields even for intermediate mass galaxies . for massive galaxies even the widest survey ( cosmos ) has considerable uncertainties due to cosmic variance . it is not very useful to compare the cosmic variance at different redshifts for a fixed bin size , because the volumes are not the same . since cosmic variance strongly depends on the volume of the sample under consideration , we choose redshift bins such that all redshift bins have the same volume . this allows us to directly compare the cosmic variance at different redshifts . the results of this analysis are given in table [ t : cvsurveyvol ] for the goods , gems and cosmos surveys . c c c c c c c c c + & & & + @xmath110 & @xmath111 & @xmath43 & & & & & + & & & 8.75&9.25&9.75&10.25&10.75&11.25 + + 0.00 & 1.12 & 0.081 & 0.099 & 0.103 & 0.107 & 0.116 & 0.126 & 0.168 + 1.12 & 1.58 & 0.084 & 0.133 & 0.143 & 0.148 & 0.163 & 0.194 & 0.291 + 1.58 & 1.99 & 0.078 & 0.149 & 0.163 & 0.174 & 0.192 & 0.241 & 0.374 + 1.99 & 2.39 & 0.073 & 0.165 & 0.183 & 0.204 & 0.225 & 0.295 & 0.464 + 2.39 & 2.78 & 0.069 & 0.186 & 0.208 & 0.243 & 0.267 & 0.365 & 0.573 + 2.78 & 3.17 & 0.065 & 0.209 & 0.235 & 0.290 & 0.316 & 0.446 & 0.696 + 3.17 & 3.58 & 0.060 & 0.231 & 0.262 & 0.339 & 0.368 & 0.534 & 0.823 + 3.58 & 4.00 & 0.056 & 0.260 & 0.296 & 0.404 & 0.436 & 0.647 & 0.983 + + 0.00 & 1.12 & 0.063 & 0.077 & 0.080 & 0.083 & 0.091 & 0.098 & 0.130 + 1.12 & 1.58 & 0.060 & 0.096 & 0.103 & 0.107 & 0.118 & 0.140 & 0.210 + 1.58 & 1.99 & 0.055 & 0.105 & 0.114 & 0.122 & 0.135 & 0.169 & 0.262 + 1.99 & 2.39 & 0.050 & 0.114 & 0.126 & 0.140 & 0.155 & 0.203 & 0.319 + 2.39 & 2.78 & 0.047 & 0.126 & 0.141 & 0.165 & 0.181 & 0.247 & 0.389 + 2.78 & 3.17 & 0.043 & 0.140 & 0.158 & 0.194 & 0.212 & 0.299 & 0.466 + 3.17 & 3.58 & 0.040 & 0.153 & 0.173 & 0.225 & 0.244 & 0.354 & 0.545 + 3.58 & 4.00 & 0.037 & 0.171 & 0.194 & 0.265 & 0.286 & 0.425 & 0.645 + + 0.00 & 1.12 & 0.036 & 0.044 & 0.046 & 0.048 & 0.052 & 0.057 & 0.075 + 1.12 & 1.58 & 0.030 & 0.047 & 0.051 & 0.053 & 0.058 & 0.069 & 0.104 + 1.58 & 1.99 & 0.026 & 0.049 & 0.054 & 0.058 & 0.064 & 0.080 & 0.124 + 1.99 & 2.39 & 0.023 & 0.052 & 0.057 & 0.064 & 0.070 & 0.093 & 0.145 + 2.39 & 2.78 & 0.021 & 0.056 & 0.063 & 0.074 & 0.081 & 0.110 & 0.173 + 2.78 & 3.17 & 0.019 & 0.061 & 0.069 & 0.085 & 0.093 & 0.131 & 0.204 + 3.17 & 3.58 & 0.017 & 0.066 & 0.075 & 0.097 & 0.105 & 0.153 & 0.235 + 3.58 & 4.00 & 0.016 & 0.073 & 0.083 & 0.113 & 0.122 & 0.181 & 0.275 + [ t : cvsurveyvol ] we would expect the cosmic variance of dark matter to decrease monotonically with increasing redshift , because it depends on the clustering which is lower for high redshift . however , we see , that for the udf @xmath43 increases , reaches a maximum and then decreases with increasing redshift , while for the other surveys @xmath43 decreases with increasing redshift . the reason for this is that although the different redshift bins have the same comoving volume , their geometry is not the same . this is due to the pencil beam geometry for which the volume of the low redshift bins are long but narrow , while the higher redshift bins become shorter and wider . thus with increasing redshift the ratio of the area and the depth of the redshift bins increases . in the lower redshift bins the galaxies are thus on average more separated than in the higher redshift bins . this causes the cosmic variance to increase with redshift . at high redshift and for very wide surveys the ratio decreases again with redshift as the line of sight distance in a redshift bin becomes smaller than the transverse distance . both effects , the decrease of @xmath43 with redshift due to lower clustering amplitudes and the growth of @xmath43 because of the geometry of the redshift bins , affect @xmath112 . the redshift at which @xmath43 is maximal depends on the area of the survey . for small areas ( e.g. udf ) it is higher ( @xmath113 ) while for large areas ( e.g. cosmos ) it is lower ( @xmath114 ) . the cosmic variance for the galaxy samples we consider , however , always increases with redshift . this is due to the large biases of these samples , which are always larger than unity . since the bias increases strongly with redshift , the cosmic variance for galaxies thus also increases with redshift . figure [ fig : cvsurvey ] plots the cosmic variance as a function of redshift with redshift bins of constant volume . the four different panels show @xmath115 for dark matter and three stellar mass bins . the lines are for different surveys . we see that except for the udf , the cosmic variance for dark matter decreases with redshift while for all surveys the cosmic variance for galaxies increases with redshift . as we have shown , cosmic variance depends strongly on the geometry of the survey . in section [ sec : redshift ] we investigated the geometric effect that arises because of the ratio between line - of - sight distance and the transverse distance in the redshift bins . now we investigate the effects that arise due to different angular geometry on the sky . for this we assume a survey with a fixed total area of @xmath116 . we vary the ratio between the two observation angles @xmath117 from @xmath118 to @xmath119 . this is done for several mean redshifts with a bin size of @xmath5 . figure [ fig : ratio ] shows the geometry effects on the cosmic variance . it plots @xmath43 as a function of the ratio between the two observation angles . the result agrees with our explanation in section [ sec : redshift ] : for a low ratio the cosmic variance is very low . if one increases the ratio the cosmic variance quickly increases and reaches its maximum at @xmath119 . here the mean distance between the observed galaxies is the smallest , so that many galaxies are likely to be correlated , resulting in large cosmic variance . this effect can also be seen by comparing the cosmic variance for gems and egs at a fixed redshift bin . we find that although the area of the gems field is larger than that of the egs , gems has a higher cosmic variance . however , we note that unless the ratio between the observation angles is smaller than @xmath120 , this effect is small . in agreement with @xcite , we find that the size of the survey area is much more important than the axis ratio . inferences about galaxy number densities and related quantities such as luminosity and stellar mass functions are subject to uncertainties due to cosmic variance . it can be a significant source of uncertainty , especially for surveys which cover only small areas and for massive high - redshift galaxies . we have derived a simple recipe to compute cosmic variance for five surveys as a function of mean redshift @xmath1 , redshift bin size @xmath2 and the stellar mass of the galaxy population @xmath3 . for this recipe we first calculated the dark matter cosmic variance as a function of mean redshift and for a reference bin size of @xmath121 . this was done by integrating the dark matter correlation function obtained from cdm theory for a pencil beam geometry in a redshift bin . we provide fitting functions for five surveys ( udf , goods , gems , aegis , cosmos ) . we then showed that the dependence of cosmic variance on bin size @xmath2 is nearly independent of mean redshift and survey geometry . this means that cosmic variance for a different bin size can be estimated from the value computed with the reference bin size using a simple conversion factor . we have used the galaxy bias predictions by @xcite ( for six stellar mass bins and six thresholds ) and the dark matter cosmic variance to compute the cosmic variance for galaxies . we also presented a formula to compute the overall cosmic variance for a volume - weighted average of multiple separated fields . applying this recipe to goods , gems and cosmos at @xmath4 and @xmath5 , the relative cosmic variance for galaxies with @xmath6 ( @xmath10 ) is @xmath95 ( @xmath122 ) for goods , while it is @xmath8 ( @xmath123 ) for gems and @xmath9 ( @xmath124 ) for cosmos . at @xmath99 and @xmath5 we found a relative cosmic variance for galaxies with @xmath6 ( @xmath10 ) of @xmath101 ( @xmath125 ) for goods , @xmath126 ( @xmath127 ) for gems and @xmath104 ( @xmath128 ) for cosmos . we find that for all fields the cosmic variance is much more significant for massive galaxies , which is due to the strong dependence of the galaxy bias on stellar mass . our results imply that cosmic variance is a significant source of uncertainty at @xmath4 for small fields and massive galaxies , while for larger fields and intermediate mass galaxies , cosmic variance is less serious . at @xmath99 cosmic variance can become serious in small fields even for intermediate mass galaxies . for massive galaxies even the widest survey ( cosmos ) has significant uncertainties due to cosmic variance .
deep pencil beam surveys ( @xmath0 ) are of fundamental importance for studying the high - redshift universe . however , inferences about galaxy population properties ( e.g. the abundance of objects ) are in practice limited by cosmic variance . this is the uncertainty in observational estimates of the number density of galaxies arising from the underlying large - scale density fluctuations . this source of uncertainty can be significant , especially for surveys which cover only small areas and for massive high - redshift galaxies . cosmic variance for a given galaxy population can be determined using predictions from cold dark matter theory and the galaxy bias . in this paper we provide tools for experiment design and interpretation . for a given survey geometry we present the cosmic variance of dark matter as a function of mean redshift @xmath1 and redshift bin size @xmath2 . using a halo occupation model to predict galaxy clustering , we derive the galaxy bias as a function of mean redshift for galaxy samples of a given stellar mass range . in the linear regime , the cosmic variance of these galaxy samples is the product of the galaxy bias and the dark matter cosmic variance . we present a simple recipe using a fitting function to compute cosmic variance as a function of the angular dimensions of the field , @xmath1 , @xmath2 and stellar mass @xmath3 . we also provide tabulated values and a software tool . we find that for goods at @xmath4 and with @xmath5 the relative cosmic variance of galaxies with @xmath6 is @xmath7 , while it is @xmath8 for gems and @xmath9 for cosmos . for galaxies of @xmath10 the relative cosmic variance is @xmath11 for goods , @xmath12 for gems and @xmath13 for cosmos . this implies that cosmic variance is a significant source of uncertainty at @xmath4 for small fields and massive galaxies , while for larger fields and intermediate mass galaxies cosmic variance is less serious .
the extragalactic background ir field potentially contains a wealth of information relevant to both cosmology and particle physics ( see for example @xcite ) , but has so far eluded any conclusive detection . direct measurement of the characteristics of this radiation are frustrated by the dominance of local , galactic ir sources . tev gamma - ray astronomy provides a means to study the ir background indirectly , free of such complications , by looking for modifications to high energy gamma ray spectra due to interactions with this background field via the process @xmath9 . this idea was first noted by gould and schrder @xcite and has been more recently restated by stecker and de jager @xcite . previous ir studies making use of this phenomenon have generally used spectral data from the active galactic nucleus ( agn ) mrk 421 taken by the egret @xcite and whipple @xcite experiments at gev and tev energies , respectively . these studies have relied on extrapolating flux measurements between these two energy regimes in an effort to determine the inherent shape of the source spectrum @xcite @xcite @xcite @xcite . this approach has several difficulties : * 1 ) * since the flux level is known to be highly variable at tev energies , the relative normalization between egret and whipple flux measurements is not known ; * 2 ) * the flux of ir photons can not be assumed to be zero in the intervening energy range ; and * 3 ) * the shape of the inherent source spectrum is not known and can not be assumed to necessarily follow a single power law from gev to tev energies . indeed , current best fit values for mrk 421 at tev energies indicate a steeper spectrum than is reflected by the egret data @xcite , suggesting that the spectral shape changes in the intervening energy range . in addition , most of the analyses have assumed a particular model to describe the shape of the ir photon spectrum . even a more recent attempt at a less model - dependent limit using data from mrk 501 nevertheless assumes a power - law source spectrum , an arbitrary ir flux normalization and constant ir energy densities over energy ranges spanning up to an order of magnitude in extent @xcite . limits derived in this manner may not apply if the background ir spectrum is of a different shape and , in particular , can not necessarily be used to constrain mechanisms which might produce sharp features in the spectrum on a scale smaller than that of the assumed model . this current paper will attempt to define more robust limits based on recent high energy observations of agn which do not rely on extrapolation from lower energies or the assumption of a strict power - law source spectrum and are independent of the ir model down to scales corresponding to a factor of 2 change in ir energy . these limits will then be used to constrain significantly the possible contribution of very massive objects ( vmos ) to the dark matter problem and also to place improved bounds on the radiative decay of massive neutrinos in the mass range above @xmath10 ev , which is in the regime of interest to the atmospheric neutrino problem . two active galaxies , mrk 421 and mrk 501 , have thus far been seen to produce tev gamma - ray emission by independent experimental groups . these active galaxies are both of the blazar class and are located at redshifts of 0.031 and 0.033 , respectively . both objects have exhibited significant flux variability at tev energies , changing by an order of magnitude or more with characteristic timescales of less than an hour ( in the case of mrk 421 , as short as 15 minutes ) . recent analyses of data from mrk 421 by the whipple group @xcite @xcite indicate a differential spectrum with an @xmath11 power law extending to energies beyond 5 tev , possibly as high as 8 tev . similarly , the hegra collaboration has measured the `` high - state '' spectrum of mrk 501 @xcite and has found it to be described by an @xmath12 power law extending to energies as high as 10 tev . for the purposes of this paper , the available data on active galaxies at a redshift of @xmath13 will thus be taken to indicate a spectral shape which is bounded at the 90% confidence level between the power laws @xmath14 and @xmath15 and normalized by an arbitrary flux constant . it will be assumed that the measured spectra continue up to energies of 10 tev . however , due to possible systematic uncertainties in energy scale of @xmath1630% , more conservative limits will also be computed assuming a maximum gamma - ray energy of 6 tev . finally , it will be assumed that the inherent differential source spectrum is steeper than an @xmath17 power law , although the specific shape need not be a power law itself . the limits derived are relatively insensitive to this latter choice . for example selecting @xmath18 as a bound was found to change the inferred limits by less than 30% . for ease of computation , the high energy spectrum is taken to consist of differential flux measurements at the following discrete energies : 0.4 , 0.75 , 1.0 , 1.5 , 2.0 , 3.0 , 4.0 , 6.0 and 10.0 tev . the spectrum of ir photons is approximated by a series of separate intervals spanning the energies 0.025 - 0.042 , 0.042 - 0.083 , 0.083 - 0.167 , 0.167 - 0.33 , 0.33 - 0.625 and 0.625 - 1.0 ev . within each interval , the value of @xmath19 is taken as constant , where @xmath20 is the energy of the ir photon and @xmath21 is the number density of photons ( _ i.e. _ photons/@xmath22/ev ) . the value of @xmath19 among these intervals can thus be varied to reflect different background ir models , and the corresponding effect of @xmath0-@xmath0 absorption on tev gamma - rays calculated according to the prescription described by previous papers @xcite , @xcite . for these calculations , a conservative value for the hubble constant of 85 km / s / mpc is assumed @xcite . to calculate upper limits then , the inherent agn differential spectrum was allowed to assume any shape steeper than @xmath17 and the values of @xmath19 among ir intervals were allowed to take on any values subject to the following constraints : * 1 ) * that these values are lower than the directly determined upper bounds imposed by other experiments ( shown in figure 1 ) ; and * 2 ) * that the attenuation implied by the ir model would not cause the shape of the observed tev spectrum to deviate outside of the power - law bounds previously defined based on the data . this approach essentially anchors the lower energy tev data to the appropriate observational upper bounds and then extends these bounds based on the shape and extent of the agn spectra at higher energies . from these considerations , the maximum value of energy density permitted in each ir interval was determined . the resulting limits are summarized in table 1 and are also plotted in figure 1 for assumed maximum gamma - ray energies of 6 tev and 10 tev . note that this choice of maximum energy primarily effects only the two lowest energy ir intervals . .derived upper limits to @xmath19 [ cols= " < , < , < " , ] in the range of @xmath23 ev , the limits presented here are up to 2 orders of magnitude more restrictive than those of dirbe @xcite . note that these limits apply only to models which would produce features in the ir spectrum on a scale comparable to or larger than the interval widths used in the calculation . monoenergetic line features , for example , are not necessarily constrained by these limits . under certain cosmological scenarios , density fluctuations in the early universe at a redshift of @xmath161000 could give rise to the pregalactic formation of massive stars ( @xmath24 ) at redshifts in the range of @xmath16100 - 300 @xcite . such very massive objects ( vmos ) would collapse to black holes after their main - sequence phase without any significant metal ejection . it has been suggested that the remnants of vmos , which would ultimately cluster with galaxies , might provide an explanation for the dark matter associated with galaxy rotation curves @xcite , @xcite . such rotation curves indicate that the amount of dark matter is in excess of 5 percent of the critical density , while other studies of galaxy motion and hot gas in clusters of galaxies suggest a value as high as 20 percent @xcite . while invisible now , vmos would have produced a significant flux of ir photons during their burning phase . the peak number density of these photons is given @xcite by : @xmath25 @xmath26 where @xmath27 is the ir photon energy ( in ev ) corresponding to the peak of the vmo contribution , @xmath28 is the photon number density due to vmos in @xmath29 , @xmath30 is the mass - density of vmos expressed as a fraction of the critical density , @xmath31 is the formation redshift of the vmos , and the hubble constant is given by @xmath32 km / s / mpc . thus , the ir limits derived previously may be used to place upper bounds on @xmath30 . these bounds are shown in figure 2 as a function of the assumed vmo formation redshift for @xmath33=0.65 ( which is conservative for this limit ) . note that for either maximum gamma - ray energy assumed , the limit on @xmath30 is less than 0.05 for the redshift range of 100 - 200 and less than 0.065 for the range 200 - 300 . if the maximum gamma - ray energy is taken to be 10 tev , the limit is less than 0.05 for the the entire 100 - 300 range . hence , assuming minimal reprocessing by dust grains , these limits appear to rule out vmo models as providing the explanation for observed galaxy rotation curves . several experiments have observed an apparent descrepency in the ratio of @xmath34 to @xmath35 fluxes expected from cosmic ray interactions in the atmosphere @xcite @xcite @xcite . a possible explanation for this effect may involve 2-component neutrino oscillations with @xmath36ev@xmath37 and , thus , a neutrino mass in the regime of 0.1 ev ( provided the mass of the lighter neutrino is comparable to or smaller than this value ) . if neutrinos have mass and different flavor eigenstates may mix , then massive neutrinos might decay via the process @xmath38 , where @xmath39 . for neutrino masses in the @xmath16ev regime , such a radiative decay process would contribute to the infrared background . although mono - energetic photons would be initially produced from a 2-body decay , the photon spectrum would be broadened for relic neutrinos which decay over a range of redshifts such that the background density can be constrained by the limits presented previously . some unstable relic neutrinos with a lifetime longer than that of the universe would still be decaying now and would have produced a background photon density given by : @xmath40 where @xmath41 is half the rest mass of the unstable neutrino in ev , @xmath42 is the photon number density due to neutrino decay in @xmath29 , @xmath43 is the branching ratio for the radiative decay mode and @xmath44 is the neutrino lifetime in years @xcite . implied lower bounds for @xmath45 in the neutrino mass range @xmath46 are shown in figure 3 based on the ir limits presented in this paper for @xmath47 . previous constraints in this mass range from the work of ressell and turner @xcite are shown for comparison . this represents an improvement of the constraints in some cases by more than 2 orders of magnitude . similarly , neutrinos with lifetimes shorter than the age of the universe but longer than the age at which decoupling of matter and radiation occurred will result in a background photon density given by : @xmath48 where , in this case , @xmath49 where @xmath50 is the mass - density of the universe in units of the critical density . ir limits given here have thus been used to define exclusion contours which can be conveniently characterised by @xmath51 for the region defined by @xmath52 where @xmath53 for an assumed maximum gamma - ray energy of 10 tev , and @xmath54 for a maximum energy of 6 tev . once again , this represents an improvement in branching ratio sensitivity of approximately two orders of magnitude . recent data from very high energy gamma - ray observations of active galaxies has been used to set improved upper limits to the infared background radiation which are robust and relatively model - independent . these limits cast doubt on vmo models as providing the explanation for dark matter in galactic halos and tighten constraints on the radiative decay of relic neutrinos in the mass range near 0.1 ev by approximately two orders of magnitude . substantial improvement in the ability to study the infrared background via @xmath0-@xmath0 interactions is expected with the next generation of instruments which will allow the behavior of more distant sources to be examined . in fact , the procedure used to derive the limits given in table 1 may also be used to determine lower limits to the maximum distance out to which tev gamma - ray telescopes can be expected to probe . this has been done for the gamma - ray energy range of 0.4 - 10 tev , assuming that a source remains `` visible '' out to an optical depth of 2 and that the majority of the ir was produced prior to the epoch of the sources under study . ir and uv upper bounds from direct observations in the energy range greater than 2 ev have also been used to calculate this minimum distance for gamma - ray energies of 0.05 and 0.1 tev . the results are shown in figure 4 . the current generation of ground - based , gamma - ray telescopes have a typical lower energy threshold of @xmath160.5 tev and are thus expected to be able to see sources possessing redshifts up to or beyond @xmath55=0.1 . the next generation of instruments is expected to have an energy threshold in the region of 0.05 to 0.1 tev , and will therefore be able to see out to a redshift of at least @xmath55=0.5 . this represents a substantially larger volume of space than has so far been explored at tev energies . this , and the known preponderance of agn at larger redshifts , indicates a bright future for ground - based , gamma - ray astronomy and strongly suggests a drive towards higher duty cycles , larger apertures and greater dynamic ranges for such instruments to explore this regime thoroughly .
from considering the effect of @xmath0-@xmath0 interactions on recently observed tev gamma - ray spectra , improved limits are set to the density of extragalactic infrared ( ir ) photons which are robust and essentially model - independent . the resulting limits are up to two orders of magnitude more restrictive than direct observations in the 0.025 - 0.3 ev regime . these limits are used to improve constraints on radiative neutrino decay in the mass range above 0.05ev and on very massive objects ( vmos ) as providing the dark matter needed to explain galaxy rotation curves . = 8.0 in = 5.5 in new limits to the ir background : + bounds on radiative neutrino decay and on vmo contributions to the dark matter problem .5 in s.d . biller,@xmath1 j. buckley,@xmath2 a. burdett,@xmath3 j. bussons gordo,@xmath4 d.a . carter - lewis,@xmath5 d.j . fegan,@xmath4 j. finley,@xmath6 j.a . gaidos,@xmath6 a.m. hillas,@xmath3 f. krennrich,@xmath5 r.c . lamb,@xmath7 r. lessard,@xmath6 j.e . mcenery,@xmath4 g. mohanty,@xmath5 j. quinn,@xmath4 a.j . rodgers,@xmath3 h.j . rose,@xmath3 f. samuelson,@xmath5 g. sembroski,@xmath6 p. skelton,@xmath3 t.c . weekes,@xmath8 j. zweerink@xmath5 + _ @xmath1 oxford university , oxford ( uk ) + @xmath2 washington university , st . louis , mo ( usa ) + @xmath3 university of leeds , leeds ( uk ) + @xmath4 university college , dublin ( ireland ) + @xmath5 iowa state university , ames , ia ( usa ) + @xmath6 purdue university , west lafayette , in ( usa ) + @xmath7 california institute of technology , california(usa ) + @xmath8 whipple observatory , amado , az ( usa ) + _ .5 in 0.5 in _ to appear in physical review letters _
the study of hardy spaces originated in the 1910 s and at the very beginning was confined to fourier series and complex analysis in one variable . since 1960 s , it has been transferred to real analysis in several variables , or more generally to analysis on metric measure spaces . there are many different equivalent definitions of hardy spaces , which involve suitable maximal functions , the atomic decomposition , the molecular decomposition , singular integrals , square functions etc . see , for instance , the classical references @xcite . more recently , a lot of work has been devoted to the theory of hardy spaces associated with operators , see for example , @xcite and the references therein . in @xcite , auscher , mcintosh and russ studied hardy spaces with respect to the hodge laplacian on riemannian manifolds with the doubling volume property by using the davies - gaffney type estimates . they defined hardy spaces of differential forms of all degrees via molecules and square functions , on which the riesz transform is @xmath0 bounded for @xmath4 . comparing with the lebesgue spaces , it holds that @xmath5 for @xmath6 and @xmath7 for @xmath8 . moreover , under the assumption of gaussian heat kernel upper bound , @xmath0 coincides @xmath1 for @xmath9 . in @xcite , hofmann , lu , mitrea , mitrea and yan further developed the theory of @xmath2 and @xmath10 spaces adapted to a metric measure space @xmath11 with the volume doubling property endowed with a non - negative self - adjoint operator @xmath12 , which generates an analytic semigroup @xmath13 satisfying the so - called davies - gaffney estimate : there exist @xmath14 such that for any open sets @xmath15 , and for every @xmath16 with @xmath17 , @xmath18 , @xmath19 . the authors extended results of @xcite by obtaining an atomic decomposition of the @xmath2 space . more generally , instead of , if @xmath20 satisfies the davies - gaffney estimate of order @xmath21 with @xmath22 : for all @xmath23 and for all @xmath24 , @xmath25 kunstmann and uhl @xcite defined hardy spaces via square functions and via molecules adapted to , where the two @xmath2 spaces are also equivalent . here and in the sequel , @xmath26 denotes the ball of centre @xmath27 and radius @xmath28 and @xmath29 . in addition , if the @xmath30 off - diagonal estimates of order @xmath21 holds : for all @xmath23 and for all @xmath24 , @xmath31 with @xmath32 the conjugate of @xmath33 , then the hardy space @xmath0 defined via square functions coincides with @xmath1 for @xmath34 . however , there are natural examples where , one finds a mix of both behaviours and , in which case the previous hardy space theory does nt apply . for example , on fractal manifolds , the heat kernel behaviour is locally gaussian and at infinity sub - gaussian ( see section [ hk estimates ] for more details ) . we aim to develop a proper hardy space theory for this setting . an important motivation for our hardy spaces theory is to study the riesz transform on fractal manifolds , where the weak type @xmath35 boundedness has recently been proved in a joint work by the author with coulhon , feneuil and russ @xcite . in this paper , we work on doubling metric measure spaces endowed with a non - negative self - adjoint operator which satisfy the doubling volume property and the @xmath36 off - diagonal estimate with different local and global decay ( see below ) . the specific description will be found below in section [ setting ] . we define two classes of hardy spaces in this setting , via molecules and via conical square functions , see setion [ definitions ] . both definitions have the scaling adapted to the off - diagonal decay . in section 3 , we identify the two different @xmath2 spaces . the molecular @xmath2 spaces are always convenient spaces to deal with riesz transform and other sub - linear operators , while the @xmath0 , @xmath37 , spaces defined via conical square functions possess certain good properties like real and complex interpolation . the identification of both spaces gives us a powerful tool to study the riesz transform , littlewood - paley functions , boundary value problems for elliptic operators etc . in section 4 , we compare the hardy spaces defined via conical square functions with the lebesgue spaces . assuming further an @xmath38 off - diagonal estimate for some @xmath39 with different local and global decay for the heat semigroup , we show the equivalence of our @xmath0 spaces and the lebesgue spaces @xmath1 for @xmath40 . we also justify that the scaling for the hardy spaces is the right one , by disproving this equivalence of @xmath0 and @xmath1 for @xmath41 close to @xmath42 on some fractal riemannian manifolds . as far as we know , no previous results are known in this direction . in section 5 , we shall apply our theory to prove that the riesz transform is @xmath43 bounded on fractal manifolds . the proof is inspired by @xcite ( see @xcitefor the original proof in the discrete setting ) , where the integrated estimate for the gradient of the heat kernel plays a crucial role . in the following , we will introduce our setting , the definitions and the main results more specifically . * notation * throughout this paper , we denote @xmath44 if @xmath45 and @xmath46 , where @xmath46 means that there exists a constant @xmath47 ( independent of the important parameters ) such that @xmath48 . for a ball @xmath49 with radius @xmath28 and given @xmath50 , we write @xmath51 as the ball with the same centre and the radius @xmath52 . we denote @xmath53 , and @xmath54 for @xmath55 . we shall assume that @xmath20 is a metric measure space satisfying the doubling volume property : for any @xmath27 and @xmath56@xmath57 and the @xmath36 davies - gaffney estimate with different local and global decay for the analytic semigroup @xmath13 generated by the non - negative self - adjoint operator @xmath12 , that is , @xmath58 , @xmath59 where @xmath60 and @xmath61 recall a simple consequence of : there exists @xmath62 such that @xmath63 it follows that @xmath64 therefore , @xmath65 if @xmath20 is non - compact , we also have a reverse inequality of ( see for instance @xcite ) . that is , there exists @xmath66 such that @xmath67 also notice that in , if necessary we may smoothen @xmath68 as @xmath69 with @xmath70 for @xmath71 , which we still denote by @xmath68 . since @xmath72 for @xmath73 and @xmath74 for @xmath75 , we have in a uniform way @xmath76 we say that @xmath20 satisfies an @xmath30 off - diagonal estimate for some @xmath77 if @xmath78 and a generalized pointwise sub - gaussian heat kernel estimate if for all @xmath79 , @xmath80 examples of fractal manifolds satisfy with @xmath81 and @xmath82 , see section 2 below for more information . recall that [ mol ] let @xmath83 and an integer @xmath84 be an integer such that @xmath85 , where @xmath86 is in . a function @xmath87 is called a @xmath88molecule associated to @xmath12 if there exist a function @xmath89 and a ball @xmath90 with radius @xmath91 such that 1 . @xmath92 ; 2 . it holds that for every @xmath93 and @xmath94 , we have @xmath95 [ molh ] we say that @xmath96 is a molecular @xmath88representation of @xmath97 if @xmath98 , each @xmath99 is a molecule as above , and the sum converges in the @xmath36 sense . we denote the collection of all the functions with a molecular representation by @xmath100 , where the norm of @xmath101 is given by @xmath102 the hardy space @xmath103 is defined as the completion of @xmath104 with respect to this norm . consider the following conical square function @xmath105 where the cone @xmath106 . we define first the @xmath107 adapted hardy space @xmath108 as the closure of the range of @xmath109 in @xmath107 norm , i.e. , @xmath110 . the hardy space @xmath111 , @xmath112 is defined as the completion of the set @xmath113 with respect to the norm @xmath114 . the @xmath111 norm is defined by @xmath115 . for @xmath116 , the operator @xmath117 is bounded on @xmath107 . indeed , for every @xmath118 , @xmath119 note that the second step follows from fubini theorem and in section 2.3 . the third step is obtained by using the fact : @xmath120 . the last one is a consequence of spectral theory . [ rem : sf ] the above definitions are similar as in @xcite ( also @xcite for @xmath121-forms on riemannian manifolds ) and @xcite . the difference is that we replace @xmath122 or @xmath123 by @xmath68 in and . in the case when @xmath124 , we denote @xmath125 by @xmath126 , that is , @xmath127 and denote @xmath128 by @xmath129 . we first obtain the equivalence between @xmath2 spaces defined via molecules and via square functions . [ h1equiv ] let @xmath20 be a metric measure space satisfying the doubling volume property and the @xmath36 off - diagonal heat kernel estimate . then @xmath130 , which we denote by @xmath131 . moreover , @xmath132 now compare @xmath111 and @xmath1 for @xmath133 . recall that on riemannian manifold satisfying the doubling volume property and the gaussian upper bound for the heat kernel of the operator , we have @xmath134 , @xmath135 , see for example ( * ? ? ? * theorem 8.5 ) for hardy spaces of @xmath136forms on riemannian manifold . however , in general , the equivalence is not known . it is also proved in @xcite that if the @xmath30 off - diagonal estimates of order @xmath21 holds , then the hardy space @xmath137 ( see remark [ rem : sf ] ) coincides with @xmath1 for @xmath34 . our result in this direction is the following : [ equihl ] let @xmath20 be a non - compact metric measure space as above . let @xmath138 and @xmath139 be as above . suppose that @xmath20 satisfies and . then @xmath140 for @xmath141 . if one assumes the pointwise heat kernel estimate , then theorems 1.5 and 1.6 yield the following . let @xmath20 be a non - compact metric measure space satisfying the doubling volume property and the pointwise heat kernel estimate . then @xmath130 , and @xmath140 for @xmath9 . in the following theorem , we show that for @xmath142 , the equivalence may not hold between @xmath1 and @xmath0 defined via conical square function @xmath126 with scaling @xmath122 . the counterexamples we find are certain riemannian manifolds satisfying and two - sided sub - gaussian heat kernel estimate : and its reverse , with @xmath81 and @xmath143 . notice that in this case , @xmath12 is the non - negative laplace - beltrami operator , which we denote by @xmath144 . for simplicity , we denote by @xmath145 and the two sided estimate by @xmath146 . also , we denote by @xmath147 the @xmath2 space defined via molecules @xmath148 , @xmath149 the @xmath0 space defined via square functions @xmath150 . [ noequiv ] let @xmath20 be a riemannian manifold with polynomial volume growth @xmath151 as well as two - sided sub - gaussian heat kernel estimate ( @xmath152 ) with @xmath153 , that is , @xmath145 and the matching lower estimate . then @xmath154 does nt hold for @xmath155 . as an application of this hardy space theory , we have [ thm2 ] let @xmath20 be a manifold satisfying the doubling volume property and the heat kernel estimate @xmath145 , @xmath156 , that is , the upper bound of @xmath157 . then the riesz transform @xmath158 is @xmath159 bounded . recall that under the same assumptions , it is proved in @xcite that the riesz transform is of weak type @xmath35 and thus @xmath1 bounded for @xmath142 . let us first give some examples that satisfy with @xmath160 . more examples of this case are metric measure dirichlet spaces , which we refer to @xcite for details . [ fm ] fractal manifolds . fractal manifolds are built from graphs with a self - similar structure at infinity by replacing the edges of the graph with tubes of length @xmath121 and then gluing the tubes together smoothly at the vertices . for instance , see @xcite for the construction of vicsek graphs . for any @xmath161 such that @xmath162 and @xmath163 , there exist complete connected riemannian manifolds satisfying @xmath164 for @xmath165 and with @xmath81 and @xmath143 in ( see @xcite and @xcite ) . [ cs ] cable systems ( quantum graphs ) ( see @xcite , ( * ? ? ? * section 2 ) ) . given a weighted graph @xmath166 , we define the cable system @xmath167 by replacing each edge of @xmath168 by a copy of @xmath169 joined together at the vertices . the measure @xmath170 on @xmath167 is given by @xmath171 for @xmath172 in the cable connecting @xmath173 and @xmath174 , and @xmath170 assigns no mass to any vertex . the distance between two points @xmath173 and @xmath174 is given as follows : if @xmath173 and @xmath174 are on the same cable , the length is just the usual euclidean distance @xmath175 . if they are on different cables , then the distance is @xmath176 ( @xmath177 is the usual graph distance ) , where the minimum is taken over all vertices @xmath178 and @xmath179 such that @xmath173 is on a cable with one end at @xmath178 and @xmath174 is on a cable with one end at @xmath179 . one takes as the core @xmath180 the functions in @xmath181 which have compact support and are @xmath182 on each cable , and sets @xmath183 let @xmath12 be the associated non - negative self - adjoint operator associated with @xmath184 and @xmath13 be the generated semigroup . then the associated kernel may satisfies . for example , the cable graph associated with the sierpinski gasket graph ( in @xmath185 ) satisfies @xmath186 . the following are some useful lemmas for the off - diagonal estimates . we first observe that @xmath187 @xmath187 for @xmath188 . indeed , [ bk ] let @xmath11 be a metric measure space satisfying the doubling volume property . let @xmath12 be a non - negative self - adjoint operator on @xmath189 . assume that holds . then for all @xmath190 , we have @xmath191 the estimate is equivalent to the @xmath192 off - diagonal estimate @xmath193 we refer to @xcite for the proof . in fact , we also have [ bk ] let @xmath11 satisfy . let @xmath12 be a non - negative self - adjoint operator on @xmath189 . assume that holds . then for all @xmath190 and @xmath194 , we have 1 . for any ball @xmath49 with radius @xmath28 , and any @xmath195 , @xmath196 2 . for all @xmath197 such that @xmath198 , @xmath199 we recall definitions and properties related to tent spaces on metric measure spaces with the doubling volume property , following @xcite , @xcite . let @xmath20 be a metric measure space satisfying . for any @xmath27 and for any closed subset @xmath200 , a saw - tooth region is defined as @xmath201 . if @xmath202 is an open subset of @xmath20 , then the tent " over @xmath202 , denoted by @xmath203 , is defined as @xmath204^c=\{(x , t)\in m \times ( 0,\infty ): d(x , o^c)\geq t\}.\ ] ] for a measurable function @xmath205 on @xmath206 , consider @xmath207 given @xmath208 , say that a measurable function @xmath209 if @xmath210 for simplicity , we denote @xmath211 by @xmath212 from now on . therefore , for @xmath213 and @xmath214 , write @xmath215 , we have @xmath216 consider another functional @xmath217 we say that a measurable function @xmath218 if @xmath219 . suppose @xmath220 , let @xmath221 be the conjugate of @xmath41 . then the pairing @xmath222 realizes @xmath223 as the dual of @xmath212 . denote by @xmath224_\theta $ ] the complex method of interpolation described in @xcite . then we have the following result of interpolation of tent spaces , where the proof can be found in @xcite . [ inter ] suppose @xmath225 , with @xmath226 and @xmath227 . then @xmath228_\theta = t_2^{p}(m).\ ] ] next we review the atomic theory for tent spaces which was originally developed in @xcite , and extended to the setting of spaces of homogeneous type in @xcite . a measurable function @xmath229 on @xmath230 is said to be a @xmath231atom if there exists a ball @xmath232 such that @xmath229 is supported in @xmath233 and @xmath234 [ decomp1 ] for every element @xmath235 there exist a sequence of numbers @xmath236 and a sequence of @xmath231atoms @xmath237 such that @xmath238 moreover , @xmath239 , where the implicit constants depend only on the homogeneous space properties of @xmath20 . finally , if @xmath240 , then the decomposition also converges in @xmath241 . in this section , we shall prove theorem [ h1equiv ] . that is , under the assumptions of and , the two @xmath2 spaces : @xmath103 and @xmath242 , are equivalent . we denote @xmath243 since @xmath103 and @xmath244 are completions of @xmath245 and @xmath246 , it is enough to show @xmath247 with equivalent norms . in the following , we will prove the two - sided inclusions seperately . before proceeding to the proof , we first note the lemma below to prove @xmath248 boundedness of an operator , which is an analogue of lemma 4.3 in @xcite . [ crit ] assume that @xmath249 is a linear operator , or a nonnegative sublinear operator , satisfying the weak - type @xmath250 bound @xmath251 and that for every @xmath252molecule @xmath253 , we have @xmath254 with constant @xmath47 independent of @xmath253 . then @xmath249 is bounded from @xmath245 to @xmath255 with @xmath256 consequently , by density , @xmath249 extends to be a bounded operator from @xmath103 to @xmath255 . for the proof , we refer to @xcite , which is also applicable here . we have the following theorem : let @xmath20 be a metric measure space satisfying the doubling volume property and the heat kernel estimate . then @xmath257 and @xmath258 first observe that @xmath259 . indeed , by definition [ mol ] , any @xmath260-molecule belongs to @xmath261 . thus any finite linear combination of molecules belongs to @xmath261 . since @xmath262 is the @xmath107 limit of finite linear combination of molecules , we get @xmath263 . it remains to show @xmath264 , that is , @xmath117 is bounded from @xmath265 to @xmath255 . note that @xmath117 is @xmath36 bounded by spectral theory ( see ) , it follows from lemma [ crit ] that it suffices to prove that , for any @xmath266-molecule @xmath253 , there exists a constant @xmath47 such that @xmath267 . in other words , one needs to prove @xmath268 , where @xmath269 assume that @xmath253 is a @xmath260-molecule related to a function @xmath270 and a ball @xmath90 with radius @xmath271 , that is , @xmath272 and for every @xmath273 and @xmath94 , it holds that @xmath274 similarly as in @xcite , we divide @xmath229 into four parts : @xmath275 here @xmath276 denotes the characteristic function and @xmath277 , @xmath278 . it suffices to show that for every @xmath279 , we have @xmath280 . firstly consider @xmath281 . observe that @xmath282 is supported on @xmath283 . indeed , denote by @xmath284 be the center of @xmath90 , then @xmath285 . also , it holds that @xmath286 here the second and the third inequalities follow from and the definition of molecules respectively . now applying the cauchy - schwarz inequality , then @xmath287 secondly for @xmath288 . for each @xmath278 , we have @xmath289 . in fact , @xmath290 . then @xmath291 we estimate @xmath292 with @xmath293 and @xmath294 respectively . firstly assume that @xmath294 . using again , we have @xmath295 assume now @xmath293 . note that @xmath296 . then it follows from lemma [ bk ] that @xmath297 the last inequality follows from . it follows from above that @xmath298 where @xmath299 depends on @xmath300 . therefore @xmath301 we estimate @xmath302 in a similar way as before except that we replace @xmath253 by @xmath303 . note that for each @xmath304 , we have @xmath305 . indeed , @xmath306 then @xmath307 when @xmath294 , by spectral theorem we get @xmath308 . and when @xmath293 , it holds @xmath309 . then we estimate @xmath310 in the same way as for , @xmath311 here @xmath299 in the second and the third lines are different . we can carefully choose @xmath299 in the second line to make sure that @xmath299 in the third line is positive . hence @xmath312 and @xmath313 it remains to estimate the last term @xmath314 . for each @xmath278 , we still have @xmath315 then we obtain as before that @xmath316 in fact , due to the doubling volume property , as well as the definition of molecules , we get @xmath317 hence @xmath318 this finishes the proof . we closely follow the proof of theorem 4.13 in @xcite and get [ cnt ] let @xmath20 be a metric measure space satisfying and . if @xmath320 , then there exist a sequence of numbers @xmath321 and a sequence of @xmath88molecules @xmath322 such that @xmath97 can be represented in the form @xmath323 , with the sum converging in @xmath107 , and @xmath324 where @xmath47 is independent of @xmath97 . in particular , @xmath325 . for @xmath326 , denote @xmath327 . then by the definition of @xmath328 , we have @xmath329 . from theorem [ decomp1 ] , we decompose @xmath205 as @xmath330 , where @xmath236 , @xmath237 is a sequence of @xmath231atoms supported in a sequence of sets @xmath331 , and the sum converges in both @xmath332 and @xmath241 . also @xmath333 for @xmath334 , by functional calculus , we have the following caldern reproducing formula " @xmath335 denote @xmath336 , then @xmath323 . since for @xmath337 , we have @xmath338 . thus we learn from lemma 4.12 in @xcite that the sum also converges in @xmath107 . we claim that @xmath339 are @xmath88molecules up to multiplication to some uniform constant . indeed , note that @xmath340 , where @xmath341 now we estimate the norm @xmath342 , where @xmath343 is the radius of @xmath344 . for simplicity we ignore the index @xmath345 . consider any function @xmath346 with @xmath347 , then for @xmath93 , @xmath348 in the last inequality , we apply hlder inequality as well as . we continue to estimate by using the definition of @xmath231atoms and the off - diagonal estimates of heat kernel . for @xmath349 , the above quantity is dominated by @xmath350 next for @xmath195 , the above estimate is controlled @xmath351 in the first inequality , we use lemma [ bk ] . since @xmath352 , the last inequality always holds for any @xmath83 . therefore , @xmath353 in this section , we will study the relations between @xmath354 , @xmath355 and @xmath356 under the assumptions of and . we first show that @xmath354 and @xmath355 are equivalent . next we give some examples such that @xmath354 and @xmath356 are not equivalent . more precisely , the inclusion @xmath357 may be false for @xmath142 . we will prove theorem [ equihl ] . that is , if @xmath20 satisfies and , then @xmath358 for @xmath141 . our main tool is the caldern - zygmund decomposition ( see for example ( * ? ? ? * corollaire 2.3 ) ) . [ c - z ] let @xmath11 be a metric measured space satisfying the doubling volume property . let @xmath359 and @xmath360 . let @xmath361 . then there exists a decomposition of @xmath97 , @xmath362 so that 1 . @xmath363 for almost all @xmath27 ; 2 . there exists a sequence of balls @xmath364 so that each @xmath365 is supported in @xmath366 , @xmath367 3 . @xmath368 ; 4 . @xmath369 and @xmath370 ; 5 . there exists @xmath371 such that each @xmath27 is contained in at most @xmath372 balls @xmath366 . due to the self - adjointness of @xmath12 in @xmath107 , we get @xmath373 , where the sum is orthogonal . under the assumptions and , we have @xmath374 and thus @xmath375 . indeed , for any @xmath376 , it holds @xmath377 as a consequence of lemma [ bk ] , we have that for all @xmath27 and @xmath378 , @xmath379 now letting @xmath380 , we obtain that @xmath381 . it suffices to prove that for any @xmath382 with @xmath141 , @xmath383 with this fact at hand , we can obtain by duality that @xmath384 for @xmath385 . indeed , for @xmath386 , write the identity @xmath387 where the integral @xmath388 converges to @xmath97 in @xmath107 as @xmath389 . then for @xmath382 , we have @xmath390 here @xmath391 and @xmath392 . the second line s equivalence is due to the doubling volume propsimeqerty . by an approximation process , the above argument holds for @xmath393 . for @xmath8 , the @xmath1 norm of the conical square function is controlled by its vertical analogue ( for a reference , see @xcite , where the proof can be adapted to the homogenous setting ) , which is always @xmath1 bounded for @xmath141 by adapting the proofs in @xcite and @xcite ( if @xmath13 is a symmetric markov semigroup , then it is @xmath1 bounded for @xmath9 , according to @xcite ) . hence holds . it remains to show for @xmath394 . in the following , we will prove the weak @xmath395 boundedness of @xmath117 by using the caldern - zygmund decomposition . since @xmath117 is also @xmath36 bounded as shown in , then by interpolation , holds for every @xmath396 . the proof is similar to ( * ? ? ? * proposition 6.8 ) and ( * ? ? ? * theorem 3.1 ) , which originally comes from @xcite . we take the caldern - zygmund decomposition of @xmath97 at height @xmath397 , that is , @xmath398 with @xmath399 . since @xmath117 is a sublinear operator , write @xmath400 here @xmath401 is chosen to be larger than @xmath402 , where @xmath86 is as in . then it is enough to prove that @xmath403 we treat @xmath404 in a routine way . since @xmath117 is @xmath36 bounded as shown in , then @xmath405 now for the second term . note that @xmath406 , it is enough to show that for every @xmath407 , @xmath408 note the following slight improvement of : for every @xmath409 and for every @xmath410 , we have @xmath411 here @xmath412 if @xmath413 , otherwise @xmath414 . indeed , it is obvious for @xmath415 and @xmath416 . for @xmath417 , that is , @xmath418 , then @xmath419 . with the above preparations , we can show now . write @xmath420 by a duality argument , @xmath421 applying cauchy - schwarz inequality , and , we get @xmath422 here @xmath423 denotes the hardy - littlewood maximal operator : @xmath424 where @xmath90 ranges over all balls containing @xmath173 . then @xmath425 the third inequality is due to the finite overlap of the caldern - zygmund decomposition . in the last line , for the first inequality , we use kolmogorov s inequality ( see for example @xcite ) . therefore , we obtain @xmath426 for the third term , we have @xmath427 from the caldern - zygmund decomposition and doubling volume property , we get @xmath428 it remains to show that @xmath429 as a consequence of the chebichev inequality , @xmath430 is dominated by @xmath431 for the estimate of @xmath432 . due to the bounded overlap of @xmath433 , we can put the sum of @xmath434 out of the square up to a multiplicative constant . that is , @xmath435 for the second inequality , note that for every @xmath434 , @xmath436 means @xmath437 . then @xmath438 and @xmath439 imply that @xmath440 . thus the integral is zero for every @xmath434 if @xmath441 . we obtain the third inequality by using the fubini theorem and . then by using , it follows @xmath442 for the second inequality , we use the reverse doubling property . the third inequality follows from the @xmath443 boundedness of the operator @xmath444 ( see lemma [ bk ] ) . then by using the @xmath445 boundedness of the heat semigroup , we get the fourth inequality . now for the global part . we split the integral into annuli , that is , @xmath446 in order to estimate the above @xmath36 norm , we use an argument of dualization . take the supremum of all functions @xmath447 with norm @xmath121 , then @xmath448 denote @xmath449 . let @xmath450 . then @xmath451 we will estimate @xmath452 by functional calculus . the notation is mainly taken from ( * ? ? ? * section 2.2 ) . for any fixed @xmath172 and @xmath271 , then @xmath453 is a holomorphic function satisfying @xmath454 for all @xmath455 with any @xmath456 . since @xmath12 is a nonnegative self - adjoint operator , or equivalently @xmath12 is a bisectorial operator of type @xmath457 , we can express @xmath458 by functional calculus . let @xmath459 , we have @xmath460 where @xmath461 is the half - ray @xmath462 and @xmath463 with @xmath464 being the half - ray @xmath465 . then for any @xmath466 , @xmath467 in the second inequality , the constant @xmath468 depends on @xmath469 and @xmath470 . indeed , @xmath471 . since @xmath472 , then @xmath473 and @xmath474 with @xmath475 . also it is obvious to see that @xmath476 . thus the second inequality follows . in the third inequality , let @xmath477 , we have @xmath478 . in addition , we dominate @xmath479 by @xmath480 . we choose @xmath469 appropriately such that @xmath481 for @xmath466 , then for any @xmath482 fixed , @xmath483 applying lemma [ bk ] , then @xmath484 in the second and the third lines , @xmath485 is originally defined in . in fact , it should be @xmath486 . since @xmath487 and @xmath488 are unanimously larger or smaller than one , we always have @xmath489 . hence , by minkowski inequality , we get from and that @xmath490 it remains to estimate the two integrals @xmath491 and @xmath492 . we claim that @xmath493 estimate first @xmath491 . since @xmath494 , we obtain @xmath495 it follows that @xmath496 in the first inequality , we dominate the exponential term by polynomial one for the first integral , where @xmath299 in the second line is chosen to be larger than @xmath497 . now estimate @xmath498 . write @xmath499 . on the one hand , @xmath500 on the other hand , we also have @xmath501 in fact , @xmath502 now we split the integral into two parts in the same way and control them by using and seperately . then @xmath503 therefore , it follows from and that @xmath504 now for the integral @xmath505 . take @xmath506 , then @xmath507 where @xmath423 is the hardy - littlewood maximal function . following the route for the proof of , we get from and that @xmath508 here the supremum is taken over all the functions @xmath509 with @xmath510 . since @xmath511 , the sum @xmath512 converges and we get the second inequality . the fourth one is a result of kolmogorov s inequality . thus we have shown @xmath513 . before moving forward to the proof of theorem [ noequiv ] , let us recall the following two theorems about the sobolev inequality and the green operator . [ hksob ] let @xmath515 be a @xmath516finite measure space . let @xmath517 be a semigroup on @xmath518 , @xmath519 , with infinitesimal generator @xmath520 . assume that @xmath517 is equicontinuous on @xmath3 and @xmath521 . then the following two conditions are equivalent : 1 . there exists @xmath522 such that @xmath523 , @xmath524 . @xmath525 is from @xmath3 to @xmath521 and for @xmath526 , @xmath527 such that @xmath528 where @xmath529 and @xmath530 . [ gr ] let @xmath20 be a complete non - compact manifold . then there exists a green s function @xmath531 which is smooth on @xmath532 satisfying @xmath533 for a proof , see for example @xcite . we also observe that [ lb ] let @xmath20 be a riemannian manifold satisfying the polynomial volume growth and the two - sided sub - gaussian heat kernel estimate @xmath146 . let @xmath90 be an arbitrary ball with radius @xmath534 . then there exists a constant @xmath468 depending on @xmath177 and @xmath21 such that for all @xmath172 with @xmath535 , @xmath536 note that for any @xmath537 , we have @xmath538 . then ( @xmath152 ) yields @xmath539 let @xmath540 be a cut - off function as follows : @xmath541 and for some @xmath542 , @xmath543 for simplicity , we denote @xmath544 by @xmath545 . taking @xmath546 , theorem [ gr ] says that @xmath547 . on the one hand , we apply theorem [ hksob ] by choosing @xmath548 . indeed , @xmath549 is markov hence bounded on @xmath1 , equicontinuous on @xmath550 and satisfies @xmath551 where @xmath552 . then taking @xmath553 and @xmath554 , it follows that @xmath555 where @xmath530 , that is , @xmath556 . using the fact that @xmath547 and @xmath557 , we get @xmath558 in particular , @xmath559 . on the other hand , @xmath560 since @xmath561 , it follows from the markovian property of the heat semigroup that @xmath562 by using lemma [ lb ] , it holds that @xmath563 if @xmath564 . then we get @xmath565 observe also that , for @xmath566 and @xmath567 , we have @xmath568 as long as @xmath569 is large enough . then the volume growth gives us a lower bound in terms of @xmath569 . that is , @xmath570 comparing the upper bound of @xmath571 in for @xmath554 , we obtain @xmath572 where @xmath554 . assume @xmath573 , i.e. @xmath574 , we have @xmath575 . then for @xmath576 , since @xmath156 , @xmath577 thus ( [ noeq ] ) implies that @xmath578 is not true for @xmath579 , i.e. @xmath580 , where @xmath581 . our conclusion is : for any fixed @xmath582 , according to and , there exists a family of functions @xmath583 such that @xmath584 , @xmath585 and @xmath586 as @xmath569 goes to infinity . therefore @xmath126 is not @xmath1 bounded for @xmath582 and the inclusion @xmath587 does nt hold for @xmath582 . more generally , a slight adaption of theorem [ noequiv ] plus theorem [ equihl ] yields the following result . [ neq - gen ] let @xmath20 be a riemannian manifold satisfying and ( @xmath152 ) as above . let @xmath588 . then for any @xmath589 , @xmath590 if and only if @xmath591 . if @xmath591 , theorem [ equihl ] says that @xmath592 . conversely , by doing a slight adjustment for the above proof , we can show that @xmath587 is false for @xmath582 , where @xmath581 and @xmath593 . this section is devoted to an application of the hardy space theory we introduced above . let @xmath11 be a riemannian manifold satisfying the doubling volume property ( @xmath594 ) and the sub - gaussian estimate @xmath145 . note that we could as well consider a metric measure dirichlet space which admits a `` carr du champ '' ( see , for example , @xcite ) . let @xmath20 be a manifold satisfying the doubling volume property and the heat kernel estimate @xmath145 , @xmath156 . then , the riesz transform is weak @xmath35 bounded and bounded on @xmath1 for @xmath595 . our aim here is to prove theorem [ thm2 ] . more specifically , we will show that the riesz transform is @xmath599 bounded . due to theorem [ h1equiv ] , it is @xmath600 bounded . the method we use is similar as in ( * ? ? ? * theorem 3.2 ) . note that the pointwise assumption simplifies the proof below . to prove the lemma , it is enough to show that the following two estimates : @xmath607 and @xmath608 then follows by adapting the proof of ( * ? ? ? * lemma 2.3 ) . note that the first estimate can be obtained by using stein s approach , similarly as the proof of lemma [ estimatekernelcor ] . the second estimate is a direct consequence of and the analyticity of the heat semigroup ( see @xcite for its discrete analogue ) . we omit the details of the proof here . denote by @xmath609 . it suffices to show that , for any @xmath252molecule @xmath253 associated to a function @xmath270 and a ball @xmath90 with radius @xmath91 , there exists a constant @xmath47 such that @xmath610 . we first estimate @xmath613 . it holds that @xmath614 using the cauchy - schwarz inequality and the @xmath36 boundedness of @xmath249 and @xmath615 , it follows that @xmath616 now we claim : @xmath617 combining and , we obtain that @xmath613 is bounded . we estimate @xmath627 as follows : @xmath628 estimate the first term inside the sum by cauchy - schwarz and the fact that @xmath629 . then @xmath630 for the second term inside the sum , we use lemma [ estimatekernelcor ] again . then @xmath631 it remains to estimate @xmath632 . using the same method as for @xmath627 , then @xmath633 for the first term inside the sum , we estimate by using cauchy - schwartz inequality and the spectral theory . then @xmath634 for the second term inside the sum , we use lemma [ time derivative ] , then @xmath635 this finishes the proof . * acknowledgements : * this work is part of the author s phd thesis in cotutelle between the laboratoire de mathmatiques , universit paris - sud and the mathematical science institute , australian national university . the author would like to thank pascal auscher for suggesting this topic , and to thank thierry coulhon for many discussions and suggestions . she is also grateful to alan mcintosh and dorothee frey for helpful discussions . the author was partially supported by the anr project `` harmonic analysis at its boundaries '' anr-12-bs01 - 0013 and the australian research council ( arc ) grant dp130101302 . alex amenta . tent spaces over metric measure spaces under doubling and related assumptions . in _ operator theory in harmonic and non - commutative analysis _ , volume 240 of _ oper . theory adv . _ , pages 129 . birkhuser / springer , cham , 2014 . e. russ . the atomic decomposition for tent spaces on spaces of homogeneous type . cma / amsi research symposium `` asymptotic geometric analysis , harmonic analysis , and related topics '' _ , volume 42 of _ proc . centre math . _ , pages 125135 . , canberra , 2007 .
hardy space theory has been studied on manifolds or metric measure spaces equipped with either gaussian or sub - gaussian heat kernel behaviour . however , there are natural examples where one finds a mix of both behaviour ( locally gaussian and at infinity sub - gaussian ) in which case the previous theory does nt apply . still we define molecular and square function hardy spaces using appropriate scaling , and we show that they agree with lebesgue spaces in some range . besides , counterexamples are given in this setting that the @xmath0 space corresponding to gaussian estimates may not coincide with @xmath1 . as a motivation for this theory , we show that the riesz transform maps our hardy space @xmath2 into @xmath3 .
clusters of galaxies are the biggest virialized structures in the universe , that form at a relatively late time ( @xmath11 ) . in the hierarchical scenario , the cosmic structures evolve hierarchically from the primordial density fluctuations , that are amplified and then collapse and merge to form larger systems under the action of gravity . the cosmic baryons fall into the gravitational potential of the cluster dark matter halo formed in this way , while the collapse and subsequent shocks heat up the intra - cluster medium ( icm ) to the virial temperature ( @xmath12 kev ) . in the adiabatic scenario , the gravity is the only responsible of the physical properties of the galaxy clusters , once they are rescaled with respect to their masses and epoch of formation ( @xcite ) . x - ray properties of galaxy clusters show , however , some deviations form this scenario , breaking up the self - similarity predicted by the adiabatic model ( see recent work in @xcite and reference therein ) . in particular , in the last years , the studies about x - ray scaling relations ( @xcite ; @xcite ; @xcite ; @xcite ) and observations of the entropy profiles ( @xcite ; @xcite ) in groups and clusters of galaxies , and the analysis of simulated sources with an extra non - gravitational energy injection ( @xcite ) have suggested that we have to account for further non - gravitational feedback beyond the gravitational energy . the gas entropy records the thermodynamic history of the icm as the product of both gravitational and non - gravitational processes , shaping its observed structure accordingly ( @xcite ) . the measurements of the gas entropy at 0.1 @xmath13 ( hereafter @xmath14 ) showed that the observed value of @xmath15 is higher than the expected one from the adiabatic scenario @xcite , where @xmath15 should scale simply with the mean temperature of the virialized systems . instead , an excess in the entropy , with respect to the prediction of the adiabatic model , is observed in the inner regions of groups and poor clusters at some fraction of @xmath13 . this excess sets a minimum value of the entropy , labeled as entropy `` floor '' or `` ramp '' , associated to the ambient gas . the presence of this minimum level of entropy calls for some energetic mechanism , not referable to the gravity only , that falls into three main classes : preheating , where the gas collapsing into the dark matter potential well is preheated by some sources , before clusters were assembled at an early epoch @xcite ; local heating by , e.g. , agn activity , star formation or supernovae @xcite ; cooling , which seems to be able to remove low - entropy gas in the centre of the clusters , producing a similar effect to non - gravitational heating @xcite . in the present work , we aim to confront the model of preheating , feedback and cooling with the observed properties of the gas and of the dark matter in x - ray luminous galaxy clusters , by putting constraints on the sources of non - gravitational heating . to do that , we have considered the sample of 24 clusters presented in @xcite , hereafter paper i. to quantify the excess of energy stored in the icm with respect to the amount available from the gravitational potential , we compare the gas and dark matter temperature profile and measure the energy feedback as a function of the radial distance . moreover , the clusters in our sample span a wide range of redshift ( @xmath16 ) and have different state of relaxation . we can thus investigate the dependence of the extra - gravitational energy feedback on the cosmic time of differently evolved structures . this paper is constructed in this way : in sect . [ dataan ] , we describe the x - ray data reduction and analysis ; in sect . [ scales2 ] , we present our results about the entropy distribution in our hot ( @xmath0 kev ) clusters , studying its relation with the gas temperature and metallicity , its radial profile and how it relates to excess of energy measured by comparing gas and dark matter temperatures . we discuss our results in sect . [ snnen ] and summarize our findings in sect . [ conclusion ] . hereafter we have assumed a flat @xmath17 cosmology , with matter density parameter @xmath18 , cosmological constant density parameter @xmath19 , and hubble constant @xmath20 . unless otherwise stated , we estimated the errors at the 68.3 per cent confidence level . in paper i , we describe our dataset and the analysis applied to study their x - ray and sunyaev - zeldovich properties . here , we remind the main characteristics of the sample and of the x - ray analysis adopted to recover the radial distribution of the icm entropy investigated in the present work . we consider 24 galaxy clusters in the redshift range 0.140.82 , emission - weighted temperature between 6 and 12 kev and x - ray bolometric luminosity @xmath21 erg / s , with exposures available in chandra archive . assuming a spherically symmetric emission , the electron density and temperature profiles are obtained by deprojecting both the surface brightness profile put in hydrostatic equilibrium with a functional form of the dark matter ( dm ) profile and the best - fit results obtained in the spatially - resolved x - ray spectral analysis by fitting a single thermal component . in particular , from the surface brightness profile resolved in a number of radial bins between 24 and 239 , we obtain directly from the geometrical deprojection the electron density @xmath22 in each @xmath23-th spherical shell . the deprojected gas temperature , @xmath24 , is obtained by integration of the hydrostatic equilibrium equation once a functional form of the dark matter density profile , @xmath25 , is assumed , where @xmath26(scale radius , concentration parameter ) and the gas pressure @xmath27 at the x - ray boundary @xmath28 are free parameters . to parameterize the cluster mass distribution , we have considered two dm models : the universal density profile proposed by @xcite ( hereafter nfw ) and the one suggested by @xcite ( hereafter rtm ) . in this study , we adopt the rtm model . our results are not affected if a nfw functional form is used . to constrain the 3 free parameters @xmath29 , we define a grid of values and proceed with a @xmath30 minimization of the merit function that compares the observed temperature profile with the projection of @xmath24 by applying the spectroscopic - like temperature definition @xcite . the best - fit values of @xmath29 are the ones corresponding to the minimum @xmath30 , @xmath31 . the associated errors are estimated at the 68.3 per cent confidence level and are computed by looking to the regions in the parameter space where @xmath32 is smaller than a given threshold , fixed according to the number of degrees of freedom ( e.g. , @xmath33 for 1 , 2 and 3 d.o.f . , respectively ; see paper i and * ? ? ? the value of @xmath29 and the related errors are quoted in paper i. furthermore , we deproject the best - fit results of the x - ray spectral analysis , spatially resolved in a lower number of bins ( between 4 and 10 ) than the surface brightness profile as requested from the higher counts statistic needed to constrain adequately the measurements of the temperature . however , in each @xmath34-th shell , the electron density @xmath35 and temperature @xmath36 are then recovered without any assumption of the hydrostatic equilibrium and provide a direct verification of the validity of this assumption once they are compared to the measures of @xmath22 and @xmath37 described above . the spectral deprojection of the observed projected temperature @xmath38 has been performed in a set of @xmath39 annuli selected to collect at least 2000 net counts by inverting the following equation : @xmath40 where the operator @xmath41 indicates the matrix product ( rows by columns ) , @xmath42 is the effective volume described in appendix of paper i , and @xmath43 using the spectroscopic - like temperature definition @xcite . in the following analysis , we divide our sample in three categories , depending on the strength of the central cooling - core ( see table [ tabdon ] ) : * * strong cooling core ( scc ) * clusters are the 6 objects in which the central cooling time is significantly less than the age of the universe at the cluster redshift ( @xmath44 ) . they show very low central temperature ( @xmath45 kev ) and strong spike of luminosity in the brightness profile , and a very pronounced drop of the temperature near the boundary of the observation , about a factor two compared with the peak of the temperature . the temperature profile is very regular , suggesting a relaxed dynamical state . * * intermediate cooling core ( icc ) * clusters have a central cooling time with values @xmath46 . the six objects in our sample show a less prominent spike of brightness than scc clusters and a mild drop of the temperature in the cooling region ( @xmath47 ) . * the * non - cooling core ( ncc ) * sources ( 12 objects in our sample ) have central cooling time higher than @xmath48 and do not present any evidence of the central drop in the temperature profile . both the temperature profile and surface brightness map are less regular than the ones observed in cc systems , showing hints of substructures and merging activity . the gas and dm density profiles ( right panel of figure [ entps3 ] ) have similar slopes over the entire radial range in the scc clusters , whereas less self - similarity is present in the icc and especially in the ncc clusters : the gas density profile is here flatter than the @xmath49 one , supporting the scenario in which the icm has been affected by some form of non - gravitational energy . we discuss the physical interpretation of these observational results in section [ snnen0 ] . the high level of relaxation of the scc sources is also confirmed by the study of the polytrophic index @xmath50 , is calculated as @xmath51 by linear fit in the @xmath52 plane by considering the spectral deprojected density @xmath53 and temperature @xmath54 described in sect . [ dataan ] . ] that has values near 1 with a very low scatter for the scc sources , whereas is more scattered in ncc sources at @xmath55 : @xmath56 , @xmath57 , @xmath58 . within @xmath59 , we measure @xmath60 , @xmath61 and @xmath62 , with a clear increase as a function of the morphological type and a very high scatter for the ncc sources . in the present paper , we have extended the above analysis by estimating the entropy profile in each cluster by using ( i ) the gas pressure @xmath63 and density @xmath22 profile in the equation @xmath64 and ( ii ) the deprojected spectral results @xmath65 . the errors on the entropy profiles are obtained by error propagation of the uncertainties on the single quantity and/or best - fit parameters . we note that the dependence of @xmath66 over @xmath27 , the gas pressure value at the x - ray spectral boundary , can be checked by comparing it with the entropy measured once @xmath27 is fixed to the value measured in the spectral analysis : we find a totally negligible variation at @xmath67 and a change @xmath68 per cent at @xmath69 . in figure [ entps23bb ] , we present a comparison of the entropy and temperature profiles recovered with the two methods in the inner regions of three representative cases , zw3146 ( scc ) , a1914 ( ncc ) and a2218 ( ncc ) . we obtain good agreement between the entropy measurements in zw3146 and a1914 , the former being an example of a typical cc source where the profile decreases moving inward , whereas the latter shows the most evident case of flattening , with hints of inversion , in the central entropy values . given the good agreement between @xmath70 and @xmath66 even in the internal regions , we believe that this inversion is not due to our approach , but it is real in a1914 ( similar behaviour is found in a773 ) . we note that , if we use the entropy recovered by using the proper cooling function in eq . [ aa1 ] instead of the functional @xmath71 , this inversion is less pronounced ( see fig . [ entps23bb ] ) for @xmath70 . in a2218 ( ncc ) , we observe a marginal disagreement between @xmath66 and @xmath70 : @xmath66 shows an inversion in the core , whereas @xmath70 appears flatter . we draw similar conclusion for a370 , a520 , a2163 , and rxj2228 + 2037 . nevertheless we observe that for the latter sources the low spatial resolution of @xmath70 in the central regions ( @xmath72 kpc ) does not allow to sample properly @xmath73 on scales of @xmath74 kpc , where the inversion occurs . we note that the larger deviations between @xmath37 and @xmath54 are observed in ncc clusters within 100 kpc , where we expect higher relative contribution from non - thermal effects due to , e.g. , merging activity . therefore , even though the most prominent substructures identified in the cluster images were masked , implying that we have reduced their effects in the temperature reconstruction under the hydrostatic equilibrium equation , the sampled gas might be still subjected to ongoing merging processes . the higher value of @xmath73 compared to @xmath70 in the cluster centre is likely due to a very flat density profile that induces a higher temperature value ( once the hydrostatic equilibrium equation is applied ) than the spectral deprojected temperature . indeed , unresolved mergers could lead to this very flat density profile ( they are clearly visible in a520 and a2163 ) , if the gas at @xmath75 kpc is not wholly relaxed and in hydrostatic equilibrium . for the other ncc sources , that do not show clearly ongoing merging processes , nevertheless we noted a disturbed morphology , as indicated , for example , from the fact that the centroid of symmetry does not coincide with the peak of brightness . on larger scales , that involve larger cluster volumes , local deviations from the hydrostatic equilibrium are washed out even in the most unrelaxed objects , making tenable the hypothesis upon which @xmath66 is obtained . this is also confirmed from ( i ) the agreement between @xmath73 and @xmath76 ( @xmath77 and @xmath54 ) measured in these sources , ( ii ) the results of hydrodynamical numerical simulations @xcite , and ( iii ) the analysis presented in paper i ( sect . 4.1.1 ) , where we show how the relation between @xmath78 and the mass - weighted temperature for our sample is in agreement with the results coming from simulations including feedback and radiative processes , supporting our overall mass and temperature @xmath37 reconstruction . in the following analysis , we evaluate the entropy at 0.1 and @xmath79 , i.e. at radii well beyond the region where the central inversion of @xmath80 is observed in few ncc objects . given that , and the good agreement on larger scales between the reconstructed profiles , we define @xmath81 hereafter to fully exploit the spatial resolution available in estimating the entropy radial profile . we examine the @xmath82 relation at fixed overdensities , comparing our results with the ones available in literature for nearby systems . we investigate , then , the radial entropy profile , studying the behaviour of its gradient and its dependence upon the state of relaxation of the system . finally , we implement an analysis of the temperature of the icm and of the dm to quantify the excess of energy associated to the gas and its radial distribution . we have determined the entropy - temperature relation at different fraction @xmath83 of the virial radius @xmath13 ( @xmath84 and @xmath85 ) . we have fitted a power - law model of the form : @xmath86 where @xmath87^{1/2}$ ] and @xmath88 is the total cool - core corrected ( by masking the central @xmath89 kpc region ) emission - weighted temperature in units of 7 kev ( see paper i ) . the fit has been performed by adopting the bces ( bivariate correlated errors and intrinsic scatter ) @xmath90 method @xcite ( see paper i for further details on this approach ) . we quote our best - fit results in table [ tabscal ] and show the distribution of the entropy values at different fractions of @xmath13 in figure [ entps1a ] . we note that scc clusters show higher normalization ( @xmath91 and @xmath92 kev @xmath93 at @xmath84 and @xmath94 , respectively ) than icc and ncc objects , with a larger deviation in the inner region ( @xmath84 ) which can be explained invoking different relaxation states of the clusters as discussed in sect . [ metallicity ] . the best - fit slopes , within the error - bar at @xmath95 , are in agreement with the self - similar prediction ( @xmath96 ) and steeper than the slopes of @xmath97 observed in local samples of galaxy groups and clusters @xcite . for comparison , we present in table [ tabscal ] also the normalizations measured by fixing @xmath96 ( self - similar expectation ) and @xmath98 @xcite and plot in figure [ entps1a ] the best - fit results obtained by @xcite and @xcite from their analyses of relaxed groups and clusters at low redshift . @xcite measure @xmath99 and @xmath100 for @xmath84 and @xmath101 , respectively , with a clear departure from the self - similar expectation ( @xmath96 ) . @xcite at @xmath84 measure @xmath102 by fixing @xmath103 . once these results are compared with what we measure in our sample of very massive systems , we observe that our normalizations are on average higher by 20 - 60 per cent , with slopes that are steeper and closer to the self - similar prediction than the values measured locally ( see also @xcite ) . this result is in agreement with the fact that we are measuring the entropy distribution in massive clusters with cool - core corrected temperatures in the range @xmath104 kev . these systems are definitely less affected from extra - gravitational , feedback processes that , on the contrary , are so relevant in groups and low - mass clusters representing the bulk in the sample of objects studied in , e.g. , @xcite and @xcite . moreover , by parameterizing the evolution in redshift using a @xmath105 dependence ( see paper i for further details on this approach ) , we did not observe any hints of evolution of the entropy - temperature relation within our sample . a very significant positive evolution ( @xmath106 but with a reduced @xmath30 of 5 ; see fig . [ entps1a ] ) is instead measured in the relation between the entropy estimated at @xmath107 and @xmath108 when our cc ( scc+icc ) objects are compared to the best - fit local results in pratt et al . although the local best - fits refer to objects distributed over a wider range in temperature , the systematic larger values measured at higher redshift is noticeable and definitely more evident at @xmath107 than at @xmath94 where we measure @xmath109 with a reduced @xmath30 of about 1 . [ tabscal ] to characterize the gas entropy profile , we follow @xcite and fit two different models . the first one reproduces the radial entropy profile with a power law plus a constant @xmath110 : @xmath111 in the second functional form , we set @xmath112 , modeling the entropy profile with a pure power law . the best fit parameters on the radial profile are determined by applying the @xmath30 statistic to the eq . [ eq1 ] over the radial entropy profile between @xmath113 and @xmath79 , whereas bces(y@xmath114x ) is used when @xmath110 is fixed to zero and the region within @xmath113 is excluded from the fit because it is strongly affected by the cooling process . the outermost bins of the fit are excluded by the fit , being noisy and likely affected by systematic errors due to subtraction of the noise in the data reduction ( see paper i ) . our best - fit results are quoted in table [ tabdon ] . the entropy profiles show a regular behaviour ( see fig . [ entps1 ] ) , once the quantities are rescaled to the characteristic value @xmath115 at the overdensity of 2500 is @xmath116 , i.e. @xmath117 kpc . ] for adiabatic clusters ( see , e.g. , eq . 2 in @xcite ) . profiles of cc clusters are similar down to the inner resolved regions , whereas ncc systems show large deviations in the central parts . these profiles are well reproduced by the functional form with a power - law plus a constant for which we obtain a @xmath118 always less than or of the order of unity , apart from a2390 . in particular , scc sources show a very tight range of values of the entropy pedestal @xmath119 ( @xmath120 ) in agreement with the value found by donahue et al . , and a power law behaviour which is roughly preserved on the entire range of the radial entropy profile , even in the cooling region ( see right panel of figure [ entps1 ] ) . the average slope determined from the second method ( @xmath121 is very similar to the theoretical value of 1.1 predicted by @xcite by using analytic models of shock dominated spherical collapse . concerning @xmath14 , it shows values in the range @xmath122 : if we adopt the definition of @xmath123 in donahue et al . as the normalization at 100 kpc , we have @xmath124 , mildly lower than than the range found by them ( @xmath125 ) . the icc clusters show higher and wider range of @xmath119 , with a typical value of @xmath126 . the power - law behaviour is preserved just on large scale , i.e. outside the cooling region . the average slope is still in agreement with above theoretical predictions ( @xmath127 ) , but it is a little lower than the value measured in scc clusters . in ncc objects , we observe a more scattered radial profile , which is likely self - similar beyond the central regions ( @xmath128 kpc ) . in the inner regions , we notice a very high dispersion on the entropy pedestal value ( @xmath129 ) , larger than the values found in the cc clusters . the average slope is mildly lower than the one determined in the cc - only subsample ( @xmath130 ) . we point out that @xmath131 rises by considering ncc , icc and scc sources , respectively ( @xmath132 , @xmath133 and @xmath134 for the power - law model ) . as we will see in sect . [ snnen0 ] , this trend is probably due to the effect of non - gravitational sources on large scale , which justify the flatter radial behaviour of the entropy profile in ncc clusters . it is worth noticing the behaviour of the entropy pedestal @xmath14 in the different subsamples : @xmath135 ; @xmath136 ; @xmath137 . the trend of the gas density and temperature profile ( see central and left panel of figure [ entps3 ] ) can justify the progressively greater value of the entropy in the inner regions by considering scc , icc and ncc clusters , respectively . we observe higher normalization of the entropy in scc sources ( see figure [ entps1 ] and the value of the parameter @xmath138 in table [ tabdon ] ) . this behaviour is due to the fact that the scc sources show steeper density profiles , i.e. at the same fraction of @xmath13 , as long as we consider radii greater than @xmath67 , the density of the scc sources is lower . even though the temperature profiles in the scc sources are a bit steeper than in the icc and ncc objects , the overall effect is that the gas entropy tends to be higher in scc clusters . we note that the unrelaxed morphology of the ncc sources can not account for systematic changes in , e.g. , the determination of @xmath139 . on the evolution with redshift of the best - fit parameters of eq . [ eq1 ] , we note that only for @xmath140 we obtain a marginal evidence of negative evolution ( spearman s rank coefficient @xmath141 for 22 d.o.f with probability of null correlation @xmath142 ) . on the contrary , @xmath131 shows a positive evolution for the icc clusters : @xmath143 , with @xmath144 , while for the scc and ncc sources there is not apparent evolution . we have calculated the weighted average value of the slopes of the best - fit parameters of the local sample of clusters determined by @xcite , so as to compare it with our estimate at higher redshift : they measure @xmath145 ( when @xmath146 ) and @xmath147 ( by accounting for @xmath110 ) , while we obtain @xmath148 and @xmath149 , respectively , by applying their procedure . these results , confirmed also including in the sample the icc sources , suggest that entropy profiles in nearby cc systems are slightly flatter than in cc clusters at higher redshift , providing some marginal hints on the evolutionary trends present in the entropy distribution . in figure [ entps23 ] , we plot the break radius @xmath150 present in the entropy profile @xmath80 , i.e. the radius where @xmath151 in eq . [ eq1 ] , as a function of the redshift . we found the following average values for @xmath152 : @xmath153 , @xmath154 and @xmath155 . the ncc sources show value of @xmath152 definitely higher than the cc clusters , defining the scale where the non - gravitational energy breaks the self - similarity ( @xmath156 kpc ) . we do not observe significant evolution for @xmath150 , except for the cc objects ( spearman s rank coefficient of @xmath157 , probability of null correlation @xmath158 ) . in this section , we define a temperature associated to the dark matter component following the method presented in @xcite . we define the temperature of the dark matter halo , @xmath7 , as : @xmath159 where @xmath160 is the mean molecular weight of the icm , @xmath161 is the proton mass , @xmath162 and @xmath163 are the 1-dimensional tangential and radial velocity dispersions of the dark matter . the radial velocity dispersions has been obtained by solving the jeans equation : @xmath164 where a velocity anisotropy parameter is defined , @xmath165 . n - body simulations for a variety of cosmologies shows that @xmath166 has roughly an universal radial profile @xcite , which is given by the following relation : @xmath167 where @xmath168 , and @xmath169 @xcite . the dark matter profile is estimated as @xmath170 , where @xmath171 and @xmath172 has been determined from the analysis in paper i. we will compare the dark matter temperature to the gas temperature , @xmath173 , recovered by applying the hydrostatic equilibrium equation . , @xmath174 and @xmath175 . ] we solve eq . [ eq : jeans1 ] for @xmath176 , corresponding to the case of isotropy of the dm and to the central value of the above - mentioned range , respectively , to recover @xmath177 and therefore @xmath7 . as boundary condition in eq . [ eq : jeans1 ] , we assume @xmath7 equal to @xmath173 at @xmath28 . we have checked that uncertainties on the dm temperature assumption at @xmath28 are almost negligible on the dm temperature profile in the inner and central regions ( @xmath178 ) , being @xmath179 , making our results up to @xmath180 reliable and not affected from the assumed value at the boundary . the errors are estimated by looking to the regions of the parameter space that satisfy the condition @xmath181 after the analysis described in paper i. examples of the gas and dm temperature profiles for scc , icc and ncc objects are shown in fig . [ entps3b ] . because only the baryonic component is expected to be prone to non - gravitational energy effects in galaxy clusters , the difference between @xmath173 and @xmath7 , @xmath182 , is a powerful tool to trace the thermal history of the icm . we show in fig . [ entps3b ] how @xmath182 varies as a function of the radius . the ncc clusters show a clear trend of @xmath182 , with values always greater than zero : @xmath183 outside the central region ( @xmath184 kpc ) and it is a few kev in the inner region . near the cluster observed boundary , dm anisotropies might make @xmath7 roughly in agreement with @xmath173 , even though large statistical errors are present and our boundary condition holds . a similar trend is observed in icc clusters , where a less significant disagreement between @xmath173 and @xmath7 is however observed . in scc clusters , on the contrary , @xmath173 is well in agreement with @xmath7 , especially in the inner and central regions . we notice here that the strong negative evolution measured in the scaling relations between @xmath185 and the x - ray / sz quantities presented in paper i , where @xmath185 is the integrated compton parameter over a fixed angular distance , supports the observed radial behaviour of @xmath182 . indeed , the measured sz effect within a fixed angular size samples larger physical region at higher redshifts . this indicates that the effect of non - gravitational processes is relatively more pronounced if the sz flux is measured within smaller physical radii , indicating the physical scale over which the non - gravitational processes are more relevant . when we perform , instead , the same analysis integrating the compton parameter within a physical radius ( as done with the quantity @xmath186 ) , we observe definitely lower negative evolution . we have also estimated the global excess of energy @xmath187 defined in this way : @xmath188 we find that @xmath189 in ncc sources ( corresponding to about 15 - 20 per cent of the total thermal energy ) , that is a factor between 4 and 10 higher than the measured excess in scc clusters . we refer to sect . [ snnen ] for a discussion of the observational evidence presented in these two last sections . the icm iron mass is a key observable to constrain the cumulative past star formation history in galaxy clusters . its relations with other observables such as the cluster optical light , total cluster mass , stellar mass and gas entropy , together with its redshift evolution , allow to study the enrichment processes . moreover , while the production of metals is linked to processes of star formation , its radial profile is determined by different physical processes , such ram - pressure stripping , galactic winds powered by supernovae and agn activity , merger mechanism @xcite . following the work of @xcite on local clusters , we present measured iron abundances in the icm , their evolution with the redshift @xmath190 and their correlation with the entropy . we adopt the solar abundance ratios from @xcite with @xmath191 by number . we first have determined measures of projected metallicity profiles @xmath192 , ( in units of @xmath193 , that is the solar abundance of iron ) , where @xmath194 and @xmath195 are the iron and hydrogen densities ( by number ) respectively . notice @xmath196 has been integrated up to @xmath28 without masking the cooling region , to compare our results with the literature available . in our sample , we find hints of possible negative evolution with redshift , with spearman s rank coefficient of @xmath197 for 22 d.o.f ( probability of null correlation @xmath198 ) , in rough agreement with @xcite , whose sample covers a wider range of @xmath190 . after the deprojection of the spectral results ( see paper i ) , we have calculated the iron mass enclosed within a sphere of radius @xmath199 by integrating the iron mass density , @xmath200 , over the cluster volume . the total iron mass in solar units can be then written as : @xmath201 where @xmath202 is the atomic weight of iron and @xmath203 is the atomic unit mass . to integrate the observed profiles at any radius , we have linearly interpolated the metallicity mass profiles within overdensities @xmath204 , @xmath205 and @xmath206 , which roughly correspond to @xmath207 , @xmath208 and @xmath209 kpc for our sample , respectively . no significant evolution with @xmath190 of @xmath210 ( probability of null correlation @xmath211 ) and @xmath212 ( @xmath213 ) is measured . instead , we observe a strong segregation between ssc , icc and ncc sources , with scc clusters that tend to have higher metallicity mass by a factor of @xmath214 within @xmath180 and by an order of magnitude within @xmath215 , which roughly corresponds to the cooling region . the iron mass excess associated with cool core regions could be entirely produced by the brightest cluster galaxy ( bcg ) , which is always found at the centre of cool core clusters , via sn- or agn- induced winds @xcite . moreover , we confirm the existence of a correlation between @xmath216 and @xmath217 . in fig . [ entps3bw ] , we present the correlation between the @xmath218 as a function of the entropy pedestal @xmath110 ( see sect . [ snnen00 ] ) : we can see an anti - correlation between the two quantities , as expected in a picture where the cooling is the likely predominant physical process in the cooling region . enrichment from recent supernovae type ia in the cd galaxies can explain the central metal abundance excess observed in cooling core clusters @xcite . on the contrary , outside the cooling region , we observe again nearly self - similar relation between @xmath219 and @xmath14 ( see fig . [ entps3bw ] ) as we have seen in sect . [ scales1 ] for the @xmath220 relation . assuming a synthesized iron mass per snia event @xmath221 of 0.74 @xmath222 @xcite and an energy output of @xmath223 , we estimate that @xmath224 snia events in the region inside @xmath180 are required to produce @xmath225 observed in ncc clusters . this number of sne corresponds to a global energy output of @xmath226 over the entire lifetime of the cluster , that is lower by a factor 2 - 4 than the excess of energy @xmath187 estimated in sect . [ snnen0 ] ( @xmath227 ) , suggesting the action of other sources of non - gravitational energy to fully account for this observed excess . the main results emerging from our study of the entropy profiles in hot ( @xmath0 kev ) galaxy clusters at @xmath228 are that these profiles , although similar in the outskirts where they behave as a power - law with slope @xmath2 , are remarkably discrepant in the central regions , with scc objects that show a power - law behaviour down to the innermost spatially resolved regions and ncc clusters having profiles that flatten to a constant value at @xmath229 ( fig . [ entps1 ] ) . accordingly , the comparison between gas and dark matter temperature profiles ( fig . [ entps3 ] ) reveals that scc clusters do not present any significant energy excess at any radius , whereas icc and , more dramatically , ncc objects show @xmath230 larger than few kev in the cooling region and above . note that the situation near the boundary of the sources is unclear , because the statistical errors are very large , the effect of possible anisotropies in the dm are there more prominent and we are assuming some constraints on the dark matter temperature at @xmath28 . this excess of energy with respect to the gravitational energy floor associated to the dm temperature profile can be interpreted as an indication of the presence of some form of non - gravitational energy that can constrain the mechanisms affecting the icm thermal history and the observed entropy profiles . indeed , in agreement with @xmath231 over the entire radial range , we observe that scc clusters have very low entropy pedestal values @xmath119 of few kev @xmath93 , while the higher and more scattered values of @xmath119 measured in icc and , particularly , in ncc systems can be justified by an injection of energy @xmath9 of 1 - 10 kev , that , distributed over scales @xmath232 kpc , explains also their flatter entropy profiles . the regular behaviour of the entropy profiles outside @xmath113 is also in agreement with the fact that @xmath9 is low at these radii , where we have to consider the limitations of our analysis near @xmath28 as mentioned above . this scenario is also supported from our results on the @xmath82 relation , where we observe an higher normalization , more significant in the inner regions ( @xmath84 ; see table [ entps1a ] ) of the scc subsample with respect to icc and ncc sources . we note hints of larger entropy values at higher redshift when our measurements in cc clusters are compared to the best - fit results obtained in nearby samples , with a more significant deviations observed at @xmath113 than at @xmath79 , suggesting that cores in our cc objects are not yet well defined from the radiative processes . moreover , the observed mild differences in the slopes of the entropy profile , with @xmath131 that becomes slightly higher by considering ncc , icc , and scc sources , respectively ( @xmath233 , @xmath133 and @xmath134 for the powerlaw model ) can be explained by looking at the temperature and density profiles ( figure [ entps3 ] ) , which are a bit flatter for non - cooling core systems : this trend can be justified with small energy excess ( @xmath234 1 - 2 kev ) at large scale in the ncc objects compared to the ncc and scc ones . the radial behaviour of @xmath235 is also confirmed by the analysis made in paper i , where we noticed a strong negative evolution in the @xmath236x - ray and @xmath236sz scaling relations ( see sect . [ snnen0 ] ) . all our systems are the products of the hierarchical scenario , how is suggested from the similar behaviour of the gas temperature , density , entropy and dark matter profiles in the regions above the cores . on the contrary , the cooling region characterizes scc , icc , and ncc systems . in particular , continuous interplay between cooling and some form of ( pre-)heating can explain the variety of the properties observed , with scc dominated from the cooling phase and , on the other end , ncc still subjected to some effects of heating . theoretical models must predict the magnitude of the observed @xmath235 , and the impact of the non - gravitational processes associated to this excess in the central regions . these models fall into three main classes : preheating , where the gas collapsing into the dark matter potential well is preheated by some mechanism , before clusters were assembled at an early epoch @xcite ; local heating by agn activity , star formation or supernovae @xcite ; cooling , which seems to be able to remove low - entropy gas in the centre of the clusters , producing a similar effect to non - gravitational heating @xcite . hereafter , we review the main characteristics of these models and discuss how they are consistent with our observational constraints . models of pre - heating , where a constant energy input is injected either prior of the cluster collapse ( @xmath237 kev per particle , e.g. @xcite ) or after the cluster formation ( @xmath238 kev per particle ( e.g. @xcite , could justify , only partially in ncc objects , the observed magnitude of @xmath235 , but not its radial behaviour . nevertheless , as pointed out by @xcite by studying hydrodynamical simulated clusters , there is no possibility to inject a large quantity of energy per particle ( @xmath239 kev ) , unless a large isentropic core is produced in the entropy profile , core that is not observed in our profiles in agreement with other work @xcite . @xcite suggest that any raise of the temperature and/or decrease of the density in the gas inside the primordial structures due to preheating can get largely raised by the accretion shock . following the model of @xcite , @xcite estimated that a mild raise of the entropy of the gas confined to filaments ( @xmath240 , corresponding to a temperature of @xmath241 kev ) can be boosted by the accretion shock to the observed value of @xmath15 ( @xmath242 ) . they point out that an interplay between shock and smoothing of the primordial gas due to a preheating can justify the observed properties of the gas entropy , given the above upper limits on the energy budget of the preheating and being the slope of the entropy profile close to the value predicted from shock heating . preheating prior of the cluster collapse should be a energetically favorable mechanism compared to _ in situ _ heating to cast further energy into the gas before it is concentrated in the gravitational potential well of the dm halo , since less energy is required to increase the entropy of the gas by a given amount when its density is lower as in the filaments . in fact we observe that in the shock dominated collapse scenario , a mild injection of energy through preheating can greatly amplify the final energy @xmath243 of the post - shocked particles , being @xmath244 , with @xmath245 the initial energy . @xcite show that smoothing the accretion pattern by preheating in the case of simulations without radiative physics amplifies the entropy generation out to the radius where the accretion shock acts . nevertheless , the effect seems to be substantially reduced when cooling is also taken into account . however , our estimates of @xmath235 show not a constant profile but instead a declining one outwards . this behaviour can not be explained by any preheating mechanism , either prior or after cluster collapse , even though entropy is amplified through subsequent shock heating . results for x - ray bright nearby objects by @xcite support this conclusions , because their scaled entropy profiles show increasing scatter in the inner regions , with a dispersion ( @xmath246 per cent ) definitely higher than the value found in simulations including filamentary preheating ( @xmath5 30 per cent , @xcite ) . in principle , heating can amplify the boosting of the entropy out to the radius where accretion shocks are taking place , especially in low mass system , since they are accreted by smaller subhalos where the gas is more smoothed by the extra heating . as pointed out by @xcite , local heating due to star formation activity is not able alone to prevent overcooling and to reproduce the predicted star formation as low as measured @xcite and the observed entropy profile : maybe we have to appeal to further sources of non - gravitational energy , like agn , not taken into account in such hydrodynamical simulations , or different physical mechanisms to distribute the energy inside the icm . the need of this further source of non - gravitational energy is also confirmed by the analysis made in sect . [ metallicity ] , where we observe that the number of supernovae we require to reproduce the observed metallicity is not able to account for all the excess of energy @xmath187 . a gentle , transonic heating process , such as the weak shocks detected in the perseus cluster @xcite , can provide a framework by which one can explain all the observed properties , like the flattening of the entropy profile in the innermost regions ( @xmath247 a few tens of kpc ) even of scc clusters , and the spikes of metallicity measured in the centre of scc sources ( see fig . [ entps3bw ] ) . weak shocks are indeed likely not able to prevent metals accumulation in the innermost regions . @xcite pointed out that the central cooling time of the scc galaxies ( @xmath248 yr ) is consistent with the time scale of the activity of radio sources ( @xmath249 yr ) at the centre of clusters . energy casted by the radio jet ( @xmath250 erg / s ) can then produce the observed flattening of the entropy profiles on scale of a few tens of kpc . gasdynamical models of jets flows proceeding from a central supermassive black hole and entering surrounding gas may heat the icm by casting mass and energy outwards till scale @xmath5 hundreds of kpc , possibly lowering the cooling rate @xcite . nevertheless the above picture does not explain the excess of energy @xmath9 in the regions outside the core , where we have probably to require some other form of heating , like e.g. shocks induced from merging activity . whatever sources of non - gravitational energy we have to appeal , they must fuel energetically the icm in such a way to reproduce the magnitude and the radial behaviour of @xmath235 , casting energy ( in the order of few kev ) and metals preferentially in the innermost regions . cooling plays a key role to explain the observed excess of energy @xmath235 . in fact , @xmath235 is higher in the central regions moving from scc to icc and to ncc systems . cooling can easily account for this trend through radiative losses of the accumulated thermal energy . in the scc clusters , @xmath173 is roughly equal to @xmath7 suggesting either that a perfect balance between cooling and heating is established , permitting the radiative losses of the only amount of energy in excess with respect to the one associated to the dm , or that heating is episodic and we are observing structures in their undisturbed phase . in a similar manner , @xcite have argued that the entropy responsible for the break of the self similarity is not a global property of the icm , but rather a property set by radiative cooling : they point out that the observed entropy value at the core radii of groups and clusters is near to the entropy at which @xmath251 . a model to explain the observed features in the entropy profile is that proposed by @xcite . they show how cooling and supernovae heating act to eliminate high compressible gas with @xmath252 , being @xmath253 the cooling threshold , from the x - ray phase . much of the condensation and the feedback is prior of the epoch of clusters formation , balancing these processes reciprocally : more the cooling is effective , more the star formation is active with release of energy to the icm and consequent reduction of the cooling itself . this picture is likely not wholly adequate , leading to a very large isentropic core in the entropy profiles , which are not observed . we have presented _ chandra _ observations of the entropy profiles and scaling properties of a sample of 24 galaxy clusters spanning the redshift range 0.140.82 and classified accordingly to their central cooling time in strong ( scc ) , intermediate ( icc ) and non - cooling core ( ncc ) systems . we have performed a spatially resolved spectral analysis and recovered the gas density , temperature @xmath254 and entropy @xmath15 profiles at high spatial resolution and in a non - parametric way . we have shown that those entropy profiles are remarkably similar outside the central regions with a typical entropy level at @xmath3 of @xmath4 , and have a central entropy plateau covering a wide range of values ( @xmath5 a few@xmath6 ) , with the highest values associated to ncc objects . the cc clusters show larger values of the entropy than the one measured in nearby luminous systems , with a more significant deviations observed at @xmath113 than at @xmath79 with respect to the best - fit results in pratt et al . ( 2006 ) , suggesting that the core in our cc objects are not yet well defined from the cooling processes . we have studied the radial behaviour of the temperature of the gas ( @xmath173 ) and of the dark matter ( @xmath7 ) . we have found that @xmath173 is always higher than @xmath7 : for the scc clusters , the difference of temperature @xmath255 is negligible , while it is large for the non - cooling core clusters ( up to @xmath510 kev per particle ) , with @xmath256 that declines outwards . we conclude that none of the models of ( pre-)heating and cooling discussed in literature is able to explain alone the observed trends of the entropy profiles and of @xmath235 . likely , we have to require an interplay of these processes . a scenario with shock dominated collapse and preheating in the primordial filaments might account for most of the extra - gravitational energy , as confirmed by the slopes of the entropy profiles near the theoretical value of 1.1 expected in the accretion shock picture @xcite . as described in sect . [ preheat ] , this should be an energetically favorable mechanism compared to the _ in situ _ heating , amplifying significantly the final energy @xmath243 of the post - shocked particles . on the other side , gentle , sub - sonic heating processes , e.g. supported from agn s jets , can account for many of the observed properties , but not for excess of energy still measured outside the core ( see sect . [ snnen0 ] ) . the relaxed environment of scc clusters is required to enhance the central metal abundance and total iron mass in correspondence of low entropy regions ( see sect . [ metallicity ] ) . there , the cooling is so efficient to remove on short time scale the excess of energy per particle of the icm , permitting ( i ) @xmath173 to approach the dark matter value @xmath7 and ( ii ) the other physical parameters used in describing the entropy profile , like , e.g. , @xmath110 , @xmath150 , to vary . overall , the different observed behaviour of the entropy profiles of scc , icc and ncc massive clusters suggest that we are observing the end products of the hierarchical model for structure formation . they represent different stages of the relative relevance of heating and cooling in regulating the feedback that shapes the icm distribution : galaxy clusters are identified either as ncc objects when heating , probably due to a residual merging activity and feedback from agns triggered from the merger itself , is predominant , or as scc systems when the radiative losses are energetically prominent , being icc objects an intermediate case between the two . we thank the anonymous referee for a careful reading of the manuscripts and suggestions that have improved the presentation of our work . we thank g. w. pratt , a. finoguenov , r. piffaretti and f. brighenti for useful discussion . we acknowledge the financial support from contract asi - inaf i/023/05/0 and from the infn pd51 grant .
the entropy distribution of the intracluster gas reflects both accretion history of the gas and processes of feedback which provide a further non - gravitational energy besides the potential one . in this work , we study the profiles and the scaling properties of the gas entropy in 24 hot ( @xmath0 kev ) galaxy clusters observed with _ chandra _ in the redshift range 0.140.82 and showing different states of relaxation . we recover the gas density , temperature and entropy profiles in a non - parametric way , just relying on the assumption of a spherically symmetric emission in the deprojection of the best - fit results of the spatially resolved x - ray spectral analysis . adding the hydrostatic equilibrium hypothesis , radial profiles are also obtained from the deprojection of the surface brightness , allowing to verify whether the hydrostatic equilibrium is a tenable hypothesis by comparison with the spectral measurements . we confirm that this is the case on scales larger than @xmath1 kpc and discuss the deviations observed in few non - cooling core clusters in the inner regions . we show that the entropy profiles are remarkably similar outside the core and can be described by simple power - laws with slope of @xmath2 . we measure an entropy level at @xmath3 of @xmath4 and a central plateau which spans a wide range of value ( @xmath5 a few@xmath6 ) depending on the state of relaxation of the source . the entropy values resolved at given fraction of the virial radius are proportional to the gas temperature in these hot systems and appear larger at higher redshift once they are compared to the local estimates . to characterize the energetic of the central regions , we compare the radial behaviour of the temperature of the gas with the temperature of the dark matter @xmath7 by estimating the excess of energy @xmath8 . we point out that @xmath9 ranges from @xmath10 in typical cooling - core clusters to few kev within @xmath1 kpc in non - cooling core systems . we also measure a significant correlation between the total iron mass and the entropy outside the cooling region , whereas in the inner regions they anti - correlate strongly . we find that none of the current models in literature on the extra - gravitational energy is able to justify alone the evidences we obtained on the entropy , metallicity and gas+dark matter temperature profiles . galaxies : cluster : general x - ray : galaxies intergalactic medium cosmology : observations .
the semimetal - semiconductor ( sm - sc ) transition is of particular interest , since the excitonic insulator ( ei ) might be realized in its vicinity at low enough temperatures @xcite . promising materials for the experimental verification are tmse@xmath0te@xmath1 @xcite , 1@xmath2-tise@xmath3 @xcite , ta@xmath3nise@xmath4 @xcite , or a double bilayer graphene system @xcite . the ei constitutes an exciton condensate as an equilibrium phenomenon in contrast to optically created exciton condensates @xcite . approaching , respectively , the ei from the sm or the sc side , a bcs - type transition of electron - hole pairs or a bose - einstein condensation ( bec ) of preformed excitons occurs . hence , the ei is discussed in view of a bcs - bec crossover @xcite . a different nature of the electron - hole pairs below the critical temperature @xmath5 might also show up in terms of precursor effects above @xmath5 @xcite . close to @xmath5 free excitons appear possibly in a large number , and the region surrounding the ei , in ref . @xcite termed halo " , constitutes a three - component plasma consisting of electrons , holes , and excitons . whereas the ei transition scenario itself and the presence of the halo phase above @xmath5 was analyzed in previous works @xcite , the correlation effects caused by the excitons have not been studied so far . we address this issue in this work . the falicov - kimball model extended by a finite @xmath6-bandwidth , in the following shortly denoted as extended falicov - kimball model ( efkm ) , is the minimal model describing the sm - sc transition and the ei formation @xcite . expressing the orbital degree of freedom by a pseudospin variable @xmath7 , the efkm can be written as @xmath8 where @xmath9 annihilates ( creates ) an electron with momentum @xmath10 in the band labeled by @xmath11 , and @xmath12 characterizes the @xmath7-electron occupation of the wannier site @xmath13 . hereafter , @xmath14 denotes the valence band and @xmath15 denotes the conduction band . the bare band dispersions are given by @xmath16 , the coulomb interaction strength is denoted by @xmath17 , and @xmath18 is the total number of lattice sites . in order to investigate the sm - sc transition we consider the efkm at half filling , @xmath19 where @xmath20 . the scenario obtained within mean - field approximation agrees qualitatively with the result from more accurate approaches @xcite . the mean - field ground - state phase diagram is even in quantitative accordance with the constraint path monte carlo data @xcite . however , correlation effects are widely ignored by mean - field approaches , and for the detailed analysis of the bound states in the normal phase a more sophisticated approach is necessary . in this work we will apply the projection technique for green functions developed by plakida @xcite to describe the correlations in the electron - hole - exciton system " in terms of a self - energy . the paper is organized as follows . in sec . [ sec : projecttech ] we outline the method . in sec . [ sec : ehcorrs ] we analyze the electron - hole correlations and determine the excitonic susceptibility , the density of bound electrons , and the density of electrons that are correlated but not bound . our numerical results are presented in sec . [ sec : numres ] . section [ sec : summary ] summarizes this work . the projection approach starts with the decomposition of the time derivatives of the electron operators into a part proportional to themselves and an irreducible part ( throughout the paper we set @xmath21 ) , @xmath22_+\rangle = 0 . \label{time_derive}\ ] ] introducing the spectral moments @xmath23_+\rangle$ ] and @xmath24_+\rangle$ ] , we have @xmath25 and @xmath26 . to derive an exact expression for the self - energy , we consider the equation of motion for the green function @xmath27 . differentiating @xmath28 successively with respect to time @xmath29 and @xmath30 , we obtain a system of equations which in the fourier representation reads @xmath31 with the scattering matrix @xmath32 where @xmath33 and @xmath34 . introducing the zeroth - order green function @xmath35 , we can rewrite eqs . and as @xmath36 the self - energy is defined by the dyson equation @xmath37 from eqs . and we get the relation between the self - energy and the scattering matrix , @xmath38 this equation shows that the self - energy is the proper part " of the scattering matrix , which has no parts connected by a single zeroth - order green function , i.e. , @xmath39 the spectral representation of the self - energy is given by @xmath40 with @xmath41 then the spectral function of the electrons reads @xmath42 ^ 2 + \big[\sigma_{{{\bf k}}\sigma}''(\omega)\big]^2 } \ ; , \label{spectral_func}\ ] ] where we have introduced the shorthand notation @xmath43 and @xmath44 . the renormalized dispersion is given by @xmath45 let us emphasize that the equations up to eq . are exact . however , approximations become necessary if @xmath46 is calculated . in eq . , a three - particle correlation function enters the self - energy . to proceed , we perform a two - time decoupling , because the proper part " of the correlation function must be considered . following the martin - schwinger decoupling scheme @xcite , we identify four contributions , @xmath47 with @xmath48 since we are concentrating on excitonic electron - hole fluctuations , described by the term @xmath49 , we further decouple @xmath50 and @xmath51 into products of single - particle correlation functions . this leads to @xmath52 . then , @xmath53 , and we end up with @xmath54 { { \rm im}}\chi_{{{\bf k}}-{{\bf k}}'}^{-\sigma,\sigma}(\bar\omega-\omega ' ) { { \rm im}}g_{{{\bf k}}'-\sigma}(\omega ' ) , \label{sf_se}\ ] ] where @xmath55 is the fermi function , @xmath56 is the bose function , and @xmath57 is the excitonic susceptibility . we will concentrate on the presence of excitons and their effect on the unbound electrons by including only the according contribution in eq . . the ladder approximation @xcite of the excitonic susceptibility , illustrated in fig . [ fig : ladder ] , is suitable to describe electron - hole bound states . taking into account only the nearly - free part of the electron green function , @xmath58 , we find @xmath59 where @xmath60/(\omega+{\tilde\varepsilon}_{{{\bf k}}-\sigma}-{\tilde\varepsilon}_{{{\bf k}}+{{\bf q}}\sigma})$ ] . . the solid lines represent the electron green functions , and the dashed line represents the coulomb attraction . ] the pole of the excitonic susceptibility located outside the electron - hole continuum , @xmath61 with @xmath62 , describes the bound state of an electron from the @xmath7-band with a hole from the band with index @xmath63 . note that @xmath61 increases with increasing temperature . however , for very high temperatures @xmath64 does not exhibit a pole outside the continuum : consequently excitons do not exist . since an exciton contains one electron , the number of bound electrons equals the number of excitons . chosen @xmath15 to be the index for the conduction band , the exciton dispersion is @xmath65 the binding energy of an exciton can be obtained from @xmath66 , where @xmath67 . provided that @xmath68 exists , we follow ref . @xcite and obtain the momentum - resolved exciton density as @xmath69 where the spectral weight of the excitons becomes @xmath70^{-1 } = z^{\uparrow,\downarrow}(\omega_x^{\uparrow,\downarrow},{{\bf q } } ) = z^{\downarrow,\uparrow}(\omega_x^{\downarrow,\uparrow},{{\bf q } } ) , \label{sp_weight}\ ] ] and the exciton density ( density of bound electrons ) is given by @xmath71 expanding the spectral function , eq . , for small damping @xcite , we can decompose the density into a nearly - free part ( with renormalized band dispersion ) and a part , where the exciton formation enters , @xmath72 where @xmath73 , and the density of the correlated electrons reads @xmath74 straightforward calculation shows that @xmath75 now , it is easy to see that @xmath76 , and the chemical potential is exclusively determined by the nearly - free part of the densities ( see also @xcite ) : @xmath77 note that @xmath78 , i.e. , the considered correlation effects go beyond the simple binding of electrons ( see below ) . moreover , the presence of excitons leads to a decrease of the density of valence - band electrons , because @xmath79 . let us first recapitulate the basic properties of the efkm . thereby we restrict ourselves to a square lattice and a direct band - gap situation , i.e. , valence - band maximum and conduction - band minimum lie at the brillouin - zone center . we choose @xmath80 to be the unit of energy , and the other model parameters are @xmath81 , @xmath82 , and @xmath83 . in a first approximation we neglect the exciton - induced band renormalization in the numerics , i.e. , @xmath84 , where the mean - field hartree - shift is contained in @xmath85 . we analyze the density of bound electrons and the density of correlated electrons along the red and blue lines ( a ) , ( b ) , and ( c ) . the points 1 , 2 , 3 , and 4 mark the points where we analyze the momentum - resolved exciton density and the momentum - resolved density of correlated but unbound electron - hole pairs in fig . [ x_corr ] . ( left panel ) band structure and exciton dispersion for a semimetal . ( right panel ) band structure and exciton dispersion for a semiconductor . in the outer panels the renormalized band dispersions @xmath85 , the chemical potential @xmath86 , the exciton energies @xmath87 , and the boundary of the electron - hole continuum @xmath88 are shown . the blue dotted line shows @xmath89 . the binding energy of an exciton is given by @xmath90 . , title="fig : " ] we analyze the density of bound electrons and the density of correlated electrons along the red and blue lines ( a ) , ( b ) , and ( c ) . the points 1 , 2 , 3 , and 4 mark the points where we analyze the momentum - resolved exciton density and the momentum - resolved density of correlated but unbound electron - hole pairs in fig . [ x_corr ] . ( left panel ) band structure and exciton dispersion for a semimetal . ( right panel ) band structure and exciton dispersion for a semiconductor . in the outer panels the renormalized band dispersions @xmath85 , the chemical potential @xmath86 , the exciton energies @xmath87 , and the boundary of the electron - hole continuum @xmath88 are shown . the blue dotted line shows @xmath89 . the binding energy of an exciton is given by @xmath90 . , title="fig : " ] we analyze the density of bound electrons and the density of correlated electrons along the red and blue lines ( a ) , ( b ) , and ( c ) . the points 1 , 2 , 3 , and 4 mark the points where we analyze the momentum - resolved exciton density and the momentum - resolved density of correlated but unbound electron - hole pairs in fig . [ x_corr ] . ( left panel ) band structure and exciton dispersion for a semimetal . ( right panel ) band structure and exciton dispersion for a semiconductor . in the outer panels the renormalized band dispersions @xmath85 , the chemical potential @xmath86 , the exciton energies @xmath87 , and the boundary of the electron - hole continuum @xmath88 are shown . the blue dotted line shows @xmath89 . the binding energy of an exciton is given by @xmath90 . , title="fig : " ] the middle panel of fig . [ pd_bandstructure ] shows the phase diagram of the efkm in the @xmath17-@xmath2 plane . the ei separates the sm phase and the sc phase at low temperatures . valence band and conduction band overlap in a sm and , as a result , a well - defined fermi surface exists . in this case , only finite momentum excitons can develop , see fig . [ pd_bandstructure ] ( left panel ) . on the other hand , in a sc the valence band and the conduction band are separated by an energy gap , and excitons with an arbitrary momentum can form , in particular zero - momentum excitons , which undergo a bec at the sc - ei transition , see fig . [ pd_bandstructure ] ( right panel ) . electrons and holes may bind to excitons , and subsequently the remaining unbound electrons and holes may scatter as well on these quasiparticles . both effects , the exciton binding and the influence on the unbound electrons , are considered in @xmath91 . how strongly the electrons are affected is mainly determined by the spatial extension of the excitons , and the underlying band structure . let us start by keeping the temperature constant and raising @xmath17 , crossing thereby the sm - sc transition , see fig . [ nx_ncorr](a ) . on the sm side , excitons are loosely bound objects with a relatively large extension . as a result , many electrons are influenced by a single exciton . an increasing coulomb attraction @xmath17 causes a tighter binding of electrons and holes . consequently the exciton radius shrinks , and @xmath92 decreases . at the sm - sc transition both the density of bound electrons and the density of correlated electrons reach their maximum . the kink of @xmath93 and @xmath94 will be less pronounced for a three - dimensional system , c.f . @xcite . in the strong coupling regime , @xmath94 tends to @xmath93 . here , the large coulomb attraction gives rise to a very small distance between the bound electrons and holes : the frenkel - limit of excitons is realized . since there is only a very limited number of unbound electrons and holes , the correlation effects apart from those leading to the formation of excitons by itself are rather weak . let us now analyze temperature effects . on the sm side , excitons are less important for the ei formation @xcite and only a few electrons are bound or correlated , see fig . [ nx_ncorr](b ) . with increasing temperature the broadening of the fermi and bose function will provide more electrons and holes to form excitons , which overcompensates the exciton dissociation . however , for very high temperatures the situation reverses , and the exciton dissociation dominates : @xmath93 and @xmath95 decrease with growing temperature until excitons disappear completely at @xmath96 for @xmath97 [ not shown in fig . [ nx_ncorr](b ) ] . the higher kinetic energy at high temperatures enables the electrons to avoid correlations , and the ratio @xmath98 decreases with increasing temperature . on the sc side , other effects prevail , see fig . [ nx_ncorr](c ) . the sc - ei transition at @xmath5 is driven by the bec of zero - momentum excitons . close to @xmath5 the proliferation of @xmath99 excitons leads to a large density of electrons that are bound or correlated . in contrast to the sm side , the ratio @xmath98 increases with increasing temperature . here , the shrinking of the band gap makes conduction - band electrons available which can participate in the correlations . note that @xmath98 is now significantly smaller than on the sm side , suggesting that the exciton formation resembles more the creation of local electron - hole pairs . for @xmath100 [ coulomb strength in fig . [ nx_ncorr](c ) ] the sc - sm transition is crossed at @xmath101 , and the total density of excitons vanishes at @xmath102 . although the constraint for the chemical potential reduces to the mean - field condition , the electron densities differ significantly from their mean - field values , see eq . . having examined the composition of the halo phase , we turn to the momentum distribution of bound and unbound electron - hole pairs . by the replacement @xmath103 in eq . we can write @xmath104 , with @xmath105 } { ( \omega_x + { \tilde\varepsilon}_{{{\bf k}}\uparrow}-{\tilde\varepsilon}_{{{\bf k}}+{{\bf q}}\downarrow})^2 } + p(\omega_x ) . \label{corrnumbr}\ ] ] we separate @xmath106 into bound pairs ( excitons ) and pairs which contain conduction - band electrons and valence - band holes that are correlated but unbound . we will shortly denote these objects as correlated pairs . the momentum - resolved exciton density @xmath107 is given by eq . , and the momentum - resolved density of correlated pairs is given by @xmath108 and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] + and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] + and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] + and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] + and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] + and ( lower panels ) the corresponding momentum - resolved densities of correlated pairs @xmath109 for @xmath110 ( cf . with marks in fig . [ pd_bandstructure ] ) . the @xmath111-axis is equal within a vertical subfigure , but differs between the subfigures.,title="fig : " ] excitons can produce correlated pairs with the same momentum only , see fig . [ x_corr ] . the ratio @xmath112 modulates with @xmath113 , suggesting that the size of the excitons depends on their momentum . to analyze which electrons become correlated in the course of exciton formation , we resolve @xmath94 according to the momenta of the electrons , @xmath114 , with @xmath115 + f({\tilde\varepsilon}_{{{\bf k}}\downarrow } ) \big[f({\tilde\varepsilon}_{{{\bf k}}-{{\bf q}}\uparrow})- 1 \big ] } { ( \omega_x+{\tilde\varepsilon}_{{{\bf k}}-{{\bf q}}\uparrow}-{\tilde\varepsilon}_{{{\bf k}}\downarrow})^2 } . \label{ncorr_k}\ ] ] the absolute values of @xmath116 can differ substantially . to compare the situations on the sm side with that on the sc side , we scale @xmath116 with its largest contribution , @xmath117 . on the sm side , @xmath118 shows a complicated structure . let us recall that in a sm only finite - momentum excitons and correlated pairs exist , see fig . [ x_corr](a ) . here , the nesting property for finite momenta is crucial . electrons become correlated only if holes with an adequate momentum are available . this is only the case at certain momenta close to the fermi surface , see fig . [ fig : ncorr_k ] ( @xmath119 ) . moreover , @xmath118 is sharply peaked , that is , the mean distance between the correlated electrons is large . on the sc side , the zero - momentum correlations are most relevant ( see @xmath120 in fig . [ x_corr ] ) and electrons close to the conduction - band minimum are predominantly correlated , as can be seen in fig . [ fig : ncorr_k ] ( @xmath120 ) . the structure of @xmath118 is broader than on the sm side , suggesting a more local nature of the correlations . this corresponds to the situation within the ei phase , see the coherence length in ref . in this work we studied the normal phase close to the excitonic insulator ( ei ) phase , the so - called halo " phase . we focused on the correlations that are induced by the presence of excitons and calculated the self - energy by the projection technique for green functions . this enabled us to separate the electron densities into a nearly - free part , which is in principle uncorrelated ( only the band dispersion is renormalized ) , and a correlation part , where the influence of the excitons enters . on the semimetal ( sm ) side excitons are loosely bound objects , whose radius is relatively large . here , bound electron - hole pairs ( excitons ) only exist with finite momentum , and correlated electrons and holes form due to the presence of these excitons . the sm - ei transition is driven by critical electron - hole fluctuations . due to the nesting property of the fermi surface for finite momenta , correlated electrons are concentrated at four distinguished points close to the fermi momenta . with increasing coulomb interaction the transition from the sm to the semiconductor ( sc ) is crossed . thereby , the band structure and consequently the electron - hole pair spectrum changes dramatically . in a sc no fermi surface exists , and the electrons close to the conduction - band minimum are involved in the electron - hole correlations . the excitons on the sc side are tighter bound , which suggests that their radius is smaller than on the sm side . besides those with finite momentum , zero - momentum electron - hole pairs exist , which represent the largest contribution to the electron - hole pair spectrum . the bec of these excitons drive the sc - ei transition . the qualitative difference in the electron - hole correlations on the sm side and on the sc side anticipates the bcs - bec crossover within the ei phase . let us finally stress that the present work exclusively addresses the formation of excitons and the question how they affect unbound charge carriers . the exciton - exciton interaction and particularly the formation of biexcitons is another interesting subject , which deserves further study in the framework of our approach . the formation of excitonic molecules could be relevant near the sm - sc transition , where a collective biexcitonic phase might possibly compete with the ei @xcite . this work was supported by the deutsche forschungsgemeinschaft through sfb 652 , project b5 . 10 url # 1#1urlprefix[2][]#2 mott n f 1961 _ philos . mag . _ * 6 * 287
the region surrounding the excitonic insulator phase is a three - component plasma composed of electrons , holes , and excitons . due to the extended nature of the excitons , their presence influences the surrounding electrons and holes . we analyze this correlation . to this end , we calculate the density of bound electrons , the density of electrons in the correlated state , the momentum - resolved exciton density , and the momentum - resolved density of electron - hole pairs that are correlated but unbound . we find qualitative differences in the electron - hole correlations between the weak - coupling and the strong - coupling regime .
for more than a decade now , much effort has been devoted to understanding the properties of a variety of insulating one - dimensional heisenberg antiferromagnets . by now , the basic facts about these systems are very well established : heisenberg antiferromagnetic ( haf ) chains with integer spins at each site exhibit a gap in their excitation spectrum while those with half- integer spins are gapless @xcite . among the spin-@xmath11 ladder compounds , those with an even number of legs exhibit a gap just like the integer spin chains @xcite while ladders with an odd number of legs are gapless analogous to the half - integer chains @xcite . theoretically , the universal low - energy properties of the gapped systems are well described by the one dimensional quantum @xmath12 non - linear @xmath5-model ( nl@xmath13 m ) without any topological term @xcite . a lot is known exactly about this field theory @xcite and this is directly useful in understanding the gapped systems . the spectrum of the @xmath5-model consists of a triplet of massive spin-@xmath14 particles as the lowest energy excitations followed by multiparticle continua with no bound states . many zero temperature ( @xmath0 ) properties of the gapped systems , including low frequency dynamic correlations , can be explained using the exact information available on the @xmath5-model @xcite . on the other hand , until very recently exact results for @xmath15 were restricted to static , thermodynamic properties @xcite while many experimental observables ( such as the inelastic neutron scattering ( ins ) crosssection and nmr relaxation rates ) directly probe dynamical correlations at non - zero temperature . ( the universal low - energy properties of the gapless systems have been treated via a mapping to a certain critical field theory @xcite . in contrast to the nl@xmath5 m , powerful techniques that exploit the conformal invariance of the theory can be used to determine exactly for @xmath15 some dynamical correlators that are directly probed by nmr experiments @xcite . similar methods have been used to obtain results for @xmath15 on static properties as well @xcite . transport properties have also been studied recently @xcite , with results that are quite different from those we shall obtain here for gapful systems . ) this paper shall deal exclusively with the @xmath15 dynamical properties of gapped heisenberg spin chains . a portion of our results have appeared earlier in a short report @xcite , where we presented them in the context of the continuum nl@xmath5 m , but did not fully discuss their range of applicability . here we shall take a more general point of view of working directly with lattice heisenberg antiferromagnets . the main , and essentially only , requirement on the spin chain being studied is that it have an energy gap and that its low - lying excitations consist of a triplet of spin-1 particles with the dispersion @xmath16 here @xmath17 is being measured from an antiferromagnetic wavevector @xmath18 ( @xmath19 , and @xmath20 is the lattice spacing ) , and we have introduced a velocity @xmath21 to parametrize the mass of the particles as @xmath22 . this is in keeping with the ` relativistic ' spectrum of the @xmath4 @xmath5-model @xmath23 , although most of our results will _ not _ rely on this relativistic form . gapped spin chains with a spontaneously broken translational symmetry ( spin - peierls order ) can have spin-1/2 particle excitations : we shall not deal with this case explicitly , although we believe most of our results can also be extended to these systems . the energy gap @xmath1 is an important energy scale which shall play a central role in our analysis . the majority of our results will be in the regime @xmath2 ( we shall use units with @xmath24 throughout ) which we now discuss . in this regime , there is a dilute gas of excited particles present , and their motion and collisions dominate the dynamical properties we study . in particular their spacing @xmath25 is much larger than their thermal de - broglie wavelength @xmath26 . as argued in ref @xcite , these particles can be treated classically except when two of them collide . such two - particle collisions need to be treated quantum - mechanically and are described by an @xmath27-matrix , which is , in general , a complicated function of the particle momenta and spin orientations . conservation of total momentum and energy implies that momenta before and after collisions have to be the same , and @xmath4 invariance and unitarity impose further constraints , but a fairly complex structure is still permitted we will see some explicit examples in this paper . however , the r.m.s . thermal velocity of a particle @xmath28 as @xmath29 and thus we need the @xmath3-matrix only in the limit of vanishing incoming ( and outgoing ) momenta . one of the central ingredients in our computations will be our claim that in this limit , all of the complexity disappears , and the @xmath3-matrix has a _ super - universal _ form ; for the scattering event shown in fig [ fig1 ] we have ( here @xmath30 are the three possible values of the @xmath4 spin label ) @xmath31 notice especially the opposite pairing of momentum and spin labels : crudely speaking , the momenta go `` through '' the collision , while the spins `` bounce off''this dichotomy will be crucial to our considerations . we dub this limiting value of the @xmath3-matrix ` super - universal ' as it requires only that the lowest lying excitations above the gap satisfy ( [ dispersion ] ) at low momenta . the value , however , does not depend on parameters such as @xmath21 and @xmath1 . moreover , we expect this limiting result to hold even at the lattice level for generic microscopic models of one - dimensional antiferromagnets with massive spin one excitations ; we shall see one explicit example that bears out this expectation later in the paper . with this simple form of the @xmath3-matrix in hand , we will use the semiclassical techniques of ref @xcite to analyze dynamical properties of spin fluctuations near @xmath32 and @xmath33 in terms of the motion of the dilute gas of quasiparticles . we will begin by discussing the properties of the @xmath3-matrix for two - particle scattering in the limit of low momenta in section [ sec : begin ] . in section [ sec : sigma ] we consider the @xmath3 matrix for the @xmath4 non - linear sigma model . this has been computed for all momenta by zamolodchikov and zamolodchikov , and we shall show that the zero momentum limit does indeed satisfy ( [ smatrix ] ) . however , the @xmath5-model is a continuum theory ; it would be much more satisfactory to be able to directly see that ( [ smatrix ] ) holds for some specific microscopic model , and explicitly verify that lattice effects do not affect the simple structure of this limit . one such model is the @xmath6 , two - leg heisenberg antiferromagnetic ladder with inter - chain exchange @xmath34 and intra - chain exchange @xmath35 . the properties of the ladder can be analysed using a ` strong - coupling ' expansion @xcite in powers of @xmath36 for the microscopic lattice hamiltonian of the system . in section [ zerot ] we shall explicitly verify ( [ smatrix ] ) for vanishing velocities in this lattice model within this strong - coupling expansion . parenthetically , we note that our strong - coupling analysis also allows us to make predictions about interesting features in the @xmath8 dynamic structure factor @xmath37 which are specific to the system considered . in particular , we find that , to second order in @xmath36 , a two - particle @xmath7 bound state gives rise to a _ second _ peak ( in addition to the usual peak coming from the stable single particle excitations of the system ) in @xmath38 for a range of values of @xmath39 around @xmath40 . this should be of relevence to inelastic neutron scattering ( ins ) experiments on the ladder compounds and it is hoped that they experimentally verify the existence of this effect . in section [ fuku ] we study the complementary ` weak - coupling ' expansion in powers of @xmath41 for the two - leg ladder . as was shown in ref @xcite , this expansion leads to a description of the low - energy , long - distance properties of the ladder in terms of an effective field theory of a triplet of massive majorana fermions . the hamiltonian for the majorana fermions also has a four - fermion coupling which has generally been ignored in previous treatments . in the absence of this scattering , the majorana fermions are free , and the resulting @xmath3-matrix does _ not _ obey ( [ smatrix ] ) . in this paper , we consider the effect of the four - fermion coupling in perturbation theory . we show that this expansion suffers from severe infra - red problems which have to be resolved by an infinite - order resummation . the structure of the divergences is very similar to those also present in the large-@xmath42 expansion of the @xmath5-model above , and we find that the resulting resummed @xmath3-matrix of the majorana fermions does indeed obey the analogue of ( [ smatrix ] ) . so neglecting the four - fermion coupling @xcite , ( or even treating it in an unresummed manner at finite order in perturbation theory ) is a very bad approximation at low momenta , and we expect that corresponding divergences in the perturbative evaluation of the spin - spin correlation function invalidate the dynamical results of ref @xcite at low @xmath0 . in section [ sec : lowt ] we shall turn to a discussion of the dynamical properties in the regime @xmath43 . our results apply universally to all gapped one - dimensional antiferromagnetic systems with spin one quasiparticles ; indeed they rely _ only _ on the dispersion ( [ dispersion ] ) and the @xmath3-matrix in ( [ smatrix ] ) . all our results will be expressed solely in terms of the parameters @xmath21 and @xmath1 , the temperature , @xmath0 and the external field @xmath44 . in section [ finitet ] we study the dynamics of the staggered component ( with wavevector @xmath39 close to @xmath40 ) of the fluctuations in the spin density . more precisely , we study the dynamical structure factor @xmath37 for @xmath39 close to @xmath40 . apart from some overall factors , this directly gives the ins crosssection at the corresponding values of momentum and energy transfer . at @xmath8 , the dynamical structure factor has a sharply defined @xmath45-function peak at @xmath46 for @xmath39 near @xmath40 . this peak can be thought of as arising from the ballistic propogation of the stable quasiparticle of the system . at non - zero temperatures , the peak broadens as the quasiparticle suffers collisions with other thermally excited particles . the main objective of section [ finitet ] is to describe the precise lineshape of the quasiparticle peak in the dynamic structure factor for @xmath15 . in the @xmath5-model approach , the staggered components of the spin density are represented by the antiferromagnetic order parameter field @xmath47 . we will use the semiclassical method of ref-@xcite to calculate the space and time dependent @xmath48-point correlation function of the @xmath47 field for @xmath15 . this allows us to calculate the thermal broadening of the single particle peak in the dynamical structure factor @xmath38 for wavevectors @xmath39 near @xmath49 . in particular , we find that the dynamic structure factor in the immediate vicinity of the quasiparticle peak at ( @xmath33 , @xmath50 ) may be written in a reduced scaling form as @xmath51 where @xmath52 , @xmath53 is the ( non - universal ) quasiparticle amplitude of the spin one excitations of the system , @xmath54 is the typical time spent by a thermally excited quasiparticle between collisions with other particles , and @xmath55 is a _ completely universal _ function that we determine numerically in this paper . notice that temperature enters this scaling form only through @xmath56 . we claim , though this is not rigorously established , that these results for the broadening are asymptotically exact for @xmath57 : all corrections to the line - width are expected to be suppressed by positive powers of @xmath58 . some evidence for the exactness of our results emerges from consideration of simpler systems where exact results for the line - broadening are available from the quantum inverse scattering method @xcite ; as we shall see in section [ finitet ] , our semiclassical results are in perfect agreement @xcite with these . it is hoped that experimental studies of the temperature dependence of the ins cross - section in this regime will confirm these results , particularly the simple scaling form ( [ scalinspik ] ) . in section [ hdep ] we turn to the correlations of the conserved magnetization density , or dynamic fluctuations near @xmath32 , for @xmath2 . unlike the staggered case , the overall magnitude of the magnetization density fluctuations is universal and given by @xmath59 , where @xmath60 is the uniform susceptibility of the system ( the non - universal overall scale of the staggered component is reflected for instance by the presence of the overall constant @xmath53 in ( [ scalinspik ] ) ) . in this temperature regime , we have the well - known result for @xmath60 @xcite : @xmath61 we shall study the dynamics of the magnetization density in section [ hdep ] @xcite . we shall show that the long - time correlations of an effective semiclassical model are characterized by _ spin diffusion _ , and obtain the following result for its low @xmath0 spin - diffusion constant @xmath62 : @xmath63 using the einstein relation for the spin conductivity @xmath64 , we obtain from ( [ chilowt ] ) and ( [ dslowt ] ) @xmath65 notice that the exponentially large factor @xmath66 has dropped out , and @xmath67 diverges with an inverse square - root power in @xmath0 as @xmath68 . the semiclassical model is possesses an infinite number of local conservation laws : in appendix [ integrable ] , we discuss how the existence of spin diffusion can be compatible with these local conservation laws . however , these results do not rigorously allow us to conclude that the ultimate long - time correlations of the underlying gapped quantum spin chain are diffusive . this has to do with a subtle question of order of limits : we computed the @xmath3 matrix ( [ smatrix ] ) in the limit @xmath29 , and then used it to evaluate the long - time limit of correlations of the magnetization , whereas in reality the limits should be taken in the opposite order . what we can claim is that our results will apply for all times upto a time scale which is larger than the collision time @xmath56 by a factor which diverges with a positive power of @xmath69 as @xmath68 ; there is a substantial time window in this regime where we have established that the spin correlations are diffusive . for the generic gapped quantum spin chain , we can reasonably expect that the ultimate long - time correlations are indeed diffusive , and the only consequences of the omitted terms in the @xmath3 matrix are subdominant corrections to the value of @xmath62 in ( [ dslowt ] ) which are suppressed by powers of @xmath58 . for the continuum nl@xmath5 m with a relativistically invariant regularization ( this is unphysical for any experimental application ) , the issue is a little more subtle : this model does possess additional non - local conserved quantities @xcite , but we consider it unlikely that these will modify the long time limit @xcite . on the experimental side , however , diffusive behaviour of the magnetization density has already been convincingly demonstrated in the @xmath7 one - dimensional antiferromagnet agvp@xmath70s@xmath71 by the nmr experiments of takigawa _ et.al . _ @xcite . as has been argued earlier @xcite , the dynamic fluctuations near @xmath32 provide the dominant contribution to the nmr relaxation rate @xmath9 for @xmath2 . thus , knowing the space and time dependent two - point correlation function of the conserved magnetization density , we are able to compute the field and temperature dependent @xmath9 in this regime . we shall see that the overall scale of @xmath9 is set by the ratio @xmath72 . as was pointed out to us by m. takigawa @xcite , this immediately leads to an activation gap for @xmath9 given as @xmath73 this difference between the activation gaps for @xmath60 and @xmath9 appears to clear - up puzzling discrepencies in the experimental literature @xcite for the value of the energy gap in these systems obtained from knight - shift susceptibilty measurements on the one hand , and @xmath9 nmr relaxation rates on the other ; a systematic tabulation of the activation gaps for a large number of gapped spin chains @xcite does indeed show a trend consistent with ( [ nmrgap ] ) . the crucial factor of @xmath74 clearly arises from the exponential divergence in @xmath62 . this diffusive behaviour we find arises entirely from intrinsic inelastic scattering between the quasi - particles . in real systems there will also be contributions from elastic scattering off inhomogenities which will eventually saturate the divergence of @xmath62 as @xmath68 . however , because of the strong spin scattering implied by ( [ smatrix ] ) the effects of inelastic scattering is particularly strong in @xmath75 , and can easily dominate inhomogenities in clean samples . we will give a detailed account of the calculations leading up to our expression for @xmath9 ( some details on the method used are relegated to appendix [ jep ] ) and then go on to compare the theoretical predictions for the field dependence of @xmath9 with the extensive experimental data of takigawa _ et.al . _ @xcite . we will see that our results ( _ without _ any adjustable parameters , except for a field independent background rate ) agree with the data extremely well for a range of intermediate temperatures . at the lowest temperatures for which data is available , the quality of the fit deteriorates significantly and the @xmath76 divergence predicted at small fields seems to get cutoff , presumably by some spin - dissipation mechanism present in the real system . at the present time , we are unable to incorporate this dissipation in any serious way in our approach . however , following ref @xcite , we can phenomenologically introduce some spin - dissipation in our results for the long - time limit of the autocorrelation function and obtain a corresponding expression for the field dependence of @xmath9 . this allows us to fit the data at the lowest temperatures with a phenomenological form that has one additional adjustible parameter corresponding to the spin - dissipation rate . we also present results for the temperature dependence of this effective rate . finally , in section [ hight ] we will turn to the regime @xmath77 . we will do this in the context of the continuum @xmath4 @xmath5-model only . any continuum theory is applicable to real lattice experiments only below some energy scale , and a natural choice for this energy scale is a typical exchange constant @xmath78 . so more specifically , we shall be studying the regime @xmath79 . for @xmath80 we expect the spins to behave independently , and the system exhibits a curie susceptibility . it is an open question whether the window of temperatures @xmath79 with universal behavior exists at all in any given system , and the answer will surely depend upon details of the microscopics . it is unlikely to be present for @xmath7 spin chains , but appears quite possible for @xmath81 spin chains @xcite . the static properties of this regime were first studied by joliceour and golinelli @xcite using the @xmath82 limit of the @xmath83 @xmath5 model of ref @xcite . we shall present here an exact treatment of static and dynamic properties for the case of general @xmath83 ; the numerical values of the @xmath84 static results are significantly different from the earlier @xmath82 results . we shall show that the antiferromagnetic correlations decay with a correlation length @xmath85 , which to leading logarithms in @xmath69 is given at @xmath84 exactly by @xmath86 where @xmath87 is euler s constant . we also obtain the exact uniform susceptibility @xmath88 ( notice the argument of the logarithm differs slightly from ( [ xihight ] ) ) . it is interesting to compare the two asymptotic results ( [ chilowt ] ) and ( [ chihight ] ) , and we have done that in fig [ fig2 ] . it is reassuring to find that the two results are quite compatible for @xmath89 . this suggests that one of either the @xmath2 or @xmath77 asymptotics are always appropriate . we shall also consider the nature of spin transport in the @xmath79 regime , and show that it is related to transport in a certain classical statistical problem of deterministic non - linear waves . we have not established whether spin diffusion exists or not in this classical problem ; if the correlations were diffusive , however , we are able to precisely predict the @xmath0 dependence of the spin diffusivity : @xmath90^{3/2}}{[3\chi_{u } ( t)/2]^{1/2}}.\ ] ] here @xmath91 is an undetermined universal number , and @xmath92 , @xmath93 are given in ( [ xihight],[chihight ] ) . notice the complementarity in the two @xmath0 regimes discussed above : the description for @xmath2 was in terms of semiclassical particles , while that for @xmath79 is in terms of semiclassical non - linear waves . the primary purpose of this section will be to establish the @xmath3-matrix by a variety of methods . we will begin in section [ sec : sigma ] by using the relativistic @xmath4 @xmath5 model . in section [ zerot ] we will consider the strong - coupling expansion of the two - leg ladder in powers of @xmath36 . this section will also present supplementary results on some interesting features in the @xmath8 ins cross - section of the strongly coupled ladder arising from the presence of a @xmath7 , two - particle bound state in its spectrum . finally , section [ fuku ] will consider the complementary @xmath41 expansion . let us begin with a brief review of the @xmath5-model as an effective field theory for the low - energy properties of the gapped systems ( for a more extensive discussion see @xcite and references therein ) . the imaginary time ( @xmath94 ) action of the @xmath5-model is @xmath95 \label{sigmaaction}\ ] ] where @xmath96 is the spatial co - ordinate , @xmath97 are @xmath4 vector indices over which there is an implied summation , @xmath98 is the totally antisymmetric tensor , @xmath21 is a velocity , @xmath99 is an external magnetic field ( we have absorbed a factor of the electronic magnetic moment , @xmath100 , into the definition of the field @xmath44 . ) , and the partition function is obtained by integrating over the unit vector field @xmath101 , with @xmath102 . the dimensionless coupling constant @xmath103 is determined by the underlying lattice antiferromagnet at the momentum scale @xmath104 to be @xmath105 where @xmath106 is the spin at each site in the original lattice system . the @xmath5-model is used to make statements about physics at length scales @xmath107 and time scales @xmath108 ; this physics is _ universally _ characterized by the dimensionful parameters @xmath21 , @xmath44 , @xmath0 , and @xmath1 , the energy gap at @xmath109 . though the magnitude of @xmath1 is determined by non - universal lattice scale quantities ( @xmath110 for small @xmath103 ) , the long distance physics of the @xmath5-model depends on these lattice scale effects only through the value of @xmath111 , and has no direct dependence on @xmath103 or @xmath112 . also , the energy - momentum dispersion of the stable particle - like excitation of this model is given by @xmath113 , and there is a triplet of them . the conserved density of this model corresponding to its @xmath114 symmetry is the magnetization density @xmath115 . in the hamiltonian formalism , this is represented by the operator @xmath116 . finally , to make contact with the lattice antiferromagnet , we must have a prescription for representing the spin - density operator @xmath117 of the lattice system in terms of the operators of the @xmath5-model . it is most convenient to do this in terms of fourier components . we have : @xmath118 ( recall that @xmath119 ) and @xmath120 for @xmath121 , @xmath122 much smaller than some microscopically determined scale @xmath123 . the missing proportionality constant in the first relation is non - universal and related to the magnitude of the spin at each site in the original lattice system . thus , the @xmath5-model allows us to represent spin fluctuations near @xmath33 ( these being the low - energy degrees of freedom ) _ and _ near @xmath32 . this is of course because the @xmath32 component of the spin density is the conserved charge corresponding to the @xmath114 symmetry of the system , and as such must be included in any description of the slow modes . the exact @xmath3 matrix of the collision of two particles in the @xmath5-model was computed in a seminal paper of zamolodchikov and zamolodchikov . for the scattering event shown in fig [ fig1 ] it is ( recall @xmath124 are the three possible values of the @xmath4 spin label ) : @xmath125 \label{smatrix1}\end{aligned}\ ] ] where @xmath126 is the ` rapidity ' , @xmath127 for @xmath128 , and @xmath4 invariance guarantees a total lack of @xmath44 dependence in the result . the functions @xmath5 in ( [ smatrix1 ] ) are @xmath129 now notice the structure of the limit @xmath130 which is important for our purposes in the region @xmath2 : we find that @xmath131 , while @xmath132 . this establishes the key result ( [ smatrix ] ) for this continuum model . in this section we concentrate on the properties of a particular model system , the spin-@xmath11 , @xmath48-leg heisenberg antiferromagnetic ladder , to which the low - energy phenomenology of the preceeding section is expected to apply . the hamiltonian of the system may be written as @xmath133 here , the @xmath134 and @xmath135 are spin-@xmath11 operators at site i along the two chains that make up the ladder , @xmath103 is a dimensionless coupling constant equal to the ratio of the antiferromagnetic bond along the individual chains @xmath136 to the bond along the rungs of the ladder @xmath137 and we have set the bond strength @xmath138 along the rungs to be unity ; this defines our unit of energy . we will analyze this model in the limit of small @xmath103 ; this ` strong - coupling ' expansion @xcite is expected to be qualitatively correct for all @xmath103 . for @xmath139 , we just have a system of isolated rungs with the two spins on each rung coupled antiferromagnetically . the ground state is a product state with each rung in a singlet state . the lowest lying excited states form a degenerate manifold with precisely one rung promoted to the triplet state . perturbative corrections in @xmath103 would presumably make this triplet ` particle ' hop around producing a single - particle band of triplet excitations as the lowest lying excited states . thus we expect that our perturbative analysis will be most conveniently performed in a representation that directly describes the state of _ individual rungs _ of the ladder . with this in mind , we switch to the ` bond - operator ' formalism introduced in ref @xcite . following ref @xcite , we write the spin operators as : @xmath140 where @xmath141 , @xmath142 , and @xmath87 are vector indices taking the values @xmath96,@xmath143,@xmath144 , repeated indices are summed over , and @xmath145 is the totally antisymmetric tensor . @xmath146 and @xmath147 are respectively creation operators for singlet and triplet bosons at site @xmath148 ( in the previous section we had used @xmath149 to denote the spin density of the lattice system ; here we shall use @xmath150 to denote the same and reserve @xmath151 for the singlet boson operator ) . the restriction that physical states on a rung are either singlets or triplets leads to the following constraint on the boson occupation numbers at each site : @xmath152 the spin density is given by @xmath153 it is also convenient to define @xmath154 the hamiltonian in terms of these operators is given as @xmath155 where @xmath156 and @xmath157 in this representation , the ground state for @xmath139 is just the state with every site occupied by a singlet boson . to zeroth order in @xmath103 , the lowest excited states form a degenerate manifold with a triplet boson ( of arbitrary polarization ) replacing the singlet particle at precisely one site . higher excited states also form degenerate manifolds labelled by the number of singlet particles that are replaced by triplet bosons.in what follows , we will describe states by the number ( which can only be zero or one ) and polarization of the triplet particles at each site , the singlet occupation numbers being determined by the constraint . thus we will loosely refer to the state with no triplet particles as the ` vacuum ' . at this order in @xmath103 , the physical particle - like excitation of the system is created at site @xmath148 by the action of @xmath158 on the vacuum , and thus coincides with the bare triplet particle . in general at higher orders in @xmath103 , we expect that the physical single particle states of the system will contain an admixture of states with more than one bare particle present . similarly , the physical vacuum will also have a component with non - zero bare particle number . in fact , it is quite convenient to make a canonical transformation ( determined order by order in @xmath103 ) to an auxillary problem in which the physical particle states do not contain admixtures of states with different bare particle number . the hamiltonian of the auxillary problem is related to the original one by a similarity transformation . the energy eigenvalues obtained in this manner of course give the energy levels of the original hamiltonian . however , to recover the corresponding wavefunctions , one has to undo the effects of the canonical transformation . we will use this convenient formulation of perturbation theory below as we discuss the strong - coupling expansion . the auxillary hamiltonian in this approach is given by @xmath159 where @xmath160 is the hermitian operator that generates our canonical transformation . we choose @xmath160 to meet the following criteria : * the matrix elements of @xmath161 between states with differing numbers of bare particles should be zero to a given order in @xmath103 . note that this implies that the elementary excitations of the auxillary hamiltonian are just the bare particles . however their dynamics , and their mutual interactions ( in multiparticle sectors ) are determined by the restriction of @xmath161 to the appropriate subspace of definite particle number.this restriction gives the corresponding energy levels of the original hamiltonian @xmath162 correct to that order in @xmath103 . this then serves as our effective hamiltonian for the corresponding sector of the original problem . * this does not completely specify @xmath160 . we therefore also require that @xmath160 have zero matrix elements to a given order in @xmath103 between any two states populated by the same number of bare particles . these criteria fix @xmath160 uniquely order by order in @xmath103 and in general we have an expansion for @xmath160 that reads : @xmath163 . it is quite straightforward to use this procedure to generate an expansion in @xmath103 for the effective hamiltonians in the one and two particle sector of the original problem ( the ` effective hamiltonian ' in the physical vacuum sector is just a constant equal to the ground state energy calculated to the relevant order in @xmath103 ) . solving for the eigenstates and eigenvalues of these effective hamiltonians is just a simple exercise in elementary quantum mechanics . if the eigenstates of the spin - ladder are of interest ( as they will be when we calculate @xmath37 perturbatively ) , we will have to obtain them from the eigenstates @xmath164 of the effective hamiltonian using @xmath165 after this preamble , we turn to the actual calculations . as we have mentioned earlier , the scattering matrix in the low - energy limit is a crucial input to the semiclassical calculations at non - zero temperature , and it is therefore interesting to have results for it in our microscopic model . so , to begin with , let us look at the two - particle sector and work out the scattering properties of the physical particles . first we need to find the effective hamiltonian for the two - particle sector . to first order in @xmath103 , this is just given by the restriction of @xmath166 to the two particle subspace . instead of introducing a lot of cumbersome notation to write this down , we will just list the amplitudes of the various processes that are allowed in this two body problem : * each particle can hop one site to the left or the right with amplitude @xmath167 except when the neighbouring site in question is occupied by the other particle . * when the two particles are at neighbouring sites , there is a non - zero amplitude for spin rotation . consider the state @xmath168 which has one particle at @xmath148 with polarization @xmath169 ( which can be any one of @xmath96,@xmath143,@xmath144 ) and another particle at @xmath170 with polarization @xmath171 . the amplitude to make a transition from this to the state is . to solve for the scattering states of this two - body problem , it is more convenient to work in a basis in which we label the spin part of the two particle states by the total angular momentum @xmath78 and the value of its @xmath144 component @xmath172 . the spin rotation amplitude now becomes just a @xmath78 dependent nearest neighbour potential which takes the values @xmath167 , @xmath173 , and @xmath174 for @xmath175 , @xmath14 , and @xmath176 respectively . note that the potential energy is independent of @xmath172 as one would expect from rotational invariance . it is now quite simple to find the scattering eigenstates in each channel . the spatial wavefunction in channel @xmath78 may be written as @xmath177 where @xmath178 is the symmetrizing operator for @xmath175 , @xmath176 and the antisymmetrizing operator for @xmath179 , and @xmath180 is the reflection coefficient that completely specifies the scattering properties of the particles . for the @xmath180 we have : @xmath181 where @xmath182 , @xmath183 , and @xmath20 is the lattice spacing along the length of either of the two chains that make up the ladder system . note that @xmath184 and @xmath185 both range over the interval @xmath186 . the energy of the scattering state labelled by @xmath187 ( the energy of the ground state being set to zero ) is given by @xmath188 this is consistent with the first order result @xcite for the single particle dispersion relation : @xmath189 . the next step is to use these results for the reflection coefficients to obtain the @xmath3-matrix for this two - body problem in the limit of low velocities . low velocities imply values of @xmath190 and @xmath191 in the vicinity of the band minimum at @xmath192 , i.e @xmath17 close to zero and @xmath193 close to @xmath194 . both @xmath195 and @xmath196 have the limiting value @xmath197 as @xmath198 , @xmath199 . however , @xmath200 is singular in the vicinity of @xmath201 , @xmath202 ; its value depends on the order in which the two limits @xmath198 and @xmath199 are taken . this is somewhat disconcerting as we expect a well - defined low - velocity limit which agrees with the predictions of the @xmath4 nl@xmath5 m field theory . to identify the source of our problem , let us look more closely at the expression for @xmath203 . we notice that @xmath203 , considered a function of the complex variable @xmath17 , has a pole in the upper half - plane for a range of values of @xmath193 . this indicates the presence of a bound state in the @xmath204 channel for the corresponding values of @xmath193 . this bound state hits threshold , i.e its binding energy goes to zero , as @xmath205 . it is the presence of a bound state at threshold that causes the singular behaviour of the reflection coefficient in the limit @xmath198 , @xmath199 . clearly , if there were a range of @xmath193 around @xmath194 for which there was no singlet bound state , then we would not have this difficulty . it turns out ( as we shall briefly outline later ) that extending our calculation to the next order in @xmath103 leads us to precisely this conclusion and gives a well - defined limiting value of @xmath197 for @xmath203 as @xmath198 , @xmath199 . this result can now be used to obtain the @xmath3-matrix of our auxillary two - body problem . we are interested , however , in the @xmath3-matrix that describes the scattering of the physical particle - like excitations of the spin - ladder . thankfully , it is quite easy to see that though the wavefunctions of the two problems are related by a canonical transformation , the purely ` off - diagonal ' form of @xmath160 implies that the two are the same at least to first order in @xmath103 . transforming to the basis used in ( [ smatrix ] ) , we see that the @xmath3-matrix in the low velocity limit is indeed given by ( [ smatrix ] ) thus , this super - universal form of the @xmath3-matrix holds for our lattice model and lends support to the idea that it is a generally valid consequence of just the slow motion of the particles and is in no way dependent on the special properties of the continuum @xmath5-model . to wind up this part of our discussion , let us now summarize the calculation of the reflection coefficients to first order in @xmath103 . we need to find the effective hamiltonian of our auxillary two - body problem to second order in @xmath103 . this involves first working out @xmath206 and then using this to obtain the effective two - body hamiltonian . to @xmath207 , we generate in this manner an additional next - nearest neighbour hopping term and some additional @xmath78 dependent nearest - neighbour interactions . we skip the details as they are somewhat tedious and not particularly illuminating . the @xmath204 reflection coefficient ( correct to @xmath208 ) obtained in this manner is given as : @xmath209 from this , it is easy to see that there is no pole in the upper half @xmath17-plane as long as @xmath210 . this means that there is no singlet bound state possible in this range of @xmath193 . this is consistent with our expectation that at the very lowest energies , the two - particle spectrum should be free of bound states in order to match the predictions of the @xmath5-model . moreover , ( [ r0 g ] ) has a well - defined low - velocity limit of @xmath197 as claimed earlier . the foregoing analysis has shown that the two particle sector has a spin @xmath211 bound state which leads to some interesting threshold singularities for the scattering matrix . examining our expressions for the reflection coefficients , we notice that there is in fact a bound state in the @xmath7 channel as well ( actually , there is also a @xmath81 anti - bound state ; we will not delve further into that aspect of the spectrum here ) . now , a @xmath7 excited state can have observable consequences for the ins cross - section of a system and we might expect to see some interesting features in the same as a result of this . with this motivation , let us turn to the perturbative calculation of the dynamic structure factor at @xmath8 . we pick a coordinate system in which the two chains that make up our ladder are parallel to the @xmath96 axis and have @xmath143 coordinates of @xmath212 and @xmath213 respectively , where @xmath214 is the distance between them ( for simplicity , we are assuming here that the rungs of the ladder are perpendicular to its legs ) . the spins along a chain are located at @xmath96 values equal to integer multiples of @xmath20 . we denote the position of each spin in the @xmath96-@xmath143 plane by @xmath215 . we define @xmath216 . the @xmath8 dynamic structure factor may be written as @xmath217 where @xmath218 is the ground state of the system , @xmath219 is the length of each chain , and @xmath220 denotes the spin operator at @xmath215 in the heisenberg representation . our strategy is to write down the usual spectral representation for ( [ sforladd ] ) and then evaluate it perturbatively . actually , a complete calculation of the second order contribution would involve the eigenstates with more than two - particles present ; below we will ignore this complication and confine ourselves to calculating the contribution of the one and two - particle sectors , correct to the appropriate order in @xmath103 . the spin operator at any site is a sum of two terms : a single particle piece coming from the @xmath221 , and a two particle part coming from the spin density operator @xmath222 . from the structure of the strong - coupling expansion , it is clear that the single particle part does not have matrix elements between the ground state and any state in the two particle sector ; similarly the two - particle piece does not have matrix elements between the ground state and any state in the single - particle sector . thus , keeping only the contributions from the one and two particle sectors , we can write to second order in @xmath103 : @xmath223 where @xmath224 and @xmath225 denote one and two particle states respectively , and @xmath226 and @xmath227 denote the discrete fourier transforms of @xmath228 and @xmath229 . let us digress for a moment and think in terms of the inelastic neutron scattering cross - section for a process with momentum transfer @xmath230 and energy tranfer @xmath231 ; this coincides with the dynamic structure factor apart from some geometrical factors . this scattering can of course produce a single spin - one particle in the spin system . but there is also a non - zero amplitude for producing a pair of these particles close to each other ( as is clear from the actual calculations described later ) . this is the origin of the second term in ( [ spectralform ] ) . now , these two pieces contribute to the structure factor over very distinct intervals along the frequency axis . while it is in principle possible to calculate both terms correct to @xmath207 , we will confine ourselves below to calculating the leading perturbative correction for each value of @xmath231 . thus , we will calculate the single particle piece only to first order in @xmath103 , while doing a full second order calculation for the two - particle piece . below , we give a brief outline of the calculation and then discuss our final results . to calculate the single particle piece , we first need to determine the ground state and the physical one - particle state wavefunctions correct to @xmath208 . this involves using @xmath160 correct to first order to obtain the physical wavefunctions from the wavefunctions of the corresponding auxillary problem ( for the one - particle sector , these are just plane waves to all orders in @xmath103 ; this follows from translational invariance ) . a simple calculation then gives the one - particle piece as @xmath232 where @xmath233 . turning to the two - particle piece , we see that one can actually ignore the distinction between the physical @xmath48-particle wavefunction and the wavefunction of the auxillary two - body problem . moreover , it suffices to consider the auxillary problem to first order in @xmath103 . also , since the ground state has spin zero and we are looking at the matrix elements of a vector operator , we need to consider only the triplet ( @xmath179 ) channel of the auxillary problem . the only subtlety lies in the fact that we need to consider the bound state contribution as well as the usual contribution of the scattering states . from ( [ r1 ] ) , we see that this bound state exists for @xmath234 and for @xmath235 ( remember @xmath193 ranges from @xmath176 to @xmath236 ) . thinking in terms of an inelastic neutron scattering event with momentum transfer @xmath237 in the fundamental domain ( @xmath176 , @xmath194 ) , we see that this bound state can be excited for @xmath238 _ and _ for @xmath239 . in the latter case , momentum is conserved modulo a reciprocal lattice vector of @xmath194 . of course , in addition to the bound state contribution there is a background term coming from the scattering states in this channel . again , the two particles can be created in the scattering state either with total momentum @xmath193 equal to the momentum transfer @xmath237 , or with the two differing by a reciprocal lattice vector of @xmath194 . the actual calculations are quite elementary and we proceed directly to the results for the two - particle contributions . the bound state contribution for @xmath240 may be written as : @xmath241 where @xmath242 . on the other hand , the scattering states give rise to the following background contribution for @xmath243 : @xmath244 note that for @xmath245 or @xmath246 , there is a square - root divergence at the lower threshold to the continuum in @xmath231 ; these are precisely the values of @xmath237 for which the binding energy of the triplet bound - state goes to zero . this enhanced scattering can thus be thought of as arising from the presence of the triplet bound state at threshold . the salient features of these results are summarized in fig [ fig3 ] and fig [ fig4 ] . fig [ fig3 ] is a plot of the positions along the @xmath231 axis of the single particle peak , the bound state peak , and the bottom of the two - particle continuum as a function of @xmath237 . in fig [ fig4 ] , we show the spectral weight in the single particle and bound state peaks as a function of @xmath237 . thus , we see that that the existence of a triplet bound state of two elementary spin - one excitations leads to some interesting features in the dynamic structure factor . actually , qualitatively similar features , again arising from a triplet bound - state , had been predicted earlier @xcite in the alternating one - dimensional heisenberg antiferromangnetic chain . recent ins experiments @xcite on @xmath247vo)@xmath70p@xmath70o@xmath248 do indeed see a second sharply defined peak in the dynamical structure factor for a range of values of @xmath237 . while this compound had been previously thought to be a good example of a spin - ladder @xcite , more recent work @xcite has favoured the alternating chain model @xcite and the ins results have been interpreted @xcite in terms of the additional bound state contribution predicted in ref @xcite . thus , our results may not be of direct relevance to this particular experimental system . however , our work does predict that a second peak in the ins cross - section should be seen in strongly coupled ladder systems and it is quite possible that the feature persists to all orders in the perturbation expansion we have employed . it would be interesting to confirm this effect by looking at other systems that are more convincingly modelled by a simple ladder hamiltonian and it is hoped that future experiments do indeed see the effects coming from the bound state . in this section , we analyze the ladder system ( [ 2legham ] ) in the complementary weak - coupling limit : @xmath249 . an elegant mapping developed by shelton _ _ @xcite allows one to express the low - energy , long - distance properties of the model in terms of a continuum theory of weakly - interacting massive majorana ( real ) fermions . we will analyze the low - energy scattering properties of the spin one excitations of the weakly - coupled ladder by working in this majorana fermion representation . we begin with a brief review of the majorana fermion representation . we will not attempt here to describe in any detail the procedure used @xcite to arrive at this field - theoretic representation . instead , we will be content with a rather telegraphic summary of the principal steps involved . to begin with , one writes down the usual , free , massless bosonic theory @xcite for the low - energy properties of each of the two @xmath6 heisenberg antiferromagnetic chains that make up the ladder . the interchain exchange @xmath35 is then turned on , introducing a local , isotropic ( in spin space ) coupling between the spin - density operators of each chain in the bosonic representation . this has two pieces to it : one coupling the staggered parts of the spin densities with each other and the other doing the same for the uniform component . now , one works with symmetric and antisymmetric combinations of the two boson fields ( one for each chain ) and transcribes everything to a fermionic representation , introducing one dirac fermion for the symmetric combination and another for the antisymmetric combination in the usual manner ( for a readable account of the relevant machinery of abelian bosonization , see for instance the review @xcite by shankar ) . the last step is to write each dirac fermion as two majorana fermions . if one leaves out the uniform part of the coupling to begin with , the theory in terms of the majorana fermions is , remarkably enough , a free - field theory . the staggered part of the coupling just provides a mass @xmath1 to each of the two majorana fermions obtained from the symmetric combination of the bosons , while the two majorana fermions obtained from the antisymmetric combination acquire masses @xmath1 and@xmath250 respectively ( the actual energy gap is given by the absolute value of the mass ) . the three majorana fermions with mass @xmath1 form the spin one triplet we expect on general grounds , and the fourth majorana fermion represents a high - energy singlet mode that will not be very important for our purposes . the mass parameter @xmath1 of the theory is proportional to @xmath35 with the proportionality constant being non - universal . finally , turning on the coupling between the uniform part of the spin densities gives us a four - fermion interaction term between these massive majorana fermions which will play a crucial role in our analysis of the @xmath3-matrix . the procedure outlined above gives us the following hamiltonian for the effective field theory written in terms of majorana fermions : @xmath251 here the @xmath252 and @xmath253 are majorana fermion fields with anticommutation relations given as @xmath254 with all other anticommutators being equal to zero , @xmath255 is defined in general as @xmath256 with @xmath257 and the interaction term @xmath258 may be written as @xmath259 with @xmath260 . note that each majorana fermion is a two component object , the two components being labelled with the subscripts @xmath261 and @xmath219 to denote the ` right - moving ' and ` left - moving ' parts . to make contact with the original spin - ladder , we also need a prescription for expressing the spin - operators of the ladder in terms of the majorana fermions . in sharp contrast to the @xmath5-model , _ only _ the uniform part of the spin - density operator has a local representation in terms of the fermions ; the components of the spin - density near @xmath33 can be expressed only in terms of _ highly non - local _ functions of the fermi - fields @xcite . we have the following expressions @xcite for the uniform parts , @xmath262 and @xmath263 , of the spin density on each chain : @xmath264 where the index @xmath265 takes on values @xmath261 or @xmath219 and repeated indices are summed over . note that the field @xmath253 corresponding to the non - universal high - energy singlet mode drops out of the expression for the uniform part of the total spin - density of the ladder which can then be expressed entirely in terms of the spin one triplet fields . we shall find it convenient , when it comes to actually doing any calculations , to rewrite all of the foregoing in terms of fermionic creation and annihilation operators . these are defined as follows : let @xmath266 and @xmath267 denote the fourier transforms of @xmath268 and @xmath269 respectively . we write @xmath270 where @xmath271 and @xmath272 are the fermionic annihilation operators corresponding to the triplet and singlet modes respectively and @xmath273 and @xmath274 are complex - valued functions of @xmath275 which we specify below . these creation and annihilation operators obey the usual anticommutation relations : @xmath276 with all other anticommutators equal to zero . in terms of these operators , the non - interacting part of the hamiltonian reads @xmath277 where @xmath278 , @xmath279 , and the repeated index @xmath20 is summed over . the functions @xmath280 and @xmath281 are actually chosen to ensure that the non - interacting hamiltonian has this simple diagonal form in terms of the creation and annihilation operators ; this choice guarantees that the operators @xmath282 and @xmath283 , as defined in ( [ defsandt ] ) , create the true quasiparticles of the non - interacting system . the expressions for @xmath280 and @xmath281 are best written as follows : @xmath284 here the functions @xmath285 and @xmath286 are defined in general as @xmath287 with the angle @xmath288 being specified by @xmath289 , @xmath290 . now , we can rewrite the interaction term in normal ordered form with respect to these singlet and triplet creation and annihilation operators . the quadratic terms so generated give the first order correction to the masses of the singlet and triplet modes ( this correction has already been calculated in ref @xcite by other means ) . the quartic term left over , has , in addition to the usual , normal - ordered , particle - number conserving piece , other pieces that involve pair creation and destruction . the full expressions are somewhat messy and we refrain from displaying them here . however , and this is key , we will need only a very simple part ( corresponding to the low momentum limit of the particle - number conserving piece ) of this quartic term for the calculation of the @xmath3-matrix in the low - momentum limit ; our method of writing everything in terms of the creation and annihilation operators has the advantage of identifying and isolating this piece at the very outset . finally , as an aside , we note that the total spin operator of the system may be written in terms of the triplet operators as @xmath291 this confirms that the triplet creation operator @xmath283 does indeed create a single spin one quasiparticle ( with polarization @xmath20 ) of the non - interacting system . with all of this in mind , let us turn to the analysis of the scattering properties of this model . as we are hoping to calculate the @xmath3-matrix perturbatively in the coupling @xmath103 , it is convenient to write @xmath292 . the ` transition - matrix ' @xmath293 can then be calculated perturbatively using the standard field - theoretic prescription that relates it to the corresponding amputated , connected green s functions of the theory . let us make this precise for the case we are interested in : namely , a scattering process in which the initial state consists of two particles , one with momentum @xmath190 and spin polarization @xmath294 , and the other with momentum @xmath191 and spin polarization @xmath295 , and the final state has two particles labelled by @xmath296 and @xmath297 . note that we are now _ not _ talking about the bare particles of the non - interacting theory , but the actual physical quasiparticle states of the system , correct to the relevant order in the perturbative expansion in @xmath103 . the corresponding matrix element , @xmath298 , may then be written as @xmath299 where @xmath300 and @xmath301 are the final and initial energies respectively , @xmath302 and @xmath303 are the total momenta in the final and initial states respectively , and @xmath304 is the ` reduced ' matrix element ( with energy and momentum conserving @xmath45 functions removed ) for the process under consideration . we now specialize to the case @xmath305 , @xmath306 ( @xmath307 ) ; this special case allows us to make our basic point ( regarding the infrared divergences present in a perturbative calculation of the scattering properties ) while keeping the calculations simple . in this case , we may decompose the scattering matrix as follows : @xmath308~. \label{decomps}\end{aligned}\ ] ] now , energy and momentum conservation in one dimension provide enough constraints on the two - body problem to ensure that the allowed final states have the same set of momentum labels as the initial state . this allows us to convert the overall energy and momentum conserving @xmath45 functions in the second term of ( [ defm ] ) to @xmath45 functions that identify @xmath309 with @xmath17 and @xmath310 with @xmath311 . in the process , we of course introduce additional kinematic factors coming from the jacobian ( we are basically using @xmath312 ) . using this , we can write @xmath313 where @xmath314 , @xmath315 , and @xmath316 are defined in terms of the following decomposition for @xmath304 : @xmath317~. \label{decompm}\end{aligned}\ ] ] the relations ( [ relsm ] ) are useful because there is a simple diagrammatic prescription for the perturbative evaluation of @xmath304 . according to this standard field theoretic prescription @xcite , @xmath318 is proportional to the sum of all ` amputated ' ( factors corresponding to external legs omitted ) , fully connected , one particle irreducible diagrams contributing to the time ordered four - point function with two incoming external lines and two outgoing external lines . the incoming lines must carry momenta @xmath190 and @xmath191 , frequencies @xmath319 and @xmath320 set to their respective ` on - shell ' values of @xmath321 and @xmath322 , and spin labels @xmath294 and @xmath295 respectively . the outgoing lines must carry momenta @xmath309 and @xmath310 , frequencies again set to their on - shell values of @xmath323 and @xmath324 , and spin labels @xmath325 and @xmath326 respectively . denoting the sum of all such diagrams schematically by @xmath327 , we can write @xmath328 where the field - strength renormalization factor @xmath329 comes into play because the singlet and triplet creation operators @xmath282 and @xmath330 create the bare particles , while we are asking questions about the scattering properties of the physical quasiparticle excitations . we will not be very careful here about the precise definition of @xmath329 ; it will soon become apparent that this does not play any role in the calculation we do . before we set about calculating @xmath327 , we need to specify our conventions regarding the diagrammatic representation of perturbation theory . as shown in fig [ fig5 ] , we denote the propogator of the triplet particle by a solid line with an arrow carrying momentum @xmath17 , frequency @xmath231 and spin label @xmath331 ; this has a factor of @xmath332 associated with it . it turns out that we do not need to consider any diagrams that have lines corresponding to singlet particles and we will not bother to introduce a diagrammatic representation for their propogator . we also display our diagram convention for the four point vertices of the theory in the same figure ; again , only the particle number conserving vertices in which all four lines involved correspond to triplet particles have been assigned a diagram as the others will not play a role in what follows . one type of vertex , labelled ( a ) in fig [ fig5 ] , depicts a process in which two particles of momentum @xmath333 and @xmath334 , both with spin label @xmath335 scatter into a final state populated by two particles with momenta @xmath336 and @xmath337 , and spin label @xmath338 . the full momentum dependent factor associated with this diagram is also shown below it . we will need only a very simple low - momentum limit of this expression in most of what follows . the other kind of vertex , labelled ( b ) in fig [ fig5 ] , shows incoming particles with labels @xmath339 and @xmath340 scattering into a final state populated by particles with labels @xmath341 and @xmath342 respectively . again , the full momentum dependent factor is displayed alongside for completeness . we will mostly need only the value of this factor when all four momenta equal zero ; this is given simply by @xmath343 . of course , all other vertices of the same type , but having different spin labels that can be obtained from these using the @xmath4 symmetry of the problem , have the same factors associated with them . we are now in a postition to do some calculations . we begin by noting that , apart from the overall factor of @xmath344 which we are ignoring for now , @xmath345 , @xmath346 and @xmath347 are equal to @xmath348 , @xmath349 , and @xmath350 respectively . it is quite simple to calculate these three quantities to leading order in @xmath103 . the diagrams contributing to @xmath351is shown in fig [ fig6 ] , while those contributing to @xmath352 and @xmath353 are shown in fig [ fig7 ] . evaluating these ` tree - level ' amplitudes , we obtain @xmath354 as long as we are interested in only the first order result for @xmath3 , we can set @xmath355 and directly use these expressions to get the following results for the leading low @xmath17 behaviour of @xmath356 , @xmath357 , and @xmath358 correct to first order in @xmath103 : @xmath359 we immediately see that the perturbative expansion can not be trusted in the low - momentum limit because of the infrared divergences present in the expressions for @xmath357 and @xmath358 . the structure of this first order result is seen to be qualitatively similar to the @xmath360 result for the two - particle @xmath3-matrix of the @xmath83 @xmath5-model @xcite . in the latter case , we know that the exact value of the @xmath3-matrix is perfectly well - behaved in the @xmath361 limit and is in fact given by the super - universal expression ( [ smatrix ] ) . to obtain the correct result in this limit for our problem , we need to identify the _ leading _ infrared divergences at _ each order in _ @xmath103 and perform a resummation . now , we do not expect any infrared divergences in the perturbation expansion of @xmath329 and as a result the prefactor of @xmath344 in the expression for @xmath304 does not contribute at all to the terms that need to be resummed ; we will forget about this factor from now on . let us now try and identify the leading infrared divergent diagrams at each order in perturbation theory . first of all , it is clear , purely from frequency and momentum conservation at each vertex , that no diagrams involving pair creation or annihilation can provide the leading divergence at any order . moreover , only internal loops in which both propogators involved point in the same direction give a nonzero result on doing the integral over the frequency running through the loop . a little thought should convince the reader that these two constraints allow us to conclude that the ladder series shown in fig [ fig8 ] give the leading infrared divergent terms in @xmath357 and @xmath358 to all orders in @xmath103 . turning our attention to @xmath356 , we see immediately that fermi statistics guarantees that each vertex in the analogous ladder series for @xmath356 has enough factors of momentum associated with it to rule out any infrared divergence appearing in @xmath356 . our task is thus reduced to evaluating the two series shown in fig [ fig8 ] . to do this , we note that as far as the coefficient of the divergent piece is concerned , we can ignore the momentum dependence of each vertex and simply replace it with a factor of @xmath343 . each crossing of the fermion propogators gives a factor of @xmath197 and each loop integral gives @xmath362 . putting all this together and summing the resultant geometric series , we obtain the following _ non - perturbative _ results for the low momentum behaviour of @xmath357 and @xmath358 : @xmath363 an interesting feature of these results is the pole in the upper - half @xmath17 plane at @xmath364 which seems to suggest the presence of a bound state . however , this region of @xmath17 space is definitely beyond the domain of validity of ( [ resum12 ] ) and it is not clear what significance , if any , to ascribe to this curious fact . turning to firmer ground , we see that the foregoing implies that the low - momentum limit of the two particle @xmath3-matrix is perfectly well - defined and is in fact given by @xmath365 note that apart from an overall factor of minus one , this is exactly the super - universal form ( [ smatrix ] ) . the relative sign is simply a consequence of fermi statistics and our choice of phase for the final state of the scattering process . in any case , we will see that when we use the super - universal form of the @xmath3-matrix for discussing spin transport , the overall phase is immaterial . on the other hand , the overall factor of @xmath197 in the superuniversal form ( [ smatrix ] ) will be crucial when we work out the correlators of the staggered component of the magnetization density . this may seem worrisome at first sight . however , as we do not have any local representation of the staggered component of the spin density in terms of the majorana fermion operators , there is no contradiction at all . in fact , the semiclassical techniques used in section [ finitet ] may also be applied to the problem of calculating the finite temperature green function of the fermions ; this would correspond to calculating the finite temperature correlators of some _ highly non - local _ string operators of the original spin system . however , as it is difficult to see how these may be accessible at all to any experimental probes , we do not pursue this line of thought any further . thus , we see that the low - momentum behaviour of the @xmath3-matrix in this fermionic representation of the weakly - coupled ladder is consistent with the super - universal form ( [ smatrix ] ) , although this behaviour is definitely not accessible to perturbation theory . this leads us to believe that similar infrared divergences would invalidate any perturbative calculation of dynamical properties at finite temperature ( when there will be a dilute gas of thermally excited particles present ) that uses this representation . in particular , this appears to indicate that the results of ref @xcite for the nmr relaxation rate @xmath9 are incorrect at low @xmath15 . the results of this section are expected to apply to all gapped one - dimensional antiferromagnets with massive spin - one quasiparticles . we will develop , what we believe is an exact semiclassical theory of dynamics and transport for @xmath2 . we will consider fluctuations near @xmath33 in section [ finitet ] , and near @xmath32 in section [ hdep ] . in this section , we present calculations leading up to our results for the thermal broadening of the single particle peak in the dynamics structure factor . the inelastic neutron scattering cross - section provides a direct measure @xcite of the dynamical structure factor @xmath38 which is defined as @xmath367 where @xmath368 is the heisenberg representation operator corresponding to the component of the spin density at wavevector @xmath39 , the expectation values are with respect to the usual equilibrium density matrix and summation over the repeated index @xmath141 is implied ( note that we are assuming rotational invariance in spin space and working at @xmath369 ) . we are interested in the structure factor for @xmath39 close to @xmath192 . in this case we have @xmath370 where @xmath371 . to get a feel for what ( [ strucfac1 ] ) looks like at @xmath8 , let us consider a particular lattice regularization of the @xmath5-model , defined by the quantum rotor hamiltonian @xmath372 where @xmath373 is the angular momentum operator of the rotor at site @xmath148 , @xmath374 is the unit vector that denotes the position of the rotor at site @xmath148 and we have temporarily set @xmath375 . it is not hard to analyze the properties of this model in a large @xmath103 , strong coupling expansion ; moreover this is expected to be qualitatively correct for all @xmath103 in one dimension @xcite . to lowest order , we can easily see that the ground state would just be a product state with each site being in an eigenstate of @xmath376 with zero eigenvalue . the lowest excited states would be a degenerate manifold corresponding to promoting any one site to the @xmath377 state and thereby creating a ` particle ' at that site . to first order in @xmath378 , a hopping term would be generated in the effective hamiltonian for the single - particle sector , resulting in a band of one particle excitations . to this order , @xmath379 is just a sum of creation and annihilation operators for the stable particle - like excitation of the system . at higher orders in @xmath378 , @xmath379 acting on the vacuum will also produce multiparticle states , but there will always be some single particle component . reverting back to our continuum theory , we see that ( [ strucfac1 ] ) evaluated at @xmath8 would have a contribution @xmath380 associated with the stable particle . the next contribution is actually a continuum above the @xmath381-particle threshold @xcite . following @xcite , we shall now focus exclusively on how this one - particle peak broadens as @xmath0 becomes non - zero . let us define @xmath382 where the repeated index @xmath141 is summed over . let @xmath383 denote @xmath384 evaluated at @xmath8 keeping only the single particle contributions . we have @xmath385 here @xmath386 is a ` form factor ' . for our lorentz invariant continuum model , @xmath387 where @xmath53 is a non - universal quasiparticle residue . this gives @xmath388 , with @xmath389 the modified bessel function . now let us evaluate @xmath384 for non - zero temperatures using the semiclassical method of @xcite . first , it is convenient to switch to operators @xmath390 , @xmath391 and @xmath392 , defined as @xmath393 and @xmath394 @xmath395 is a sum of a creation operator for particles with @xmath144-component of spin @xmath396 equal to @xmath397 and an annihilation operator for particles with @xmath396 equal to @xmath197 . @xmath398 is a sum of creation and annihilation operators for particles with @xmath396 equal to @xmath176 . in the absence of an external field , we may write @xmath399 we represent ( [ czero ] ) as a ` double time ' path integral , with the @xmath400 factor coming from the heisenberg operator generating paths that move forward in time , and the @xmath401 producing paths that move backward in time . we begin with an initial state which is populated by thermally excited particles , the density of particles being @xmath402 @xmath403 and their mean spacing being much larger than the thermal de - broglie wavelength @xmath402 @xmath404 . as argued in @xcite , this means that the particles can be treated semiclassically . in this semiclassical limit the dominant contribution to the feynman sum comes about when the paths going backward in time are exactly the time - reversed counterparts of those going forward and all particles follow their classical trajectories between collisions @xcite . whenever two particles collide , energy and momentum conservation is sufficient to determine the final momenta . however , one can not entirely ignore quantum effects of the collisions . the spins of the particles after the collision as well as the phase picked up by the wavefunction of the system as a result of the collision is determined by the quantum mechanical scattering matrix ( @xmath3 ) . for @xmath2 , the particles all move very slowly and we need only the super - universal low - momentum limit ( [ smatrix ] ) . all this leads to the following description of @xmath384 in this asymptotic limit @xcite : at time @xmath405 we begin with an initial state populated equally ( for @xmath369 ) with three species ( corresponding to the three values of spin projection @xmath396 ) of particles each uniformly distributed in space with density @xmath406 , where the total density @xmath253 is given as @xmath407 the velocities are distributed according to the classical maxwell - boltzmann distribution function @xmath408 each particle in the initial state is assigned one of the three values of @xmath396 with equal probability ( assuming @xmath369 ) . the operator @xmath409 acting on this initial state creates at time @xmath405 one extra particle at @xmath410 with spin value equal to @xmath176 ( the annihilation part of @xmath411 gives a contribution which is exponentially suppressed and is ignored here ) . these particles follow their classical trajectories forward in time . at every collision , we pick up a factor of @xmath197 from the @xmath3-matrix . at time @xmath412 , a particle with spin projection of zero is annihilated at @xmath96 by @xmath413 . the resulting state is then propogated backward in time to @xmath405 and its overlap with the initial state calculated . @xmath384 is then given by the average of this overlap over the ensemble specified earlier . a typical example of a space - time configuration of trajectories that leads to a non - zero value for this overlap is shown in fig [ fig9 ] . all trajectories in the figure except the dotted line denote space - time paths that are traversed both forward and backward in time . the dotted line is traversed only forward in time as the particle travelling on it is destroyed at time @xmath412 by @xmath413 . a little thought convinces one that this overlap is non - zero only when all particles colliding with a particle travelling on the dotted trajectory have the same spin @xmath396 ( equal to zero ) as it does . moreover , when this condition is satisfied , the value of the overlap is just @xmath414 where @xmath415 is the number of collisions that the dotted trajectory suffers . the factor of @xmath416 comes from the scattering matrix at each collision between a particle on the dotted trajectory and other particles . all other collisions occur in pairs ( the second member of the pair coming from the evolution backward in time ) and thus do not contribute any phase factor . the factor of @xmath383 is just the relativistic amplitude for the propogation of a single particle from @xmath410 at @xmath405 to position @xmath96 at time @xmath412 . all this implies that we can write @xmath417 which defines the ` relaxation function ' @xmath261 . for the case where the particle has only _ one _ allowed value of its spin label , @xmath396 , it is possible to compute @xmath418 analytically @xcite , and the resulting expression ( [ rdefine ] ) then agrees precisely with a computation using very sophisticated quantum inverse scattering method @xcite : this agreement gives us confidence that the physical approach developed here is asymptotically exact at low temperatures . let us now turn to the calculation of @xmath261 for the case of interest here . we begin by writing a formal expression for @xmath261 based on the foregoing semiclassical description . let @xmath419 be the positions of the thermally excited particles at time @xmath405 . let @xmath420 be their initial velocities . here @xmath17 is an index running from @xmath14 to @xmath42 , the total number of particles present in the initial state in a system of size @xmath219 . we label the initial positions with the convention that @xmath421 for @xmath422 . let @xmath423 denote the @xmath424 space - time trajectory ( note that this is quite distinct from the position of the @xmath424 particle at time @xmath412 ) . let @xmath425 denote the spin projection value of the particle travelling along the @xmath424 trajectory at time @xmath412 . the spin projections are randomly assigned to each particle at time @xmath405 as described earlier and @xmath425 at later times depends on which particle is travelling on the @xmath424 trajectory at any given time . we have the following expression for @xmath261 : @xmath426 with @xmath427 where @xmath428 and @xmath429 the angular brackets in ( [ rexpress ] ) denote averaging over the ensemble of initial conditions specified earlier . unfortunately , it does not seem possible to do the ensemble average analytically . using methods of refs @xcite , it is possible to develop a ` cumulant ' expansion for the logarithm of @xmath261 @xcite . this expansion , however , is essentially a short time expansion which is not uniformly convergent , and thus not very useful for our purposes as we eventually need to fourier transform @xmath384 . it is also possible to develop a ` mean - field ' approximation to this classical model that ignores the complicated correlations between the @xmath430 at different times ( see appendix [ mf ] ) . this proves to be reasonably accurate at least for the @xmath431 , though the high - accuracy numerics we describe next show clear deviations from the mean - field results . so , although we have an asymptotically exact formulation for the non - zero temperature @xmath384 at distances much larger than the thermal de - broglie wavelength and times much larger than @xmath432 , we need to numerically determine the relaxation function @xmath261 to actually calculate anything accurately . this is what we turn to next . an important property of @xmath418 , which follows directly from ( [ rexpress ] ) is that it can be written in a scaling form as @xmath433 with @xmath434 and @xmath435 where @xmath436 and @xmath437 thus it is most convenient for the numerics to measure length in units of @xmath438 and time in units of @xmath56 and directly calculate @xmath439 . we start with a system size of @xmath440 ( in units of @xmath438 ) and impose periodic boundary conditions . the density in these units is unity and so the initial state is populated by @xmath441 particles with their initial positions drawn from a uniform ensemble . this system size is large enough that finite - size effects are negligible for our purposes . each particle is assigned a velocity from the classical thermal ensemble . in these new units this implies that we choose velocities from the distribution @xmath442 an important advantage of our method is that we do the average over the spin values analytically . to do this , we note that it is possible to reformulate the calculation of @xmath439 by writing @xmath443 where @xmath444 denotes a given space time configuration of trajectories , the angular brackets denote averages _ only _ over the initial positions and velocities that define this configuration , and @xmath445 is defined as @xmath446 here , @xmath447 is the total number of collisions involving a particle travelling on the dotted trajectory of fig [ fig9 ] and @xmath448 is the number of _ different _ thermally excited particles that have had collisions with a particle travelling on the dotted trajectory . now , @xmath449 for all configurations @xmath444 in which the presence of the extra particle ( that starts out on the dotted trajectory ) affects the evolution of the various spin values @xmath425 . so we might as well forget about the particle travelling on the dotted trajectory and consider an auxillary spacetime diagram that now involves only the thermally excited particles . we now agree to ignore the spin label on the dotted line of fig [ fig9 ] ; the dotted line now does @xmath450 denote the trajectory of any particle . in terms of this picture we can define @xmath447 as the number of times any solid line crosses the dotted line , and @xmath448 as the number of _ different _ thermally excited particles that cross the dotted line . with this new formulation in hand , calculating @xmath445 reduces to some simple book - keeping that keeps track of these two integers for a given configuration @xmath444 . we implement the ensemble average by averaging over @xmath451 configurations drawn from the appropriate distribution . the combined absolute error ( statistical and finite - size ) in @xmath452 for values of @xmath453 , @xmath454 of interest to us is estimated to be less than about @xmath455 . with @xmath439 available , it is a simple matter to numerically fourier transform the resulting @xmath384 and obtain the dynamic structure factor @xmath37 . details of the numerical procedure used are relegated to appendix [ a ] . here we only comment on some conceptual issues involved and discuss our results . there is an important subtlety associated with doing the fourier transform that needs to be first addressed . as discussed in ref @xcite , the semiclassical result for @xmath384 is valid as long as both @xmath96 and @xmath412 are not very small ; the results break down when @xmath456 and @xmath457 ( @xmath458 being the thermal de - broglie wavelength ) . however , the fourier transform of @xmath459 ( at wavevector @xmath371 ) is an asymptotically valid approximation to @xmath37 _ only _ for @xmath231 close to @xmath460 . the reason for this can be understood by noting that the long - time asymptotics of our form for @xmath384 have an oscillatory character with oscillations on the scale of @xmath461 . put another way , it is the spectral weight in the one - particle peak that plays a dominant role in determining the long - time , large - distance asymptotics of @xmath384 and so we can learn only about this feature in the spectral weight by fourier transforming our form for @xmath459 . with this caveat in mind , we have @xmath462 where @xmath371 . we have not attempted to exhaustively map out @xmath37 , although it would be quite straightforward to get more extensive numerical results should they be of interest in some experimental context . below we confine ourselves to discussing our results for @xmath37 for a couple of sample values of @xmath39 . fig [ fig10 ] shows scans in frequency across the quasiparticle peak in @xmath37 for @xmath33 at four different values of temperature . it is interesting to note that when rescaled by @xmath56 and plotted against a rescaled frequency variable @xmath463 , the three curves for @xmath464 , @xmath465 , and @xmath466 seem to collapse on top of one another within our numerical errors ( which are conservatively estimated to be a few percent at the most ) . in fig [ fig11 ] , we show a scan in wavevector across the same peak for @xmath467 , again at the same four values of temperature . the curves at the lower temperatures again show scaling collapse ; when rescaled by @xmath56 and plotted against the rescaled variable @xmath468 , they seem to all fall on top of one another . moreover , the scaling curve in fig [ fig11 ] , when plotted as a function of the independent variable @xmath469 coincides within our numerical error with the scaling function of fig [ fig10 ] for @xmath470 ; this is displayed in fig [ fig12 ] . while we do not have any reason to expect that this scaling is generally true , all three observations may be put together in terms of a scaling form that is valid _ locally _ in the neighbourhood of the quasiparticle peak for @xmath33 ; more formally we write @xmath471 we also investigated @xmath37 in the vicinity of the quasiparticle peak corresponding to @xmath472 ; for this to be meaningful , we of course need @xmath473 to be much less than the microscopic scale @xmath474 beyond which our continuum theory does not work . we again tried to check if the analogous scaling form , @xmath475 is approximately valid . fig [ fig13 ] shows scans in frequency across the peak with @xmath17 held fixed at @xmath473 , for @xmath476 , @xmath381 , @xmath465 , and @xmath466 . we see that the curves do not really appear to scale . in fig [ fig14 ] , we show scans in wavevector , with @xmath231 held fixed equal to @xmath477 for the same values of the ratio @xmath69 . we plot the data ( rescaled by @xmath56 ) against the rescaled variable @xmath478 ( note the difference in the choice of rescaling of the independent variable from the earlier case ) . again , in sharp contrast to the @xmath479 peak , we see that the curves do not show any signs of scaling ; our local scaling form is not a very good way of organizing the data in this case . these scaling properties are best understood as follows : imagine developing @xmath418 in an expansion about @xmath410 for constant @xmath412 and then calculating the fourier integral in ( [ broads ] ) . the zeroth order term clearly gives us a result for @xmath37 which is compatible with the scaling form we have postulated for asymptotically low temperatures . however , before we can trust this result , we need to check that the corrections to the leading behaviour go to zero in the limit @xmath480 . this is where the difference between the two peaks we looked at becomes apparent . it is easy to see that this is true only for values of @xmath39 such that @xmath481 as @xmath68 and this explains why the scaling form ( [ scale1 ] ) does not work . now consider the peak at @xmath33 : the zeroth order scaling result has most of its weight in the region @xmath482 . for @xmath483 , the corrections to this leading result do indeed go to zero and this establishes the scaling form ( [ scale0 ] ) . an interesting feature of this result is that the scaling function @xmath55 is _ completely determined _ by the @xmath410 part @xmath431 of the relaxation function . more precisely , we have @xmath484 a useful check on all of our calculations is thus to compare the scaling function obtained in fig [ fig10 ] and fig [ fig12 ] with ( [ dirscalfn ] ) evaluated numerically ( it is possible to do this to a high accuracy ; details may be found in appendix [ a ] ) . the results of such a comparison are shown in fig [ fig15 ] and the agreement is seen to be quite good . while the numerical results for @xmath431 show a clear deviation from the simple exponential decay predicted by the ` mean - field ' theory referred to earlier , we do find that the corresponding simple lorentzian form for the fourier transform : @xmath485 ( with @xmath486 ) provides an excellent approximation to the line - shape ( the ` mean - field ' theory , however , gives a value of @xmath487 for @xmath141see appendix [ mf ] ) . we thus have results for the thermally broadened quasiparticle peak in @xmath37 ; the accuracy of these in the asymptotic regime ( @xmath2 ) is limited only by the computer time spent in numerically evaluating the relaxation function and doing the fourier transform . these results , especially the scaling properties in the vicinity of the peak corresponding to @xmath33 , should be of relevance to neutron scattering experiments on gapped one - dimensional heisenberg antiferromagnets performed at temperatures @xmath2 and it is hoped that this study provides a useful paradigm for organizing the experimental results . in this section , we shall present a detailed comparison of our results @xcite for the field ( @xmath44 ) and temperature ( @xmath0 ) dependence of the nmr relaxation rate @xmath9 ( in the regime @xmath0 , @xmath488 ) with the experimental data of ref @xcite on the nmr relaxation rate in the compound agvp@xmath70s@xmath71 which is thought to be a @xmath7 one - dimensional antiferromagnet with a large gap @xmath489 k and single - ion anisotropy energy of about @xmath490 k @xcite . we will ignore this anisotropy for the most part in our theoretical analysis ( although we are forced to phenomenologically introduce spin - dissipation into our theorerical results in order to fit the data of ref @xcite at low temperatures , we do not have any theory that gives the detailed temperature dependence of this spin dissipation rate starting from the anisotropic coupling term in the hamiltonian ) . for completeness , let us begin with a detailed review of the calculations leading up to our expression for @xmath9 . the nmr relaxation rate is given in general by an expression of the form @xmath491 where @xmath492 is the fourier transform of the spin - spin correlation function ( the subscripts refer to the _ o_(3 ) indices of the spin operators ) , @xmath493 is the nuclear larmor frequency ( @xmath494 is the nuclear gyromagnetic ratio ) , the field @xmath44 points in the @xmath144 direction and @xmath495 are the hyperfine coupling constants . the @xmath39 integral in ( [ gennmrrate ] ) is dominated by values of @xmath39 near @xmath176 @xcite and we can thus work out the field and temperature dependence of @xmath9 knowing the @xmath15 correlators of the conserved magnetization density of the _ o_(3 ) nl@xmath5 m field theory . this is what we turn to next . we define the correlation functions @xmath496 here the angular brackets denote averaging over the usual equilibrium density matrix , @xmath497 is the heisenberg representation operator corresponding to the @xmath144 component of the magnetization density , and @xmath498 are operators corresponding to the circularly polarized components of the magnetization density defined as @xmath499 . as argued in refs @xcite , these correlation functions in the asymptotic regime may be evaluated by writing down a double - time path integral representation for them and evaluating it semiclassically . this leads to the following prescription @xcite for @xmath500 : at time @xmath405 we begin with an initial state populated with three species ( corresponding to the three values of spin projection @xmath396 ) of particles each uniformly distributed in space with densities given respectively by @xmath501 and with velocities distributed according to the classical maxwell - boltzmann distribution function @xmath408 each particle in the initial state is assigned one of the three values of @xmath396 with probability @xmath502 . the operator @xmath503 merely keeps track of the local value of the @xmath144 component of the spin . acting on the initial state , it measures the @xmath144 component of the magnetization density in the initial state at position @xmath410 . these particles then follow their classical trajectories forward in time . at every collision , the particles retain their spin labels . in addition , the state picks up a factor of @xmath197 from the @xmath3-matrix at each collision . at time @xmath412 , the operator @xmath504 measures the value of the @xmath144 component of magnetization density at position @xmath96 . the state is then propogated backward in time to @xmath405 and its overlap with the initial state calculated . @xmath500 is then given by the average of this overlap over the ensemble specified earlier . as all collisions have a time - reversed counterpart , the phase of the scattering matrix does not matter here and the overlap we are interested in equals the two - point correlation function of the classical observable @xmath505 where we are labelling particles consecutively from left to right with an index @xmath17 , @xmath506 denotes the position of the @xmath424 particle at time @xmath412 , and @xmath507 is the @xmath144 component of the spin of the @xmath424 particle . this correlation function is calculated using the ensemble of initial conditions outlined above . the dynamics governing the time evolution of the @xmath508 is just that of particles moving ballistically except for elastic collisions in which they retain their spin values . thus we can write @xmath509 here the angular brackets refer to averaging over the ensemble of spin labels @xmath507 , initial velocities @xmath510 , and initial positions @xmath511 specified earlier . now as the spin - projections @xmath507 are not correlated with the initial positions or velocities , the averages factorize . the correlators of the @xmath507 are easily evaluated as : @xmath512 where @xmath513 and @xmath514 are simple , dimensionless , known functions of @xmath515 only . using ( [ mmcorr ] ) we have @xmath516 where @xmath517 is the spacetime dependent total density , all averages are now with respect to initial positions and velocities , and @xmath518 . the two - point correlators of @xmath519 are also easy to evaluate : if the spin labels are neglected , the collisions have no effect and correlators of the total denstiy can be obtained by considering an ideal gas of point particles . the second correlator in ( [ ab ] ) , multiplying @xmath520 , is more difficult : it involves the self two - point correlation of a given particle @xmath17 , which follows a complicated trajectory . fortunately , precisely this correlator was considered three decades ago by jepsen @xcite and a little later by others @xcite ; they showed that , at sufficiently long times , this correlator has a brownian motion form . in appendix [ jep ] , we give a self- contained summary of jepsen s calculation . here we just write down the final results @xcite for the correlation function : @xmath521 \label{czz1}\ ] ] where @xmath522 is the connected density correlator of a classical ideal gas in @xmath75 , @xmath523 and @xmath524 is the correlator of a given labeled particle @xcite , @xmath525 e^{-(g_2(u ) + g_2 ( -u ) ) \tilde{t } } \label{czz3}\end{aligned}\ ] ] with @xmath526 , @xmath527 , and @xmath528 . for @xmath529 , the function @xmath530 has the ballistic form @xmath531 , while for @xmath532 it crosses over to the _ diffusive _ form @xmath533 in the original dimensionful units , ( [ larget ] ) implies a spin diffusion constant , @xmath62 , given exactly by @xmath534 this result has been obtained by the solution of a classical model which possesses an infinite number of local conservation laws : in appendix [ integrable ] , we explicitly show how the existence of these local conservation laws is not incompatible with diffusive spin dynamics . it must be noted that the result ( [ diffres ] ) does not imply that we have rigorously established that the ultimate long - time corelations of the quantum model are also diffusive : the reasons for this and related comments were made earlier in section [ intro ] below ( [ dslowt ] ) . let us now summarize the calculation of the correlator of the transverse components of the magnetization density . the semiclassical prescription for evaluating @xmath535 is again quite straightforward : we begin with an initial state chosen from the same ensemble as before . @xmath536 acting on the initial state gives zero unless there is a particle at @xmath410 with spin label @xmath537 , @xmath197 , in which case it raises the @xmath396 value of that particle by @xmath14 and multiplies the state by a factor of @xmath538 ( coming from the usual properties of raising operators for the spin - one representation of the angular momentum algebra ) . the resulting state is then propogated forward in time with all the particles moving along their classical trajectories as before . at time @xmath412 , the operator @xmath539 acting on this state gives zero unless there is a particle at @xmath96 with spin label @xmath537 , @xmath14 , in which case it lowers the spin value of that particle by @xmath14 and again multiplies the state by a factor of @xmath540 . this state is then propogated backward in time and its overlap with the initial state calculated . @xmath535 is given by this overlap averaged over the ensemble of initial conditions . here , as before , the phase factor of @xmath197 coming from each collision does not matter as each collision has a time reversed counterpart . also , it is easy to see that in this case the overlap with the initial state is zero _ unless _ @xmath541 _ lowers the spin of precisely the particle whose spin was raised by _ lastly , we see that there is an overall factor of @xmath542 coming from the unitary time evolution as the total spin of the state during its evolution forward in time is greater than the total spin during its evolution backward in time by precisely one . similar considerations apply to @xmath543 . putting all of this together we see that @xmath544 where @xmath545 . now , we may express the nmr relaxation rate in terms of the correlation functions of the conserved magnetization density as @xmath546 where the local dynamic structure factor @xmath547 is defined as @xmath548 note that we have neglected the @xmath39 dependence of the hyperfine couplings and ignored the contribution of the antiferromagnetic spin fluctuations to the integral over @xmath39 in ( [ gennmrrate ] ) . at this point we have to address an important subtlety that arises in calculating the local dynamic structure factor from the autocorrelation function . we are treating the spin dynamics semiclassically to arrive at our expressions for the correlation functions . this gives rise to a characteristic @xmath549 divergence at short times in the corresponding autocorrelation functions . this is basically a signature of _ classical _ ballistic spin transport ; at these short time scales collisions play no role . as a result , the integral as written is logarithmically divergent at short times . our semiclassical expressions for the correlation functions do not make sense for very short times . this is natural as our whole approach has been geared towards calculating these correlations at time scales much larger than @xmath550 and length scales much larger than the thermal de broglie wavelength ; our method fails when both these conditions are simultaneously violated @xcite . the semiclassical expressions for @xmath551 are thus only valid for @xmath552 where @xmath553 is a short time cutoff @xmath554 . introducing this short time cutoff will give a well- defined result for @xmath555 at the price of introducing an arbitrary scale @xmath556 ; this does not seem very promising as our results for @xmath557 ( @xmath558 ) will depend sensitively upon @xmath553 except for very small fields such that we are in the collision dominated diffusive regime : @xmath559 ( @xmath560 ) . note that the range of @xmath44 for which the results are insensitive to the cutoff differs for the transverse components of the local dynamic structure factor because of the overall oscillatory factor of @xmath561 in the corresponding autocorrelation functions ( this factor always dominates as @xmath562 ) . however , we can still use our approach to compute the @xmath555 . the point is that , at very short times , the collisions between the thermally excited particles do not matter , and the spin dynamics is ballistic . this means that @xmath555 , for high frequencies @xmath563 ( such that @xmath563 is much larger than the mean collision rate @xmath564 ) , may be calculated exactly by doing a full quantum calculation for a gas of non - interacting spin - one particles @xcite . now , we can expand our semiclassical result ( obtained by using a cutoff @xmath553 ) for @xmath565 and _ match the leading term in this large @xmath44 expansion _ with the _ small @xmath44 limit _ ( for @xmath566 ( @xmath567 ) this would be the regime @xmath568 ( @xmath569 ) ) of the quantum calculation of ref @xcite . this , then , will uniquely fix @xmath553 and give us results for the @xmath555 that will work reasonably well even for @xmath570 ( though strictly speaking they are valid only in the range @xmath571 ( @xmath569 ) for @xmath566 ( @xmath567 ) ) . let us now use all of this to make contact with the experimental results of ref @xcite . for this particular experimental setup , the expression for @xmath9 simplifies and to a very good approximation we can write @xcite @xmath587 here the relevant hyperfine coupling constant is known @xcite to have the value @xmath588 ( note that we have used units such that @xmath24 in our computation of the correlation functions and thus time is being measured in inverse kelvins ) . to begin with , we straightforwardly attempt to fit the field dependent @xmath9 with our results . we use the values @xmath589 k and @xmath590 ( we are working in units where the lattice constant @xmath20 is set to one ) extracted from the susceptibility data @xcite . in actual fact , we introduce an additional , _ field - independent _ background rate @xmath591 that we add on to our theoretical result for @xmath9 . this serves as our fitting parameter ; we choose it at each temperature to achieve the best agreement with the results of ref @xcite . we show the resulting fits for @xmath592 , @xmath593 , and @xmath594 k in fig [ fig16 ] . we see that the theoretical curves account for the field dependence of @xmath9 _ extremely _ well in this temperature range ( of course the agreement for @xmath592 k should not be taken too seriously as our theory is valid only for temperatures smaller than the gap ) . in particular , the data seems to clearly exhibit the theoretically predicted @xmath76 divergence at low fields which is a characteristic of diffusive spin dynamics . in fig [ fig17 ] and fig [ fig18 ] , we compare the theoretical predictions with the experimental data at @xmath595 , @xmath596 , @xmath597 , @xmath598 , @xmath599 , and @xmath600 k. at these lower temperatures this divergence seems to get cut off below some threshold field and the quality of the fit deteriorates rapidly . this indicates the presence of some spin - dissipation mechanism which becomes significant at these lower temperatures and rounds off the diffusive @xmath601 divergence in the local dynamic structure factor . both inter - chain coupling and single - ion anisotropy of the intra - chain coupling are expected to contribute to the spin dissipation rate . however , we do not have any real theory that can work out the effects of these terms in the hamiltonian on the field and temperature dependence of @xmath9 . we can only attempt to phenomenologically introduce some spin dissipation in our theoretical results for the spin correlators . following @xcite , we do this by simply introducing an exponential cutoff to the long - time tail of the autocorrelation function ; thus we write @xmath602 . it is straightforward , though somewhat tedious to work out the corresponding local dynamic structure factor by doing the fourier transform and we will spare the reader the details . this now gives us a phenomenological result for @xmath9 with an additional tunable parameter @xmath603 . we choose this spin - dissipation rate at each temperature to achieve the best fit with the data . the resulting curves are shown in fig [ fig19 ] and fig [ fig20 ] for a few representative temperature values . we see that it is possible to fit the data moderately well ; discrepencies are however clearly visible and it is not clear how much significance to attach to the sharp increase in @xmath603 as the temperature is lowered . the quality of our fit seems at first sight to be much worse than the corresponding fit to a purely classical diffusive form employed in ref @xcite . however , it is important to note that the phenomenological model of ref @xcite used the diffusion constant as an additional fitting parameter ; we do not have any such freedom . moreover , both the diffusion constant and the constant background rate extracted from the fit in ref @xcite take on unphysical values below about @xmath596 k @xcite . this is because , at these lower temperatures , we are in the _ ballistic _ regime of spin - transport for a significant portion of the @xmath44 axis and the contribution from the ` free - boson logarithms ' can not be neglected . as the crossover to the ballistic regime is already incorporated in our form , the present results for the background rate do not suffer from any such obvious problems ( the diffusion constant of course is just given by ( [ diffres ] ) in our approach ) . in fig [ fig21 ] and fig [ fig22 ] , we plot the corresponding values of the spin - dissipation rate @xmath87 and the background rate @xmath591 as a function of temperature . the spin dissipation rate is seen to increase rapidly as the temperature is decreased . on the other hand , we see that the temperature dependence of @xmath591 may be fit approximately by an activated form with activation gap close to @xmath604 . we consider here the possibility that it may be possible to find gapped spin chains which satisfy @xmath606 , where @xmath78 is a typical exchange constant . in this case , it becomes possible to access a higher temperature regime where a continuum field theory description is possible in the regime @xmath607 . in particular , we expect that the continuum @xmath5-model to apply in such a regime @xcite . it is our hope that such a universal high @xmath0 regime can be experimentally accessed in @xmath81 spin chains @xcite . moreover , the study of such a high @xmath0 regime is of importance as matching its results with the @xmath57 theory can , in principle , help us estimate the values of @xmath0 to which the low @xmath0 results can be applied . an important property of this regime @xcite is that equal - time two - point correlator of @xmath608 , @xmath609 ( eqn ( [ defc ] ) ) decays at large @xmath96 with a correlation length @xmath610 we will shortly determine the exact values of the prefactor and the argument of the logarithm in ( [ hight1 ] ) . at distances of order or shorter than this correlation length we may crudely expect that a weak - coupling , spin - wave picture will hold , and excitations will have energy of order or smaller than @xmath611 , which is logarithmically smaller than the thermal energy @xmath0 ; in other words @xmath612 so the occupation number of these spin - wave modes will then be @xmath613 the last occupation number is precisely that appearing in a classical description of thermally excited spin waves , which suggests that a classical _ wave _ description should yield an appropriate picture of the dynamics of this high @xmath0 region . however , notice that classical thermal effects are only logarithmically preferred , and any predictions of a classical dynamical theory will only be correct to leading logarithms . having obtained the classical model ( [ 1drot28 ] ) for the equal - time correlations , and the precise value of the coupling @xmath85 in ( [ 1drot37 ] ) , we now turn to an examination of unequal time correlations in the high @xmath0 region @xmath77 . we employ an approach related to that used in the study of the quantum @xmath5-model in @xmath637 in ref @xcite in a _ low _ @xmath0 region ; unlike ( [ 1drot28 ] ) , the equal time correlations in @xmath637 were described by a theory that was not ultraviolet finite , and this will lead to significant differences in the analysis and physical properties here . to obtain classical equations of motions we clearly need to extend the classical hamiltonian in ( [ 1drot28 ] ) by including a kinetic energy term , expressed in terms of a canonical conjugate momentum to @xmath608 . the obvious approach is to take the quantum equations of motion , and to simply treat the variables as @xmath21-number classical degrees of freedom . in particular , we treat the rotor - angular momentum @xmath219 as a classical variable , and augment the classical hamiltonian by the kinetic energy of rotational motion . the moment of inertia of the rotor is related to the response of the system to a magnetic field @xmath44 , and we therefore need to study the behavior of @xmath60 in the @xmath77 regime . we will determine @xmath60 by strategy similar to that employed above in the computation of @xmath85 : first integrate out the non - zero frequency modes , and then perform the average over the zero frequency fluctuations . we choose an @xmath44 which rotates @xmath608 in the 12 plane , and define @xmath638 where @xmath617 , @xmath639 are a set of @xmath42 mutually orthogonal vectors in spacetime , and @xmath640 represent the finite frequency degrees of freedom which must be integrated out . we expand the partition function to quadratic order in @xmath44 , drop all terms proportional to the spatial gradients of @xmath617 or @xmath639 ( these can be shown to be logarithmically subdominant to the terms kept ) , and find that the @xmath44 dependent terms in the free energy density are @xmath641 \label{1drot38}\end{aligned}\ ] ] evaluating the expectation values of the @xmath642 fields , and using orthonormality of the vectors @xmath608 , @xmath643 , the expression ( [ 1drot38 ] ) simplifies to @xmath644 \label{1drot39}\end{aligned}\ ] ] finally to obtain the susceptibility @xmath60 , we have to evaluate the expectation value of the zero frequency field @xmath608 under the partition function ( [ 1drot28 ] ) . this simply yields @xmath645 . the first frequency summation is precisely the same as that evaluated earlier for @xmath85 in ( [ 1drot29 ] ) , while the second is explicitly finite in @xmath75 and can directly evaluated ; in this manner we obtain our final result for @xmath60 : @xmath646 % % @ \nonumber \\ & = & \frac{(n-2)}{n \pi c } \ln \left ( \frac{{\cal g } t}{\lambda_{\overline{ms } } e } \right ) \label{1drot40}\end{aligned}\ ] ] we have omitted the form of the subleading logarithms , which are the same as those in ( [ 1drot37 ] ) . this result was quoted earlier in ( [ chihight ] ) . we have now assembled all the information necessary to describe the effective classical dynamics in the region @xmath77 . the classical partition function is given by the following phase - space functional integral , which generalizes ( [ 1drot28 ] ) ( and we will now specialize the remainder of the discussion to the special case @xmath84 ) : @xmath647 \label{1drot44}\end{aligned}\ ] ] where @xmath648 is the classical angular momentum density , and @xmath648 , @xmath608 are classical commuting variables . the second term in @xmath649 was absent in ( [ 1drot28 ] ) , and represents the kinetic energy of the classical rotors : integrating out @xmath648 we obtain ( [ 1drot28 ] ) . by evaluating linear response to a field under which @xmath650 we find @xmath651 with @xmath84 ( we have given , without proof , the expression for general @xmath42 ) ; the factor of @xmath652 comes from the contraint @xmath653 . using ( [ 1drot40 ] ) , we then have the value of @xmath654 . we can finally specify the manner in which time - dependent correlations have to be computed in this effective classical model . the classical equations of motion are the hamilton - jacobi equations of the hamiltonian @xmath649 , with poisson brackets which are the continuum classical limit of the quantum commutation relations : @xmath655 the equations of motion are @xmath656 the classical correlation functions are obtained by averaging these deterministic equations over an ensemble of initial conditions specified by ( [ 1drot44 ] ) . note also that simple dimensional analysis of the differential equations ( [ 1drot45a ] ) shows that disturbances travel with a characteristic velocity @xmath657 given by @xmath658 notice from ( [ 1drot37 ] ) and ( [ 1drot40 ] ) that to leading logarithms @xmath659 , but the second term in the first equation of ( [ 1drot40 ] ) already shows that exact equality does not hold . we complete the relationship of the quantum to the classical model , by noting that there is also an additional wave - function renormalization of the @xmath608 field @xcite which appears when the non - zero frequency modes are integrated out . our final result for the correlator @xmath459 in ( [ defc ] ) then takes the form @xmath660^{\frac{(n-1)}{(n-2 ) } } \left\langle n_{\alpha } ( x , t ) n_{\beta } ( x , t ) \right\rangle_c \label{1drot48}\ ] ] the subscript @xmath21 represents the classical average specified by ( [ 1drot44 ] ) and ( [ 1drot45a ] ) . the constant @xmath53 is the @xmath8 quasi - particle residue which appeared in ( [ defcala ] ) . the constant @xmath661 is an unknown universal number which can not be obtained by the present methods . it could , in principle , be obtained from the bethe - ansatz solution . there is no similar renormalization of the correlator of the magnetization density , @xmath662 in ( [ defczzpm ] ) , which is precisely equal to the two - point correlator of @xmath648 under ( [ 1drot44 ] ) and ( [ 1drot45a ] ) . it is now possible to perform a simple rescaling and to show that the classical dynamics problem above is free of any dimensionless couplings , and is a unique , parameter - free theory . this will allow us to completely specify the @xmath0 dependence of observables upto unknown numerical constants . let us perform the following rescalings on ( [ 1drot44 ] ) and ( [ 1drot45a ] ) @xmath663 then the partition function ( [ 1drot44 ] ) is transformed to @xmath664 \label{1drot50}\end{aligned}\ ] ] while the equations of motion become @xmath665 notice that coupling constants and parameters have been scaled away , and ( [ 1drot50],[1drot51 ] ) constitute a unique problem that must be solved exactly . the @xmath0 and @xmath1 dependencies of all quantities arise only through the rescalings defined in ( [ 1drot49 ] ) and the results ( [ 1drot37 ] ) and ( [ 1drot40],[1drot41 ] ) for @xmath85 and @xmath654 given earlier . complete description of the correlators now requires exact solution of ( [ 1drot50],[1drot51 ] ) . the equal - time correlations are of - course known from ( [ 1drot50 ] ) : @xmath666 even though the equations of motion constitute an integrable system with an infinite number of non - local conservation laws @xcite , it is not known how to analytically compute correlations averaged over the initial conditions of a thermal ensemble , or whether the correlator @xmath667 has a diffusive form at long times and distances . if diffusion did exist in the continuum equations ( [ 1drot53 ] ) , the present analysis does allows us to completely specify the @xmath0 dependence of the diffusion constant ; by a simple dimensional analysis of ( [ 1drot49 ] ) , we get @xmath668^{3/2}}{[\chi_{u \perp } ( t)]^{1/2 } } \label{1drot54}\ ] ] where @xmath91 is an unknown universal number , and the @xmath0 dependencies of @xmath85 and @xmath654 are in ( [ 1drot37 ] ) and ( [ 1drot40],[1drot41 ] ) . in this context , it is interesting to note that recent measurements @xcite of the field dependence of @xmath9 in the compound ( vo)@xmath70p@xmath70o@xmath248 at temperatures @xmath669 seem to provide clear evidence for spin diffusion . however , the bulk of the data is at temperatures comparable to the microscopic exchange constants of the system and it is not clear if the foregoing description based on the universal high temperature properties of the continuum field theory is applicable in the temperature regime studied experimentally . it is interesting that the experimental results appear to suggest that @xmath670 , which is consistent with ( [ 1drot54 ] ) if @xmath671 ( as is the case with our results ( [ xihight ] ) and ( [ chihight ] ) to leading logarithms ) . the main results of the paper are already summarized in section [ intro ] , and here we will note some unresolved issues and directions for future work . all experimental realizations of gapped antiferromagnets have additional complications which have not been included in the model systems studied here . most important among these are the spin anisotropies away from perfect heisenberg symmetry and the inter - chain couplings which make the system only quasi - one - dimensional . consider first the consequence of anisotropy . the three - fold degenerate quasiparticle spectrum will now be lifted , and three resulting particles will have have different energy gaps and masses . further , these parameters will depend in a complicated way upon the external field . nevertheless , we expect that the simple structure of the @xmath3-matrix in ( [ smatrix ] ) will be retained , as it only depends upon simple dimensional properties of slowly moving particles with a quadratic dispersion . correlations of the particle density can probably be computed along the semiclassical lines of this paper : one has to deal with a classical gas of particles of different masses and average densities . the latter problem is considerably more complex than the equal mass case , and there is probably no alternative to numerical simulations . correlations of the spin operators appear more problematical these will invariable change the labels of the particles when they act , and therefore lead to differences in the labels in the forward and backward trajectories . combined with the complication that the masses of the different particles are different , and so their trajectories will have different velocities , we are faced with what appears to be a very complex problem with quantum and classical effects intertwined . inter - chain couplings will eventually require us to consider dynamics in two or three dimensions . if temperatures are low enough that the inter - chain motion is coherent , then we have to consider the @xmath3-matrix for scattering in higher dimensions . in this case the low - momentum behavior is quite different : in fact the @xmath0-matrix vanishes at low momenta for @xmath672 . we would then expect all scattering to be dominated by elastic scattering of impurities which would control the behavior of the spin diffusion constant and the quasi - particle broadening . on the other hand , systems with only incoherent hopping between chains will probably be dominated by the inelastic scattering along the one - dimensional chains , and display behavior qualitatively similar to that discussed in this paper . we are grateful to m. takigawa for initially stimulating our interest in the issue of spin transport in gapped antiferromagnets by informing us about the convincing experimental evidence for diffusion in ref @xcite , and for guiding us through the experimental literature . we thank g. aeppli , i. affleck , c. broholm , g. chaboussant , v. korepin , s. majumdar , b. mccoy , b. narozhny , b. normand , d. reich , t. senthil and x. zotos for helpful discussions . this research was supported by the national science foundation grant no . dmr-96 - 23181 . the computation of the spin diffusivity in section [ hdep ] was carried out using the exact solution a simple classical model of point particles in one dimension . this model is exactly solvable @xcite and possesses an infinite number of local conservation laws , as we will show explicitly below . the existence of spin diffusion then appears to run counter to the conventional wisdom that the time evolution of a integrable system is not ` chaotic ' enough to be compatible with diffusion . in particular , one might expect that any non - zero spin current produced in the system will not ultimately decay to zero because the numerous conservation laws prevent it . in this appendix we will show that this expectation does not hold for the particular model being studied , and that an important ` particle - hole'-like symmetry allows complete decay of any spin current at @xmath369 . in a finite magnetic field ( @xmath673 ) , the particle - hole symmetry is absent , and then the spin current does not decay completely : this is consistent with the presence of a purely ballistic component , @xmath674 , in ( [ czz1 ] ) which contributes only for @xmath675 ( @xmath676 at @xmath369 ) , and the arguments of zotos _ et al _ @xcite . a closely related particle - hole symmetry also played an important role in the appearance of a finite conductivity in our recent quantum transport analysis in two dimensions @xcite . the classical model of sections [ finitet ] and [ hdep ] consisted of particles @xmath677 with spins @xmath507 chosen randomly ( at @xmath369 ) from @xmath678 . at time @xmath405 the particles had uncorrelated random positions @xmath679 , and subsequently they occupied ` trajectories ' @xmath680 where @xmath681 are uncorrelated random velocities chosen from a boltzmann distribution . the position @xmath682 of particle @xmath17 was however a rather complicated function of time , and was chosen from the set of trajectories , @xmath683 , such that for all @xmath412 , @xmath684 for every @xmath422 . it is useful at this point to note two discrete symmetries of the above classical statistical problem at @xmath369 . the first is the time - reversal symmetry , @xmath685 , under which both spins and velocities change sign : @xmath686 the second is the ` particle - hole ' symmetry @xmath687 , under which only the spins reverse direction : @xmath688 these symmetries will be crucial in our discussion below . let us now explicitly identify the local conserved quantities of this classical dynamics . all of the velocities @xmath681 are clearly constants of the motion . however , we would like to work with locally conserved quantities which can be written as the spatial integrals over local observables , and which are invariant under permutation of the particle labels ; so we define @xmath689 \nonumber \\ & = & \sum_{k=1}^{n } v_k^n \label{integ1}\end{aligned}\ ] ] with @xmath690 ( notice @xmath691 , but the result holds after summation over @xmath17 because the set @xmath692 differs from the set @xmath693 only by a renumbering ) . all the @xmath694 are constants of the motion . similarly , with spins @xmath507 we can define @xmath695 with @xmath696 , as additional locally conserved quantities ( @xmath314 is the spatial integral of @xmath697 in ( [ varrho ] ) , and a similar result holds of @xmath315 ) . we can now easily work out the signature of the @xmath694 and @xmath698 under the discrete symmetries noted earlier , and tabulate the results : @xmath699 the central quantity in spin transport is the total spin current @xmath700 , which is not a constant of the motion . it is also given by an integral over a local quantity as @xmath701 \nonumber \\ & = & \sum_{j , k=1}^{n } m_j v_k a_{jk } ( t ) , \label{integ3}\end{aligned}\ ] ] where @xmath702 is defined to be equal to 1 if particle @xmath703 is on trajectory @xmath17 at time @xmath412 and 0 otherwise ; we will analytically study the function @xmath702 in appendix [ jep ] , but here we will be satisfied by a numerical simulation . again , as in ( [ integ2a ] ) it is useful to note the signature of @xmath78 under the discrete symmetries : @xmath704 as will become clear shortly , one of the central points of this appendix is that the signatures in ( [ integ3a ] ) differ from all of those of the conserved quantities in ( [ integ2a ] ) . the current @xmath700 is the sum of @xmath42 random numbers of each sign , and so is expected to be of order @xmath705 for a typical initial condition chosen from the ensemble defined above . we show the _ deterministic _ time evolution of @xmath700 for one such initial condition for a system of 400 particles in fig [ fig23 ] : notice that it is rather noisy - looking and repeatedly changes its sign . also , among the constants of the motion above , we expect @xmath694 with @xmath706 odd and @xmath314 to be of order @xmath705 ( provided @xmath706 is not too large ) , and @xmath694 with @xmath706 even and @xmath315 to be of order @xmath42 for a typical initial condition ; notice that it is only the conserved quantities of order @xmath705 that can distinguish left movers from right movers , or spin up from down . let us now create a macroscopic spin current ( of order @xmath42 ) in this system . we do this by hitting the system with a magnetic field gradient impulse at a time @xmath707 , and subsequently setting the field to zero . as a result of the impulse , the velocities of the particles with spin up are assumed to increase by @xmath708 , while those of spin down are assumed to decrease by @xmath708 . formally , this can be written as @xmath709 immediately after the impulse , @xmath700 will have a macroscopic value @xmath710 the subsequent deterministic time evolution of @xmath700 is also shown in fig [ fig23 ] : it decays in a few collision times to a value of order @xmath705 and then appears to chaotically oscillate in time ! the basic point is now easy to see . because @xmath507 is as likely to be @xmath397 or @xmath197 , the impulse on any given particle is equally likely to be @xmath711 or @xmath712 . hence the @xmath694 , with @xmath706 odd , remain of order @xmath705 even after the impulse . this is simply a manifestation of the fact that the signatures of @xmath78 under @xmath687 and @xmath685 are different from those of the @xmath694 . a non - zero @xmath78 is therefore not correlated with an induced value of a conserved quantity which could prevent the decay of @xmath78 to non - macroscopic values . in this appendix we outline the numerical method employed in calculating @xmath37 starting from the numerically determined semiclassical @xmath384 and the procedure used to directly determine the scaling function @xmath713 ( see eqn ( [ dirscalfn ] ) ) as the numerical determination of @xmath452 is the most time consuming part of the entire procedure , we calculated @xmath439 only at a predetermined grid of points in the @xmath714 plane . we chose @xmath454 values from @xmath176 to @xmath715 at intervals of @xmath716 . for each such value of @xmath454 , we chose about @xmath717 points so as to sample @xmath439 as well as possible in the region in which @xmath718 ; this choice was made to reflect the fact that our absolute error in @xmath439 was estimated to be about @xmath719 . this then defined our grid . at each @xmath454 , we fit @xmath439 as a function of @xmath720 to the form @xmath721 where @xmath722 and @xmath723 . the rationale behind our choice of the value of @xmath724 is as follows : when @xmath725 , the complicated correlations between the spin labels of a given classical trajectory at different times do not matter and @xmath261 is well approximated by our ` mean - field ' theory ( see appendix [ mf ] ) . the mean - field theory in this limit gives @xmath726 and this is what determines our choice of @xmath724 . the error in the fit was estimated to be roughly the same as the error in the original computation of @xmath439 ; thus we did not lose anything by doing the fit . having tabulated the fitting parameters for each value of @xmath454 on the grid , we evaluated the spatial fourier transform numerically . the resulting function of @xmath454 is expected to be smooth as long as @xmath727 is not too large . more precisely , we do not expect any oscillations on the scale of our grid spacing in @xmath454 as long as @xmath728 . as we are interested only in @xmath729 , we can safely interpolate the resulting function in @xmath454 . in practice we use a cubic - spline to do the interpolation . lastly , we do the @xmath454 integral numerically to obtain @xmath37 . the accuracy of both numerical integrations is quite high and so we expect that the dominant error in our calculation comes from the interpolation ; this is conservatively estimated to be a few percent at the most for the largest values of @xmath730 . let us now briefly indicate the procedure used in obtaining the fourier transform of @xmath731 needed for the calculation of the scaling function @xmath713 . the available data for @xmath731 is fit _ well by the following form : @xmath732 where the choice @xmath733 is again motivated by ` mean - field ' considerations . it is now a simple matter to do the fourier integral to a very high accuracy using this fit and we estimate the errors involved to be less than @xmath734 at the most . in this appendix , we shall attempt to give a self - contained account of the method devised by jepsen @xcite for the calculation of the tagged particle correlations in the classical model introduced in ref @xcite . we will try to adhere to the notation and conventions of @xcite as far as possible . the model is defined as follows : we begin with @xmath42 particles of mass @xmath396 distributed uniformly along a one - dimensional segment of length @xmath219 with periodic boundary conditions ( we will eventually take the thermodynamic limit @xmath735 with @xmath736 fixed to be equal to the density @xmath253 ) . at time @xmath405 each particle is assigned a velocity from the classical thermal ensemble defined by the usual maxwell - boltzmann distribution function @xmath737 . the subsequent evolution of the system is purely deterministic ; the particles travel without any change in their velocities until they collide with another particle . every collision is elastic and the particles merely exchange their velocities as a result of the collision . to begin our analysis , let us label the particles from left to right with an index @xmath148 running from @xmath176 to @xmath738 . thus the particles are initially at positions @xmath739 such that @xmath740 for @xmath741 . actually , it is convenient to identify @xmath742 with @xmath148 because of the periodic boundary conditions employed which identify the ends @xmath410 and @xmath743 of the interval . note that this labelling of the particles is left invariant by the dynamics . we also label _ trajectories _ ( which follow the straight line defined by @xmath744 on the space - time diagram representing the evolution of the system ) with an index @xmath148 , again with the convention that @xmath745 for @xmath741 ( here @xmath746 is the initial velocity of the @xmath747 particle ) . let @xmath748 denote the _ position _ of the @xmath747 particle at time @xmath412 . we wish to calculate the correlator @xmath749 where summation over the repeated index @xmath17 is implied and the angular brackets refer to averaging over the ensemble of initial conditions specified earlier . let us now consider the quantity @xmath750 , introduced in appendix [ integrable ] , which is defined to be equal to @xmath14 if particle @xmath703 is on trajectory @xmath17 at time @xmath412 and @xmath176 otherwise . another useful quantity is the number @xmath751 of ( _ signed _ ) crossings suffered by the @xmath424 trajectory upto time @xmath412 . every time this trajectory is hit from the left , @xmath751 decreases by @xmath14 and every time it is hit from the right @xmath751 increases by @xmath14 . clearly , @xmath752 for @xmath753 and zero otherwise . we may probe the dynamics a bit more by defining another quantity @xmath754 which equals @xmath14 if trajectory @xmath755 has crossed trajectory @xmath17 precisely @xmath706 times upto time @xmath412 and zero otherwise . here too , we are talking of _ signed _ crossings ; if trajectory @xmath755 crosses from the left this is a negative crossing and if it crosses from the right it is a positive crossing . clearly @xmath756 has the interpretation of a probability when averaged over any ensemble of initial conditions . let us also define the corresponding ` generating function ' as @xmath757 the reason for introducing @xmath756 and @xmath758 is that @xmath750 , which is clearly a central quantity of interest , may be very conveniently expressed in terms of @xmath758 as @xmath759 here we are using the convention that @xmath760 . this is quite easy to check from the defintions of @xmath758 and @xmath702 . moreover , it is possible to write down a fairly explicit expression for @xmath761 . this takes a slightly different form depending on whether @xmath755 is greater or less than @xmath17 . if @xmath762 , we have @xmath763~ , \label{hgreaterk}\ ] ] while if @xmath764 , we have @xmath765~. \label{hlessk}\ ] ] @xmath766 $ ] used above is defined as : @xmath767 = e^{inu}~ , \label{definebigs}\ ] ] where @xmath768 , and @xmath706 is the integer that satisfies @xmath769 . using this definition , we can write the following compact expression for @xmath750 in terms of @xmath106 : @xmath770~. \label{finala}\ ] ] with all this machinery in place , it is a relatively straightforward matter to calculate the correlation function we need . we begin by explicitly writing out the ensemble averages involved : @xmath771\left ( \prod_{l=0}^{n-1 } % % @ g(v_l)\right)\delta(x_0(t)-x)~ , \label{explicitav}\ ] ] where we have used the definitions @xmath772 and @xmath773 \equiv \int_{-\infty}^{\infty}dv_{n-1}\int_{-%%@ \infty}^{\infty}dv_{n-2}\ldots \int_{-\infty}^{\infty}dv_{0}~$ ] with @xmath774 , and it is understood that @xmath775 is set equal to @xmath176 when evaluating the right hand side of ( [ explicitav ] ) . now we can transform from particle positions to trajectories by writing @xmath776 using this and writing @xmath777 in terms of @xmath766 $ ] allows us to express our correlation function as @xmath778\int \left [ dv\right ] \left ( \prod_{l=0}^{n-1 } % % @ g(v_l)\right)\frac{1}{n}\sum_{k , l=0}^{n-1}\prod_{h=0}^{n-1}s\left[\frac{2\pi % % @ l}{n},w_{kh}\right]\delta(x_k(t)-x)~. \label{btrajrep}\end{aligned}\ ] ] here we have also used the fact that the integrand in this representation is explicitly symmetric in the spatial integration variables to change the spatial integration to @xmath779 \equiv\int_{0}^{l}dx_{n-1}\int_{0}^{l}dx_{n-%%@ 2}\ldots \int_{0}^{l}dx_{1}$ ] . in this appendix , we briefly outline our approximate ` mean - field ' theory for the relaxation function @xmath418 . we begin by noting that the classical model defined in section [ finitet ] has been solved exactly in ref @xcite for the special case in which there is only one possible value for the spin label @xmath396 . all of the difficulties we encounter in attempting to generalize this solution to the case of interest here stem from the fact that there are complicated correlations between the @xmath430 ( defined in section [ finitet ] ) at different times . our ` mean - field ' approximation consists of simply ignoring these correlation effects ( hence our choice of terminology to describe our approximation ) . having made this uncontrolled approximation , it is now a fairly straightforward matter to obtain a closed form expression for @xmath418 in analogy with the corresponding discussion in @xcite . the actual calculation proceeds as follows : let @xmath39 be the probability that any given solid line in fig [ fig9 ] intersects the dotted line . if we ignore the correlations between the @xmath430 at different times , then the probability that this line carries a spin label equal to the spin label of the dotted line is @xmath800 . so given that the line intersects the dotted line , this intersection contributes a factor of @xmath197 to @xmath418 with probability @xmath800 and a factor of @xmath176 with probability @xmath652 ( if the line does not intersect at all , we of course get a factor of @xmath14 ) . within our mean field theory , @xmath261 is just a product of such factors , one from each solid line . this gives @xmath801 , where @xmath42 is the total number of thermally excited particles in the system . now , using @xmath802 @xcite ( where the angular brackets denote averaging over the maxwell - boltzmann distribution function for @xmath803 and @xmath219 is the length of the system ) and taking the thermodynamic limit , we obtain @xmath804 . we can now do our usual rescalings and write down the main result of our mean field theory : @xmath805 where the angular brackets now denote averaging over the distribution @xmath806 and @xmath720 and @xmath454 are defined as in section [ finitet ] . in particular , note that this implies @xmath807 ; this turns out to be reasonably accurate for some purposes ( see the discussion on the approximate form of the scaling function @xmath713 in section [ finitet ] ) . by c. broholm in _ dynamical properties of unconventional magnetic systems _ , proceedings of the nato advanced study institute , geilo , norway , april 2 - 12 , 1997 , a. skjeltorp and d. sherrington eds . , kluwer academic . by r. a. cowley in _ dynamical properties of unconventional magnetic systems _ , proceedings of the nato advanced study institute , geilo , norway , april 2 - 12 , 1997 , a. skjeltorp and d. sherrington eds . , kluwer academic . the simplest non - local conserved charge in ref . @xcite , @xmath808 , does have the same quantum numbers as the spin current . the discussion in appendix [ integrable ] would then suggest that @xmath808 could prevent complete decay of the spin current . however the non - local nature of @xmath808 suggests to us that this is unlikely . in a system with @xmath42 thermally excited particles , @xmath808 consists of two components : a local piece with order @xmath42 terms which is proportional to the spin current at low momentum , and a second non - local piece with order @xmath809 terms . in a typical thermal state , the naive expectation is that the value of the local piece will be of order @xmath705 , while that of the non - local piece will be of order @xmath42 . if we now set up a spin current with an impulse , as in appendix [ integrable ] , the local piece will jump to a term of order @xmath42 . however this does not change the magnitude of the total value of @xmath808 , and it is reasonable to expect that , within a few collision times , there will be a redistribution within the two pieces of @xmath808 so that their orders of magnitude are similar to those in a typical thermal state . by s. sachdev in _ dynamical properties of unconventional magnetic systems _ , proceedings of the nato advanced study institute , geilo , norway , april 2 - 12 , 1997 , a. skjeltorp and d. sherrington eds . , kluwer academic ; report no . cond - mat/9705266 . by i. affleck in _ dynamical properties of unconventional magnetic systems _ , proceedings of the nato advanced study institute , geilo , norway , april 2 - 12 , 1997 , a. skjeltorp and d. sherrington eds . , kluwer academic . by r. shankar in _ current topics in condensed matter and particle physics _ , proceedings of the bcspin school , katmandu , may 1991 , j. pati , q. shafi , and yu lu eds . , world scientific ( 1993 ) ; also r. shankar , acta phys . pol . * 26 * , 1835 ( 1995 ) . k. pohlmeyer , comm . * 46 * , 207 ( 1976 ) ; m. luscher and k. pohlmeyer , nucl . phys . * b137 * , 46 ( 1978 ) . the discussion in also applies to the non - local conserved charges , @xmath810 , in the second paper . also the classical model has the property ( not present in the quantum model ) that all disturbances travel precisely with the velocity @xmath657 : this could lead to special singular structure in the correlation functions at @xmath811 .
we present the theory of nonzero temperature ( @xmath0 ) spin dynamics and transport in one - dimensional heisenberg antiferromagnets with an energy gap @xmath1 . for @xmath2 , we develop a semiclassical picture of thermally excited particles . multiple inelastic collisions between the particles are crucial , and are described by a two - particle @xmath3-matrix which has a _ super - universal _ form at low momenta . this is established by computations on the @xmath4 @xmath5-model , and strong and weak coupling expansions ( the latter using a majorana fermion representation ) for the two - leg @xmath6 heisenberg antiferromagnetic ladder . as an aside , we note that the strong - coupling calculation reveals a @xmath7 , two particle bound state which leads to the presence of a second peak in the @xmath8 inelastic neutron scattering ( ins ) cross - section for a range of values of momentum transfer . we obtain exact , or numerically exact , universal expressions for the thermal broadening of the quasi - particle peak in the ins cross - section , for the magnetization transport , and for the field dependence of the nmr relaxation rate @xmath9 of the effective semiclassical model : these are expected to be asymptotically exact for the quantum antiferromagnets . the results for @xmath9 are compared with the experimental findings of takigawa _ et.al . _ and the agreement is quite good . in the regime @xmath10 we argue that a complementary description in terms of semiclassical waves applies , and give some exact results for the thermodynamics and dynamics .
leptogenesis is one of the most attractive mechanisms to explain the origin of the baryon asymmetry of the universe ( bau ) @xcite . this is so because it arises naturally in simple extensions of the standard model ( sm ) which can also explain why the neutrino masses are so tiny . in this mechanism a lepton asymmetry is produced in the out of equilibrium decay of heavy majorana neutrinos , which is then partially converted into a baryon asymmetry by non - perturbative sphaleron processes ( see @xcite for a complete review ) . in the most economical model ( type i seesaw ) the heavy neutrinos @xmath0 are sm singlets with majorana masses @xmath1 , which only interact with the lepton doublets @xmath2 and higgs field @xmath3 via yukawa interactions , @xmath4 . the number of new parameters associated to the type i seesaw with three singlets , one per each sm family , is 18 . however the baryon asymmetry @xmath5 produced via @xmath6-leptogenesis depends on a few combinations of them ( here @xmath7 and @xmath8 are the baryon and entropy densities ) . when flavour effects @xcite are not relevant the main parameters are @xmath9 , which determines the epoch of leptogenesis , @xmath10 , that gives a measure of the amount of cp violation per @xmath6-decay ( see below for a precise definition ) , and the effective mass , @xmath11 ( with @xmath12 the vev of the higgs field ) , which is an appropriate measure of the intensity of the yukawa interactions of @xmath6 . if the cp asymmetry @xmath10 is constant during leptogenesis ( which is usually a good approximation ) , the final baryon asymmetry , @xmath13 , is simply proportional to @xmath10 . in this case it can be expressed as @xmath14 , with @xmath15 a numerical factor and @xmath16 is the so called efficiency , which carries the dynamical information and it is mainly a function of @xmath17 . by definition @xmath18 and the maximum efficiency is obtained when @xmath19 ev . this value is determined by the condition that the decay rate of @xmath6 equals the hubble expansion rate at a temperature @xmath20 , so that the yukawa interactions of @xmath6 are barely out of equilibrium at the time it becomes non - relativistic . this result is amazing given that the contribution of @xmath6 to the masses of the light neutrinos ( @xmath21 ) in the type i seesaw is expected to be of the same order as @xmath17 ( barring cancellations due to phases ) . in other words , an efficient leptogenesis mechanism suggests a scale for the light neutrino masses which is roughly of the correct order of magnitude . moreover , the most simple models for leptogenesis require an interesting upper bound for the masses of the light neutrinos , @xmath22 ev in the one flavour approximation @xcite and @xmath23 few ev when flavour effects are taken into account @xcite . the conditions described above provide a highly non - trivial connection between baryogenesis via leptogenesis and the low energy parameters @xmath24 . but this is not enough at all to probe leptogenesis . unfortunately in the most simple models no more generic relations can be established between leptogenesis and low energy parameters . in fact , for a very hierarchical spectrum of heavy singlet neutrinos @xmath25 , the @xmath26-violating cp asymmetry generated in the decay of the lightest singlet has an upper bound proportional to @xmath9 , the so - called davidson - ibarra ( di ) bound @xcite . this implies a lower bound @xmath27 gev for the mass of the sterile neutrinos in order for @xmath6-dominated leptogenesis to be successful . careful numerical studies show that the di bound can be evaded for moderate hierarchies , e.g. the lower bound on @xmath9 is relaxed by more than one order of magnitude with respect to the hierarchical limit one for @xmath28 @xcite . however to reach these low values of @xmath9 some unlikely cancellations are needed , which are not motivated by any underlying symmetry . flavour effects do not substantially change this result . therefore leptogenesis occurs at very high energies in these scenarios . in addition no generic relation can be made between low and high energy phases , i.e. leptogenesis can work for any value of the observable pmns phases @xcite . in conclusion , leptogenesis in the context of the type i seesaw with hierarchical heavy neutrinos provides a simple and natural explanation to the bau , but it will not be possible to test this scenario in foreseeable experiments . this has motivated research in different directions . for example in @xcite some ways to falsify ( rather than probe ) leptogenesis at the lhc were investigated . also , one can avoid the di bound resorting to resonant leptogenesis , i.e. , a resonant enhancement of the cp asymmetry which occurs when there are at least two strongly degenerated heavy neutrinos , such that @xmath29 , being @xmath30 their decay width @xcite . in this scenario , leptogenesis is feasible at much lower temperatures , @xmath31 @xcite . however it is not enough to have leptogenesis at the tev scale in order to probe it . this is so because the most crucial parameters for observing effects from the heavy neutrinos are the active - sterile neutrino mixings , which in the type i seesaw are roughly given by the ratio @xmath32 ( with @xmath33 ) , and hence are too small . therefore it is very interesting that there are well motivated seesaw models which not only yield a heavy neutrino quasi - degenerate spectrum but can also provide a large active - sterile neutrino mixing , namely those that have an approximately conserved @xmath34 @xcite ( with @xmath26 being conserved at the perturbative level ) . in these models the tiny neutrino masses are proportional to small lepton number - breaking parameters , which are technically natural since a larger symmetry is realized when they vanish . this implies that the heavy neutrinos can be much lighter than in the generic seesaw , within the energy reach of lhc . also , lepton flavour violating rare decays as well as non - unitarity of the leptonic mixing matrix are present even in the limit of conserved @xmath34 , and therefore they are unsuppressed by the light neutrino masses @xcite . as a consequence , much attention has been devoted recently to this class of low scale seesaw models , since they have a rich phenomenology both at lhc @xcite and at low energy charged lepton rare decay experiments , such as @xmath35 , and also lead to successful resonant leptogenesis @xcite . it has also been noticed that even if the heavy neutrinos that generate the bau are not quasi dirac , or the mass splitting is outside the resonant regime , in seesaw models with almost conserved @xmath34 the scale of leptogenesis can be lower than in the standard seesaw @xcite , provided flavour effects are at work . this is so because there is a @xmath26-conserving part in the flavoured cp - asymmetries which escapes the di bound . in these notes we review and summarize the results of @xcite on the possibility of having successful leptogenesis driven by the purely flavoured @xmath26-conserving contribution to the cp asymmetries , in the context of seesaw models with small violation of @xmath34 . in general there are at least two species of neutrinos involved in leptogenesis , one , called here @xmath6 , which is mainly responsible for the generation of the lepton asymmetry during its production and decay , and another one , @xmath36 , that makes the most important virtual contribution to the cp asymmetry in @xmath6 decays . if @xmath34 is only slightly violated , then each @xmath0 must satisfy one of the two following conditions : * @xmath0 is a majorana neutrino with two degrees of freedom , whose yukawa interactions violate lepton number and therefore the couplings @xmath37 must be small . * the @xmath0 is a dirac or quasi - dirac neutrino with four degrees of freedom ; this means that there are two majorana neutrinos @xmath38 and @xmath39 with masses @xmath40 and @xmath41 respectively . the parameter @xmath42 measures the amount of @xmath34 violation , so that if @xmath34 is conserved , @xmath43 and @xmath44 is a dirac fermion . the yukawa interactions can be expressed as @xmath45 where @xmath46 . the terms proportional to @xmath47 induce lepton number violation even when @xmath48 and hence they are similar in nature to the ones described in ( i ) . instead the @xmath49 can be large , because they do not vanish in the @xmath34 conserved limit : in the absence of @xmath50 and @xmath47 , a perturbatively conserved lepton number can be defined , by assigning @xmath51 to @xmath0 , and @xmath52 to the sm leptons . there are two cases that are relatively easy to analyze , * * ( iia ) @xmath53 ( dirac limit ) , and * * ( iib ) @xmath54 ( majorana limit ) . + here @xmath55 and @xmath56 are the decay widths of @xmath38 and @xmath39 respectively . a comprehensive research on leptogenesis in models with small violation of @xmath34 can be obtained considering the different possibilities ( i ) or ( ii ) for both @xmath6 and @xmath36 . since we have not considered the widely - studied case of a resonant contribution of @xmath36 to the cp asymmetry in @xmath6 decays , the optimum situation for generating a lepton asymmetry is when @xmath36 satisfies ( ii ) . in this way the cp asymmetries @xmath57 in the decays of @xmath6 into leptons of flavour @xmath58 , @xmath59 , being proportional to the yukawa couplings of @xmath36 , can be enhanced . in turn , for @xmath6 the simplest possibility is ( i ) . it can also satisfy ( iib ) , in which case @xmath60 and @xmath61 behave as two independent majorana neutrinos regarding the generation of the bau , that would roughly double with respect to case ( i ) . however if @xmath6 satisfies ( iia ) , then it is ( or effectively behaves as ) a dirac neutrino , i.e. lepton number is conserved in its decays , and therefore the only possibilities to end up with a non - zero bau is to have important washouts from the two majorana components of @xmath36 ( if @xmath62 ) or let the sphalerons freeze out during leptogenesis @xcite . motivated by the previous discussion we have considered a scenario for leptogenesis involving three fermion singlets @xmath63 ( each of them having two degrees of freedom ) , with respective masses @xmath64 and yukawa couplings given by the lagrangian @xmath65 the parameters @xmath66 violate lepton number and hence @xmath67 . as shown in @xcite it is convenient to take @xmath68 in order to obtain the lowest energy scale for leptogenesis within this framework , which corresponds to the so called @xmath6-leptogenesis . then @xmath13 is proportional to the cp asymmetries @xmath57 , which have a @xmath26-violating part suppressed by the small @xmath26-violating parameter @xmath69 and an unsuppressed @xmath26-conserving piece , @xmath70 , whose contribution to the total cp asymmetry is null , i.e. @xmath71 . in order to have a large @xmath13 ( not suppressed by @xmath69 ) , it is mandatory to have flavour effects so that there is a contribution to @xmath13 coming from @xmath70 @xcite . in turn , to have the appropriate flavour effects , it is crucial to demand that the couplings of @xmath6 and @xmath36 be small enough , such that the yukawa interactions of the @xmath72 are the strongest ones ( see @xcite for a detailed explanation ) . when this happens the density matrix of leptons is diagonal in the orthogonal basis @xmath73 , with @xmath74 and @xmath75 being determined by the fastest interaction acting in the plane perpendicular to @xmath76 . something similar occurs in the antilepton sector . then , as a first approximation , the lepton asymmetries in the flavours @xmath77 , and @xmath75 evolve independently . in this case , although @xmath71 , @xmath13 can get contributions from the individual @xmath70 . this is so because the final amount of lepton asymmetry in a given flavour also depends on how much of the produced asymmetry was erased , and this can be different for each flavour . actually the evolutions of the different lepton flavour asymmetries are not completely independent . on one hand , spectator processes @xcite effectively couple the flavour asymmetries , nevertheless we have checked that their effect on @xmath13 is at most a few tens of percent . one the other hand , there are @xmath26-conserving but @xmath78-violating scatterings @xmath79 , @xmath80 , and @xmath81 , hereafter called generically flavour changing interactions ( fci ) , which are inherent to models with an approximately conserved @xmath34 . the fci play a crucial role because they tend to equilibrate the different flavour asymmetries @xcite , effectively diminishing flavour effects and consequently @xmath13 . the cross sections of the fci have been calculated in @xcite , finding important differences with previous literature . summarizing , in order to determine the bau generated in models for leptogenesis with small violation of @xmath34 , it is very important to consider the fci and the adequate conditions for having flavour effects . the set of boltzmann equations ( be ) taking into account these elements can be read in @xcite . for the case @xmath82 the be are like the ones typically found in the literature . instead , if @xmath83 then @xmath84 and @xmath85 combine to form a dirac neutrino @xmath86 , and therefore there is an asymmetry generated among the degrees of freedom of @xmath36 which has to be taken into account @xcite . the relevant parameters for leptogenesis are @xmath9 , @xmath87 , @xmath88 , @xmath89 , the projectors @xmath90 , and @xmath69 . we have determined the minimum value of @xmath9 compatible with successful leptogenesis as a function of @xmath87 , maximizing @xmath13 over the remaining parameters . to obtain the baryon asymmetry we have solved numerically the appropriate set of be scatterings @xcite . however we have checked that their inclusion modifies the results by at most a few tens of percent . ] , and to get successful leptogenesis we have required @xmath91 @xcite . the result is represented with the thick continuous curves in fig . [ fig:1 ] , the red line corresponding to the case @xmath82 and the green one to @xmath83 . yielding successful leptogenesis as a function of @xmath87 . the red curves are for the case @xmath82 and the green ones for @xmath83 . the thick continuous curves give the physically correct bound , while the thin dashed ones show the result that would be obtained if the yukawa couplings of @xmath36 were allowed to take values as large as 1 for all values of @xmath87 . ] as can be seen it is possible to have neutrino masses as low as @xmath92 gev , i.e. around three orders of magnitude below the lower bound for the standard case of type i seesaw with hierarchical heavy neutrinos . such lower bound on @xmath9 in turn yields a lower bound for the reheating temperature , @xmath93 , of the same order , since to thermally produce the neutrinos @xmath94 @xcite . an interesting consequence is that the bound @xmath95 gev can be compatible with the upper bound on @xmath93 required to avoid the gravitino problem in sugra models @xcite . moreover , @xmath9 values around @xmath96 gev can be achieved for a wide range of @xmath36 masses and also for different values of the yukawa couplings ( see @xcite for details on this point as well as on the relation between the parameters defined above and the light neutrino masses ) . an important issue for obtaining the bound on @xmath9 has been to determine how large the yukawa couplings of @xmath36 can be without violating the condition that the rates of processes involving @xmath36 be slower than the rates of the @xmath72-yukawa interactions . for comparison we have also plotted in fig . [ fig:1 ] the -wrong- bound that would be obtained if @xmath89 were allowed to be as large as 1 . it is clear that as @xmath97 approaches @xmath9 the requirement of an upper bound for @xmath89 becomes very relevant . finally let us comment that for simplicity the results depicted in fig . [ fig:1 ] were obtained assuming that @xmath98 is perpendicular to the decay eigenstates of @xmath6 and @xmath36 , so that only two flavour asymmetries are generated . we have checked that in the more general three flavour case it is possible to lower the bound on @xmath9 by a factor up to almost 4 with respect to the two flavour case @xcite . seesaw models with an almost conserved @xmath34 are an interesting alternative to explain the smallness of neutrino masses because they can lead , in principle , to large active - sterile neutrino mixings . we have found another merit of these models , namely that leptogenesis is possible for @xmath99 gev , i.e. around three orders of magnitude below the standard type i seesaw case , without resorting to the resonant enhancement of the cp asymmetry for strongly degenerate heavy neutrinos . however , it is also clear that such energy scale is too large to have both , successful non - resonant leptogenesis and active - sterile neutrino mixings large enough to produce observable effects . the author wishes to thank his collaborators on this subject , concha gonzlez - garca , manuel pea and nuria rius . + this work has been supported by the spanish mineco subprogramme juan de la cierva and it has also been partially supported by the spanish mineco grants fpa-2007 - 60323 , fpa2011 - 29678-c02 - 01 , and consolider - ingenio cup ( csd2008 - 00037 ) . in addition we acknowledge partial support from the european union fp7 itn invisibles ( marie curie actions , pitn- ga-2011- 289442 ) .
we review the motivations and some results on leptogenesis in seesaw models with an almost conserved lepton number . the paper is based on a talk given at the 5th international symposium on symmetries in subatomic physics , ssp2012 . + * preprint * : ific/12 - 58
recently the production of @xmath1 and @xmath4 meson by @xmath4 . ] mesons in @xmath3 collisions at @xmath5 tev was studied with the cdf detector at the fermilab tevatron @xcite . the @xmath1 and @xmath4 mesons were reconstructed from their @xmath6 decay modes . the inclusive production cross sections for both charmonium states were measured as functions of the transverse momentum ( @xmath7 ) in the central region , corresponding to the rapidity ( @xmath8 ) range @xmath9 . the cdf collaboration was able to extract individual cross sections for @xmath1 and @xmath4 mesons originating from weak decays of @xmath2 hadrons and from prompt production . in this way , rather accurate data for the inclusive @xmath1 and @xmath4 cross sections coming from the decays of @xmath0 mesons or other hadrons containing @xmath2 quarks were obtained for @xmath7 values between 5 and 20 gev . these cross sections were compared with theoretical predictions based on next - to - leading ( nlo ) calculations in quantum chromodynamics ( qcd ) with massive @xmath2 quarks @xcite and subsequent fragmentation of the @xmath2 quarks into @xmath2 hadrons . the @xmath2-hadron decays to @xmath10 and @xmath11 were described by a parametrization of the momentum distribution measured by the cleo collaboration @xcite . it was found that the data for @xmath1 ( @xmath4 ) production were higher than the qcd prediction by a factor of 24 ( 34 ) depending on the @xmath7 of the @xmath1 ( @xmath4 ) meson . however , for larger @xmath7 values , _ @xmath12 gev , the experimental cross sections could be reproduced by the theoretical calculations if the scale @xmath13 , the @xmath2-quark mass @xmath14 , and the parameter @xmath15 in the peterson @xcite fragmentation function ( ff ) were simultaneously reduced from @xmath16 to @xmath17 , from @xmath18 gev to @xmath19 gev , and from @xmath20 to @xmath21 , respectively . even with this choice of parameters , the measured cross section for @xmath1 production was still a factor of 2 above the prediction at @xmath22 gev . in the case of @xmath4 production , the cross section predicted with the modified parameters is still below the data , although the discrepancy was only about one standard deviation at large @xmath7 . for the nominal @xmath1 and @xmath4 predictions , it was assumed that the nonperturbative part of the fragmentation of @xmath2 quarks into @xmath2 hadrons can be described by a peterson ff with @xmath23 . this value for @xmath15 was extracted more than ten years ago from a global analysis of data on @xmath0 production in @xmath24 annihilation at petra and pep @xcite , based on monte carlo ( mc ) models which were in use at that time . due to the nonperturbative nature of the peterson ff , the choice of @xmath15 must be backed up by other independent data , _ e.g. _ on @xmath25 , which must be analysed within the very theory that is used for the interpretation of the cdf data , _ i.e. _ nlo qcd with massive quarks ( @xmath26 ) and fixed order in @xmath27 ( massive scheme @xcite ) . in @xmath24 annihilation , this theory is only reliable just above threshold , where almost no data exist , except from the argus and cleo experiments at the @xmath28 resonance . in this case , however , the @xmath0 mesons are only produced in pairs and not in the fragmentation mode . therefore , the underlying description of @xmath29 fragmentation in the massive scheme , on which the comparison with the cdf data is based , is to a large extent _ ad hoc _ and not supported by the analysis of independent data within the same nlo perturbative scheme . reliable information on the ff for @xmath2 hadrons can only be gained from the high - statistics experiments at cern lep 1 . at lep 1 , the production of @xmath2 quarks is enhanced as compared with the lower @xmath24 energies , so that the produced @xmath2 hadrons can be identified more easily through their weak decays . the fragmentation into @xmath0 mesons was measured by the opal collaboration at lep 1 @xcite . based on these data , we recently constructed ff s for @xmath0 mesons using three different forms for the ff at the starting scale @xmath30 , including the one by peterson _ et al . _ @xcite , which yielded the lowest @xmath31 values at leading order ( lo ) and nlo @xcite . in ref . @xcite , the @xmath29 ff was obtained using the so - called massless scheme @xcite . in this scheme , the parton - level cross sections are calculated with @xmath32 , and the appearing collinear final - state singularities are factorized into the ff s according to the modified minimal - subtraction ( @xmath33 ) scheme . a nonvanishing value for @xmath14 only appears in the initial conditions for the ff s . this scheme provides the appropriate approach for describing the fragmentation of @xmath2 quarks into @xmath0 mesons at the @xmath34-boson resonance , since the @xmath2 quarks in the reaction @xmath35 typically have large momenta . a large - momentum @xmath2 quark essentially behaves like a massless particle , radiating a large amount of its energy in the form of hard , collinear gluons . this leads to the well - known logarithms of the form @xmath36 originating from collinear radiation in a scheme where @xmath14 is taken to be finite . these terms appear in all orders of perturbation theory . the method for summing them is to introduce ff s and to absorb the @xmath14-dependent logarithms into their evolution up to the factorization scale of order @xmath37 . if all terms of @xmath38 are neglected and the @xmath33 subtraction scheme is adopted , then this approach is equivalent to the massless scheme , where one puts @xmath32 from the beginning . in the massless scheme , a nonperturbative ff is easily incorporated . this allows one to transfer the information on the fragmentation process in one reaction to other processes , _ e.g. _ @xmath39 . this was done in ref . @xcite , where the @xmath2-quark ff s obtained from fits to the opal data were used to predict the differential cross section @xmath40 of @xmath0 production in @xmath3 scattering at @xmath5 tev . these predictions were compared with data from the cdf collaboration @xcite and found to agree with them within errors . in this paper , we use these results for @xmath0 production to predict the distribution @xmath40 of @xmath1 and @xmath4 mesons originating from @xmath2-hadron decay . for this purpose , we need a realistic description of the inclusive @xmath0 decays into @xmath1 and @xmath4 mesons . we adopt the parton - model description from ref . @xcite , which allows one to nicely interpret the cleo data @xcite . while only @xmath41 and @xmath42 mesons and their antiparticles , which we collectively call @xmath0 mesons in the following , are produced at cleo energies , the cdf @xmath1 and @xmath4 samples also contain contributions from @xmath43 mesons , @xmath44 mesons , @xmath45 baryons , etc . we may safely ignore @xmath44 production , whose rate is expected to be about @xmath46 of the total @xmath2-hadron production rate @xcite . the @xmath47 quark in the initial state should enhance the branching ratio to charmonium , but @xmath44 production is expected to account for @xmath48 of all events of charmonium production from @xmath2-hadron decay at the tevatron @xcite . unfortunately , there exist no measurements of the differential cross sections of @xmath43 and @xmath45 production at lep 1 , which could be used to extract ff s for these hadrons . moreover , there are no data on the momentum distributions of the @xmath1 and @xmath4 mesons inclusively produced in @xmath43 and @xmath45 decays . only the branching fractions of @xmath49 and @xmath50 ( here @xmath45 stands for a collection of @xmath2 baryons ) are known , which are approximately 10% each @xcite . thus , the dominant channels for @xmath2-quark fragmentation are @xmath51 and @xmath52 . due to the lack of detailed knowledge of the fragmentation into @xmath43 mesons and @xmath45 baryons and their inclusive @xmath1 and @xmath4 decay properties , we must estimate the @xmath43 and @xmath45 contributions using information on @xmath29 fragmentation and on the momentum distributions of the inclusive @xmath53 and @xmath54 decays . for simplicity , we assume that the @xmath49 and @xmath50 ff s are proportional to those of @xmath51 @xcite , and scale the latter by the factor @xmath55 and similarly for @xmath50 adopting the values for these branching ratios from ref . furthermore , we approximate the momentum distributions of the @xmath1 ( @xmath4 ) mesons from the @xmath43 and @xmath45 decays by the one of @xmath53 ( @xmath54 ) . this paper is organized as follows . in sect . 2 , we recall the framework for calculating the cross section of @xmath0 production in @xmath3 collisions closely following our earlier work @xcite . in sect . 3 , the momentum distributions of @xmath1 , @xmath56 , and @xmath4 mesons from @xmath0 decay , which enter the predictions of charmonium production from @xmath2-hadron decay at the tevatron , are described in the framework of nonrelativistic qcd ( nrqcd ) @xcite . in sect . 4 , the relevant charmonium matrix elements are determined from cdf @xcite and cleo @xcite data on prompt charmonium production . in sect . 5 , we calculate the @xmath7 spectra of @xmath1 and @xmath4 mesons originating from @xmath2-hadron decay at the tevatron and compare them with available cdf @xcite and d0 @xcite data . section 6 summarizes our conclusions . @xcite , we presented lo and nlo predictions for the inclusive cross section of @xmath0 production in @xmath3 collisions with @xmath5 tev at the tevatron . this analysis provides the basis for the theoretical description of the inclusive production of @xmath1 and @xmath4 mesons from @xmath2 hadrons at the tevatron . having obtained the inclusive @xmath0 cross section @xmath40 , this cross section is convoluted with the appropriately boosted longitudinal - momentum distributions of the @xmath1 and @xmath4 mesons from @xmath0 decay , which are considered in the next section . in this way , we obtain the bulk of the cross section , more than 80% of it , from which the additional contributions originating from the production of @xmath43 mesons and @xmath45 baryons may be estimated . before we come to these points , we shortly recapitulate the input that was assumed in ref . the formalism used in ref . @xcite is very similar to ref . @xcite , where inclusive light - meson production in @xmath3 collisions was studied in the qcd - improved parton model . we work at nlo in the @xmath33 scheme with @xmath57 massless flavors . in this respect , we differ from any of the massive calculations @xcite , where only @xmath58 active flavors are taken into account . for the proton and antiproton parton density functions ( pdf s ) we use the set cteq4 m @xcite with asymptotic scale parameter @xmath59 mev or the more recent set mrst @xcite of the durham - oxford - rutherford group with @xmath60 mev , which corresponds to @xmath61 mev and a @xmath2-quark mass of @xmath62 gev @xcite . the authors of ref . @xcite presented new fits to the relevant deep - inelastic - scattering data incorporating a more realistic description of the heavy - quark pdf s , with an improved extrapolation from scales near threshold to higher scales . for the @xmath0 ff s , we adopt set nlo p @xcite , which uses the peterson form @xcite for the @xmath29 ff at the initial scale and yielded the best fits to the opal data @xcite . the two alternative sets s and b , which are provided in ref . @xcite , lead to almost identical predictions for inclusive @xmath0 hadroproduction . the evolution of the ff s is performed with @xmath63 mev @xcite . the strong coupling constant @xmath64 is evaluated from the two - loop formula adopting the @xmath65 value from the selected set of proton pdf s . we identify the factorization scales associated with the proton , antiproton , and @xmath0 meson and collectively denote them by @xmath66 . we choose the renormalization and factorization scales to be @xmath67 , where @xmath68 is a number of order unity and @xmath69 is the transverse mass of the produced charmonium state , which we generically denote by @xmath70 . unless otherwise stated , we put @xmath71 . we recall that , in the case of inclusive @xmath0 production discussed in ref . @xcite , we instead took @xmath72 to be the @xmath0 transverse momentum . in the present case , we consider it more appropriate to relate @xmath72 to the final - state @xmath70 state . the rationale is that we combine the @xmath29 fragmentation and @xmath73 decay processes into one single process . this scale convention also considerably simplifies the computations , since the evolution of the @xmath0 ff s are performed numerically in @xmath74 space @xcite . later on , we study the scale dependence of our results in order to get some handle on the theoretical uncertainty related to this arbitrariness in convention . we adopt all kinematic conditions from ref . @xcite . when we present lo results , they are consistently computed using set cteq4l @xcite ( mrstlo @xcite ) of proton pdf s , set lo p of @xmath0 ff s @xcite , the one - loop formula for @xmath64 with @xmath75 mev @xcite ( @xmath76 mev @xcite ) and the lo hard - scattering cross sections . the inclusive decay @xmath53 has three sources : the prompt production of @xmath1 mesons and the two feed - down modes @xmath54 followed by @xmath77 and @xmath78 , with @xmath79 , followed by @xmath80 . in the recent cleo measurements @xcite , these three sources were disentangled and found to have the branching fractions @xmath81 , @xmath82 , and @xmath83 , respectively . the total inclusive @xmath1 branching fraction was measured to be @xmath84 @xcite . the inclusively produced @xmath4 mesons are all believed to be promptly produced . the cleo collaboration also measured the momentum distribution of the @xmath53 decay including all three sources and the one of the @xmath85 decay . furthermore , they presented the momentum distribution of the prompt @xmath53 decay , _ i.e. _ with the contributions due to the two feed - down channels subtracted . in order to predict the inclusive cross section of @xmath1 mesons from @xmath0 decay in @xmath3 collisions , we need the total @xmath1 momentum distribution , _ i.e. _ the sum of the prompt and the two feed - down contributions . so , the easiest way to incorporate the cleo information would be to boost back the cleo data , measured in the cleo laboratory frame , into the @xmath0 rest frame . in this system , the distribution is assumed to be isotropic . this distribution must then be boosted from the @xmath0 rest frame along the momentum direction of the produced @xmath0 meson to the tevatron laboratory frame . this procedure can easily be implemented into an event generator to be used for the analysis of the experimental data @xcite . our procedure of incorporating the cleo information is somewhat different . we first calculate , within nrqcd @xcite , the momentum distribution of @xmath1 mesons promptly produced from @xmath0 decay in the @xmath0 rest system . from this we derive the longitudinal - momentum distribution of the @xmath1 mesons for @xmath0 mesons in flight by performing an appropriate boost and integrating over the transverse momentum components . to adjust the input parameters , the momentum distribution in the @xmath0 rest system is boosted to the cleo laboratory frame and compared with the cleo data . this already accounts for 71% of the @xmath1 mesons coming from @xmath0 decay . the same procedure can be applied to the @xmath54 decay mode . as we shall see in sect . 4 , in the case of @xmath78 , the nrqcd prediction falls short of the cleo data @xcite , so that its normalization must be adjusted by hand . for simplicity , the feed - down from @xmath4 and @xmath56 mesons to @xmath1 mesons is accounted for by including the respective branching fractions and assuming that the @xmath1 mesons receive the full momentum of the primary charmonium states . the momentum spectrum of the prompt , inclusive @xmath53 decay has been investigated just recently by palmer , paschos , and soldan @xcite in the nrqcd approach @xcite . in this approach , the decay @xmath86 is represented by a sum of products , each of which consists of a short - distance coefficient for the creation of a @xmath87 pair in a specific angular - momentum ( @xmath88 ) and color ( @xmath89 ) configuration and a nonperturbative nrqcd matrix element @xmath90\right\rangle$ ] , which parametrizes the subsequent evolution of the intermediate @xmath91 $ ] state into the physical @xmath1 state ( plus light hadrons ) via the emission of soft gluons . in their work , the transition @xmath92 is described by two alternative models , the parton model and the model based on fermi momentum smearing @xcite . these models differ in the description of the long - distance transition @xmath92 . in the parton model approach , the transition @xmath92 is parametrized by a distribution function similar to the one used in deep - inelastic scattering . in the fermi motion approach , the bound - state corrections to the free-@xmath2-quark decay are incorporated by giving the spectator quark a fermi motion inside the @xmath0 meson . since both models rely on unknown parameters , which are fixed by the cleo data on the @xmath53 prompt , inclusive decay , we expect that the fine details of describing the @xmath92 transition are irrelevant for our purposes . in particular , as we shall see , the momentum distribution of the @xmath1 mesons boosted to the tevatron laboratory frame only depends on the bulk properties of the momentum distribution in the @xmath0 rest system . in the following , we adopt the parton - model description of @xmath92 . in the perturbative nrqcd framework , one first calculates the decay rate @xmath93+q)$ ] of a @xmath2 quark going to a @xmath87 pair in state @xmath94 $ ] plus a light quark @xmath95 , where @xmath96 . then , one obtains from this the @xmath86 decay rate by exploiting the nrqcd factorization theorem . in the @xmath2-quark rest frame , the final formula reads @xcite @xmath97 , \label{eq : par}\ ] ] where @xmath98 is fermi s constant , @xmath99 is an element of the cabibbo - kobayashi - maskawa matrix , and @xmath100 and @xmath2 are nrqcd coefficients given by @xmath101\right\rangle } { 3m_c^2 } + ( c_++c_-)^2 \nonumber\\ & & { } \times \left [ \frac{\left\langle{\cal o}^{j/\psi}[\,\underline{8},{}^3\!s_1]\right\rangle } { 2m_c^2 } + \frac{\left\langle{\cal o}^{j/\psi}[\,\underline{8},{}^3\!p_1]\right\rangle } { m_c^4}\right ] , \nonumber\\ b&=&(c_++c_-)^2 \frac{\left\langle{\cal o}^{j/\psi}[\,\underline{8},{}^1\!s_0]\right\rangle } { 2m_c^2}.\end{aligned}\ ] ] here , it is assumed that @xmath102 and @xmath103 . @xmath104 are the short - distance coefficients appearing in the effective weak - interaction hamiltonian that describes the @xmath105 transition . we observe that the @xmath86 decay rate essentially depends on two parameters , @xmath100 and @xmath2 , which are given in terms of the charm - quark mass @xmath106 , the short - distance coefficients @xmath104 , and the nrqcd color - singlet and color - octet matrix elements . in our numerical analysis , we use @xmath107 gev@xmath108 , @xmath109 , @xmath110 , where @xmath111 gev @xcite is the @xmath1 mass , @xmath112 , and @xmath113 @xcite . the nrqcd matrix elements are specified in sect . the inclusive decay of the @xmath0 meson is described by introducing the structure function @xmath114 for the transition @xmath92 . it characterizes the distribution of the @xmath2-quark momentum inside the @xmath0 meson and is usually parametrized by the peterson form @xcite , @xmath115 ^ 2},\ ] ] where @xmath15 is a free parameter and @xmath116 is determined so that @xmath114 is normalized to 1 . it is well known that the ff for the transition @xmath29 peaks at large @xmath74 . since we expect a similar behaviour for the inverse transition @xmath92 , the peterson ansatz should be well suited . specifically , we adopt the value @xmath117 , which was determined in the lo analysis of ref . @xcite . using the formalism of ref . @xcite , we obtain the following formula for the inclusive @xmath53 decay width , differential in the @xmath1 three - momentum @xmath118 , in an arbitrary frame , where the @xmath0 meson has three - momentum @xmath119 : @xmath120 \left[f(x_+)+f(x_-)\right]\right . \nonumber\\ & & { } + \left.(a+b)\frac{m_b^2}{2m_\psi^2}(x_++x_-)(x_+-x_- ) \left[f(x_+)-f(x_-)\right]\right\ } , \label{eq : cov}\end{aligned}\ ] ] where @xmath121 , @xmath122 , and @xmath123 here , a prime is introduced to discriminate quantities referring to the system with finite @xmath119 from their counterparts in the @xmath0 rest frame . a similar formula applies to the inclusive @xmath54 decay . equation ( [ eq : cov ] ) is used in two ways . first , it is applied in the cleo laboratory frame , where @xmath124@xmath125 gev , after integration over the direction of @xmath118 , to constrain the theoretical input entering @xmath100 and @xmath2 by comparison with the cleo data @xcite . for the prompt @xmath53 decay , this was already done in ref . @xcite . a convenient alternative to doing the angular integration in the cleo laboratory system is to carry it out in the @xmath0 rest frame and then to perform a boost to the cleo laboratory system . in the @xmath0 rest frame , the decay width differential in @xmath126@xmath124 is easily obtained as @xcite @xmath127 \left[f(x_+)+f(x_-)\right]\right . \nonumber\\ & & { } + \left.(a+b)\frac{m_b^2}{2m_\psi^2}(x_++x_-)(x_+-x_- ) \left[f(x_+)-f(x_-)\right]\right\ } , \label{eq : pps}\end{aligned}\ ] ] where @xmath128 is now given by @xmath129 here , @xmath130 varies in the interval @xmath131 , where @xmath132 in the cleo laboratory frame , we then have @xmath133 where @xmath134 here , @xmath135 varies in the interval @xmath136 , where @xmath137 notice that @xmath138 on the left - hand side of eq . ( [ eq : res ] ) refers to the @xmath0 rest frame , so that integrating eq . ( [ eq : res ] ) over @xmath135 yields the proper @xmath53 decay width . the second application of eq . ( [ eq : cov ] ) concerns the calculation of the @xmath53 decay distribution in the component @xmath139 of the @xmath1 three - momentum that is parallel to @xmath119 . for this purpose , we must integrate eq . ( [ eq : cov ] ) over the orthogonal momentum component @xmath140 . this can be directly done leading to @xmath141 \left[f(x_+)+f(x_-)\right]\right . \nonumber\\ & & { } + \left.(a+b)\frac{m_b^2}{2m_\psi^2}(x_++x_-)(x_+-x_- ) \left[f(x_+)-f(x_-)\right]\right\ } , \label{eq : lon}\end{aligned}\ ] ] where @xmath142 and @xmath143 . for most of our applications , @xmath144 is satisfied . in this limit , an asymptotic formula for @xmath145 , where @xmath146 , can be derived . defining @xmath147 and @xmath148 , we have @xmath149 \right . \nonumber\\ & & { } \times\left.\left[f(x_+)+f(x_-)\right ] + \frac{a+b}{2}(x_+-x_-)^2\left[f(x_+)-f(x_-)\right]\right\ } , \label{eq : inf}\end{aligned}\ ] ] where @xmath150 and @xmath151 . an alternative way of calculating the distribution in eq . ( [ eq : lon ] ) is to transform the @xmath140 integration into an integration over @xmath130 in the @xmath0 rest frame , in a way similar to eq . ( [ eq : res ] ) . this leads to @xmath152 where @xmath153 and @xmath154 are given in eqs . ( [ eq : pps ] ) and ( [ eq : max ] ) , respectively , and @xmath155 with @xmath156 . as already mentioned in sect . 2 , we incorporate the @xmath157 decay in our calculation of inclusive @xmath0 hadroproduction by defining effective ff s for the transition of the partons @xmath158 that come out of the hard scattering to the @xmath70 mesons as @xmath159 where @xmath160 are the nonperturbative ff s determined in ref . @xcite , @xmath161 is the total @xmath0 decay width , and @xmath162 is obtained from eqs . ( [ eq : lon ] ) or ( [ eq : alt ] ) . for given @xmath70 transverse momentum @xmath7 and rapidity @xmath8 , @xmath163 in eq . ( [ eq : eff ] ) is given by @xmath164 , where @xmath72 is the @xmath70 transverse mass . for reasons explained above , we choose the factorization scale to be @xmath165 . in our numerical analysis , we use the @xmath166 average values @xmath167 gev and @xmath168 ps @xcite for the @xmath0 mass and lifetime , respectively . in the remainder of this section , we discuss the decays @xmath78 , with @xmath79 , within the nrqcd framework . they were not considered in ref . they contribute via radiative feed - down to the yield of @xmath1 mesons from @xmath2-hadron decay measured at the tevatron and must , therefore , be included in our analysis . we find that eq . ( [ eq : par ] ) and the subsequent equations derived from it carry over to these cases if we substitute everywhere @xmath70 by @xmath56 , insert @xmath169\right\rangle}{2m_c^2 } , \nonumber\\ a&=&(2c_+-c_-)^2\frac{2\left\langle{\cal o}^{\chi_{c0 } } [ \,\underline{1},{}^3\!p_0]\right\rangle}{m_c^4 } + ( c_++c_-)^2\frac{3\left\langle{\cal o}^{\chi_{c0 } } [ \,\underline{8},{}^3\!s_1]\right\rangle}{2m_c^2 } , \nonumber\\ a&=&(c_++c_-)^2 \frac{5\left\langle{\cal o}^{\chi_{c0 } } [ \,\underline{8},{}^3\!s_1]\right\rangle}{2m_c^2},\end{aligned}\ ] ] for @xmath79 , respectively , and put @xmath170 in all three cases . in this section , we determine the leading nrqcd colour - octet matrix elements of the @xmath1 , @xmath4 , and @xmath56 mesons from fits to the cdf data of their prompt hadroproduction @xcite , imposing the requirement that the branching fractions of the inclusive @xmath0 decays into these charmonium states , calculated as described in sect . 3 , agree with the cleo results @xcite whenever this is possible . this complements our previous analysis @xcite , where we determined the @xmath1 and @xmath56 colour - octet matrix elements from the cdf data @xcite alone . we start by repeating the @xmath1 analysis of ref . @xcite for the @xmath4 mesons . we extract @xmath171\right\rangle$ ] from the measured @xmath172 decay width @xcite using the qcd - improved formula ( 3 ) of ref . we then determine the leading colour - octet matrix elements from the cdf data sample of prompt @xmath4 hadroproduction @xcite , which contains 5 data points . similarly to the @xmath1 analysis reported in ref . @xcite , we obtain @xmath173\right\rangle$ ] from the upper part of the @xmath7 spectrum ( last 2 data points ) working in the fragmentation picture , where the @xmath87 bound state is created from a single high - energy gluon , charm quark or antiquark which is close to its mass shell . in a second step , we extract the linear combination @xmath174\right\rangle + \frac{r}{m_c^2 } \left\langle{\cal o}^{\psi^\prime}[\,\underline{8},{}^3\!p_0]\right\rangle , \label{eq : mr}\ ] ] from the lower part of the @xmath7 spectrum ( first 4 data points ) adopting the fusion picture , where the @xmath87 bound state is formed within the primary hard - scattering process . here , @xmath175 is chosen in such a way that the superposition of these two channels is insensitive to precisely how they are weighted relative to each other . the results obtained with set cteq4l @xcite of proton pdf s are listed in table [ t1 ] together with our previous lo results for @xmath1 mesons @xcite . our result for @xmath176 agrees very well with the value @xmath177 gev@xmath178 found by cho and leibovich @xcite from a fit to earlier cdf data , while our result for @xmath173\right\rangle$ ] is somewhat larger than their value @xmath173\right\rangle= ( 4.6\pm1.0)\cdot10^{-3}$ ] gev@xmath178 . figure [ f1 ] illustrates , for the fusion picture , how the theoretical cross section of prompt @xmath4 hadroproduction compares with the corresponding cdf data @xcite and how it is decomposed into its colour - singlet and colour - octet components . the total cross section in the fragmentation picture , which is complementary to the fusion picture , is not shown in fig . [ f1 ] because it would be hardly distinguishable from the solid line . it is almost entirely due to the @xmath179 $ ] channel , while the @xmath180 $ ] contribution is approximately down by a factor of 60 . the goodness of the fits in the fusion and fragmentation pictures is measured in terms of the @xmath31 per degree of freedom , @xmath181 . these values and their combinations are also given in table [ t1 ] . .values of the @xmath1 and @xmath4 matrix elements resulting from the minimum-@xmath31 fits to the cdf data @xcite . @xmath182 is defined in eq . ( [ eq : mr ] ) . [ cols="^,^,^",options="header " , ] we now turn to the @xmath78 decays . the relevant matrix elements for their description in the nrqcd framework are @xmath183\right\rangle$ ] and @xmath184\right\rangle$ ] . for our analysis , we adopt the value @xmath183\right\rangle= ( 8.80\pm2.13)\cdot10^{-2}$ ] gev@xmath185 from ref . @xcite , where is was determined from the measured hadronic decay widths of the @xmath56 mesons . using the central value of this result , we then find @xmath184\right\rangle = ( 1.39\pm0.17)\cdot10^{-3}$ ] gev@xmath178 from a lo fit to the cdf data on prompt @xmath56 hadroproduction @xcite . using the nrqcd formalism described in sect . 3 with this input , we evaluate the @xmath78 branching fractions for @xmath79 to be @xmath186 , @xmath187 , and @xmath188 , respectively . these values are incompatible with the cleo results , @xmath189 and @xmath190 @xcite , where the first error is statistical and the second systematic . @xmath191 decays were not observed by cleo . obviously , the cdf data on @xmath192 and the cleo data on @xmath78 can not be simultaneously interpreted to lo in the nrqcd framework . the inclusion of the nlo corrections to the @xmath78 decay width does not remove this discrepancy @xcite . in want of a theoretically satisfactory solution to this problem , we choose a purely phenomenological description . we calculate the momentum spectrum of the @xmath1 mesons from @xmath0 decay via the @xmath56 states to lo in nrqcd , using the above matrix elements , an average @xmath56-meson mass of 3.495 gev , and the values of @xmath193 from ref . @xcite , and adjust its normalization in accordance with the cleo result , to be @xmath194 @xcite . without this normalization factor , we would have @xmath195 . the cleo collaboration did not specify experimental data on the momentum spectrum of the @xmath1 mesons originating from the @xmath56 states . however , they published the measured decay spectrum for the combined @xmath1 sample , including prompt production and feed - down from @xmath196 , @xmath197 , and @xmath4 states . in fig . [ f2](c ) , these data are compared with the superposition of our nrqcd results . notice that both experimental and theoretical results have the same normalization . in this section , we present our predictions for the inclusive production of @xmath1 and @xmath4 mesons originating from @xmath2-hadron decay in @xmath3 collisions at the tevatron and compare them with available cdf @xcite and d0 @xcite data . as described in sect . 3 , the cross section of this process emerges from the one of inclusive @xmath2-hadron production by convolution with the longitudinal - momentum distribution of the @xmath1 and @xmath4 mesons from @xmath2-hadron decay appropriately boosted along the @xmath2-hadron flight direction . in fig . [ f3 ] , we investigate how the @xmath53 branching fraction @xmath198 differential in @xmath146 , as given by eqs . ( [ eq : lon ] ) or ( [ eq : alt ] ) , is distorted by a boost to the frame where the @xmath0 meson has nonvanishing three - momentum @xmath163 . specifically , we consider the cases @xmath199 , 10 , and 20 gev . for comparison , also the asymptotic line shape according to eq . ( [ eq : inf ] ) , which refers to the infinite - momentum frame , is shown . we observe that the finite-@xmath163 results rapidly converge towards the asymptotic form . in order to obtain the effective @xmath200 ff s via @xmath0 decay , we need to convolute the @xmath53 decay distribution shown in fig . [ f3 ] with our @xmath201 ff s @xcite according to eq . ( [ eq : eff ] ) . having fixed all relevant input from fits to the opal data on @xmath25 @xcite , the cdf data on @xmath202 via prompt production @xcite , and the cleo data on @xmath73 @xcite , we are now in a position to make absolute predictions for charmonium production via @xmath2-hadron decay in @xmath3 collisions . in figs . [ f4](a ) and ( b ) , we consider @xmath1 production , summing over the prompt channel and the feed - down channels via @xmath56 and @xmath4 states . in fig . [ f4](a ) , the @xmath7 distribution @xmath40 measured by cdf @xcite in the central region of the detector , for @xmath9 , is compared with our lo and nlo predictions based on the cteq4 @xcite and mrst @xcite pdf s and scale choice @xmath71 . in fig . [ f4](b ) , the @xmath68 dependence of these predictions is analyzed for three representative values of @xmath7 . as a rule , the theoretical uncertainty may be estimated from three criteria : ( i ) the difference between the lo and nlo predictions ; ( ii ) the shift due to scale variations ; and ( iii ) the variation between different pdf sets . all these points may be studied with the help of figs . [ f4](a ) and ( b ) . we find good agreement with the experimental data , especially in the upper @xmath7 range , where the theoretical uncertainty is smallest . as expected , the nlo results are there more stable under scale variations than the lo ones . at the lower end of the @xmath7 spectrum , the nlo results undershoot the data , but exhibit sizeable normalization uncertainties . the cteq and mrst pdf s lead to very similar results , except at very low @xmath7 . in fig . [ f5 ] , the @xmath1 analysis of fig . [ f4](a ) is repeated for @xmath4 mesons . here , the agreement between experimental data and nlo predictions is also good at low @xmath7 , where the theoretical uncertainty is largest . the cdf @xcite and d0 @xcite collaborations also published their full data samples on @xmath1 production without discriminating between prompt production , feed - down from charmonium states with higher masses , and @xmath2-hadron decay . these cross sections are both differential in @xmath7 , but are complementary in the sense that the cdf data are integrated over the central region , with @xmath9 , while the d0 data are sampled in the forward and backward directions , with @xmath203 . in figs . [ f6](a ) and ( b ) , we compare these data with our predictions obtained from the formalism explained here and in ref . @xcite with the nonperturbative charmonium matrix elements determined in sect . 4 . we work at lo using the cteq4l pdf s @xcite . apart from the total contributions , also the partial contributions due to prompt production , feed - down from @xmath56 mesons , feed - down from @xmath4 mesons , and @xmath2-hadron decay ( summing over the prompt and feed - down channels ) are shown . in the case of cdf , there is good agreement over the full @xmath7 range considered . this is not surprising because the contributing matrix elements were actually determined from the prompt @xmath1 , @xmath56 , and @xmath4 data samples taken in the same experiment . furthermore , we already know from fig . [ f4](a ) that the predicted @xmath2-hadron decay contribution agrees well with the cdf data . so far , the d0 collaboration did not present separate cross sections for the individual channels of @xmath1 production . the agreement between their combined cross section and the theoretical prediction , which is essentially tuned to fit the cdf data , is remarkably good , except at the upper end of the @xmath7 spectrum . in a way , [ f6](b ) represents an indirect comparison between the cdf and d0 data on @xmath1 production , which refer to different kinematic regimes . we notice that , at large @xmath7 , the cross sections of the prompt and @xmath2-hadron channels are of the same order for the cdf central production [ see fig . [ f6](a ) ] , whereas , in the case of d0 forward ( backward ) production , the prompt channel plays the dominant rle [ see fig . [ f6](b ) ] . this means that the cross sections of these two channels have different rapidity dependences . this could be checked experimentally if these cross sections were measured separately also in the forward ( backward ) direction . we considered the hadroproduction in @xmath3 collisions of @xmath2 hadrons which subsequently decay to charmonia and presented theoretical predictions for the @xmath7 distribution of the latter . the formation of the @xmath2 hadrons was described in the qcd - improved parton model with ff s fitted to @xmath24 data on @xmath0-meson production @xcite . in want of detailed experimental information of the ff s of @xmath43 mesons and @xmath45 baryons , we approximately accounted for their contributions by appropriately adjusting the normalization of our @xmath0 ff s @xcite . the @xmath0 decays to the various charmonium states were treated in nrqcd @xcite with nonperturbative matrix elements determined from the cdf @xcite and cleo @xcite data on prompt - charmonium production in @xmath3 collisions and @xmath0 decay , respectively . in the case of the @xmath78 decays , the nrqcd results for their branching fractions came out considerably smaller than the cleo results @xcite , so that we had to include a phenomenological magnification factor . furthermore , we assumed that the decays of @xmath43 mesons and @xmath45 baryons to charmonia , for which no data exist so far , can be described on the same footing as the @xmath0 decays . we compared the cdf data on @xmath1 and @xmath4 production from @xmath2-hadron decay @xcite with our theoretical predictions and found good agreement . we estimated the theoretical uncertainty by comparing lo and nlo predictions , by scale variation , and by varying the pdf s and found that it was of order @xmath204 for @xmath205 gev . we also compared the full cdf @xcite and d0 @xcite data samples on @xmath1 production without distinguishing between prompt , feed - down , and @xmath2-hadron decay channels with our predictions . again , we found good agreement . since the cdf and d0 data refer to different ranges of rapidity , this may also be regarded as an indirect comparison between them . p. nason , s. dawson , and r. k. ellis , nucl . * b303 * , 607 ( 1988 ) ; * b327 * , 49 ( 1989 ) ; for a review , see s. frixione , m. l. mangano , p. nason , and g. ridolfi , in _ heavy flavors ii _ , edited by a. j. buras and m. lindner ( world scientific , singapore , 1998 ) , p. 609 .
using information on @xmath0-meson fragmentation functions from cern lep 1 and adopting the nonrelativistic qcd factorization formalism proposed by bodwin , braaten , and lepage , we predict the transverse - momentum distribution of @xmath1 mesons originating from the inclusive decays of @xmath2 hadrons produced in @xmath3 collisions at the fermilab tevatron . we determine the relevant colour - octet charmonium matrix elements from fits to cdf data on prompt charmonium hadroproduction and to cleo data on charmonium production from @xmath0-meson decay . our predictions are found to agree well with recent cdf and d0 data . pacs numbers : 13.60.-r , 13.85.ni , 13.87.fh , 14.40.lb 16.cm 22.cm @=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta
thermal , surface emission from neutron stars ( nss ) has the potential of providing invaluable information on the physical properties and evolution of nss ( equation of state at super - nuclear densities , cooling history , magnetic field , surface composition , different populations ; see , e.g. , yakovlev & pethick 2004 ; harding & lai 2006 ) . with x - ray telescope such as _ chandra _ and _ xmm - newton _ , the last decade has seen significant observational progress , revealing the surface magnetic field geometry of isolated pulsars with phase - resolved spectroscopy , and constraining the cooling physics from thermal emission of young nss in supernova remnants ( e.g. , kaspi et al . in addition , thermal emission from seven isolated , radio - quiet nss has been studied in detail , revealing absorption features in their spectra in many cases ( see , e.g. , van kerkwijk & kaplan 2007 ; kaplan 2008 ) . it has been recognized that in addition to imaging , timing and spectroscopy , x - ray polarimetry provides a new way to study many high - energy astrophysical sources , particularly magnetic nss . recent advances in detector technology suggest that polarimetry study of x - ray sources holds great promise in the future ( e.g. , costa et al . 2008 ; swank et al . 2008 ) . the surface emission from magnetized nss ( with @xmath0 g ) is highly polarized ( e.g. , gnedin & sunyaev 1974 ; meszaros et al . 1988 ; pavlov & zavlin 2000 ) for the following reason . in the magnetized plasma that characterizes ns atmospheres , x - ray photons propagate in two normal modes : the ordinary mode ( o - mode , or @xmath1-mode ) is mostly polarized parallel to the @xmath2-@xmath3 plane , while the extraordinary mode ( x - mode , or @xmath4-mode ) is mostly polarized perpendicular to the @xmath2-@xmath3 plane , where @xmath2 is the photon wave vector and @xmath3 is the external magnetic field ( e.g. , meszaros 1992 ) . this description of normal modes applies under typical conditions , when the photon energy @xmath5 is much less than the electron cyclotron energy @xmath6kev [ where @xmath7 , @xmath5 is not too close to the ion cyclotron energy @xmath8ev ( where @xmath9 and @xmath10 are the charge number and mass number of the ion ) , the plasma density is not too close to the vacuum resonance ( see below ) and @xmath11 ( the angle between @xmath2 and @xmath3 ) is not close to zero . under these conditions , the x - mode opacity ( due to scattering and absorption ) is greatly suppressed compared to the o - mode opacity , @xmath12 ( e.g. lodenquai et al . 1974 ; meszaros 1992 ; potekhin & chabrier 2003 ) . as a result , the x - mode photons escape from deeper , hotter layers of the ns atmosphere than the o - mode photons , and the emergent radiation is linearly polarized to a high degree ( e.g. , pavlov & zavlin 2000 ; ho & lai 2001,2003 ; van adelsberg & lai 2006 ) . measurements of x - ray polarization , particularly when phase - resolved and measured in different energy bands , could provide unique constraints on the ns magnetic field strength and geometry . it has long been predicted from quantum electrodynamics ( qed ) that in a strong magnetic field the vacuum becomes birefringent ( e.g. , schwinger 1951 ; adler 1971 ) . while this vacuum polarization effect makes the photon index of refraction deviate from unity only when @xmath13 , where @xmath14 g is the critical qed field strength , it can significantly affect the spectra of polarization signals from magnetic nss in more subtle way , at much lower field strengths ( see section 2 of lai & ho 2003a for a qualitative explanation ) . in particular , the combined effects of vacuum polarization and magnetized plasma gives rise to a `` vacuum resonance '' , at which the contributions from these two effects ( plasma and vacuum polarization ) to the dielectric tensor `` compensate '' each other ( gnedin et al . 1978 ; meszaros & ventura 1979 ; pavlov & gnedin 1984 ; lai & ho 2002 ) . a photon may convert from the high - opacity mode to the low - opacity one and vice verse when it crosses the vacuum resonance region in the inhomogeneous ns atmosphere . for @xmath15 g , this vacuum resonance phenomenon tends to soften the hard spectral tail due to the non - greyness of the atmospheric opacities and suppress the width of absorption lines , while for @xmath16 g , the spectrum is unaffected ( see lai & ho 2002,2003a ; ho & lai 2003 ; van adelsberg & lai 2006 ) . the qed - induced vacuum birefringence influences the x - ray polarization signals from magnetic nss in two different ways . ( i ) _ photon mode conversion in the ns atmosphere : _ since the mode conversion depends on photon energy and magnetic field strength , this vacuum resonance effect gives rise to a unique energy - dependent polarization signal in x - rays : for `` normal '' field strengths ( @xmath17 g ) , the plane of linear polarization at the photon energy @xmath18kev is perpendicular to that at @xmath19kev , while for `` superstrong '' field strengths ( @xmath20 g ) , the polarization planes at different energies coincide ( lai & ho 2003b ; van adelsberg & lai 2006 ) . ( ii ) _ polarization mode decoupling in the magnetosphere : _ the birefringence of the magnetized qed vacuum decouples the photon polarization modes , so that as a polarized photon leaves the ns surface and propagates through the magnetosphere , its polarization direction follows the direction of the magnetic field up to a large radius ( the so - called polarization limiting radius ) . the result is that although the magnetic field orientations over the ns surface may vary widely , the polarization directions of the photon originating from different surface regions tend to align , giving rise to large observed polarization signals ( heyl & shaviv 2002 ; van adelsberg & lai 2006 ) . in this paper we examine in detail the photon polarization evolution in the magnetized qed vacuum of a ns magnetosphere [ effect ( ii ) in the last paragraph ] . we are particularly interested in photon propagation across the _ quasi - tangential _ region ( qt region ) : as the photon travels through the magnetosphere , it may cross the region where its wave vector is aligned or nearly aligned with the magnetic field ( i.e. , @xmath11 is zero or small ) . in such a qt region , the two photon modes ( @xmath1 and @xmath4 modes ) become ( nearly ) identical , and mode coupling may occur , thereby affecting the polarization alignment . our previous analytical ( or semi - analytical ) treatment of the alignment effect ( lai & ho 2003b ; van adelsberg & lai 2006 ) focused on the region far away from the ns surface , thus did not not include the qt region ( which typically lies within a few stellar radii ) . the numerical ray integration presented in heyl & shaviv ( 2002 ; see also heyl et al . 2003 ) should in principle have included such qt region , but no systematic characterization of the qt propagation effect on the final photon polarization was attempted there . as we show in this paper , polarization evolution through the qt region is sufficiently subtle ( e.g. the effect varies on small length scales across the emission region ) that a careful examination of its effect is necessary . the purpose of this paper is to study the evolution of high - energy ( x - ray ) photon polarization in ns magnetospheres and to quantitatively assess the qt propagation effect . the remainder of our paper is organized as follows : section 2 summarizes the basic equations for studying photon polarization evolution in magnetized qed vacuum . in section 3 we examine the general behavior of the polarization evolution across a qt region in a generic magnetic geometry . in section 4 we present detailed calculations in the case of dipole magnetic field and consider emissions from both the polar cap and the other regions of the ns surface . we provide a simple prescription ( see section 4.3 ) for including the qt effect in the calculations of the observed polarization signals . in section 5 we discuss the implications of our results for the x - ray polarization signals from magnetic nss and the prospect of using x - ray polarimetry to probe strong - field qed . the magnetospheres of pulsars and magnetars consist of relativistic electron - positron pairs streaming along magnetic field lines . the lorentz factor @xmath21 of the streaming motion and the plasma density @xmath22 are uncertain . for the open field line region of radio pulsars , pair cascade simulations generally give @xmath23 and @xmath24 ( e.g. , daugherty & harding 1982 ; hibschman & arons 2001 ; medin & lai 2009 ) , while recent theoretical works suggest that the corona of magnetars consist of pair plasma with @xmath21 up to @xmath25 and @xmath26 ( where @xmath27 is the stellar radius ; thompson et al . 2002 ; beloborodov & thompson 2006 ) , where @xmath28 is the goldreich - julian density . in general , both the plasma and vacuum polarization affect the photon modes in the magnetosphere . for a given photon energy @xmath5 , @xmath21 and @xmath29 , the vacuum resonance occurs at the density ( wang & lai 2007 ) @xmath30 where @xmath31 , @xmath11 is @xmath2-@xmath3 angle , @xmath32 , @xmath33 , @xmath34 , @xmath35 is equal to unity for @xmath36 and is at most of order a few for @xmath37 g . for the typical photon energy of interest in this paper , the magnetosphere plasma density @xmath22 is much less than @xmath38 , and vacuum birefringence dominates over the plasma effect . to put it in another way , at given @xmath29 , @xmath21 and density @xmath22 ( or @xmath39 ) , we can define the vacuum resonance photon energy : @xmath40}^{1/2}\,{\rm kev}. \label{eq : nuv}\ ] ] where @xmath41 is the ns spin period in units of 1 second . throughout this paper , we shall be interested in photon energies @xmath42 , so that the wave modes are determined by the vacuum polarization effect . in this section we summarize the key equations for studying the polarization evolution of x - rays in ns magnetospheres . the dielectric tensor and the inverse permeability tensor of a magnetized qed vacuum take the form : @xmath43 where @xmath44 is the unit tensor and @xmath45 is the unit vector along @xmath3 . in the low frequency limit , @xmath46 , the general expressions for the vacuum polarization coefficients @xmath47 , @xmath48 , and @xmath49 are given in adler ( 1971 ) and heyl & hernquist ( 1997 ) . for @xmath50 g , they are given by @xmath51 with @xmath52 here @xmath53 is the fine structure constant . for @xmath54 , simple expressions for @xmath47 , @xmath48 , @xmath49 are given in ho & lai ( 2003 ) ( see also potekhin et al . 2004 for general fitting formulae ) . using the relations @xmath55 , @xmath56 and the maxwell equations , we obtain the equation for plane waves with @xmath57 @xmath58}\right\ } e_j=0 , \label{eq : nrefract}\ ] ] where @xmath59 is the refractive index and @xmath60 . the dielectric tensor are given by eq . ( [ eq : dt ] ) . in the coordinate system @xmath61 ( with @xmath2 along the @xmath62-axis and @xmath3 in the @xmath63-@xmath62 plane , such that @xmath64 ) , we can solve equation ( [ eq : nrefract ] ) to obtain the two eigenmodes : the @xmath1-mode ( or ordinary mode , polarized in the @xmath2-@xmath3 plane ) and @xmath4-modes ( or extraordinary mode , polarized perpendicular to the @xmath2-@xmath3 plane ) . the refractive indices and polarization states of these two modes are @xmath65 the difference between the two refractive indices is ( for @xmath66 ) @xmath67 a general polarized electromagnetic wave with frequency @xmath68 traveling in the @xmath62-direction can be written as a superposition of the two modes : @xmath69 note that both @xmath70 , @xmath71 and @xmath72 , @xmath73 depend on @xmath62 . substituting equation ( [ field ] ) into the wave equation @xmath74 we obtain the amplitude evolution equations ( see adelsberg & lai 2006 ) @xmath75 where @xmath76 . here @xmath77 is the azimuthal angle of @xmath3 in the fixed frame @xmath78 with the @xmath9-axis parallel to the line of sight @xmath79 ( note that the @xmath61-frame introduced in section 2.1 rotates around the @xmath9-axis since @xmath3 changes along the ray ) . in deriving equation ( [ eq : me ] ) , we have used the geometric optical approximation @xmath80 . the condition for the adiabatic evolution of wave modes is @xmath81 here @xmath82 is the adiabaticity parameter which changes along the photon ray . in the adiabatic limit , the photon modes are _ decoupled _ , and the photon will always stay in the initial mode although the mode itself changes along the ray following the direction of @xmath83 ( the projection of @xmath3 in the xy plane ) . substituting eq . ( [ eq : dn ] ) into eq . ( [ eq : gamma_ad ] ) , we have @xmath84 where @xmath85 is equal to unity for @xmath36 and is at most of order a few for @xmath37 g ( see fig . 1 of ho & lai 2003 ) we will use @xmath86 hereafter . the unit of @xmath87 is km@xmath88 and the factor @xmath89 is determined by magnetic field geometry along the ray . once the mode amplitudes @xmath70 and @xmath71 are known , we can obtain the stokes parameters in the fixed @xmath78-frame from ( see van adelsberg & lai 2006 ) @xmath90 alternatively , we can use equation ( [ eq : me ] ) to obtain the evolution equation for the stokes parameters ( see van adelsberg & lai 2006 ) : @xmath91 and @xmath92 . here we examine the general behavior of the polarization evolution when a photon crosses the quasi - tangential ( qt ) region . an exact tangential point is where the photon wave vector @xmath2 is aligned with @xmath3 , or @xmath93 . across the tangential point , the azimuth of the magnetic field , @xmath77 , changes by @xmath94 . for a general magnetic field geometry , not all photon rays have such an exact tangential point . however , there exits a _ quasi - tangential _ point where the @xmath2-@xmath3 angle @xmath11 reaches a local minimum . without loss of generality , the magnetic field around the qt point can be modeled as @xmath95 in the fixed @xmath78-frame with @xmath96 . here @xmath29 is the magnitude of @xmath3 , @xmath97 is the curvature radius of the projected magnetic field line in @xmath98-plane , @xmath99 measures the distance along the @xmath9-axis ( with the qt point at @xmath100 ) . the polar angle and azimuthal angle of @xmath3 are given by @xmath101 the minimum @xmath11 occurs at the qt point @xmath102 we also have @xmath103 thus the adiabaticity parameter along the ray is @xmath104 where @xmath105 . c + before and after the qt point ( where @xmath100 ) , when @xmath106 , the adiabaticity parameter @xmath82 increases rapidly with @xmath107 and the mode evolution is generally adiabatic ( @xmath108 ) . however at the qt point , @xmath82 reaches its minimum value : @xmath109 the value of @xmath110 determines the mode evolution characteristics across the qt region . figure [ fig : sketch ] shows the qualitative behaviors in two limiting cases : the adiabatic limit ( @xmath111 ) and the non - adiabatic limit ( @xmath112 ) . in the adiabatic case , the photon polarization direction follows the variation of the transverse magnetic field @xmath83 . since the final direction of @xmath83 is opposite to the initial direction , the final polarization direction is the same as the initial one . in the non - adiabatic case , @xmath77 changes rapidly , so that the polarization direction can not follow @xmath83 and remains constant across the qt region . thus , in both limiting cases , the polarization direction is unchanged when the photon traverses the qt point . c + c + c + obviously , a non - trivial change of photon polarization across the qt point occurs only in the intermediate case , @xmath113 . to obtain the quantitative behaviors of the polarization evolution for general values of @xmath110 , we integrate eq . ( [ eq : me ] ) numerically . figures [ fig : single_nonad ] [ fig : single_middle ] show three examples of single photon mode evolution in the qt region , corresponding to @xmath114 , @xmath111 and @xmath113 , respectively . for the @xmath114 ( non - adiabatic ) case ( fig . [ fig : single_nonad ] ) , the photon polarizations ( stokes parameters ) are constant throughout the region . for the @xmath111 ( adiabatic ) case ( fig . [ fig : single_ad ] ) , the photon polarizations ( stokes parameters ) change around the qt region , following the variation of @xmath3 , but the final polarizations are very close to the initial values . the most interesting case occurs for @xmath113 ( fig . [ fig : single_middle ] ) . in this intermediate regime , partial mode conversion takes place , so that after crossing the qt point , the photon becomes a mixture of two modes ( even when it is in a pure mode prior to qt crossing ) . the polarization state of the photon is therefore significantly changed by the qt effect . based on the above results , we can use @xmath115 to define the parameter regime for which propagation through the qt region gives rise to an appreciable change in the photon polarization , @xmath116 note that when @xmath117 , the qt effect is also negligible . this equation effectively maps out the domain where the qt propagation effect must be carefully considered . outside this domain , one can ignore the qt effect in determining the final observed polarization signals . to translate this effective parameter domain into the physical size of the emission region requires a knowledge of the global ns magnetic field structure . the smallness of @xmath118 for typical parameters ( e.g. , @xmath119 and @xmath120 ) indicates that the `` affected '' region is a small fraction of the ns surface . but as far as the observed polarization signals are concerned , it is more relevant to compare the size of the `` affected '' region to the size of the photon emission area . we consider the special case of dipole magnetic field in the next section . in this section we assume that the ns has a pure dipole field . we focus on x - ray emission from the polar cap region of the star ( sections 4.2 - 4.3 ) , but also consider more general emission regions on the ns surface ( section 4.4 ) . [ cols="^,^ " , ] to calculate the observed polarized radiation signals , we set up a fixed coordinate system @xmath121 with the @xmath122-axis along the line - of - sight ( pointing from the ns toward the observer ) and the @xmath123-axis in the plane spanned by the @xmath122-axis and @xmath124 ( the spin angular velocity vector ) ( see fig . [ fig : embeam ] ) . the angle between @xmath124 and @xmath2 is @xmath125 . the magnetic dipole @xmath126 rotates around @xmath124 , and @xmath127 is the angle between @xmath124 and @xmath126 . consider a photon emitted at time @xmath128 ( corresponding to the ns rotation phase @xmath129 ) from the position @xmath130 on the ns surface . at this emission time , the magnetic dipole moment is @xmath131 . as the photon propagates along the @xmath9-axis , its position vector changes as @xmath132 where @xmath133 is the photon displacement from the emission point . neglects the effect of gravitational light bending , which can be incorporated in a straightforward manner ( e.g. beloborodov 2002 ; van adelsberg & lai 2006 ) . this effect amounts to shifting @xmath134 in the direction perpendicular to the @xmath9-axis , and does not appreciably change our result . ] in the meantime , @xmath126 rotates around @xmath124 , and changes according to @xmath135},\ ] ] where the rotation phase @xmath136 is ( we set @xmath137 when @xmath126 lies in the @xmath138 plane ) @xmath139 with @xmath140 the radius of the light cylinder . the polar angles @xmath141 , @xmath142 of @xmath126 in the @xmath121 frame are given by @xmath143 ( similar expressions hold for the polar angles of @xmath131 , @xmath144 and @xmath145 , with @xmath136 replaced by @xmath146 . ) the changing magnetic field as `` seen '' by the photon is obtained from @xmath147 when discussing the polarization result of a given rotation phase ( at emission ) @xmath129 , it is convenient to introduce another fixed coordinate system @xmath78 ( see fig . [ fig : embeam ] ) , such that @xmath96 and @xmath148 in the @xmath98-plane with @xmath149 . in the @xmath78 frame , @xmath145 is given by @xmath150}.\ ] ] the magnetic field ( [ eq : b ] ) as `` seen '' by the photon is inclined at an angle @xmath11 with respect to the line - of - sight , and makes an azimuthal angle @xmath77 in the @xmath151-plane . the angle @xmath11 , @xmath77 can be obtained from equation [ eq : b ] via : @xmath152 the observed polarized radiation is the incoherent sum of photons from the emission region on the ns surface . for each emission point , @xmath153 , we integrate the mode evolution equation ( [ eq : me ] ) along the photon ray from @xmath134 to a large radius , beyond the _ polarization limiting radius _ , to determine the final polarization state of the photon . the polarization limiting radius , @xmath154 , is where the two photon modes start recoupling to each other , and is determined by the condition @xmath155 . at large distance ( @xmath156 ) , the magnetic field is simply @xmath157 , and @xmath154 is given by @xmath158 where @xmath159 is the polar magnetic field at the stellar surface in units of @xmath160 g ( see van adelsberg & lai 2006 for a more detailed expression ) . beyond @xmath154 , the photon polarization state is frozen . as mentioned in section 1 , the calculations presented in van adelsberg & lai ( 2006 ) and lai & ho ( 2003b ) did not consider the possibility that the photon polarization may change appreciably when crossing the qt point , which typically lies at a much smaller radius than @xmath154 . in general , the radiation emerging from the ns atmosphere at @xmath134 includes both the @xmath1-mode and the @xmath4-mode , with the intensities @xmath161 depending on the field strength , photon energy and emission angle ( see lai & ho 2003b and van adelsberg & lai 2006 ) . in the absence of the qt effect , the radiation at @xmath162 will consist of approximately the same @xmath163 and @xmath164 , with a small mixture of circular polarization generated around @xmath154 ( see van adelsberg & lai 2006 ) . this simple result should be modified if there is a significant polarization change when the photon crosses the qt region . since we are interested in understanding the qt effect , in the following we will assume that at the emission point the radiation is in the @xmath1-mode , with @xmath165 and @xmath166 . c + c + c + figure [ fig:2d ] gives an example of the 2-dimensional polarization map of the final stokes parameters of photons emitted from the ns polar cap region . this map is produced for a specific set of parameters : surface magnetic field @xmath167 , photon energy @xmath5 , and the angle between @xmath148 and @xmath2 , @xmath144 . [ for a rotating ns , the map also depends on the angles @xmath127 , @xmath125 , the rotation period @xmath168 and emission phase @xmath129 , but the dependence is mainly through @xmath144 see eq . ( [ eq : thetaphi_mu ] ) . see below . ] without the qt effect , the final polarization from the polar cap region will be @xmath169 . we see from fig . [ fig:2d ] that the qt effect gives rise to fine features ( with length scale much less than the polar cap size ) in the polarization map . in particular , there are two narrow bands ( which are parallel to the projected magnetic dipole axis ) in the emission region , where the linear polarization @xmath170 is significantly reduced because the original @xmath1 mode is converted across the qt region into a mixture of @xmath1-mode and @xmath4-mode . in these two narrow belts , non - zero pattens of stokes @xmath171 and @xmath172 are produced . however , these @xmath173 patterns are anti - symmetric with respect to the @xmath174 line ( i.e. , the @xmath2-@xmath148 plane ) , so that the observed values of @xmath171 and @xmath172 almost equal zero when photons from the whole emission region are added . in fig . [ fig:2d_3 t ] we show three sections of fig . [ fig:2d ] along the @xmath175-axis for a fixed value of @xmath176 [ @xmath177 or @xmath178 for panel ( a ) , @xmath179 or @xmath180 for panel ( b ) , and @xmath181 or @xmath182 for panel ( c ) ] . we also plot the adiabaticity parameter @xmath110 at the qt point for the photon rays from each emission point . the @xmath110 profile is useful for understanding the results for the stokes parameters . as the analysis in section [ sec : qte ] shows , photon propagation through the qt region changes the photon polarization significantly only when @xmath113 . from fig . [ fig:2d_3 t ] , we see that the photon stokes parameters are modified by the qt effect only in two narrow bands , confined in the region @xmath183 , with the boundaries defined approximately by @xmath184 ( the two vertical dotted lines ) . outside this region , @xmath185 , the mode evolution is adiabatic around the qt point , so that the photon polarization state follows the magnetic field direction and the stokes parameters are unchanged across the qt region . in the middle of the band ( i.e. , around @xmath174 ) , @xmath186 , mode evolution is non - adiabatic , and the photon polarization state also does nt change across the qt point ( see fig . [ fig : sketch ] ) . figure [ fig:1d_b ] depicts other examples of the 1d polarization map for different values of stellar magnetic field @xmath167 ( @xmath187 and @xmath188 g ) and spin period @xmath168 . as in fig . [ fig:2d_3 t ] , we see that the final photon polarizations are determined by the value of @xmath110 , the qt effective region is confined to @xmath183 defined by @xmath189 ( the region between the two vertical dotted lines ) . a careful examination of figs . [ fig:2d_3 t ] [ fig:1d_b ] shows that due to the ns rotation , the @xmath110 profile and stokes profiles shift in @xmath175 by the amount @xmath190 , where @xmath191 is the distance of the qt point from the emission point . since typically @xmath191 is less than a few ns radii , this shift @xmath192 is negligible . according to eq . ( [ eq : gamma_t ] ) , @xmath110 is proportional to @xmath193 , thus for smaller @xmath167 ( and smaller @xmath5 ) the effective width @xmath194 is larger , as seen in fig . [ fig:1d_b ] . in fig . [ fig : wd_b ] we show how @xmath194 changes with varying @xmath193 . our numerical result for @xmath195 can be fitted by @xmath196 here @xmath197 is a dimensionless function of @xmath144 [ which in turn depends on @xmath127 , @xmath125 and @xmath129 , see eq.([eq : thetaphi_mu ] ) ] : @xmath198 and varies from 1 to 1.7 for different values of @xmath144 ( see fig . [ fig : f_thetamui ] ) . the scaling relation in eq . ( [ eq : w ] ) is derived in the appendix . note that the polar cap width ( diameter ) is given by @xmath199 where @xmath200 . then we have @xmath201 equation ( [ eq : w / wcap ] ) implies that the size of the effective qt region ( where the qt effect changes the photon polarization ) can be comparable to the polar cap size for some parameters ( e.g. , low photon energy and low magnetic field strength ) . c + c + as discussed above , when linearly polarized radiation with @xmath202 traverses the qt region with @xmath189 , its stokes parameters will be changed ( so that @xmath170 will become less than unity , and @xmath173 will be nonzero ; see figs . [ fig:2d_3 t ] [ fig:1d_b ] ) . however , when adding up radiation from a finite emission region , we find @xmath203 and @xmath204 . thus the net effect of the qt propagation is to reduce the degree of the linear polarization of the photon . in general , if @xmath205 is the flux of linearly polarized radiation prior to passing the qt region , then after traversing the qt region , the linearly polarized radiation flux @xmath206 can be obtained by @xmath207 where @xmath208 is the width of the emission region . as seen in section [ sec : sim ] ( see figs . [ fig:2d_3 t ] [ fig:1d_b ] ) , the @xmath209 profiles for different parameters are similar , so that @xmath210 depends only on @xmath211 . our numerical result for @xmath210 as a function of @xmath211 is shown in fig . [ fig : fl_w ] . for a given set of parameters ( @xmath167 , @xmath5 , @xmath144 ) , we can use eq . ( [ eq : w ] ) and fig . [ fig : f_thetamui ] to obtain @xmath194 , and then use fig . [ fig : fl_w ] to read off the ratio @xmath210 . if @xmath212 , we find that @xmath210 is approximately given by @xmath213 note that for @xmath214 , the effective qt region is much smaller compared to the emission region so that @xmath215 ; for @xmath216 , the photon mode evolution across the qt region is non - adiabatic so that @xmath210 is also close to unity . thus in both @xmath217 and @xmath218 limits , @xmath219 , i.e. , the linear polarization is unchanged by the qt effect . the minimum value of @xmath210 occurs at @xmath220 . c + c + suppose the emission is coming from polar cap region with width @xmath221 , then according to eq . ( [ eq : w / wcap ] ) , @xmath222 is proportional to @xmath223 . thus , for given @xmath224 , @xmath225 and @xmath144 , we can translate fig . [ fig : fl_w ] into a spectrum for @xmath210 this spectrum is shown in fig . [ fig : spec ] . we see that at sufficiently low energies and high energies , @xmath215 , and the minimum reduction of the degree of linear polarization due to the qt effect occurs at @xmath226 , corresponding to the photon energy @xmath227 now we consider the ( energy - dependent ) polarization light curve produced by the polar cap emission of a rotating ns . as the ns rotates , the @xmath2-@xmath148 angle @xmath144 changes with the rotation phase @xmath129 [ see eq . ( [ eq : thetaphi_mu ] ) ] . according to eq . ( [ eq : w / wcap ] ) and figs . [ fig : f_thetamui ] , [ fig : fl_w ] , different @xmath144 will give different @xmath210 , which means @xmath210 will evolve with the rotation phase @xmath129 . moreover , @xmath228 and @xmath210 depend on the photon energy . figs . [ fig : profile1 ] and [ fig : profile2 ] present two examples of the phase evolution of @xmath228 and @xmath210 , for @xmath229 g and @xmath187 g , respectively . as noted before , @xmath210 reaches a minimum at @xmath230 ( see fig . [ fig : fl_w ] ) . in the case depicted in fig . [ fig : profile1 ] , @xmath231 for @xmath232kev and 0.5kev at all phases , thus @xmath210 is larger for lower photon energies ( since @xmath233 increases with @xmath5 ) . in fig . [ fig : profile2 ] , @xmath234 for all the three photon energies ( @xmath235kev ) , so that @xmath210 is larger for higher energies . c + c + to produce the observed polarized radiation fluxes , using the results present above , we will need input @xmath205 . in general , the emergent radiation from the ns atmosphere ( before passing through the qt region ) is linearly polarized , with @xmath205 dependent on @xmath5 and the rotation phase ( and @xmath167 @xmath127 , @xmath125 ) . in the @xmath78 frame , the radiation has polarized fluxes @xmath236 ( which can be either positive or negative ) , @xmath237 and @xmath238 . using @xmath205 from an atmosphere model and our result for @xmath210 , we can compute @xmath239 , the polarized flux after passing through the qt region . then the _ observed _ polarized radiation fluxes in the fixed @xmath121 frame are given by @xmath240 where @xmath241 is the azimuthal angle of the magnetic field , @xmath77 , evaluated at the polarization limiting radius @xmath242 . for @xmath243 , or for spin frequency @xmath244 hz , @xmath245 ( see lai & ho 2003b ; van adelsberg & lai 2006 ) , thus @xmath246 and @xmath247 . in the previous subsections we have focused on emission from the polar cap region of the ns . in reality , the `` hot spot '' on the ns surface may be larger . the qt effect also exists outside the polar cap region . as discussed above , the qt effective region is confined to @xmath189 . for a given dipole field geometry , we can calculate @xmath110 for photon rays emerging from different points on the stellar surface . [ fig:2d_theta_mu15 ] and [ fig:2d_theta_mu5 ] give two examples of the effective qt region as defined by @xmath189 , for the @xmath2-@xmath148 angle @xmath248 and @xmath249 , respectively . for each emission point , if we ignore the qt effect , the final photon polarization is dominated by the wave mode coupling effect and the polarization position angle ( pa ) is approximately determined by @xmath77 at polarization limiting radius @xmath154 ( again , assuming at emission the photon is in the @xmath1-mode ) . generally we have @xmath250 , so @xmath251 . therefore , the final polarization direction of the x - ray photon from any point of the star surface are always parallel to the @xmath2-@xmath126 plane . however , in the qt effective region , the final pa may be modified when the photon propagates through qt region ( see the lower panel of fig . [ fig:2d_theta_mu15 ] ) . note that the final polarization angles are always symmetric with respect to the @xmath2-@xmath126 plane . thus , as discussed before , when adding up radiation from a finite emission region , the net effect of qt propagation is to reduce the degree of linear polarization without changing the polarization angle . special caution must be taken when @xmath144 is small . we see from fig . [ fig:2d_theta_mu5 ] that for @xmath252 , the effective qt region may cover a significant part of the stellar surface ( in fig . [ fig:2d_theta_mu5 ] , the hatched region only shows the effective qt region that satisfies @xmath253 . ) this can be understood as follows . for @xmath254 the transverse parts of the dipole magnetic field are [ see eq . ( [ eq : b ] ) ] @xmath255 with @xmath256 ( neglecting light bending ) and @xmath257 , and @xmath258 are the @xmath259 components of the emission position @xmath134 . for @xmath260 , we have @xmath261 , @xmath262 , thus outside the qt region , the final polarization is aligned with the @xmath263 plane . however , for @xmath264 , we no longer can neglect @xmath265 relative to @xmath266 . thus for sufficiently small @xmath144 , polarization alignment will not be achieved for most emission points . figure [ fig:1d_theta_mu5_diffthetar ] shows the 1d profiles of the final @xmath170 , @xmath171 , @xmath172 and @xmath110 for a fixed @xmath267 and varying @xmath268 , all with @xmath252 . in the left and middle panels where @xmath269 and @xmath270 , the profiles are similar to the cases examined before ( see figs . [ fig:2d_3 t ] and [ fig:1d_b ] ) . for the @xmath271 case ( the right panels of fig . [ fig:1d_theta_mu5_diffthetar ] ) , @xmath110 is always less that 3 . the reason is that when @xmath267 is large , the qt point lies far away from the star @xmath272 ( and it is even possible that @xmath273 for sufficiently large @xmath267 ) . in this case ( small @xmath144 and emission from the region far away from the magnetic pole ) , the simple prescription we have presented in section 4.2 4.3 to account for the qt effect can not be used , and numerical ray integration from each emission point is necessary . c + + c + ccc + we have studied the evolution of photon polarization in a neutron star magnetosphere whose dielectric property is dominated by vacuum birefringence . we have focused on x - rays because of the potential of using x - ray polarimetry to constrain neutron star magnetic fields and to probe strong - field qed ( see heyl & shaviv 2002 ; heyl et al . 2003 ; lai & ho 2003b ) . if one neglects the qt propagation effect studied in this paper , then it is straightforward to obtain the obserevd polarized x - ray fluxes ( stokes parameters ) from the fluxes at the emission region , at least approximately , without integrating the polarization evolution equations in the magnetosphere ( lai & ho 2003b ; van adelsberg & lai 2006 ) . for a given ( small ) emission region of projected area @xmath274 , one need to know the intensities of the two photon modes at emission , @xmath164 and @xmath163 . in the case of thermal emission , these can be obtained directly from atmosphere / surface models . as the radiation propagates through the magnetosphere , the photon mode evolves adiabatically , following the variation of the magnetic field , until the polarization limiting radius @xmath154 , at which point the polarization is frozen . thus the polarized radiation flux beyond @xmath154 is @xmath275 , and @xmath276 is not exactly zero because of the neutron star rotation and because mode recoupling does not occur instantly at @xmath154 ; see van adelsberg & lai ( 2006 ) . ] , where @xmath277 is the distance of the source , @xmath205 and @xmath278 are defined in the coordinate system such that the stellar magnetic field at @xmath154 lies in the @xmath98 plane ( with the @xmath9-axis pointing towards the observer ) . since @xmath154 is much larger than the stellar radius , the magnetic fields as `` seen '' by different photon rays are aligned and are determined by the dipole component of the stellar field , one can simply add up contributions from different surface emission areas to @xmath205 to obtain the observed polarization fluxes . note that the description of the polarization evolution in the last paragraph is valid regardless of the possible complexity of the magnetic field near the stellar surface . this opens up the possibility of constraining the _ surface _ magnetic field of the neutron star using x - ray polarimetry . for example , the polarization light curve ( particularly the dependence on the rotation phase ) depends only on the dipole component of the magnetic field , while the intensity lightcurve of the same source depends on the surface magnetic field . on the other hand , the linear polarization spectrum ( i.e. , its dependence on the photon energy ) depends on the magnetic field at the emission region ( lai & ho 2003b ; van adelsberg 2006 ) ; thus it is possible that a ns with a weak dipole field ( @xmath279 g ) may exhibit x - ray polarization spectrum characteristic of a @xmath280 g ns ( see section 1 ) . the qt propagation effect studied in this paper complicates the above picture somewhat . when a photon passes through the qt region ( which typically lies within a few stellar radii ) , its polarization modes can be temporarily recoupled , and this can give rise to partial mode conversion . thus after passing through the qt point , the mode intensities change to @xmath281 and @xmath282 . the observed polarization flux is then @xmath283 . in the most general situations , to account for the qt propagation effect , it is necessary to integrate the polarization evolution equations ( see section 2.2 ) in order to obtain the observed radiation stokes parameters . however , we show in this paper that for generic near - surface magnetic fields , the effective region where the qt effect leads to significant polarization changes covers only a small area of the neutron star surface [ see section 3 , particularly eq . ( 3.22 ) ] . for a given emission model ( and the size of the emission region ) and magnetic field structure , one can use our result in section 3 to evaluate the importance of the qt effect . in the case of surface emission from around the polar cap region of a dipole magnetic field , we have quantified the effect of qt propagation in detail ( section 4 ) and provided a simple , easy - to - use prescription to account for the qt effect in determining the observed pollarizatin fluxes . our key results are presented in sections 4.2 4.3 , particularly fig . 11 and equations ( 4.32 ) ( together with fig . 10 ) and ( 4.35 ) ( 4.36 ) . as discussed in section 4.3 , the net effect of qt propagation is to reduce the degree of linear polarization , so that @xmath284 , with the reduction factor depending on the photon energy , magnetic field strength , geometric angles , rotation phase and the emission area . the largest reduction is about a factor of two , and occurs for a particular emission size ( @xmath220 ; see fig . 11 ) . obviously , for emission from a large area of the stellar surface , the qt effect is negligible . overall , the qt effect will have a small or at most modest effect on the observed x - ray polarization signals from magnetized neutron stars . thus the the description of x - ray polarizaton signals given in section 5.1 remains largely valid . in most situations ( the exceptions are discussed in section 4.4 ) , the qt effect can be accounted for in a straightforward manner using the result presented in our paper . this work has been supported in part by nasa grant nnx07ag81 g , nsf grants ast 0707628 , and by _ chandra _ grant tm6 - 7004x ( smithsonian astrophysical observatory ) . chen wang is also supported by the national natural science foundation ( nnsf ) of china ( 10833003 ) and the initialization fund for president award owner of chinese academy of sciences . longestkeymustbeshorterthanthis99 adler , s.l . 1971 , ann . phys . , 67 , 599 beloborodov , a. m. , 2002 , apj , 565 , 808 beloborodov , a. m. & thompson , c. , 2006 , 36th cospar scientific assembly . held 16 - 23 july 2006 . costa , e . , bellazzini , r. , & bregeon , j. et al . 2008 , in `` space telescopes and instrumentation 2008 : ultraviolet to gamma ray '' . edited by turner , martin j. l. ; flanagan , kathryn a. proceedings of the spie , volume 7011 . arxiv:0810:2700 daugherty , j. k. & harding , a. k. , 1982 , apj , 252 , 337 gnedin yu.n . and sunyaev r.a . , 1974 , a&a . 36 , 379 gnedin , yu.n . , pavlov , g.g . , & shibanov , yu.a . 1978 , sov . lett . , 4(3 ) , 117 harding , a.k . , & lai , d. 2006 , rept . phys . , 69 , 2631 heyl , j.s . & hernquist , l. 1997 , j. phys . a , 30 , 6485 heyl , j.s . , & shaviv , n.j . 2002 , phys . d66 , 023002 heyl , j.s . , shaviv , n.j . , & lloyd , d. 2003 , mnras , 342 , 134 hibschman , j. a. & arons , j. , 2001 , apj , 560 , 871 ho , w. c. & lai , d. 2001 , mnras , 327 , 1081 ho , w. c. & lai , d. 2003 , mnras , 338 , 233 kaplan , d.l . 2008 , arxiv:0801.1143 kaspi , v.m . , roberts , m. , & harding , a.k . 2006 , in `` compact stellar x - ray sources '' eds . w. lewin & m. van der klis ( cambridge univ . press ) , p279 . lai , d. & ho , w. c. g. 2002 , apj , 566 , 373 lai , d. & ho , w. c. g. 2003a , apj , 588 , 962 lai , d. & ho , w. c. 2003b , phrvl , 91 , 1101 lodenquai , j. , canuto , v. , ruderman , m. & tsuruta , s. 1974 , apj , 190 , 141 medin , z. & lai , d. , 2009 , in preparation mszros , p. , high - energy radiation from magnetized neutron stars , univ . chicago press , chicago , 1992 mszros , p. , et al . 1988 , apj , 324 , 1056 mszros , p. & ventura , j. 1979 , phys . d 19 , 3565 pavlov , g.g . & gnedin , yu.n . 1984 , sov . e : astrophys . space phys . 3 , 197 pavlov , g.g . & zavlin , v.e . 2000 , apj , 529 , 1011 potekhin a.y . & chabrier g. , 2003 , apj , 585 , 955 potekhin , a. y. , lai , d. , chabrier , g. , & ho , w.c.g . , 2004 , apj , 612 , 1034 swank , j. , kallman , t. , & jahoda , k. 2008 , presentation at the 37th cospar scientific assembly , p.3102 schwinger , j. 1951 , phys . , 82 , 664 thompson , c. , lyutikov , m. , & kulkarni , s. r. 2002 , apj , 574 , 332 van adelsberg , m. & lai , d. 2006 , mnras , 373 , 1495 van kerkwijk , m.h . & kaplan , d.l . 2007 , ap&ss , 308 , 191 wang , c. & lai , d. 2007 , mnras , 377 , 1095 yakovlev , d.g . , & pethick , c.j . 2004 , araa , 42 , 169 ; consider the emission from the polar cap region of a non - rotating ns . in the @xmath78-frame ( see fig . [ fig : embeam ] , @xmath285 , @xmath148 in @xmath98-plane ) , the initial magnetic momentum is @xmath286 and the photon position is @xmath287 where @xmath99 is the displacement from the emission point , and @xmath288 . we want to evaluate @xmath110 at the qt point @xmath289 as a function of @xmath175 . the magnetic field is given by eq . ( [ eq : b ] ) , and has components : @xmath290 ( here we set @xmath291 ) . for a given @xmath175 , the qt point ( @xmath292 ) is defined by @xmath293 we can expand @xmath266 and @xmath265 near @xmath294 using taylor expansion : @xmath295 with @xmath296 . note that @xmath297 , @xmath298 , @xmath299 , @xmath300 all depend on @xmath301 . substitute eq . ( [ eq : b_xy_te ] ) into eq . ( [ eq : qtp ] ) , we have @xmath302 from eq . ( [ eq : bxt ] ) we see that @xmath303 thus @xmath304 where @xmath305 , @xmath306 , @xmath307 are constants . . the azimuthal angle of @xmath3 is determined by @xmath308 . therefore @xmath87 at @xmath309 is @xmath310 we also have @xmath311 substitute eq . ( [ eq : dphi_b ] ) and ( [ eq : sintheta_b ] ) into eq . ( [ eq : gamma_ad2 ] ) , the adiabaticity parameter at qt point is then given by @xmath312 the qt effective width @xmath194 , defined by @xmath184 , is then @xmath313 .
x - ray photons emitted from the surface or atmosphere of a magnetized neutron star are highly polarized . however , the observed polarization may be modified due to photon propagation through the star s magnetosphere . for photon frequencies much larger than the typical radio frequency , vacuum birefringence due to strong - field quantum electrodynamics dominates over the plasma effect . we study the evolution of photon polarization in the magnetized qed vacuum of a neutron star magnetosphere , paying particular attention to the propagation effect across the quasi - tangential ( qt ) point , where the photon momentum is nearly aligned with the magnetic field . in agreement with previous studies , we find that in most regions of the magnetosphere , the photon polarization modes are decoupled due to vacuum birefringence , and therefore a large net linear polarization can be expected when the radiation escapes the magnetosphere . however , we show that the x - ray polarization may change significantly when the photon passes through the qt region . when averaging over a finite emission area , the net effect of qt propagation is to reduce the degree of linear polarization ; the reduction factor depends on the photon energy , magnetic field strength , geometry , rotation phase and the emission area , and can be more than a factor of two . we derive the general conditions under which the qt propagation effect is important , and provide an easy - to - use prescription to account for the qt effect for most practical calculations of x - ray polarization signals from magnetic neutron stars . for a neutron star with a dipole magnetic field , the qt effect can be important for emission from the polar cap for certain magnetic field and energy ranges , and is negligible for emission from the entire stellar surface . = 0.0truein [ firstpage ] plasmas polarization waves star : magnetic fields pulsars : general x - rays : stars
weakly interacting massive particles ( wimps ) are currently favoured as candidates for dark matter , which makes up about 20@xmath0 percent of the total matter - energy content of the universe . although the lightest supersymmetric particle , which is the lightest neutralino in many susy models , would make a very good wimp if susy with r - parity conservation is realised in nature , most extensions of the standard model of particle physics contain wimps . in fact the more general requirement is that some new conserved quantum number makes the lightest of the new particles stable . the international linear collider ( ilc ) offers the possibility to look for wimps in a rather model - independent way , which first has been pointed out in @xcite . here , the expected sensitivity to such a wimp signal , the achievable mass resolution and the influence of beam polarisation are studied using a full detector simulation . assuming that the cosmic relic density of wimps is determined by pair annihilation of wimps , and that an unknown branching fraction @xmath1 of these annihilations proceeds into an @xmath2 pair(@xmath3 ) , crossing relations can be used to derive an expected cross - section for the reverse process , i.e. @xmath4 . this cross - section contains as free parameters : * the @xmath2 branching fraction @xmath1 * the mass of the wimp @xmath5 * the spin of the wimp @xmath6 * the angular momentum of the annihilation s dominant partial wave @xmath7 . in order that this process be observable in a collider detector , where the wimps themselves leave no signature , an additional photon from initial state radiation is required : @xmath8 . the main standard model background process for this reaction is neutrino pair production , again with an isr photon : @xmath9 . at energies significantly above the @xmath10 pole , this reaction is dominantly mediated by @xmath11-channel @xmath12-exchange and can thus be reduced significantly by choosing the appropriate polarisations for the electron and position beams . therefore we consider three possible scenarios for the helicity structure of the wimp coupling to electrons : * the same as the sm charged current weak interaction , i.e. only @xmath13 is nonzero * parity and helicity conserving , i.e. @xmath14 * opposite to sm charged current weak interaction , i.e. only @xmath15 is nonzero . especially in the last case a significant enhancement of the signal over background ratio is expected . the neutrino background has been generated with nunugpv @xcite , which is a dedicated generator for neutrino pair production with up to three photons @xmath16 . @xmath17 events with at least one photon with an energy @xmath18 8 gev @xmath19 250 gev and a polar angle @xmath20 15@xmath21 were generated at a center - of - mass energy of @xmath22 , corresponding to an integrated luminosity of 500 fb@xmath23 . these events serve not only to describe the irreducible standard model background , but are also reweighted w.r.t . energy and polar angle of the photon according to the wimp cross - section . the benefit of the is method is that in this way a signal sample can be obtained for all mass and spin hypotheses to be tested without applying the detector simulation again , which reduces the processing time tremendously . this sample has been subjected to the full ldc detector simulation , using the detector model ldc01sc with a 4 t magnetic field and mokka 6.1 @xcite . the events were then reconstructed with marlinreco @xcite , using the wolf @xcite algorithm for particle flow and a simple selection demanding @xmath24 10 gev and 20@xmath25 . for the energy resolution studies described in the following , also an angular match to the generated photon was required . the influence of the detector s energy resolution has been studied at two levels : besides the full reconstruction , the so called cheated reconstruction makes use of the simulation information for associating calorimeter clusters with particles . this way the pure detector resolution and calibration can be disentangled from confusion effects , noise and so on . figure [ fig : egamma ] shows the generated ( left ) as well as the cheated and fully reconstructed ( right ) energy spectra of the most energetic photon candidate of the events . while the peak at @xmath26 gev from radiative returns to the @xmath10 resonance is extremely clear at the generator level , it his heavily smeared after reconstruction , even for the cheated case . the full reconstruction yields even less photons at high energies . the reason for this effect can be seen in the left plot of figure [ fig : ncluseres ] , which shows the mean number of photon candiates per generated photon as a function of the generated photon s energy . it is clearly visible that at high photon energies the wolf algorithm tends to split clusters stemming from one photon into several photon candidates . therefore we apply a merging procedure to neighboring photon candidates , after which the fully reconstructed photon energy distribution is practicaly identical to the cheated spectrum shown in the right hand plot of figure [ fig : egamma ] . the energy resolution obtained after the recombination procedure is shown in right plot of figure [ fig : ncluseres ] as a function of @xmath27 . with being roughly constant at about @xmath28 , it seems to be significantly worse than the @xmath29 aimed for at the ilc . this effect has been meanwhile tracked down to an imperfect calibration , which means that the results presented in the next section are expected to improve further once a better calibration procedure is applied . .,title="fig : " ] .,title="fig : " ] currently , the following scenarios have been investigated : * wimp spin : p - wave annihilation ( j=1 ) for @xmath30 and @xmath31 * wimp couplings : @xmath32 , @xmath33 and @xmath34 * polarisation : unpolarised beams , @xmath35 polarisation only ( @xmath36 ) and additional @xmath37 polarisation ( @xmath36 and @xmath38 ) . for each combination of these parameters , the reach of the ilc with an integrated luminosity of 500 fb@xmath23 at @xmath39 gev for a 3@xmath40 observation of wimps has been determined as a function of the wimp mass . due to the high irreducible background from standard model neutrino production , the sensitivity has been obtained statistically by using fractional event counting @xcite as implemented in the root class tlimit . observation reach of the ilc for a spin-1 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] observation reach of the ilc for a spin-1 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] observation reach of the ilc for a spin-1 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] observation reach of the ilc for a spin-@xmath44 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] observation reach of the ilc for a spin-@xmath44 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] observation reach of the ilc for a spin-@xmath44 wimp in terms of wimp mass and @xmath1 for three different assumptions on the chirality of the electron - wimp coupling , see text . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . regions above the curves are accessible.,title="fig : " ] figure [ fig : case1 ] shows the expected ilc sensitivity for spin-1 wimps in terms of the minimal observable branching fraction to electrons @xmath1 as a function of the wimp mass . the leftmost plot shows the case where the wimps couple only to lefthanded electrons and righthanded positrons ( @xmath13 ) , the middle plot shows the parity and helicity conserving case ( @xmath14 , while the right plot is dedicated to the case that the wimps couple to righthanded electrons and lefthanded positrons ( @xmath45 ) . the regions above the curves are accessible , where the full line gives the result for unpolarised beams , the dotted line for @xmath36 and the dashed line for @xmath36 and @xmath38 . in the latter two coupling scenarios polarised beams increase the reach significantly , especially the additional positron polarisation increases the accessible range in @xmath1 by about a factor of 2 . figure [ fig : case2 ] shows the same for a spin-@xmath44 wimp . here the sensitivity is somewhat worse , but again beam polarisation extends the observable part of the parameter space significantly . if wimps are observed at the ilc , their mass can be determined from the recoil mass distribution of the photons : @xmath46 figure [ fig : mass ] shows an example for the recoil mass distribution for a 150 gev spin-1 wimp with both beams polarised . the wimp signal shown in dark grey kicks in at @xmath47 316 gev . r0.5 from this distribution , the wimp mass can be reconstructed for example by a template method . for this procedure , only 200 fb@xmath23 of the available mc sample have been analysed as dataset , the rest is used for the templates . figure [ fig : mcase1 ] shows the obtained @xmath48 as function of the reconstructed wimp mass for a 150 gev spin-1 wimp for @xmath49 . the left plot shows the helicity and parity conserving case , the right plot the case that the wimps couple to righthanded electrons and lefthanded positrons ( @xmath45 ) . again the full line gives the result for unpolarised beams , the dotted line for @xmath36 and the dashed line for @xmath36 and @xmath38 . without any beam polarisation , the mass resolution is about 4 gev , which is reduced to about 1.2 gev by switching on the electron polarisation . positron polarisation improves the resolution by another factor 2 to about 0.6 gev . from mass determination by a template method for a spin-1 wimp with @xmath5=150 gev . left : parity and helicity conserving couplings , right:@xmath50 . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . , title="fig : " ] from mass determination by a template method for a spin-1 wimp with @xmath5=150 gev . left : parity and helicity conserving couplings , right:@xmath50 . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43 . , title="fig : " ] figure [ fig : mcase2 ] shows the analoguos results for a 180 gev spin-@xmath44 wimp . as for the observation reach , the situation is slightly worse than in the spin-1 case , but again the use of beam polarisation leads to a significant gain in resolution . from mass determination by a template method for a spin-@xmath44 wimp with @xmath5=150 gev . left : parity and helicity conserving couplings , right:@xmath50 . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43.,title="fig : " ] from mass determination by a template method for a spin-@xmath44 wimp with @xmath5=150 gev . left : parity and helicity conserving couplings , right:@xmath50 . full line : @xmath41 , dotted line : @xmath42 , dashed line : @xmath43.,title="fig : " ] the study of model - independent wimp production at the ilc presented here is one of the first examples of analyses performed using the full simulation of the ldc detector . a reconstruction for these fully simulated events exists and usable for analysis , but needs further improvements in parallel to the optimization of the detector concept . already with the current level of sophistication this study shows that there is a good chance to detect wimps in this model - independent way at the ilc and to measure their mass with a precision of about 1 gev . both the range in phase space as well as the mass resolution improve significantly when polarised beams are assumed . typically the use of 80% electron polarisation gives improvements of a factor of two over unpolarised beams , whereas an additional positron polarisation of 60% yields another factor of two . the results presented here will be improved in the near future by applying a more appropriate detector calibration and by using other particle flow algorithms and photon finders . furthermore reducible backgrounds as well as beamstrahlung have to be included in the study , before finally different detector concepts can be compared . the authors acknowledge the support by dfg grant li 1560/1 - 1 . a. birkedal , k. matchev and m. perelstein , phys . rev . d * 70 * ( 2004 ) 077701 [ arxiv : hep - ph/0403004 ] . g. montagna , o. nicrosini and f. piccinini , comput . commun . * 98 * , 206 ( 1996 ) . ` http://polywww.in2p3.fr/activites/physique/geant4/tesla/www/mokka/mokka.html ` ` http://ilcsoft.desy.de/marlin `
we investigate the possibility to detect wimps at the ilc in a model - independent way using events with single photons . the study is done with a full detector simulation of the ldc detector and marlinreco . it turns out that wimps are observable this way at the ilc if their coupling to electrons is not too small . beam polarisation can increase the accessible phase space significantly .
strongly correlated electrons have been in recent years intensively investigated theoretically , mainly in connection with the open question of the mechanism of superconductivity as well as the unusual metal properties in cuprates . here we will discuss only the anomalous normal state properties of cuprates at @xmath9 . these are generally attributed to correlation effects , nevertheless a unifying theory still remains a challenge . in this article we will address several questions related to the theoretical understanding of the normal - state properties of cuprates : \a ) can strong correlations alone account for the unusual normal - state properties , as manifested by the charge response : optical conductivity @xmath2 and d.c . resistivity @xmath10 , and the spin response : dynamical spin susceptibility @xmath3 , related to the nmr relaxation @xmath11 , etc . ? \b ) what is the origin of the unconventional universality within the ` normal ' metallic state of cuprates , in particular in the ` optimum - doping ' regime , obtained by doping the reference antiferromagnetic ( afm ) insulator with the mobile hole - type charge carriers ? \c ) how many relevant energy scales exist in the doped afm ? \d ) are the simplest prototype models , such as the @xmath0 model ( and analogous hubbard model ) , enough to reproduce qualitatively and quantitatively experimental results in cuprates ? \f ) can we learn something by studying small correlated systems at @xmath1 ? from the recent experiments testing the normal - state properties such as the d.c . resistivity @xcite , optical conductivity @xcite , neutron scattering @xcite , and the nmr and nqr relaxation @xcite , a unifying picture seems to emerge @xcite . these properties mainly depend on the extent of doping , and the regimes have been classified into the underdoped , the ` optimal ' ( intermediate - doping ) , and the overdoped , respectively . whereas the ` optimal ' doping refers in a narrow sense usually to samples with the highest @xmath12 , the same broader region of doping , with effective hole concentration approx . @xmath13 @xcite , also exhibits most universal properties . down to the lowest @xmath14 resistivity follows the remarkable @xmath15 law , the uniform susceptibility @xmath16 is varying monotonously with @xmath4 , etc . on the other hand , the underdoped materials ( effective @xmath17 ) are characterized by the onset of the ` pseudogap ' at @xmath18 , which shows up in the kinks in @xmath19 , @xmath8 etc . we investigate in this review mainly the intermediate - doping ( optimal ) regime . we discuss finite-@xmath4 ( normal - state ) properties of the doped afm within the @xmath0 model @xcite @xmath20 where @xmath21 are projected fermionic operators , prohibiting double occupancy of sites . the ground state of the @xmath0 model and related properties have been intensively studied both by analytical @xcite and numerical methods @xcite , nevertheless some basic questions , e.g. concerning possible pairing in the ground state and the origin of anomalous metallic properties at finite @xmath1 , remain unsolved . normal metals are expected to follow the concept of the * landau fermi liquid*. the excitations in this case should be quasiparticles with well defined @xmath22 and energy @xmath23 near the fermi energy , with the damping @xmath24 . consequences are well known , e.g. for the dynamical susceptibility @xmath25 leading to the korringa law @xmath26 . also , the drude form is expected for @xmath2 with @xmath27 etc . the normal - state properties of cuprates , however , do not follow the landau scenario . experiments have been mainly described within two alternative concepts . the * nearly antiferromagnetic fermi liquid * has been introduced by moriya and millis et al . @xcite to describe the spin fluctuations in the doped afm , mainly in connection with the anomalous nmr relaxation phenomena in cuprates . the low-@xmath6 spin response is modeled with the overdamped spin fluctuations peaked at the afm wavevector @xmath28 , @xmath29 ^ 2 } .\label{eq32}\ ] ] afm correlation length @xmath30 is assumed to be critical in the doped regime , in particular with @xmath31 in the @xmath32 dynamical rpa - type theory @xcite , while the postulated mapping to the quantum critical regime of the nonlinear sigma model would imply @xmath33 @xcite . alternative scenario has been summarized within the * marginal fermi liquid * ( mfl ) hypothesis @xcite . here , the spin and charge susceptibility have been postulated in the anomalous form @xmath34 leading to the marginality of the quasiparticle self energy @xmath35 consistent with eq.([eq34 ] ) , the low - frequency @xmath2 has been written in the generalized drude form introducing an anomalous effective relaxation rate @xmath36 . @xmath1 properties of the @xmath0 model have been so far calculated only by the high - temperature series expansion @xcite and via full or partial diagonalization of quite small systems . recently the present authors introduced a novel numerical method , based on the lanczos diagonalization method combined with random sampling @xcite . the latter method has the advantage that it allows the study of dynamical and static response functions within the most challenging regime @xmath37 , where one could hope for a universal behavior , as observed in the experiments on cuprates . it should be noted that analogous hubbard model seems more feasible for @xmath1 studies , due to the possible application of the quantum monte carlo methods @xcite . nevertheless , in the doped case the well known minus sign problem so far mainly prevented calculations within the regime @xmath38 @xcite . let us present briefly our method for @xmath1 calculations , showing in more detail the evaluation of the statical expectation value of an operator @xmath39 , @xmath40 where @xmath41 ( we use @xmath42 furtheron ) and the sum runs over the chosen complete basis set of orthonormal wavefunctions @xmath43 , spanning the hamiltonian @xmath44 . if we could perform the full diagonalization of the problem and find all eigenstates @xmath45 and corresponding energies @xmath46 , we would express the result in a usual way , @xmath47 we choose an alternative approach . with each basis function @xmath48 we start a lanczos procedure @xmath49 , generating an orthonormal set of functions @xmath50 , @xmath51 for the chosen number of lanczos steps @xmath52 we then diagonalize the tridiagonal matrix of coefficients @xmath53 to find the energies @xmath54 and the corresponding eigenfunctions @xmath55 . within the restricted basis of these functions one can write the approximation for eq.([eq52 ] ) as @xmath56 it is evident that the average evaluated via eq.([eq54 ] ) is equivalent to eq.([eq52 ] ) for the sampling over the full basis set @xmath57 and @xmath58 . our claim is that very accurate results can be obtained via eq.([eq54 ] ) even for a severly reduced number of lanczos steps @xmath59 and for a partial random sampling of the basis states @xmath60 , instead of the full sampling . let us first investigate the method for the full sampling @xmath57 . we note that the lanczos procedure eq.([eq53 ] ) represents an iterative action of the operator @xmath44 on the initial function @xmath48 . performing the ( high temperature ) expansion in @xmath61 of the numerator and of the denominator of eq.([eq54 ] ) , it is easy to prove that both power series are correct up to the order @xmath62 . on the other hand , @xmath63 evaluated via eq.([eq54 ] ) is accurate also for @xmath64 ( provided @xmath65 ) , since it is a standard experience that the lanczos procedure converges quite rapidly , typically in @xmath66 steps , from an arbitrary initial function @xmath48 ( not orthogonal to the ground state ) to the ground state energy @xmath67 and the corresponding wavefunction @xmath68 . hence , at full sampling our lanczos - based method represents an optimum interpolation scheme between the high temperature @xmath69 and the ground state @xmath64 result . the second important ingredient is the reduction of the sampling to the partial random one with @xmath60 . this step can not be justified rigorously , but seems plausible in analogy with the statistical monte carlo methods . the most severe test is expected to be the result at @xmath64 , which can be anyhow tested via an alternative direct evaluation with the known @xmath68 . the general experience is that good convergence is achieved for @xmath60 . even more appealing is the application of the method to the calculation of dynamical quantities . we choose as an example the optical conductivity , expressed within the linear response theory as @xmath70 in order to get the result at arbitrary @xmath61 a double sum over all eigenstates @xmath71 is required . instead , we use the approach described above for @xmath63 generalized to dynamical @xmath72 , @xmath73 here , @xmath74 and @xmath75 are generated by the lanczos procedure analogous to eq.([eq53 ] ) via the orthonormal basis set @xmath76 , but with the initial condition @xmath77 while for @xmath78 the expression ( [ eq58 ] ) is equivalent to the exact one , the arguments for using @xmath60 and @xmath59 are analogous to those described above for the static case . e.g. , for the full sampling @xmath57 eq.([eq58 ] ) yields correct series in @xmath79 and @xmath80 , respectively , up to the @xmath81 order . accordingly , @xmath2 has correct frequency moments @xmath82 , for @xmath83 etc . again for @xmath65 the method yields meaningful result also for @xmath64 , which can be tested via the usual lanczos method for the @xmath64 dynamical quantities @xcite . it is important to stress that computational requirements of the method for a system of the given size are comparable to those for the ground state evalutions via the lanczos method , whereby the cpu time clearly increases with the number of random samples @xmath84 . the cpu time and memory requirements are also enhanced due to the necessary reorthogonalization of lanczos functions and due to the evaluation of matrix elements in eq.([eq58 ] ) . let us finally comment on the finite size effects , which due to the restricted system sizes still remain the main obstacle to achieve reliable results for low @xmath4 . it is a plausible observation that the method yields macroscopic - like results for @xmath85 , where @xmath86 is roughly determined by the average level separation in the low - energy many - body spectrum , being dependent on the system size . for @xmath87 the finite - size effects become pronounced , e.g. dynamical quantities in general reveal ( spurious ) peak structures etc . it is however very important to note that within the @xmath0 model , the @xmath86 is smallest within the intermediate ( optimum ) doping regime @xmath88 , where we typically reach @xmath89 for systems @xmath90 . the @xmath86 is larger both within the underdoped and the overdoped region ( for the fixed system size ) . let us consider the dynamical spin response , as given by the susceptibility @xmath91 and the corresponding dynamical spin correlation function @xmath92 @xmath93 it has been shown by the present authors @xcite that @xmath94 shows the coexistence of the high - frequency ( @xmath95 ) free - fermion - like contribution and the low-@xmath6 spin - fluctuation contribution . at the same time , @xmath96 reveals a pronounced @xmath4 dependence , consistent with the mfl form , and ( even quantitatively ) with the nmr relaxation in cuprates . it appears quite helpful to concentrate on the local spin correlation function @xmath97 @xcite and its symmetric part @xmath98 @xmath99 it should be pointed out that @xmath97 and the related susceptibility @xmath100 are directly measured ( in cuprates ) by neutron scattering @xcite . the nmr relaxation as well yields the information on @xmath101 @xcite , provided that the afm spin fluctuations @xmath102 are dominant . an important restriction for @xmath103 is the sum rule @xmath104 where @xmath105 is the hole concentration . we perform the evaluation of @xmath103 via eq.([eq3 ] ) by calculating @xmath106 , using the finite-@xmath4 diagonalization method for small systems , in this case for the @xmath0 model on the square lattice with @xmath107 sites . we fix @xmath108 to remain in the regime of cuprates @xcite . we stress again that the method yields macroscopic - like results for @xmath109 . in fig . 1 we display the @xmath103 for @xmath110 and several @xmath4 in the range @xmath111 @xcite it is immediately evident that @xmath103 at ` optimal ' doping @xmath112 is essentially @xmath4-independent in a wide @xmath4-range , although one crosses the exchange - energy scale @xmath113 . for the underdoped case @xmath114 the behavior is analogous for higher @xmath115 , consistent with the quantum critical regime within the afm @xcite . deviations at lower @xmath116 ( where the renormalized classical regime @xcite is expected in the afm ) could be an indication for the onset of a ` pseudogap ' . to follow the doping dependence we present in fig . 2 the variation with @xmath117 at fixed @xmath118 , plotting the integrated intensity @xmath119 again , we notice that for the chosen @xmath4 the results are most reliable at the intermediate doping . the most striking message is that the initial slope of @xmath120 and consequently @xmath101 are nearly doping independent for @xmath121 . this is well consistent with the nmr ( nqr ) relaxation rates @xmath122 as measured in @xmath123 in the range @xmath124 @xcite . only for the overdoped systems with @xmath125 the low - frequency behavior changes qualitatively , where the latter part is strongly suppressed as expected in ( more ) normal fermi liquids . @xmath126 is doping dependent even for @xmath127 , consistent with the @xmath117-dependence of the sum rule , eq.([eq4 ] ) . in addition , at the intermediate doping @xmath103 decreases smoothly ( see fig . 1 ) up to @xmath128 , this being the consequence of a free - fermion - like component @xcite . on the other hand , in the underdoped regime the dynamics is restricted to @xmath129 . how can one explain the universality of @xmath103 at intermediate @xmath117 ? first , we note that up to @xmath130 the dominant scale of spin fluctuations remains related to @xmath131 . from the explicit expression in terms of the eigenstates of the system @xmath132 one would conjecture ( quite generally ) a @xmath4-independence of response for @xmath133 . although such plausibility arguments have been used previously , their validity clearly depends on the character and on the density of low - lying many - body states . to explain also the @xmath4-independence of @xmath134 , we only need to recall the sum rule eq.([eq4 ] ) and to assume that there is no characteristic scale @xmath135 , which could introduce an additional low-@xmath6 structure in @xmath98 . a natural scale for the afm is the ( gap ) frequency @xmath136 , where @xmath137 is the spin wave velocity and @xmath138 the afm correlation length . here originates the essential difference between the undoped and the ` optimally ' doped afm . while for an afm in the renormalized classical regime @xmath138 is exponentially large for @xmath139 , and consequently @xmath140 @xcite , in the doped case @xmath141 is determined predominantly by @xmath117 , so @xmath138 is rather @xmath4-independent for @xmath142 , excluding @xmath143 . in other words , the observed anomalous low-@xmath6 behavior within the @xmath0 model does not originate in a strong @xmath4-dependence of @xmath138 , as conjectured by the nearly afm fermi liquid , eq.([eq32 ] ) . we conclude the discussion of spin dynamics by the consequences of the universality of @xmath103 . the local susceptibility is given by @xmath144 since neutron scattering probes only @xmath145 , one can simplify eq.([eq7 ] ) further by @xmath146 . such form has been recently used to describe experiments @xcite . it is also qualitatively consistent ( but as well more restricted in the form ) with the mfl ansatz , eq.([eq33 ] ) @xcite . let us investigate in an analogous way the dynamical conductivity @xmath2 , eq.([eq56 ] ) . results for this quantity by the present authors @xcite , obtained by the finite-@xmath4 method , can be summarized as follows : a ) in the intermediate - doping regime @xmath2 shows a non - drude fall - off , consistent with the @xmath6-dependent relaxation rate within the mfl concept eq.([eq34 ] ) @xcite , deduced from experiments in cuprates @xcite , and b ) qualitative as well as quantitative results for @xmath2 and @xmath19 agree well with the experimental ones . unlike @xmath103 , @xmath5 does not obey a @xmath4-independent sum rule . nevertheless , motivated by the universal spin dynamics we reexamine in an analogous manner our results on @xmath2 @xcite . in fig . 3 we present the corresponding integrated spectra @xmath147 at fixed doping in the ` optimal ' regime for various @xmath148 @xcite . we establish that for @xmath149 spectra @xmath150 are essentially independent of @xmath4 , at least for the available @xmath85 . at the same time the slope of @xmath151 is nearly constant , i.e. @xmath152 in a wide @xmath6-range , @xmath153 being weakly @xmath131-dependent ( tested for @xmath154 ) . we note that such @xmath155 implies a nonanalytic behavior of @xmath156 , starting with a finite slope at @xmath157 . moreover , with @xmath158 we can claim a simple universal form for @xmath159 @xmath160 at first sight , it appears rather surprising that this form can be well fitted ( for @xmath161 ) with a drude - type form with an mfl - like effective relaxation rate @xmath162 with specific @xmath163 and @xmath164 @xcite . the form eq.([eq10 ] ) for @xmath2 trivially reproduces the remarkable linear law @xmath165 in cuprates @xcite , as well as the non - drude falloff at @xmath166 . a quantitative comparison ( assuming @xmath167 and @xmath168 ) of @xmath2 with the experimental results @xcite is quite remarkable @xcite , realizing the simplicity of the model . it is however evident that the expression ( [ eq10 ] ) is more restricted than the mfl ansatz , hence it should be retested in more detail with experiments . it should be mentioned that @xmath169 has been derived for a single hole conductivity within the retraceable path approximation @xcite , with a restricted validity for @xmath170 ( or possibly @xmath171 ) . we find this behavior only for the intermediate doping , hence new arguments are needed . we can follow the analysis analogous to @xmath98 , eq.([eq6 ] ) , expressing @xmath5 in terms of the eigenstates . as before , the @xmath4-independence of @xmath172 seems plausible . the fact that @xmath173 requires , however , that the current relaxation is very fast , i.e. determined only by the incoherent hopping and by the interhole collisions . the spin fluctuation scale @xmath131 does not enter directly , i.e. even for @xmath139 the spin system only serves as a random bath for the charge degrees of freedom ( holons ) . this conclusion remains valid as far as there is no characteristic frequency @xmath174 ( e.g. the ` pseudogap ' ) in the system . on the other hand , our results also indicate the possible qualitative changes when entering the underdoped regime . e.g. the @xmath175 doping seems to represent a crossover to the latter behavior , where in contrast ( e.g. for @xmath114 ) we find a pronounced @xmath4- and @xmath6-dependence of @xmath5 at @xmath176 . this qualitative change can be again attributed to a ` pseudogap ' appearing at larger @xmath4 in the underdoped systems @xcite . above arguments , both for @xmath97 and @xmath5 , require a large density ( degeneracy ) of the low lying many - body states , apparently being a crucial feature for the most challenging intermediate - doping regime . to quantify this statement we calculate the entropy density @xmath177 , using again the finite-@xmath4 diagonalization method , being less space and time consuming for static quantities @xcite . in fig . 4 we present the results for @xmath178 , obtained for @xmath179 and @xmath180 at different @xmath4 . the main lesson from fig . 4 is that the ` optimal ' cases with @xmath181 are characterized by the largest entropy @xmath182 at low @xmath142 , e.g. @xmath183site for @xmath184 , being almost one half of @xmath185 for an afm . this implies a very large degeneracy of the low - lying states , which could be attributed to the spin subsystem , frustrated by the hole motion . calculated entropy , large even at moderate @xmath4 and strongly dependent on doping , seems to be ( also quantitatively ) consistent with the specific heat measurements on cuprates @xcite . our results for the @xmath0 model seem to indicate that the anomalous , but universal , dynamics in the intermediate - doping regime of correlated systems is a consequence of an extreme degeneracy and a collapse of the low - lying quantum states ( introduced by doping the magnetic insulator ) . this leads to a diffusive - like charge and spin response , where @xmath4 represents the only relevant energy ( frequency ) scale . on the other hand , our results do not exclude the possible onset of coherence ( related to ` pseudogaps ' and superconductivity ) at lower @xmath186 . it is evident that the ` optimal ' doping regime is characterized by the competing influence of two quantum processes : afm fluctuations and hole hopping . outside this regime , either one of the two processes should dominate . to show that such physics is indeed incorporated within the @xmath0 model , we present our results for the uniform susceptibility @xmath8 at various hole dopings @xmath117 . from fig.5 we establish that @xmath8 is most monotonous within the intermediate - doping regime . on the other hand , for @xmath187 the susceptibility starts to reveal a shoulder at @xmath188 , which could be interpreted as an increase of the pseudogap scale , but as well as a sign of the dominating role of the afm fluctuations . g. shirane , physica c * 185 - 189 * , 80 ( 1991 ) ; s.m . hayden _ rev . lett . * 66 * , 821 ( 1991 ) ; t.e . mason , g. aeppli , and h.a . mook , phys . * 68 * , 1414 ( 1992 ) ; b. keimer _ et al . _ , b * 46 * , 14034 ( 1992 ) ; r.j . _ , z. phys . b * 87 * , 15 ( 1992 ) ; m. matsuda _ b * 49 * , 6958 ( 1994 ) . for a review , see e.g. t.m . rice , in _ high temperature superconductivity _ , proc . of the @xmath189 scottish universities summer school in physics , edited by d.p . turnball and w. barford ( adam hilger , 1991 ) , p.317 .
recent results for the finite - temperature static and dynamical properties of the planar @xmath0 model , obtained by a novel numerical method for small correlated systems , are reviewed . of particular interest are @xmath1 charge and spin response functions : optical conductivity @xmath2 and spin susceptibility @xmath3 , which show very universal features at intermediate doping . in spin dynamics the universality is best seen in the local spin correlation function , which appears to be @xmath4-independent and only weakly doping dependent . results apply directly to the inelastic neutron scattering and to the nmr relaxation experiments in cuprates . analysing @xmath2 we find that the current correlation function @xmath5 is nearly @xmath4- and moreover @xmath6-independent in the regime @xmath7 , leading to the @xmath2 behavior , very close to the marginal fermi liquid concept and experiments . these properties are interpreted in connection with the large entropy , related to the large degeneracy of the low - lying states and quantum frustration in the doped antiferromagnets . on the other hand , results for the uniform susceptibility @xmath8 , as well as for the @xmath2 , reveal qualitative changes on entering the underdoped regime .
the discovery@xcite in 1986 of high critical - temperature ( high-@xmath15 ) superconductivity in cuprate oxides has revived intensive and extensive studies on strong electron correlations because it occurs in the vicinity of the mott - hubbard metal - insulator ( m - i ) transition or crossover . when no electrons or _ holes _ are doped , cuprates are insulators ; they exhibit antiferromagnetism at low temperatures . when enough electrons or holes are doped , they become metals ; they exhibit superconductivity at low temperatures . the m - i transition or crossover and the transition between antiferromagnetic and paramagnetic phases as a function of doping concentrations are asymmetric between electron - doped and hole - doped cuprates @xcite ; the insulating phase is much wider in electron - doped ones than it is in hole - doped ones , and the nel temperature is relatively higher in electron - doped ones than it is in hole - doped ones . it is plausible that a relevant theory can explain not only high-@xmath15 superconductivity but also the asymmetry of the m - i transition or crossover and that of the nel temperature . in 1963 , three distinguished theories on electron correlations in a single - band model , which is now called the hubbard model , were published by kanamori , @xcite hubbard , @xcite and gutzwiller . @xcite two theories among them are related with the m - i transition or crossover . according to hubbard s theory , @xcite the band splits into two subbands called the lower and upper hubbard bands . according to gutzwiller s theory , @xcite with the help of the fermi - liquid theory , @xcite a narrow band of quasiparticle appears on the chemical potential ; we call them gutzwiller s band and gutzwiller s quasiparticles . the combination of the two theories implies that the density of states must be of a three - peak structure , gutzwiller s band between the lower and upper hubbard bands . this speculation was confirmed in a previous paper . @xcite the mott - hubbard splitting occurs in both metallic and insulating phases as long as the onsite repulsion @xmath16 is large enough , and gutzwiller s band is responsible for metallic behaviors . then , we can argue that a metal in the vicinity of the m - i transition can become an insulator only when a gap opens in gutzwiller s band or that it can behave as a _ bad _ metal or an insulator when lifetime widths of gutzwiller s quasiparticles are larger than their bandwidth . not only hubbard s @xcite and gutzwiller s @xcite theories but also the previous theory@xcite are within the single - site approximation ( ssa ) . this fact implies that local electron correlations must be responsible for the three - peak structure . local electron correlations are rigorously considered in ssa that includes all the single - site terms ; such an ssa is rigorous for _ paramagnetic phases with no order parameter _ in infinite dimensions . @xcite it is reduced to solving the anderson model , @xcite which is one of the simplest effective hamiltonians for the kondo problem . this approximation is often called the dynamical mean - field theory ( dmft ) . @xcite the kondo problem has been almost settled or almost completely clarified @xcite . the kondo temperature @xmath2 or @xmath3 is defined as an energy scale of local quantum spin fluctuations , with @xmath4 the boltzmann constant . gutzwiller s band between the lower and upper hubbard bands corresponds to the so called kondo peak between two sub - peaks ; its bandwidth @xmath5 is about @xmath17 . one of the most essential physics involved in the kondo problem is that a magnetic moment is quenched by local quantum spin fluctuations so that the ground state is a singlet @xcite or a normal fermi liquid . @xcite a strongly correlated electron liquid on a lattice is a normal fermi liquid at @xmath18 , where magnetic moments are quantum - mechanically quenched , while it is a nondegenerate fermi liquid at @xmath19 , where magnetic moments are thermally quenched . local - moment magnetism occurs at @xmath19 , and itinerant - electron one occurs at @xmath20 ; superconductivity can only occur at @xmath21 . the m - i transition or crossover is related with the crossover between local - moment magnetism and itinerant - electron magnetism . the coherence of quasiparticles is destroyed by not only thermal fluctuations but also disorder . when @xmath22 or @xmath23 , with @xmath7 the quasiparticle lifetime width , quasiparticles are never well - defined ; the liquid behaves as a bad metal or an insulator . when @xmath24 and @xmath25 , on the other hand , they are well - defined ; the liquid behaves as a metal . disorder can also play a significant role in the m - i transition or crossover . kondo - lattice theories @xcite have been developed so far mainly in order to elucidate strong electron correlations in typical or canonical kondo lattices of lanthanide and actinide alloys , which are also called dense kondo systems or heavy - fermion systems . in these canonical kondo lattices , almost localized @xmath26 or @xmath27 electrons and itinerant conduction electrons coexist and conduction electrons can play an important role not only in the quenching of magnetic moments of almost localized electrons but also in the appearance of an intersite exchange interaction mediated by conduction electrons such as the ruderman - kittel - kasuya - yosida ( rkky ) exchange interaction @xcite and the opening of the so called hybridization gaps or pseudogaps . in this paper , we are interested in electron correlations in non - canonical kondo lattices such as transition - metal alloys in the vicinity of the m - i transition or crossover , where the coexistence of almost localized electrons and conduction electrons is absent or it plays no essential role even if it is present . a kondo - lattice theory developed in this paper is a _ perturbation _ theory starting from a starting or an _ unperturbed _ state constructed in a _ non - perturbative _ method , that is , ssa or dmft ; intersite terms are perturbatively considered in terms of intersite exchange interactions . this perturbation is nothing but @xmath28 expansion , with @xmath29 being spatial dimensionality . @xcite it has already been applied to typical issues on electron correlations in non - canonical kondo lattices such as the mott - hubbard m - i transition , @xcite the curie - weiss law of itinerant - electron magnets , @xcite itinerant - electron antiferromagnetism @xcite and ferromagnetism , @xcite and related issues to high-@xmath30 superconductivity , @xcite and so on . early papers @xcite on a mechanism of @xmath31-wave high-@xmath15 superconductivity can also be regarded within the framework of the kondo - lattice theory . the interplay between correlations and disorder in a paramagnetic phase is investigated for the hubbard model with the just half filling . @xcite one of the purposes of this paper is to study the interplay between correlations and disorder in the crossover between local - moment magnetism and itinerant - electron magnetism as a function of electron fillings in the @xmath0-@xmath1 model . this paper is organized as follows : the kondo - lattice theory is reviewed in [ seckondo - lattice ] . a paramagnetic unperturbed state is constructed in [ secfock ] . antiferromagnetic instability of the unperturbed state is studied in [ sectn ] . discussion is given in [ secdiscussion ] . conclusion is given in [ secconclusion ] . the selfenergy of quasiparticles in disordered kondo lattices is studied in appendix . we consider the @xmath0-@xmath1 model @xcite or the @xmath0-@xmath32-@xmath1-@xmath33 model with @xmath34 on a quasi - two dimensional lattice composed of simple square lattices : @xmath35 with @xmath0 the transfer integral between nearest neighbors @xmath36 , @xmath32 between next - nearest neighbors @xmath37 , @xmath38 , with @xmath39 , @xmath40 , and @xmath41 being the pauli matrices , and @xmath42 . infinitely large onsite repulsion @xmath33 is introduced to exclude doubly occupied sites . the @xmath0-@xmath1 model ( [ eqt - jmodel ] ) itself can only treat less - than - half fillings . when the hole picture is taken , it can also treat more - than - half fillings . we consider two models : a _ symmetric _ one with @xmath43 , where physical properties are symmetric between less - than - half and more - than - half fillings , and an _ asymmetric _ one with @xmath44 , whose precise definition is made in [ secfock ] . effects of disorder and weak three dimensionality are phenomenologically considered in this paper . we assume @xmath45 ; @xmath46 ev and @xmath47-@xmath48 ev for cuprate oxide superconductors . we follow the previous paper @xcite to treat @xmath49 . the selfenergy is divided into the single - site one @xmath50 and the multisite one @xmath51 . the single - site one @xmath52 is given by that of the anderson model with the same @xmath49 as that of the @xmath0-@xmath1 model ( [ eqt - jmodel ] ) ; other parameters should be selfconsistently determined to satisfy @xcite @xmath53 with @xmath54 the number of sites , and @xmath55 and @xmath56 the green functions of the anderson and @xmath0-@xmath1 models , respectively . here , @xmath57 is the chemical potential and @xmath58 , with @xmath59 @xmath60 with @xmath61 the lattice constant of the simple square lattices . note that the mapping condition ( [ eqmap ] ) with eq . ( [ eqgreen ] ) depends on the multisite selfenergy , which should be perturbatively and selfconsistently calculated . there can be various approximations of singles - site selfenergies or dynamical mean fields depending on it . as is shown in [ secfock ] , the multisite selfenergy includes an energy independent part , which is denoted by @xmath62 , because the superexchange interaction constant does not depend on energies in the @xmath0-@xmath1 model . in this paper , we include only @xmath62 in eq . ( [ eqgreen ] ) ; the green function that includes @xmath62 instead of @xmath63 is denoted by @xmath64 . the kondo temperature @xmath2 is defined by @xmath65_{t\rightarrow 0 } , \ ] ] where @xmath66 is the spin susceptibility of the anderson model , which does not include the conventional factor @xmath67 , with @xmath68 and @xmath69 being the @xmath68 factor and the bohr magneton , respectively . the single - site selfenergy @xmath50 is expanded for @xmath70 as @xmath71 , with @xmath72 and @xmath73 infinitesimally small spin - dependent chemical potential shifts . the wilson ratio is defined by @xmath74 , with @xmath75 . for almost half fillings , charge fluctuations are suppressed so that @xmath76 . for such fillings , @xmath77 so that @xmath78 . the green function is divided into coherent and incoherent parts : @xmath79 , with @xmath80 with @xmath81 and @xmath82/\tilde{\phi}_\gamma$ ] . we introduce a phenomenological lifetime width @xmath7 . although @xmath7 depends on energies even if it is due to disorder , as is discussed in appendix [ secdisorder ] , its energy dependence is ignored . @xcite according to the fermi - surface sum rule , @xcite the number of electrons per site for @xmath83 and @xmath84 is given by @xmath85 with @xmath86 ^ 2 + \gamma^2 } , \qquad\ ] ] @xmath87 , \ ] ] with @xmath88 the di - gamma function . note that @xmath89 $ ] . we assume eq . ( [ eqfsumrule ] ) even for nonzero @xmath8 and @xmath7 . the parameter @xmath90 or @xmath91 can be determined from eq . ( [ eqfsumrule ] ) as a function of @xmath92 . in kondo lattices , local spin fluctuations at different sites interact with each other by an exchange interaction . following this physical picture , we define an intersite exchange interaction @xmath93 by @xmath94 where @xmath95 is the spin susceptibility of the @xmath0-@xmath1 model to be studied . following the previous paper , @xcite we obtain @xmath96 with @xmath97 and @xmath98 the multi - site part of the irreducible polarization function in spin channels ; @xmath99 is examined in [ secexc ] and [ secmode ] . according to the ward relation , @xcite the irreducible single - site three - point vertex function in spin channels , @xmath100 , is given by @xmath101 for @xmath102 and @xmath103 . we approximately use eq . ( [ eqthreel ] ) for @xmath104 and @xmath105 . a mutual interaction between quasiparticles mediated by spin fluctuations is given by @xmath106 ^ 2 f_{\rm s}({\rm i } \omega_l,{\bm q } ) = \tilde{\phi}_{\rm s}^2 \frac{1}{4 } i_{\rm s}^ * ( { \rm i } \omega_l , { \bm q } ) , \ ] ] with @xmath107 @xmath108 the single - site term is subtracted in @xmath109 because it is considered in ssa . the expansion coefficient @xmath110 appears as an effective or reducible single - site vertex function in eq . ( [ spinflucti ] ) . because the spin space is isotropic , the interaction in the transversal channels is also given by these equations . in the kondo - lattice theory , intersite effects are perturbatively considered in terms of @xmath93 , @xmath111 , or @xmath112 depending on each situation . in this section , we restrict the hilbert space within paramagnetic states . there are two types of selfenergies linear in the exchange interaction : hartree - type and fock - type terms . the hartree - type term vanishes in paramagnetic states . we consider the fock - type term of the superexchange interaction : @xmath113 the factor 3 appears because of three spin channels . when only the coherent part is considered , @xmath114 with @xmath115 .\ ] ] the dispersion relation of quasiparticles is given by @xmath116 where @xmath117 should be selfconsistently determined to satisfy @xmath118 and @xmath119 is simply given by @xmath120 . for the symmetric model , @xmath43 so that @xmath121 . in order to examine how crucial role the shape of the fermi surface plays in the asymmetry , we consider a phenomenological asymmetric model with @xmath122 . of the asymmetric model for @xmath123 . see also the caption of fig . [ t - star ] . , width=302 ] expansion parameters @xmath124 and @xmath110 are given by those of the mapped anderson model . however , we approximately use those for the anderson model with a constant hybridization energy . according to appendix of the previous paper @xcite , @xmath125 @xmath126 where @xmath127 is the concentration of dopants , _ holes _ ( @xmath128 ) or electrons ( @xmath129 ) . these are consistent with gutzwiller s theory . @xcite figures [ t - star ] and [ t - star - asym ] show @xmath117 of the symmetric and asymmetric models as a function of @xmath130 for various @xmath7 . it is interesting that @xmath117 is nonzero even for @xmath131 if @xmath7 and @xmath132 are small enough . for the symmetric model ( @xmath133 ) , eq . ( [ eqxi ] ) can be analytically calculated for @xmath134 k , @xmath135 and @xmath136 @xmath137 in such a way that @xmath138 . it follows that @xmath139_{\delta \rightarrow 0 } \rightarrow -\left ( 3\tilde{w}_{\rm s}^2j / 2\pi^2 t \right ) = 0.18 $ ] for @xmath140 . if @xmath7 or @xmath132 is large enough , on the other hand , @xmath141 and @xmath117 vanish for @xmath131 . figure [ rho_n_mu_fs ] shows physical properties of the unperturbed state of the symmetric model . those of the unperturbed state of the asymmetric model can be found in fig . 2 of ref . . in [ sectn ] , we study antiferromagnetic instabilities of these paramagnetic unperturbed states . according to eq . ( [ eqkondosus ] ) , the nel temperature @xmath13 is determined by the competition between the kondo effect and the intersite exchange interaction : @xmath142 where @xmath143 is an ordering wave number to be determined . effects of local spin fluctuations in ssa or those of dynamical mean fields in dmft are included in the local term of @xmath144 ; @xmath145 corresponds to the conventional weiss s mean field , which is an intersite effect and can only be included beyond ssa . the local susceptibility @xmath146 is almost constant at @xmath147 , while it obeys the curie law at @xmath19 . we use an interpolation between the two limits : @xmath148 here , @xmath2 is an averaged one over @xmath2 s at different sites . in disordered systems , the mapping conditions are different from site to site so that @xmath2 s are also different from site to site . such disorder in @xmath2 causes energy - dependent lifetime widths , as is studied in appendix . however , lifetime widths due to the disorder in @xmath2 are small on the chemical potential in case of non - magnetic impurities . according to eqs . ( [ eqmap ] ) and ( [ eqdeftk ] ) together with the fermi - liquid relation , @xcite it follows that @xmath149_{t=0\mbox{~\scriptsize k } } = 2 \tilde{w}_{\rm s } \left[\rho_{\gamma } ( \mu^ * ) \right]_{\gamma \rightarrow 0},\ ] ] in the absence of disorder . then , a mean value of @xmath2 in disordered systems is approximately given by eq . ( [ eqtk2 ] ) with nonzero but small @xmath7 . it follows from eq . ( [ eqtk2 ] ) that @xmath150 with @xmath151 a numerical constant depending on @xmath92 . as is shown in fig . [ rho_n_mu_fs](a ) , @xmath152 for @xmath153 and @xmath154 . we assume that @xmath151 is independent of @xmath92 for the sake of simplicity : @xmath155 . we are only interested in physical properties that never drastically change when @xmath151 slightly changes . the second term of eq . ( [ eqis ] ) is the sum of an exchange interaction arising from the virtual exchange of pair excitations of quasiparticles , @xmath156 , and the mode - mode coupling term , @xmath157 : @xmath158 when higher - order terms in intersite effects are ignored , @xmath159 ^ 2 \left [ p({\rm i } \omega_l , { \bm q } ) - p_0({\rm i } \omega_l ) \right],\ ] ] with @xmath160 the local contribution @xmath161 is subtracted because it is considered in ssa . the static component is simply given by @xmath162 \!-\ ! f_\gamma \bigl[\xi({\bm k } ) \!-\ ! \mu^*\bigr ] } { \xi({\bm k } ) -\xi({\bm k } + { \bm q } ) -{\rm i } 0 } .\ ] ] the magnitude of @xmath163 is proportional to @xmath3 or the quasiparticle bandwidth . |t^*|$ ] of the asymmetric model : @xmath164 . a solid line shows a result for @xmath165 , dashed lines results of electron doping cases such as @xmath166 , 1.2 and 1.3 , and broken lines results for hole doping cases such as @xmath167 , 0.8 , and 0.7 . , width=245 ] figures [ pol - sym ] and [ pol - asy ] show @xmath168 $ ] of the symmetric and asymmetric models . the polarization function at @xmath169 is relatively larger in electron - doping cases than it is in hole - doping cases . following previous papers , @xcite we consider mode - mode coupling terms linear in intersite spin fluctuations @xmath170 given by eq . ( [ eqf ] ) : @xmath171 the first term @xmath172 is a _ local _ mode - mode coupling term , which includes a single _ local _ four - point vertex function , as is shown in fig . 4 of ref . . both of @xmath173 and @xmath174 are _ intersite _ mode - mode coupling terms , which include a single _ intersite _ four - point vertex function ; a single @xmath175 appears as the selfenergy correction to the single - particle green function in @xmath173 while it appears as a vertex correction to the polarization function in @xmath174 , as are shown in figs . 3(a ) and 3(b ) , respectively , of ref . . their static components are given by @xmath176 @xmath177,\ ] ] @xmath178 with @xmath179 ^ 2 g_\sigma^{(0)}({\rm i } \varepsilon_n+{\rm i } \omega_l,{\bm k}+{\bm p } ) % \nonumber \\ & & - \left[r_\sigma^{(0)}({\rm i } \varepsilon_n)\right]^3 r_\sigma^{(0)}({\rm i } \varepsilon_n+{\rm i } \omega_l ) \biggr\ } , \end{aligned}\ ] ] @xmath180 ^ 2 \ ! \left[r_\sigma^{(0)}({\rm i } \varepsilon_n \!+\ ! { \rm i } \omega_l)\right]^2 \biggr\ } , \end{aligned}\ ] ] with @xmath181 . because the selfenergy correction linear in @xmath182 is considered in [ secfock ] , @xmath183 is subtracted in eq . ( [ lambda_s ] ) . in this paper , weak three dimensionality in spin fluctuations is phenomenologically included . because @xmath182 has its maximum value at @xmath184 and the nesting vector of the fermi surface in two dimensions is close to @xmath185 for almost half filling , we assume that the ordering wave number in three dimensions is @xmath186 , with @xmath187 depending on interlayer exchange interactions . on the phase boundary between paramagnetic and antiferromagnetic phases , where eq . ( [ eqafcondition ] ) is satisfied , the inverse of the susceptibility is expanded around @xmath143 and for small @xmath188 in such a way that @xmath189_{t = t_{\rm n } } = a({\bm q } ) + \alpha_\omega |\omega_l| + \cdots,\ ] ] with @xmath190 ^ 2 .\ ] ] here , @xmath191 is the lattice constant along the @xmath192 axis . because @xmath193 diverges in the limit of @xmath194 and @xmath195 on the phase boundary , @xmath196 ^ 2 % % \nonumber \\ & & \times \left[g_\sigma^{(0)}({\rm i } \varepsilon_n,{\bm k } ) \right]^2 \!-\!\left[r_\sigma^{(0 ) } ( { \rm i } \varepsilon_n)\right]^4 \biggr\ } , \hspace{1.1cm}\end{aligned}\ ] ] can be approximately used for @xmath197 in eq . ( [ lambda_s ] ) and @xmath198 in eq . ( [ lambda_v ] ) . then , it follows that @xmath199 with @xmath200 \!\ ! % \nonumber\\ & & \hspace*{5 cm } \times \frac1{[\xi({\bm k})-\xi({\bm k}+{\bm q})]^2 } , \end{aligned}\ ] ] @xmath201 f_\gamma(\varepsilon-\mu^ * ) , \ ] ] @xmath202 \ ! % % \nonumber \\ & & \times \biggl\{\tan^{-1}\ ! \left [ \frac{a_\parallel ( q _ { \rm c } a)^2 + a_z ( q_zc)^2 } { 4\alpha_\omega\omega}\right ] % % \nonumber \\ & & \qquad - \tan^{-1}\!\left [ \frac{a_z ( q_zc)^2 } { 4\alpha_\omega\omega}\right ] \biggr\ } . \nonumber \\ & & \end{aligned}\ ] ] in eq . ( [ eqcf ] ) , @xmath203 is the derivative of @xmath204 defined by eq . ( [ eqfgamma ] ) . in eq . ( [ eqcl ] ) , @xmath205 is denoted by @xmath206 , and @xmath207 in eq . ( [ eqphi ] ) , the summation is restricted to @xmath208 and @xmath209 . we assume that @xmath210 and @xmath211 is given by a larger one of @xmath212 and @xmath11 . as was shown in the previous paper , @xcite @xmath213 for @xmath214 and @xmath215 . physical properties for @xmath216 or @xmath217 scarcely depends on @xmath218 . then , we assume @xmath219 for any @xmath8 and @xmath7 in this paper . the kondo temperature @xmath2 is low in the limit of @xmath131 for nonzero @xmath8 or @xmath7 . when @xmath216 or @xmath217 , the instability condition ( [ eqafcondition ] ) becomes simple : @xmath220 when we ignore @xmath221 and we assume @xmath222 , eq . ( [ eqafcondition2 ] ) gives @xmath223 . this @xmath13 is nothing but @xmath13 in the mean - field approximation for the heisenberg model , to which the @xmath0-@xmath1 model with the half filling is reduced . on the other hand , @xmath224 diverges at non zero temperatures for @xmath225 ; no magnetic instability occurs at nonzero temperature in complete two dimensions . @xcite the reduction of @xmath13 by @xmath226 or by quasi - two dimensional thermal spin fluctuations depends on the anisotropy of @xmath227 . when only the superexchange interaction is considered , @xmath228 . although @xmath229 also contribute to the @xmath230-quadratic term , we assume @xmath231 for the sake of simplicity . figure [ tn - asym ] show @xmath13 as a function of the exponent @xmath232 of @xmath233 for the half filling @xmath234 . when we take @xmath235 , for example , we obtain @xmath236 or @xmath237 . we assume this anisotropy factor @xmath235 in the following part of this paper . determined by eq . ( [ eqafcondition2 ] ) for @xmath238 and @xmath239 as a function of the exponent @xmath232 of the anisotropy factor defined by @xmath233 . , width=283 ] of the symmetric model as a function of @xmath240 . from the bottom , solid , dashed , broken , chain , and chain double - dashed lines show @xmath13 for @xmath241 , 0.3 , 0.4 , 1 , and 2 , respectively . the topmost solid line shows @xmath13 determined from eq . ( [ eqafcondition2 ] ) for comparison . the solid line with a positive slope shows @xmath242 . antiferromagnetic states whose @xmath13 are much above and below this line are characterized as local - moment ones and itinerant - electron ones , respectively , according to discussion in [ secdiscussion ] . , width=283 ] of the asymmetric model as a function of @xmath130 . see also the caption of fig . [ neel - tn ] . note that @xmath13 is almost symmetric between @xmath129 and @xmath128 . , width=283 ] figures [ neel - tn ] and [ neel - tn - asym ] show @xmath13 of the symmetric and asymmetric models as a function of @xmath243 for various @xmath7 . the antiferromagnetic phase becomes narrower as @xmath7 decrease . as long as @xmath7 is almost symmetric between @xmath128 and @xmath129 , @xmath13 is also almost symmetric between them . when @xmath23 or @xmath22 , @xmath13 is determined by the competition between the superexchange interaction @xmath1 and two quenching effects , @xmath144 and @xmath226 . the reduction of @xmath13 for small @xmath244 is mainly due to @xmath226 . on the other hand , the critical concentration @xmath245 below which antiferromagnetic ordering appears at @xmath246 k is determined by the competition between @xmath1 and @xmath247 , because thermal spin fluctuations vanish at @xmath246 k. because the kondo effect is rather weak , @xmath245 is as large as @xmath248 , as is shown in figs . [ neel - tn ] and [ neel - tn - asym ] . when @xmath25 and @xmath249 , well - defined quasiparticles can play roles in both of the enhancement and suppression of @xmath13 . antiferromagnetism is enhanced by @xmath250 , in which the nesting of the fermi surface can play a significant role . however , the difference of the fermi surfaces can not give any significant asymmetry of @xmath13 . the reduction of @xmath251 by the kondo effect is large . the intersite mode - mode coupling term , @xmath252 , also suppresses @xmath13 in addition to the local terms , @xmath3 and @xmath226 . the quenching effects overcome the enhancement effect , and @xmath13 decrease with decreasing @xmath7 . one of the most serious assumptions in this paper is that the homogeneous life - time width @xmath7 of quasiparticles , which is phenomenologically introduced in this paper , can be defined . this is relevant when there are many weak impurities and their distribution is totally random . when disorder is so small that quasiparticles are well defined , electrons behave as landau s fermi liquid . the assumption is obviously relevant in such a case . on the other hand , when disorder is so large in the vicinity of the mott transition or crossover that quasiparticles are never well - defined , electrons behave as localized magnetic moments . because @xmath7 plays no role , the theoretical framework of this paper must also be relevant in such a case . then , an issues is whether the assumption is relevant or irrelevant in the crossover regime between the two regimes , for example , in the regime where the anderson localization of quasiparticles occurs . because the homogeneous lifetime width can be defined in the so called weakly localized regime of the anderson localization , @xcite we expect that the results of this paper apply to such a regime . the assumption is irrelevant when scattering potentials of dopants are strong and the concentration of dopants is rather small . the strong anderson localization , where the localization length is short , is out of scope in this paper . the crossover between the mott - hubbard insulator and the anderson insulator , which is examined in ref . , is also out of scope in this paper . one may suspect that the theoretical framework of this paper can not apply to the limit of @xmath253 or to vanishingly small @xmath244 . according to the combination of gutzwiller s theory @xcite and the fermi - liquid theory , @xcite the bandwidth of quasiparticles is vanishingly small but is still nonzero such as @xmath254 . the fermi liquid state is formed , at least , at @xmath255 ; the spectral weight of quasiparticles is vanishingly small . it is obvious that when the renormalization of quasiparticles due to the fock - type term of the superexchange interaction is taken into account the bandwidth becomes as large as @xmath11 , at least , at @xmath255 . we show in this paper that these quasiparticles are stable not only for @xmath256 and @xmath257 but also for @xmath258 and @xmath259 . the existence of elementary excitations or quasiparticles whose bandwidth is of the order of @xmath11 are implied by several studies on the @xmath0-@xmath1 model on small lattices . @xcite then , we can argue that when the hilbert space is restricted within paramagnetic phases no divergence of the effective mass of quasiparticles or no m - i transition can occur except for @xmath260 , @xmath131 , and @xmath261 k. what can occur for nonzero @xmath1 is a m - i crossover rather than a m - i transition , as is shown in figs . [ t - star ] and [ t - star - asym ] . the coexistence phase of a metal and a mott - hubbard insulator is obtained for the hubbard model with the just half filling and finite onsite @xmath16 . @xcite it is interesting to examine whether or not the coexistence phase remains when the renormalization of quasiparticles by an intersite exchange interaction such as the superexchange interaction is included . it is also interesting to examine whether or not the coexistence phase remains for non - half fillings . quasiparticles are well - defined for @xmath20 and @xmath262 . according to refs . and , the exchange interaction @xmath263 , which arises from the virtual exchange of pair excitations of quasiparticles , shows an almost @xmath8-linear dependence consistent with the curie - weiss law at @xmath20 in a small region of @xmath264 , with @xmath143 being the nesting wavenumber , for itinerant - electron antiferromagnets where the nesting of the fermi surface is significant . @xcite on the other hand , the @xmath8-linear dependence of @xmath144 at @xmath19 is responsible for the curie - weiss law of local - moment magnets or insulating magnets . then , magnetism with @xmath265 is characterized as local - moment one , while magnetism with @xmath14 is characterized as itinerant - electron one . local - moment magnetism appears for almost half fillings , and itinerant - electron magnetism appears for fillings substantially away from the half filling . because no order is possible at non - zero temperatures in two dimensions , @xcite it is likely that @xmath13 is substantially reduced by critical thermal fluctuations in quasi - two dimensions . as is examined in [ secalmost ] , the reduction of @xmath13 is large ; @xmath237 for the anisotropy of @xmath266 . this explains observed @xmath267 k for cuprate oxides , when we take @xmath268 ev . the exponent @xmath269 seems to be little larger than actual ones . we should consider the reduction of @xmath13 by thermal critical fluctuations more properly than we do in this paper . the nesting of the fermi surface becomes less sharp as @xmath7 becomes larger , as is shown in figs . [ pol - sym ] and [ pol - asy ] . one may argue that antiferromagnetism is weaken by disorder in the itinerant - electron regime where the nesting of the fermi surface plays a crucial role . however , this effect is overcome by the reduction of @xmath2 by nonzero @xmath7 . the renormalization of the quasiparticle bandwidth by the superexchange interaction becomes small with increasing @xmath7 so that the kondo temperature @xmath3 becomes small . on the other hand , the superexchange interaction is not reduced by nonzero @xmath7 . then , antiferromagnetism is strengthen with increasing @xmath7 even in the itinerant - electron region . an antiferromagnetic states appears in a narrow range of @xmath270 , while it appears in a wide range of @xmath271@xmath272 in electron - doped cuprate oxides @xmath273 . tohyama and maekawa @xcite argued that the asymmetry between hole - doped and electron - doped oxides must arises from the difference of the fermi surfaces , and that the @xmath0-@xmath32-@xmath1 model should be used . they showed that the intensity of spin excitations is relatively stronger in electron doping cases than it is in hole doping cases . according to fig . [ pol - asy ] , the polarization function at @xmath274 is relatively larger in electron doping cases than it is in hole doping cases . this asymmetry is consistent with that of spin excitations studied by tohyama and maekawa . however , the difference of the fermi surfaces can not explain the asymmetry of @xmath13 , as is shown in fig . [ neel - tn - asym ] . the condensation energy at @xmath246 k of the asymmetric model is also quite asymmetric ; @xcite it is consistent with the asymmetry of spin excitations and the polarization function discussed above . on the other hand , @xmath13 is significantly reduced by quasi - two dimensional spin fluctuations as well as local spin fluctuations of the kondo effect . this large reduction of @xmath13 arises from the renormalization of normal states ; not only the nel states but also paramagnetic states just above @xmath13 are largely renormalized . it is plausible that the asymmetry of the condensation energy of paramagnetic states just above @xmath13 is similar to that of the nel states at @xmath246 k. it is interesting to examine by comparing the condensation energy of the nel states and that of paramagnetic states just above @xmath13 whether @xmath13 is actually almost symmetric as is shown in this paper . electrical resistivities of electron - doped cuprates are relatively larger than those of hole - doped cuprates are . @xcite then , we can argue that disorder must be relatively larger in electron - doped cuprates than it is in hole - doped cuprates , and that the asymmetry of @xmath13 can arise , at least partly , from the difference of disorder between them . it is interesting to examine whether or not the symmetric behavior of @xmath13 can be restored by preparing hole - doped and electron - doped cuprates with similar degree of disorder to each other . when nonmagnetic impurities of zn ions are introduced on cuo@xmath275 planes , antiferromagnetic moments appear in the neighborhood of zn ions . @xcite because the configuration of cu ions is @xmath276 and that of zn ions is @xmath277 , the phase shift at the chemical potential due to scatterings by zn ions must be about @xmath278 according to the friedel sum rule . then , scatterings by zn ions must be very strong . one of the plausible scenarios is that the renormalization of @xmath3 by the fock - type term of the superexchange interaction is small in the neighborhood of zn ions because of strong scatterings , that is , the quenching of magnetic moments by local quantum spin fluctuations are weak there so that antiferromagnetism is enhanced there . it is interesting to develop a microscopic theory that can treat such an inhomogeneous effect . of the symmetric model as a function of @xmath130 for various @xmath7 and @xmath279 ; @xmath280 or @xmath281 is assumed instead of @xmath282 or @xmath283 ( see text ) . from the top , solid , dashed , broken , and chain line show @xmath117 for @xmath284 , 0.1 , 0.2 , and 0.4 , respectively . for the sake of comparison , @xmath242 is also shown by a bottom solid line . , width=283 ] of the symmetric model as a function of @xmath130 for various @xmath7 ; @xmath280 or @xmath281 is assumed instead of @xmath282 or @xmath283 ( see text ) . from the bottom , solid , dashed , broken , and chain lines show @xmath13 for @xmath284 , 0.1 , 0.2 , and 0.4 , respectively . for comparison , @xmath13 determined from eq . ( [ eqafcondition2 ] ) is shown by a topmost solid line . , width=283 ] the three - point vertex function @xmath110 and the mass - renormalization factor @xmath124 are those in ssa ; @xmath285 as is discussed in [ secrenomalizedssa ] . when we take @xmath282 , on the other hand , the superexchange interaction constants as small as @xmath286 ev , which are much smaller than experimental ones @xmath287 ev , give @xmath30 as high as @xmath288 k , @xcite which are as high as observed @xmath30 . this implies that the renormalization of the vertex function and the mass renormalization factor by intersite effects such as antiferromagnetic and superconducting fluctuations should be included . taking it into account , we treat @xmath289 as another phenomenological parameter following the previous paper,@xcite where @xmath290 is used instead of @xmath282 in order to explain observed @xmath30 and observed coefficients of @xmath8-linear resistivities when @xmath287 ev are used . then , we replace @xmath289 by @xmath291 , with @xmath292 a numerical constant smaller than unity ; eq . ( [ eqt * ] ) is replaced by @xmath293 and eqs . ( [ eqtk2 ] ) and ( [ eqtkdef ] ) are replaced by those with @xmath291 instead of @xmath289 . @xcite figures [ tstar-1 ] and [ neel - tn3 ] show @xmath117 and @xmath13 , respectively , of the symmetric model as a function of @xmath130 for @xmath281 and various @xmath7 . the antiferromagnetic region extends with decreasing @xmath292 . an antiferromagnetic state in the range of @xmath294 of hole - doped la@xmath295m@xmath296cuo@xmath297 ( m= sr or ba ) is characterized as a local - moment one . the so called spin - glass or kumagai s phase@xcite appears in the range of @xmath298 . we characterize the tail part of the solid line in fig . [ neel - tn3 ] as a critical line to the kumagai phase ; @xcite @xmath299 are needed to reproduce kumagai s phase . according to fig . [ tstar-1 ] , @xmath300 for @xmath301 . then , we can argue that @xmath302 , with @xmath303 the fermi wave number and @xmath304 the mean free path , must be 48 in kumagai s phase . according to fig . [ pol - sym](b ) , the nesting of the fermi surface is substantial at least for @xmath305 ; the nesting can not be ignored for @xmath306 . kumagai s phase must be a spin density wave ( sdw ) state in a disordered system rather than a spin glass . the divergence@xcite of the nuclear quadrupole relaxation ( nqr ) rate at @xmath13 supports this characterization . the @xmath0-@xmath1 model with @xmath307 on a quasi - two dimensional lattice is studied . first , an _ unperturbed _ state is constructed in ssa or dmft . the kondo temperature @xmath2 or @xmath3 is defined as the energy scale of local quantum spin fluctuations . the unperturbed state is a normal degenerate or almost degenerate fermi liquid at @xmath308 and is a non - degenerate fermi liquid at @xmath309 . no metal - insulator transition is possible within paramagnetic phases ; what occurs is a metal - insulator crossover . the bandwidth @xmath5 of quasiparticles , which is about @xmath310 , is renormalized by the fock - type selfenergy due to the superexchange interaction @xmath1 , so that that @xmath5 is larger for lower @xmath8 and smaller @xmath7 , with @xmath7 being the phenomenologically introduced lifetime width of quasiparticles . the bandwidth @xmath5 is approximately given by the sum of terms of @xmath311 and @xmath312 , with @xmath130 being the concentration of dopants added to the just half - filled system ; the term of @xmath311 is consistent with gutzwiller s theory , and the coefficient of the term of @xmath312 is @xmath313 for @xmath314 and @xmath315 and is vanishingly small for @xmath316 or @xmath317 . next , antiferromagnetic instability of the unperturbed state is examined perturbatively in terms of intersite exchange interactions . this perturbation theory can treat both of local - moment magnetism and itinerant - electron magnetism . when @xmath265 , with @xmath13 the nel temperature , or local thermal spin fluctuations are dominant over local quantum spin fluctuations at @xmath13 , magnetism is characterized as local - moment magnetism ; it appear for almost half fillings of electrons . in the opposite case where @xmath318 or local quantum ones are dominant over local thermal ones at @xmath13 , magnetism is characterized as itinerant - electron magnetism ; it appears for fillings away from the half filling . the reduction of @xmath13 by quasi - two dimensional thermal critical fluctuations is large when the anisotropy of the exchange interaction constants is large . for example , @xmath319 in complete two dimensions , as is expected . the nel temperature @xmath267 k observed in cuprates with no doping can be explained , when we assume @xmath320 ev and @xmath321 , with @xmath1 and @xmath322 the exchange interaction constants between nearest neighbors within a cuo@xmath275 plane and between cuo@xmath275 planes , respectively . when @xmath323 or @xmath324 , @xmath325 or @xmath326 so that the quenching of magnetic moments by local quantum spin fluctuations is rather weak . because thermal spin fluctuations vanish at @xmath246 k , an antiferromagnetic state is stabilized in a wide range of @xmath130 . when @xmath25 and @xmath24 , quasiparticles are well - defined . because the nesting of the fermi surface is substantial , the exchange interaction arising from the virtual exchange of pair excitations of quasiparticles is also responsible for antiferromagnetic instability in addition to the superexchange interaction . on the other hand , the quenching of magnetic moments by local quantum spin fluctuations is strong . not only quasi - two dimensional thermal critical fluctuations but also the quenching of magnetic moments by local quantum fluctuations make @xmath13 substantially reduced or they destroy antiferromagnetic ordering . an antiferromagnetic state is stabilized only for small @xmath244 ; the critical concentration @xmath245 below which antiferromagnetic ordering appears is smaller for smaller @xmath7 . this result implies that @xmath13 is lower and a magnetic phase is narrower in a cleaner system . the difference or asymmetry of disorder between electron - doped and hole - doped cuprates must be , at least partly , responsible for that of @xmath13 and antiferromagnetic phases between them . we characterize the so called kumagai s phase as a spin density wave ( sdw ) state in disordered system rather than a spin glass . the author is thankful for discussion to m. ido , m. oda , n. momono , and k. kumagai . this work was supported by a grant - in - aid for scientific research ( c ) grant no . 13640342 from the ministry of education , cultures , sports , science and technology of japan . in disordered kondo lattices , the mapped anderson models are different from site to site . the selfenergy for the @xmath327th one is expanded in such a way that @xmath328 . when the energy dependence of the hybridization energy @xmath329 is ignored , it follows according to shiba @xcite that @xmath330 , \ ] ] with @xmath331 and @xmath332 being the localized - electron level and the number of electrons with spin @xmath333 , respectively . it follows from eqs . ( [ eqphig ] ) and ( [ eqphis ] ) that @xmath334 and @xmath335 , with @xmath336 , for almost half filling @xmath337 . we assume non - magnetic impurities so that @xmath338 . we denote the average number of electrons and their mean - square deviation by @xmath339 and @xmath340 , respectively , where @xmath341 stands for an ensemble average over disordered systems . we assume that there is no correlation between different sites : @xmath342 for @xmath343 . we consider the site - dependent part of @xmath344 as a static but energy - dependent scattering potential ; it is approximately given by @xmath345 here , the averages of @xmath346 and @xmath347 are simply dented by @xmath124 and @xmath348 . when we treat @xmath349 in the second - order ssa or the born approximation , the coherent part of the ensemble averaged green function is given by @xmath350 with @xmath351 ^ 2 \!\!\delta n^2 \rho_\gamma ( \varepsilon ) , \ ] ] for @xmath352 . the bandwidth of quasiparticles is about @xmath353 ; typical lifetime width is as large as @xmath354 the energy - independent term can be ignored because @xmath355 for almost half fillings . it is quite likely that @xmath356 and @xmath357 for almost half fillings . it is straightforward to extend the above argument to a system in the presence of magnetic fields and a system with magnetic impurities . in such cases , @xmath358 can be different from site to site ; @xmath359 can be large even for @xmath360 . bednortz and k.a . mller : z. phys . b * 64 * ( 1986 ) 189 . see for example , h. takagi , y. tokura and s. uchida : physica c * 162 - 164 * ( 1989 ) 1001 . j. kanamori : prog . theor . phys . * 30 * ( 1963 ) 275 . j. hubbard : proc . london ser . a * 276 * ( 1963 ) 238 ; a * 281 * ( 1964 ) 401 . gutzwiller : phys . * 10 * ( 1963 ) 159 ; phys . a * 134 * ( 1963 ) 293 ; a * 137 * ( 1965 ) 1726 . luttinger and j.c . ward : phys . rev . * 118 * ( 1960 ) 1417 . luttinger : phys . rev . * 119 * ( 1960 ) 1153 . ohkawa : j. phys . * 58 * ( 1989 ) 4156 . w. metzner and d. vollhardt : phys . * 62 * ( 1989 ) 324 . ohkawa : phys . b * 44 * ( 1991 ) 6812 . f.j . ohkawa : j. phys . * 60 * ( 1991 ) 3218 ; * 61 * ( 1992 ) 1615 . a. georges and g. kotliar : phys . b * 45 * ( 1992 ) 6479 . see , for example , g. kotliar and d. vollhardt : phys . today * 57 * ( 2004 ) 53 . k. yosida : phys . rev . * 147 * , 223 ( 1966 ) . p. w. anderson : j. phys . c * 3 * ( 1970 ) 2436 . k. g. wilson : rev . * 47 * ( 1975 ) 773 . p. nozires : j. low . temp . * 17 * ( 1974 ) 31 . k. yamada : prog . . phys . * 53 * ( 1975 ) 970 ; * 54 * ( 1975 ) 316 . k. yosida and k. yamada : prog . * 53 * ( 1970 ) 1286 . c. lacroix : j. magn . magn . mat . , * 100 * ( 1991 ) 90 . see , for example , this and the following references @xcite for canonical kondo lattices . g. zwicknagl : adv . phys . * 41 * ( 1992 ) 203 . p. coleman and n. andrei : j. phys . matter * 1 * ( 1989 ) 4057 . f. j. ohkawa : j. phys . * 67 * ( 1998 ) 525 . according to this reference , the rkky interaction works only at high temperatures such as @xmath19 or in the so called localized regime where almost localized electrons such as @xmath26 and @xmath27 ones behave as localized magnetic moments . at low temperatures such as @xmath20 or in the so called itinerant electron regime where they behave as itinerant electrons , it is replaced by the exchange interaction arising from the virtual exchange of pair excitations of quasiparticles , which are admixtures of almost localized electrons and conduction electrons . the exchange interaction corresponds to @xmath361 in this paper . therefore , the hubbard or @xmath0-@xmath1 model can be a relevant effective hamiltonian for canonical kondo lattices where the hybridization gap plays no essential role . ohkawa : j. phys . . jpn . * 67 * ( 1998 ) 535 . e. miyai and f.j . ohkawa : phys . b * 61 * ( 2000 ) 1357 . f.j . ohkawa : phys . b * 66 * ( 2002 ) 014408 . f.j . ohkawa : phys . b * 65 * ( 2002 ) 174424 . f.j . ohkawa : phys . b * 69 * ( 2004 ) 104502 . f.j . ohkawa : phys . b * 70 * ( 2004 ) 184514 . ohkawa : jpn . * 26 * ( 1987 ) l652 . ohkawa : j. phys . * 56 * ( 1987 ) 2267 . k. byczuk , w. hofstetter and d. vollhardt : phys . * 94 * ( 2005 ) 056404 . hirsch : phys . * 54 * ( 1985 ) 1317 . zhang and t.m . rice : phys . b * 37 * ( 1988 ) r3759 . because of this simplification , we can carry out analytically summations over matsubara s frequencies in many equations of this paper . j.c . ward : phys . rev . * 68 * ( 1950 ) 182 . a. kawabata : j. phys . f * 4 * ( 1974 ) 1447 . t. moriya : _ spin fluctuations in itinerant electron magnetism _ ( springer series in solid - state sciences 56 , springer - verlag , berlin heidelberg new york tokyo , 1985 ) . k. miyake and o. narikiyo : j. phys . soc . jpn . * 63 * ( 1994 ) 3821 . ohkawa : phys . b * 57 * ( 1998 ) 412 . mermin and h. wagner : phys . * 17 * ( 1966 ) 1133 . see , for example , _ anderson localization _ , ed . y. ngaoka and h. fukuyama ( springer series in solid - sate science 39 , springer - verlag , berlin heidelberg new york , 1982 ) . t. tohyama , p. horsch and s. maekawa : physica c * 235 * -*240 * ( 1994 ) 2231 . a. moreo , s. haas , a.w . sandvik and e. dagotto : phys . b * 51 * ( 1995 ) r12045 . wei - cheng lee , t. k. lee , chang - ming ho and p.w . leung , phys . * 91 * ( 2003 ) 057001 . in itinerant - electron ferromagnets where the density of state has a sharp peak around the chemical potential , the exchange interaction @xmath263 shows an almost @xmath8-linear dependence consistent with the curie - weiss law in a small region of @xmath362 @xcite . in itinerant - electron antiferromagnets where the nesting of the fermi surface is never significant and itinerant - electron ferromagnets where the density of states is almost constant around the chemical potential , on the other hand , the mode - mode coupling term shows an almost @xmath8-linear dependence consistent with the curie - weiss law @xcite . t. tohyama and s. maekawa : phys . b * 49 * ( 1994 ) 3596 . h. yokoyama , t. tanaka , m. ogata and h. tsuchiura : cond - mat/0308264 ( unpublished ) . a. sawa , m. kawasaki , h. takagi and y. tokura : phys . b * 66 * ( 2002 ) 014531 . h. alloul , p. mendels , h. casalta , j.f . marucco and j. arabski : phys . * 67 * ( 1991 ) 3140 . in eqs . ( [ eqtk2 ] ) and ( [ eqtkdef ] ) , we should use a multiplication factor slightly different from @xmath292 because @xmath110 appearing there is a purely single - site property . in the first term of eq . ( [ eqreplace ] ) , we should also use another one slightly different from @xmath292 because no @xmath110 appears there . however , we assume the same @xmath292 in every equation for the sake of simplicity ; this corresponds to an approximation where only the renormalization of the mass - enhancement factor is considered . k. kumagai , i. watanabe , h. aoki , y. nakamura , t. kimura , y. nakamichi and h. nakajima : physica * 148*b ( 1987 ) 480 . the dependence of @xmath7 on @xmath130 should be included in order to explain more precisely the dependence of @xmath13 on @xmath130 . h. shiba : prog . * 54 * ( 1975 ) 967 .
a kondo - lattice theory is applied to the @xmath0-@xmath1 model on a quasi - two dimensional lattice . the kondo temperature @xmath2 or @xmath3 is defined as an energy scale of local quantum spin fluctuations , with @xmath4 the boltzmann constant . the bandwidth @xmath5 of quasiparticles , which is approximately given by @xmath6 , is renormalized by the fock - type term of the superexchange interaction . when the lifetime width @xmath7 of quasiparticles is so small and temperature @xmath8 is so low that @xmath9 and @xmath10 , the renormalization is large . even in the limit of the half filling , for example , the bandwidth @xmath5 is nonzero and of the order of @xmath11 , with @xmath1 the superexchange interaction constant between nearest neighbors . the kondo temperature @xmath3 , which also gives a measure of the strength of the quenching of magnetic moments by local quantum spin fluctuations , increases away from the half filing ; it is smaller for larger @xmath7 . therefore , local - moment magnetism , which is characterized by @xmath12 , with @xmath13 the nel temperature , appears for almost half fillings , and itinerant - electron magnetism , which is characterized by @xmath14 , appears for fillings away from the half filling ; @xmath13 is higher for larger @xmath7 . the antiferromagnetic region as a function of electron fillings is wider for larger @xmath7 ; this result implies that it is wider in more disordered systems than it is in less disordered ones . the difference or asymmetry of disorder between electron - doped and hole - doped cuprate oxide superconductors must be , at least partly , responsible for that of antiferromagnetic phases between them .
the large hadron collider ( lhc ) at cern will soon provide a great opportunity for exploring physics at the tev scale . as a proton - proton collider with a total center - of - mass energy of 14 tev , the lhc is truly a quantum chromodynamics ( qcd ) machine . we therefore wish to study color exotics , since any accessible new physics in the strong interaction sector will appear in the early stages of lhc operation . many models of physics beyond the standard model ( sm ) naturally require the presence of color exotics , such as gluinos and squarks in supersymmetric extensions of standard model , @xmath9-gluons and @xmath9-quarks in extra dimensional models , or the top - prime in little higgs or twin higgs models . all of these are either quark or gluon partners which belong to the fundamental and adjoint representations of the qcd gauge group @xmath10 respectively . it is natural to consider colored particles in other representations ; in this paper , we focus on a scalar which is in the sextet(6 ) representation of @xmath10 . color sextet particles have been widely discussed in nuclear physics as diquark condensate states ; in the ssc era , sextet quarks were discussed in @xcite . color sextet scalars are naturally present in partial unification @xcite , grand unification @xcite and composite models ; in some cases they may be present around the weak scale . for instance , in a supersymmetric pati - salam @xmath11 model , light color sextet scalars can be realized around the weak scale , even though the scale of @xmath12 symmetry breaking is around @xmath13 gev due to existence of accidental symmetries with the masses of color sextet higgs arising only through high dimension operators @xcite . in this case , the introduction of a color sextet higgs will not lead to proton decay but only to neutron - anti - neutron ( @xmath14 ) oscillation and is fully compatible with present limits @xcite . in a similar framework , light color sextet scalars also help in post - sphaleron baryogenesis @xcite . in this paper , however , we will take a purely phenomenological approach toward the sextet scalar without assuming any model a priori . among all the color exotics , the color sextet scalar is unique in its coupling to quarks . in group theory language , the color sextet lies in @xmath15 as a symmetric @xmath16 rank tensor under @xmath10 . the lorentz structure for this scalar coupling to quarks is given by @xmath17 , where @xmath18 is a dirac spinor and @xmath19 is the scalar . under the sm gauge group @xmath20 , the sextet scalar can be @xmath21 , a @xmath22 adjoint @xmath23 ; @xmath24 , a @xmath22 singlet @xmath25 ; @xmath26 , a @xmath22 singlet @xmath27 ; or @xmath28 , a @xmath22 singlet @xmath29 . the color sextet scalars are also charged under the global baryon symmetry @xmath30 and the electromagnetic symmetry @xmath31 . to avoid breaking @xmath31 , these scalar fields should not develop a non - zero vacuum expectation value . this condition removes any possibility of @xmath14 oscillation in the minimal model involving color sextet scalars . we may write down the flavor independent lagrangian of such a minimal model by only considering sm gauge invariants and keeping @xmath31 unbroken , @xmath32-m^2_\delta \text{tr}[\delta_6^\dagger \delta_6]+ f_\delta q_{l } ^t c^{-1 } \tau_2 \delta^\dagger_{6 } q_{l}\nonumber\\ & + & ( d_\mu \phi_6)^\dagger ( d^\mu \phi_6)- m^2_\phi \phi_6^\dagger\phi_6 + f_\phi u^t_r c^{-1 } u_r \phi^\dagger_6\nonumber\\ & + & ( d_\mu \phi_6)^\dagger ( d^\mu \phi_6)- m^2_{\phi } { \phi_6}^\dagger\phi_6 + f_\phi d^t_r c^{-1 } d_r { \phi}^\dagger_6\nonumber\\ & + & ( d_\mu \delta_6)^\dagger ( d^\mu \delta_6)- m^2_{\delta_6 } { \delta_6}^\dagger\delta_6 + f_\delta d^t_r c^{-1 } u_r \delta^\dagger_6\nonumber\\ & - & \lambda_\delta(\text{tr}[\delta_6^\dagger \delta_6])^2-\lambda_\phi(\phi_6^\dagger\phi_6)^2- \lambda_\phi({\phi_6}^\dagger\phi_6)^2-\lambda_\delta ( { \delta_6}^\dagger\delta_6)^2\nonumber\\ & -&\lambda^\prime_\delta\text{tr}[\delta_6^\dagger \delta_6 \delta_6^\dagger \delta_6]-\text{tr}[\delta_6^\dagger \delta_6](\lambda_1{\phi_6}^\dagger\phi_6+\lambda_2{\phi_6}^\dagger\phi_6+\lambda_3 { \delta_6}^\dagger\delta_6)\nonumber\\ & -&\lambda_4 { \phi_6}^\dagger\phi_6 { \phi_6}^\dagger\phi_6 -\lambda_5 { \phi_6}^\dagger\phi_6 { \delta_6}^\dagger\delta_6 -\lambda_6 { \phi_6}^\dagger\phi_6 { \delta_6}^\dagger\delta_6~,\end{aligned}\ ] ] where the qcd covariant derivative is defined as @xmath33 , and the @xmath34 are the representation matrices for the sextet ; @xmath35 , @xmath36 and @xmath37 are all positive - definite model parameters . if we consider the @xmath22 adjoint sextet scalar @xmath21 , there will be three physical sextet scalar states that couple to up - type quark pairs , down - type quark pairs , and up - down type quark pairs . when the sextet scalar decays into light quark states , the existing search strategies for massive octet scalars or vectors @xcite may be employed . @xmath21 , @xmath24 and @xmath28 may all contribute to the single top plus jet singal and @xmath38 signal from pair production . here we consider the scenario in which a color sextet scalar decays into a top quark pair so that one can use the leptons from the top quark decay to determine the features of the sextet . signature that contains multi - top final states has been discussed in the context of many new physics models as resonance decaying into top quarks or top composite@xcite . to illustrate this and simplify our search , our study will focus on the color sextet @xmath39 singlet scalar @xmath24 that only couples to righthanded up - type quarks . the decay of the @xmath24 depends on its mass , @xmath40 , and its couplings to quarks , @xmath41 . to illustrate our reconstruction algorithm in the discussion of discovery , we consider the case where @xmath42 gev and the @xmath24 decays into two onshell top quarks ; other mass ranges are discussed in the conclusion section . above threshold , the general expression for the decay partial widths of the sextet scalar are @xmath43 where @xmath44 and @xmath45 . by far , the most stringent bounds on these parameters come from @xmath46 mixing , to which @xmath24 would make a tree level contribution proportional to @xmath47 . the off - diagonal coupling @xmath48 will contribute to flavor violation processes , for instance @xmath49 which is proportional to @xmath50 . the current bounds require that f_11 f_22 10 ^ -6 ; f_11 f_12 10 ^ -2 , for @xmath40 of a few hundred gev to tev mass range @xcite . one will also expect less stringent constraints from one loop process as @xmath51 . to escape from the bound , for acessible values of @xmath40 we expect at least one of the couplings , @xmath52 or @xmath53 , to be negligible . however from our purely phenomenological perspective , we take the decay branching fraction @xmath54 to be a completely free parameter whose value may be determined at the lhc . because the sextet is a colored object , we need to consider the possibility of it hadronizing before decaying . for example , it may form a tetraquark - like bound state with @xmath55 , such as @xmath56 , @xmath57 , or @xmath58 , with charges @xmath59 respectively . if the total width is less than @xmath60 , then the colored object will hadronize before it decays . to determine the constraint imposed by the possibility of hadronization , in fig . [ decay ] we plot the contour for which decay width of @xmath24 is equal to @xmath61 as a function of the couplings and the mass . setting @xmath62 , @xmath63 , and eliminating any coupling to @xmath64 , we see the possibility that a large portion of our parameter space will be protected from the risk of hadronization . in the mass and coupling plane.,width=302 ] because it carries color , @xmath24 can be produced directly through the qcd strong interaction at the lhc . the pair of @xmath65 is produced from gluon - gluon fusion or @xmath2 annihilation : @xmath66 the total production cross section depends only on the mass of @xmath24 , since the vertex is just the strong coupling , @xmath67 , as shown in eq . ( [ eq : feynman_rule ] ) . by comparison , the electroweak production of @xmath65 is small enough to be neglected in our search . from the scalar qcd gauge interaction ( d__6)^(d^_6 ) , g^a_t^a , one may obtain the feynman rules @xmath68 the momenta are assigned according to @xmath69 and all momenta are out - going . in group theory langauge , this is @xmath70 . the parton level cross sections for color a sextet pair production are given by ( q|q|_6_6)= c(3)c(r)d_8d_3 ^ 2 ^2_s 3 s ^3 = 1027 s^2_s ^3 and @xmath71\nonumber\\ & = & { 5\pi\over 96 s } \alpha^2_s[\beta(89 - 55\beta^2)+\text{ln } |{\beta+1\over \beta-1}|(11\beta^4 + 18 \beta^2 -29)]~,\end{aligned}\ ] ] where @xmath72 is the total energy , @xmath73 and @xmath74 is @xmath75 with the normalization factor @xmath76 and casimir @xmath77 satisfying [ t^a_r t^b_r]=c(r)^ab t^a_r t^a_r = c_2(r)1 . we list the values for different representations under @xmath78 as in table [ tab : grp1 ] . .normalization factor @xmath79 and quadratic casimir @xmath80 for @xmath81 under @xmath78 . [ cols="^,^,^,^ " , ] the qcd production cross sections for the color sextet scalar pair @xmath65 at both lhc and tevatron are plotted in fig . [ total ] with factorization scale @xmath82 , renormalization scale @xmath83 and the cteq6l @xcite parton distribution function ( pdf ) . the matrix elements in our calculations here and elsewhere are generated by susy - madgraph @xcite with modified color factors . for comparison , we also show the pair production cross sections for @xmath10 triplet and octet scalars at the lhc . as we can see , the total production cross section of the sextet scalar is similar to that of the octet scalar , but is about one order maginitude larger than that of the triplet scalar , which can be understood from values of @xmath76 and @xmath77 for different reprensentations in table [ tab : grp1 ] . at the lhc and tevatron with @xmath82 , fixed scale @xmath84 with @xmath83 . the pdf set cteq6l has been used in all calculation . , title="fig:",width=302 ] at the lhc and tevatron with @xmath82 , fixed scale @xmath84 with @xmath83 . the pdf set cteq6l has been used in all calculation . , title="fig:",width=302 ] as discussed in the introduction , the color sextet scalar @xmath24 only couples to the righthanded up - type quark quark pair . thus we may also have single production of a @xmath24 through uu ( cc)_6 . however , the production cross section is proportional to the coupling @xmath85 and @xmath86 , and may therefore be suppressed due to the @xmath46 mixing constraint . some studies of the single @xmath24 production at the tevatron and the lhc have been done in ref . as discussed in the previous sections , the most distinct feature of the color sextet scalar is its decay mode @xmath87 , which leads to a same - sign dilepton signature in the final state if both top quarks decay semileptonically , i.e. @xmath88 . to avoid ambiguities in lepton assignments during reconstruction , we require the anti - top quark pair from the @xmath89 to decay hadronically . hence , the final state of @xmath65 is pp|_6_6tt|t|t4b+^^+_t+nj , where @xmath90 and @xmath91 to allow initial and final state qcd radiation . in our study , however , the qcd radiation is not included . to get this final state , the decay branching ratio will be = ^2(_6tt)(29)^2(23)^22 , where the situation that top quark decays hadronically and anti - top quark decays semileptonically is also included . figure [ total ] also clearly shows that a color sextet with @xmath92 gev will not be bounded by tevatron data as the same - sign dilepton plus multi - jet final state from @xmath93 will be less than one event for @xmath94 luminosity . to illustrate the kinematic features of the color sextet scalar pair , we consider the decay process @xmath95 and take @xmath96 gev . the leading and second - leading jet @xmath97 distributions are shown in fig . the typical hardness of these jets is the basis for one of our selection cuts introduced later in this section . in order simulate the detector effects on the energy - momentum measurements , we smear the electromagnetic energy and the muon momentum by a gaussian distribution whose width is parameterized as @xcite @xmath98 the jet energies are also smeared using the same gaussian formula as in eq . ( [ ecal ] ) , but with @xcite @xmath99 and @xmath100 , width=302 ] we first reconstruct the two on - shell hadronically decaying @xmath101 s . our procedure is to consider all dijet invariant masses except for those containing one of the two tagged b - jets , since we require b - tagging in the event selection discussed later , and choose the two closest @xmath102 combinations , which we then require to lie within the mass window |m_jj - m_w| < 15 . from this , we get the two reconstructed @xmath101 momenta . we then consider all combinations of reconstructed @xmath103 with all jets and again choose the two closest invariant masses @xmath104 . in this way , we reconstruct the two hadronically decaying anti - top quarks . the distributions of these reconstructed invariant masses are shown in fig . [ mbjj ] . once we have the reconstructed two anti - top quarks , the reconstruction of the sextet @xmath105 can be done using the 6-jet invariant mass @xmath106 for the two hadronic anti - top quarks . although the production of neutrinos prevents us from fully reconstructing the sextet which produces the leptonic decays , we may reconstruct the transverse mass @xmath107 for the remaining two jets plus same - sign dilepton and @xmath108 as m_t= . and @xmath107.,width=302 ] as seen in fig . [ m6j ] , our reconstruction shows a clear resonance in both the @xmath106 and @xmath107 distributions . between reconstructed top momentum and reconstructed sextet momentum . dashed(solid ) line shows the distribution without(with ) smearing effects and kinematic cuts . , width=302 ] finally , since the two anti - top quarks may be fully reconstructed , we can boost back to the rest frame of the @xmath89 and study its spin . as shown in fig . [ cos ] , the angular distribution of the anti - quark clearly shows that the @xmath89 is a scalar . since there are two missing neutrinos from decay of two top quarks , it is challenging to fully reconstruct top quark s momentum , and study the spin information of top quarks , which can be used to check this model since @xmath24 only decays into a right - handed top quark pair . we leave this for future work . we next consider the backgrounds for our signal . we require at least 2 tagged b - jets plus a same - sign dilepton and multijet . the irreducible sm background for this final state consists of @xmath109 , @xmath110 and @xmath111 . we estimate the qcd @xmath110 background by computing @xmath112 production . this is only 14 fb , and one expects the @xmath110 is about three orders lower and therefore @xmath113 fb . the sm 4-top @xmath111 is less than 0.1 fb to start with . the leading background thus comes from @xmath114 with one hadronic top decay and one semileptonic top decay with the same sign as @xmath115 leptonic decay . we propose the following selection cuts : * @xmath116 gev , @xmath117 gev , @xmath118 gev , @xmath119 * same sign dilepton with @xmath120 gev , @xmath121 * @xmath122 * at least two b - tagged jets * @xmath123 gev since the production rate of our signal only depends on the mass @xmath40 and branching ratio of @xmath24 decay to a top quark pair , we scan these two parameters to study the discovery potential . we summarize our results in fig . [ final ] as the signal production rate for @xmath124 from @xmath125 with sm @xmath114 background included . we use a factor of @xmath126 in both plots in fig . [ final ] for tagging two @xmath127 jets with @xmath128 effeciency to tag each @xmath127-jet . the sm background is taken as 1 fb in the significance contour . as we can see in the left plot of fig . [ final ] , for 100 fb@xmath8 luminosity , the statistical significance can surpass the @xmath129 level for @xmath130 if br(@xmath131 ) is about 0.5 . also note that the mass of the sextet scalar can be determined by reconstructing two hadronically decaying top ( or anti - top ) quarks , and the branching ratio of @xmath87 can be roughly estimated from the total signal event rate if the one can understand the backgorund sufficiently well . no reconstruction selection has been implented since we did not simulate the events with initial state / final state radiation and the reconstruction efficiency is thus unknown . in principle , we expect that the @xmath132 can be further improved by including reconstruction . and significance contour . dashed(solid ) curves in left plot represent production before(after ) selection cuts.,title="fig:",width=302 ] and significance contour . dashed(solid ) curves in left plot represent production before(after ) selection cuts.,title="fig:",width=302 ] in this paper , we discuss the production of a new exotic particle , a color sextet scalar , at the cern large hadron collider . taking a purely phemenological approach , we discuss the discovery of the color sextet scalar through its decay into a top - top quark pair . the unique feature of same sign dilepton plus multijet makes it easy to identify and reconstruct the color sextet scalar object . due to its large qcd production , it is possible to cover the color sextet scalar up to a mass range of 1 tev for 100 fb@xmath8 integrated luminosity . in the text , we only consider the case of @xmath133 , where @xmath24 decays into two on - shell top quarks . in the case @xmath134 , there is a possiblity of a @xmath135 decaying into a sextet pair , since @xmath24 carries a @xmath136 charge , which we expect is highly constrained by lep data . we also expect to find strong constraints from tevatron data . for example , for @xmath40 just above @xmath137 threshold , the @xmath65 signal will directly contribute to @xmath138 sample as the offshell top decay products are soft . kw would like to thank rabi mohapatra for initiating this project , carefully reading the manuscript and providing valuable suggestions . we would also like to thank qing - hong cao , kaoru hagiwara , tao han , hitoshi murayama , mihoko nojiri , chris quigg , t.t . yanagida and c.p . yuan for useful discussion . this work is supported by the world premier international research center initiative ( wpi initiative ) , mext , japan . kw is supported in part by the project of knowledge innovation program ( pkip ) of chinese academy of sciences , grant no . kjcx2.yw.w10 and would like to acknowledge the hospitality of kavli institute for theoretical physics china ( kitpc ) while the work was in progress . s. errede and s. h. h. tye , summary report of the stable / exotic particles working group on ssc , snowmass , 1984 . h. tanaka and i. watanabe , int . j. mod . phys . a * 7 * , 2679 ( 1992 ) . j. c. pati and a. salam , phys . d * 10 * , 275 ( 1974 ) [ erratum - ibid . d * 11 * , 703 ( 1975 ) ] . r. n. mohapatra and r. e. marshak , phys . lett . * 44 * , 1316 ( 1980 ) [ erratum - ibid . * 44 * , 1643 ( 1980 ) ] . p. fileviez perez , phys . b * 654 * , 189 ( 2007 ) [ arxiv : hep - ph/0702287 ] . z. chacko and r. n. mohapatra , phys . d * 59 * , 055004 ( 1999 ) [ arxiv : hep - ph/9802388 ] . r. n. mohapatra , n. okada and h. b. yu , phys . d * 77 * , 011701 ( 2008 ) [ arxiv:0709.1486 [ hep - ph ] ] . see for example : v. barger , t. han and d. g. e. walker , phys . lett . * 100 * , 031801 ( 2008 ) [ arxiv : hep - ph/0612016 ] . b. lillie , l. randall and l. t. wang , jhep * 0709 * , 074 ( 2007 ) [ arxiv : hep - ph/0701166 ] . a. l. fitzpatrick , j. kaplan , l. randall and l. t. wang , jhep * 0709 * , 013 ( 2007 ) [ arxiv : hep - ph/0701150 ] . b. lillie , j. shu and t. m. p. tait , jhep * 0804 * , 087 ( 2008 ) [ arxiv:0712.3057 [ hep - ph ] ] . t. han , arxiv:0804.3178 [ hep - ph ] . y. bai and z. han , arxiv:0809.4487 [ hep - ph ] . t. plehn and t. m. p. tait , arxiv:0810.3919 [ hep - ph ] . k. s. babu , r. n. mohapatra and s. nasri , phys . rev . lett . * 97 * , 131301 ( 2006 ) [ arxiv : hep - ph/0606144 ] . l. j. hall and a. e. nelson , phys . b * 153 * , 430 ( 1985 ) . r. s. chivukula , m. golden and e. h. simmons , nucl . b * 363 * , 83 ( 1991 ) . m. i. gresham and m. b. wise , phys . d * 76 * , 075003 ( 2007 ) [ arxiv:0706.0909 [ hep - ph ] ] . b. a. dobrescu , k. kong and r. mahbubani , arxiv:0709.2378 [ hep - ph ] . m. gerbush , t. j. khoo , d. j. phalen , a. pierce and d. tucker - smith , phys . d * 77 * , 095003 ( 2008 ) [ arxiv:0710.3133 [ hep - ph ] ] . c. kilic , t. okui and r. sundrum , jhep * 0807 * , 038 ( 2008 ) [ arxiv:0802.2568 [ hep - ph ] ] . m. v. martynov and a. d. smirnov , arxiv:0807.4486 [ hep - ph ] . fileviez perez , r. gavin , t. mcelmurry and f. petriello , arxiv:0809.2106 [ hep - ph ] . c. kilic , s. schumann and m. son , arxiv:0810.5542 [ hep - ph ] . j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) [ arxiv : hep - ph/0201195 ] . g. c. cho , k. hagiwara , j. kanzaki , t. plehn , d. rainwater and t. stelzer , phys . d * 73 * , 054002 ( 2006 ) [ arxiv : hep - ph/0601063 ] .
taking a phenomenological approach , we study a color sextet scalar at the lhc . we focus on the qcd production of a color sextet pair @xmath0 through @xmath1 fusion and @xmath2 annihilation . its unique coupling to @xmath3 allows the color sextet scalar to decay into same - sign diquark states , such as @xmath4 . we propose a new reconstruction in the multijet plus same sign dilepton with missing transverse energy samples ( @xmath5 , @xmath6 ) to search for on - shell @xmath7 final states from sextet scalar pair production . thanks to the large qcd production , the search covers the sextet mass range up to 1 tev for 100 fb@xmath8 integrated luminosity .
in order to measure @xmath7 , observed neutrino events must be separated into charged current ( cc ) and neutral current ( nc ) categories . both cc and nc neutrino interactions initiate a cascade of hadrons in the target that is registered in both the scintillation counters and drift chambers . however , muon neutrino cc events are distinguished by the presence of a final state muon , which typically penetrates well beyond the hadronic shower and deposits energy in a large number of consecutive scintillation counters . these differing event topologies enable the statistical separation of cc and nc interactions based solely on event length ( _ i.e. _ , on the presence or absence of a muon in an event ) . events with a long length ( spanning more than 20 counters ) are identified as cc candidates ; those with a short length ( spanning less than 20 counters ) as nc candidates . the experimental quantity measured in both neutrino and antineutrino modes is the ratio : @xmath13 the ratios of short to long events ( @xmath14 ) measured in the nutev data are 0.4198 @xmath15 0.0008 in the neutrino beam and 0.4215 @xmath15 0.0017 in the antineutrino beam . a standard model value of @xmath7 can be directly extracted from these measured ratios by using a detailed monte carlo simulation of the experiment . the monte carlo must include the integrated neutrino fluxes , the neutrino cross section , and a detailed description of the nutev detector . more detailed information on the specific components of the monte carlo simulation can be found elsewhere @xcite . using our separate high - purity neutrino and antineutrino data sets , nutev measures the following linear combination of @xmath16 and @xmath17 : @xmath18 the value for _ x _ is selected using the monte carlo in order to minimize uncertainties related to charm quark production . the single remaining free parameter in the monte carlo , @xmath7 , is then varied until the model calculation of @xmath19 agrees with what is measured in the data . the preliminary result from the nutev data sample for @xmath20=175 gev and @xmath21=150 gev is : @xmath22 having chosen the convention @xmath23 , and given the well - determined z mass from lep , our result implies : @xmath24 our measurement is in good agreement with standard model expectations , and is consistent with current measurements from w and z production as well as from other neutrino experiments ( figures [ fig : mw ] , [ fig : mt - mw ] ) . the data tend to collectively favor a light higgs mass . the central value from recent global fits to all precision data is @xmath25 gev with an upper bound of @xmath26 gev at 95% cl @xcite . outside the standard model , deviations between electroweak measurements in @xmath27 scattering and those from other processes are sensitive to new physics . in these proceedings we discuss two such possibilities : neutrino oscillations and extra neutral vector gauge bosons . + the presence of neutrino oscillations will directly shift our measured ratios , @xmath28 , from their standard model predictions . since @xmath29 s oscillating to either @xmath30 s or @xmath31 s would be less likely to produce a final state muon , we would expect to observe an excess of short events . since no such excess is observed , single mode ( @xmath32 or @xmath33 ) limits can be set for both @xmath34 and @xmath35 oscillations . one advantage to this type of search is that the paschos - wolfenstein quantity , r@xmath11 , is particularly sensitive to cp - violating oscillations because it is formed from a difference in neutrino and antineutrino rates . as a result , nutev is presently the only experiment with direct limits on @xmath36 . more details on this analysis can be found elsewhere @xcite . + extra z bosons ( z@xmath37 ) are of interest not only because they are predicted by many grand unified theories and superstring models , but also in light of recent experimental developments . erler and langacker have shown in a recent global fit that the precision electroweak data are better described if an extra tev - scale z boson is included @xcite . of course , a large portion of this improvement arises from the fact that the 2.5 @xmath38 deviation of the new atomic parity violation ( apv ) measurement @xcite from the standard model prediction can be explained by including an additional z boson @xcite . in our case , extra z bosons would manifest themselves as shifts in the neutrino - quark couplings away from their standard model values . these shifts can arise from both pure - z@xmath37 exchange as well as from z - z@xmath37 mixing contributions . if we consider constraints on extra z bosons in @xmath39 models , then these coupling shifts are well - determined @xcite . in this case , the lightest extra z boson is a linear combination of the so(10 ) singlet @xmath40 and the su(5 ) singlet @xmath41 : @xmath42 expressed in terms of a free parameter @xmath43 . our data tend to disfavor the inclusion of additional z bosons , so we set a 95% cl lower limit on the mass of such an extra @xmath44 plotted as a function of @xmath43 ( figure [ fig : zprime1 ] ) . limits are displayed for two cases : the @xmath44 has no mixing with the standard model z , and the more realistic case that allows for some level of mixing . for the latter , we input the z pole data constraint that the level of z-@xmath44 mixing is small ( @xmath45 ) @xcite . note that the inclusion of z-@xmath44 mixing , even at this small level , weakens the limits . at 95% cl and assuming a 10@xmath46 level of z-@xmath44 mixing , we set lower mass limits of 675 and 380 gev for the @xmath40 and @xmath47 respectively . direct searches have already excluded masses below @xmath12 600 gev @xcite . as can be seen from figure [ fig : zprime1 ] , our maximum sensitivity is to the @xmath40 , and our exclusion peaks in the most viable region suggested by the apv data . figure [ fig : zprime2 ] shows a direct comparison of the region nutev excludes in z@xmath48 mass-@xmath43 space to the favored central value from the apv data analyses @xcite . nutev has successfully completed its data taking and has extracted a value of @xmath7 . the precision of this result represents a factor of two improvement over previous measurements in @xmath27 scattering , because of reduced uncertainties associated with measuring the paschos - wolfenstein ratio , @xmath19 . interpreted within the framework of the standard model , this result is equivalent to a determination of the w mass and is consistent with direct measurements of @xmath49 . outside the standard model , this measurement can be used to set limits on neutrino oscillations as well as extra neutral vector gauge bosons .
the nutev experiment at fermilab presents a determination of the electroweak mixing angle . high purity , large statistics samples of @xmath0 and @xmath1 events allow the use of the paschos - wolfenstein relation , a technique which considerably reduces systematic errors associated with charm production and other sources . within the standard model , this measurement of @xmath2 indirectly determines the w boson mass to a precision comparable to direct measurements from high energy @xmath3 and @xmath4 colliders . nutev measures @xmath5 , which implies @xmath6 gev . outside the standard model , this result can be used to explore the possibility of new physics ; in particular , we present limits on both neutrino oscillations and the presence of extra neutral vector gauge bosons . in deep inelastic neutrino - nucleon scattering , the weak mixing angle can be extracted from the ratio of neutral current ( nc ) to charged current ( cc ) total cross sections . a method for determining @xmath7 that is much less dependent on sources of model uncertainty and the the details of charm production ( the largest source of uncertainty in the previous neutrino measurement @xcite ) employs the paschos - wolfenstein relation @xcite : @xmath8 where @xmath9 , and @xmath10 . unfortunately , the substantially reduced uncertainties come at a price : r@xmath11 is a more difficult quantity to measure experimentally because neutral current neutrino and antineutrino events have identical observed final states . the two samples can only be separated by knowing the incoming neutrino beam type . + high - purity neutrino and antineutrino beams were provided by the sign selected quadrupole train ( ssqt ) at the fermilab tevatron during the 1996 - 1997 fixed target run . neutrinos are produced from the decay of pions and kaons resulting from interactions of 800 gev protons in a beo target . dipole magnets immediately downstream of the proton target bend pions and kaons of specified charge in the direction of the nutev detector , while wrong - sign and neutral mesons are stopped in beam dumps . the resulting beam is almost purely neutrino or antineutrino depending on the selected sign of the parent mesons ( opposite particle contamination is @xmath120.1% ) . in addition , the beam is almost purely muon neutrinos with a small ( @xmath121% ) contamination of electron neutrinos . neutrino interactions are then observed in the nutev detector @xcite , which is located approximately 1.5 km downstream of the proton target . the detector consists of an 18 m long , 690 ton steel - scintillator target followed by an instrumented iron - toroid spectrometer . the target calorimeter is composed of 168 3 m x 3 m x 5.1 cm steel plates interspersed with liquid scintillation counters and drift chambers . the scintillation counters provide triggering information as well as a determination of the longitudinal event vertex , event length and visible energy deposition . the mean position of hits in the drift chambers help establish the transverse event vertex . the toroidal spectrometer , which determines muon sign and momentum , is not directly used for this analysis . in addition , because the detector was continuously calibrated through exposure to a wide energy range of test beam hadrons , electrons and muons , many systematics related to detector effects were substantially reduced .
core collapse supernovae play important roles , for example , in star formations , evolutions of galaxy and the entire universe , and accelerations of cosmic ray particles owing to its nucleosynthesis , energetic shock waves and high luminosity . recently the phenomena are put in the spotlight because the gamma - ray bursts are likely to be associated with them . unfortunately , however , the mechanism of supernova explosion is still unresolved [ e.g. @xcite for reviews ] . the supernova explosions are commenced by gravitational collapse of massive stars at the end of their lives . when the nuclear saturation density is reached at the center of core , the bounce occurs , producing a shock wave that starts to propagate outwards in the core . it is almost a consensus among the researchers that this shock wave does not yield the supernova explosion directly , since the shock loses energy by photo - dissociations of nuclei as well as neutrino - cooling and stalls somewhere inside the core . it is widely expected that the shock will be revived by the irradiation of neutrinos diffusing out of the proto - neutron star , the scenario originally proposed by @xcite . although many researchers have studied this scenario intensively and extensively , the state - of - the - art numerical simulations have shown so far that , as long as collapse is spherically symmetric , the stalled shocks do not revive @xcite . these simulations are concerned with the dynamical evolution of the entire core from the onset of the collapse till long after the stagnation of the shock . on the contrary , @xcite took a different approach , focusing on the revival of the stalled shock wave . after the shock is stagnated due to the energy losses mentioned above , the accretion flows through the standing shock wave are quasi - steady ( see @xcite for the criticism on this point ) . they approximated these flows with steady state solutions for constant mass accretion rates and neutrino luminosities . varying these constant values , they found that for a given mass accretion rate there is a critical neutrino luminosity , above which there exists no steady solutions . using this fact , they argued that the revival of stalled shock occurs when the neutrino luminosity exceeds this critical value . the existence of the critical luminosity was also observed in the numerical experiments done by @xcite , where they artificially increased neutrino luminosities in their failed explosion models and saw what happened to the stalled shocks . merits of these steady state calculations are not only that they are computationally simpler but also that they can clearly demonstrate , if qualitatively , what is the cause of the failure of explosions . that is , according to this theory , an inappropriate combination of the neutrino luminosity and the mass accretion rate . so far , we have been talking about the studies of spherically symmetric collapse . however , several observations suggest that the supernova explosions are intrinsically asymmetric in general . for example , @xcite and @xcite observed a few percent of linear polarizations for photons from collapse - driven supernovae and argued that the stellar envelopes are globally asymmetric , oblate or prolate with an aspect ration of @xmath2 . we also know that supernova remnants are asymmetric in general . we can easily imagine a couple of possible causes for this asymmetry , e.g. hydrodynamical instabilities , rotation and magnetic fields . in fact , young pulsars are thought to have rather large angular momenta @xcite or magnetic fields @xcite , which are probably taken over from the pre - collapse stars . it is also believed that some convective motions occur in the supernova core . the large scale non - spherical oscillations of the standing accretion shocks may also cause the stellar envelope globally asymmetric@xcite . among these effects , we pay particular attention to the stellar rotation in this paper . note , however , that the rotation of the core may not be so rapid after all . recent evolution models by heger et al . @xcite suggest that the transport of angular momentum during the quasi - static evolutionary phase of the progenitor deprives the core of substantial fraction of its angular momentum , particularly when the magnetic torque is taken into account @xcite . if this is really the case , the rotation will play no significant role in dynamics of core - collapse as shown by @xcite ( see also @xcite ) . we had better bear in mind , however , that the evolution models are based on 1d calculations and have some uncertainties in the mechanism and treatment of angular momentum transport . the dynamics of rotational collapse has been explored numerically by some authors . for example , @xcite and @xcite demonstrated that rapid rotation tends to weaken the prompt shock wave because the centrifugal force prevents the core from contracting sufficiently . @xcite found that the efficiency of energy transfer by convection was reduced by the decrease of effective gravity near the equatorial plane due to the centrifugal force and , as a result , the shock did not propagate as much as in the non - rotation case . they also asserted that the resultant anisotropy of energy transfer would have made explosion jet - like if it had been successful . @xcite estimated the anisotropy of neutrino radiation from the rotational cores . they claimed that the deformed neutrino - sphere , oblate in general , yields the neutrino flux and the heating rate that are larger near the rotation axis than near the equatorial plane . given the results by @xcite and the results by @xcite that such anisotropic neutrino irradiation could lead to successful jet - like explosion , we may well expect that the rotation is helpful for the explosion in the neutrino heating scenario . in order to see the effect of rotation more clearly , if qualitatively , we followed the lead by @xcite . we solved time - independent euler equations describing steady accretion flows through a standing shock , taking rotation into account . we paid particular attention to how the critical luminosity of neutrinos is altered by the stellar rotation and demonstrated that the rotation assist the revival of the stalled shock . this paper is organized as follows . in the next section , we discuss the spherically symmetric case . we find two branches of steady solutions and discuss the revival of the shock wave employing the critical luminosity . we examine in section 3 the stability of the two branches of solutions obtained in section 2 . in section 4 , we analyze the steady axisymmetric solutions with rotation and discuss the effect of rotation on the revival of the stalled shock . the final section is devoted to summary and some more discussion . in this and next sections , we discuss spherically symmetric flows , solving one dimensional hydrodynamical equations . the calculations in this section are similar to those of @xcite . after the shock is stalled , mass accretion rates or neutrino luminosities change rather slowly and the flows become quasi - steady . idealizing this situation , we seek in this section steady solutions of the accretion flow . since our purpose of this paper is not to determine quantitatively the critical luminosity of neutrinos , we make following assumptions and simplifications : i ) we do not consider the flows outside the shock wave in detail , simply assuming a free fall , and restrict our calculations to the inside of the shock surface . ii ) newtonian formulations are adopted because the region of our interest is outside of the neutron star and general relativistic effects are negligible there . iii ) we do not solve neutrino transfer equations assuming that the luminosity and energy of the neutrinos are independent of radius . it is also assumed that the neutrinos are thermal with a fixed value of temperature ( @xmath3 ) , for simplicity . iv ) a simplified equation of state and rates of neutrino heating and cooling are employed . v ) photo dissociations of nuclei , magnetic fields and convections are neglected , for simplicity , although there should exist regions where flows are convectively unstable . then , the basic equations describing steady spherically symmetric accretion flows are given in the spherical coordinate as , @xmath4 @xmath5 @xmath6 where @xmath7 , @xmath8 , @xmath9 , @xmath10 denote radial velocity , density , pressure and specific internal energy , respectively ; @xmath11 , @xmath12 , @xmath13 , @xmath14 are mass accretion rate , gravitational constant , neutron star mass and net heating rate by neutrino irradiation , respectively . as for the neutron star mass , we adopt take @xmath15 . pressure and specific internal energy of nucleons , photons , and relativistic particles are approximately written as @xmath16 @xmath17 where @xmath18 , @xmath19 , @xmath20 , @xmath21 are boltzmann constant , speed of light , planck constant , and nucleon mass , respectively ; @xmath22 is temperature of matter and photons . as for the heating and cooling processes , we take into account only the absorption and emission of neutrinos by nucleons . we adopt the formulae given in @xcite for degeneration factor @xmath23 ; that is @xmath24 where @xmath25 is the luminosity of electron neutrinos , which is assumed to be equal to that of anti - electron neutrinos . the outer and inner boundaries of calculations are set at the shock surface and the neutrino - sphere , respectively . we impose the rankine - hugoniot relations at the shock surface . the flow into the shock is assumed to be a free fall with the velocity of @xmath26 and the temperature of @xmath27 . thus , the outer boundary conditions are written as @xmath28 @xmath29 where @xmath30 is upstream density at the shock surface . the continuity equation is automatically satisfied by imposing equation ( [ eq1 ] ) on each side of the shock . we impose the condition that the density be @xmath31 @xmath32 at the inner boundary ; this is approximately corresponding to the condition that the optical depth from the neutrino - sphere to infinity equals to @xmath33 . the latter condition was adopted by @xcite . we note that the position of the inner boundary is determined by the relation @xmath34 where @xmath35 is the stefan - boltzmann constant and @xmath36 is the radius of the neutrino - sphere , that is , the inner boundary . on the other hand , the position of the outer boundary is determined so as to satisfy the above jump conditions . thus , the problem is regarded as an eigen - value problem . we solve these equations for a wide range of values of the mass accretion rate and the neutrino luminosity . the results of the calculations are shown in figures [ fig1 ] and [ fig2 ] . in figure [ fig1 ] , we can see there are two solutions for a given mass accretion rate and a neutrino luminosity when the neutrino luminosity is below a certain critical value . the shock radii of these two solutions differ . from now on , we refer to the solution with a smaller shock radius as the inner solution and to the other solution as the outer solution . as shown in figure [ fig2 ] , as the neutrino luminosity is raised with the mass accretion rate fixed , the shock radius for the inner solutions becomes larger , whereas that for the outer solutions becomes smaller . two solutions coincide with each other when the luminosity reaches the critical value . for the luminosity over the critical value , there is no solution . this critical value of neutrino luminosity is shown in figure [ fig8 ] . note that when the luminosity is close to the critical value , the shock radii are insensitive to the luminosity . although our formulation is different in detail from burrows & goshys , the results are in good agreement [ cf . figure 1 of @xcite ] . as mentioned , the shock radius for the critical luminosity is almost independent of the mass accretion rate , and is about @xmath37 km according to our calculations . this behavior of the solutions is understood if one remind of the adiabatic subsonic bondi - hoyle flow ( cf . @xcite ) . here , we introduce @xmath38 , as the post shock velocity that we would obtain if the shock radius were @xmath39 . the velocity @xmath38 is shown as dotted lines in figure [ fig1 ] . the subsonic adiabatic flows accelerate outside a certain radius , whereas they decelerate inside that radius . the critical radius is determined by the balance of the gravitational energy and thermal energy and is close to the sonic point in the transonic accretion flow . on the other hand , the free fall velocity decreases monotonically . furthermore , the acceleration of the adiabatic flow is larger than that of the free fall outside the critical radius owing to the pressure gradient force . therefore , if any , there are two radii where the accretion flow joins with the free fall stream via the rankine - hugoniot relations . in other words , the solution curve of the adiabatic accretion flow @xmath40 intersects with @xmath38 at two points . although the solution curves are somewhat modified if one takes account of the neutrino irradiation , this behavior is not altered , for the amplitudes of neutrino luminosity of our interest . the irradiation tends to heat and push out the infalling matter . hence , as the luminosity gets larger , the infall velocity gets smaller in the whole range and the solution curves go down . when the luminosity exceeds a critical value , there exists no point where the subsonic accretion flow joins with the free fall stream , via the rankine - hugoniot relations ; the solution curves of the accretion flow @xmath40 never intersects with @xmath38 . from the above explanation , we can also understand another important nature of the solutions ; as the luminosity is raised with the mass accretion rate fixed , the radial derivative of the velocity of the accretion flow at the shock radius @xmath41 becomes larger for the inner solutions , whereas it becomes smaller for the outer solutions . if the flow is adiabatic , these derivatives coincide with that of @xmath42 at the shock radius for the critical luminosity . the neutrino heating , however , makes the former slightly smaller than the latter . this characteristics will be utilized in section 4 . finally , let us discuss what happens when the luminosity exceeds the critical value . since there is no steady solution , we should discuss this issue with time - dependent calculations . however , we can infer at least the initial response of the shock with a simpler discussion . we can find in figure [ fig1 ] that when the luminosity exceeds the critical value , the downstream velocity at the shock wave becomes inevitably smaller than @xmath38 . in addition , the downstream entropy should be larger than that for the critical luminosity , because of the higher luminosity . we can regard this situation as a riemann problem that has a discontinuity with a velocity slightly lower and an entropy slightly larger in the downstream than those satisfying the rankine - hugoniot relations . the solution of the riemann problem tells us that , a shock wave will propagate outwards , while a weak rarefaction wave or another weak shock wave will propagate inwards at the same time @xcite ; in our words , the shock wave revives . thus , we can conclude that the condition for the shock to revive is that the neutrino luminosity exceeds the critical value . in the previous section , we found that there are two branches of steady solutions for the accretion flow with a shock . in this section , we discuss which type of the solutions is realized , examining the stability of these solutions by global linear analysis . we consider the perturbations of neutrino luminosity , that is , non - adiabatic effects are taken into account . in this paper , only the stability against radial perturbations is considered , this is because we did not take account of the effect of convection in obtaining the unperturbed flows in the previous section . there is , in general , a region where heating prevails over cooling under the irradiation of neutrinos ; in this region , entropy increases in the direction of gravity and the flow is convectively unstable . thus our solutions are convectively unstable against non - radial perturbations , which fact is of no interest to us here , though . since we consider the perturbations accompanying variations of the positions of inner and outer boundaries , it is convenient to introduce a new variable @xmath43 , defined as , @xmath44 where @xmath45 denotes the shock radius . then , the basic equations describing the time - dependent flows are written as , @xmath46 @xmath47 @xmath48 since we assume that the flow outside the shock is steady , the rankine - hugoniot relations at the outer boundary are expressed as , @xmath49 @xmath50 @xmath51 where @xmath52 is the shock velocity , which can be written as @xmath53 in calculating the perturbations of heating and cooling rates , we assume , for simplicity , that the effective temperature of neutrinos is not perturbed . we perform a linear analysis , assuming that all the perturbed quantities have the time - dependence of @xmath54 . in the rest of this section , the suffix @xmath55 is attached to the perturbed quantities and the suffix @xmath27 to the unperturbed ones . then , the equations describing the perturbations become @xmath56 @xmath57 @xmath58 where , @xmath59 , @xmath60 and @xmath61 are defined as @xmath62 @xmath63 @xmath64 all the variables , @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 and an eigen - value @xmath71 can be complex . the outer boundary conditions ( @xmath72 ) become @xmath73 @xmath74 @xmath75 where @xmath76 , @xmath77 are density and velocity of the free - fall at the shock in the unperturbed state . since we set the neutrino - sphere as the inner boundary ( @xmath78 ) and the density is fixed there by definition , we impose the condition @xmath79 at the inner boundary . in addition , we assume @xmath80 at the inner boundary , supposing that the infalling matter comes to rest on the neutron star . since the neutrino temperature is assumed to be unperturbed , the variation of the luminosity is related with that of the neutrino - sphere radius as , @xmath81 above equations are solved as follows . we first take trial values of @xmath71 , @xmath69 , ( @xmath68 is set to be unity ) and solve the conditions ( [ eq24 ] ) , ( [ eq25 ] ) , ( [ eq26 ] ) , and equations ( [ eq18 ] ) , ( [ eq19 ] ) , ( [ eq20 ] ) from the outer boundary to the inner boundary . the solution obtained this way does not , in general , meet the inner boundary conditions ( [ eq27a ] ) , ( [ eq27b ] ) . then we improve the value of @xmath71 , @xmath69 and repeat the procedure until ( [ eq27a ] ) , ( [ eq27b ] ) are satisfied at the inner boundary . we found for a given unperturbed solution many modes with complex eigen - values ( probably an infinite number of overtones , in fact ) and only one mode with a real eigen - value . all modes with complex eigen - values are damped , namely , the real part of the eigen - value is negative , owing to the thermal smearing effect of the neutrino irradiation . we show only the results for the mode with a real eigen - value . the growth rates are shown in figure [ fig3 ] . we can see that the inner solutions are always stable , while the outer solutions are always unstable . further , when the luminosity reaches the critical value , both solutions become neutral . the eigen - functions are shown in figure [ fig4 ] . the ratios of the perturbation of luminosity to that of shock radius are shown in figure [ fig5 ] . since the amplitude of luminosity perturbation is twice as much as that of neutrino - sphere radius , we can see that the perturbations of inner and outer boundaries are in phase for the inner solutions , whereas they are in opposite phase for the outer solutions . we should note again that we investigated the stability only against radial perturbations . as several authors pointed out , spherical accretion shocks are , in general , unstable against non - radial perturbations @xcite . in this sense , we can not say that the inner solutions are stable . detailed studies of the stability against non - radial perturbations will be necessary with the effect of convection and neutrino irradiation properly taken into account . in this section , we discuss the effects of rotation on the revival of shock . we extend the one dimensional analysis in section 2 to the two dimensional case . a brief summary of our results in this section has previously appeared @xcite . since as we found in the previous section , the radii of the outer solutions are unstable , we consider only the inner solutions in this section . the assumptions and simplifications mentioned in section 2 are also adopted here except for the assumptions for the flows outside the shock , which will be mentioned later . assuming axial symmetry , we use the spherical coordinates . then the basic equations describing the steady flow inside the shock are written as , @xmath82 @xmath83 @xmath84 @xmath85 @xmath86 where @xmath87 denotes velocity and the other notations are the same as those in section 2 . since we seek solutions with axial and equatorial symmetry , we impose the following conditions , @xmath88 at @xmath89 , and @xmath90 at @xmath91 . the outer and inner boundaries are set at the shock surface and neutrino - sphere , respectively . the neutrino - sphere will be oblate due to centrifugal force and , as a result , the neutrino flux will be anisotropic @xcite . such effect may have some influence on the explosion @xcite . since our concern here is the hydrodynamical effects of rotation on the accretion flow , we assume that the neutrino - sphere is spherical and that the neutrino flux and temperature are isotropic . unlike in section 2 , the shock surface is not spherical and oblique to the flow , in general . as in the one dimensional calculations , we impose the condition that the density is @xmath31 @xmath32 , at the inner boundary . the rankine - hugoniot relations for oblique shocks are imposed at the outer boundary . as for the flows outside the shock , we take the following approximations : i ) the flows are radial except for rotation and free fall . ii ) the density is independent of latitude , thus the accretion isotropic outside the shock . iii ) the rotation frequency is independent of latitude and the specific angular momentum is conserved along each stream curve . although these assumptions are artificial and not self - consistent , since the centrifugal force is small compared with the gravity or the inertia , the radial component of the velocity must be predominant in reality . we define the parameter @xmath92 as the rotational frequency at the radius of @xmath93 km . we performed the calculations for @xmath94 @xmath95 and @xmath96 @xmath95 , which correspond to the specific angular momenta averaged over the whole sphere outside the shock wave surface of @xmath97 and @xmath98 , respectively . ( the definition of the specific angular momentum is employed by @xcite in their calculations of stellar evolutions ) . in figure [ fig6 ] , we show the stream curves for the parameters @xmath99 @xmath100 , @xmath101 @xmath102 , and @xmath103 @xmath95 . we can see that the flow is pushed toward the equatorial plane owing to centrifugal force . the stream curves bend rather abruptly near the neutrino - sphere because the infalling matter is decelerated as it lands onto the neutron star surface and the radial velocity falls rapidly there . it is also noted that the shock surface is slightly deformed into an oblate configuration , and the flows are directed toward the rotational axis at large radii . the angular dependence of the radial velocity inside the shock is shown in figure [ fig7 ] . at large radii , the velocity is larger near the rotation axis than the equatorial plane , whereas the opposite is true near the inner boundary . this behavior is attributed to the variation of cross section of the flow . the theory of laval nozzle tells us that subsonic flows are decelerated when the cross section of nozzle gets larger , vice versa @xcite . as we saw in figure [ fig6 ] , the flow is bent toward the equatorial plane near the neutrino - sphere but toward the rotational axis near the shock . as a result , at small radii , the cross section of flow becomes larger near the axis , whereas it gets smaller near the equatorial plane , and vice versa at large radii . thus , we obtain the observed angular dependence of velocity . next we discuss the effect of rotation on the critical luminosity . in figure [ fig8 ] , we show the dependence of the critical luminosity on the mass accretion rate . we can see that the smaller the mass accretion rate is , or the larger the rotation frequency is , the more reduced the critical luminosity is . according to the results of our calculations , if @xmath92 is 0.03 @xmath95 , and the accretion rate is 0.1 @xmath104/s , the critical value is about @xmath105 smaller than that of the spherically symmetric flow with the same mass accretion rate . if the rotation parameter is 0.1 @xmath95 the effect is more remarkable . for the mass accretion rates of 1.0 @xmath104/s and 0.5 @xmath104/s , we get the critical luminosities reduced by about @xmath106 and @xmath107 respectively , from the value for the spherically symmetric flow . finally , we discuss the fate of the shock wave when the critical luminosity is reached . in section 2 , we discussed the movement of the steady shock for the inner solutions ; the radial derivative of the infall velocity increases when the neutrino luminosity is raised , and it becomes nearly equal to that of @xmath38 when the luminosity reaches the critical value . the angular dependence of the velocity at the critical luminosity is shown in figure [ fig7 ] . we can see that the derivative is largest at @xmath89 . this suggests that the steady flow cease to exist first at the axis . thus , we can infer that the stalled shock is revived and start to propagate outwards first at the rotational axis and that the jet - like explosion is followed , although time - dependent numerical simulations are necessary to confirm this . this break up of shock at the rotational axis and the resulting reduction of the critical luminosity are also attributed to the changes of radial flow velocities , which can be understood based on the theory of laval nozzle , as we saw above . the variation of flow - cross - section causes the acceleration of the radial flow at large radii near the rotation axis whereas it leads to the deceleration near the equatorial plane . as a result , the revival of the stalled shock is more easily attained at the rotational axis for lower neutrino luminosities than in the spherically symmetric flows . in this paper , we investigated the behavior of the stalled shock in the supernova core and the effects of rotation on its revival . the spherically symmetric accretion flows were re - investigated first . we found that there exist two branches of solutions when the luminosity of the irradiating neutrinos is below the critical value . the steady solution does not exist when the luminosity is higher than the critical value . in order to consider which solution is realized below the critical luminosity , we then examined the stability of the spherical accretion flows against radial perturbations , and found that the inner solutions are stable while outer solutions are unstable . in fact , for small luminosities expected for the early phase of the shock stagnation , the radii of the outer solutions are extremely large ( figure [ fig2 ] ) . thus the outer solutions will not be realized in reality . the inner solutions are always stable until the neutrino luminosity reaches the critical value , where the two branches merge and the solution becomes neutrally stable . it should be noted , however , that the stability of the outer solutions may be affected by rotation , which will be a future work . we did not discuss the stability against non - radial perturbations , since our model are convectively unstable . several authors showed the spherical accretion shock is unstable against non - radial perturbations . @xcite demonstrated the instability for adiabatic perturbations by numerical simulations . @xcite examined such a stability taking into account the cooling processes whose rates are determined locally . in the present situation , however , the heating rates are not determined locally because the flow is irradiated by neutrinos coming out of the proto - neutron star ; the perturbations of the flow affect the neutrino luminosity , and the latter then affects the former , by varying the heating and cooling rates . thus , the problem is different from those they treated . the detailed analysis in such a situation is a future work . furthermore , the stability analysis with rotation taken into account is also an interesting issue . finally , we discussed the effects of rotation on the revival of the stalled shock . we showed that rotation lowers the critical luminosity . however , we claimed that the shock revival would first take place at the rotation axis . we took the rotation frequency of 0.03 and 0.1 @xmath95 at the radius of 1000 km and found that the rotation effect is substantial . the rotation frequency of newly born neutron stars guessed from the spin down rate of young pulsars is about several ten per second @xcite . thus the frequency we adopted here is one order of magnitude higher than the value inferred from the observations of young pulsars . it might be possible , however , that this discrepancy is removed by some mechanisms such as magnetic breaking in the late phase of explosion , which could take away the excessive angular momentum from the proto - neutron stars . in fact , the values we adopted are consistent with those obtained in calculations of the evolution of rotating stars @xcite . anyway , the angular momentum distribution inside massive stars are not yet well understood because of our poor knowledge on the mechanism of angular momentum transform ( cf . if the angular momentum of the infalling matter is large enough , our results suggest that rotation assists the shock revival . in this paper , we employed various assumptions and simplifications in the formulations . we assumed that the neutrino - sphere is spherical and the neutrino flux is isotropic . if the collapsing star is rotating , the proto - neutron star will have angular momentum . @xcite investigated how the neutrino - sphere is affected by rotation . they showed that the neutron star becomes oblate and , as a result , the emerging flux is more concentrated to the rotation axis . one the other hand , @xcite studied how the explosion is affected when the neutrino flux is greater near the rotation axis , and claimed that such neutrino anisotropy can trigger a jet - like explosion in otherwise failed models . such effects were neglected in this paper just for simplicity and are expected to further reduce the critical luminosity , which we will study in the future . the convection should be also affected by rotation . @xcite studied such effects by numerical simulations . they found that the centrifugal force reduce the effective gravity , and , thus , the efficiency of convection is lowered near the equatorial plane . as a result , the explosion energy was also decreased . the explosion , however , becomes jet - like . in order to investigate the effect of convection in the time - independent models , it is necessary to introduce some simplification , such as a mixing length theory or other formulation to smooth the entropy gradient this is currently being undertaken . finally the collapsing stars may have magnetic fields , which are then amplified by rotation . the recent works suggest that the magnetic fields are amplified by the differential rotation or magneto - rotational instability @xcite . if the magnetic energy is amplified to be comparable to the rotational energy , it must affect the accretion flow and the behavior of the stalled shock as they suggested . this issue is also a future work . akiyama , s. , wheeler , j. c. , meier , d. l. , & lichtenstadt , i. 2003 , , 584 , 954 ardeljan , n. v. , bisnovatyi - kogan , g. s. , & moiseenko , s. g. 2004 , , in press , ( astro - ph/0410234 ) blondin , j. m. , mezzacappa , a. , & demarino , c. 2003 , , 584 , 971 burrows , a. , & goshy , j. 1993 , , 416 , l75 buras , r. , rampp , m.,janka , h. t. , & kifonidis , k. 2003 , , 90 , 241101 courant , r. , & friedrichs , k. o. , 1976 , in supersonic flow and shock waves ( new york : springer - verlag ) foglizzo , t. 2002 , , 392 , 353 fryer , c. l. , & heger , a. 2000 , , 541 , 1033 heger , a. , langer , n. , & woosley , s. e. 2000 , , 528 , 368 heger a , woosley s e , langer n and spruit h c 2003 to appear in _ proc . iau 215 stellar rotation _ , ( astro - ph/0301374 ) herant , m. , benz , w. , & colgate , s. 1992 , , 395 , 642 houck , j. c. , & chevalier , r. a. 1992 , , 395 , 592 janka , h .- th . 2001 , , 368 , 527 janka , h .- th . and mller , e. 1996 , , 306 , 167 kaspi , v. m. , manchester , r. n. , siegman , b. , johnston , s. , & lyne , a. g. 1994 , , 422 , 83 kotake , k. , yamada , s. , & sato , k. 2003 , , 595 , 304 kotake , k. , sawai , h. , yamada , s. , & sato , k. 2004 , , 608 , 391 kulkarni , s. r. , 1992 , philos . london , a , 341 , 77 landau , l. d. , and lifshitz , e. m. 1987 , in fluid mechanics ( oxford : pergamon press ) leonard , d. c. , filippenko , a. v. , barth , a. j. , & matheson , t. 2000 , , 536 , 239 liebendrfer , m. , mezzacappa , a. , thielemann , f. k. , messer , o. e. b. , hix , w. r. , & bruenn , s. w. , 2001 , , 63 , 103004 liebendrfer , m. , rampp , m. , janka , h. t. , & mezzacappa , a. , 2004 , , in press , ( astro - ph/0310662 ) mnchmeyer r m and m@xmath108ller e 1989 , in nato asi series , timing neutron stars , ed . h. gelman & e.p.j van der heuvel ( new york : asi ) mller e , rampp m , buras r and janka h - th 2004 , 603 , 221 shapiro , s. l. , & teukolsky , s. a. 1983 , in black holes , white dwarfs , and neutron stars ( new york : john wiely & sons , inc . ) shimizu , t. m. , ebisuzaki , t. , sato , k. , & yamada , s. 2001 , , 552 , 756 spruit h c 2002 , 381 , 923 thompson t a , burrows a and pinto p a 2003 , 592 , 434 van der swaluw , e. , & wu , y. 2001 , , 555 , l49 wang , l. , wheeler , j. c. , hflich , p. , khokhlov , a. , baade , d. , branch , d. , challis , p. , filippenko , a. v. , fransson , c. , garnavich , p. , kirshner , r. p. , lundqvist , p. , mcray , r. , panagia , n. , pun , c. s. j. , phillips , m. m. , sonneborn , g. , & suntzeff , n. b. 2002 , , 579 , 671 wheeler , j. c. , 2003 , _ am . _ , 71 , 11 wilson , j. r. , 1982 , proc . illinois meeting on numerical astrophsics yamada , s. , kotake , k. , & yamasaki , t. 2004 , _ new j. phys . _ , 6 , 79 yamada , s. , & sato , k. 1994 , , 434 , 268
in order to infer the effects of rotation on the revival of a stalled shock in supernova explosions , we investigated steady accretion flows with a standing shock . we first obtained a series of solutions for equations describing non - rotating spherically symmetric flows and confirmed the results of preceding papers that , for a given mass accretion rate , there is a critical luminosity of irradiating neutrinos , above which there exists no steady solution . below the critical value , we found two branches of solutions ; one is stable and the other is unstable against radial perturbations . with a simple argument based on the riemann problem , we can identify the critical luminosity as the one , at which the stalled shock revives . we also obtained the condition satisfied by the flow velocity for the critical luminosity , which can be easily applied to the rotational case . if a collapsing star rotates , the accretion flow is non - spherical due to centrifugal forces . flows are accelerated near the rotation axis whereas they are decelerated near the equatorial plane . as a result , the critical luminosity is lowered , that is , rotation assists the revival of a stalled shock . according to our calculations , the critical luminosity is @xmath0% lower for the mass accretion rate of 1m@xmath1/sec and the rotation frequency of 0.1 hz at a radius of 1000 km than that of the spherically symmetric flow with the same mass accretion rate . we found that the condition of the flow velocity at the critical luminosity is first satisfied at the rotation axis . this suggests that the shock revival is triggered on the rotation axis and a jet - like explosion ensues .
we understand a complex system as a system with a large number of interacting components whose aggregated behaviour is non - linear and undetermined from the behaviour of the individual components @xcite . if we now consider these components as nodes of a network , and the underlying physical interaction between any two nodes as links , one way to understand these complex systems is by studying its topological structure , namely , the network connectivity . in natural complex systems , the connectivity of the components is often unknown or is difficult to detect by physical methods due to large system - sizes . hence , it is of interest to infer the network structure that represents the physical interaction between time - series collected from the dynamics of the nodes . although network inference in non - linear systems has been extensively studied in recent years using cross - correlation or mutual information @xcite , recurrences @xcite , functional dynamics @xcite , and granger causality @xcite , to name a few , it still presents open challenges . the fundamental reason is that non - linearities , even in the absence of noise , produce behaviour that hinders the correct identification of existing or non - existing underlying direct physical dependence between any pair of nodes . in this paper , we introduce an information - based methodology to infer the structure of complex systems from time - series data . our methodology is based on a normalized form of an estimated mutual information rate ( mir ) , the rate by which information is exchanged per unit of time between any two components . mir is an appropriate measure to quantify the exchange of information in systems with correlation @xcite . in particular , authors in ref . @xcite show how to calculate mir in the case a markov partition is attainable , which is generally extremely difficult to find or unknown . here , we first show how mir can be approximately calculated for time - series data of finite length and low - resolution . then , we propose a normalization of the estimated mir that allows for a successful inference about the dependence structure of small networks of interacting dynamical systems , when markov partitions are unknown . our findings show that the estimated normalized mir allows for a successful inference of the structure of small networks even in the presence of additive noise , parameter heterogeneities and different coupling strenghts . moreover , our normalized estimated mir outperforms the use of mutual information ( mi ) based inference when different time - scale dynamics are present in the networks . the paper is organized as follows . in sec . [ section_methods_and_material ] , we introduce two information - based measures , the mi and the mir . we discuss the theoretical aspects of their definitions and show how they are related to each other . in sec . [ sec : section_models ] , we introduce the models used to create the complex system dynamics studied in this work . in sec . [ sec : methodology ] , we explain our methodology to calculate an approximation value of mir and introduce the normalized mir . section [ sec : results ] shows how we apply our methodology to different coupled maps and to a neural network in which the dynamics of the nodes is described by the hindmarsh - rose neuron model @xcite . finally , in sec . [ sec : conclusions ] we discuss our work and discuss our findings . information can be produced in a system and it can be transferred between its different components @xcite . if transferred , at least two components that are physically interacting by direct or indirect links should be involved . in general , these components can be time - series , modes , or related functions of them , defined on subspaces or projections of the state space of the system . in this work , we study the amount of information transferred per unit of time , i.e. , the mutual information rate ( mir ) , between any two components of a system , to determine if a link between them exists . the existence of a link between two units means there is a bidirectional connection between them due to their interaction . the mutual information ( mi ) @xcite between two random variables , @xmath0 and @xmath1 , of a system is the amount of uncertainty one has about @xmath0 ( @xmath1 ) after observing @xmath1 ( @xmath0 ) . specifically , mi is given by @xcite @xmath2 where @xmath3 and @xmath4 are the marginal entropies of @xmath0 and @xmath1 ( shannon entropies ) respectively , and @xmath5 is the joint entropy between @xmath0 and @xmath1 . @xmath6 is the probability of a random event @xmath7 to happen in @xmath0 , @xmath8 is the probability of a random event @xmath9 to happen in @xmath1 , and @xmath10 is the joint probability of events @xmath7 and @xmath9 to occur simultaneously in variables @xmath0 and @xmath1 . @xmath11 is the number of random events in both variables , @xmath0 and @xmath1 . in particular , eq . can be written equivalently as @xmath12 this equation can be interpreted as the strength of the dependence between two random variables @xmath0 and @xmath1 @xcite . when @xmath13 , the dependence strength between @xmath0 and @xmath1 is null , consequently , @xmath0 and @xmath1 are independent . the computation of @xmath14 from time - series is a subtle task . firstly , it requires the calculation of probabilities computed on an appropriate probabilistic space on which a partition can be defined . secondly , @xmath14 is a measure suitable for the comparison between pairs of components of the same system but not between different systems . the reason is that different systems can have different correlation decay times , @xcite hence , different characteristic time - scales . there are three main approaches to compute mi , and the variation resides in the different ways to compute the probabilities involved in eq . . the first one is the bin or histogram method , which finds a suitable partition of the 2d space on equal or adaptive - size cells @xcite . the second one employs density kernels , where a kernel estimation of the probability density function is used @xcite . the last one computes mi by estimating probabilities from the distances between closest neighbours @xcite . in this work , we adopt the first method and compute probabilities in a partition of equally - sized cells in the probabilistic space generated by two variables @xmath0 and @xmath1 . it is well known that this approach , proposed in @xcite and studied in @xcite , overestimates the value of @xmath14 for random systems or non - markovian partitions @xcite . in particular , the authors explain two basic reasons for the overestimation of mi . the finite resolution of a non - markovian partition and the finite length of the recorded time - series . according to @xcite , these errors are systematic and are always present in the computation of mi for an arbitrary non - markovian partitions . here , we avoid these systematic error by creating a novel normalization when dealing with the mir . for the numerical computation of @xmath14 [ eq . ] , we use the approach reported in refs . we define a probabilistic space @xmath15 , where @xmath15 is formed by the time - series data observed from a pair of nodes , @xmath0 and @xmath1 , of a complex system . then , we partition @xmath15 into a grid of @xmath16 fixed - sized cells . the length - side of each cell , @xmath17 , is then set to @xmath18 . consequently , the probability of having an event @xmath7 for variable @xmath0 , @xmath6 , is the fraction of points found in row @xmath7 of the partition @xmath15 . similarly , @xmath8 is the fraction of points that are found in column @xmath9 of @xmath15 , and @xmath19 is the joint probability computed from the fraction of points that are found in cell @xmath20 of the same partition , where @xmath21 . we emphasize here that @xmath14 depends on the partition considered for its calculation as @xmath22 , @xmath23 , and @xmath24 attain different values for different cell - sizes @xmath17 . due to the issues arising from the definition of mi in terms of its partition dependence , the authors in ref . @xcite have demonstrated how to calculate the mir for two time - series of finite length irrespective of the partitions , instead of using the mi . this quantity is invariant with respect to the resolution of the partition @xcite . in particular , and for infinitely long time - series , mir is theoretically defined as the mutual information exchanged per unit of time between @xmath0 and @xmath1 @xcite . specifically , @xmath25 where @xmath26 represents the mi of eq . between random variables @xmath0 and @xmath1 , considering trajectories of length @xmath27 that follow an itinerary over boxes in a grid with an infinite number of cells @xmath11 . since @xmath28 is a symmetric function with respect to @xmath0 and @xmath1 , @xmath29 . we also note that the term @xmath30 tends to zero in the limit of infinitely long trajectories , @xmath31 . the authors in ref . @xcite show that if a partition with @xmath11 cells is a markov partition of order @xmath32 , then mir can be estimated from finite - length and low - resolution time - series ( since the limits in eq . ) by using @xmath33 where both @xmath34 and @xmath11 are finite quantities . notice that an order @xmath32 partition can only generate statistically significantly probabilities if there is in each cell a sufficiently large amount of points ( see eq . ( [ noc ] ) ) . besides , points in a cell must spread over the probabiliistic space @xmath15 after @xmath32 iterations . so , the length of the time - series must be reasonably larger than @xmath32 . in sec . [ sec : methodology ] , we make a novel demonstration of eq . ( [ mir_definition_epsilon ] ) , from which it becomes clear why mir can be estimated from finite - length and low - resolution time - series . in this equation , @xmath14 is the mi between @xmath0 and @xmath1 , considering probabilities that are calculated in a markov partition , and @xmath34 represents the shortest time for the correlation between @xmath0 and @xmath1 to be lost for that particular markov partition . @xmath34 also represents the time after which the evolution of a chaotic system is unpredictable . moreover , this time is of the order of the shortest poincar return - time @xcite and is related to the order @xmath35 markov partition , where @xmath35 indicates that the future state of a random variable @xmath0 is independent on its @xmath36s previous states and is independent on the states of @xmath0 for an order @xmath37 . we adopt various topologies for the networks and various dynamics for the components of the complex systems considered . hence , the network inference , which represents the detection of the topological structure of the component s interactions , is done from the time - series that are recorded for each component . in particular , we divide the analysis on discrete and on continues time - series components . the dynamics of the class of discrete complex systems that are of interest here are described by the following equation @xcite @xmath38 where @xmath39 is the @xmath40-th iterate of map @xmath7 , where @xmath41 and @xmath42 is the number of maps ( nodes ) of the system , @xmath43 $ ] is the coupling strength , @xmath44 is the binary adjacency matrix ( with entries @xmath45 or @xmath46 , depending on whether there is a connection between nodes @xmath7 and @xmath9 or not , respectively ) that defines the structural connectivity in the network , @xmath47 is the dynamical parameter of each map , @xmath48 is the node - degree , and @xmath49 is the considered map . particularly , we use @xmath50 for the logistic map @xcite of eq . , we use @xmath51 ( if it is not explicitly mentioned ) , that corresponds to fully developed chaos , whereas for the circle map @xcite of eq . we use @xmath52 and @xmath53 , following ref . @xcite , for the same reason . figure [ fig : networks]*(a ) * shows the network topology described by the adjacency matrix @xmath44 used to create a network where the dynamics of each node is described either by logistic or circle maps . we will use these networks to study the robustness of our methodology for different coupling strengths , observational noise , and data - length . we also use small - size networks with discrete dynamics , with different decay of correlation times for the nodes to test our methodology ( see fig . [ fig : networks]*(b ) * ) . in those networks , the dynamics of the nodes is given by logistic maps . in particular , we construct a network formed by two clusters of 3 nodes each . the clusters are connected by a small - coupling strength link . specifically , the dynamics of fig . [ fig : networks]*(b ) * for the cluster formed by the nodes @xmath54 is constructed by using @xmath51 , and the dynamics of the cluster formed by the nodes @xmath55 is given by a third - order composition of the logistic map , i.e. , @xmath56 , with @xmath57 . consequently , both clusters are constructed by time - series with different correlation decay times , creating a good example to understand how a clustered network with different time - scales can affect the inference capabilities of mi- or mir - based methodologies . .,title="fig : " ] .,title="fig : " ] + .,title="fig : " ] we consider continuous dynamics for the nodes of a network described by the hindmarsh - rose ( hr ) neuron model @xcite . the particular network we choose is shown in fig . [ fig : networks]*(c)*. the hr model is given by @xmath58 , \end{aligned}\ ] ] where @xmath59 is the membrane potential , @xmath60 is associated with the fast currents ( @xmath61 or @xmath62 ) , and @xmath40 with the slow current , for example @xmath63 . the rest of the parameters are defined as @xmath64 , @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 and @xmath70 , for which the system exhibits a multi - scale chaotic behaviour with spike bursting . parameter @xmath71 modulates the slow dynamics of the system . the neural networks of @xmath42 neurons connected by electrical ( linear coupling ) synapses is described in @xcite , and corresponds to having @xmath72,\;i=1,\ldots , m,\end{aligned}\ ] ] where @xmath42 is the number of neurons and @xmath73 . in eq . , @xmath74 is the strength of the electrical synapses . we use as initial conditions for each neuron @xmath7 : @xmath75 , @xmath76 , @xmath77 , and @xmath78 , where @xmath79 is a uniformly distributed random number in @xmath80 $ ] for all @xmath81 , following @xcite . @xmath82 is a laplacian matrix and accounts for the way neurons are electrically ( diffusively ) coupled . particularly , @xmath83 where @xmath84 is the binary adjacency matrix of the electrical connections and @xmath85 is the nodes degree diagonal matrix based on @xmath84 . if @xmath86 then neuron @xmath9 perturbs neuron @xmath7 with an intensity given by @xmath74 to infer the topology of a network using mir [ eq . ] , we need to compute the correlation decay time @xmath34 . @xmath34 is difficult to calculate in practical situations since it depends on quantities such as lyapunov exponents and expansion rates , which demand a high computational cost @xcite . here , we estimate it by the number of iterations that takes to points in cells of @xmath15 to expand and completely cover @xmath15 . this is a necessary condition to determine the shortest time for the correlation to decay to zero . in particular , we are introducing a novel way to calculate @xmath34 from the diameter of a network @xmath87 , which is based on the dynamics of points mapped from one cell of @xmath15 to another , namely , a network with the connectivity given by the transitions of points from cell to cell of @xmath15 or an itinerary network . we construct @xmath87 as follows . we assume that each equally - sized cell in @xmath15 , occupied by at least one point , represents a node in @xmath87 . then , following the dynamics of points moving from one cell to another , we create the connections between nodes , i.e. , the links in @xmath87 . specifically , a link between nodes @xmath7 and @xmath9 exists if points in @xmath15 travel from cell @xmath7 to cell @xmath9 . if the link exists the weight is equal to 1 , if it is absent , then it is equal to @xmath46 , therefore , @xmath87 is defined as a binary matrix with elements @xmath88 . in this framework , a uniformly random time - series with no correlation results in a complete network , namely , an all - to - all network . we define @xmath34 as the diameter of @xmath87 . the reason is that @xmath34 is the minimum time that takes for points inside any cell of @xmath15 to spread to the whole extent of @xmath15 . by definition , the diameter of a network is the maximum length for all shortest - paths , i.e , the minimum distance required to cross the entire network . hence , our approach transforms the calculation of @xmath34 into the calculation of the diameter of @xmath87 . in particular , for the estimation of the network diameter we use johnson s algorithm @xcite . to estimate mir from finite - length low - resolution time - series data , we truncate the summation in eq . up to a finite size @xmath11 , depending on the resolution of data , and consider small trajectory pieces of the time - series with a length @xmath27 , which depends on the total length of the time - series and on eq . ( [ noc ] ) , such that , @xmath89 . \label{truncate}\ ] ] in eq . ( [ truncate ] ) , left - hand and right - hand sides would be equal if the partition , where probabilities are being calculated , is markov . the length @xmath27 represents also the largest order @xmath32 that a partition that generates statistically significant probabilities can be constructed from these many trajectory pieces . assuming that the order of the partition constructed is @xmath90 ( which also represents the time for the correlation in the partition to decay to zero , if the partition would be markov ) , then eq . ( [ truncate ] ) becomes @xmath91.\ ] ] now , taking two partitions , @xmath92 and @xmath93 , with different correlation decay times , @xmath94 and @xmath95 respectively , and different number of cells , @xmath96 and @xmath97 respectively , with @xmath98 , we have @xmath99 . moreover , @xmath92 generates @xmath93 in the sense that @xmath100 , where @xmath101 is the evolution operator and @xmath102 means the pre - iteration of partition @xmath92 . then , @xmath103 hence , we can write eq . as , @xmath104\vspace{0.25cm}\\ & \cong&\frac{1}{t_1}\sum_{i=1}^{t_1}[i_{xy}(i,\lambda_2)-i_{xy}(i,\lambda_1 ) ] . \label{eq : mir_partitions}\end{aligned}\ ] ] when the partition is a markov generating partition , its properties @xcite fulfil @xmath105 then , if our partition is close to a markov partition , eq . results in @xmath106\\ & \equiv & \frac{1}{t_1}i_{xy}(1,\lambda_{t_1}),\end{aligned}\ ] ] which is our demonstration for the validity of eq . . therefore , in order to use eq . ( [ eq : mir_markov_p ] ) , we must have partitions for which eq . ( [ relate - markov ] ) is approximately valid . this condition can be reached for partitions constructed with a sufficiently large number of equally - sized cells of length @xmath107 , exactly the type of partition considered here . notice , however , that partitions will typically not be markov nor generating , causing systematic errors in the estimation of mir . to correct these errors , we propose the normalizations in eqs . ( [ mir_epsilon_normalized ] ) and ( [ mir_en2 ] ) . it is important to notice that @xmath108 is always a partition - independent quantity , if and only if , the partitions are markov . in order to calculate @xmath109 , we use eq . , which requires the calculation of probabilities in @xmath15 . fulfilling the inequality @xmath110 where @xmath111 is the mean number of points inside all occupied cells of the partition of @xmath15 , eq . guarantees that the probabilities are unbiased . for our analysis , using a non - markovian partition allows us to simplify the calculations of @xmath108 , however , taking this kind of partitions into consideration would make the mir values to oscillate around an expected value . moreover , mir for different non - markovian partitions , not only has a non - trivial dependence with the number of cells in the partition , but also presents a systematic error @xcite . therefore , since @xmath108 for a non - markovian partition of @xmath112 equally - sized cells [ estimated by eq . ] , is expected to be partition - dependent , we propose here a way to obtain a measure , computed from @xmath113 , that is partition independent and that is suitable for network inference . to infer the structure of a network , we calculate the mir for the @xmath114 different pairs of nodes in the network , which is all we need due to the symmetric property of mir . we also discard the mir values for the same variable , i.e. , mir@xmath115 , because we are interested in the exchange of information between different variables . we compute the @xmath108 exchanged between any two nodes in a network by taking the expected value over different partition sizes @xmath116 , i.e. , @xmath117 , where @xmath118 is the expected value of @xmath0 . in order to remove the systematic error @xcite in this calculation , we perform instead a weighted average , where the finer partitions ( larger @xmath11 ) contribute more to the @xmath108 value than the coarser ones ( smaller @xmath11 ) . the reason is that a smaller @xmath11 is likely to create a partition that is further away from a markovian one than a partition of larger @xmath11 . consequently , we resolve the systematic error by weighing differently the different partitions . therefore , we propose a novel normalization for the mir as follows . first , we use an equally - sized grid of size @xmath11 , we subtract from mir@xmath119 , calculated for all pairs of nodes , its minimum value and denote the new quantity as @xmath120 . theoretically , a pair that is disconnected should have a mir value close to zero , however , in practice , the situation is different because of the systematic errors coming from the use of a non - markovian partition , as well as , from the information flow passing through all the nodes in the network . for example , the effects of a perturbation in one single node will arrive to any other node in a finite amount of time . this subtraction is proposed to reduce these two undesired overestimations of mir . after this step , we remain with mir as a function of @xmath11 . normalizing then by @xmath121 , where again the maximum and minimum are taken over all different pairs , we construct a relative magnitude @xmath122 , namely , @xmath123 where @xmath113 is the mir between nodes @xmath0 and @xmath1 and @xmath124 is the minimum with respect to the @xmath114 pairs and @xmath125 is the maximum with respect to all @xmath114 pairs . this magnitude is still a function of @xmath11 , however , we can now perform an average over different values of @xmath11 without the systematic error . next , we apply eq . for different grids sizes @xmath126 to obtain @xmath127 , where @xmath128 is the maximum number of cells per axis , resulting in a grid of @xmath129 cells , and fulfilling at the same time eq . . then , similarly to the idea used for eq . , we make a second normalization over @xmath130 to obtain @xmath131 where the maximum is being taken now over the @xmath128 grids . finally , applying eq . to each pair @xmath132 , we obtain its average value , @xmath133 . the higher the value of @xmath133 , the higher the amount of information exchanged between @xmath0 and @xmath1 per unit of time . this allows us to identify pairs of nodes that exchange larger rates of information than others . in order to perform the network inference from the mir , we fix a threshold in @xmath134 $ ] and create a binary adjacency matrix @xmath135 , where the entry @xmath136 is @xmath45 if @xmath137 is higher than the threshold , and @xmath46 otherwise . @xmath135 is then compared with the adjacency matrix @xmath84 used to construct the dynamics of the nodes in sec . [ sec : section_models ] . recording the threshold used to create @xmath135 , and varying it in @xmath134 $ ] , we obtain different inferred networks . our results show that there is an interval of thresholds within @xmath134 $ ] that fulfil @xmath138 , i.e. , a band that represents a @xmath139 successful network inference . in general , the effectiveness of our network inference methodology is measured by the absolute difference between the real topology and the one inferred for different threshold values . we find that whenever there is a band of threshold values , there is successful inference without errors . in practical situations , where the underlying network is unknown and the absolute difference is impossible to compute , the ordered values of the mir or other similarity measures @xcite show a plateau which corresponds to the band of thresholds aforementioned . in particular , if the plateau is small , the authors in ref . @xcite propose a method to increase the size of the plateau by `` silencing '' the indirect connections , hence , allowing for a more robust reconstruction of the underlying network . we now present our results for network inference using the three models introduced in sec . [ sec : section_models ] . here we study the performance of eq . for network inference in the case where the dynamics of each node is described by a circle or a logistic map . the network structure that comprises our small - network of interacting discrete - time systems is given in fig . [ fig : networks]*(a)*. here , we analyze the effectiveness of the inference as the coupling strength , @xmath140 , between connected nodes , is varied . in ref . @xcite , the authors have shown that , for the logistic [ eq . ] and circle maps [ eq . ] , assuming the same topology , the dynamics is quasi - periodic for @xmath141 and chaotic for @xmath142 . we , therefore , choose the coupling strength @xmath140 in eq . to be equal to @xmath143 and @xmath144 , both values corresponding to chaotic dynamics . figure [ fig : coupling_strength_16_logistic_maps ] shows the network inference results using @xmath133 . the wideness of the red band represents all possible values a threshold can take to perform a 100% success network inference , i.e. the correct identification of all physical and non - physical links . the wider the band , the bigger the probability to perform a complete reconstruction , therefore the reconstruction is more robust . when we deal with experimental data , and the correct topology is unknown , the optimal threshold can be determined by the range of consecutive thresholds for which the inferred topology is invariant , see ref . @xcite . an error in the percentage of reconstruction comes from links that were not inferred ( false negatives ) or inferred erroneously ( false positives ) . in our current study we avoid the distinction between them and we categorize both as reconstruction errors . then , the reconstruction percentage can decrease by inferring non - existent links ( non physical links ) or by missing them . each time this happens , we decrease the percentage by an amount @xmath145 , where @xmath146 is the number of real links in the original network . we start by analysing the effectiveness of @xmath133 for different time - series lengths , using the dynamics of the logistic map for each node and a coupling strength @xmath147 $ ] . in fig . [ fig : param]*(a ) * , we observe that for @xmath140 closer to @xmath148 , a relatively short length ( of about @xmath149 points ) is enough to infer correctly the original network , which is generated by the adjacency matrix @xmath84 of sec . [ sec : section_models ] . however , when @xmath140 is close to @xmath143 , a larger time - series ( of about @xmath150 points ) is needed to achieve @xmath139 successful reconstruction . values of @xmath151 and @xmath152 $ ] are considered to test the effectiveness of the values @xmath133 in the case of nodes being totally independent and in the case of nodes having periodic dynamics . in these regions , @xmath133 is expected to be zero , a situation evidenced in both panels of the figure . our results so far suggest that the successful reconstruction for short - length time - series depends on the intensity of the coupling strength . however , it is surprising to see that exact inference can always be achieved for this dynamical regime if a sufficiently large time - series is available . next , we apply our methodology for network inference considering noisy time - series data . in particular , we introduce additive normally - distributed noise to the logistic map , i.e. , @xmath153 where @xmath154 is the noisy dynamics , @xmath155 is a random number drawn from the normal distribution with @xmath46 mean and standard deviation of @xmath45 , i.e. @xmath156 , and @xmath157 $ ] is the noise strength . since @xmath158 , the noise strength is the standard deviation in the normal distribution . fig . [ fig : param]*(b ) * shows the parameter space for different coupling strengths versus @xmath159 . we observe perfect inference for noise strengths @xmath160 , i.e. for @xmath156 . moreover , the best reconstruction using @xmath133 is for coupling strengths in @xmath161 $ ] , a dynamical regime where chaotic behaviour is prevalent . we also apply our methodology for the study of network inference in the case of continuous dynamics given by the hr system . we use two electrical couplings , @xmath162 and @xmath163 , both considered for time - series of length @xmath164 . figure [ fig : hindmarsh - rose_recons ] shows the band for 100% successful network inference , where panel * ( a ) * corresponds to @xmath162 and panel * ( b ) * to @xmath165 . this figure shows that @xmath133 is able to infer the correct network structure , in this case , for small networks of continuous - time interacting components . and @xmath165 , respectively . the red bands show the range of thresholds for which the original network is inferred with a 100% success.,title="fig : " ] and @xmath165 , respectively . the red bands show the range of thresholds for which the original network is inferred with a 100% success.,title="fig : " ] finally , we compare mi and @xmath133 to assess the effectiveness of our proposed methodology for network inference . we apply the same normalization process used for mir , eq . , to mi to have an appropriate comparison . in particular , we infer the network structure of the system described in sec . [ sec : section_models ] with the network shown in fig . [ fig : networks]*(b)*. as we have explained in sec . [ sec : section_models ] , this system has two clusters of nodes with different dynamics . the dynamics in the left cluster is given by the 3rd - order composition of the logistic map , whereas the dynamics of the right cluster is given by ordinary logistic map dynamics . the different dynamics of the two groups produces different correlation decay times , @xmath34 , for nodes @xmath0 and @xmath1 , in particular when the pair of nodes comes from different clusters . the different correlation decay times produce a non - trivial dynamical behaviour that challenges the mi performance for network inference . . panel * ( a ) * plots @xmath28 of eq . for all links . this is a case where a complete network inference can not be achieved ( indicated by the absence of any red band ) . panel * ( b ) * is the same as before but for @xmath108 . the color code corresponds to the same color code identifying different nodes in fig . [ fig : networks]*(b)*. the darkest color is the link connecting the two clusters.,title="fig : " ] . panel * ( a ) * plots @xmath28 of eq . for all links . this is a case where a complete network inference can not be achieved ( indicated by the absence of any red band ) . panel * ( b ) * is the same as before but for @xmath108 . the color code corresponds to the same color code identifying different nodes in fig . [ fig : networks]*(b)*. the darkest color is the link connecting the two clusters.,title="fig : " ] figure [ fig : composed_system ] shows the results obtained for the normalized mi , @xmath166 , and our normalized mir , @xmath133 , for each of the possible pairs of nodes . the purple bars correspond to the pairs of nodes @xmath45 , @xmath167 and @xmath168 of the first cluster , the orange bars correspond to the pairs of nodes @xmath169 , @xmath170 and @xmath171 of the second cluster ( 3rd order composed dynamics ) and the black bar corresponds to the link between clusters ( notice that due to the small coupling strength between the two clusters this link is not detected using any of the two methods ) . nevertheless , mir identifies correctly all intra links of the network where mi fails to do so . we conclude that the normalized mir is preferable over the normalized mi when it comes to the detection of links in a complex system with different correlation decay times . the reason is that the normalized mir takes into consideration the correlation decay time associated to each pair of nodes , contrary to the mi . in this paper we have introduced a new information based approach to infer the network structure of complex systems . mir is an information measure that computes the information transferred per unit of time between pairs of components in a complex system . @xmath133 , our novel normalization for the mir that is introduced in eq . , is a measure based on mir and developed for network inference . we find that @xmath133 is a robust measure to perform network inference in the presence of additive noise , short time - series , and also for systems with different coupling strengths . since mir and @xmath133 depend on the correlation decay time @xmath32 , they are suitable for inferring the correct topology of networks with different time - scales . in particular , we have explored the effectiveness of mir versus mi in terms of how successful they are in inferring exactly the network of our small complex systems . in general , we find that mir outperforms mi when different time - scales are present in the system . our results also show that both measures are sufficiently robust and reliable to infer the networks analyzed whenever a single time - scale is present . in other words , small variations in the dynamical parameters , time - series length , noise intensity , or topology structure , maintain a successful inference for both methods . it remains to be seen the types of errors that are found in these measures when perfect inference is missing or impossible to be done . ebm , msb and cga acknowledge financial support provided by the epsrc `` ep / i032606/1 '' grant . cga contributed to this work while working at the university of aberdeen and then , while working at the university of essex , united kingdom . nr acknowledges the support of pedeciba , uruguay . 10 n. rubido , a. c. mart , e. bianco - martinez , c. grebogi , m. s. baptista , and c. masoller , `` exact detection of direct links in networks of interacting dynamical units '' , new j. phys . * 16 * , 093010 ( 2014 ) . m. s. baptista , f. m. kakmeni , and c. grebogi , `` combined effect of chemical and electrical synapses in hindmarsh - rose neural networks on synchronization and the rate of information . '' , phys . rev .. e * 82*(3 ) , 036203 ( 2010 ) . g. benettin , l. galgani , a. giorgilli , and j. m. strelcyn , `` lyapunov characteristic exponents for smooth dynamical systems and for hamiltonian systems ; a method for computing all of them . part 1 : theory '' , meccanica * 15 * , 9 - 20 ( 1980 ) . g. benettin , l. galgani , a. giorgilli , and j. m. strelcyn , `` lyapunov characteristic exponents for smooth dynamical systems and for hamiltonian systems ; a method for computing all of them . part 2 : numerical application '' , meccanica * 15 * , 21 - 30 ( 1980 ) . j. m. stinnett - donnelly , n. thompson , n. habel , v. petrov - kondratov , d. d. correa de sa , j. h. bates , and p. s. spector , `` effects of electrode size and spacing on the resolution of intra - cardiac electrograms '' , coronary artery dis . * 23*(2 ) , 126 - 32 ( 2012 ) . n. k. chen , c. c. dickey , s. s. yoo , c. r. guttmann , and l. p panych , `` selection of voxel size and slice orientation for fmri in the presence of susceptibility field gradients : application to imaging of the amygdala '' , neuroimage * 19*(3 ) , 817 - 25 ( 2003 ) .
this work uses an information - based methodology to infer the connectivity of complex systems from observed time - series data . we first derive analytically an expression for the mutual information rate ( mir ) , namely , the amount of information exchanged per unit of time , that can be used to estimate the mir between two finite - length low - resolution noisy time - series , and then apply it after a proper normalization for the identification of the connectivity structure of small networks of interacting dynamical systems . in particular , we show that our methodology successfully infers the connectivity for heterogeneous networks , different time - series lengths or coupling strengths , and even in the presence of additive noise . finally , we show that our methodology based on mir successfully infers the connectivity of networks composed of nodes with different time - scale dynamics , where inference based on mutual information fails . = 1 * the mutual information rate ( mir ) measures the time rate of information exchanged between two non - random and correlated variables . since variables in complex systems are not purely random , mir is an appropriate quantity to access the amount of information exchanged in complex systems . however , its calculation requires infinitely long measurements with arbitrary resolution . having in mind that it is impossible to perform infinitely long measurements with perfect accuracy , this work shows how to estimate mir taking into consideration this fundamental limitation and how to use it for the characterization and understanding of dynamical and complex systems . moreover , we introduce a novel normalized form of mir that successfully infers the structure of small networks of interacting dynamical systems . the proposed inference methodology is robust in the presence of additive noise , different time - series lengths , and heterogeneous node dynamics and coupling strengths . moreover , it also outperforms inference methods based on mutual information when analysing networks formed by nodes possessing different time - scales . *
in the past few years , considerable progress has been made in our understanding of the nature of high redshift galaxies , driven mostly by the availability of large samples that have been photometrically selected using rest - frame uv spectral features . the largest existing sample at present is at @xmath8 , where it is efficient to use ground - based imaging through optical filters designed to isolate the lyman limit spectral discontinuity at 912 in the rest - frame , and where the spectroscopic follow - up has proved to be straightforward because many important spectral features are well - placed at wavelengths where optical spectrographs are most sensitive ( steidel _ et al . _ 1996a , steidel _ et al . _ very deep hubble space telescope ( hst ) imaging such as that provided by the hubble deep fields has also proved to be very effective , using similar techniques for @xmath15 ( steidel _ et al . _ 1996b , madau _ et al . _ 1996 , lowenthal _ et al . _ the hst data reach much deeper into the high redshift galaxy uv luminosity function , but are confined to very small areas on the sky ; as such , the ground - based and space - based surveys have been largely complementary ( cf . et al . _ 1999 ) . to date , most of the work on these high redshift galaxy samples has focused on the large - scale clustering properties of the galaxies ( e.g. , adelberger _ et al . _ 1998 , giavalisco _ et al . _ 1998 , steidel _ et al . _ 1998 , giavalisco & dickinson 2001 ) , on the properties of individual galaxies as deduced from their spectra ( e.g. , pettini _ et al . _ 1998 , 2000 , 2001 ) , or on inferences concerning the universal star - formation history as deduced from the global uv luminosity density ( e.g. , madau _ et al . _ 1996 , steidel _ et al . _ 1999 , meurer , heckman , & calzetti 1999 , adelberger & steidel 2000 ) . the strong clustering of the bright lyman break galaxies ( lbgs ) has generally been interpreted as indirect evidence that the observed galaxies reside in relatively massive dark matter halos , but how these galaxies are linked to present - day galaxies is far from clear , and has been quite controversial . while considerable , the directly - observed uv luminosity of lbgs is clearly very significantly modified by extinction ( see adelberger & steidel 2000 , hereafter as2000 , for an extensive discussion of this topic ) , and at best provides information on the instantaneous formation rate of o and b stars . while the far - uv properties of the lbgs are the most straightforward to study observationally , it is also of interest to explore additional properties of these galaxies that require longer wavelength observations , to be used in concert with the existing far - uv measurements . for example , one would like to determine the distribution of rest - frame optical luminosities , and the star - formation histories , ages , dust content , and stellar masses all necessary for understanding the range of objects selected with the typical lyman break photometric criteria , and the relationship of these objects to galaxies in the lower redshift universe . indeed , different models provide quite divergent descriptions of the nature and fate of the objects which are identified as lbgs . according to one model , the galaxies selected as lbgs are bright in the rest - frame uv because they are experiencing merger - induced starburst events . the intense starburst events occur on relatively short timescales ( less than 100 myr ) , and produce 10 - 100 times less stellar mass than what is seen in a typical @xmath16 galaxy . such low - mass bursting objects would then be the precursors of local low - mass spheroids , unless they merge with similar objects to form more massive systems ( lowenthal _ et al . _ 1997 ; sawicki & yee 1998 ; somerville , primack , & faber 2001 ) . alternatively , lbgs are the central objects in relatively massive dark matter halos , which form stars steadily but relatively quiescently over longer than 1 gyr timescales , accumulate @xmath17 by @xmath0 , and evolve eventually into the ellipticals and spiral galaxies at the bright end of the local luminosity function ( steidel _ et al . _ 1996a ; baugh _ et al . _ 1998 it is not possible to determine the nature of lbg stellar populations and thereby distinguish between the above scenarios with only optical observations probing the rest - frame uv . both an aging stellar population and increasing dust extinction result in redder rest - frame uv colors , so that the effects of age and dust are degenerate without longer wavelength data . recent studies ( sawicki & yee 1998 ; dickinson 2000 ; papovich , dickinson , & ferguson 2001 ) have shown that the addition of near - ir photometric measurements at rest wavelengths longer than the location of an age - sensitive spectral break at @xmath18 removes some of the degeneracy between dust and age in the modeling of lbg stellar populations . here we present the results of a moderately large near - ir survey of optically selected lbgs , designed to determine the distribution of rest - frame optical luminosities of lbgs , and , where possible , information about their star - formation histories . in 2 , we present the details of the optical and near - ir observations and data reduction . 3 summarizes the distribution of optical - ir colors of lbgs . in 4 , we describe the derivation of the lbg rest - frame optical luminosity function , and compare it to local galaxy luminosity functions . 5 presents the procedure and results of population synthesis modeling of the lbg spectral energy distributions between 900 and 5500 in the rest - frame , using the measured optical / ir colors . extensions of the model results to rest - frame uv spectroscopy are discussed in 6 , and an evolutionary sequence for lbgs is proposed in 7 . our general conclusions are summarized in 8 . optical images were obtained for all the fields included here as part of our extensive survey for @xmath8 galaxies ; the field centers are given in table 1 . the optical imaging data were collected during the interval 1995 - 1998 at the william herschel telescope ( 3c 324 , b2 0902 + 34 , and cdfa ) , the palomar 200 inch hale telescope ( cdfa , dsf2237a , dsf2237b , hdf , q0201 , q0256 , ssa22a , and ssa22b ) , and the kitt peak mayall 4 meter telescope ( westphal ) . the details of our survey have been presented elsewhere ( steidel _ et al . _ 1996a , giavalisco _ et al . _ 1998 , steidel _ et al . _ 1999 ) , so here we present only a summary of its relevant features . in deep @xmath19 , @xmath20 , @xmath4 ( steidel & hamilton 1993 ) images , with typical 1@xmath21 surface brightness limits of 29.1 , 29.2 , 28.6 @xmath22 magnitudes per arcsec@xmath1 , respectively , we identify @xmath8 galaxy candidates by their distinctive colors . to select the objects in the sample centered at @xmath8 , the photometric criteria consist of the following : @xmath23 these criteria effectively isolate star - forming galaxies whose redshifts place the lyman continuum discontinuity within the @xmath19 band , and result in a redshift distribution that is reasonably well - described by a gaussian with @xmath24 and a standard deviation of @xmath25 . the various biases inherent in such photometric selection have been discussed extensively by steidel _ et al . _ ( 1999 ) . at the time of this writing , there are approximately 1000 lbgs in this redshift range that have been spectroscopically confirmed using the low resolution imaging spectrometer ( lris ; oke _ et al . _ 1995 ) at the w. m. keck observatory . the full optical survey and spectroscopic catalogs will be presented elsewhere . to explore the range of optical / near - ir colors present in our sample of high redshift galaxy candidates and spectroscopically confirmed @xmath26 galaxies , a subset of objects was selected for follow - up near - ir imaging . objects bright in @xmath4 were preferentially selected for near - ir imaging , as can be seen by the relative distributions of @xmath4 apparent magnitude in the near - ir sample and that in our @xmath8 sample as a whole ( figure 1a ) . we deliberately selected objects whose rest - uv - inferred extinction properties ( see steidel _ et al . _ 1999 and as2000 for a discussion ) spanned the entire range seen in the full lbg survey , from zero to several magnitudes of extinction in the rest - frame uv . within the lbg @xmath8 selection function , galaxies with higher redshifts suffer greater absorption by intergalactic h i , resulting in systematically redder @xmath27 colors for a fixed intrinsic spectral energy distribution ( sed ) . figure 1b shows the distributions of @xmath28 , the @xmath27 color statistically corrected for igm absorption , for both the nirc lbg sample and the lbg @xmath8 sample as a whole . this figure shows that , in an effort to target galaxies with extreme values of implied extinction with nirc , we over - sampled the reddest @xmath28 bins by roughly a factor of 2 relative to the rest of the distribution . all of the near - ir data were obtained with the facility near - ir camera ( nirc ) ( matthews & soifer 1994 ) on the w.m . keck i telescope during the course of 9 separate observing runs in the interval 1997 may - 1999 may . nirc has a @xmath29 insb array , with a pixel scale of @xmath30 , resulting in a @xmath31 field . this field size represents less than 1% of the large optical pointings used to find lbgs ( typically 9 by 9 ) which contain 100 - 150 @xmath8 candidates to @xmath32 . to maximize efficiency , we preferentially targeted objects having at least one other lbg within @xmath33 . a typical pointing included 23 lbgs , but this number varied from 1 to 7 lbgs per pointing . each nirc pointing contained at least one primary target galaxy with a spectroscopic redshift , and , in many cases , additional lbg candidates for which spectra have not yet been obtained . there are spectroscopic redshifts for about 75% of the full @xmath5 sample . we observed at @xmath34 and @xmath35 , using the standard @xmath3 and @xmath2 filters . using integration times of @xmath36 seconds @xmath37 and @xmath38 seconds @xmath39 per exposure , we dithered the telescope between exposures to form a 9-point box pattern on the sky , each box position separated by a few arcseconds . in most cases , we aimed to complete 6 or more sets of the 9-point dither pattern for each nirc pointing . during the dither pattern the telescope was guided with an off - set ccd camera , with the orientation of the nirc detector chosen so that a suitable guide star would fall on the guider chip . table 1 summarizes the coverage of the nirc survey with respect that of the optical survey for lbgs , listing the angular area and number of lbgs covered in @xmath2 and @xmath3 for each optical survey field . the optical fields listed in table 1 do not represent the entire optical lbg survey , but only the fields which contain nirc pointings . table 2 summarizes all of the nirc target pointings , the objects contained in each pointing , and the total integration times for each pointing . the data were reduced using standard procedures with the aid of the dimsum package . on clear nights , we determined the photometric zeropoints for the @xmath3 and @xmath2 bandpasses with observations of faint ( 11th - 12th magnitude ) near - infrared standard stars from the list of persson _ et al . _ ( 1998 ) . during 5 of our 9 nirc runs , we experienced stable , photometric conditions , while the remaining 4 runs contained variable cirrus . on runs with variable conditions , we carefully calibrated observing sequences with observations obtained when the conditions were judged to be photometric . as measured from the fwhm of standard stars , the seeing during our several nirc runs ranged from @xmath40 in both @xmath3 and @xmath2 , with fwhm@xmath41 being typical . figure 2 shows two different examples of @xmath4 and @xmath2 images of lbgs surveyed with nirc . westphal - mmd11 has the reddest @xmath5 color and brightest @xmath2 magnitude in the nirc lbg sample , while b20902-c6 is in the bluer and fainter half of the nirc lbg sample . prior to the measurement of optical / near - ir colors , each nirc frame was smoothed with a gaussian to match the image quality in the corresponding optical images , which generally had @xmath42 . the photometry was then performed in a manner analogous to that used for @xmath8 galaxy searches , described elsewhere ( steidel _ et al . _ 1995 ; steidel & hamilton 1993 ) . in brief , using a modified version of the focas ( valdes 1982 ) image detection and analysis routines , a catalog of isophotal and `` total '' object detections was generated in the @xmath43 ( where `` total '' refers to the focas definition , which is the flux measured within an aperture grown to twice the area of the initial detection isophote ) . the @xmath43 isophotal detection apertures were applied to the @xmath3 and @xmath2 images , to measure @xmath44 and @xmath5 colors . table 3 lists the @xmath4 _ total _ magnitudes , @xmath27 , @xmath44 , @xmath5 _ isophotal _ colors , and redshifts for all objects in the nirc sample . optical magnitudes ( @xmath20 and @xmath4 ) and colors are referenced to the @xmath22 system , whereas the near - ir magnitudes ( @xmath3 , and @xmath2 ) are on the vega system . magnitudes , we adopt the transformations : @xmath45 , @xmath46 . the standard relation between @xmath22 magnitude and flux - density , @xmath47 , is : @xmath48 where @xmath47 is in units of ergs s@xmath10 @xmath49 hz@xmath50 . ] to quantify the uncertainties in the measured @xmath44 and @xmath5 colors , we ran monte carlo simulations mimicking the actual process used to measure magnitudes and colors from the real data . artificial galaxies , with a reasonable range of intrinsic sizes and with @xmath4 magnitudes and @xmath5 colors drawn randomly from the observed range of both quantities , were added to the images after convolution with the seeing disk . we then produced detection catalogs in @xmath4 and measured the @xmath5 colors with matched apertures in the @xmath2 image . the process of adding fake galaxy sets ( the number of galaxies added in a given trial was kept small enough so as not to alter the systematics of the observed data frames ) and recovering the magnitudes and colors using the same procedures used for the real data was repeated until enough detections were obtained to study the scatter between true and recovered @xmath5 color as a function of recovered @xmath4 and @xmath5 . the monte - carlo generated photometric uncertainties are significantly larger than those that would result from the application of simple poisson counting statistics that neglect systematics . the detections of artificial objects were binned in 0.5 mag steps in measured @xmath4 magnitude and 0.2 mag steps in measured @xmath5 color . the mean and standard deviation were then computed for the distribution of @xmath51 in each @xmath52 bin . in general we did not find a significant systematic offset between the measured and true colors , so the color uncertainties are treated as symmetric . accordingly , we adopted the standard deviation of the @xmath53 distribution as the @xmath54 color uncertainty , @xmath55 for each @xmath52 bin . to be conservative , we initially treated each nirc pointing independently for the estimation of photometric uncertainties in order to explore possible variations due to slightly different depths , seeing conditions , and sky background . we subsequently determined that the trend of @xmath55 with @xmath52 was similar enough in all @xmath2 nirc pointings that we simply combined the simulation results from all pointings to calculate an average trend of @xmath55 with @xmath52 for our entire @xmath5 sample . uncertainties were then assigned to the measured @xmath56 colors in the nirc sample , based on the magnitude and color of each object . an identical procedure was used to generate uncertainties in the @xmath44 colors of all of the observed galaxies . the @xmath27 uncertainties were computed from a separate , but similar , set of monte carlo simulations ( adelberger 2001 ) . all the color uncertainties are listed along with the optical / ir photometry in table 3 . the distribution of measured optical - ir colors for the 118 galaxies with nirc @xmath2 data is shown in figure 3 . 107 of these measurements are @xmath2 detections significant at the @xmath57 level , while 11 of them are only upper - limits , whose @xmath58 limits correspond to the typical @xmath59 detection limit of @xmath60 in our images . we find a mean @xmath58 color of @xmath61 and a standard deviation @xmath62 . of these 118 galaxies , 81 have measured redshifts , with a mean of @xmath63 . there are @xmath3 band measurements for 63 of the galaxies in the @xmath56 sample . the mean @xmath64 color is @xmath65 with @xmath66 . at @xmath8 , the @xmath3 and @xmath2 filters correspond to @xmath67 , and @xmath68 in the rest - frame , respectively , whereas the optical passbands @xmath19 , @xmath20 , and @xmath4 sample the galaxies at rest - frame @xmath69 900 , 1200 , and 1700 , respectively . clearly , there is a wide range of optical / near - ir color among the rest - uv selected @xmath8 lbgs ; the bluest galaxies in @xmath5 have colors that are essentially flat in @xmath47 units from the far - uv to the optical in the rest - frame , whereas the reddest lbgs approach the colors of `` extremely red objects '' . the typical @xmath58 color of @xmath8 lbgs is more than 2.5 magnitudes redder than that expected for an unreddened instantaneous burst of star - formation at the mean redshift of our sample ( @xmath58 = 0.30 ) , but is significantly bluer than most galaxies in the present - day universe ( cf . et al . _ unfortunately , in the absence of high quality spectra covering the same large wavelength range , the interpretation of the colors of lbgs in the context of understanding their stellar populations and extinction must rely heavily on models . in 5 , we use the full set of optical - ir colors ( @xmath70 ) of lbgs in the nirc sample in an attempt to disentangle the degenerate effects of dust and age on lbg broadband spectral energy distributions . the distribution of redshifts and observed @xmath4 and @xmath71 magnitudes have been used to construct rest - frame far - uv luminosity functions for lbgs at @xmath8 and @xmath72 respectively ( steidel _ et al . _ 1999 ) . with near - ir @xmath2 magnitudes and optical - ir @xmath58 colors for a sample of 118 lbgs , we have the necessary information to construct the rest - frame _ optical _ luminosity function of @xmath8 lbgs , which is much more easily compared with galaxies in the present - day universe . optical emission is much less attenuated than far - uv light by the presence of dust , and expected to be less directly linked to the instantaneous star - formation rate , as the stars giving rise to the optical emission sample a larger swath of the main sequence and include stars with longer lifetimes than the o / b stars producing the far - uv light . thus , one might hope that the optical luminosities provide more information on the integrated stellar populations of the lbgs than can be obtained from far - uv measurements . at @xmath73 , the mean redshift of the lbg spectroscopic sample ( and of the nirc sample as well ) , the @xmath2-band central wavelength of @xmath74 is quite a good match to the central wavelength of the standard @xmath7 optical filter . while the nirc sample contains galaxies with a range of redshifts , the @xmath2-band central wavelength falls within the rest - wavelength range @xmath75 for the bulk of the sample . furthermore , the `` photometric depth '' of the lbg redshift selection function corresponds to a difference in absolute magnitude of only @xmath76 magnitudes for a given apparent magnitude ; consequently , given our typical measurement uncertainties , we view it as a reasonable approximation for this analysis to treat all the galaxies in the nirc sample as if they were located at the mean redshift of the entire lbg spectroscopic sample . this of course allows us to use the whole nirc sample , and not just the objects with spectroscopic redshifts , for the luminosity function analysis . two pieces of information are required to construct the rest - frame optical luminosity function for lbgs at @xmath8 : the lbg rest - frame uv luminosity function ( i.e. the distribution of @xmath4 apparent magnitudes ) , and the lbg distribution of @xmath58 as a function of @xmath4 magnitude for the nirc sample . the best - fit @xmath4 apparent magnitude luminosity function parameters for lbgs ( uncorrected for the effects of dust extinction ) are a faint - end slope of @xmath77 , a characteristic apparent magnitude of @xmath78 , and an overall normalization of @xmath79 ( as2000 ) .. ] in 3 , we presented the overall distribution of @xmath58 colors for the nirc sample . a correlation with @xmath80 confidence ( better than @xmath81 ) is detected between @xmath58 color and @xmath4 magnitude , such that fainter galaxies have redder @xmath58 colors ( see figure 4 ) . this correlation probability was computed including the 11 upper limits in the sample of 118 galaxies with @xmath56 measurements , so the result should not be biased by our typical detection limit of @xmath60 . the trend of @xmath5 with @xmath4 is included in the luminosity function analysis by using the relationship implied by the best - fit regression slope to the correlation : @xmath82 . there is a lot of scatter around this regression slope , but it provides a means of encoding the trend between the two variables . we generated a large sample ( 15000 ) of lbg @xmath4 apparent magnitudes between @xmath83 and @xmath84 , by randomly drawing luminosities from the lbg @xmath8 rest - frame uv luminosity function placed at z=2.972 , the mean redshift of the current @xmath0 lbg spectroscopic sample . an @xmath58 color was then assigned to each of the @xmath4 magnitudes , drawn randomly from the distribution of @xmath58 colors for galaxies in the nirc sample with @xmath85 , and then shifted by the amount , @xmath86 , according to the correlation between @xmath4 and @xmath5 . the fiducial @xmath87 distribution was restricted to @xmath88 because this @xmath89 magnitude range contained the largest number of @xmath58 measurements ( 36 ) . also , the @xmath58 measurements for @xmath88 were virtually all detections , rather than mixture of upper limits and detections . combining each randomly generated pair of @xmath4 and @xmath58 measurements , we obtained an ensemble of @xmath2 magnitudes , which were grouped into 0.5 magnitude bins . to determine how the photometric uncertainties of both the @xmath4 magnitudes and @xmath58 colors translated into uncertainties in the derived @xmath2 luminosity function , the procedure of generating an ensemble of random ( @xmath4 , @xmath58 ) pairs was repeated a large number ( 10000 ) of times . in each trial , we perturbed the sample of actual ( @xmath4 , @xmath58 ) measurements by random amounts consistent with the photometric uncertainties . in so doing , we assumed that the errors in @xmath4 and @xmath56 were uncorrelated , which is a valid approximation since the error in @xmath56 for galaxies in our sample is dominated by the @xmath2 photometric error . due to the slight differences in the sample of @xmath4 magnitudes drawn from the lbg @xmath4 apparent luminosity function , and the more significant differences of the perturbed ( @xmath4 , @xmath58 ) measurements , a different @xmath2 distribution was produced in each trial . the analytic schechter function ( 1976 ) was fit to each perturbed distribution of @xmath2-band apparent magnitudes . we restricted the fitted region to span only @xmath90 , reflecting the @xmath2 range for which we have actual measurements . the average number from all the trials was adopted for each 0.5 @xmath2 magnitude bin , and the adopted @xmath91 uncertainty was the standard deviation among the realizations ( which does not include the uncertainties in the lbg @xmath4-band luminosity function best - fit parameters ( as2000 ) ) . the schechter function was then fit to the average luminosity function values . the best - fit parameters we obtain are an overall normalization of @xmath92 , a faint end slope of @xmath93 , and a characteristic apparent magnitude of @xmath94 . the uncertainties on the fitted parameters represent the confidence intervals generated by fitting the schechter function to each perturbed @xmath2 distribution . since it was assumed that all the galaxies comprising the @xmath2 luminosity function were at the mean redshift of the lbg spectroscopic sample , it is straightforward to convert the distribution of @xmath2 apparent magnitudes into a @xmath95 ( i.e. rest - frame @xmath7-band ) absolute magnitude distribution . with our adopted @xmath96 cosmology , @xmath97 corresponds to @xmath98 . the overall shape of the rest - frame optical luminosity function is determined by the way in which the @xmath56 distribution as a function of @xmath4 magnitude redistributes @xmath4 magnitudes into @xmath2 magnitudes . accordingly , the faint end slope of the lbg rest - frame optical luminosity function is steeper than that of the uv luminosity function , due to the positive correlation between @xmath4 and @xmath56 . figure 5 shows the apparent @xmath2 luminosity function ( and the absolute optical magnitudes to which it corresponds ) , as well as the confidence intervals on the best - fit schechter luminosity function parameters . immediately apparent from figure 5 is the fact that the well - constrained bright end of the lbg luminosity function greatly exceeds locally determined optical luminosity functions . to calculate the rest - frame optical co - moving luminosity density for lbgs brighter than @xmath99 , we integrate over the luminosity - weighted luminosity function in the appropriate magnitude range to obtain @xmath100hz@xmath101 . we use the @xmath7-band absolute magnitude of the sun , @xmath102 ( binney & merrifield 1998 ) , to express this quantity in solar units : @xmath103 . if the luminosity function were integrated to @xmath104 , the derived luminosity density would be more than 2 times larger than the value determined down to the typical survey detection limit of @xmath99 . however , such an extrapolation depends sensitively on the value of the faint end slope of the luminosity function , which is poorly constrained by our observed sample . et al . _ 1998 summarize recent local determinations of the @xmath105-band luminosity density , most of which are in the range @xmath106 . thus , the lbg @xmath7-band optical luminosity density down to only @xmath99 is within a factor of 2 of local measurements of the @xmath105-band optical luminosity density , which were obtained by integrating over the entire @xmath105-band luminosity function . the optical luminosity function of local galaxies has been computed from numerous surveys over the past thirty years . some recent determinations from magnitude limited redshift surveys include the two - degree field galaxy redshift survey ( 2dfgrs ) @xmath107 luminosity function ( folkes _ et al . _ 1999 ) , the las campanas redshift survey ( lcrs ) @xmath108-band luminosity function ( lin _ et al . _ 1996 ) , and the sloan digital sky survey ( sdss ) luminosity functions in five optical bands ( blanton _ et al . _ these local optical luminosity functions have much shallower faint end slopes @xmath109 , fainter characteristic luminosities @xmath110 , and higher overall normalizations @xmath111 than the rest - frame optical luminosity function of lbgs . to demonstrate these differences , we plot the 2dfgrs @xmath107 luminosity function in figure 5 along with the lbg rest - frame optical luminosity function . the best - fit parameters of the 2dfgrs luminosity function are @xmath112 , @xmath113 , and @xmath114 . for a direct comparison with the lbg luminosity function , which specifically probes the rest - frame @xmath7 band , the 2dfgrs luminosity function has been shifted 0.5 magnitudes brighter ( i.e. @xmath115 ) , reflecting the typical @xmath116 color of local galaxies . at the brightest absolute magnitudes , where the lbg @xmath7-band luminosity function is well - constrained by our observations , the @xmath8 lbg @xmath7-band luminosity function greatly exceeds the local function , and has a different shape . for this reason , we caution against over - interpreting the differences in the best - fit schechter function parameters relative to other luminosity functions ; for example , the very steep faint end slope , very bright value of @xmath117 , and relatively low value of @xmath118 may well result from fitting the function over a small range of luminosity that does not actually extend as faint as the true `` knee '' in the luminosity distribution . in practice , it will be extremely difficult to extend this distribution to much fainter near - ir magnitudes from the ground , but it should be quite possible from space . the obvious implication of the bright end of the @xmath7-band luminosity function of lbgs is that the mass - to - light ratio @xmath119 of lbgs must be very different from that of galaxies in the local universe . since the co - moving optical luminosity density in the @xmath8 universe is evidently as large or larger than that in the @xmath120 universe despite the fact that most of the stars observed in the local universe probably formed after @xmath0 , the @xmath7 band light at @xmath8 is apparently dominated by relatively massive stars associated with greatly enhanced current star - formation rates , and carries little information about total stellar mass . unfortunately , this also means that no observations from the ground are likely to act as a direct proxy for stellar mass in objects that are actively forming stars at high redshift . indeed , these observations , at @xmath121 @xmath122 m , are still dominated by the light from ongoing star - formation . in any case , _ the @xmath7-band luminosity density of lbgs at @xmath123 probably exceeds that of all stars in the present - day universe_. given photometry that extends from @xmath124 in the rest - frame for a reasonably large sample of @xmath8 lbgs , it is of interest to use the data to constrain the star - formation histories and extinction properties of the galaxies with reference to models . because the galaxy luminosity at even the longest wavelengths observed in our sample is still dominated by stars with small mass - to - light ratios , it is impossible to constrain the integrated stellar mass of galaxies without understanding something about the star - formation history . as it turns out , our sensitivity to the star - formation history using optical and near - ir data is coarse , at best at the corresponding rest - frame far - uv to optical wavelengths , active star - formation can easily mask the evidence for previous generations of stars ( see , e.g. , papovich _ et al . _ 2001 ) , and dust extinction has a significant effect on the emergent sed . in order to interpret the color information of the lbgs in terms of interesting constraints on physical parameters , we must rely on models of both dust extinction _ and _ the intrinsic spectral energy distributions of stellar populations as a function of metallicity , initial mass function ( imf ) , star - formation history , and age . there is a large range of possible parameterizations of the extinction law and the intrinsic stellar population , and in general these can not be well - constrained independently from fits to the galaxy seds . consequently , in what follows , we adopt simple models for the attenuation law , and for the metallicity , imf , and star - formation history of the stellar population , which external observations have shown to be reasonable for star - forming galaxies in the nearby and high redshift universe . where relevant , we point out where different assumptions yield significantly different results . extensive grids of models have been fitted to ( generally fainter ) lbgs observed in the hdf - north by papovich _ et al . _ 2001 ; we refer the reader to that paper for a more detailed discussion of the effects of varying the modeling assumptions . with only optical ( @xmath19 , @xmath20 , @xmath4 , and @xmath125 ) photometry for @xmath8 galaxies , it is not possible to distinguish between a young galaxy spectral energy distribution ( @xmath126 myr ) which is considerably reddened by dust , and the spectral energy distribution of a galaxy which has been continuously forming stars over a much longer time scale ( @xmath127 gyr ) , but with much less dust extinction ( sawicki & yee 1998 ) . figure 6 shows the results of fitting only the @xmath27 color of the galaxy 3c324-c2 with bruzual & charlot ( 1996 ) constant star - formation galaxy spectral templates of different ages . as the template age increases , the amount of dust reddening necessary to fit the observed @xmath27 color decreases . both the 1 myr model with @xmath128 and the 1 gyr model with @xmath129 fit the observed optical photometry equally well ( where @xmath130 is magnitudes of extinction at @xmath131 rest - frame ) . indeed , the model rest - frame uv slope can be modified to match the observed ( @xmath19 , @xmath20 , @xmath4 , @xmath125 ) colors , by simultaneously tuning its age and dust - content . the degeneracy results from the fact that , at @xmath8 , commonly used optical filters probe a short - wavelength region of the rest - frame spectrum which does not contain any age - sensitive features that can be easily recognized from broad - band photometry . however , there is an age - sensitive break in the spectral energy distributions of actively star - forming galaxies that occurs near @xmath132 , often referred to as the `` balmer break '' . the balmer break feature , which is is age sensitive over star - formation timescales of @xmath133 myr@xmath134 gyr , can not be described equally well by any combination of stellar population aging and dust - content . longer wavelength photometry which includes the discontinuity can in principle be used to break the degeneracy between model fits . the balmer break feature in galaxy spectra is due to the stellar bound - free absorption coefficient , which has a sharp edge at @xmath132 , ( the wavelength of the @xmath135 transition in neutral hydrogen ) , causing photons with wavelengths shortward of the edge to be absorbed much more efficiently than those with longer wavelengths . the prominence of this feature in a stellar spectrum depends on the temperature and gravity of the stellar atmosphere , which determines the form of hydrogen that is prevalent in the stellar atmosphere , and therefore which component of the total stellar absorption coefficient is dominant . in cooler late - type stars ( such as f , g , and k ) , @xmath136 opacity dominates the total absorption coefficient , and the balmer decrement is not very pronounced . in the early - type massive o and b stars the high temperatures result in the increased ionization of hydrogen , so again , the balmer absorption coefficient is reduced . the balmer break is most prominent in a - stars with temperatures around @xmath137 ( gray 1976 ) . at the onset of star - formation , a galaxy spectrum is dominated in the rest - frame uv and blue by the most - massive ( o and b ) stars , most of which have lifetimes shorter than 10 myr . in a continuous star - formation model , after roughly 10 myr , the rate at which o stars are being formed balances the rate at which they are dying , and the number of o stars levels off to a constant value ( leitherer _ et al . _ 1999 ) . meanwhile , the lifetimes of less massive a stars range from several hundred myr to 1 gyr therefore , as the o star population levels off , the number of a stars continues to increase . at the age where the increasing number of a stars relative to the constant number of o stars causes the galaxy spectrum at @xmath138 to be dominated by an a - star spectrum , we start to see an increasing balmer break in the galaxy spectrum . after 100 myr the magnitude difference on either side of the balmer break in an unreddened galaxy model is 0.3 mag , and by 1 gyr , the break has increased to a 0.6 magnitudes . thus , the strength of the balmer break can be used as a rough guide to the duration of the current episode of star - formation on timescales @xmath139 . the details will depend on the star - formation history ( e.g. , whether the star - formation rate ( sfr ) is an increasing , or declining , function of time ) , the shape of the initial mass function , and the metallicity of the stellar population . because the balmer break even at its strongest is a relatively subtle spectral feature , and because only broad - band photometry is available for the galaxies , a measurement of the balmer break strength requires removing the effects of continuum reddening . this requirement introduces some covariance into the simultaneous estimates of extinction and the age of the star - formation episode ; we discuss this covariance extensively in 5.3.2 . in order to examine the range of stellar population properties present in the nirc sample of lbgs , we considered only the subsample of galaxies with measured redshifts and near - ir detections at least in @xmath2 . the spectroscopic sample so defined consists of 81 galaxies . we included the @xmath27 , @xmath140 , and @xmath5 colors of these galaxies in the modeling ; the @xmath19 band is strongly affected by h i absorption in both the galaxy and the igm . a substantial fraction of the spectroscopic sample also has available i band photometry , but for these high redshift objects the i band adds little information because of its proximity to @xmath4 and because the measurements are generally noisier than in other bands . nevertheless , @xmath19 and @xmath71 data points ( when they exist ) are included in figures indicating the best - fit spectral template with respect to the observed lbg photometry , the model galaxy templates used to fit the lbg photometry include no lyman @xmath141 emission , and typically contain lyman @xmath141 absorption with observed equivalent widths of @xmath142 ( @xmath143 in the rest - frame ) . however , the sample of lbgs for which there are rest - frame uv spectra exhibit a wide range of lyman @xmath141 equivalent widths , some of which are ten times larger than the model line widths , and appear in either absorption , emission , or a combination of both . the _ observed _ lyman @xmath141 equivalent width was used to correct the @xmath27 color for line emission for each of the 81 galaxies in the spectroscopic sample . figure 7 displays the distribution of lyman @xmath141 observed equivalent widths for the lbgs with both nirc data and keck lris spectra ( cf . et al . _ 2000 nebular line emission redshifted into the @xmath2-band might also bias the results of population synthesis modeling . the [ oiii ] @xmath144 doublet and @xmath145 @xmath146 line all fall within the @xmath2-band window for @xmath147 . in the nirc lbg spectroscopic sample , 49 out of 81 galaxies ( @xmath148 ) have @xmath149 such that at least [ oiii ] @xmath150 , the strongest of the three lines , is located within the @xmath2-band . nebular line strengths have been measured for only a small sample of lbgs at the current time ( see pettini _ et al . _ 2001 ) , nine of which are in the nirc sample . the combined equivalent width from all the nebular lines affecting the @xmath2 magnitude was calculated by comparing the line fluxes from pettini _ et al . _ 2001 to the continuum flux , measured with nirc . for each galaxy , the total nebular equivalent width in the @xmath2 filter was compared to the effective width of the @xmath2 filter , @xmath151 , to evaluate the necessary correction to the @xmath2 magnitude and @xmath5 color . in every case , the indicated corrections for nebular emission reduce the @xmath2 continuum flux , resulting in a bluer @xmath5 color . the @xmath5 corrections ranged from @xmath152 magnitudes with a median @xmath153 . in order to determine the effect of these @xmath5 color corrections on best - fit model parameters , we fit bruzual & charlot ( 1996 ) constant star - formation models to the colors of the nine galaxies with measured nebular equivalent widths . models were fit to both the uncorrected colors and the colors corrected for nebular line emission . the @xmath5 correction resulted in little or no change to the best - fit @xmath154 value ( since most of the lever arm for extinction estimation comes from the rest - uv data ) . the effect on the best - fit age was more significant , in that the nebular line correction will result in a systematically younger inferred age . the median correction for the nebular line contamination results in stellar population ages that are @xmath15535% younger than if no corrections for line emission are applied . it should be emphasized that nebular line contamination is negligible for the 40% of the spectroscopic nirc sample with @xmath156 . with such a small sub - sample having near - ir spectroscopy at the time of this writing , there is not enough information available to determine how corrections for emission line contamination should be implemented for objects without near - ir spectroscopic measurements . we have chosen , therefore , not to apply these corrections , but to emphasize that in some cases the inferred ages may be influenced by the emission line contamination of the broad - band @xmath2 measurements . the two main theoretical elements which we used were ( 1 ) model galaxy spectra generated with the bruzual & charlot 1996 , ( hereafter bc96 ) population synthesis code , and ( 2 ) a dust effective attenuation law empirically derived by calzetti ( 1997 ) for local starbursts . , in units of @xmath157 , to the intrinsic flux , @xmath158 , with the relation : @xmath159 , where @xmath160 is a decreasing function of wavelength , reflecting the fact that shorter wavelengths suffer more extinction than longer wavelengths . the attenuation curve is parameterized by an overall normalization , @xmath154 . ] there are few constraints on the appropriate reddening model to use for high redshift star forming galaxies , and so we have adopted the form that works quite well for local starburst galaxies , which are arguably the best local analogs of lbgs . the validity of this reddening / extinction law at high redshift has yet to be tested adequately , due to the difficulty in measuring bolometric star - formation luminosities at high redshift ( see meurer _ et al . _ 1999 ; as2000 ) . we simply use it as the most realistic representation of the attenuation in in @xmath8 lbgs ; we will briefly discuss how the results depend on the assumed attenuation law in 3.2 . at the time this work began , the newest bruzual and charlot software available to us was the bc96 version , so that the results discussed in detail are based on bc96 model fits . in order to verify that the population synthesis fits did not change significantly with the most recent version of the bruzual and charlot package , bc2000 ( charlot 2000 , private communication ; liu , charlot , & graham 2000 ) , we compared the bc96 and bc2000 results for a subset of models . bc96 results were also compared with those from the starburst99 models ( leitherer _ et al . _ 1999 ) . models with both solar and sub - solar metallicity were fit to the observed colors of lbgs , but we discuss results for only solar metallicity models . recent measurements of lbg metallicities are not very well constrained , ranging from @xmath161 ( pettini _ et al . _ 2001 ) , indicating that these galaxies are more metal - rich than damped lyman @xmath141 systems at the same redshift , but have slightly lower metallicites than h ii regions in the milky way near the sun . even so , we prefer to use solar metallicity stellar models , since they are the only models which have been directly calibrated against empirical stellar spectra of many spectral types . specifically , in the grid of theoretical stellar spectra compiled by lejeune , buser , and cuisinier ( 1996 , 1997 ) , which was used for the bc96 modeling , the only spectra which have been extensively tested against empirical stellar spectra are those with solar abundance . et al . _ 2001 report that the effects of using @xmath162 rather than solar metallicity models to fit lbg colors include best - fit ages systematically younger by a factor of 2 , best - fit uv extinction factors systematically higher by a factor of 3 , and derived stellar masses smaller by a factor of 2 . these effects stem from the fact that the uncalibrated sub - solar metallicity models have intrinsically bluer uv continuum slopes and slightly larger balmer breaks than solar metallicity models , for a given stellar population age . however , leitherer _ et al . _ 2001 recently presented observations of o stars in the large and small magellanic clouds with an average metallicity of @xmath163 , showing that metallicity does not have very drastic effects on the empirical rest - frame uv spectra of hot stars . et al . _ concluded that modeling the spectra of @xmath164 galaxies with solar metallicity models is an approximately valid approach . until theoretical stellar libraries are updated with empirically calibrated sub - solar metallicity models of many spectral types , we believe that solar metallicity models are preferable . for all of the bc96 models , a salpeter imf extending from @xmath165 was assumed . et al . _ ( 2001 ) have investigated the effects of varying the imf on the estimated stellar population parameters . acceptable model fits are obtained for imfs with various slopes and lower mass cutoffs , with slightly different results . for example , use of the steeper scalo imf results in younger ages , lower attenuation values , higher star - formation rates , and larger formed stellar masses . the effect of changing the imf on the formed stellar mass is small , however , compared with the associated uncertainties . we choose not to include the form of the imf as another parameter in our model , although clearly the shape of the imf , if allowed to vary arbitrarily , can have a very significant effect on the stellar population parameters . evolutionary tracks distributed by the `` padova school '' ( the default for bc96 ) were used to describe the evolution of all the formed stars through the h - r diagram . several different simple star - formation histories were considered : a continuous star - formation rate , @xmath166 ; and exponentially declining star - formation rates , @xmath167 , with @xmath168 . the set of time constants , @xmath169 , was chosen to span the range between the two extremes of an instantaneous burst and continuous star - formation . for each star - formation history , two parameters were allowed to vary : 1 ) the _ extinction _ , parameterized by @xmath154 , and 2 ) the _ age _ since the onset of the most recent episode of star - formation , @xmath170 . the model dust - attenuated , rest - frame galaxy spectra were shifted to the measured redshift of the actual galaxy , and further attenuated in a manner simulating absorption by the intergalactic medium of neutral hydrogen ( madau 1995 ) . each internally reddened , redshifted , igm - absorbed galaxy spectrum was then integrated through the ( @xmath171 , @xmath3 , and @xmath2 ) filter transmission curves to calculate its predicted colors . for each @xmath172 combination , the predicted ( @xmath27 , @xmath140 , and @xmath5 ) colors were compared with the observed colors using the @xmath173 statistic . the best - fit @xmath172 combination was chosen to minimize @xmath173 , and the intrinsic instantaneous star - formation rate was determined by normalizing the best - fit model to the observed @xmath4 magnitude ( taking into account the best - fit extinction ) . by our definition , the `` formed stellar mass '' @xmath174 is simply the integral of the star - formation rate and the age of the star - formation episode obtained from the best - fit model . once the best - fit @xmath154 , @xmath170 , @xmath175 , and @xmath174 , were obtained for the galaxies in the spectroscopic sample , we computed the @xmath176 and @xmath177 confidence intervals associated with each parameter . for each galaxy we generated a large sample of fake `` observed '' colors , by perturbing the observed colors in a manner consistent with the photometric errors . we assumed the errors were gaussian , based on the results of monte carlo simulations ( 2.4 ) . we also assumed that the color errors were uncorrelated , a valid assumption given that the near - ir uncertainties tend to be much larger than the @xmath4 uncertainty . the large sample of fits to the perturbed colors indicated the region of @xmath178 parameter space allowed by each galaxy s observed colors and photometric uncertainties . as described above , a range of bc96 star - formation histories was considered , and the differences between bc96 , bc2000 , and starburst99 were investigated . for the purpose of this analysis , and given our relatively large photometric uncertainties , the bc96 and bc2000 model fits are virtually indistinguishable . for the solar metallicity constant star - formation models , the bc2000 model seds have slightly bluer rest - frame far - uv slopes and slightly larger balmer breaks than the bc96 seds for a given star - formation history . therefore , best - fit ages from bc2000 are typically @xmath179% as old , while extinction values are larger ( @xmath180 ) . the bc2000 best - fit extinction - corrected star - formation rates @xmath175 are typically @xmath181 larger than in the bc96 models , and so the bc2000 @xmath174 is typically @xmath182 of the bc96 @xmath174 . these differences are insignificant , given the large confidence intervals allowed for the best - fit and derived parameters for any given set of models . the differences between the bc96 and the starburst99 constant star - formation models were also explored . the starburst99 models use stellar evolutionary tracks from the `` geneva school '' , and the model atmosphere grid of lejeune _ et al . for the sake of consistency , we chose solar metallicity starburst99 templates with a salpeter imf , which we re - normalized by roughly a factor of @xmath89 in luminosity , to remove the effects of the larger minimum mass cutoff in the starburst99 salpeter imf @xmath183 . as in the case of bc2000 , the starburst99 best - fit @xmath154 and @xmath170 parameters tended towards younger ages ( @xmath184% of the bc96 @xmath170 ) , and slightly larger values of @xmath154 , ( @xmath185 ) . the higher @xmath154 values in the starburst99 fits lead to extinction - corrected @xmath175 values which are @xmath186 larger than the bc96 values on average , and therefore the @xmath174 values for the starburst99 models were typically @xmath187 of the bc96 @xmath174 values . the differences between the bc96 and starburst99 @xmath154 and @xmath170 parameters are also small compared with the allowed confidence intervals . for all three sets of models : bc96 , bc2000 , and starburst99 , the relative ordering of @xmath154 and @xmath170 is preserved galaxies best described by higher @xmath154 [ @xmath170 ] galaxies in the bc96 models also have higher @xmath154 [ @xmath170 ] according to the starburst99 and bc2000 models . finally , we note that for each galaxy in our sample , the bc96 , bc2000 , and starburst99 constant star - formation models all provide statistically indistinguishable fits i.e. there are no cases where one set of models has significantly lower @xmath173 values than the others . evidently , even for identical star - formation histories , there are significant differences in the predictions of each of the models that can amount to a factor of 2 in inferred age and a factor somewhat less than 2 in formed stellar mass . thus , it would be a mistake to take the results from any of the models as anything but approximate until both the models and the data improve . the model results for different bc96 star - formation histories were also compared in detail . we did not consider a very large number of star - formation histories , but enough to bracket a range of star - formation decay time @xmath189 parameter space . for the majority of the galaxies successfully modeled , _ all _ of the star - formation histories considered yielded best - fit models with @xmath173 values allowed at the @xmath177 level . for 8 of the galaxies we were able to rule out the models with time constants smaller than @xmath190 myr this sub - sample is comprised of galaxies which are simultaneously blue in @xmath27 and red in @xmath5 , and which have among the largest inferred continuous star - formation ages . for the rest of the galaxies , whose star - formation histories can not be usefully constrained , the best - fit @xmath154 and , especially , @xmath170 , parameters demonstrate quite a large spread depending on which star - formation history is chosen to fit the photometry . in contrast , the formed stellar mass , @xmath174 , derived from such best - fit parameters is a more robust quantity ( cf . et al . _ we investigated the @xmath174 implied by each best - fit @xmath172 combination as a function of star - formation history , and find that for the @xmath191 and @xmath192 models , the @xmath174 derived is typically within a factor of 2 of the @xmath174 derived for the constant star - formation model . for the @xmath193 models ( these are effectively equivalent to an `` instantaneous burst '' model ) , the derived @xmath174 is typically a factor of four smaller than the constant star - formation prediction this relatively large difference is caused mostly by the degeneracy between an aging starburst and one which is relatively strongly reddened by dust . star - formation histories with star - formation rates that are _ increasing _ functions of time pose even more significant problems for the models ; for example , papovich _ et al . _ ( 2001 ) have shown that an underlying maximally old stellar population ( hidden by the most recent episode of star - formation ) can contain up to @xmath194 the stellar mass derived from the rest - frame uv - to - optical sed . in this sense , as emphasized by papovich _ _ , the inferred @xmath174 based on the modeling should be treated as lower limits on the total stellar mass of the galaxies . in short , broad - band photometry is not particularly effective in constraining star - formation histories the root cause of this problem is that the far - uv flux is determined solely by the instantaneous rate of o and b star - formation , whereas the rest - frame optical flux is dominated by previous star - formation if the sfr is declining , and by the instantaneous star - formation if the rate is increasing . there is of course no good _ a priori _ reason to favor declining star formation rates over increasing ones . the simplest assumption is that the observed instantaneous star - formation rate is representative of the rate since the onset of the current episode of star - formation . under these simplifying assumptions , short - lived `` bursts '' of star - formation for which the sfrs are much higher than the past average , will appear `` young '' , and protracted periods of star - formation with relatively constant sfrs on timescales @xmath195 myr will have significant balmer breaks that can be recognized in the broad - band photometry . we summarize the best - fit and derived parameters for each galaxy assuming the bc96 constant star - formation models in table 4 . in figure 8a , the measured photometry for three nirc lbgs is plotted along with the best - fit constant star - formation models for the galaxies . these examples demonstrate the range of stellar population parameter space which is probed with nirc lbgs . figure 8b shows the @xmath196 confidence regions for these three galaxies . most of the nirc lbg spectroscopic sample was adequately described by the simple models detailed above . statistically acceptable fits were obtained for 74 of the 81 lbgs in the nirc sample with redshifts . three of the remaining seven galaxies have suspect @xmath3 and @xmath2 photometry upon closer inspection , and fourth galaxy has a clear seyfert-2-like spectrum . the remaining galaxies have colors that defy the simple models considered above . two of the most anomalous galaxies are westphal - mmd11 and dsf2237b - md81 . mmd11 and md81 are the two reddest galaxies in @xmath56 and @xmath44 in the entire nirc lbg sample , with @xmath5 colors of 4.54 and 4.27 , respectively , and @xmath44 colors of 2.67 and 2.23 . mmd11 is also in the reddest quartile of igm - absorption corrected @xmath27 measurements for the sample , while md81 is in the bluest quartile ( see figure 1b ) . neither mmd11 nor md81 has a lris rest - frame uv spectrum with anomalously strong lyman @xmath141 emission or absorption , or notable interstellar or stellar features ( though the signal - to - noise ratio in these spectra does not enable a quantitative examination ) . as discussed by as2000 , at @xmath197 ( chapman _ et al . _ 2000 ) , the @xmath198 flux of mmd11 is the strongest detection from the sample of optically selected lbgs that have been observed with scuba . furthermore , mmd11 has a complex [ oiii ] 5007 nebular line profile , extended into two components along the spatial direction . while there is substructure present , the [ oiii ] equivalent width is one of the smallest which has been measured in lbgs ( pettini _ et al . _ 2001 ) , and could not have contaminated the @xmath5 colors by any significant amount . md81 is not detected at 850@xmath122 m at the @xmath199mjy level ( chapman _ et al . _ , in preparation ) , and has not been studied with near - ir spectroscopy . when we attempt to fit simultaneously the @xmath27 , @xmath44 , and @xmath5 colors for either mmd11 or md81 , we are unsuccessful , mainly due to the extremely red optical - ir colors . since mmd11 and md81 are so bright in the near - ir , their well - determined colors have more power to discriminate the quality of a model fit than the more poorly determined optical - ir photometry for the majority of the nirc sample however , even if mmd11 and md81 had @xmath44 and @xmath5 uncertainties which were more typical of the nirc lbg sample , no satisfactory model fits would be obtained for md81 , and only marginally acceptable fits would be obtained for mmd11 . figure 9 shows various attempts to fit the colors of these anomalous galaxies . estimates of dust reddening are obtained from fitting only the @xmath200 color , assuming bc96 constant star - formation , and ages of 1 myr and 1 gyr . clearly , the @xmath3 and @xmath2 points for both galaxies are significantly brighter than the predictions for even the 1 gyr model . while a model with an age older than 1 gyr might be able to fit the @xmath2 points , the balmer break present between @xmath3 and @xmath2 in such an old template spectrum would not be consistent with the well - determined , and bright , @xmath3 measurements . for most of the galaxies in the nirc lbg sample , this is not the case even when fitting the @xmath27 color alone , it is easy to find a model with the appropriate age and dust extinction to describe all of the photometry within the errors . if we neglect the @xmath201 color , and fit @xmath44 and @xmath5 alone , the best - fit models predict more than 7 magnitudes of extinction in the rest - frame uv and would significantly under - predict the rest - frame far - uv fluxes . the very red optical - ir colors of these galaxies may indicate a problem with the calzetti extinction law in cases where dust extinction is very extreme . in these cases , as in the case of local ultra - luminous infrared galaxies ( ulirgs ) , the details of the geometry and distribution of the dust relative to where star - formation is occurring become more important and the simple recipe for extinction breaks down ( trentham , kormendy , and sanders 1999 ; meurer and seibert 2001 ) . for example , based on the @xmath27 color alone , we would infer a moderate to low amount of dust extinction in md81 ( which may still be true ) . but such a small amount of extinction , with any age or star - formation history considered in this work , can not account for the very red @xmath44 and @xmath56 colors of md81 . alternatively , the inability to model these unusual galaxies might be due to the over - simplified nature of the star - formation histories considered . using the sample of 74 galaxies for which acceptable bc96 continuous star - formation fits were obtained , we explore the trends present among observed and modeled quantities . these galaxies , all of which have redshifts , are brighter by an average of @xmath202 magnitudes in @xmath4 than the nirc lbg sample as a whole , and brighter by @xmath203 mag than the larger optical lbg photometric sample . it should be emphasized that the sample of hdf - n lbgs recently analyzed by papovich _ et al . _ ( 2001 ) includes only four ( of 33 ) galaxies brighter than @xmath16 in the rest - frame uv lbg luminosity function , whereas @xmath204% of the galaxies in the sample analyzed here are brighter than @xmath16 . differences between the two samples might then be ascribed to uv - luminosity - dependent effects , although we see only subtle trends with uv luminosity within our full ground - based sample spanning a factor of @xmath205 in uv luminosity . steidel _ et al . _ 1999 and as2000 have computed the distribution of lbg @xmath154 values based on igm - absorption - corrected @xmath27 colors alone , assuming a fixed intrinsic stellar population that corresponds to a bc96 continuous star - formation model with @xmath206 . with the addition of near - ir data , we have been able to model the extinction and variations in the stellar populations simultaneously . the inferred distribution of reddening , assuming the starburst attenuation relation of calzetti ( 1997 ) and a suite of bc96 continuous star - formation models , has a median @xmath207 , corresponding to @xmath208 magnitudes , similar to the median extinction presented in the earlier works . this reflects the fact that the previously - assumed galaxy sed is a reasonable approximation for the bulk of the sample in the rest - frame uv , and that most of the information on reddening comes from the far - uv data points even for a relatively `` gray '' extinction law . the median age for our sample is @xmath209 . more than @xmath210 of the galaxies have @xmath211 , while @xmath212 have @xmath213 . there were nine galaxies with formal best - fit @xmath170 values smaller than @xmath214 . such small ages are not physically plausible for episodes of star - formation , given the dynamical timescales of the star - forming regions ( see 7 ) . furthermore , the conversion between extinction - corrected uv luminosity and star - formation rate becomes highly non - linear at ages shorter than @xmath214 . on these grounds , we restricted the best - fit constant - star - formation @xmath170 parameter space to ages of at least @xmath214 , which still provides acceptable fits for the nine youngest galaxies . we also applied the constraint that best - fit @xmath170 values could not be older than the age of the universe at @xmath0 ( assuming an @xmath215 , @xmath216 , @xmath217 cosmology ) , which affected the fits for seven galaxies whose formal best - fit @xmath170 values were then too old . these galaxies , however , all had colors which were statistically consistent with being younger than the age of the universe at @xmath0 . as discussed in 5.2.3 , the @xmath170 values for the nirc lbg sample are affected in a systematic way by varying the star - formation history or the population synthesis models , but the sample as a whole should broadly represent the significant range in these parameters spanned by the lbgs at @xmath8 . we have noted that the distribution of @xmath170 values does not reflect corrections to the @xmath58 color for nebular line emission for the @xmath148 of the galaxies in the nirc lbg sample which are at high enough redshift that the @xmath2 measurement could be affected . however , there is no apparent systematic offset towards higher @xmath170 values for the higher - redshift portion of the sample as might be expected if the higher redshift objects were significantly biased due to nebular line contamination . histograms of best - fit @xmath170 and @xmath154 are shown in figure 10a . the median derived star - formation rate for the sample is @xmath218 , and the distribution of values is quite broad , ranging from @xmath219 to as high as @xmath220 . multiplying the best - fit @xmath170 and @xmath175 yields a distribution of @xmath174 , with a median of @xmath221 , smaller but comparable to the present - day @xmath16 stellar mass , @xmath222 ( cole _ et al . _ 2000 ; papovich _ et al . _ 2001 ) . however , we find that about @xmath223 of the sample have @xmath174 smaller than @xmath224 ( roughly the same sample with best - fit @xmath225 ) . histograms of the inferred @xmath174 and @xmath175 distributions are shown in figure 10b . as mentioned above , a similar modeling technique for extracting physical parameters from observed colors of lbgs was used in recent work by papovich _ et al . _ ( 2001 ) ; their sample of 33 galaxies in the hdf - n is a superset of earlier work by sawicki & yee ( 1998 ) on 17 hdf - n galaxies . the papovich _ et al . _ hdf study , while containing all of the sawicki & yee galaxies , finds somewhat different results from sawicki & yee , including lower typical extinction factors , older ages , and larger formed stellar masses . the authors attribute some of these differences to their higher - quality hst nicmos @xmath3 and @xmath226 data ( compared to the sawicki & yee ground - based @xmath3 and @xmath226 ) , more accurate matching of optical and infrared photometry to compute colors , and larger galaxy sample . here and above we have focused primarily on the more recent hdf - n analysis of papovich _ et al . _ ( 2001 ) . the main difference between our modeling procedure and that of papovich _ et al . _ is that the latter work allowed @xmath169 , the exponential star - formation decay time constant , to vary as a free parameter in fitting the models . as discussed above , there were few cases in which our data could discriminate between different values of @xmath169 that were significant compared to the inferred value of @xmath170 . however , a direct comparison of the model results is possible for 19 of the 33 galaxies in the papovich _ et al . _ work because the best - fit @xmath169 values are either longer than or much shorter than the inferred @xmath170 , so that the star - formation history is essentially indistinguishable from a constant star - formation model . for these 19 hdf - n galaxies , the median inferred age is @xmath227 , and the median @xmath228 . the median @xmath170 is thus higher , while similar , to the nirc lbg median @xmath170 , and the median @xmath154 is slightly smaller than that of the nirc lbg sample . the remaining 14 hdf - n galaxies have best - fit @xmath229 , and @xmath170 values which are systematically lower than the @xmath230 sample . however , the @xmath169 values for these galaxies are not very well constrained , and larger @xmath169 values also provide acceptable fits . the older @xmath170 values resulting from larger @xmath169 parameters would be roughly consistent with the @xmath170 values for the other 19 galaxies . thus , given the uncertainties , it is probably fair to compare broadly the whole papovich _ _ sample to the nirc - lbg sample . our results are consistent with the inferred parameters for the small subset of the brightest hdf galaxies , including two galaxies in common between the two samples ( see table 3 ) . generally speaking , the uv - brighter nirc lbg sample has a higher proportion of objects with large inferred extinction and with large values of the inferred stellar mass ( these are generally distinct sets of objects , as discussed further below ) , both of which may be due to the different range of uv luminosity spanned within the two samples . using observed optical and far - infrared ( fir ) data , as2000 and meurer _ et al . _ ( 1999 ) demonstrated that at both low and high redshifts , there appears to be a correlation between dust - obscuration and bolometric luminosity ( @xmath231.5ex@xmath232 ) . the results of our modeling of the lbg seds support the assertion that more heavily obscured galaxies have larger bolometric luminosities ; however , because many of our assumptions are similar most notably , the adopted starburst attenuation relation , our new results can not be taken as independent evidence . as in as2000 , we find that the rest - frame uv luminosity is uncorrelated with the inferred value of @xmath154 , so that applying the inferred extinction correction naturally results in a strong correlation between bolometric luminosity and extinction . as emphasized by as2000 , inferences about extinction in the high redshift galaxies are very difficult to test observationally , but the trends are very similar to what is observed for star - forming galaxies in the local universe and , so far , the starburst attenuation relation is consistent with available cross - checks at high redshift . we note in passing that internally consistent application of the inferred extinction corrections to the rest - frame optical luminosities results in a similar correlation of dust obscuration and optical luminosity , and a net correction to the @xmath7-band luminosity density of a factor of @xmath233 compared to the uncorrected numbers presented in 4 . the correlations of the uv and optical luminosities with best - fit @xmath154 are shown in figure 11 for bc96 continuous star - formation models . these correlations hold for the all of the star - formation histories used in this work to model the galaxy colors , and would hold for any assumed extinction relation that is correlated with rest - uv color . in figure 12 we plot the best - fit @xmath154 and @xmath170 parameters for the 74 galaxies successfully fit with bc96 constant - star - formation models . there is a very strong anti - correlation between @xmath154 and @xmath170 in the sense that younger galaxies are more heavily extinguished than older galaxies . given the strong correlation between @xmath175 and @xmath154 discussed in 5.3.1 , the relationship between extinction and age translates into a link between age and star - formation rate , where the youngest galaxies also have the highest star - formation rates . the correlation in the sample between @xmath154 and @xmath170 falls roughly in the same direction as the observed covariance of the ( @xmath234 ) parameters for each individual galaxy . the covariance of @xmath154 and @xmath170 is dominated in most cases by response of the fitted parameters to perturbations in the @xmath27 color . typically , a positive perturbation to the @xmath27 results in a _ positive _ perturbation to the best - fit @xmath154 and a _ negative _ perturbation to the best - fit @xmath170 , the latter due to the effect of the extinction estimate on the predicted @xmath235 color for a given @xmath170 . the covariance of the two parameters is clearly demonstrated by the nirc lbg confidence regions shown in figure 8b . given the strong covariance between best - fit @xmath154 and @xmath170 parameters inherent in the modeling procedure , further tests are necessary to evaluate whether the trend of increasing extinction with decreasing @xmath170 is significant . in other words , does the apparent correlation arise solely due to objects covariant confidence regions in @xmath196 space , because measurement errors can scatter intrinsically uncorrelated points along the same direction as an @xmath196 correlation ? in order to test the strength of such an apparent correlation , we constructed a sample of 74 uncorrelated @xmath154 and @xmath170 pairs ( to match the size of the observed sample of 74 galaxies ) . this sample was generated by randomly selecting @xmath154 and @xmath170 values independently from the _ intrinsic _ marginal distributions of each of the parameters . these _ intrinsic _ marginal distributions were estimated by assuming that the _ observed _ marginal distributions ( shown in figure 10a ) represented the _ intrinsic _ marginal distributions broadened by photometric measurement errors . in addition to @xmath154 and @xmath170 , @xmath4 apparent magnitude and @xmath236 were randomly selected from the observed distributions of @xmath4 and @xmath236 for nirc lbgs , in order to compute the colors that would be observed from a galaxy with the randomly selected @xmath154 and @xmath170 parameters . a fake galaxy was retained only if its @xmath154 and @xmath170 parameters implied a @xmath27 color satisfying the lbg selection criterion of @xmath237 , and a @xmath2 magnitude brighter than the typical nirc lbg detection limit of @xmath60 . photometric uncertainties were assigned to each fake galaxy s predicted set of colors , based on the galaxy s @xmath4 magnitude and colors , and the previously determined functions of @xmath238 , @xmath239 , and @xmath240 ( see 2.4 ) . the sample of fake galaxies was then `` observed '' a large number ( 1000 ) of times . each time , the colors of the fake galaxies were perturbed in a manner consistent with their photometric errors ( a process similar to the one used to construct the confidence intervals for the 74 real observed galaxies ) . the result of observing the fake galaxies numerous times with measurement errors was a large distribution of best - fit @xmath154 and @xmath170 values ( the number of fake galaxies ( 74 ) multiplied by the number of trials ( 1000 ) ) . the marginal distributions of @xmath154 and @xmath170 for this large simulated distribution matched the observed marginal distributions of @xmath154 and @xmath170 in figure 10a which verifies that our estimate of the intrinsic marginal distribution was reasonable . we then randomly selected a large number of groups of 74 best - fit @xmath196 pairs from the sample of 74,000 `` observed '' galaxies , and computed the correlation coefficient between @xmath154 and @xmath170 for each randomly selected sample . in 1000 random samples of 74 fake best - fit @xmath196 values , _ no _ sample had a correlation coefficient as strong as the one seen in the real sample . it appears that the age / extinction correlation for nirc lbgs has less than a @xmath241 chance of being the result of correlated measurement errors alone . _ a related question is whether the detection limits and selection criteria for the sample could have created an apparent dearth of young , unreddened galaxies , and old , dusty galaxies , which would mimic a real correlation between @xmath154 and @xmath170 . for the redshift range of the sample , the bc96 constant star - formation models with @xmath242 and @xmath243 have @xmath244 within the range for lbgs ( @xmath245 ) , but also very blue @xmath5 colors ( @xmath246 ) . a detection limit in @xmath2 translates into a limit in @xmath5 as a function of @xmath4 . however , since there are @xmath5 limits rather than detections for only @xmath247 out of @xmath248 galaxies with redshifts in the nirc sample , we conclude that our @xmath2 detection limit has not prevented us from detecting a significant population of young , unreddened galaxies . at any redshift in the nirc sample , old and reddened galaxies , with @xmath249 and @xmath250 , have predicted @xmath251 , which means that all such galaxies with @xmath252 , should be detected in @xmath56 . at the median redshift of the nirc sample , the lbg selection limit of @xmath253 implies that we should be able to detect old galaxies with @xmath254 , which is more than one standard deviation to the red of the mean of the sample . at increasing redshift , the upper limit on @xmath154 decreases until at the highest redshift in the nirc sample , @xmath255 , it is only possible to select a 1 gyr model with @xmath256 as a lbg , based on @xmath27 . such a redshift dependent effect discriminates against old and very dusty objects at the high redshift end of the lbg redshift selection function . to test the significance of such an effect , the sample of 74 galaxies was divided into low and high redshift subsamples , using the median redshift as a discriminator . all of the galaxies in the low redshift sample have @xmath257 , at which redshift a @xmath258 stellar population could be detected with @xmath259 . despite the fact that @xmath154 is barely restricted for the low redshift sample , a strong correlation between @xmath154 and @xmath170 is recovered from this sample , as well as from the high redshift and total samples . we conclude that the redshift - dependent bias against detecting old and dusty galaxies can not account for the strong correlation found between @xmath154 and @xmath170 in the nirc sample either . the inferred strong correlation between star - formation age and extinction does depend on the details of the attenuation law applied to correct observed galaxy colors and magnitudes to their intrinsic , unobscured values . at low redshift , relatively `` gray '' starburst galaxy attenuation relations such as that of calzetti ( 1997 ) provide a much better predictor of bolometric luminosity from far - uv observations than the reddening curves derived from observations of single stars , such as the smc , lmc , and galactic reddening curves ( meurer _ et al . _ there are very few observations to constrain the nature of dust extinction in galaxies at high redshift ; however , the limited information which exists favors the starburst attenuation relation over a much steeper curve such as the smc law ( meurer _ et al . _ 1999 , nevertheless , if the smc law is used with the bc96 constant star - formation models , instead of the calzetti relation , the best - fit extinction values at 1600 @xmath260 are systematically smaller , by a median factor of 2.7 , and the @xmath170 values are systematically larger . the ratio between smc and calzetti @xmath170 values increases as a function of increasing extinction , but the median ratio is 3.7 . qualitatively , the larger smc @xmath170 values result from interpreting a larger fraction of the @xmath5 color as due to the balmer break ( aging stellar population ) , and a smaller fraction as due to the effects of dust extinction . larger when using the smc rather than calzetti law . ] the result of using the smc law is that the correlation between inferred extinction and inferred age is very much diminished . the formed stellar mass , @xmath174 , for the constant star - formation models is simply the product of the best - fit @xmath170 and @xmath175 . figure 13 shows the relationship between @xmath174 and observed and intrinsic ( extinction - corrected ) luminosity values . when the sample is restricted to galaxies with best - fit @xmath261 ( the median for nirc lbgs ) , there are significant correlations between both uv and optical extinction - corrected luminosities and @xmath174 . however , when the whole nirc lbg sample is considered , the extinction - corrected uv and optical luminosities are only weakly correlated with @xmath174 , indicating that , for the wide range of extinction and age parameters found in the nirc lbg sample , even rest - frame optical luminosities ( the longest accessible wavelengths for @xmath8 galaxies until the space infrared telescope facility ( sirtf ) flies ) often have more to do with current star - formation rates than with formed stellar masses . thus , as emphasized by papovich _ et al . _ ( 2001 ) , the estimates of the formed stellar mass associated with the most recent star - formation episode must rely on the imf - dependent population synthesis modeling until longer wavelength observations are possible . however , papovich _ et al . _ also explored the dependence of inferred @xmath174 on the modeling assumptions for a given observed galaxy sed , concluding that , while variation of the modeling parameters yields highly variable results for parameters such as age and extinction , the combination of best - fit pairs of these parameters generally results in much more tightly constrained @xmath174 estimates for the modeled star formation episode . while exploring a smaller volume of parameter space , we confirm this trend for the formed stellar mass values ( cf . 5.2.3 ) . our results should be compared directly with the solar metallicity , salpeter imf models of papovich _ et al . _ ( 2001 ) ( their figure 17 ) , where the characteristic formed stellar mass is @xmath262 a few @xmath263 m@xmath264 for objects with uv luminosities of @xmath265 in the far - uv luminosity function of steidel _ et al . _ ( 1999 ) . our nirc sample , which has a median luminosity of somewhat brighter than @xmath265 , has a median inferred stellar mass of @xmath266 m@xmath264 using the @xmath217 cosmology adopted by papovich _ et al . thus , in the small region of overlap , the ground - based sample yields results similar to the brightest hdf - n galaxies . in contrast to papovich _ _ , who found that the uv luminosity was well correlated with the inferred stellar mass , we find that the uv luminosity is uncorrelated with the inferred @xmath174 . this may reflect a real difference between the brighter and fainter samples . one possible explanation could be that we see a wider range of extinction among the nirc lbgs than is found at fainter uv luminosities ( e.g. , objects of a given uv luminosity can be either heavily extinguished very luminous objects , or modestly extinguished , much less luminous objects ) such that the relationship between uv luminosities uncorrected and corrected for extinction is much less tight ( see as2000 ) . because of the very small overlap in uv luminosity between the two samples , the differences are at present statistically insignificant . there are 9 objects from the nirc sample which also have nebular line width measurements from near - ir spectroscopy ( pettini _ et al . _ 2001 we use the nebular line widths and the measured near - ir half - light radii to compute the dynamical mass enclosed within the half - light radius . the mass inferred from the nebular line widths is probably not indicative of the total mass in the dark - matter halo containing the lbg , but is likely to represent a lower limit ( see pettini _ et al . _ 2001 ) . similarly , since the inferred stellar mass estimates are only sensitive to the most recent episode of star - formation , they should also represent lower limits on the total stellar mass present . however , the near - infrared spectroscopic and photometric measurements generally span the same physical region , which contains not only stellar mass , but also gas and dark matter . the spectroscopic line widths should be sensitive to all this matter , while the photometry only probes the luminous stellar matter . therefore , the mass inferred from the nebular line widths should represent a rough upper limit to the @xmath174 values derived from the bc96 model fits . we would suspect a problem in the population synthesis modeling technique if the derived @xmath174 values were much larger and statistically discrepant with the inferred dynamical masses . as shown in figure 14 , for the majority of the galaxies , the @xmath174 value is consistent with the inferred dynamical mass ; in two cases where it is not , the best - fit stellar mass is significantly smaller than the inferred dynamical mass . we refrain from drawing conclusions from figure 14 , given the very large uncertainties in both mass estimates , but there is no evidence that the two mass scales are wildly inconsistent with one another . one of the more striking results from the modeling of lbg stellar populations is the strong correlation between extinction and age , such that the galaxies best fit by younger stellar populations are also best described by larger amounts of extinction and reddening . as discussed in 5.3.2 , we believe that this correlation is significant , despite the tendency for the individual ( @xmath234 ) covariance intervals to lie in a similar direction to the observed sample correlation , and despite a @xmath27 selection criterion which prevents a galaxy older than @xmath258 with significant amounts of dust extinction from being classified as a lbg at the highest redshifts in the sample . assuming that the correlation is real , there are implications for interpreting the range of lbg best - fit stellar populations and extinction parameters . it appears that the distributions of lbg star - formation ages , extinction , and inferred stellar masses fall along a continuum in the parameter space , the extremes of which might be viewed as separate `` populations '' . to illustrate this point , we isolate 16 galaxies with best - fit @xmath250 , and 16 galaxies with @xmath267 . in the discussion which follows , the @xmath268 gyr subsample is referred to as the `` old '' sample , and the @xmath269 subsample is labeled `` young . '' the `` young '' sample has @xmath270 , @xmath271 , and @xmath272 . in contrast , the `` old '' sample has @xmath273 , @xmath274 , and @xmath275 . the young sample is characterized by 2 magnitudes more dust extinction in the rest - frame uv than the old sample , star - formation rates at least an order of magnitude higher ( even when restricting the best - fit @xmath170 values to be above a reasonable minimum age ) , and formed stellar masses more than an order of magnitude smaller . to investigate further the distinctions between galaxies in the young , dusty subsample and those in the older , less reddened group , we drew from our database of lbg rest - frame far - uv spectra , collected by our group between october 1995 and november 1999 ( steidel _ et al . _ 1996a , b , 1998 , 1999 ) . by combining individual rest - frame uv spectra of galaxies in the `` young '' and `` old '' subsamples described above , we constructed `` young '' and `` old '' composite spectra following a method similar to that described in steidel , pettini , & adelberger ( 2001 ) . the young and old composite spectra are plotted in figure 15 , and figure 16 shows an expanded view of four specific regions of the spectra for more detailed comparison . there are several clear differences to note , the most dramatic of which is in the relative lyman @xmath141 line profiles . the old spectrum has strong lyman @xmath141 emission with rest - frame @xmath276 , whereas the young spectrum has a broad absorption trough plus much weaker emission which combine to give roughly @xmath277 . the radiative transfer of lyman @xmath141 photons in galaxies is a complex process which depends not only on the amount of dust present in the ism but also , and perhaps more importantly , on the geometry and kinematics of the neutral interstellar hydrogen gas ( charlot & fall , 1993 ; chen & neufeld 1994 ; kunth _ et al . _ 1999 ; tenorio - tagle _ et al . _ however , due to resonant scattering , lyman @xmath141 photons will on average traverse a much longer path before escaping the galaxy than nearby continuum photons , so that they are more prone to absorption by dust in the ism . therefore , the amount of dust present in the ism should have a non - negligible effect on the emergent lyman @xmath141 profile . the fact that the composite `` young '' spectrum exhibits a lyman @xmath141 profile with a combination of absorption and only weak emission , while the composite `` old '' spectrum has very strong lyman @xmath141 emission , offers independent support for a scenario in which the youngest galaxies have significantly dustier interstellar media than more mature lbgs . additionally , the young spectrum has much stronger lyman @xmath278 absorption , stronger low - ionization interstellar metal absorption lines of si ii @xmath279 , c ii @xmath280 , fe ii @xmath281 , and a stronger p - cygni c iv @xmath282 profile . the only interstellar metal absorption line which is stronger in the old spectrum is al ii @xmath283 . while the s / n ratios of the composite spectra represent large improvements over that of the individual spectra , the significance of the differences in line strengths between the old and young spectra is still difficult to quantify , given the noise in the composite spectra and the relatively small sample of objects whose spectra were combined . we therefore choose to make more qualitative observations . the low - ionization interstellar absorption lines are probably optically thick , and therefore on the flat part of the curve of growth , where equivalent width is determined mainly by velocity width , and not column density ( steidel _ et al . _ 1996a , gonzalez delgado _ et al . _ 1998 , heckman & leitherer 1997 ) . accordingly , the stronger interstellar absorption lines in the young spectrum may indicate interstellar medium velocity widths larger than those in the old spectrum . larger velocity widths are consistent with a scenario in which a higher supernova rate ( based on the higher @xmath175 ) deposits larger amounts of mechanical energy into the ism , accelerating the interstellar gas to higher velocities . a picture begins to emerge from the range of stellar populations observed in the nirc lbg sample . at one extreme of the continuum , there is a group of galaxies with best - fit ages younger than or equal to the dynamical timescale associated with the luminous portions of lbgs . based on nirc and hst - wfpc2 observations , the typical lbg angular half - light radius is @xmath284 ( giavalisco _ et al . _ 1996 ; pettini _ et al . _ this angular size translates into @xmath285 kpc at @xmath286 . furthermore , from near - ir spectroscopic measurements of a sample of 15 lbgs , the typical nebular line widths are @xmath287 km s@xmath10(pettini _ et al . _ 2001 ) . the associated dynamical timescale is @xmath288 . with typical values for @xmath289 and @xmath21 , this calculation yields @xmath290 myr . a starburst can not occur on timescales shorter than the dynamical timescale of the region experiencing star - formation , for reasons of causality . since a constant star - formation model with an age less than or equal to the dynamical timescale is indistinguishable from a bursting mode with a decay time , @xmath169 , limited to be greater than or equal to the dynamical timescale , the young , dusty galaxies in the nirc lbg sample might well be described as galaxies undergoing bursts of star - formation . whatever the cause of the burst of star - formation , we see intense star - formation and enhanced dust extinction in the youngest galaxies . at the other extreme are the galaxies whose best - fit constant star - formation @xmath170 values are close to a hubble time at @xmath0 , whose dust extinction is only moderate , and whose star - formation rates are more quiescent . the nirc lbg sample also contains galaxies with ages , dust - extinction , and star - formation rates which are intermediate between these two extremes . since the range of nirc lbg stellar populations is controlled by two strong correlations one between dust and age , and another between dust and intrinsic star - formation rate we use these correlations to construct a unified evolutionary model for lbgs at high redshift . according to this model , the `` young '' and `` old '' samples discussed in this section constitute different evolutionary stages of the same population , which are both selected as lbgs . the `` young '' galaxies represent objects in the first stage , where the star - formation rate is very active ( @xmath291 ) , large amounts of dust obscure the sites of star - formation , and a stellar mass on the order of @xmath292 is formed in a dynamical timescale . as time passes , dust is either destroyed or blown out , ( or both ) , resulting in lower extinction , and the star - formation rate decays on a @xmath293 myr timescale to reach the more quiescent rates seen in the `` old '' sample ( @xmath294 ) . the `` old '' sample then represents the most advanced stage of the evolution which we observe in @xmath0 lbgs , by which stage a stellar mass of @xmath295 has formed . as discussed earlier , for most `` old '' lbgs , it is not possible to distinguish from a strong balmer break whether constant star - formation has continued for at least 1 gyr , or whether exponentially declining star - formation has continued well past the age of the time constant , @xmath169 . if the relatively quiescent star - formation rates of the old galaxies are extrapolated back to the ages of the young / dusty sample assuming an exponentially declining star - formation rate with a time constant on the order of the dynamical timescale for lbgs , the star - formation rates obtained are on the order of the active rates in the young - dusty sample . such a simple evolutionary picture for lbgs is qualitatively consistent with the strong correlations we see in the nirc lbg stellar population parameters . this evolutionary picture predicts a relatively flat distribution of @xmath170 values , which is broadly consistent with the observed @xmath170 histogram ( figure 10a ) except for the smallest @xmath170 bin , which is `` over - sampled '' in the nirc lbg data set by about a factor of two ( 2.2 ) . while correcting the youngest bin by this amount makes the @xmath170 distribution flatter , the youngest bin still contains more galaxies than the other bins , implying that some fraction of the young , dusty , actively star - forming galaxies may fade to star - formation rates which are too low to be detected with the ( ground - based ) lbg criteria . there are several interesting issues which accompany the above scenario . first , there is the question of what process produces the initial , rapid star - formation phase . some obvious possibilities are major or minor merger events ( e.g. , somerville _ et al . _ 2001 ; mihos & hernquist 1996 ) , or even the initial collapse of the baryonic component of the galaxy . second , there is the question of how the large amount of dust in the significantly reddened `` young '' galaxies forms , given the relatively short timescales implied by the ages of these galaxies how much of the dust is formed in the current star - formation episode , and how much originated in previous generations of stars ? another important issue is the process by which dust is depleted as `` young '' lbgs evolve into `` old '' galaxies . there are a few proposed mechanisms for the removal of dust from galaxies . the first is radiation pressure , by which dust grains ( especially larger ones @xmath296 in size ) can be expelled by the general radiation field of a galaxy et al . _ ( 1998 ) have numerically modeled the removal of dust from a present - day @xmath16 spiral galaxy by radiation pressure , and have shown that up to @xmath177 of the dust formed in such a galaxy may be ejected from the disk . while the relevant physical conditions might not be identical in local spiral galaxies and the high redshift lbgs , radiation pressure provides one viable mechanism by which dust can be removed from galaxies . perhaps more relevant to the conditions in lbgs is the observational evidence for galactic - scale outflows , so - called `` superwinds , '' in actively star - forming local galaxies . these outflows are due to the combined mechanical energy input from supernovae and winds from massive stars , which sweeps up interstellar material and accelerates it to speeds of several hundred @xmath297 or greater ( heckman _ et al . _ 1990 , et al . _ 2000 there is strong kinematic evidence for such outflows in lbgs at high - redshift as well ( franx _ et al . _ 1997 ; steidel _ et al . _ 1998 ; pettini _ et al . _ 1998 , 2000 , 2001 ) . relative to the nebular emission lines representing the systemic velocity of the galaxy , low - ionization interstellar absorption lines in lbg rest - frame uv spectra are blue - shifted by @xmath298 , and lyman-@xmath141 emission is redshifted by @xmath299 ( pettini _ et al . _ the implied typical outflow velocities are @xmath300 km s@xmath10 , and in the best studied case ( pettini _ et al . _ 2000 ) the implied mass outflow rate equals or exceeds the current star - formation rate ( similar to what is seen in local starburst galaxies e.g. martin 1999 ) . recently , there has been direct observational evidence that dust , as well as gas , has been swept up by starburst superwinds and is contained in the material outflowing from local starburst galaxies ( heckman _ et al . _ 2000 ; alton , davies , & bianchi 1999 ) . et al . _ ( 2000 ) determine a typical outflowing dust mass of @xmath301 with a typical dust outflow rate of @xmath302 for the starbursts in their sample . while there are no direct observations of dust in lbg outflows , the presence of dust in the ism of lbgs is required by their broad - band colors . it seems very reasonable that the dust in the ism of lbgs becomes entrained in the outflows implied by the kinematics of uv interstellar and nebular spectral features . the process of dust being expelled in superwinds from starburst galaxies at high redshift is qualitatively consistent with the observed trend between extinction and star - formation age in lbgs . indeed , the superwind phenomenon has important implications not only for the enrichment of the intergalactic medium , but also for the evolution of dust opacity in galaxies . we have presented the results of a near - ir survey of optically selected @xmath0 lbgs . the survey includes 118 galaxies with @xmath2 measurements , 63 of which also have @xmath3 measurements . we combined the new near - ir data with previously obtained optical data to compute the distributions of optical - ir colors of lbgs . we have used the photometric data , spanning typical rest - frame wavelengths of 9005500 to obtain the following results : \1 . the rest - frame optical luminosity function of uv - selected lbgs greatly exceeds locally determined luminosity functions at the bright end . a schechter function fit formally yields a very steep ( @xmath303 ) faint - end slope , but part of this could be due to the fact that at our detection limit of @xmath99 ( @xmath304 at @xmath8 ) the data do not extend significantly beyond the knee in the luminosity function and the overall shape of the luminosity distribution is quite different from the local functions . independent of the fitted parameters , the @xmath0 optical luminosity function has a large excess of luminous galaxies relative to locally determined luminosity functions . down to our detection limit of @xmath305 ( or a luminosity equivalent to @xmath306 in the present - day @xmath7-band luminosity function ) , the rest - frame optical co - moving luminosity density of lbgs ( uncorrected for dust extinction ) is @xmath307 . this value is already within a factor of 2 of values determined by integrating local luminosity functions down to arbitrarily small luminosities . thus , the optical light from lbgs at @xmath8 likely exceeds that produced by all stars in the local universe . we have modeled the stellar populations and extinction of the lbgs with both spectroscopic redshifts and near - ir measurements ( 81 galaxies ) using relatively simple prescriptions . most of the galaxies have star - formation histories , star - formation rates , and star - formation ages that are only weakly constrained by the photometric observations without limiting parameter space using external constraints on the assumed extinction law and relatively simple forms for the star - formation history . we have chosen to report the results assuming the solar - metallicity bruzual and charlot ( 1996 ) models , a salpeter imf , the calzetti ( 1997 ) starburst attenuation relation , and a constant star - formation history . the systematic effects of changing the any of these assumptions have been summarized . when the interpretation of lbg stellar populations is restricted to bc96 constant star - formation models , we find a large spread of inferred ages , ranging from several myr to more than 1 gyr . the median best - fit constant star - formation age is @xmath209 . in model - independent terms , the distribution of best - fit ages indicates that a significant fraction of lbgs have been forming stars on timescales long enough that a detectable balmer break exists in their spectral energy distributions . although the specific age assigned to the balmer break depends on the star - formation history used to fit the colors , we note that very recent work indicates that winds from lbgs similar to those modeled above have shock - heated the intergalactic medium on physical scales of @xmath308 kpc ( adelberger _ et al . _ 2001 , in preparation such a large sphere of influence , with typical outflow velocities of a few hundred km s@xmath10 , suggests star - formation timescales of at least a few hundred million years for typical lbgs . this constitutes completely independent support for the timescales resulting from the stellar population modeling presented in this work . the bc96 constant star - formation distribution of best - fit @xmath154 ranges from @xmath309 with a median of 0.155 . this distribution depends less on star - formation history than the distribution of best - fit ages . the median implied extinction for this sample corresponds to a factor of @xmath310 attenuation in the rest - frame uv ( 1600 ) and a factor of @xmath1552 in the rest - frame optical . derived from the distributions of @xmath154 and @xmath170 , the best - fit star - formation rate , @xmath175 , ranges from @xmath311 , with a median of @xmath312 . the distribution of formed stellar masses , @xmath174 , ranges over more than 2 orders of magnitude with a median of @xmath313 . we note that the above parameter distributions and median values may not be completely general to lbgs at @xmath314 , since the way in which the nirc lbgs were selected tended to enhance the number of young , dusty , intensely star - forming , low @xmath174 objects , relative to the full lbg sample . regardless of the star - formation history used in this work to model the colors of the nirc lbgs , we find a strong correlation between dust extinction and intrinsic luminosity , such that dustier galaxies are intrinsically more luminous . this correlation has been reported by other authors ( as2000 ; meurer _ et al . _ 1999 ) , but here the extinction estimates are supported by observations over a wider wavelength baseline , spanning on average from @xmath315 in the rest - frame . this dependence of extinction on luminosity is consistent with a similar relation seen for star - forming galaxies in the local universe , and has the implication that , for the range of galaxies which we select as lbgs , those which are more heavily obscured by dust are as easy to detect in the rest - frame uv and optical as galaxies which are less dusty , because they are also intrinsically more luminous . \5 . for most galaxies , we find that the formed stellar mass derived from the integral of the best - fit star - formation rate over time depends relatively weakly on the star - formation history applied to fit the observed colors ( cf . et al . _ regardless of star - formation history , the inferred formed stellar mass is only weakly correlated with extinction corrected uv and optical luminosities for our total ground - based sample . this fact , together with the results on the rest - frame optical luminosity function of lbgs , shows that even at observed wavelengths of 2.15 @xmath122 m , the luminosities are poor proxies for stellar mass in general . therefore , inferences on stellar mass from optical / near - ir photometry are highly imf - dependent and will apply only to the stars that have formed in the last several hundred myr prior to observation . longer - wavelength observations with sirtf should improve the situation considerably . we find a strong correlation between inferred star - formation age and extinction in lbgs , such that younger galaxies are dustier than older galaxies , and have higher star - formation rates . establishing the significance of this correlation is difficult because of the covariance of @xmath154 and @xmath170 in our fits , but we have been unable to reproduce a similarly strong observed correlation from any intrinsically uncorrelated distributions of @xmath154 and @xmath170 . the correlation might be made to vanish with a suitable choice of star - formation histories or dust attenuation curve ( i.e. smc ) , but its existence appears secure for the most plausible choices of both . we also present evidence that this correlation is not an artifact of either the lbg @xmath27 selection criteria , or of the typical detection limit of @xmath99 for the nirc observations . composite rest - frame uv spectra of `` young '' and `` old '' subsamples of nirc lbgs exhibit differences , most dramatically in their respective lyman-@xmath141 profiles , offering independent evidence that the youngest galaxies have more dust extinction and/or more active interstellar media . we interpret the dust - age correlation in the context of a unified picture of lbgs in which the younger , dustier , more actively star - forming galaxies evolve into the older , less reddened , and more quiescent galaxies . the evolution of the dust extinction towards lower values is probably governed by the outflow of dust which is entrained in a `` superwind '' powered by the energy from supernovae explosions and the winds from massive stars . objects at various points along this evolutionary sequence are identified by the lbg optical selection criteria . the timescale for the evolution of the young - dusty galaxies into the older - less - reddened galaxies is on the order of @xmath316 , while the more quiescent phase can last a gyr or more . we would like to thank the staffs of the palomar , la palma , kitt peak , and keck observatories for their assistance in both optical and near - infrared observations . we would also like to thank matthew hunt , for assistance with the near - ir data reduction , and stephane charlot for helpful discussions and assistance in implementing the most up - to - date bruzual and charlot spectral synthesis codes . finally , we wish to extend special thanks to those of hawaiian ancestry on whose sacred mountain we are privileged to be guests . without their generous hospitality , most of the observations presented herein would not have been possible . ccs , kla , and aes have been supported by grants ast95 - 96229 and ast-0070773 from the u.s . national science foundation and by the david and lucile packard foundation . adelberger , k.l . 2001 , ph . d. thesis adelberger , k.l . , & steidel , c.c . 2000 , apj , 544 , 218 adelberger , k.l . , steidel , c.c . , giavalisco , m. , dickinson , m.e . , pettini , m. , & kellogg , m. 1998 , apj , 505 , 18 alton , p.b . , davies , j.i . , & bianchi , s. 1999 , a&a , 343 , 51 baugh , c. m. , cole , s. , frenk , c. s. , & lacey , c. g. 1998 , apj , 498 , 504 blanton , m. et al . 2001 , aj , accepted ( astro - ph/0012085 ) binney , j. & merrifield , m. 1998 , galactic astronomy ( princeton : princeton university press ) prevot - burnichon , m. l. 1985 , a&a , 149 , 330 bruzual , g. , & charlot , s. 1996 , bc96 manual ( private communication ) calzetti , d. 1997 , aj , 113 , 162 chapman , s. et al . 2000 , mnras , 319 , 318 charlot , s. , & fall , s.m . 1993 , apj , 415 , 580 chen , w.l . , & neufeld , d.a . 1994 , apj , 432 , 567 cole , s. et al . 2000 , mnras in press ( astro - ph/0012429 ) davies , j.i . , alton , p.b . , bianchi , s. , trewhella , m. 1998 , mnras , 300 , 106 dickinson , m.e . , 2000 , phil . a , 358 , 2001 folkes , s. et al . 1999 , mnras , 308 , 459 franx , m. , illingworth , g. d. , kelson , d. d. , van dokkum , p. g. , & tran , k .- v . 1997 , apj , 486 , l75 fukugita , m. , hogan , c.j . , peebles , p.j.e . 1998 , apj , 503 , 518 giavalisco , m. , & dickinson , m. e. 2001 , apj , 550 , 177 giavalisco , m. , steidel , c. c. , adelberger , k. l. , dickinson , m. e. , pettini , m. , & kellogg , m. 1998 , apj , 503 , 543 giavalisco , m. , steidel , c. c. , & macchetto 1996 , apj , 470 , 189 gonzalez delgado , r.m . , leitherer , c. , heckman , t.m . , lowenthal , j.d . , ferguson , h.c . , & robert , c. 1998 , apj , 495 , 698 gray , d. f. 1976 , the observation and analysis of stellar photospheres ( new york : wiley ) heckman , t.m . , lehnert , m.d . , strickland , d.k . , armus , l. 2000 , apjs , 129 , 493 heckman , t.m . , & leitherer , c. 1997 , aj , 114 , 69 heckman , t. m. , armus , l. , & miley , g. k. 1990 , apjs , 74 , 833 kunth , d. , mas - hesse , j. m. , terlevich , e. , terlevich , r. , lequeux , j. , & fall , s. m. 1998 , a&a , 334 , 11 leitherer , c. , leao , j.r.s . , heckman , t.m . , lennon , d.j . , pettini , m. , robert , c. 2001 , apj 550 , 724 leitherer , c. , et al . 1999 , apjs , 123 , 3 lejeune , t. , cuisinier , f. , buser , r. 1997 , a&as , 125 , 229 lejeune , t. , buser , r. , cuisinier , f. 1996 , gissel96 cd - rom lin , h. , kirshner , r.p . , schechtman , s.a . , landy , s.d . , oemler , a. , tucker , d.l . , schechter , p.l . 1996 , apj , 464 , 60 liu , m.c . , charlot , s. , & graham , j.r . 2000 , apj , 543 , 644 lowenthal , j. d. , koo , d. , guzman , r. , gallego , j. , phillips , a. , faber , s.m . , vogt , n.p . , illingworth , g. , & gronwall , c. 1997 , apj , 481 , 673 madau , p. ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , mnras , 283 , 1388 madau , p. 1995 , apj , 444 , 18 martin c. l. 1999 , apj , 513 , 156 matthews , k. & soifer , b.t . 1994 , in infrared astronomy with arrays : the next generation , ed . i. mclean ( dordrecht : klewer ) , 239 meurer , g.r . , & seibert 2001 , in starburst galaxies near and far eds . d. lutz & l. tacconi ( springer ) ( astro - ph/0101479 ) meurer , g.r . , heckman , t.m . , & calzetti , d. 1999 , apj , 521 , 64 mihos , j.c . & hernquist , l. 1996 , apj , 464 , 641 oke , j. b. et al . 1995 , pasp 107 , 3750 papovich , c. , dickinson , m.e . , & ferguson , h.c . 2001 , apj , accepted ( astro - ph/0105087 ) persson , e. , murphy , d. , krzeminski , w. , roth , m. , & rieke , m. 1998 , aj , 116 , 2475 pettini , m. , shapley , a.e . , steidel , c.c . , cuby , j - g . , dickinson , m.e . , moorwood , a. , adelberger , k.l . , giavalisco , m. 2001 , apj , 554 , 981 pettini , m. , steidel , c.c . , adelberger , k.l . , dickinson , m. , & giavalisco , m. 2000 , apj , 528 , 96 pettini , m. , kellogg , m. , steidel , c.c . , dickinson , m.e . , adelberger , k.l . , & giavalisco , m. 1998 , apj , 508 , 539 sawicki , m. , & yee , h. k. c. 1998 , aj , 115 , 1329 schechter , p. 1976 , apj , 203 , 297 somerville , r.s . , primack , j.r . , & faber , s.m . 2001 , mnras , 320 , 504 steidel , c.c . , adelberger , k.l . , & pettini , m. 2001 , apj , 546 , 665 steidel , c.c . , adelberger , k.l . , giavalisco , m. , dickinson , m.e . , & pettini , m. 1999 , apj , 519 , 1 steidel , c.c . , adelberger , k. l. , dickinson , m.e . , giavalisco , m. , pettini , m. & kellogg , m. 1998 , apj , 492 , 428 steidel , c.c . , giavalisco , m. , pettini , m. , dickinson , m.e . , & adelberger , k.l . 1996a , apjl , 462 , 17 steidel , c.c . , giavalisco , dickinson , m.e . , & adelberger , k.l . 1996b , aj , 112 , 352 steidel , c.c . , pettini , m. , & hamilton , d. 1995 , aj , 110 , 2519 steidel , c.c . , & hamilton , d. 1993 , aj , 105 , 2017 steidel , c.c . , & hamilton , d. 1992 , aj , 104 , 941 tenorio - tagle , g. , silich , s. a. , kunth , d. , terlevich , e. , & terlevich , r. 1999 , mnras , 309 , 332 trentham , n. , kormendy , j. , & sanders , d. b. 1999 , aj , 117 , 2152 valdes , f. 1982 , focas user s manual ( noao , tucson ) williams , r. e. , _ et al . _ 1996 , aj , 112 , 1335 lcccccc cdfa & 00 53 23.7 + 12 34 00 & 78.42 & 1.72 & 7 & 1.18 & 5 q0201 & 02 03 51.6 + 11 34 09 & 75.77 & 1.35 & 3 & 1.21 & 3 q0256 & 02 59 08.6 + 00 11 41 & 72.28 & 1.17 & 6 & & b2 0902 + 34 & 09 05 30.2 + 34 07 55 & 41.93 & 2.63 & 10 & 1.77 & 6 hdf & 12 36 52.3 + 62 12 59 & 75.32 & 2.96 & 14 & 2.29 & 10 westphal & 14 17 14.5 + 52 24 36 & 226.93 & 9.03 & 30 & 4.12 & 15 3c324 & 15 49 49.6 + 21 29 07 & 44.29 & 1.85 & 6 & 0.56 & 2 ssa22a & 22 17 34.2 + 00 15 01 & 77.75 & 5.75 & 22 & 1.73 & 6 ssa22b & 22 17 34.2 + 00 06 18 & 77.60 & 0.59 & 2 & & dsf2237a & 22 40 08.5 + 11 52 34 & 83.38 & 0.64 & 1 & 0.61 & 1 dsf2237b & 22 39 34.3 + 11 51 44 & 81.62 & 2.43 & 17 & 1.71 & 15 total & & 935.29 & 30.12 & 118 & 15.18 & 63 cdfa - c8 & c8,md10 & 2400 & cdfa - c1 & c1,md2 & 3780 & 1620cdfa - c22 & c22,m19,c19 & 3780 & 4620q0201-c6 & c6,mmd21 & 3480 & 3780q0201-b13 & b13 & 2700 & 2400q0256-m13 & m13,c15,md22 & 5220 & q0256-m17 & m17,md32,md34 & 4260 & b20902-c6 & c6,m11,md31 & 3780 & 3240b20902-md21 & md21 & 3240 & 3240b20902-d11 & d11,d12,md32 & 1080 & 1620b20902-md11 & md11,md14,md18 & 2160 & hdf - mm23 & mm23,oc37,omd49 & 6000 & 4680hdf - dd15 & dd15,mm28 & 3780 & 1620hdf - cc24 & cc24,oc38,om5,mm25 & 2160 & hdf - dd3 & dd3,mm9,mm7 & 3120 & 3240hdf - mm18 & mm18,mm17 & 3240 & 3240westphal - cc70 & cc70,mm69,dd49 & 3240 & 3240westphal - dd8 & dd8,mmd14,mmd16,mm8 & 3240 & 3240westphal - c11 & c11 & 3240 & westphal - cc13 & cc13 & 2700 & westphal - cc32 & cc32,mmd53,mmd54 & 3240 & westphal - cc79 & cc79,mmd115 & 3000 & westphal - mmd17 & mmd17,mmd20,mm13 & 6420 & 2520westphal - cc63 & cc63,mmd91 & 3240 & 3180westphal - mmd11 & mmd11 & 1080 & 1620westphal - cc1 & cc1 & 3240 & westphal - cc43 & cc43,cc45,dd29 & 3180 & 3240westphal - mmd109 & mmd109,mmd112 & 1620 & westphal - mmd113 & mmd113,mmd11 & 1080 & westphal - mmd23 & mmd23,mm11 & 1620 & 3c324-c1 & c1,c2,md5 & 4200 & 3c324-d7 & d7,c12 & 1080 & 13803c324-c3 & c3 & 1620 & ssa22a - d14 & d14,md42,md40,m31 & 1860 & ssa22a - d3 & d3,md3 & 7800 & 2700 ssa22a - md46 & md46,md50 & 3360 & ssa22a - aug96d1 & aug96d1,m8 & 3780 & ssa22a - c16 & c16,m13,m11 & 3240 & ssa22a - c36 & c36 & 3240 & 3240ssa22a - c6 & c6,m4,md4 & 6960 & 4860ssa22a - blob1 & c11,c15,m10 & 2160 & ssa22a - blob2 & m14,md14 & 2160 & ssa22b - oct96d8 & oct96d8,md11 & 2880 & dsf2237a - c2 & c2 & 1860 & 2700dsf2237b - m20 & m20,c26,m19,md56,md57,m17,m18 & 7380 & 7920dsf2237b - md2 & md2,md10 & 3240 & dsf2237b - d28 & d28,md81,md80,c43 & 3240 & 3240dsf2237b - d3 & d3,d4,md9,c1 & 6480 & 4860 lccccccc cdfa - c22 & 00 53 09.42 & + 12 36 00.5 & 23.97 & 0.73 @xmath317 0.08 & 0.90 @xmath317 0.50 & 3.18 @xmath317 0.23 & 3.046 cdfa - m19 & 00 53 11.04 & + 12 36 12.3 & 25.29 & 0.84 @xmath317 0.15 & 0.93 @xmath317 0.75 & @xmath318 2.79 & cdfa - c19 & 00 53 11.27 & + 12 35 39.9 & 24.63 & 0.81 @xmath317 0.13 & 1.96 @xmath317 0.30 & 3.79 @xmath317 0.24 & 2.667 cdfa - c8 & 00 53 32.85 & + 12 32 11.4 & 23.72 & 0.96 @xmath317 0.08 & & 2.16 @xmath317 0.40 & 3.071 cdfa - md10 & 00 53 32.22 & + 12 31 56.3 & 25.42 & 0.48 @xmath317 0.16 & & 2.10 @xmath317 0.60 & cdfa - c1 & 00 53 34.74 & + 12 30 30.6 & 23.53 & 0.69 @xmath317 0.06 & 1.29 @xmath317 0.18 & 2.69 @xmath317 0.21 & 3.110 cdfa - md2 & 00 53 36.31 & + 12 30 31.2 & 23.88 & 0.75 @xmath317 0.06 & 1.52 @xmath317 0.23 & 2.96 @xmath317 0.24 & 2.871 q0201-c6 & 02 03 41.81 & + 11 34 41.5 & 23.92 & 0.62 @xmath317 0.07 & 0.40 @xmath317 0.55 & 2.39 @xmath317 0.42 & 3.052 q0201-mmd21 & 02 03 41.66 & + 11 34 44.8 & 24.67 & 0.51 @xmath317 0.12 & 1.46 @xmath317 0.47 & 2.72 @xmath317 0.28 & q0201-b13 & 02 03 49.23 & + 11 36 10.8 & 23.34 & 0.03 @xmath317 0.08 & 1.72 @xmath317 0.13 & 2.81 @xmath317 0.15 & 2.167 q0256-c15 & 02 59 00.02 & + 00 11 38.6 & 24.23 & 0.69 @xmath317 0.09 & & 2.46 @xmath317 0.46 & 3.385 q0256-m13 & 02 58 58.83 & + 00 11 25.4 & 24.48 & 0.96 @xmath317 0.12 & & 2.89 @xmath317 0.38 & 3.227 q0256-m17 & 02 59 19.86 & + 00 12 32.4 & 24.04 & 1.10 @xmath317 0.11 & & @xmath318 1.54 & q0256-md22 & 02 59 00.79 & + 00 11 40.3 & 24.54 & 0.74 @xmath317 0.11 & & 2.30 @xmath317 0.51 & q0256-md32 & 02 59 18.11 & + 00 12 41.5 & 23.61 & 0.89 @xmath317 0.07 & & 2.61 @xmath317 0.25 & q0256-md34 & 02 59 20.21 & + 00 13 02.9 & 24.02 & 0.72 @xmath317 0.08 & & 3.88 @xmath317 0.14 & b20902-d11 & 09 05 23.01 & + 34 09 40.1 & 22.97 & 0.29 @xmath317 0.06 & 0.89 @xmath317 0.27 & 2.97 @xmath317 0.13 & 2.837 b20902-d12 & 09 05 23.45 & + 34 09 45.0 & 25.46 & -0.12 @xmath317 0.10 & @xmath318 -0.23 & @xmath318 2.96 & b20902-md32 & 09 05 26.24 & + 34 09 28.8 & 24.03 & 0.73 @xmath317 0.10 & 1.25 @xmath317 0.37 & 2.52 @xmath317 0.41 & 2.860 b20902-c6 & 09 05 20.58 & + 34 09 07.7 & 24.13 & 0.45 @xmath317 0.08 & 0.73 @xmath317 0.54 & 2.39 @xmath317 0.45 & 3.098 b20902-m11 & 09 05 19.58 & + 34 09 04.0 & 24.19 & 1.18 @xmath317 0.16 & 1.36 @xmath317 0.43 & 2.82 @xmath317 0.32 & 3.300 b20902-md31 & 09 05 20.27 & + 34 09 29.6 & 24.74 & 0.55 @xmath317 0.12 & & 3.77 @xmath317 0.24 & b20902-md21 & 09 05 20.08 & + 34 07 19.7 & 24.18 & 1.06 @xmath317 0.16 & 1.03 @xmath317 0.49 & 2.51 @xmath317 0.32 & 3.017 b20902-md11 & 09 05 35.20 & + 34 06 25.2 & 24.27 & 1.04 @xmath317 0.16 & & 2.22 @xmath317 0.48 & 3.392 b20902-md14 & 09 05 36.58 & + 34 06 37.0 & 24.68 & 0.97 @xmath317 0.14 & & 3.44 @xmath317 0.28 & b20902-md18 & 09 05 35.23 & + 34 06 53.2 & 24.51 & 0.41 @xmath317 0.12 & & @xmath318 2.01 & 2.869 hdf - cc24 & 12 36 51.81 & + 62 15 16.3 & 24.33 & 0.76 @xmath317 0.12 & & 2.72 @xmath317 0.40 & 3.333 hdf - dd3 & 12 36 47.60 & + 62 10 54.1 & 24.25 & 0.58 @xmath317 0.09 & 1.45 @xmath317 0.30 & 2.82 @xmath317 0.32 & 2.942 hdf - dd15 & 12 36 48.97 & + 62 15 43.4 & 23.61 & 0.62 @xmath317 0.07 & 0.36 @xmath317 0.49 & 2.39 @xmath317 0.37 & 3.135 hdf - mm7 & 12 36 47.75 & + 62 10 32.9 & 24.79 & 0.67 @xmath317 0.17 & 1.02 @xmath317 0.68 & @xmath318 2.29 & 2.985 hdf - mm9 & 12 36 51.45 & + 62 10 42.7 & 24.73 & 0.78 @xmath317 0.16 & & 4.04 @xmath317 0.19 & 2.972 hdf - mm17 & 12 36 47.67 & + 62 12 56.8 & 24.46 & 1.00 @xmath317 0.15 & 1.52 @xmath317 0.45 & 3.54 @xmath317 0.25 & 2.931 hdf - mm18 & 12 36 44.01 & + 62 13 11.9 & 24.10 & 1.00 @xmath317 0.11 & 1.63 @xmath317 0.21 & 3.46 @xmath317 0.16 & 2.929 hdf - mm23 & 12 37 02.64 & + 62 14 27.0 & 24.61 & 1.09 @xmath317 0.16 & 1.58 @xmath317 0.47 & 3.65 @xmath317 0.24 & 3.214 hdf - mm25 & 12 36 50.72 & + 62 14 45.6 & 24.82 & 0.74 @xmath317 0.17 & & 2.45 @xmath317 0.57 & 3.105 hdf - mm28 & 12 36 46.64 & + 62 15 18.1 & 25.04 & 0.70 @xmath317 0.18 & 1.01 @xmath317 0.70 & 2.89 @xmath317 0.53 & 3.371 hdf - oc37 & 12 37 03.18 & + 62 14 52.1 & 25.25 & 0.40 @xmath317 0.16 & 0.99 @xmath317 0.75 & 2.14 @xmath317 0.40 & 2.925 hdf - oc38 & 12 36 48.81 & + 62 15 03.6 & 24.97 & 0.67 @xmath317 0.18 & -0.81 @xmath317 0.61 & 3.47 @xmath317 0.32 & 3.114 hdf - omd49 & 12 37 04.25 & + 62 14 47.2 & 24.78 & -0.09 @xmath317 0.11 & 2.20 @xmath317 0.24 & 4.21 @xmath317 0.19 & 2.212 hdf - om5 & 12 36 50.46 & + 62 14 45.5 & 24.98 & 0.93 @xmath317 0.16 & & 2.93 @xmath317 0.53 & 3.097 westphal - cc79 & 14 17 13.70 & + 52 36 16.5 & 24.28 & 0.87 @xmath317 0.14 & & 3.17 @xmath317 0.40 & 3.061 westphal - mmd115 & 14 17 15.48 & + 52 36 12.6 & 23.97 & 0.65 @xmath317 0.11 & & 1.62 @xmath317 0.41 & 3.203 westphal - mmd109 & 14 17 27.43 & + 52 35 49.0 & 23.94 & 0.84 @xmath317 0.11 & & 3.05 @xmath317 0.20 & 2.715 westphal - mmd112 & 14 17 26.57 & + 52 35 59.2 & 25.34 & 0.63 @xmath317 0.20 & & @xmath318 2.84 & westphal - mmd113 & 14 18 24.85 & + 52 36 10.0 & 23.29 & 0.45 @xmath317 0.08 & & 2.71 @xmath317 0.24 & 2.730 westphal - mmd111 & 14 18 23.71 & + 52 36 09.2 & 25.03 & 0.08 @xmath317 0.15 & & 3.22 @xmath317 0.49 & westphal - cc70 & 14 17 31.43 & + 52 34 25.6 & 23.75 & 0.55 @xmath317 0.08 & 1.28 @xmath317 0.24 & 2.41 @xmath317 0.31 & 2.992 westphal - mm69 & 14 17 30.62 & + 52 34 17.2 & 25.24 & 0.83 @xmath317 0.21 & 1.49 @xmath317 0.57 & 3.24 @xmath317 0.42 & westphal - dd49 & 14 17 29.23 & + 52 34 31.5 & 23.50 & 0.17 @xmath317 0.06 & 1.33 @xmath317 0.18 & 3.35 @xmath317 0.13 & 2.806 westphal - cc63 & 14 18 23.10 & + 52 32 46.9 & 23.50 & 1.14 @xmath317 0.12 & 1.83 @xmath317 0.14 & 2.60 @xmath317 0.21 & 3.133 westphal - mmd91 & 14 18 22.93 & + 52 32 35.9 & 23.84 & 0.49 @xmath317 0.09 & 1.11 @xmath317 0.35 & 2.73 @xmath317 0.32 & 2.738 westphal - cc43 & 14 17 25.47 & + 52 29 37.9 & 23.87 & 1.04 @xmath317 0.15 & 1.24 @xmath317 0.33 & 3.30 @xmath317 0.16 & 3.081 westphal - cc45 & 14 17 27.68 & + 52 29 50.4 & 24.83 & 0.67 @xmath317 0.18 & 1.59 @xmath317 0.53 & 3.03 @xmath317 0.41 & 2.758 westphal - dd29 & 14 17 25.92 & + 52 29 32.1 & 24.82 & 0.32 @xmath317 0.13 & 1.60 @xmath317 0.42 & 3.86 @xmath317 0.24 & 3.240 westphal - cc32 & 14 18 14.45 & + 52 28 04.7 & 24.17 & 0.55 @xmath317 0.11 & & 2.40 @xmath317 0.46 & 3.197 westphal - mmd53 & 14 18 12.63 & + 52 28 06.6 & 25.22 & 0.36 @xmath317 0.16 & & 3.98 @xmath317 0.25 & westphal - mmd54 & 14 18 14.46 & + 52 28 16.0 & 24.64 & 0.45 @xmath317 0.13 & & 2.98 @xmath317 0.45 & 3.017 westphal - mmd23 & 14 17 21.95 & + 52 23 39.1 & 24.22 & 0.82 @xmath317 0.14 & & 2.81 @xmath317 0.32 & 2.857 westphal - mm11 & 14 17 22.02 & + 52 23 27.0 & 25.24 & 0.61 @xmath317 0.20 & & 3.59 @xmath317 0.32 & westphal - cc13 & 14 18 02.47 & + 52 24 36.5 & 23.64 & 1.06 @xmath317 0.14 & & 2.42 @xmath317 0.37 & 3.396 westphal - mmd17 & 14 18 09.56 & + 52 23 32.1 & 24.56 & 0.91 @xmath317 0.17 & 2.04 @xmath317 0.24 & 2.49 @xmath317 0.55 & 2.869 westphal - mmd20 & 14 18 09.63 & + 52 23 37.2 & 24.59 & 0.63 @xmath317 0.16 & 1.87 @xmath317 0.30 & 2.15 @xmath317 0.50 & 2.799 westphal - mm13 & 14 18 12.15 & + 52 23 30.4 & 25.40 & 0.69 @xmath317 0.21 & & 3.96 @xmath317 0.27 & 2.856 westphal - west3-c11 & 14 18 17.63 & + 52 23 46.2 & 24.25 & 1.22 @xmath317 0.16 & & 2.69 @xmath317 0.40 & 3.137 westphal - dd8 & 14 18 25.44 & + 52 23 22.4 & 24.47 & 0.68 @xmath317 0.14 & 1.09 @xmath317 0.56 & 2.77 @xmath317 0.46 & 2.841 westphal - mmd14 & 14 18 24.72 & + 52 23 19.8 & 24.92 & 0.82 @xmath317 0.20 & 1.62 @xmath317 0.42 & 3.08 @xmath317 0.41 & westphal - mmd16 & 14 18 23.61 & + 52 23 28.8 & 24.73 & 0.92 @xmath317 0.17 & 1.88 @xmath317 0.30 & 4.06 @xmath317 0.19 & westphal - mm8 & 14 18 23.92 & + 52 23 07.7 & 24.13 & 1.04 @xmath317 0.16 & 1.74 @xmath317 0.21 & 3.25 @xmath317 0.19 & 2.829 westphal - mmd11 & 14 18 09.73 & + 52 22 01.3 & 24.05 & 1.04 @xmath317 0.16 & 2.67 @xmath317 0.10 & 4.54 @xmath317 0.11 & 2.979 westphal - cc1 & 14 18 21.98 & + 52 21 22.0 & 23.83 & 1.02 @xmath317 0.15 & & 2.78 @xmath317 0.32 & 2.984 3c324-d7 & 15 49 52.96 & + 21 30 59.1 & 23.88 & 0.52 @xmath317 0.08 & 0.78 @xmath317 0.49 & 3.01 @xmath317 0.20 & 3c324-c12 & 15 49 53.95 & + 21 31 06.7 & 25.04 & 0.54 @xmath317 0.15 & 0.78 @xmath317 0.71 & 3.34 @xmath317 0.45 & 3c324-c3 & 15 49 47.10 & + 21 27 05.5 & 24.14 & 0.85 @xmath317 0.13 & & 2.57 @xmath317 0.41 & 3.283 3c324-c1 & 15 49 54.27 & + 21 26 33.2 & 24.33 & 0.62 @xmath317 0.14 & & 3.12 @xmath317 0.26 & 2.873 3c324-c2 & 15 49 53.98 & + 21 26 35.6 & 24.26 & 0.54 @xmath317 0.10 & & 3.19 @xmath317 0.26 & 2.880 3c324-md5 & 15 49 54.46 & + 21 26 34.6 & 25.48 & -0.08 @xmath317 0.24 & & 3.93 @xmath317 0.27 & ssa22a - md46 & 22 17 27.28 & + 00 18 09.9 & 23.30 & 0.42 @xmath317 0.09 & & 2.34 @xmath317 0.29 & 3.090 ssa22a - md50 & 22 17 26.83 & + 00 18 30.2 & 25.18 & 0.20 @xmath317 0.17 & & 3.18 @xmath317 0.47 & ssa22a - d14 & 22 17 35.29 & + 00 17 24.1 & 24.32 & 0.19 @xmath317 0.14 & & 2.07 @xmath317 0.46 & 3.018 ssa22a - md42 & 22 17 35.83 & + 00 17 19.8 & 25.33 & 0.06 @xmath317 0.17 & & 4.04 @xmath317 0.25 & ssa22a - md40 & 22 17 35.96 & + 00 17 08.3 & 24.89 & 0.70 @xmath317 0.19 & & 2.21 @xmath317 0.57 & 3.015 ssa22a - m31 & 22 17 36.87 & + 00 17 12.4 & 25.41 & 0.42 @xmath317 0.18 & & 3.22 @xmath317 0.44 & ssa22a - c36 & 22 17 46.10 & + 00 16 43.1 & 24.06 & 0.78 @xmath317 0.13 & 1.23 @xmath317 0.37 & 3.07 @xmath317 0.23 & 3.065 ssa22a - c16 & 22 17 31.96 & + 00 13 16.1 & 23.64 & 0.98 @xmath317 0.10 & & 2.86 @xmath317 0.20 & 3.061 ssa22a - m13 & 22 17 31.46 & + 00 12 55.2 & 25.46 & 0.74 @xmath317 0.24 & & @xmath318 2.96 & ssa22a - m11 & 22 17 31.77 & + 00 12 51.3 & 25.35 & 0.50 @xmath317 0.18 & & @xmath318 2.85 & ssa22a - m14 & 22 17 39.05 & + 00 13 30.1 & 25.47 & 0.75 @xmath317 0.24 & & 2.99 @xmath317 0.62 & 3.091 ssa22a - md14 & 22 17 37.93 & + 00 13 44.2 & 24.14 & 0.86 @xmath317 0.14 & & 1.96 @xmath317 0.43 & ssa22a - c11 & 22 17 25.68 & + 00 12 35.3 & 24.20 & 0.47 @xmath317 0.12 & & 2.82 @xmath317 0.32 & 3.108 ssa22a - c15 & 22 17 26.12 & + 00 12 55.3 & 25.19 & 0.55 @xmath317 0.19 & & 3.50 @xmath317 0.32 & 3.092 ssa22a - m10 & 22 17 26.79 & + 00 12 21.1 & 24.45 & 1.03 @xmath317 0.18 & & 2.32 @xmath317 0.51 & 3.098 ssa22a - aug96d1 & 22 17 24.00 & + 00 12 02.8 & 23.60 & 0.32 @xmath317 0.09 & & 2.79 @xmath317 0.12 & 2.202 ssa22a - m8 & 22 17 25.11 & + 00 11 56.8 & 24.72 & 0.89 @xmath317 0.19 & & 3.64 @xmath317 0.24 & ssa22a - d3 & 22 17 32.42 & + 00 11 33.0 & 23.37 & 0.97 @xmath317 0.10 & -0.60 @xmath317 0.40 & 2.24 @xmath317 0.32 & 3.086 ssa22a - md3 & 22 17 31.89 & + 00 11 38.3 & 24.60 & 0.15 @xmath317 0.13 & 1.40 @xmath317 0.47 & 2.17 @xmath317 0.50 & 2.483 ssa22a - c6 & 22 17 40.93 & + 00 11 26.0 & 23.44 & 0.79 @xmath317 0.09 & 0.36 @xmath317 0.44 & 2.01 @xmath317 0.37 & 3.099 ssa22a - m4 & 22 17 40.92 & + 00 11 27.9 & 24.83 & 0.76 @xmath317 0.19 & 0.93 @xmath317 0.68 & 2.68 @xmath317 0.56 & 3.091 ssa22a - md4 & 22 17 39.95 & + 00 11 39.6 & 24.25 & 0.24 @xmath317 0.12 & 1.03 @xmath317 0.49 & 2.92 @xmath317 0.32 & 2.611 ssa22b - md11 & 22 17 23.07 & + 00 03 42.2 & 25.23 & 0.37 @xmath317 0.15 & & 3.14 @xmath317 0.47 & ssa22b - oct96d8 & 22 17 23.52 & + 00 03 57.3 & 23.53 & 0.77 @xmath317 0.08 & & 2.87 @xmath317 0.18 & 3.323 dsf2237a - c2 & 22 40 08.30 & + 11 49 04.9 & 23.55 & 1.13 @xmath317 0.10 & 1.47 @xmath317 0.20 & 3.02 @xmath317 0.17 & 3.318 dsf2237b - d28 & 22 39 20.25 & + 11 55 11.4 & 24.46 & 0.32 @xmath317 0.08 & 1.34 @xmath317 0.49 & 2.42 @xmath317 0.52 & 2.934 dsf2237b - md81 & 22 39 21.72 & + 11 55 10.5 & 24.16 & 0.31 @xmath317 0.07 & 2.23 @xmath317 0.17 & 4.27 @xmath317 0.14 & 2.819 dsf2237b - md80 & 22 39 22.23 & + 11 55 08.3 & 25.37 & 0.94 @xmath317 0.20 & 1.16 @xmath317 0.77 & 3.08 @xmath317 0.37 & dsf2237b - c43 & 22 39 21.57 & + 11 54 44.9 & 24.35 & 0.70 @xmath317 0.11 & 1.73 @xmath317 0.34 & @xmath318 1.85 & 2.885 dsf2237b - m20 & 22 39 38.88 & + 11 52 22.1 & 24.20 & 1.12 @xmath317 0.13 & 1.33 @xmath317 0.43 & 2.96 @xmath317 0.20 & 3.156 dsf2237b - c26 & 22 39 38.02 & + 11 52 11.4 & 24.73 & 0.57 @xmath317 0.12 & 1.03 @xmath317 0.60 & 2.64 @xmath317 0.50 & 3.251 dsf2237b - m19 & 22 39 38.04 & + 11 52 20.6 & 24.80 & 0.96 @xmath317 0.18 & 1.47 @xmath317 0.53 & 2.79 @xmath317 0.56 & 3.259 dsf2237b - md56 & 22 39 39.00 & + 11 52 19.2 & 25.43 & 0.65 @xmath317 0.17 & 1.81 @xmath317 0.41 & 3.10 @xmath317 0.39 & dsf2237b - md57 & 22 39 39.92 & + 11 52 27.2 & 24.56 & 0.73 @xmath317 0.13 & 0.95 @xmath317 0.66 & 2.51 @xmath317 0.55 & 3.331 dsf2237b - m17 & 22 39 39.87 & + 11 52 06.5 & 24.67 & 1.03 @xmath317 0.18 & 1.06 @xmath317 0.60 & 2.53 @xmath317 0.55 & 3.020 dsf2237b - m18 & 22 39 40.16 & + 11 52 16.0 & 25.33 & 0.76 @xmath317 0.16 & 0.78 @xmath317 0.72 & @xmath318 2.83 & dsf2237b - md2 & 22 39 29.89 & + 11 47 12.5 & 24.17 & 0.90 @xmath317 0.12 & & 2.28 @xmath317 0.36 & 2.505 dsf2237b - md10 & 22 39 30.58 & + 11 47 34.3 & 25.04 & 1.01 @xmath317 0.19 & & 2.87 @xmath317 0.53 & 3.115 dsf2237b - d3 & 22 39 18.91 & + 11 47 40.4 & 24.19 & 0.51 @xmath317 0.09 & 1.07 @xmath317 0.49 & 2.02 @xmath317 0.37 & 2.931 dsf2237b - d4 & 22 39 18.75 & + 11 47 40.5 & 24.35 & 0.17 @xmath317 0.10 & 1.13 @xmath317 0.56 & 2.20 @xmath317 0.29 & 2.935 dsf2237b - md9 & 22 39 17.68 & + 11 47 39.2 & 24.97 & 0.45 @xmath317 0.07 & 0.31 @xmath317 0.56 & @xmath318 2.47 & dsf2237b - c1 & 22 39 16.90 & + 11 47 47.5 & 24.99 & 0.37 @xmath317 0.12 & 1.60 @xmath317 0.51 & 2.95 @xmath317 0.32 & 3.064 lcccc cdfa - c1 & 0.08 & 905 & 35 & 10.5 cdfa - c19 & 0.38 & 203 & 156 & 10.5 cdfa - c22 & 0.13 & 1278 & 36 & 10.7 cdfa - c8 & 0.26 & 10 & 275 & 9.4 cdfa - md2 & 0.20 & 321 & 65 & 10.3 q0201-c6 & 0.08 & 321 & 23 & 9.9 q0201-b13 & 0.04 & 2500 & 14 & 10.5 q0256-c15 & 0.02 & 905 & 12 & 10.0 q0256-m13 & 0.15 & 404 & 32 & 10.1 b20902-c6 & 0.05 & 509 & 15 & 9.9 b20902-d11 & 0.09 & 1278 & 53 & 10.8 b20902-m11 & 0.21 & 143 & 80 & 10.1 b20902-md11 & 0.17 & 30 & 69 & 9.3 b20902-md21 & 0.23 & 35 & 99 & 9.5 b20902-md32 & 0.24 & 35 & 110 & 9.6 hdf - cc24 & 0.08 & 806 & 19 & 10.2 hdf - dd15 & 0.10 & 227 & 42 & 10.0 hdf - dd3 & 0.12 & 719 & 23 & 10.2 hdf - mm17 & 0.30 & 321 & 104 & 10.5 hdf - mm18 & 0.32 & 203 & 179 & 10.5 hdf - mm23 & 0.18 & 2000 & 36 & 10.8 hdf - mm25 & 0.09 & 360 & 11 & 9.6 hdf - mm28 & 0.05 & 1800 & 8 & 10.1 hdf - mm9 & 0.24 & 2100 & 49 & 11.0 hdf - oc37 & 0.11 & 114 & 9 & 9.0 hdf - oc38 & 0.15 & 2000 & 18 & 10.5 hdf - om5 & 0.29 & 55 & 81 & 9.6 westphal - cc1 & 0.34 & 10 & 516 & 9.7 westphal - cc13 & 0.11 & 227 & 51 & 10.1 westphal - cc32 & 0.06 & 454 & 16 & 9.9 westphal - cc43 & 0.23 & 454 & 104 & 10.6 westphal - cc45 & 0.23 & 255 & 33 & 9.9 westphal - cc70 & 0.10 & 321 & 32 & 10.0 westphal - cc79 & 0.18 & 641 & 45 & 10.4 westphal - dd49 & 0.17 & 1139 & 68 & 10.9 westphal - cc63 & 0.36 & 10 & 942 & 10.0 westphal - dd8 & 0.32 & 15 & 175 & 9.4 westphal - mm8 & 0.37 & 40 & 324 & 10.1 westphal - mmd109 & 0.41 & 10 & 662 & 9.8 westphal - mmd113 & 0.23 & 81 & 156 & 10.1 westphal - mmd115 & 0.13 & 10 & 71 & 8.9 westphal - mmd23 & 0.32 & 20 & 219 & 9.6 westphal - mmd54 & 0.10 & 1278 & 14 & 10.3 westphal - mmd91 & 0.16 & 286 & 42 & 10.1 westphal - west3-c11 & 0.29 & 15 & 223 & 9.5 3c324-c1 & 0.12 & 1434 & 20 & 10.4 3c324-c2 & 0.12 & 1700 & 22 & 10.6 3c324-c3 & 0.07 & 641 & 19 & 10.1 ssa22a - aug96d1 & 0.17 & 321 & 37 & 10.1 ssa22a - c11 & 0.08 & 1139 & 18 & 10.3 ssa22a - c15 & 0.12 & 2000 & 11 & 10.3 ssa22a - c16 & 0.21 & 181 & 109 & 10.3 ssa22a - c36 & 0.17 & 571 & 48 & 10.4 ssa22a - c6 & 0.20 & 10 & 204 & 9.3 ssa22a - d14 & -0.01 & 571 & 6 & 9.6 ssa22a - m10 & 0.30 & 10 & 212 & 9.3 ssa22a - m14 & 0.15 & 571 & 12 & 9.8 ssa22a - m4 & 0.13 & 321 & 17 & 9.7 ssa22a - d3 & 0.29 & 10 & 540 & 9.7 ssa22a - md3 & 0.24 & 15 & 54 & 8.9 ssa22a - md4 & 0.14 & 641 & 21 & 10.1 ssa22a - md40 & 0.21 & 20 & 45 & 9.0 ssa22a - md46 & 0.01 & 806 & 22 & 10.2 ssa22b - oct96d8 & 0.06 & 1900 & 32 & 10.8 dsf2237a - c2 & 0.15 & 719 & 76 & 10.7 dsf2237b - c1 & 0.05 & 2100 & 6 & 10.1 dsf2237b - c26 & 0.05 & 1139 & 9 & 10.0 dsf2237b - d28 & 0.06 & 719 & 10 & 9.9 dsf2237b - d3 & 0.16 & 30 & 46 & 9.1 dsf2237b - d4 & -0.04 & 1278 & 5 & 9.8 dsf2237b - m19 & 0.20 & 161 & 40 & 9.8 dsf2237b - m20 & 0.21 & 227 & 69 & 10.2 dsf2237b - md10 & 0.29 & 35 & 88 & 9.4 dsf2237b - md2 & 0.39 & 10 & 377 & 9.6
we present the results of a near - infrared imaging survey of @xmath0 lyman break galaxies ( lbgs ) . the survey covers a total of 30 arcmin@xmath1 and includes 118 photometrically selected lbgs with @xmath2 band measurements , 63 of which also have @xmath3 band measurements , and 81 of which have spectroscopic redshifts . using the distribution of optical @xmath4 magnitudes from previous work and @xmath5 colors for this sub - sample , we compute the rest - frame optical luminosity function of lbgs . this luminosity function is described by an analytic schechter fit with a very steep faint end slope of @xmath6 , and it strikingly exceeds locally determined optical luminosity functions at brighter magnitudes , where it is fairly well constrained . the @xmath7-band luminosity density of only the observed bright end of the @xmath8 lbg luminosity function already approaches that of all stars in the local universe . for the 81 galaxies with measured redshifts , we investigate the range of lbg stellar populations implied by the photometry which generally spans the range 9005500 in the rest - frame . the parameters under consideration are the star - formation rate as a function of time , the time since the onset of star - formation , and the degree of reddening and extinction by dust . while there are only weak constraints on the parameters for most of the individual galaxies , there are strong trends in the sample as a whole . with a wider wavelength baseline than most previous studies at similar redshifts , we confirm the trend that intrinsically more luminous galaxies are dustier . we also find that there is a strong correlation between extinction and the age of the star - formation episode , in the sense that younger galaxies are dustier and have much higher star - formation rates . the strong correlation between extinction and age , which we show is unlikely to be an artifact of the modeling procedure , has important implications for an evolutionary sequence among lbgs . a unified scenario which accounts for the observed trends in bright lbgs is one in which a relatively short period of very rapid star - formation ( hundreds of m@xmath9 yr@xmath10 ) lasts for roughly 50100 myr , after which both the extinction and star - formation rate are considerably reduced and stars are formed at a more quiescent , but still rapid , rate for at least a few hundred myr . in our sample , a considerable fraction ( @xmath11% ) of the lbgs have best - fit star - formation ages @xmath12gyr , implied stellar masses of @xmath13 m@xmath9 , and are still forming stars at @xmath14 m@xmath9 yr@xmath10 .
the hot intracluster medium ( icm ) is the most massive baryonic component in galaxy clusters , comprising @xmath012% of the total cluster mass , or @xmath010@xmath910@xmath10 m@xmath2 in rich clusters . the bulk of the icm is very low density ( @xmath11 @xmath12 ) , and will require @xmath010 gyr to cool via thermal bremsstrahlung radiation . in the centers of clusters , however , this situation is reversed . the icm in the inner @xmath0100 kpc can reach high enough density that the cooling time is short relative to the age of the cluster ( @xmath13 gyr ) . this ought to result in a runaway cooling flow , fueling a massive , 1001000 m@xmath2 yr@xmath4 starburst in the central cluster galaxy ( for a review , see * ? ? ? however , despite the fact that roughly a third of all galaxy clusters have short central cooling times ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , such massive starbursts are extremely rare ( e.g. , * ? ? ? * ; * ? ? ? the vast majority of clusters which ought to host runaway cooling flows show only mild amounts of star formation @xcite , suggesting that much of the predicted cooling at high temperature is not , in fact , occurring . by comparing the cooling rate of intermediate - temperature ( @xmath010@xmath14 k ) gas ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) to the ongoing star formation rate in central cluster galaxies , one can quantify what fraction of the cooling icm is able to condense and form stars . these studies typically find that only @xmath01% of the predicted cooling flow is ultimately converted into stars . recently , @xcite showed that this inefficient cooling can be divided into two contributing parts : cooling from high ( @xmath010@xmath15 k ) to low ( @xmath010 k ) temperature at @xmath010% efficiency , and an additionally @xmath010% efficiency of converting cold gas into stars . the former term is generally referred to as the `` cooling flow problem '' , and suggests that some form of feedback is preventing gas from cooling out of the hot phase . the most popular solution to the cooling flow problem is that `` radio - mode '' feedback from the central supermassive blackhole is offsetting radiative losses in the icm ( see reviews by * ? ? ? * ; * ? ? ? radio jets from the active galactic nucleus ( agn ) in the central cluster galaxy can inflate bubbles in the dense icm , imparting mechanical energy to the hot gas ( e.g. , * ? ? ? * ; * ? ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the ubiquity of radio jets in so - called `` cool core clusters '' ( e.g. , * ? ? ? * ) suggests that the two are intimately linked , while the correlation of the feedback strength ( e.g. , radio power , mechanical energy in bubbles ) with the x - ray cooling luminosity provides evidence that this mode of feedback is , on average , sufficient to fully offset cooling ( e.g. , * ? ? ? * ; * ? ? ? further , recent studies of newly - discovered high-@xmath16 clusters in the south pole telescope 2500 deg@xmath17 survey @xcite suggest that this energy balance between icm cooling and agn feedback has been in place since @xmath18 @xcite . the study of outliers in a population can often provide new insights into the physical processes governing said population . amongst cool core clusters , there is perhaps no more extreme system than the phoenix cluster ( spt - clj2344 - 4243 ; * ? ? ? * ; * ? ? ? this massive ( @xmath19 m@xmath2 ) system is the most x - ray luminous cluster yet discovered ( l@xmath20 erg s@xmath4 ) , with a predicted cooling rate of @xmath02000 m@xmath2 yr@xmath4 . however , contrary to the norm , the central galaxy in the phoenix cluster harbors a massive starburst ( @xmath0800 m@xmath2 yr@xmath4 ; * ? ? ? * ) , a massive reservoir of molecular gas ( m@xmath21 m@xmath2 ; * ? ? ? * ) , and a dusty type-2 quasar ( qso ; * ? ? ? * ; * ? ? ? * ) . in @xcite , we argued that the only feasible way to bring such a vast supply of cold gas into the center of a rich cluster was via a runaway cooling flow . however , @xcite showed that radio - mode feedback is operating at a level that ought to offset radiative cooling in this system , although whether this energy has had time to couple to the icm is uncertain . the fact that the central agn is simultaneously providing strong radiative ( quasar - mode ) and mechanical ( radio - mode ) feedback suggests that it may be in the process of transitioning from a qso to a radio galaxy , and that the starburst is being fueled by gas that cooled before the first radio outburst . however , this interpretation hinges on a single , shallow x - ray observation with the _ chandra x - ray observatory _ ( 10 ks ) , in which both the x - ray nucleus and cavities are detected at low significance . in an effort to provide a more complete picture of this system , we have obtained deep far - uv spectroscopy and x - ray imaging spectroscopy using the _ hubble space telescope _ cosmic origins spectrograph ( hst - cos ) and the advanced ccd imaging spectrometer ( acis ) on the _ chandra x - ray observatory _ , respectively . these data provide a detailed picture of the young stellar populations , the intermediate - temperature gas ( ovi ; 10@xmath22k ) , and the hot intracluster medium . these new x - ray data represent a factor of @xmath610 increase in exposure time over previously - published observations . we have also obtained deep optical and radio data via megacam on the magellan clay telescope and the giant meterwave radio telescope ( gmrt ) , respectively . combined , these data will provide a much more detailed view of the complex interplay between cooling and feedback in this extreme system . we describe the reduction and analysis of these new data in 2 , along with supporting data at ir and optical wavelengths . in 3 we present new results derived from these data , focusing on the central agn , the starburst , and the cluster core . in 4 we discuss the implications of these results and present a coherent picture of the physical processes at work in this unique system . we conclude in 5 with a summary of this and previous work , with a look towards the future . throughout this work , we assume h@xmath23 km s@xmath4 mpc@xmath4 , @xmath24 , and @xmath25 . we assume @xmath26 for the phoenix cluster , which is based on optical spectroscopy of the member galaxies @xcite . below , we summarize the acquisition and reduction of new data used in this study , along with a brief summary of supporting data published in previous works . the phoenix cluster ( spt - clj2344 - 4243 ) was observed by _ chandra _ in 2011 ( obsid : 13401 ; pi : garmire ) for a total of 11.9ks as part of a large x - ray survey of spt - selected clusters ( pi : benson ) . to understand in detail this extreme system , we have obtained an additional 117.4ks ( obsids : 16135 , 16545 ; pis : mcdonald , garmire ) , resulting in a combined exposure time of 129.3ks ( see figure [ fig : chandra ] ) and a total of 88,042 counts in the central 1mpc and in the energy range 0.58.0 kev . all observations were performed with acis - i . the data from each observation were individually processed in the standard manner , including cleaning and filtering for background flares , using ciao 4.6 and caldb 4.6.3 . the x - ray background was modeled using three distinct components . first , the re - scaled acis blank - sky background observations provide an adequate representation of both the particle - induced and unresolved cosmic x - ray background . a second component , modeled as a soft ( 0.18kev apec ) excess , accounts for galactic interstellar medium emission @xcite . finally , a third component , modeled as a hard ( absorbed 40kev bremss ) excess , accounts for the fact that a larger - than - normal fraction of the cxb may be unresolved in shorter exposures . the latter two components were simultaneously fit to the on - source spectrum and an off - source spectrum extracted from a blank region of the target field , at a physical separation of @xmath63mpc from the cluster center . scaledwidth=47.0% ] in order to search for structure in the x - ray surface brightness , images were made in several bandpasses including a soft ( 0.52.0 kev ) and hard ( 4.08.0 kev ) band , which trace the hot icm and central agn , respectively . to search for faint structure in the soft image , we subtract a two - dimensional model comprised of three beta functions with a shared center , position angle , and ellipticity . the residual image after this model was subtracted was smoothed both adaptively , using csmooth , and with fixed - width gaussians . the former technique provides the highest - quality image over large scales , while the latter has a lower ( but non - zero ) likelihood of artificially creating features out of noise . thermodynamic profiles were measured by extracting on - source spectra in concentric annuli , where we strive for @xmath05000 x - ray counts per annulus . these spectra were empirically deprojected using dsdeproj @xcite . deprojected spectra were background subtracted using off - source regions ( the cluster only occupies one of the four acis - i chips ) and fit over the range 0.510.0kev using xspec @xcite . we model the x - ray spectrum with a combination of galactic absorption ( phabs ) and an optically - thin plasma ( mekal ) , freezing the absorbing column ( @xmath27 ) to the galactic value @xcite and the redshift to @xmath26 . in the inner two annuli , we add an additional absorbed powerlaw component to account for emission from the central x - ray - bright agn . the cosmic origins spectrograph ( cos ) is a medium - to - high resolution uv spectrograph on hst with a 2.5@xmath28 diameter aperture . the broad uv coverage and sensitivity of this instrument make it ideal for simultaneously detecting coronal line emission ( e.g. , ovi ) and the stellar continuum in young stellar populations . we designed our observational setup ( i d 13456 , pi : mcdonald ) to be able to detect ovi@xmath51032,1038 over the full extent of the central starburst . this required two offset pointings , as depicted in figure [ fig : cos_regions ] . following the cos instrument handbook , we model the throughput across the aperture , yielding a two - dimensional estimate of the throughput over the extent of the central cluster galaxy . our setup yields @xmath675% throughput over the full extent of the starburst , with @xmath650% throughput out to the edge of the central galaxy . figure [ fig : cos_regions ] shows our estimate of the two - dimensional throughput map compared to optical and near - uv images from hst wfc3-uvis and cos of the central cluster galaxy . we use the g160 m grating with a central wavelength of 1577 , providing coverage from 13861751 . this corresponds to rest - frame 8701100 at @xmath26 , which covers the ly@xmath29 , ovi@xmath51032,1038 , and heii@xmath301085 lines . each of the two pointings were observed for 8 orbits , totaling @xmath020ks on - source for each spectrum . this exposure time was chosen to allow simultaneous modeling of the ovi lines and the uv continuum , to sufficient depth to rule out a 1% efficient cooling flow in the absence of an ovi detection . the observed data were binned to @xmath00.8 / pix to improve the signal - to - noise . geocoronal emission and residual background was subtracted using a deep blank - sky spectrum with the same observational setup . the resulting spectra were modeled using the latest version of starburst99 ( v7.0.0 ; * ? ? ? these models have sufficient spectral resolution ( @xmath31 ) over 1000 @xmath32 1100 to constrain the age and metallicity of the starburst in the core of phoenix . we are restricted to the empirically - calibrated models in this wavelength range , which have metallicities of @xmath33 ( galactic ) and @xmath34 ( lmc / smc ) . we use the latest models from @xcite and @xcite , which include rotation of massive stars . the fits were performed by grid searching over a variety of populations ( continuous / instantaneous star formation , low / high metallicity ) , with the ages ranging from 050 myr in steps of 0.5 myr . for each grid position , we perform a least - squares fit of the model to the data , allowing the normalization , reddening , redshift , stellar dispersion , and emission line fluxes ( ly@xmath29 , ovi@xmath51032,1038 , and heii@xmath301085 ) to vary . the best - fitting model was chosen to minimize @xmath35 . the best - fitting models are shown in figures [ fig : fullspec ] and [ fig : sb99model ] , and will be discussed in 3.1 and 4.1 . the phoenix cluster ( spt - clj2344 - 4243 ) was observed on 30 august 2013 with the megacam instrument on the clay - magellan telescope at las campanas observatory , chile @xcite . exposures were taken in @xmath36 ( 4@xmath37200 sec ) , @xmath38 ( 4@xmath37600 sec + 2@xmath37120 sec ) , and @xmath39 ( 4@xmath37400 sec ) passbands in standard operating mode and cover the inner 12@xmath40 ( 5mpc ) of the cluster . for this work we consider only the inner @xmath00.5@xmath40 the remaining data will be published in an upcoming weak lensing study ( applegate et al . in prep ) . images were processed and stacked using the standard megacam reduction pipeline at the smithsonian astrophysical observatory ( sao ) telescope data center . color photometry was calibrated using stellar locus regression ( slr ; * ? ? ? for a more detailed description of the data reduction and calibration , the reader is directed to @xcite . giant metrewave radio telescope ( gmrt ) 610 mhz observations of spt - cl j23444243 were taken on june 14 and 15 , 2013 . the total on source time resulting from these two observing runs was about 10 hrs , with a usable bandwidth of 29 mhz . the data were reduced with the astronomical image processing system ( aips ) , parseltongue @xcite and obit @xcite , following the scheme detailed in @xcite . reduction steps include flagging of radio frequency interference ( rfi ) , bandpass and gain calibration . following that , several cycles of self - calibration were carried out to refine the calibration solutions . direction - dependent gain solutions were then obtained towards several bright sources within the field of view . the data was imaged by dividing the field of view into smaller facets @xcite . this corrects for the non - coplanar nature of the array and the direction - dependent calibration . for more details about the data reduction the reader is referred to @xcite and @xcite . we also include in this paper optical multi - band imaging from hst wfc3-uvis @xcite and optical imaging spectroscopy @xcite . for a detailed description of a specific dataset , we direct the reader to the aforementioned papers . relevant features of the various datasets are summarized below . optical imaging was acquired on the cluster core for @xcite in five wfc3-uvis bands : f225w , f336w , f475w , f625w , and f814w . while relatively shallow , these data provided the first resolved view of the starburst in the central cluster galaxy . additionally , large - area ground - based data at @xmath16-band was obtained with mosaic - ii on the blanco 4-m telescope ( for details , see * ? ? ? * ; * ? ? ? optical imaging spectroscopy of the central galaxy was obtained from gemini - s gmos in ifu mode , and presented in @xcite . these data span rest - frame 30006000 , and reveal complex emission - line nebulae in and around the central galaxy . the morphology of this gas is , for the most part , consistent with the near - uv morphology . to allow a direct comparison of hst - cos and gmos spectroscopy , we extract aperture spectra from the gmos ifu data , matching the cos throughput as a function of radius within the aperture . this allows us to simultaneously model the uv and optical continuum , as well as compare measured emission line fluxes in the uv and optical . below , we summarize the main results that emerge from the analysis of these deep uv , x - ray , optical , and radio data . we defer a discussion of these results in the greater context of the cluster s evolution to 4 . , ovi @xmath51032,1038 , [ sivi]@xmath301049 , and heii@xmath301085 ( blue ) and a blank - sky background component ( grey ) . the best - fitting stellar population model is described in the upper right of each panel . in general , these spectra are consistent with a highly - reddened , young ( @xmath415 myr ) population of stars.,scaledwidth=48.0% ] using aperture - matched spectra from both hst - cos and gemini - s gmos @xcite , combined with broadband photometry spanning the spectral gap at 12003500 @xcite , we can constrain the relative contribution of young and old stellar populations to the uv+optical continuum , along with the effects of reddening on the spectrum over nearly an order of magnitude in wavelength . in figure [ fig : fullspec ] , we show the full uv+optical spectrum for the combined area of the two hst - cos apertures ( figure [ fig : cos_regions ] ) . we overplot a series of best - fitting model spectra , comparing different reddening models @xcite and star formation histories . all models include a variable - mass old ( 6 gyr ) stellar population . we note that the extinction curve from @xcite was calibrated using observations of stars in our galaxy and the magellanic clouds , while @xcite was calibrated on nearby starburst and blue compact galaxies . we find , as did @xcite , that a relatively gray , `` feature - less '' extinction curve provides the best fit to the data , suggesting that the size distribution of the dust is skewed towards larger grains . the data also suggest a lack of the characteristic 2175 bump , which is the strongest absorption feature in the interstellar medium but tends to be missing in starburst galaxies @xcite . @xcite suggest that high levels of turbulence can both flatten the curvature of the extinction law and wipe out the 2175 absorption feature . this is consistent with @xcite , where we show that the ionized gas has significant velocity structure ( @xmath42 km s@xmath4 ) and is consistent with being ionized primarily by shocks . assuming the flatter extinction curve of @xcite , we find an overall good fit ( @xmath43 ) to the data spanning rest - frame 8006000 . the best - fitting model ( figure [ fig : fullspec ] ; purple curve ) is a 4.5myr - old starburst with a total zero - age mass of @xmath44 m@xmath2 . _ this corresponds to @xmath411% of the total stellar mass of the central galaxy . _ the fit quality is only slightly reduced ( @xmath45 ) if we assume a continuous star formation history , with the best - fit model representing a @xmath0317 m@xmath2 yr@xmath4 over the past @xmath4615 myr ( figure [ fig : fullspec ] ; orange curve ) . these two models disagree on the strength of the balmer jump at @xmath03650 ( see inset of figure [ fig : fullspec ] ) , with the data preferring the instantaneous star formation model . this provides marginal evidence that star formation has been quenched in the past @xmath05 myr . when the far - uv spectrum ( 1000 @xmath41 @xmath47 @xmath41 1100 ) is considered on its own , the data prefer a younger population ( @xmath012 myr ) , as we show in figure [ fig : sb99model ] . it is worth noting that the age is only weakly constrained in this region of the spectrum , due to the relatively short wavelength range . there is little difference in stellar populations and reddening between the northern and southern cos apertures . this may be due to the considerable overlap between the two regions ( see figure [ fig : cos_regions ] ) or may indicate a uniformity in the starburst . with only two options for metallicity when synthesizing high - resolution spectra at far - uv ( * ? ? ? * ; * ? ? ? * see also 2.2 ) , we can only weakly constrain the metallicity from this spectrum . we note that the solar metallicity spectrum provides a significantly worse fit than the spectrum for the @xmath48 population , with the former having significantly ( factor of @xmath05 ) stronger absorption lines which are not present in the observed spectrum . we stress that , despite the extremely high star formation rate inferred by the uv spectrum , the young population constitutes @xmath411% of the total stellar mass in the central galaxy . this is consistent with our picture of bcg formation , since this mode of star formation _ must _ be suppressed to prevent central cluster galaxies from growing too massive ( and luminous ) by @xmath49 . in summary , the combined uv - through - optical spectrum of the central galaxy in the phoenix cluster reveals a highly - reddened ( e(b - v ) @xmath0 0.5 ) , young stellar population , consistent with being recently quenched ( @xmath505 myr ) . [ cols="^ " , ] the emissivity curve of ovi@xmath51032,1038 peaks at @xmath010@xmath22k , making it an excellent probe of gas at intermediate temperatures between the `` hot '' ( @xmath610@xmath15k , probed by x - ray emission lines ) and `` warm '' ( @xmath010@xmath51k , probed by optical emission lines ) phases . emission from ovi has been detected in several nearby galaxy clusters @xcite , allowing an independent estimate of the icm cooling rate . here , we attempt a similar analysis to these earlier works on the central @xmath015kpc of the phoenix cluster ( see figure [ fig : cos_regions ] ) . in figure [ fig : sb99model ] we show the red - side uv spectra for our two hst - cos pointings . these spectra are modeled using the best - fitting young stellar population described in 2.2 and 3.1 , along with emission lines including ly@xmath29 , ovi@xmath51032,1038 , and heii@xmath301085 . while the model is , overall , a good match to the data , there are several features present in the data that are not reflected in the models . in the case of the emission at @xmath01050 this may be due to time - varying airglow emission . other features may be absent in the models due to the fact that they are empirically calibrated and lacking the full population of stars included in synthetic optical spectra ( see * ? ? ? we note that heii@xmath301085 is detected in the northern aperture only , consistent with previously - published optical ifu data @xcite which shows that the heii@xmath304686 peak is offset from the nucleus , due , most likely , to a highly - ionized wind . we measure a total flux in the ovi@xmath51032,1038 doublet of @xmath52 erg s@xmath4 @xmath53 and @xmath54 erg s@xmath4 @xmath53 for the northern and southern apertures , respectively . the placement of these apertures was chosen so that the spectra could be added ( i.e. , the combined throughput never sums to @xmath61 ; see figure [ fig : cos_regions ] ) , meaning that the total ovi luminosity in the central @xmath015kpc of the phoenix cluster is l@xmath55 erg s@xmath4 . for comparison , @xcite found l@xmath56 erg s@xmath4 for the perseus cluster , using uv spectroscopy from the fuse satellite . under the naive assumption that 100% of the ovi@xmath51032,1038 emission is a result of gas cooling through @xmath010@xmath22k , we can estimate the icm cooling rate at intermediate temperatures . using the latest cloudy code @xcite , and assuming initial plasma properties matched to the measured values in the inner @xmath0150 kpc of the cluster ( @xmath57 , kt = 10 kev ) , we find m@xmath58 erg s@xmath4 ) . for the measured luminosity quoted above , this corresponds to a cooling rate of 55,000 m@xmath2 yr@xmath4 . however , there are additional sources of ionization that may be dominating the line flux here , specifically photoionization from the central agn and heating from shocks . in @xcite we show that both of these ionization sources are contributing to the line flux in high - ionization lines such as [ oiii]@xmath305007 and heii@xmath304863 . in figure [ fig : ovi ] we show the combined ( both apertures ) aperture - matched ( between hst - cos and gmos ) line ratios for various sets of high - ionization uv and optical emission lines . the observed , extinction - corrected line ratios are consistent with having origins in either dust - free agn @xcite or fast radiative shocks @xcite . we subtract the contribution to these emission lines from star - forming regions , converting the uv - derived star formation rates to emission line fluxes for a range of stellar photoionization models @xcite . the larger uncertainties on the sf - corrected line ratios reflects our uncertainty in the ionization field of these young stars . these revised line ratios , with stellar photoionization contributions removed ( figure [ fig : ovi ] , blue points ) , remain consistent with an agn or shock - heated origin . we remove the agn contribution to the emission lines by extracting the optical spectrum from our gmos ifu data in a 1@xmath28 wide aperture centered on the nucleus . assuming a range of agn photoionization models @xcite to estimate the ovi emission , based on the observed heii@xmath304686 flux , we are able to infer the amount of , for example , ovi emission coming from the nucleus , without having any actual spatial information from the cos spectroscopy . the inferred contribution to each line from the agn was subtracted , allowing us to estimate agn - free line ratios in the optical and uv .. the resulting line ratios , which have both stellar and agn photoionization removed ( figure [ fig : ovi ] , green points ) , remain consistent with radiative shocks with @xmath59 km s@xmath4 @xcite . the large error bars , resulting from our combined uncertainty in the stellar and agn photoionization fields , are consistent with the full range of magnetic fields tested by @xcite . this is consistent with our earlier work @xcite , in which we argued that the warm ( 10@xmath51k ) gas is predominantly heated by radiative shocks , based on multiple optical line ratio diagnostics ( e.g. , [ oiii]/h@xmath29 , [ oii]/[oiii ] , as well as the gas kinematics . in this earlier work , we showed evidence for a high - velocity , highly - ionized plume of gas extending north from the central agn , along the same direction as our two cos pointings . such highly - ionized , high - velocity signatures are typically not observed in low-@xmath16 clusters , where the warm gas tends to be only weakly ionized . figure [ fig : ovi ] demonstrates that , even after removing ionization contributions from young stars and agn , the observed high - ionization line ratios are inconsistent with a pure cooling flow model from cloudy @xcite . however , we demonstrate that including a 5000 m@xmath2 yr@xmath4 cooling flow to the model ( representing only @xmath010% of the observed ovi flux , purple point in figure [ fig : ovi ] ) does not significantly change the observed line ratios , which remain consistent with shocks . further , we expect additional ionization from both particle heating ( e.g. , * ? ? ? * ) and mixing ( e.g. , * ? ? ? the former likely contributes at a low level throughout the cluster core . given that the radio flux of the agn is fairly typical of bcgs with significantly less star formation , we expect particle heating to be negligible compared to other ionization sources in the extreme environment of the phoenix cluster . on the other hand , mixing of hot and cold gas is likely contributing significantly given the abundance of multiphase gas in this system . mixing of the hot and cold gas ought to result in an intermediate - temperature phase , which would likely be near @xmath010@xmath60k @xcite . this would lead to substantial ovi emission for relatively low rates of net cooling from the hot to cold phase . thus , we conclude that , as a result of the myriad of additional high - ionization sources , including shocks ( e.g. , * ? ? ? * ) , mixing ( e.g. , * ? ? ? * ) , and particle heating ( e.g. , * ? ? ? * ) , we are unable to constrain the properties of the cooling icm , which is contributing negligibly to the total ovi flux . additional high - ionization lines such as ovii or fexiv would improve these constraints , perhaps allowing an estimate of the amount of gas cooling through @xmath010@xmath60k . in @xcite we presented high - angular - resolution broadband imaging of the cluster core , revealing complex filaments of star formation on scales of @xmath040kpc . while providing unmatched angular resolution , these data were relatively shallow ( @xmath012 m ) , making the detection of more extended , diffuse emission at large radii challenging . in contrast , the megacam data presented , while having inferior angular resolution to the hst data , are considerably deeper , providing substantially - improved sensitivity to faint , extended sources . in figure [ fig : megacam ] we show a three - color image which combines two megacam bands ( @xmath36 , @xmath38 ) and a redder band ( @xmath16 ) from the mosaic - ii camera on the blanco 4-m telescope . in the cluster core , the new data show that the star - forming filaments initially identified with hst @xcite extend significantly further than previously thought . to the south , a pair of filaments extend for @xmath040 kpc each , while to the northwest a third filament extends for @xmath065kpc . the most extended filament extends due north from the central galaxy for @xmath0100kpc ( highlighted in figure [ fig : megacam ] ) . this is the most extended star - forming filament yet detected in a cool core cluster , exceeding the well - studied filaments in the nearby perseus ( 60kpc ; * ? ? ? * ; * ? ? ? * ) , abell 1795 ( 50kpc ; * ? ? ? * ; * ? ? ? * ) , and rxj1532.9 + 3021 ( 50kpc ; * ? ? ? * ) clusters . these four filaments , particularly the northern pair , are exceptionally straight , similar to the northern filament in the perseus cluster . this morphology has been used to argue for relatively low turbulence in the core of the perseus cluster(e.g . * ) , although counter arguments can be made on the basis of icm density fluctuations ( e.g. , * ? ? . the filamentary emission is observed in both the @xmath36 and @xmath38 bands for all four filaments , providing preliminary evidence that this is continuum , rather than line , emission . at the rest frame of the phoenix cluster , the @xmath36-band spans 24003300 . this wavelength range is relatively free of strong emission lines , with mgii@xmath302798 being the only line that may be contributing significantly to the flux in this band . this line should be faint in most scenarios , with the exception of fast ( @xmath61 km s@xmath4 ) radiative shocks in a dense ( @xmath62 @xmath12 ) medium . the total amount of rest - frame uv emission in the outer filaments those that were not included in @xcite is relatively small . for example , the outer 50% of the northern filament ( 50100kpc ) contributes @xmath00.5% of the total blue emission , while the thin northwestern filament contributes an additional @xmath01% . thus , the global estimate of the star formation is not significantly altered by including these deeper data . ) 0.52.0 kev image of the central @xmath0100kpc of the phoenix cluster . this image clearly shows the pair of cavities @xmath010kpc to the north and south of the x - ray peak , despite the lack of any additional processing . lower panel : unsharp masked image of the same region as above . this residual image highlights the small - scale structure in the inner region of the cluster , showing a pair of highly - significant cavities to the north and south and overdense regions to the east and west . , scaledwidth=40.0% ] in figure [ fig : unsharp ] we show a smoothed 0.52.0 kev image of the central @xmath0100kpc in the phoenix cluster . without any additional processing , the pair of cavities in the inner @xmath010kpc , reported initially by @xcite , are evident . an unsharp masked image of the core ( lower panel of figure [ fig : unsharp ] ) reveals significant structure in the inner @xmath030kpc , with overdense regions surrounding the pair of cavities . the cavities , located @xmath010kpc to the north and south of the central agn , are detected at a significance of @xmath025@xmath63 in this residual image . the unsharp mask technique is only sensitive to substructure on a single angular scale , so it is unsurprising that we detect no large - scale asymmetries beyond the core region in this residual image . in order to look for icm substructure on multiple angular scales , we model the 0.52.0 kev surface brightness map with a sum of two beta models and a constant background , using sherpa . the two beta models , meant to represent the core and outer icm , share a common center , ellipticity , and position angle . we note that , since the x - ray spectrum of the central agn is highly obscured @xcite , there is no central point source detected in this soft energy band . during the fit , the cavities ( figure [ fig : unsharp ] ) and point sources were masked to prevent any spurious positive or negative residuals . the best - fitting model was subtracted , with the residual images shown in figure [ fig : residual ] . this residual image shows a significant amount of structure in the central @xmath0150kpc . on scales larger than @xmath0200kpc ( 0.15r@xmath8 ) there is little structure in the residual image ( see rightmost panel in figure [ fig : residual ] ) . if the cluster had recently experienced a recent minor merger , one would expect to find large - scale spiral - shaped surface brightness excesses that would be visible on @xmath6100kpc scales in the residual image ( e.g. , * ? ? ? * ; * ? ? ? . for comparison , the large - scale spiral feature in abell 2029 @xcite represents an excess surface brightness of 22% at a radius of @xmath0150 kpc . at similar radii in the phoenix cluster , the surface brightness uncertainty in the 0.52.0 kev image is @xmath015% , meaning that a similar feature to that observed in abell 2029 would be at the level of the noise . thus , we are unable to rule out a recent _ minor _ merger . in general , the cluster appears relaxed an observation corroborated by @xcite , who find a marked absence of any large - scale `` sloshing '' features in this system based on an independent analysis , labeling it as one of the most relaxed clusters in the known universe . directly to the northwest and southeast of the x - ray peak there are a pair of cavities , detected with @xmath64 , and consistent in location and size to those reported in @xcite . the northern cavity appears to be `` leaking '' to the west , while the southern cavity is slightly extended to the east . if these cavities are indeed not confined to a simple bubble morphology , then the estimates of the agn power derived in previous works @xcite are significantly underestimating the full power output of the agn . interestingly , the four most extended , linear filaments of young stars ( figure [ fig : megacam ] ) appear to trace the agn outflow , with the northern and southern cavities occupying the space between the two filaments in the respective direction . this may be due to the expansion of the cavity compressing the surrounding gas , leading to more rapid cooling along the edges of these bubbles . we note that these two x - ray cavities are at a common distance of @xmath018kpc from the x - ray peak , which is relatively small given the massive size of the cool core ( @xmath0100200 kpc ) . this may provide the first clue towards understanding the extreme level of star formation in the core of the phoenix cluster . under the assumption that these bubbles rise buoyantly , the distance of the cavity from the radio source can be linked to the age of the agn outburst . this would signal that the outburst of agn feedback in the core of the phoenix cluster is relatively recent ( @xmath010100 myr ) , and may not yet have had time to distribute the energy required to offset cooling to the surrounding icm . we will return to this line of reasoning in 4.3 . the residual x - ray images ( figure [ fig : residual ] ) also show a significant amount of excess above a smooth component . we note that this excess is not an artifact induced by the sharp negative residuals ( cavities ) , as these have been masked in the 2d fitting process . the left - most panel of figure [ fig : residual ] shows three distinct regions of excess emission ( colored blue ) . these overdensities may be gas that has been `` pushed aside '' by the inflating bubbles , infalling cool clouds , or low - mass groups that are in the midst of merging with the main cluster . surrounding the northern cavity , the largest of these overdensities extends over @xmath6100kpc ( see central panel of figure [ fig : residual ] ) . the morphology of this overdense region is reminiscent of an infalling cloud of cool gas . alternatively , the fact that both the northern bubble and the northern overdensity share the same curvature suggests that this may simply be the outer rim of a bubble expanding in a dense , cool core . to classify these with more certainty , we require deep enough x - ray data to produce temperature and metallicity maps . on larger scales , there may be a pair of `` ghost '' cavities to the north and south of the cluster center . these are located at distances of @xmath0100kpc from the x - ray peak and are detected most readily in a heavily - smoothed residual image ( right panel of figure [ fig : residual ] ) . we will return to a discussion of the significance of these cavities in 4.3 . with 600 mhz radio contours overlaid in white . the beam shape of the 600 mhz data is shown in the upper right corner . these contours highlight radio emission centered on the x - ray nucleus and elongated in the north - south direction , in the same direction as the most significant set of cavities . at large radii , the radio emission appears to bend in the direction of the `` leaky '' cavities . the extended radio emission to the southeast appears to coincide with the most extended cavity , at a distance of @xmath0100kpc . we also highlight the direction of the ionized wind ( inset ) , discussed in @xcite , with a yellow arrow.,scaledwidth=45.0% ] in @xcite , we present evidence from spatially - resolved optical spectroscopy of a highly - ionized outflow to the north of the central cluster galaxy . this outflow was identified via the presence of a peak in high - ionization optical emission lines ( e.g. , [ oiii ] , heii ) roughly 20kpc north of the galaxy nucleus . in figure [ fig : jets ] ( inset ) , we show that this highly - ionized gas lies in the same direction as the northwestern cavity in the x - ray emission . this may indicate that strong agn feedback , which is excavating the central cavities , is also heating the multiphase gas , presumably via shocks . this is similar to what is observed in abell 2052 @xcite . figure [ fig : jets ] also shows the extended morphology of the 600mhz radio data from gmrt . these data show emission centered on the x - ray peak and central galaxy , extended in the same north - south direction as the cavities . interestingly , the extended emission spreads to the southeast and northwest , similar to the `` leaking '' cavities shown in figure [ fig : residual ] . the most extended emission in the southeast appears to trace the morphology of the potential `` ghost '' cavities , identified in the previous section ( see right panel of figure [ fig : residual ] ) . this low - significance cavity may , instead , be an extension of the primary southern cavity , with the radio emission leaking out to the southeast in a similar manner to the northern cavity . this seems to be the case based on the central panel of figure [ fig : residual ] . alternatively , the presence of isolated cavities at large radii may signal two distinct bursts of agn feedback , separated by a relatively small amount of time . we will discuss these scenarios , among others , in 4 . the s - shaped morphology of the radio data may also be due to bulk motions of the icm . @xcite showed that gas motions in the core of clusters could bend mhd jets . given that the over - dense gas shares a similar morphology , it may be that both the x - ray residuals and the radio morphology can be explained by sloshing of the cool core , as we would expect if this system had undergone a recent minor merger . we note that , given the large , asymmetric beam in these radio data , it is challenging to say with any certainty how well the radio and x - ray morphologies are related . we await higher angular resolution radio observations to establish the morphological connection between the x - ray cavities and the radio emission . clusters in similar mass ranges . the central temperature , pressure , and entropy in phoenix are at the extreme end for low-@xmath16 clusters , and are even more extreme compared to other clusters at @xmath65 which tend to have less massive cool cores @xcite.,scaledwidth=47.5% ] c c c c c[htb ] 7 27 & 6.1@xmath66 & 1.114@xmath67 & 19.2@xmath68 & 0.55@xmath69 + 27 54 & 8.0@xmath70 & 0.672@xmath71 & 42.1@xmath72 & 0.40@xmath73 + 54 108 & 9.9@xmath74 & 0.332@xmath75 & 95.4@xmath76 & 0.56@xmath77 + 108 236 & 11.6@xmath78 & 0.109@xmath79 & 262@xmath80 & 0.31@xmath81 + 236 608 & 17.3@xmath82 & 0.030@xmath83 & 1200@xmath84 & 0.46@xmath85 + 6081148 & 10.8@xmath86 & 0.004@xmath87 & 1980@xmath88 & 0.35@xmath89 [ table : spectra ] in figure [ fig : profiles ] and table [ table : spectra ] we show the results of our x - ray deprojection analysis ( see 2.2 ) . the measured temperature , entropy , and pressure profiles are roughly consistent with the profiles derived based on a mass - modeling approach in @xcite . in general , the profiles look as expected for a strong cool core , with the temperature dropping by a factor of @xmath03 between the peak temperature ( @xmath90kpc , 0.4r@xmath8 ) and the central temperature . the temperature profile is similar in shape to the average profiles of low-@xmath16 @xcite and high-@xmath16 clusters @xcite , which typically peak at 0.30.4r@xmath8 . the factor - of - three drop in the temperature in the inner region is on the high end of what is observed in massive clusters @xcite . the central pressure ( 1.1 kev @xmath12 ) is exceptionally high the highest measured in any cluster to date consistent with the expectation for a strong cool core embedded in one of the most massive clusters known . the pressure drops off rapidly with radius , consistent with the expectation from the universal pressure profile ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? the pressure profile in the phoenix cluster appears more consistent with @xmath49 clusters than other massive clusters at @xmath91 ( see figure [ fig : profiles ] ) . the deprojected central ( @xmath92kpc ) entropy is 19.2 kev cm@xmath17 , consistent with what is found for nearby cool core clusters ( e.g. , * ? ? ? * ; * ? ? ? * ) . the entropy profile shows no evidence of excess entropy ( e.g. , * ? ? ? * ) in the inner region , following the `` baseline '' profile of @xcite from the central bin to the outermost bin at @xmath01mpc . there is no evidence of flattening in the outer part of the entropy profile , as is seen in many low-@xmath16 and high-@xmath16 systems ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the measured metallicity profile is consistent with the declining profile found in low-@xmath16 clusters ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? we note that , given the large uncertainties , the metallicity profile is also consistent with a constant value as a function of radius . the ratio of the cooling time to the freefall time ( @xmath94 ) in the cores of nearby clusters correlates with the presence of multiphase gas , presumably signaling a link between the warm and hot phases @xcite . following @xcite , we estimate the ratio of the cooling time to the freefall time in the core of phoenix using the following equations : @xmath95 @xmath96 ) to the free - fall time ( @xmath97 ) as a function of radius for the phoenix cluster . the shaded grey band represents our uncertainty , which is primarily driven by uncertainty in the temperature profile . we highlight @xmath98 with a horizontal dashed line , which appears to be the approximate threshold for condensation in nearby clusters @xcite . we also highlight the maximum extent of four distinct star - forming filaments ( figure [ fig : megacam ] ) with thin vertical lines , and the average extent with a thick vertical line . the range of filament radii is shown as a shaded blue region . this figure demonstrates that the condensation of the cooling icm may be fueling star formation out to radii of @xmath0100kpc in this extreme system.,scaledwidth=49.0% ] we estimate g(r ) assuming that the core is in hydrostatic equilibrium , and the cooling function ( @xmath99 ) following @xcite , assuming solar metallicity and @xmath100 . the resulting @xmath94 profile is shown in figure [ fig : tctff ] . the minimum value of @xmath101 is reached @xmath020kpc from the central agn , which corresponds to the radius within which the bulk of the star formation is contained ( see figure [ fig : megacam ] and * ? ? ? this minimum value is on par with the minimum value of @xmath94 in some nearby cool core clusters , which can vary from @xmath04 to @xmath020 @xcite . however , such low values of @xmath94 are typically reached at @xmath102kpc @xcite . interestingly , @xmath94 remains below 10 ( the approximate threshold for icm condensation in nearby clusters @xcite ) out to a radius of @xmath060kpc , which is nearly the maximum extent of the star - forming filaments shown in figure [ fig : megacam ] . this , combined with the presence of large - scale star - forming filaments , suggests that condensation of the cooling icm is important in this system out to exceptionally large radii . we will return to this discussion of the cluster s thermodynamic properties later , to quantitatively compare cooling and feedback processes in the cluster core . we have presented new x - ray , ultraviolet , optical , and radio data on the core of the phoenix cluster ( spt - clj2344 - 4243 ) , providing the most complete picture of this extreme system to date . below , we discuss the implications of these new data in the context of previous observations and interpretations @xcite . in @xcite , we calculate the star formation rate ( sfr ) in the core of the phoenix cluster based on rest - frame near - uv imaging from hst , obtaining a value of 800 m@xmath2 yr@xmath4 . in this earlier work , we assume that , in the absence of dust , the slope of the uv spectral energy distribution ( sed ) should be flat @xcite . this allowed us to compute an extinction correction based on the uv color . however , the amount of uv extinction and the age of the stellar population are degenerate with regards to the slope of the uv sed . ideally , we would like an alternate , age - free , estimate of the reddening , which we obtained via balmer line ratios in @xcite . this reddening map showed a peak value of @xmath103 at the central agn , dropping to @xmath104 in the most extended filaments . applying this reddening map to the rest - frame near - uv ( @xmath03000 ) imaging from hst , assuming an extinction curve from @xcite with @xmath105 and a relation between the uv luminosity and sfr from @xcite , yields a revised sfr of 690 m@xmath2 yr@xmath4 . this estimate of the star formation rate may be artificially high . @xcite show that starburst galaxies tend to lack the rapid rise in the uv extinction curve that is characteristic of the milky way and lmc / smc extinction curves . this is confirmed in figure [ fig : fullspec ] , where we show that a shallow , featureless extinction curve ( e.g. , * ? ? ? * ) provides a much better match to the uv - through - optical spectrum of the central galaxy in the phoenix cluster . using the best - fit instantaneous - burst model from figure [ fig : fullspec ] , and averaging over the lifetime of the burst ( 4.5 myr ) , we infer a reduced star formation rate of 490 m@xmath2 yr@xmath4 . as we show in figure [ fig : cos_regions ] , the throughput of the combined hst - cos apertures is only @xmath0100% in the inner region of the starburst . thus , an accurate estimate of the total sfr based on these data requires an aperture correction . based on rest - frame near - uv hst imaging from @xcite , we find that only @xmath026% of the uv continuum emission is missed by the combined hst - cos apertures . assuming that this missing flux originates from regions with similar extinction , we would infer an aperture - corrected star formation rate of 660 m@xmath2 yr@xmath4 . more likely , the extinction in the extended filaments is lower than in the inner few kpc . assuming a much more conservative extinction in the aperture correction ( @xmath106 , representing the minimum value measured in @xcite ) yields sfr = 570 m@xmath2 yr@xmath4 . assuming that the correct answer lies somewhere between these extrema , we infer that the aperture- and extinction - corrected , time - averaged star formation rate for the central galaxy in the phoenix cluster is @xmath3 m@xmath2 yr@xmath4 , where the uncertainty here represents only the uncertainty in the extinction and aperture corrections . we note that this uv - based estimate is consistent with the recent ir - based estimate from @xcite of @xmath107 m@xmath2 yr@xmath4 , suggesting that these different methods may be converging on the right answer . the dominant systematic uncertainty in this estimate is how , exactly , the star formation has proceeded . for example , an instantaneous burst of average age 2 myr , which is preferred by the uv - only spectral fit ( figure [ fig : ovi ] ) , would have a time - averaged star formation rate of 1200 m@xmath2 yr@xmath4 _ before aperture correction_. on the other hand , if the star formation has been proceeding at a constant level for the past 15 myr , the time - averaged rate can be as low as @xmath0300 m@xmath2 yr@xmath4 ( figure [ fig : fullspec ] ) . this uncertainty could be reduced by obtaining deep , rest - frame near - uv ( @xmath02000 ) spectroscopy . at these wavelengths , the stellar population synthesis models are considerably more mature and well - tested , which should lead to overall better fits to the data and a tightening of the allowed parameter space . the megacam data presented here , while significantly deeper , do not provide additional constraints on the global star formation rate , since the bulk of the uv emission is contained in the inner @xmath030kpc . instead , these data provide improved constraints on the extent and large - scale morphology of the star - forming filaments . in the central @xmath030kpc , the uv emission is highly asymmetric ( figure [ fig : megacam ] ) , indicative of a vigorous , turbulent starburst . however , at large radii the young stars are oriented along thin , linear filaments , akin to systems like perseus @xcite and abell 1795 @xcite . in @xcite we demonstrate that the fuel for the observed star formation must originate within the cluster core . a scenario in which cold gas is brought into the core via infalling gas - rich galaxies and/or groups is unfeasible for this system , given the extreme icm density ( efficient ram - pressure stripping ) and amount of gas needed to fuel such a starburst ( i.e. , multiple gas - rich compact groups ) . a popular scenario for the origin of star - forming filaments in nearby cool core clusters involves cool gas being drawn from the cluster core in the wake of buoyant radio - blown bubbles ( e.g. , * ? ? ? * ; * ? ? ? figures [ fig : megacam ] and [ fig : residual ] demonstrate that the most extended filaments surround , rather than trail , the most significant set of cavities , inconsistent with the uplift scenario . indeed , the fact that the star - forming filaments extend beyond these cavities in radius suggests that the cool gas was present before the current epoch of feedback . however , this does not rule out a scenario in which these filaments were uplifted by a previous episode of feedback , with the bubbles responsible for this action having risen to such radii that they are undetectable . it may be that no additional mechanism is necessary to produce the extended , star - forming filaments observed in the core of the phoenix cluster . recent work by @xcite , @xcite , and @xcite has shown that local thermodynamic instabilities can develop when the ratio of the cooling time to the free - fall time ( @xmath94 ) is less than 10 . in figure [ fig : tctff ] , we show this ratio as a function of radius for the phoenix cluster , assuming hydrostatic equilibrium in the computation of @xmath97 . the cooling time is shorter than 10 times the free - fall time over an unprecedented @xmath060 kpc in radius , suggesting that the star - forming filaments may be fueled by local thermodynamic instabilities in the hot icm , which rapidly condense and then `` rain '' down onto the central cluster galaxy @xcite further , we expect a map of @xmath94 to be asymmetric , since the gravitational potential is roughly spherically symmetric ( in the core ) , while the gas density is not ( see figure [ fig : residual ] ) . for example , the spiral - shaped overdensity at @xmath050kpc north of the cluster center will likely have @xmath108 , since the azimuthally - averaged value at that radius is @xmath109 and @xmath110 . thus , despite the fact that @xmath111 at @xmath112kpc , there are likely overdense regions in the icm with @xmath108 out to the full extent of the star - forming filaments at @xmath113kpc . if the star - forming filaments did indeed condense out of the hot icm , there ought to be evidence of cooling in the x - ray spectrum . using the latest _ data , we estimate the classical cooling rate following @xcite , using the equation : @xmath114\sum_{i^{\prime}=1}^{i^{\prime}=i-1}\dot{m}(i^{\prime})}{h(i ) + \delta\phi(i)}\ ] ] where the i and i@xmath40 indices refer to given annuli , @xmath115 is the bolometric x - ray luminosity in a given annulus , @xmath116 is the energy per particle - mass of the hot gas for a given annulus , and @xmath117 is the change in gravitational potential across shell @xmath39 . the first term on the right corresponds to the cooling rate in the absence of any additional heating , while the second term on the right is the correction factor to account for the fact that infalling gas will simultaneously be heated gravitationally . carrying out this sum in the inner 100kpc , we estimate a classical ( luminosity - based ) cooling rate of 3300 @xmath118 200 m@xmath2 yr@xmath4 , consistent with the estimate of 2700 @xmath118 700 m@xmath2 yr@xmath4 from @xcite . assuming the star formation rate of 613 m@xmath2 yr@xmath4 from 4.1 , based on the most realistic reddening and stellar population model , this implies that @xmath020% of the predicted cooling flow is converted into stars . this number is very close to the expected star formation efficiency from cool gas ( 1015% ; * ? ? ? * ; * ? ? ? * ) , leaving open the possibility that cooling at the hot phase may be @xmath0100% efficient . however , the deprojected x - ray spectra in the inner 50kpc show no evidence of a cooling flow , with a single - temperature model providing an excellent fit ( @xmath35 = 128.75 for 128 degrees of freedom ) . likewise , recent _ xmm - newton _ observations published by @xcite find relatively weak cooling signatures in the spectrum at temperatures of 0.33.0 kev , with measurements by the mos ( @xmath119 m@xmath2 yr@xmath4 ) and pn ( @xmath120 m@xmath2 yr@xmath4 instruments yielding values significantly lower than the luminosity - based estimate . this may be because agn feedback has recently halted cooling at high temperatures ( see 4.3 ) , with the star - forming filaments being fueled by previously - cooled gas . indeed , in figure [ fig : fullspec ] , we show that the optical spectrum in the vicinity of the balmer break is consistent with a recently - quenched starburst . alternatively , a significant amount of cooling can be `` hidden '' in the x - ray spectrum , since all of the strong cooling lines at low temperature ( ovii , oviii , fexvii xx ; * ? ? ? * ) are redshifted to @xmath121kev where systematic uncertainties in the _ chandra _ effective area correction are high due to contamination @xcite . further , the disagreement between the spectroscopic cooling rates derived by the pn and mos detectors on _ xmm - newton _ @xcite suggest that systematic uncertainties in our understanding of the soft x - ray response of these detectors ( on both _ chandra _ and _ xmm - newton _ ) will dominate any estimate of the spectroscopic cooling rate . thus , while there is no direct evidence for cooling in the low - resolution x - ray spectrum , we can not rule out the hypothesis that the star - forming filaments are being fueled by local thermodynamic instabilities in the icm . the _ chandra _ x - ray data presented here offer further support for a local fuel supply . the short central cooling time ( @xmath122 yr ) falls below the threshold for the onset of cooling instabilities in the hot atmosphere ( e.g. , * ? ? ? it also meetings the @xmath94 criterion for cooling instabilities ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . however , the level of star formation lies below that expected from unimpeded cooling . phoenix harbors a powerful central radio - agn that is apparently reducing the cooling rate by @xmath080% , operating in a self - regulating feedback loop . the absence of spectral signatures of cooling through @xmath123k ( i.e. , fexvii , ovii ) as in other clusters @xcite , is inconsistent with simple , isobaric cooling models , indicating that the gas may be cooling through other channels ( e.g. , mixing ) or has recently been quenched ( e.g. , * ? ? ? future observations with high - resolution x - ray grating / microcalorimeter spectrometers ( e.g. , astro - h ) will help to break this degeneracy by providing firm constraints on the rate of radiative cooling from high ( @xmath010@xmath124k ) through low ( @xmath010@xmath60k ) temperatures for a sample of nearby , cool core clusters . c c c c c c c c c c c[htb ] + north & 2.6 & 17.3 & 1.7 & 11.3 & 1.3 & 8.6 & 1.9 2.5 & 2.1 5.1 & 0.8 2.5 & 24.7 + south & 2.6 & 17.3 & 2.2 & 14.6 & 1.3 & 8.6 & 2.5 4.2 & 1.8 5.8 & 0.9 4.7 & 26.2 + + & & & & & 4.4 6.7 & & 1.7 7.2 & + + + north & 17.0 & 114.6 & 6.6 & 44.8 & 4.6 & 31.1 & 18 26 & 8.7 13 & 1.8 3.8 & 7.4 + south & 14.4 & 97.2 & 8.6 & 58.4 & 2.9 & 19.3 & 11 34 & 6.4 14 & 1.0 6.7 & 8.1 + + & & & & & 29 60 & & 2.8 10.5 & [ table : cavities ] 50 kpc of the phoenix cluster ( see figure [ fig : residual ] ) . contours highlight structure in the positive ( dashed ) and negative ( solid ) residuals . the red cross highlights the position of the central agn , while red lines show the size and shape of the ellipses used to determine cavity energetics . lower panel : similar to upper panel , but zoomed out by a factor of @xmath05 to show the positions of the extended `` ghost '' cavities . these cavities are also visible in figure [ fig : residual].,scaledwidth=45.0% ] to estimate the enthalpy released by the radio jets , @xmath125 , we measured the cavity sizes and surrounding pressures ( e.g. , * ? ? ? * ; * ? ? ? * ) . the x - ray cavities are seen in the raw image with well - defined , elliptical shapes . we measured their locations and sizes using the residual image after smoothing it with a gaussian kernel ( fwhm = 3 pixels ) . their radial centroids , projected major and minor axes , and energetics are presented in table [ table : cavities ] and shown in figure [ fig : cavs ] . we assume the cavities are ellipsoidal volumes with axes perpendicular to the plane of the sky and equal to the projected major axes ( upper limit ) and minor axes ( lower limit ) . the mean jet power assumes the cavities rose buoyantly to their current locations in the plane of the sky following @xcite . for projected distances of 17 kpc , we find a rise time of @xmath126 yr , assuming the bubbles rise at the sound speed , to a more likely @xmath127 yr , assuming the bubbles rise buoyantly @xcite . these figures depend on the local gravitational acceleration measured from the x - ray pressure profile , assuming hydrostatic equilibrium . the gas pressure and mean gas density at the locations of the cavities were found to be @xmath128 and @xmath129 , respectively . for a total cavity enthalpy of @xmath130 erg , we find a mean jet power of @xmath131 . this mechanical power places phoenix among of the most energetic agn outbursts known . the total radio luminosity ( 10 mhz 10 ghz ) , assuming a constant spectral index of @xmath132 @xcite , is @xmath133 erg s@xmath4 , or roughly one percent of the jet power , consistent with other nearby galaxy clusters @xcite . the total mechanical power output measured in the core of the phoenix cluster is @xmath134 erg s@xmath4 . this is a factor of @xmath04 times less than that quoted by @xcite . this large difference is , for the most part , a result of shallower x - ray data in the earlier study , which resulted in a factor of @xmath02 change in pressure and a factor of @xmath04 change in local gravitational acceleration . the new estimate of the mechanical power lies among the highest measured in a cluster core the core , but is still a factor of @xmath024 times less than the energy required to offset cooling ( @xmath135 erg s@xmath4 ) . given the extreme classical cooling rate in the inner 100 kpc ( 3300 m@xmath2 yr@xmath4 ) , even a small mismatch in energetics over a short time could lead to the observed star formation rate of @xmath0600 m@xmath2 yr@xmath4 in the central galaxy . in figures [ fig : residual ] and [ fig : cavs ] we show the locations of two potential `` ghost '' cavities , at radii of @xmath0100kpc . these potential cavities are detected with signal - to - noise of @xmath07 and @xmath08 ( table [ table : cavities ] ) and are along the same north - south direction as the radio jets . due to the lower detection significance , it is possible that only one ( or neither ) of these cavities are real . however , by computing the total cavity power assuming they are real , we place a rough upper limit on the total amount of power in previous outbursts . following the same procedures as above , we estimate a total cavity power of @xmath136 erg s@xmath4 , or anywhere from a third to 100% of the cooling luminosity . this range overlaps well with the energy in the ongoing outburst at small radii , suggesting that the mechanical power has been roughly constant over the past two epochs of feedback , spanning @xmath0100 myr . in summary , there is strong evidence for ongoing mechanical - mode feedback in the inner @xmath020 kpc of the phoenix cluster , and weaker , but still convincing , evidence for a prior epoch of feedback roughly 100 myr ago . these episodes of feedback have highly uncertain energetics ( factor of a few in p@xmath137 ) , but are consistent with the energy required to offset cooling losses . this scenario is further supported by the fact that the cooling time at @xmath138kpc is longer than the apparent duty cycle of this agn ( @xmath0100 myr ) , suggesting that perfect balance in a given feedback epoch is not necessary at larger radii . future studies with significantly deeper x - ray and low - frequency radio data will provide a more clear picture of both the duty cycle and energetics in these radio - mode outbursts . phoenix is unusual in that its high radio mechanical power is accompanied by a bright quasar with x - ray photon luminosity @xmath139 . this high x - ray flux is relatively stable , with no significant fluctuations observed between the initial chandra and suzaku observations in 2011 and 2012 , and the most recent chandra observations in 2014 . radio agn with comparable radiative and mechanical powers are uncommon , as are radiatively - efficient agn at the centers of clusters @xcite . the eddington luminosity for a @xmath140 black hole is @xmath141 erg s@xmath4 , which would place phoenix at a few percent of the eddington rate , close to the expected transition between radio - mechanical agn and quasars @xcite . standard agn theory posits that the form of power output depends on the specific accretion rate : objects accreting above a few percent of the eddington rate are dominated by radiation and those below are dominated by mechanical outflows . that phoenix exhibits the characteristics of both a quasar and a radio galaxy , suggesting that it may be in transition between the two states . based on their discovery of neutral iron emission from the agn using the suzaku observatory , @xcite have argued that the bcg is a type 2 quasar with a high absorption column density . assuming a bolometric correction factor of 130 @xcite , they argue that the total unabsorbed agn power would be @xmath142 erg s@xmath4 . this figure implies a black hole accretion rate exceeding @xmath143 , which would be the highest rate known in a brightest cluster galaxy . such a high bolometric correction implies a high infrared luminosity , where most of the radiation from the buried agn should be escaping . the total infrared luminosity emerging from the quasar and surrounding star formation is only @xmath144 @xcite . this luminosity is comparable to both the mechanical and x - ray energy fluxes , but lies two orders of magnitude below the bolometric luminosity quoted by @xcite . we conclude that the bolometric correction to the nuclear x - ray luminosity is at most a factor of a few to ten , consistent with the lower values reported by @xcite for high - luminosity agn . therefore the mechanical power of phoenix s radio jets is comparable to the radiation emerging from the quasar , suggesting that it may currently be undergoing a transition from `` quasar mode '' to `` radio mode '' . the 610 mhz image of the core of the phoenix cluster previously revealed the presence of a central compact radio source , associated with the bcg , and diffuse @xmath145 kpc emission surrounding this central source @xcite . the diffuse emission is classified as a radio mini - halo and , at a redshift of @xmath146 , is the most distant known . the integrated flux density of @xmath147 mjy for the mini - halo results in a 1.4 ghz radio power of @xmath148 w hz@xmath4 , scaling with a spectral index of @xmath149 . considering the lifetimes of the radio emitting electrons and the large extent of the mini - halos , a form of in - situ cosmic ray ( cr ) production or re - acceleration is required . an explanation for the presence of cr electrons is that turbulence in the cluster core re - accelerates a population of relativistic fossil electrons @xcite . these electrons could , for example , have originated from the central radio source , associated with the bcg . it has been suggested that gas sloshing , induced by a minor merger event , could generate the required turbulence . this is supported by the observed spatial correlation between the morphologies of some mini - halos and spiral - like sloshing patterns in the x - ray gas of other nearby clusters @xcite . numerical simulations ( e.g. , * ? ? ? * ) provide further support for this model . the x - ray residual maps shown in figure [ fig : residual ] do show some spiral structure reminiscent of sloshing excited by a minor merger , suggesting that phoenix cluster may have undergone a minor merger recently . the lack of any large - scale asymmetry ( rightmost panel of figure [ fig : residual ] ) implies that any merging system must be relatively small compared to the main cluster . due to the mass of this system , the necessary turbulent velocities to accelerate crs can be achieved by the relatively - smooth accretion of low - mass systems , rather than a single , recent event . since the phoenix cluster is amongst the most massive clusters known ( @xmath150 m@xmath2 ; * ? ? ? * ) , and turbulent velocities scale with halo mass ( based on structure formation theory ) , it is feasible that the phoenix cluster maintains a high enough turbulent velocity in its core to continuously re - accelerate a population of relativistic electrons . alternative models , which can also explain the presence of radio mini halos , invoke secondary electrons that are produced by collisions between cr protons and thermal protons ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? these secondary models are only successful in explaining a small range of observed properties ( e.g. , * ? ? ? as one of the most massive cool core clusters known , the phoenix cluster can be considered , in many ways , a `` scaled - up '' , but otherwise normal , system . for example , the @xmath0100kpc extent of the star - forming filaments is extreme , being the most extended cool filaments observed in a cool core cluster to date . however , we showed in @xcite that multiphase gas is typically observed to a maximum radius of @xmath151 in nearby clusters , where @xmath151 is defined at the radius within which the cooling time is less than 3 gyr . for perseus and abell 1795 , this radius corresponds to @xmath060kpc , while for the phoenix cluster we measure @xmath152kpc . thus , while extreme , the extent of these star - forming filaments are scaling as expected with the overall mass of the cluster , and do not challenge our understanding of cool core clusters . the same can be said for the extreme cooling luminosity , central gas pressure , and radio luminosity . the extreme star formation rate in the core of the phoenix cluster ( @xmath0600 m@xmath2 yr@xmath4 ) may be pointing to something unique about this cluster . this starburst accounts for @xmath020% of the classically - predicted cooling flow , suggesting that cooling may be proceeding very efficiently in this system . for comparison , @xcite found , for a sample of nearby clusters with star - forming bcgs , a typical ratio of the star formation rate to the cooling rate of 1.8@xmath153% roughly an order of magnitude lower than what we observe in the phoenix cluster . however , when the most extreme end of this population is considered clusters harboring starburst bcgs such as abell 1835 ( sfr @xmath0 200 m@xmath2 yr@xmath4 ; * ? ? ? * ) , rx j1504.1 - 0248 ( sfr @xmath0 140 m@xmath2 yr@xmath4 ; * ? ? ? * ) , and macs 1931.8 - 2634 ( sfr @xmath0 170 m@xmath2 yr@xmath4 ; * ? ? ? * ) this ratio jumps to @xmath154% . that is , it appears that there is a tail of extreme clusters that are cooling rapidly , fueling efficient star formation in the central galaxy . in this context , the phoenix cluster represents the most extreme of ( but not distinct from ) a subset of rapidly - cooling clusters the cooling properties of the icm are also , as one may expect , quite different in the phoenix cluster than in a typical galaxy cluster . in figure 3 of @xcite , the normalized density and cooling time profiles of @xmath0100 massive galaxy clusters at @xmath155 are shown . of these systems , only the phoenix cluster violates the minimum floor of @xmath98 , and it does so by a substantial margin . the violation of this boundary is thought to initiate strong agn feedback via condensation or precipitation of the cooling icm . this is also reflected in figure [ fig : profiles ] , where we show that the icm entropy profile in the phoenix cluster is consistent with the baseline entropy profile , which represents the expectation in the absence of any feedback @xcite , _ at all radii_. when considering all 165 known galaxy groups and clusters with existing _ chandra _ data and temperatures in the range 4 kev @xmath41 t@xmath156 @xmath41 15 kev , @xcite did not find a single cluster with an entropy profile as steep as phoenix over the range @xmath157 kpc . indeed , the only other cluster known with such a steep entropy profile is h1821 + 643 @xcite , which also harbors a cluster - centric qso , presumably fueled by a rapidly - cooling core . the consistency between the entropy profile in the phoenix cluster and the baseline entropy profile suggests that the icm has been allowed to cool in this unique system with relatively little resistance . perhaps providing the link between these two characteristic aspects of the phoenix cluster the extreme starburst and rapidly - cooling icm is the central agn , which is unusual in its own way . this central agn is one of only a few known cluster - centric qsos @xcite and appears to be transitioning from quasar - mode to radio - mode . the current radio outburst appears to have recently begun , with the radio - blown bubbles appearing small in size and confined to the innermost regions of the cluster , while a previous outburst may have occurred as recently as @xmath0100myr ago . this agn appears to be heavily influencing the local environment , highly ionizing the cool gas in the central @xmath010kpc , driving an ionized outflow north of the central galaxy @xcite , and inflating bubbles in the dense icm which may be re - directing the extended , star - forming filaments ( see figure [ fig : jets ] ) . understanding the extreme nature of this agn , specifically why it appears to have had transitioned from mechanical - mode to radiative - mode , and now back to mechanical - mode , will likely provide the key to understanding the other oddities in the core of the phoenix cluster . in this paper we present a detailed , multi - wavelength study of the core of the phoenix galaxy cluster ( spt - clj2344 - 4243 ) . the analysis presented here is based on a combination of new and archival data at radio ( gmrt ) , optical ( hst - wfc3 , magellan megacam ) , ultraviolet ( hst - wfc3 , hst - cos ) , and x - ray ( _ chandra_-acis - i ) wavelengths . the primary results of this latest study are summarized as follows : * complex , star - forming filaments are observed in the rest - frame ultraviolet to extend up to @xmath0100kpc from the central cluster galaxy in multiple directions . these newly - detected filaments extend a factor of @xmath02 times further than the previously - reported maximum extent of star formation . * modeling the combined uv - optical ( rest - frame 9006000 ) spectrum of the central cluster galaxy reveals a massive ( @xmath44 m@xmath2 ) , young ( @xmath04.5 myr ) stellar population . based on these data , we estimate a time - averaged , extinction- and aperture - corrected star formation rate of @xmath3 m@xmath2 yr@xmath4 . we note that this estimate can vary by a factor of @xmath02 in either direction by changing our assumptions on the star formation history . * the best - fitting dust model is significantly `` grayer '' than that measured for the milky way and smc / lmc , consistent with observations for nearby starburst galaxies and distant quasars . the data show no evidence of the 2175 bump , suggesting that there may be a process either destroying or preventing the formation of small grains and those contributing to this bump ( i.e. , pahs ) . * we detect significant ovi@xmath51032,1038 emission ( l@xmath158 erg s@xmath4 ) in the inner @xmath015 kpc of the cluster core . this emission is consistent with having origins primarily in the central agn and in shock - heated gas along a northern ionized outflow . we are unable to put constraints on what fraction of this emission may originate from the cooling icm the data are consistent with both a complete lack of cooling and a massive ( i.e. , 5000 m@xmath2 yr@xmath4 ) cooling flow . * we confirm the presence of strong ( s / n @xmath0 25 ) x - ray cavities in the inner 20kpc of the cluster core . the total mechanical energy in these cavities is p@xmath137 = 27 @xmath37 10@xmath159 erg s@xmath4 , depending on their intrinsic shape , making this one of the most powerful outbursts of radio - mode feedback known . the inferred jet power from these cavities is slightly less than the cooling luminosity ( l@xmath160 erg s@xmath4 ) in the inner 100 kpc . * we find that the bolometric x - ray luminosity of the agn ( @xmath161 erg s@xmath4 ) corresponds to a few percent of the eddington luminosity , consistent with a recent transition from quasars to radio - mode agn . the similarity of the bolometric luminosity and the mechanical power further supports this picture of an agn transitioning from a radiatively - efficient mode to a mechanical mode . * we find evidence for an additional set of x - ray cavities at larger radii ( @xmath0100 kpc ) , suggesting that there may have a been prior episode of radio - mode feedback @xmath0100 myr ago . assuming that both potential `` ghost '' cavities are real , this prior episode of feedback may have had jet powers of 2.8 10.5 @xmath37 10@xmath159 erg s@xmath4 , which is similar to what is measured in the inner set of cavities . this implies a relatively constant mechanical output of the central agn between bursts , with a duty cycle of @xmath0100 myr . * the azimuthally - averaged cooling time of the icm is shorter than the precipitation threshold ( @xmath94 = 10 ) at radii of @xmath5050 kpc , suggesting that local thermodynamic instabilities in the hot icm may be fueling both the star formation and agn feedback . dense substructures extending beyond 50kpc likely have exceeded the precipitation threshold out to much larger radii ( @xmath0100 - 150kpc ) . * we see significant substructure in the inner @xmath0200kpc ( 0.15r@xmath8 ) of the cluster . the spiral shape of this structure is reminiscent of infalling cool clouds . this may be cooling gas that has been redirected by the strong mechanical feedback , infalling cool group - sized halos , or sloshing of the cool core . the presence of a radio mini - halo supports the x - ray observations that the inner core is highly turbulent . * outside of the cool core ( @xmath162 ) , the cluster appears relaxed and has thermodynamic properties typical of other clusters at similar mass and redshift . while illuminating in many ways , these new data leave several questions unanswered . it remains unclear how such vigorous star formation is sustained in the midst of massive , radio - mode outbursts from the central agn . the absolute estimate of the star formation rate in the central galaxy hinges on understanding the star formation history , which would benefit from broader - band near - uv spectroscopy , while confirmation that star formation is being fueled by cooling of the hot icm awaits high resolution x - ray spectroscopy . confirmation of the extended cavities at large radii requires deeper x - ray follow - up , alongside high angular resolution low - frequency radio imaging . as the most extreme cool core cluster known , such follow - up studies of this systems will allow a deeper understanding of the complex and ongoing war between agn feedback and cooling in dense cluster cores . m. m. acknowledges support by nasa through contracts hst - go-13456.002a ( hubble ) and go4 - 15122a ( chandra ) , and hubble fellowship grant hst - hf51308.01-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 . the guaranteed time observations ( gto ) included here were selected by the acis instrument principal investigator , gordon p. garmire , of the huntingdon institute for x - ray astronomy , llc , which is under contract to the smithsonian astrophysical observatory ; contract sv2 - 82024 . j.e.c . acknowledges support from national science foundation grants plr-1248097 and phy-1125897 . b.r.m . acknowledges generous financial support from the natural sciences and engineering research council of canada . r.j.w . is supported by nasa through the einstein postdoctoral grant number pf2 - 130104 awarded by the chandra x - ray center , which is operated by the smithsonian astrophysical observatory for nasa under contract nas8 - 03060 . j.hl is supported by nserc through the discovery grant and canada research chair programs , as well as frqnt . d.a . acknowledges support from the dlr under projects 50 or 1210 & 1407 , and from the dfg under project ap 253/1 - 1 . a.c.e . acknowledges support from stfc grant st / i001573/1 . , m. , van langevelde , h. j. , reynolds , c. , & cotton , b. 2006 , in astronomical society of the pacific conference series , vol . 351 , astronomical data analysis software and systems xv , ed . c. gabriel , c. arviset , d. ponz , & s. enrique , 497 , b. a. , gauron , t. m. , geary , j. c. , ordway , m. p. , & roll , j. b. 1998 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 3355 , optical astronomical instrumentation , ed . s. dodorico , 477486
we present new ultraviolet , optical , and x - ray data on the phoenix galaxy cluster ( spt - clj2344 - 4243 ) . deep optical imaging reveals previously - undetected filaments of star formation , extending to radii of @xmath050100 kpc in multiple directions . combined uv - optical spectroscopy of the central galaxy reveals a massive ( @xmath1 m@xmath2 ) , young ( @xmath04.5 myr ) population of stars , consistent with a time - averaged star formation rate of @xmath3 m@xmath2 yr@xmath4 . we report a strong detection of ovi @xmath51032,1038 , which appears to originate primarily in shock - heated gas , but may contain a substantial contribution ( @xmath61000 m@xmath2 yr@xmath4 ) from the cooling intracluster medium . we confirm the presence of deep x - ray cavities in the inner @xmath010 kpc , which are amongst the most extreme examples of radio - mode feedback detected to date , implying jet powers of @xmath7 erg s@xmath4 . we provide evidence that the agn inflating these cavities may have only recently transitioned from `` quasar - mode '' to `` radio - mode '' , and may currently be insufficient to completely offset cooling . a model - subtracted residual x - ray image reveals evidence for prior episodes of strong radio - mode feedback at radii of @xmath0100 kpc , with extended `` ghost '' cavities indicating a prior epoch of feedback roughly 100 myr ago . this residual image also exhibits significant asymmetry in the inner @xmath0200 kpc ( 0.15r@xmath8 ) , reminiscent of infalling cool clouds , either due to minor mergers or fragmentation of the cooling icm . taken together , these data reveal a rapidly evolving cool core which is rich with structure ( both spatially and in temperature ) , is subject to a variety of highly energetic processes , and yet is cooling rapidly and forming stars along thin , narrow filaments .
over the last two decades spectroscopic and photometric observations of nearby early - type galaxies have shown that they obey tight scaling relations . for example more luminous galaxies are redder @xcite ; the strength of the mg - absorption ( at @xmath65175 ) increases with increasing central velocity dispersion @xcite ; and early - type galaxies populate only a band ( or ` plane ' ) in the three dimensional space of central velocity dispersion , effective radius , and mean effective surface brightness , the so called ` fundamental plane ' @xcite . the tightness of the scaling relations has traditionally been interpreted as evidence for a very homogeneous population of early - type galaxies , old stellar populations and similar dynamical make - up of the galaxies @xcite . for example the colour - magnitude relation is perhaps best explained by a correlation of the metal abundance of the stellar population and total galaxy mass with no age gradient along the sequence @xcite . however , evidence is being found that elliptical galaxies do generally have complicated dynamical structures ( decoupled cores ) and disturbed morphologies ( shells ) , suggesting an extended and complex assembly such as in a hierarchical merging picture ( cole et al . . furthermore measurements of absorption line - strength indices together with evolutionary population synthesis models suggest that the mean age of the stellar populations in early - type galaxies span a wide range @xcite . there is a hint of a connection between the detailed morphology and the mean ages of the stellar populations as ellipticals with disky isophotes show on average younger ages @xcite . this is in agreement with the studies of cluster galaxies at modest look - back times of 3 - 5 gyr where an increased fraction of blue , late - type galaxies is found @xcite . probably this change is associated with the transformation of spiral galaxies into s0 types observed over the same interval @xcite . how can we reconcile the existence of tight scaling relations for present day early - type galaxies with the evidence of an extended galaxy assembly and star formation history ? an answer to this question rests largely on our understanding of the detailed physical processes which determine the spectral or photometric properties seen in scaling relations . broad band colours and line - strength indices of integrated stellar populations are sensitive to changes in both their age and metallicity . in fact it is possible to find many different combinations of ages and metallicities which produce similar overall photometric properties ( age - metallicity degeneracy , worthey 1994 ) . in order to explain tight scaling relations , despite the presence of a substantial diversity in the integrated stellar populations , it has been suggested that the ages and metallicities of integrated stellar populations ` conspire ' such that deviations from a given scaling relation due to age variations are balanced by an appropriate change in metallicity and vice versa ; an age - metallicity anti - correlation @xcite . for example , a re - analysis of the sample by trager et al . appears to show that galaxies with a young stellar component are also more metal rich . recently , evidence for a negative age - metallicity correlation has also been found in other data sets @xcite and in literature compilations @xcite . in contrast to this kuntschner & davies ( 1998 ) and kuntschner ( 2000 ) did not find strong evidence for the existence of an age - metallicity anti - correlation in their analysis of the fornax cluster . however , to date there is a lack of large , high quality samples with which it would be possible to probe the population of early - type galaxies as a whole . in this paper we seek to further our understanding of scaling relations such as mg@xmath0 . we investigate the spread in the mean ages and metallicities _ at a given mass _ but also probe the possibility of age and metallicity trends _ along _ the relations . we use established methods such as the analysis of the scatter @xcite in the mg@xmath0 relation and line - strength age / metallicity diagnostic diagrams , and take advantage of recent advances in stellar population modeling such as improved corrections for non - solar abundance ratios @xcite . we employ a homogeneous and high - quality subset of the data collected initially for the smac peculiar motion survey @xcite . the sample is not complete but includes galaxies drawn mostly from nearby clusters and groups with some galaxies from less dense environments . the present paper is organized as follows . section [ sec : data ] describes the sample selection , observations , data reduction and the measurements of the line - strength indices . section [ sec : mg_sig ] presents the mg@xmath0 and fe@xmath0 relations with an initial investigation into the scatter and slopes . in section [ sec : age ] we present a detailed analysis of age - metallicity diagnostic diagrams including an improved treatment of non - solar abundance ratios . the effects of the non - solar abundance ratio corrections on the age - metallicity estimates and on the index@xmath0 relations are presented in sections [ sec : imp_index ] and [ sec : imp_sig ] respectively . our conclusions are given in section [ sec : conclusions ] . the data reported here is from the smac i97ma spectroscopic run ( see smith et al . 2000 for further details ) . this run was undertaken principally to provide a secure connection of the smac data to other data sets and concentrated on observing galaxies with measurements previously reported by dressler ( 1984 ) , ( 1993 ) and davies et al . accordingly the smac i97ma data shows a good overlap with the lick / ids system with 48 galaxies in common . in selecting targets , higher priority was given to galaxies with measurements from multiple sources . while the sample is dominated by ellipticals in the virgo and coma clusters , it does include a few s0s and galaxies in less dense environments ( see table [ tab : table ] of the appendix ) . it is not a complete sample in any sense and our sensitivity to environmental effects is limited . however , we did not find strong differences between galaxies from different environments or morphologies in our sample and thus throughout this paper we do not split up our sample but treat it as a whole . the observations were carried out by jrl at the 2.5 m isaac newton telescope ( march 1997 ) . the intermediate dispersion spectrograph was used in conjunction with the 23.5 cm camera , the 900v grating and a tek1k chip . with a slit width of 3 arcsec , an instrumental resolution of @xmath7 fwhm was achieved , equivalent to an instrumental dispersion ( @xmath8 ) of 98 . the spectra cover a wavelength range of 1024 , sampled at 1 pix@xmath9 from 4800 to 5824 . in total 200 galaxy spectra were obtained including many repeat observations . along the slit the central five pixels of each observation were combined giving an effective aperture of @xmath10 arcsec@xmath11 . for other basic data reduction details and the velocity dispersion measurements see smith et al . ( 2000 ) . the wavelength range and the redshift distribution of our data allows the measurement of five important line - strength indices ( , mg@xmath12 , , fe5270 , fe5335 ) in the lick / ids system ( worthey 1994 , hereafter w94 , trager et al . the indices were measured on the galaxy spectra after they had been corrected to a relative flux scale and were broadened to the lick / ids resolution and were measured from the un - broadened spectra . ] ( see also worthey & ottaviani 1997 ) . the index measurements were then corrected for internal velocity broadening of the galaxies following the method outlined in kuntschner ( 2000 ) . the formal measurement errors were rescaled to ensure agreement with the error estimates derived from the repeat observations ( the scale factors ranged from 0.80 to 1.07 for all indices but mg@xmath12 with a scale factor of 0.44 ) . in order to maximize the s / n , multiple index measurements of the same galaxy were averaged giving a sample of 140 different galaxies . the effective s / n per of these combined measurements ranges from 14 to 78 . in order to keep the errors of the index measurements reasonably low we decided to use only index measurements with an effective @xmath13 per giving a final sample of 72 galaxies ( median s / n @xmath14 per ) which we will analyze in this paper . all of our data were obtained using the same physical aperture size , but since the galaxies span a factor of @xmath610 in distance , the spectra sample different physical aperture sizes in kpc . since line indices and velocity dispersion exhibit measurable gradients with respect to radius , it is necessary to correct these parameters to a ` standard ' aperture . here we adopt a generalisation of the aperture correction due to jrgensen , franx & kjrgaard ( 1995 ) , which scales parameters to a circular aperture of diameter 1.19 @xmath15 kpc , @xmath16 mpc@xmath9 , equivalent to a circular aperture of 3.4 arcsec at the distance of coma . the strength of the correction ( the correction in the quantity per dex in aperture size ) is @xmath17 for @xmath18 , mg@xmath12 and , but @xmath19 for and zero ( no correction ) for . these correction strengths are based on the average line strength gradients observed in early - type galaxies @xcite . the correction strengths adopted in this paper are overall similar to the ones used in jrgensen ( 1997 ) , however we use @xmath19 for the index , whereas jrgensen used @xmath20 . in order to determine the systematic offsets between our line - strength system and the original lick / ids system ( on which the worthey 1994 model predictions are based ) , we compared index measurements for the 48 galaxies in common with the lick galaxy library @xcite . we note that the lick / ids aperture was @xmath21 arcsec ; so for the following comparison our data was corrected to the ( fixed ) lick / ids aperture using the formula given in jrgensen et al . ( 1995 ) . generally there is good agreement between the two data sets and only small offsets are found ( see table [ tab : lick_off ] and figure [ fig : offset_all ] ) . in order to match the lick / ids system we removed the offsets from our data . the median measurement errors in the indices , for our sample galaxies , are given in column 4 of table [ tab : lick_off ] . for a definition of non - standard line - strength indices see section [ sec : special_indices ] . notes : the first column shows the name of the galaxy . if the galaxy is found in the ngc or ic ( nilson 1973 ) catalogues , then it will have the name from the corresponding catalogue , in the order of preference given above . there are also some galaxies denoted by their name from the cgcg ( zwicky et al . 1961 , zxxx - xx ) catalogue , dressler s ( 1980 , dxxx ) catalogue and one galaxy found in the godwin , metcalfe & peach catalogue ( 1983 , gmp 1652 ) . the second column indicates the environment . if the galaxy is a member of a well known cluster or group , the name is given . for smaller associations we just indicate whether the galaxies are in a very low density environment ( `` - '' ) or belong to a group . the third column shows the effective s / n per of the line - strength measurements . the velocity dispersion @xmath22 and the line - strength indices have been aperture corrected to @xmath23kpc , equivalent to 34 at the distance of the coma cluster . the line - strength indices are calibrated onto the lick / ids system and corrected to zero velocity dispersion with the exception of the index which is calibrated to the ( 1993 ) system . mg@xmath12 is given in mag , whereas the other line - strength indices are given in .
we investigate the mg@xmath0 and @xmath0 relations in a sample of 72 early - type galaxies drawn mostly from cluster and group environments using a homogeneous data - set which is well - calibrated onto the lick / ids system . the small intrinsic scatter in mg at a given @xmath0 gives upper limits on the spread in age and metallicity of @xmath1% and @xmath2% respectively , if the spread is attributed to one quantity only and if the variations in age and metallicity are uncorrelated . the age / metallicity distribution as inferred from the _ vs _ diagnostic diagram reinforces this conclusion , as we find mostly galaxies with large luminosity weighted ages spanning a range in metallicity . using monte - carlo simulations , we show that the galaxy distribution in the _ vs _ plane can not be reproduced by a model in which galaxy age is the only parameter driving the index-@xmath0 relation . in our sample we do not find significant evidence for an anti - correlation of ages and metallicities which would keep the index@xmath0 relations tight while hiding a large spread in age and metallicity . as a result of correlated errors in the age - metallicity plane , a mild age - metallicity anti - correlation can not be completely ruled out given the current data . correcting the line - strengths indices for non - solar abundance ratios following the recent paper by trager et al . , leads to higher mean metallicity and slightly younger age estimates while preserving the metallicity sequence . the [ mg / fe ] ratio is mildly correlated with the central velocity dispersion and ranges from [ mg / fe ] @xmath3 0.05 to 0.3 for galaxies with @xmath4 . under the assumption that there is no age gradient along the index@xmath0 relations , the abundance - ratio corrected mg@xmath0 , fe@xmath0 and @xmath0 relations give consistent estimates of @xmath5}/ \delta \log \sigma \simeq 0.9 \pm 0.1 $ ] . the slope of the @xmath0 relation limits a potential age trend as a function of @xmath0 to 2 - 3 gyrs along the sequence . = = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] galaxies : abundances - galaxies : formation - galaxies : elliptical and lenticular - galaxies : kinematics and dynamics
understanding the nature of the mesonic resonances listed in ref . @xcite is an important topic of both experimental and theoretical hadron physics . in the vector kaonic sector with @xmath2 ( @xmath6 stands for isospin , @xmath7 for total spin , and @xmath8 for parity ) , the resonance @xmath9 corresponds very well to the expected quark - antiquark states @xmath10 @xmath11 @xmath12 , @xmath13 ; moreover , its spectral function is nicely described by a ( relativistic ) breit - wigner function . on the contrary , the scalar kaonic sector @xmath5 is much more complicated . two resonance are listed in the pdg below 1.5 gev : the broad but well - established @xmath14 and the very broad and light @xmath3 ( also known as @xmath15 ) , whose existence still requires definitive confirmation ( for discussions , see e.g. refs . @xcite and refs . therein ) . the inclusion of @xmath15 in the summary table of pdg would allow to complete the nonet of light scalar states in the energy region below @xmath16 gev . the aim of this work , based on ref . @xcite , is to study the nature of the vector and scalar kaonic resonances . by using qft lagrangians , we determine the coordinates of the poles on the complex plane and study their nature by using the large-@xmath0 limit . we confirm that @xmath17 and @xmath18 are standard quark - antiquark states , while @xmath3 is a dynamically generated state . in order to describe @xmath17 and @xmath14 we introduce relativistic lagrangians that couples them to one kaon and one pion:@xmath19 the expressions above contain non - derivative and derivative interaction terms . the dots stay for the sum over isospin and hermitian conjugation . the terms in eq . ( 1 ) naturally emerge as a subset of more complete mesonic models , e.g. ref . @xcite . according to our models , the decay widths of @xmath20 and @xmath21 ( as function of the running mass @xmath22 ) are : @xmath23 f_{\lambda}(m)\text { , } % \]]@xmath24 ^{2}f_{\lambda } ( m)\text { , } % \ ] ] where the form factor @xmath25 has been introduced . @xmath26 is an energy scale , @xmath27 the three - momentum of one outgoing particle , @xmath28 the kaon mass , and @xmath29 the pion mass . the on - shell decay widths are obtained by setting @xmath22 to the masses of @xmath20 or @xmath14 : this is accurate in the former case , but quite imprecise in the latter . the ( scalar part of the ) propagator of the resonances is given by @xmath30 ^{-1}$ ] , where @xmath31 is the bare mass of @xmath32 and @xmath33 is the one - loop contribution . the spectral function which determines the probability that resonance has a mass between @xmath22 and @xmath34 reads : @xmath35 spectral functions must be normalized to unity . for details of the used formalism , see ref . @xcite . the spectral functions of @xmath17 and @xmath14 are shown in fig . 1 . for the vector kaon we observe a single peak close to @xmath36 gev and a ( unique ! ) pole at @xmath37 gev . in the scalar sector , there is a broad peak at about @xmath38 gev , but no peak corresponding to the light @xmath15 ( there is only a broad enhancement in the low - energy regime ) . in this channel , it turns out that there are two poles : @xmath39 gev , which corresponds to the seed state @xmath14 , and @xmath40 gev , which corresponds to @xmath3 as an additional companion pole . the parameters for the scalar channel were determined via a fit to existing pion - kaon data , see ref . @xcite for details . we studied the change of the spectral function for different values of the number of colors @xmath0 with the rescaling @xmath41 ( and so for @xmath42 and @xmath43 ) . when @xmath0 increases , the interaction becomes smaller . in both channels we observe that the peak becomes narrower and higher . however , in the scalar sector the enhancement of the @xmath15 becomes smaller for increasing @xmath0 . in fig . 2 we show the trajectories of the poles for increasing @xmath0 . one sees that the poles of @xmath17 and @xmath14 tend to the real axis , while that of @xmath3 goes away from it and finally disappears for @xmath44 in conclusion , all these results , with special focus on fig . 2 which is the main outcome of the present proceedings , confirm that @xmath3 is a dynamically generated non - quarkonium meson . we have discussed the nature of @xmath17 , @xmath3 , and @xmath14 by using qft models presented in ref . we find that @xmath17 and @xmath14 are regular quark - antiquark mesons ( see the quark - model review in @xcite ) . in both cases , the spectral function has a well - pronounced peak ; in the large-@xmath0 limit the positions of the poles tend to the real axis , as it should for conventional mesons . on the contrary , @xmath3 does not correspond to a peak of the scalar spectral function , but there is a related pole in the complex plane . the original lagrangian contains a single scalar field , which is associated to @xmath14 , hence @xmath3 emerges as a companion pole of @xmath14 . in the large-@xmath0 limit the pole of @xmath3 disappears , confirming its non - quarkonium nature . e. van beveren , t. a. rijken , k. metzger , c. dullemond , g. rupp and j. e. ribeiro , _ z. phys . _ * c30 * , 615 - 620 ( 1986 ) . d. black , a. h. fariborz , f. sannino and j. schechter , phys . rev . d * 58 * ( 1998 ) 054012 . t. wolkanowski , m. soltysiak and f. giacosa , arxiv:1512.01071 [ hep - ph ] . d. parganlija , p. kovacs , g. wolf , f. giacosa and d. h. rischke , _ phys . _ * d87 * , 014011 ( 2012 ) ; s. janowski , f. giacosa and d. h. rischke , phys . rev . d * 90 * ( 2014 ) 11 , 114005 [ arxiv:1408.4921 [ hep - ph ] ] .
we study the spectral functions , the poles and their trajectories for increasing @xmath0 of the vector kaon state @xmath1 characterized by @xmath2 , and of the scalar kaons @xmath3 and @xmath4 characterized by @xmath5 . to this end , we use relativistic qft s lagrangians with both derivative and non - derivative terms . in the vector kaonic sector the spectral function is well approximated by a breit - wigner function : there is one single peak and , correspondingly , a single pole in the complex plane . on the contrary , in the scalar sector , although the lagrangian contains only one scalar kaonic field , we find two poles , one corresponding to a standard quark - antiquark , , seed state @xmath4 and one to a companion dynamically generate pole @xmath3 . the latter does not correspond to any peak in the scalar kaonic spectral function , but only to an enhancement in the low - energy regime .
we consider the following pde system @xmath5}(\chi_t)-\operatorname{div}({\bf d}(x,\nabla\chi))+w'(\chi ) \ni - b'(\chi)\frac{{\varepsilon({\bf u})}\mathrm{r}_e { \varepsilon({\bf u})}}2 + { \vartheta}\quad\hbox{in } \omega \times ( 0,t),\label{eqii}\end{aligned}\ ] ] which describes a thermoviscoelastic system occupying a reference domain @xmath6 , @xmath7 , supplemented with suitable initial and boundary conditions . the symbols @xmath0 and @xmath1 respectively denote the absolute temperature of the system and the vector of _ small displacements_. depending on the choices of the functions @xmath8 and @xmath9 , we obtain a model for _ phase transitions _ : in this case , @xmath2 is the order parameter , standing for the local proportion of one of the two phases ; for _ damage _ : in this case , @xmath2 is the damage parameter , assessing the soundness of the material . we will assume that @xmath2 takes values between @xmath10 and @xmath11 , choosing @xmath10 and @xmath11 as reference values : for the _ pure phases _ in phase change models ( for example , @xmath3 stands for the solid phase and @xmath4 for the liquid one in solid - liquid phase transitions , and one has @xmath12 in the so - called _ mushy regions _ ) ; for the completely _ damaged _ @xmath3 and the _ undamaged _ state @xmath13 , respectively , in damage models , while @xmath12 corresponds to _ partial damage_. let us now briefly illustrate the derivation of the pde system . we shall systematically refer for more details to @xcite , where we dealt with the case of phase transitions in thermoviscoelastic materials , and just underline here the main differences with respect to the discussion in @xcite . equation , governing the evolution of the displacement @xmath1 , is the classical balance equation for macroscopic movements ( also known as the _ stress - strain relation _ ) , in which inertial effects are taken into account as well . it is derived from the principle of virtual power ( cf . @xcite ) , which yields @xmath14 where the symbol @xmath15 stands both for the scalar and for the vectorial divergence operator , @xmath16 is the stress tensor , and @xmath17 an exterior volume force . for @xmath18 , we adopt the well - known constitutive law @xmath19 with @xmath20 the linearized symmetric strain tensor , which in the ( spatially ) three - dimensional case is given by @xmath21 , @xmath22 ( with the commas we denote space derivatives ) . hence , the explicit expression of @xmath23 depends on the form of the free energy functional @xmath24 and of the pseudopotential of dissipation @xmath25 . the former is a function of the state variables , namely @xmath2 , its gradient @xmath26 , the absolute temperature @xmath0 , and the linearized symmetric strain tensor @xmath20 . according to moreau s approach ( cf . @xcite and references therein ) , we include dissipation in the model by means of the latter potential , which depends on the dissipative variables @xmath27 , @xmath28 , and @xmath29 . we will make precise our choice for @xmath30 and @xmath31 below , cf . and . we shall supplement with a zero dirichlet boundary condition on the boundary of @xmath32 @xmath33 yielding a _ pure displacement _ boundary value problem for @xmath1 , according to the terminology of @xcite . however , our analysis carries over to other kinds of boundary conditions on @xmath1 , see remark [ rem - other - b.c . ] . following frmond s perspective , is coupled with the equation of microscopic movements for the phase variable @xmath2 ( cf . @xcite ) , leading to . let @xmath34 ( a density of energy function ) and @xmath35 ( an energy flux vector ) represent the internal microscopic forces responsible for the mechanically induced heat sources , and let us denote by @xmath36 and @xmath37 their dissipative parts , and by @xmath38 and @xmath39 their non - dissipative parts . standard constitutive relations yield @xmath40 then , if the volume amount of mechanical energy provided to the domain by the external actions ( which do not involve macroscopic motions ) is zero , the equation for the microscopic motions can be written as @xmath41 where @xmath42 and @xmath35 will be specified according to the expression of @xmath30 and @xmath31 . the natural boundary condition for this equation of motion is @xmath43 where @xmath44 is the outward unit normal to @xmath45 . thus ( cf . ) we obtain the homogeneous neumann boundary condition on @xmath2 @xmath46 finally , equation is derived from the internal energy balance @xmath47 where @xmath48 denotes a heat source and @xmath49 and @xmath50 are obtained from @xmath30 and @xmath31 by means of the standard constitutive relations @xmath51 we couple equation with a no - flux boundary condition : @xmath52 implying ( cf . ) the homogeneous neumann boundary condition @xmath53 from the above relations and the following choices for the free energy functional and of the pseudopotential of dissipation ( cf . and ) , we derive the pde system within the _ small perturbation assumption _ @xcite ( i.e. neglecting the quadratic terms @xmath54 on the right - hand side of the heat equation ) . in agreement with thermodynamics ( cf . @xcite and ( * ? ? ? * sec . 4 , 6 ) ) , we choose the volumetric free energy @xmath24 of the form @xmath55 where @xmath56 is a concave function of @xmath0 . notice that the symmetric , positive - definite elasticity tensor @xmath57 is pre - multiplied by a function @xmath9 of the phase / damage parameter @xmath2 . in particular , in the case of phase transitions in viscoelastic materials , a meaningful choice for @xmath9 is @xmath58 , or a function vanishing at @xmath11 ( * ? ? ? 4.5 , pp . 42 - 43 ) . this reflects the fact that we have the full elastic contribution of @xmath59 only in the non - viscous phase , and that such a contribution is null in the viscous one ( i.e. when @xmath4 ) ; for damage models a significant choice is instead @xmath60 ( cf . @xcite and ( * ? ? ? 6.2 , pp . 102 - 103 ) for further comments on this topic ) . the term @xmath61 represents the classical elastic contribution in which the stiffness of the material decreases as @xmath2 approaches @xmath10 , i.e. during the evolution of damage . the term @xmath62 is a _ mixture _ or _ interaction free - energy_. we shall suppose that @xmath63 is a normal integrand , such that for almost all @xmath64 the function @xmath65 is convex , @xmath66 , with @xmath67-growth , and @xmath68 . hence , the field @xmath69 , @xmath70 , leads to a @xmath67-laplace type operator in . the prototypical example is @xmath71 , yielding @xmath72 . let us point out that the gradient of @xmath2 accounts for interfacial energy effects in phase transitions , and for the influence of damage at a material point , undamaged in its neighborhood , in damage models . in this sense we can say that the term @xmath73 models nonlocality of the phase transition or the damage process , i.e. the feature that a particular point is influenced by its surrounding . in damage , this leads to possible hardening or softening effects ( cf . also @xcite for further comments on this topic ) . gradient regularizations of @xmath67-laplacian type are often adopted in the mathematical papers on damage ( see for example @xcite ) , and in the modeling literature as well ( cf . , e.g. , @xcite ) . in a different context , a @xmath67-laplacian elliptic regularization with @xmath74 has also been exploited in @xcite , in order to study a diffuse interface model for the flow of two viscous incompressible newtonian fluids in a bounded domain . in the following , we will also scrutinize another kind of elliptic regularization in , given by the _ @xmath75-laplacian operator on the sobolev - slobodeckij space @xmath76 , hereafter denoted by @xmath77 ( cf . later on for its precise definition ) . recently , fractional laplacian operators have been widely investigated ( cf . , e.g. , @xcite and the references therein ) , and used in connection with real - world applications , such as thin obstacle problems , finance , material sciences , but also phase transition and damage phenomena ( cf . , e.g. @xcite and @xcite ) . for analytical reasons , we will have to assume @xmath78 , which ensures the ( compact ) embedding @xmath79 , in the same way as @xmath80 for @xmath74 . this property will play a crucial role in the _ degenerate _ limit to complete damage , as it did in @xcite within the rate - independent context , cf . remark [ rmk : explanation ] for more details . as for the potential @xmath81 , we suppose that @xmath82 with @xmath83 $ ] convex and possibly nonsmooth , and @xmath84 smooth and possibly nonconvex . we will take the domain of @xmath85 to be contained in @xmath86 $ ] . note that , in this way , the values outside @xmath86 $ ] ( which indeed are not physically meaningful for the order parameter @xmath2 , denoting a phase or damage proportion ) are excluded . typical examples of functionals which we can include in our analysis are the logarithmic potential @xmath87 where @xmath88 and @xmath89 are positive constants , as well as the sum of the indicator function @xmath90}$ ] with a _ nonconvex _ @xmath91 . in such a case , in the derivative @xmath92 needs to be understood as the subdifferential @xmath93 in the sense of convex analysis . the term @xmath94 in accounts for the thermal expansion of the system , with the thermal expansion coefficient @xmath95 assumed to be constant ( cf . , e.g. , @xcite ) . indeed , one could consider more general functions @xmath95 depending , e.g. , on the phase parameter @xmath2 and vanishing when @xmath3 . this would be meaningful especially in damage models , where the terms associated with deformations should disappear once the material is completely damaged ( cf . , e.g. , @xcite ) . we will discuss the mathematical difficulties attached to this extension in section [ mathdiff ] . for the pseudo - potential @xmath31 , following ( * ? ? ? * sec . 4 , 6 ) we take @xmath96}(\chi_t ) + a(\chi)\frac{{\varepsilon}({\mathbf{u}}_{t})\mathrm{r}_v{\varepsilon}({\mathbf{u}}_t)}{2}\,,\end{aligned}\ ] ] where @xmath97 is a symmetric and positive definite viscosity matrix , premultiplied by a function @xmath8 of @xmath2 . in particular , for phase change models , one can take for example @xmath98 . the underlying physical interpretation is that the viscosity term @xmath99 vanishes when we are in the non - viscous phase , i.e. in the solid phase @xmath3 . also in damage models the choice @xmath100 is considered , cf . e.g. @xcite . the heat conductivity function @xmath101 will be assumed continuous ; for the analysis of system , we will need to impose some compatibility conditions on the growth of @xmath102 and of the heat capacity function @xmath103 in , see hypothesis ( ii ) in section [ ss : assumptions ] . furthermore , in @xmath104 is a non - negative coefficient : for @xmath105 we encompass in our model the _ unidirectionality _ constraint @xmath106 a.e . in @xmath107 . in fact , throughout the paper we are going to use the term _ irreversible _ in connection with the case in which the process under consideration is unidirectional , which is indeed typical of damage phenomena . with straightforward computations , from and using the form of the free energy functional and of the pseudopotential of dissipation , we derive equations , neglecting the quadratic contributions in the velocities on the right - hand side in by means of the aforementioned _ small perturbation assumption _ @xcite . this is a simplification needed from the analytical point of view in order to solve the problem . indeed , in a forthcoming paper we plan to tackle the pde system , featuring in addition these quadratic terms in the temperature equation . to do so , we are going to resort to specific techniques , partially mutuated from @xcite , however confining the analysis to some particular cases . in fact , to our knowledge only few results are available on diffuse interface models in thermoviscoelasticity ( i.e. also accounting for the evolution of the displacement variables , besides the temperature and the order parameter ) : among others , we quote @xcite . in all of these papers , the small perturbation assumption is adopted . for , without it in the spatial three - dimensional case existence results seem to be out of reach , at the moment , even when the equation for displacements is neglected ( whereas the existence of solutions to the _ full _ phase change model in the unknowns @xmath0 and @xmath2 has been obtained in @xmath108 in @xcite ) . this has led to the development of suitable _ weak solvability _ notions to handle ( the usually neglected ) quadratic terms , like in @xcite ( where however @xmath1 is still taken constant ) . also in @xcite , a pde system coupling the displacement and the temperature equation ( with quadratic nonlinearities ) and a _ rate - independent _ flow rule for an internal dissipative variable @xmath2 ( such as the damage parameter ) has been analyzed . rate - independence means that the evolution equation for @xmath2 has no longer the _ gradient flow _ structure of : the term @xmath28 therein is replaced by @xmath109 , viz . in the pseudo - potential @xmath31 , instead of the quadratic contribution @xmath110 we have the @xmath11-homogeneous dissipation term @xmath111 . in the frame of the ( weak ) _ energetic formulation _ for rate - independent systems @xcite , suitably adapted to the temperature - dependent case , in @xcite existence results have been obtained . a temperature - dependent , _ full _ model for ( rate - dependent ) damage has been addressed in @xcite as well , with local - in - time existence results . the main difficulties attached to the analysis of system are : the _ elliptic degeneracy _ of the momentum equation : in particular , we allow for the positive coefficients @xmath112 and @xmath113 to tend to zero simultaneously ; the _ highly nonlinear coupling _ between the single equations , resulting in the the quadratic terms @xmath114 , @xmath115 , and @xmath116 in the heat and phase equations and , respectively ; the _ poor regularity _ of the temperature variable , which brings about difficulties in dealing with the coupling between equations and when we consider the thermal expansion terms ( i.e. we take @xmath117 ) ; the _ doubly nonlinear _ character of , due to the nonsmooth graph @xmath118 and the nonlinear operator @xmath119 ( which on the other hand has a key regularizing role ) . furthermore , if we set @xmath105 in to enforce an irreversible evolution for @xmath2 , the simultaneous presence of the terms @xmath120 and @xmath121}(\chi_t)$ ] makes it difficult to derive suitable estimates for @xmath2 , also due to the low regularity of the right - hand side of . we now partially survey how each of these problems has been handled in the recent literature . as for * 1 ) * , in @xcite we have focused on the _ phase transition _ case , in which @xmath100 and @xmath58 . we have proved the local - in time ( in the @xmath122-setting ) and the global - in - time ( in the @xmath108-setting ) well - posedness of a system in thermoviscoelasticity analogous to ( with the laplacian instead of the @xmath123laplacian in , in the case @xmath124 and @xmath125 , and for constant heat capacity and heat conductivity in ) . the main idea in @xcite to handle the possible elliptic degeneracy of is in fact to _ prevent it_. specifically , we have shown that , if the initial datum @xmath126 stays away from the values points @xmath10 and @xmath11 , so does @xmath2 during this evolution , guaranteeing that the operators in are uniformly elliptic . separation property _ is proved by exploiting a sufficient coercivity of @xmath81 at the thresholds @xmath10 and @xmath11 , which for example holds true for the logarithmic potential . in @xcite an _ isothermal _ ( irreversible ) model for damage has been considered : therein , because of the elliptic degeneracy of , the authors only prove a local - in - time existence result . for ( isothermal ) _ rate - independent _ damage models @xcite , the results change significantly : in this realm , only poor time - regularity of the solution component @xmath2 is to be expected , because the @xmath11-homogeneous dissipation contribution in @xmath28 to @xmath31 just ensures @xmath127-estimates for the function @xmath128 . that is why , one has to resort to the aforementioned notion of _ energetic solution _ @xcite , in which no time - derivatives of @xmath2 are featured . therefore , this concept is very flexible for analysis , and has allowed for handling the ( degenerate ) case of _ complete damage _ in @xcite by means of a specially devised formulation we will refer to later . concerning problem * 2 ) * , as already mentioned existence results have been obtained in @xcite for a _ full _ model of phase transitions ( in the reversible case @xmath124 and for constant @xmath1 ) , even featuring the term @xmath129 on the right - hand side of the temperature equation . therein , a suitable notion of weak solution is addressed , consisting of the phase equation , coupled with a total energy balance and a weak entropy inequality , for which existence is proved by relying on an iterative regularization procedure . this technique can not be applied to system . nonetheless , let us mention that a key assumption in @xcite is a suitable growth of the heat conductivity @xmath101 . following @xcite , here we will combine it with conditions on the heat capacity coefficient @xmath130 to handle the quadratic nonlinearities @xmath114 and @xmath115 in . due to the lack of good " a priori estimates for @xmath0 mentioned in * 3 ) * , we will not be able to encompass in our analysis the case of a non - constant thermal expansion coefficient @xmath95 , e.g. @xmath131 , which would still be interesting for damage @xcite . indeed , such a choice would lead to an additional term of the type @xmath132 in the heat equation , which we would not be able to handle without resorting to further regularizations , and possibly proving only local - in - time existence results . nonetheless , let us stress that , especially in case of phase transition phenomena , the choice of a constant @xmath95 is quite reasonable ( cf . , e.g. , @xcite ) . as for * 4 ) * , in @xcite ( dealing with cahn - hilliard systems coupled with elasticity and damage processes ; see also @xcite ) , the authors have devised a weak formulation of ( in the irreversible case @xmath133 ) which has allowed them to circumvent its triply nonlinear character . such a formulation strongly relies on the special choice @xmath134 ( which , joint with the irreversibility constraint , still ensures that @xmath2 takes values in the meaningful interval @xmath86 $ ] , provided that @xmath135 $ ] ) . it consists of a _ one - sided _ variational inequality ( i.e. with test functions having a fixed sign ) , and of an _ energy inequality _ , see later . unlike @xcite , here we shall not enforce separation of @xmath2 from the threshold values @xmath10 and @xmath11 , and accordingly we will allow for general initial configurations of @xmath2 . then , it is not to be expected that either of the coefficients @xmath112 and @xmath113 stay away from @xmath10 , which results in the elliptic degeneracy of the displacement equation . to handle it , we shall approximate system with a non - degenerating one , where we replace with @xmath136 let us note that , to rule out the degeneracy , it is sufficient to truncate away from zero only the coefficient @xmath112 of the _ viscous _ part of the elliptic operator in the momentum equation . however , for technical reasons which will become apparent in section [ sec:5 ] ( cf . [ rmk : added ] ) , when addressing the asymptotic analysis as @xmath137 to the _ degenerate _ limit , we will need to truncate the coefficient @xmath113 as well , resulting in . in the analysis of , we will distinguish the cases @xmath138 and @xmath139 : let us stress that , in the latter , there is an additional coupling between the heat and the momentum balance equations , which needs to be carefully handled and indeed requires strengthening of some of our assumptions . furthermore , to avoid overburdening the paper we will tackle the case @xmath139 only for the _ reversible _ system ( i.e. with @xmath124 ) . more specifically , in theorems [ teor1 ] , [ teor3 ] and [ teor4 ] , we will establish global - in - time existence results for the non - degenerating system ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) with @xmath125 , both in the _ reversible _ and in the _ irreversible _ cases . we will work under quite general assumptions on @xmath140 and @xmath141 , basically requiring that @xmath140 and @xmath141 are bounded from below and above by the sum of a bounded function and function behaving like a small power of @xmath0 ( cf . hypotheses ( i ) and ( ii ) in sec . [ ss : assumptions ] ) . in theorem [ teor1bis ] , we will handle the case @xmath142 and prove the existence of global solutions to ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) , under the more restrictive assumption that @xmath141 is bounded from below and above by a function behaving like @xmath143 , with @xmath144 , cf . hypothesis ( viii ) . a continuous dependence result , yielding uniqueness of solutions , for the non - degenerating _ isothermal _ reversible system , possibly with @xmath139 , will be given in theorem [ teor2 ] . finally , we will address the degenerate limit @xmath137 in theorem [ teor5 ] in a less general setting , in particular confining ourselves to the case @xmath125 . in what follows , we give more details on thms.[teor1][teor5 ] . our first main result thm . [ teor1 ] states the existence of solutions to system ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) with @xmath125 , in the _ reversible _ case @xmath124 , with the heat equation suitably reformulated by means of an _ enthalpy _ transformation ( cf . [ ss : assumptions ] ) , switching from the temperature variable @xmath0 to the enthalpy @xmath145 . already in the proof of this global - in - time existence result , a key role is played by the aforementioned @xmath67-growth assumption on the function @xmath146 with @xmath74 . in fact , it enables us to derive an estimate for @xmath2 in @xmath147 , which in turns allows for a suitable regularity estimate on the displacement variable @xmath1 , leading to a global - in - time bound on the quadratic nonlinearity @xmath148 on the right - hand side of . for further details we refer to the proof of thm . [ teor1 ] , developed by passing to the limit in a carefully designed time - discretization scheme and exploiting boccardo&gallout - type estimates on @xmath0 . relying on the stronger hyp . ( viii ) , in the case @xmath149 we will obtain enhanced estimates on ( the sequence , constructed by time discretization , approximating ) @xmath0 , cf . also remark [ rmk : afterthm2 ] later on . these bounds and the related enhanced convergences will enable us to handle the ( passage to the limit in the time discretization of the ) thermal expansion terms in and . in this way , we will conclude the proof of the existence theorem [ teor1bis ] for system ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) in the case @xmath139 . in the reversible and _ isothermal _ case , continuous dependence of the solutions on the initial and problem data is proved in thm . [ teor2 ] under a slightly more restrictive condition on the field @xmath146 , which is however satisfied in the prototypical case of the @xmath67-laplacian operator . as already mentioned , in the _ irreversible _ case @xmath105 a major difficulty in the analysis of system ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) stems from the simultaneous presence in of the multivalued operators @xmath121}(\chi_t)$ ] and @xmath150 , ( cf . ) , as well as of the @xmath67-laplacian type operator @xmath151 , which still has a key role in providing global - in - time estimates for @xmath148 . to tackle this problem , following the approach of @xcite we restrict to the yet meaningful case @xmath152 and consider a suitable weak formulation of . it consists ( cf . definition [ def - weak - sol ] later on ) of the _ one - sided _ variational inequality [ weaksol - intro ] @xmath153 and of the following energy inequality for all @xmath154 $ ] , for @xmath155 , and for almost all @xmath156 : @xmath157 in sec . [ glob - irrev ] , several comments and remarks shed light on this weak solvability notion for . in particular , proposition [ more - regu ] shows that , if @xmath2 is regular enough , and the subdifferential inclusion are equivalent . in theorem [ teor3 ] we state the existence of global - in - time solutions to the weak formulation of system ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) with @xmath105 , consisting of the ( weakly formulated ) enthalpy equation , of and of . the proof is again carried out via a time - discretization procedure , combined with yosida - regularization techniques . finally , theorem [ teor4 ] focuses on the isothermal case , i.e. with a fixed temperature profile . in this setting , we succeed in proving enhanced regularity for @xmath2 , thus solving in a stronger sense than . in the particular case @xmath158 the crucial estimate consists in testing by @xmath159 ( where for simplicity we write @xmath160 as single - valued ) . this enables us to estimate separately the terms @xmath121}(\chi_t)$ ] ( again written as single - valued ) , @xmath161 , and @xmath162 in @xmath163 , which is the key step for proving the existence of solutions to the _ pointwise _ subdifferential inclusion . uniqueness results for the _ irreversible _ system , even in the isothermal case , do not seem to be at hand , due to the triply nonlinear character of equation , cf . also remark [ uni - irrev ] ahead . nonetheless , both in the reversible and in the irreversible case , in thms . [ teor1 ] , [ teor1bis ] and [ teor3 ] we will prove positivity of the temperature @xmath0 . in fact , under suitable conditions on the initial temperature , for @xmath105 we will also obtain a strictly positive lower bound for @xmath0 . for the analysis of the degenerate limit @xmath137 of ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) , we have carefully adapted to the present setting techniques from @xcite and @xcite . these two papers deal with _ complete damage _ in the fully rate - independent case , and , respectively , for a system featuring a rate - independent damage flow rule for @xmath2 and a displacement equation with viscosity and inertia according to kelvin - voigt rheology . in particular , we have extended the results from @xcite to the case of a _ rate - dependent _ equation for @xmath2 , also coupled with the temperature equation . following @xcite , the key observation is that , for any family @xmath164 of solutions to ( [ eq0 ] , [ eqi - delta ] , [ eqii ] ) ( where @xmath165 denotes the _ enthalpy _ ) , it is possible to deduce for the quantities @xmath166 and @xmath167 the estimates @xmath168 for a positive constant independent of @xmath169 . therefore , there exist @xmath170 and @xmath171 such that , up to a subsequence @xmath172 in @xmath173 and @xmath174 in @xmath175 as @xmath176 . according the terminology of @xcite , we refer to @xmath170 and @xmath171 , respectively , as the viscous and elastic _ quasi - stresses_. in theorem [ teor5 ] we will focus on the degenerate limit @xmath176 , confining the discussion to the case where @xmath125 and @xmath105 ( viz . the map @xmath177 is nonincreasing for all @xmath178 ) . we refer to remark [ rmk : explanation ] for a thorough justification of these choices . passing to the limit as @xmath137 in and exploiting the above convergences for @xmath179 and @xmath180 we will prove that there exist a triple @xmath181 solving the _ generalized _ momentum balance [ degen - intro ] @xmath182 such that the quasi - stresses fulfill @xmath183 in addition to , the notion of weak solution to system arising in the limit @xmath137 consists of the ( weak formulation of the ) enthalpy equation , of the _ one - sided _ variational inequality @xmath184 and of a _ generalized _ total energy inequality , featuring the quasi - stresses @xmath171 and @xmath170 . while referring to remark [ rmk : comparison - with - hk ] for more comments in this direction , we may observe here that is in fact the integrated version in terms of _ quasi - stresses _ of the variational inequality . [ [ plan - of - the - paper . ] ] plan of the paper . + + + + + + + + + + + + + + + + + + in the next section [ s : main ] we introduce the variational formulation for the initial boundary value problem associated to the pde system , as well as our main assumptions . then , we state theorems [ teor1][teor4 ] on the existence / uniqueness of solutions for the reversible and the irreversible _ non - degenerating _ systems ( i.e. @xmath185 ) . the existence thms . [ teor1 ] , [ teor1bis ] , [ teor3 ] , and [ teor4 ] rely on the time - discretization procedure of section [ s : time - discrete ] ; their proof is carried out by passing to the limit with the time discretization in sections [ ss:4.1 ] , [ sec:4.2new ] , [ ss:5.1 ] , and [ ss:5.2 ] . the continuous dependence thm . [ teor2 ] is proved in section [ cd ] . finally , section [ sec:5 ] is devoted to the passage to the degenerate limit @xmath137 . the following table summarizes our results [ cols="<,<,<",options="header " , ] [ not:2.1 ] throughout the paper , given a banach space @xmath186 we shall denote by @xmath187 its norm , and use the symbol @xmath188 for the duality pairing between @xmath189 and @xmath186 . hereafter , we shall suppose that @xmath190 we will identify both @xmath191 and @xmath192 with their dual spaces , and denote by @xmath193 the scalar product in @xmath194 , by @xmath195 both the scalar product in @xmath196 , and in @xmath192 , and by @xmath197 and @xmath198 the spaces @xmath199 for @xmath200 we will use the notation @xmath201 we standardly denote by @xmath202 and , for any @xmath203 , by @xmath204 its mean value . given a ( separable ) banach space @xmath186 , we will denote by @xmath205;x)$ ] ( by @xmath206;x)$ ] , respectively ) , the space of functions from @xmath207 $ ] with values in @xmath208 that are defined at every @xmath209 $ ] and have bounded variation on @xmath207 $ ] ( and are _ weakly _ continuous on @xmath207 $ ] , resp . ) finally , throughout the paper we shall denote by the symbols @xmath210 various positive constants depending only on known quantities and by @xmath211 ( respectively @xmath212 ) or ( whenever it turns out to be more convenient ) @xmath213 ( respectively @xmath214 ) the first ( respectively ) second partial derivatives with respect to time of a function @xmath215 . * preliminaries of mathematical elasticity . * in what follows , we shall assume the material to be homogeneous and isotropic , so that the elasticity matrix @xmath57 in equation may be represented by @xmath216 where @xmath217 are the so - called lam constants and @xmath218 is the identity matrix . in order to state the variational formulation of the initial - boundary value problem for , we need to introduce the bilinear forms related to the @xmath2-dependent elliptic operators appearing in . hence , given a _ non - negative _ function @xmath219 , let us consider the continuous bilinear symmetric forms @xmath220 defined for all @xmath221 by @xmath222 where @xmath223 is the viscosity matrix @xmath97 . now , by korn s inequality ( see eg ( * ? ? ? * thm . 6.3 - 3 ) ) , the forms @xmath224 and @xmath225 are @xmath197-elliptic and continuous . namely , there exist constants @xmath226 , only depending on @xmath227 and @xmath228 , such that such that for all @xmath229 @xmath230 we shall denote by @xmath231 and @xmath232 the linear operators associated with @xmath224 and @xmath225 , respectively , namely @xmath233 it can be checked via an approximation argument that the following regularity results hold : @xmath234 in fact , the calculations we will develop extend to the case of an anisotropic and inhomogeneous material , for which the elasticity and viscosity matrices @xmath57 and @xmath97 are of the form @xmath235 and @xmath236 , with functions @xmath237 satisfying the classical symmetry and ellipticity conditions ( with the usual summation convention ) @xmath238 clearly , ensures , whereas not only does imply , but the @xmath66-regularity also allows us to perform the third a priori estimate of section [ ss:3.2 ] rigorously . in what follows we will use the following elliptic regularity result ( see e.g. ( * ? ? ? 6.3-.6 , p. 296 ) , cf . also @xcite ) : @xmath239 finally , in the weak formulation of the momentum equation , besides @xmath240 and @xmath241 we will also make use of the operator @xmath242 [ [ useful - inequalities . ] ] useful inequalities . + + + + + + + + + + + + + + + + + + + + we recall the celebrated gagliardo - nirenberg inequality ( cf . @xcite ) in a particular case : for all @xmath243,$ ] and for all @xmath244 such that @xmath245 , there holds @xmath246 the positive constant @xmath247 depending only on @xmath248 . combining the compact embedding @xmath249 ( where for @xmath250 we mean that @xmath251 for all @xmath252 ) , with ( * ? ? ? 16.4 , p. 102 ) , we have @xmath253 we will also make use of the compact sobolev embedding @xmath254 we conclude with the following poincar - type inequality ( cf . * lemma 2.2 ) ) , with @xmath255 the mean value of @xmath145 : @xmath256 we enlist below our basic assumptions on the functions @xmath140 , @xmath141 , @xmath81 , @xmath257 in system . * hypothesis ( i ) . * we suppose that @xmath258 * hypothesis ( ii ) . * we assume that @xmath259 * hypothesis ( iii ) . * we require @xmath260.\end{aligned}\ ] ] * hypothesis ( iv ) . * we suppose that the potential @xmath81 in is given by @xmath261 where @xmath262\ , , \quad \widehat\beta : \text{\rm dom}(\widehat{\beta } ) \to \r \ , \text { is proper , l.s.c . , convex ; } \\ \label{data - gamma } & \widehat{\gamma } \in { \rm c}^{2}(\r).\end{aligned}\ ] ] hereafter , we shall denote by @xmath263 the subdifferential of @xmath264 , and set @xmath265 . * hypothesis ( v ) . * we require that there exists @xmath266 and , setting @xmath267 , the following coercivity and growth conditions hold true : @xmath268 * a generalization of the @xmath67-laplace operator . * we now consider the realization in @xmath196 of @xmath146 , i.e. @xmath269,\quad \phi(\chi):=\begin{cases } { \int_\omega}\phi(x,\nabla\chi(x ) ) \dd x & \text{if $ \phi(\cdot,\nabla\chi(\cdot ) ) \in l^1 ( \omega)$ , } \\ + \infty & \text{otherwise . } \end{cases}\ ] ] relying on ( * ? ? ? 2.5 , p. 22 ) , it is possible to prove that @xmath270 is convex and lower semicontinuous on @xmath196 , with domain @xmath271 ( due to ) ; its subdifferential @xmath272 is a maximal monotone operator . in we will take the elliptic operator @xmath273 clearly , @xmath274 is a generalization of the @xmath67-laplace operator . note that @xmath275 and ( cf . 2.4 ) ) @xmath276 in order to obtain further regularity and uniqueness results for system , we will have to assume that either of the following additional hypotheses holds true . * hypothesis ( vi ) . * we require that the function @xmath146 fulfills the @xmath67-coercivity condition @xmath277 , viz . @xmath278 [ rmk : reg - sav ] it was proved in ( * ? ? ? * thm . 2 , rmk . 3.5 ) that , if in addition to hypothesis ( v ) the function @xmath146 fulfills hypothesis ( vi ) , then @xmath279 * hypothesis ( vii ) . * we assume that @xmath146 complies with and with the following _ convexity _ requirement @xmath280 assumptions and guarantee the validity of the following inequality ( cf . @xcite for a proof ) @xmath281 . [ ex : operators ] the two @xmath67-laplacian operators @xmath282 are clearly of the form , and comply with and i , 4(iii ) ) ) . observe that is fulfilled by the @xmath67-laplacian operator @xmath283 , whereas for the _ degenerate _ operator @xmath284 , inequality holds with @xmath285 . [ [ a - nonlocal - alternative - to - the - p - laplacian - operator . ] ] a _ nonlocal _ alternative to the @xmath67-laplacian operator . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as done in @xcite , we could replace the @xmath67-laplacian - type operator @xmath274 in with a _ linear _ operator , with domain compactly embedded in @xmath286 . more precisely , as in @xcite we could choose @xmath287 in , @xmath288 denotes the sobolev - slobodeckij space @xmath76 , endowed with the inner product @xmath289 where @xmath290 indeed , since @xmath291 , we may suppose that @xmath292 . then , we denote by @xmath293 the associated operator , viz . @xmath294 observe that , for @xmath78 we have @xmath295 . [ [ enthalpy - transformation . ] ] enthalpy transformation . + + + + + + + + + + + + + + + + + + + + + + + + we now reformulate pde system in terms of the _ enthalpy _ @xmath296 , related to the absolute temperature @xmath0 via @xmath297 it follows from that the function @xmath298 is strictly increasing on @xmath299 . thus , we are entitled to define @xmath300 in terms of the enthalpy @xmath296 , the pde system rewrites as @xmath301}(\chi_t)-\operatorname{div}({\bf d}(x,\nabla\chi))+w'(\chi ) \ni - b'(\chi)\frac{|{\varepsilon({\bf u})}|^2}2 + \theta({w } ) \quad\hbox{in } \omega \times ( 0,t),\label{eqii - ental}\end{aligned}\ ] ] supplemented with the initial and boundary conditions ( where @xmath302 denotes the outward unit normal to @xmath45 ) @xmath303 the enthalpy transformation was proposed in @xcite , and further developed in @xcite , in order to deal with pde systems where a quasilinear internal energy balance analogous to is coupled with _ rate - independent _ processes . the advantage of this change of variables , is that the _ nonlinear _ term @xmath304 in is replaced by the _ linear _ contribution @xmath305 in . we will exploit this fact , when proving the existence of solutions to ( an approximation of ) system by means of a time - discretization scheme . for later use , let us observe that hyp . ( i ) implies @xmath306 a straightforward consequence of the first of is that for every @xmath307 @xmath308 moreover , hypotheses ( i ) and ( ii ) entail @xmath309 finally , in order to deal with the case @xmath117 , we will adopt the following further assumption , which we directly state in terms of the function @xmath310 instead of @xmath101 and @xmath130 , in _ replacement _ of hypothesis ( ii ) . * hypothesis ( viii ) . * we require that the function @xmath310 defined in ( where @xmath140 fulfills hyp . ( i ) and @xmath311 is a continuous function ) satisfies @xmath312 indeed , we could slightly weaken by prescribing that @xmath310 is bounded from below and above by two functions behaving like @xmath313 , and we have restricted to for simplicity only . let us stress that , if holds , @xmath310 is no longer bounded from above . [ [ problem - and - cauchy - data . ] ] problem and cauchy data . + + + + + + + + + + + + + + + + + + + + + + + + we suppose that bulk force @xmath17 and the heat source @xmath48 fulfill @xmath314 and that the initial data comply with @xmath315 [ [ variational - formulation - of - the - non - degenerating - system . ] ] variational formulation of the non - degenerating system . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we now consider the _ non - degenerate _ version of system : as already mentioned in the introduction , to rule out the elliptic degeneracy of the momentum equation , it is indeed sufficient to truncate away from zero only the coefficient @xmath112 , cf . below . in specifying the variational formulation of the initial - boundary value problem for the _ non - degenerate _ system , due to the @xmath10-homogeneity of the operator @xmath316}$ ] we will just distinguish the two cases @xmath124 and @xmath133 . we mention in advance that the @xmath317-regularity for @xmath165 derives from boccardo&gallout - type estimates @xcite on the enthalpy equation , combined with the gagliardo - nirenberg inequality . we refer to the forthcoming sec . [ ss:3.2 ] and to @xcite for all details . [ prob ] given @xmath185 , @xmath318 , find functions @xmath319;w^{1,r'}(\omega)^*)\quad \text{for every } 1 \leq r < \frac{d+2}{d+1 } , \\ & \label{reg - u } { \mathbf{u}}\in h^1(0,t;h_0^{2}(\omega;\r^d ) ) \cap w^{1,\infty } ( 0,t;{h_{0}^1(\omega;\r^d ) } ) \cap h^2 ( 0,t;l^2(\omega;\r^d ) ) , \\ & \label{reg - chi } \chi \in l^\infty ( 0,t;w^{1,p } ( \omega ) ) \cap h^1 ( 0,t;l^2 ( \omega)),\end{aligned}\ ] ] fulfilling the initial conditions @xmath320 the equations @xmath321 ; w^{1,r'}(\omega ) ) \cap w^{1 , r'}(0,t ; l^{r'}(\omega ) ) \text { and for all } t\in ( 0,t ] , \end{aligned } \\ \label{eq1d } & { \mathbf{u}}_{tt}+{\mathcal{v}\left({(a(\chi)+\delta)}{{\mathbf{u}}_t}\right)}+{\mathcal{e}\left({b(\chi)}{{\mathbf{u}}}\right ) } + { \mathcal{c}_\rho } ( \theta(w ) ) = \mathbf{f } \quad \text{in $ h^{-1}(\omega;\r^d)$ \quad a.e.\ in } ( 0,t),\end{aligned}\ ] ] and the subdifferential inclusion @xmath322}(\chi_t ) + { \mathcal{b}}(\chi)+\beta(\chi)+ \gamma(\chi ) \ni - b'(\chi)\frac{\varepsilon({\mathbf{u } } ) \mathrm{r}_e\varepsilon({\mathbf{u}})}{2}+\theta({w } ) \text { in } w^{1 , p}(\omega)^ * \text { a.e.\ in $ ( 0,t)$}. \end{aligned}\end{aligned}\ ] ] [ w - radon ] since @xmath323;w^{1,r'}(\omega)^*)$ ] , for all @xmath209 $ ] one has @xmath324 . combining this with the fact that @xmath325 , we have that @xmath326 is a radon measure on @xmath32 for all @xmath209 $ ] , which justifies the notation in the first integral term on the left - hand side of . moreover , let us note that the @xmath327-regularity w.r.t . time of the absolute temperature , which is mainly due to the presence of quadratic nonlinearities @xmath328 and @xmath329 in , is quite natural for this kind of problems ( cf . , e.g. @xcite , @xcite , and @xcite ) . [ rem - other - b.c . ] in what follows , we will refer to system with @xmath124 ( with @xmath133 , respectively ) , as the ( non - degenerating ) _ reversible full system _ ( _ irreversible full system _ , resp ) . in both cases @xmath124 and @xmath133 , we will call _ isothermal _ the ( non - degenerating ) system , where @xmath330 in is replaced by a given temperature profile @xmath331 . our first main result states the existence of a solution @xmath332 to the reversible full system under hypotheses ( i)(v ) ; under the further hypothesis ( vi ) we are able to obtain some enhanced regularity for @xmath2 . its proof will be developed in section [ sec : proof2 ] by passing to the limit in the time - discretization scheme set up in sec.[s : time - discrete ] . [ teor1 ] let @xmath124 , @xmath125 , and assume hypotheses ( i)(v ) and conditions on the data @xmath333 . then , problem [ prob ] admits a solution @xmath334 , such that there exists @xmath335 furthermore , @xmath334 satisfies the _ total energy equality _ @xmath336 if , in addition , @xmath146 complies with hypothesis ( vi ) , then there holds @xmath337 suppose that @xmath338 then , @xmath339 a.e . in @xmath340 , hence @xmath341 we are going to prove the energy equality by testing by @xmath342 , by @xmath343 , by @xmath28 , adding the resulting relations , integrating in time , and developing the calculations at the end of the proof of thm . [ teor1 ] in sec . [ ss:4.1 ] . we now turn to the case when the thermal expansion coefficient @xmath139 . as previously mentioned , to prove existence of solutions we need to replace hyp . ( ii ) with hyp . ( viii ) , which has a key role in deriving the enhanced regularity for @xmath165 , cf . remark [ rmk : afterthm2 ] later on . [ teor1bis ] let @xmath124 , @xmath344 , and assume hypotheses ( i ) and ( iii)(v ) and conditions on the data @xmath333 . suppose moreover that hypothesis ( viii ) is satisfied ( in place of hypothesis ( ii ) ) , and that @xmath345 then , problem [ prob ] admits a solution @xmath334 fulfilling - , such that @xmath145 has the further regularity @xmath346 ; w^{2,{\mathfrak{s}}}(\omega ) ' ) \quad \text{with } { \mathfrak{s}}=\frac{6q+6}{4q+5 } , \end{aligned}\ ] ] and the weak formulation of the enthalpy equation holds in the form @xmath347 for all test functions @xmath348 ; w^{2,{\mathfrak{s}}}(\omega))\cap h^1(0,t ; l^{6/5}(\omega ) ) $ ] , where @xmath349 is a primitive of @xmath310 . moreover , @xmath334 complies with the _ total energy equality _ . if , in addition , @xmath146 complies with hypothesis ( vi ) , then the further regularity result holds true . if in addition @xmath48 complies with , then holds . [ rmk : afterthm2 ] let us justify the additional regularity for @xmath165 , by developing on a purely _ formal _ level , enhanced estimates on the enthalpy equation , based on the stronger hypothesis ( viii ) . indeed , we ( formally ) choose @xmath350 as a test function for : re - integrating by parts in time and exploiting we obtain for any @xmath351 : @xmath352 now , we observe that @xmath353 and that , due to the poincar inequality and to the fact that @xmath354 , there holds @xmath355 therefore , taking into account the continuous embedding @xmath356 , for the l.h.s . of we have the lower bound @xmath357 clearly , relying on we can absorb the second term on the r.h.s . of into its left - hand side . on the other hand , using the fact that @xmath358 and taking into account the growth of @xmath359 , we can estimate the last summand on the r.h.s . by @xmath360 where the first inequality follows from the fact that @xmath361 thanks to and , and @xmath362 is chosen sufficiently small , in such a way as to absorb @xmath363 into the r.h.s . of . plugging and into we immediately deduce an estimate for @xmath165 in the space @xmath364 . observe that , as a consequence of @xmath365 , we have @xmath366 , with @xmath367 . therefore , @xmath368 with @xmath369 the conjugate exponent of @xmath370 . a comparison in entails an estimate for @xmath371 , and we conclude . in fact , the @xmath327-estimate for @xmath165 ( and accordingly , the regularity required of the test functions ) could be slightly improved by resorting to refined interpolation arguments : however , to avoid overburdening this exposition we choose not to detail this point . to prove thm . [ teor1bis ] , we will need to combine the time - discretization procedure for system , with a truncation of the function @xmath310 in the elliptic operator of , cf . problem [ prob : rhoneq0 ] later on . hence , in order to make the estimates developed in rmk . [ rmk : afterthm2 ] rigorous , we will have to pass to the limit in two phases , first with the time - step , and then with the truncation parameter , cf . the discussion in sec . [ sec:4.2new ] . for the _ isothermal _ reversible system , in both cases @xmath138 and @xmath139 , we obtain a continuous dependence result , in particular yielding uniqueness of solutions , under the additional convexity property for @xmath146 in hypothesis ( vii ) . indeed , the latter ensures the monotonicity inequality for @xmath257 , which is crucial for the continuous dependence estimate . we also need to restrict to the case in which @xmath8 is constant . [ teor2 ] let @xmath124 , @xmath372 . assume that hypotheses ( iii)(v ) and ( vii ) are satisfied , and , in addition , that @xmath373 let @xmath374 , @xmath375 , be two sets of data complying with and , and , accordingly , let @xmath376 , @xmath375 , be the associated solutions on some @xmath207 $ ] with fixed temperature profiles @xmath377 . set @xmath378 then there exists a positive constant @xmath379 , depending on @xmath380 , @xmath169 , @xmath381 , and @xmath382 , such that @xmath383 in particular , the isothermal reversible system admits a unique solution @xmath384 . the proof is postponed to section [ cd ] . [ aconst - slapl ] let us notice that , if we consider the @xmath75-laplacian instead of the @xmath67-laplacian , the continuous dependence result stated in theorem [ teor2 ] still holds true also without assumption . in this case , for any two solutions @xmath385 and @xmath386 of the isothermal reversible system , yields an estimate on @xmath387 . for further details , we refer to remark [ slaplcd ] at the end of sec . [ cd ] . * heuristics for weak solutions . * as mentioned in the introduction , the major problem in dealing with the subdifferential inclusion in the case @xmath133 is the simultaneous presence of the three nonlinear , maximal monotone operators @xmath316}$ ] , @xmath274 , and @xmath160 , which need to be properly identified when passing to the limit in the time - discretization scheme we are going to set up in section [ s : time - discrete ] . we now discuss the attached difficulties on a formal level , treating @xmath160 and @xmath316}$ ] as single - valued . it would be possible to handle @xmath388 by exploiting the strong - weak closedness ( in the sense of graphs ) of the ( induced operator ) @xmath389 . nonetheless , a @xmath390-estimate of the term @xmath162 in can not be obtained without estimating as well @xmath391 ( and hence @xmath316}(\chi_t)$ ] by comparison ) , in @xmath390 . to our knowledge , this can be proved by testing by @xmath392 ( cf . also @xcite ) . the related calculations ( which we will develop in sec.[s : time - discrete ] , on the time - discrete level , for the _ isothermal _ irreversible system ) would involve an integration by parts of the terms on the right - hand side of . thus , they would rely on an estimate in @xmath393 of the term @xmath394 . however , presently this enhanced bound for @xmath395 does not seem to be at hand due to the poor time - regularity of @xmath145 , cf . . that is why , for the temperature - dependent irreversible system we are only able to obtain the existence of solutions @xmath396 to a suitable _ weak formulation _ of , mutuated from @xcite , where we also restrict to the particular case in which @xmath397 in the present irreversible context it is sufficient to choose @xmath264 as in to enforce the constraint @xmath398 $ ] a.e . in @xmath399 indeed , starting from an initial datum @xmath400 a.e . in @xmath32 we will obtain by irreversibility that @xmath401 a.e . in @xmath32 , for almost all @xmath351 . the underlying motivation for the weak formulation of we will consider is that , due to the @xmath11-homogeneity of @xmath402}$ ] , it is not difficult to check that is equivalent to the system [ ineq - system ] @xmath403 with @xmath404 a.e . in @xmath405 ( and @xmath406 denoting the duality pairing between @xmath407 and @xmath408 , cf . notation [ not:2.1 ] ) . in order to see this , it is sufficient to subtract from , and use the definition of @xmath409 . however , for reasons analogous to those mentioned in the above lines , the proof of is at the moment an open problem . therefore , following @xcite , in the forthcoming definition [ def - weak - sol ] we weakly formulate by means of , ( an integrated version of ) , and the _ energy inequality _ below , in place of . [ def - weak - sol ] let @xmath133 . we call a triple @xmath410 as in a _ weak solution _ to problem [ prob ] if , besides fulfilling the weak enthalpy and momentum equations , it satisfies @xmath411 for almost all @xmath412 , as well as @xmath413 with @xmath414 in the following sense : @xmath415 and the energy inequality for all @xmath154 $ ] , for @xmath155 , and for almost all @xmath156 : @xmath416 the following result sheds light on the properties of this solution concept . first of all , it states the _ total energy inequality _ : from , we will deduce in sec . [ sec:5 ] suitable estimates independent of @xmath169 , which will allow us to pass to the limit in problem [ prob ] as @xmath417 for @xmath133 . furthermore , the second part of proposition [ more - regu ] ( whose proof closely follows the argument for ( * ? ? ? 4.1 ) ) shows that , if @xmath2 is regular enough , then and are equivalent to . [ more - regu ] let @xmath133 . then , any weak solution @xmath418 in the sense of def . [ def - weak - sol ] fulfills the _ total energy inequality _ for all @xmath154 $ ] , for @xmath155 , and for almost all @xmath419 @xmath420 assume now hypotheses ( iii)(v ) . let @xmath418 be as in , and suppose in addition that @xmath421 such that @xmath422 comply with and . then , @xmath423}(\chi_t ( x , t ) ) \ { \text{for a.a.}}\ , ( x , t)\in \omega \times ( 0,t ) \text { s.t . } \\ \chi_t + \zeta+ { \mathcal{b}}(\chi)+\xi+ \gamma(\chi ) = -b'(\chi)\frac{\varepsilon({\mathbf{u } } ) \mathrm{r}_e\varepsilon({\mathbf{u}})}{2}+\theta(w ) \qquad \text{a.e.~in } \ \omega \times ( 0,t ) . \end{gathered}\ ] ] in order to prove it is sufficient choose @xmath424 in , test by @xmath343 , integrate in time , perform the calculations in the proof of thm . [ teor1 ] , and add the resulting equalities with the energy inequality . the second part of the statement can be proved considering the energy functional @xmath425 , \qquad \mathscr{e}(\chi):= \phi(\chi)+\int_\omega w(\chi)\dd x.\ ] ] it follows from hypotheses ( iv ) and ( v ) , as well as the chain rule of ( * ? ? ? * lemma 3.3 ) that , if @xmath2 complies with and , then the map @xmath426 is absolutely continuous on @xmath427 and fulfills @xmath428 therefore , differentiating in time and using we conclude that @xmath422 comply with , where the duality pairing @xmath429 is replaced by the scalar product in @xmath430 . likewise , yields . again on account of and , it is not difficult to infer from that @xmath431}(\chi_t)$ ] a.e . in @xmath107 as an element of @xmath432 , and follows . as we have already pointed out in the proof of proposition [ more - regu ] , to prove that _ any _ weak solution @xmath418 in the sense of def . [ def - weak - sol ] fulfills in the irreversible case ( i.e. with @xmath133 ) the _ total energy inequality _ it is sufficient to choose @xmath424 in , test by @xmath343 , integrate in time and add the resulting equalities to . instead , the proof of the _ total energy equality _ relies on the fact that , for @xmath124 we are able to obtain the stronger , _ pointwise _ form of the subdifferential inclusion . we can now state our existence result in the case @xmath133 , with @xmath125 . its proof will be developed in sec . [ ss:5.1 ] by passing to the limit in the time - discretization scheme devised in sec . [ s : time - discrete ] . we mention in advance that , in this irreversible setting , in addition the basic assumptions of hypotheses ( i)(v ) , we also have to require the @xmath67-coercivity condition . it has a crucial role in proving _ strong _ convergence of the approximate solutions to @xmath2 in @xmath433 , which enables us to obtain . let us also highlight that , exploiting the additional feature of irreversibility , we prove a slightly more refined ( in comparison with ) positivity result for the temperature @xmath0 , cf . below . [ teor3 ] let @xmath133 , @xmath125 , and assume hypotheses ( i)(v ) with @xmath152 as in , and conditions on the data @xmath434 @xmath48 , @xmath435 @xmath436 @xmath437 @xmath126 . suppose moreover that @xmath146 complies with . then , problem [ prob ] admits a weak solution @xmath418 ( cf . [ def - weak - sol ] ) . suppose in addition that @xmath438 then @xmath439 if @xmath274 is given by the @xmath75-laplacian operator @xmath77 , the definition [ def - weak - sol ] of weak solution to the irreversible full system is obviously modified : in the term @xmath440 is replaced by @xmath441 with @xmath442 , whereas @xmath443 in now reads @xmath444 . our existence result theorem [ teor3 ] extends to such a case , cf . also the forthcoming remark [ remk : added - revision ] . in section [ sec:5 ] , we will focus on the irreversible full system with @xmath445 , and perform an asymptotic analysis of weak solutions ( in the sense of definition [ def - weak - sol ] ) in the _ degenerate _ limit @xmath137 . [ irrev - rho ] replacing hyp . ( ii ) with ( viii ) and carefully tailoring the estimates and techniques for the proof of thm . [ teor1bis ] ( cf . [ rmk : afterthm2 ] ) to the irreversible case , we could indeed prove the existence of _ weak _ ( in the sense of def . [ def - weak - sol ] ) solutions also for @xmath446 and @xmath139 . however we expect that , in the latter setting , only the weaker positivity result can be proved . indeed , the estimate yielding the lower bound ( cf.step @xmath447 in the proof of lemma [ lemma : ex - discr ] ) , can not be performed on the enthalpy equation due to the presence of term @xmath448 . we finally turn to the irreversible _ isothermal _ case , and improve the existence result of theorem [ teor3 ] . [ teor4 ] let @xmath133 . in addition to hypotheses ( iii)(v ) , assume that @xmath449,\ ] ] and suppose that the data @xmath450 comply with conditions and . suppose in addition @xmath146 fulfills and , that @xmath451 and consider a fixed temperature profile @xmath452 then , there exists a quadruple @xmath453 , fulfilling , @xmath454 and @xmath455}(\chi_t ) $ ] a.e . in @xmath107 , as well as @xmath456 satisfying equations and , with @xmath395 replaced by @xmath331 . [ uni - irrev ] uniqueness of solutions for the _ irreversible _ system , even in the isothermal case , is still an open problem . this is mainly due to the triply nonlinear character of ( cf.also @xcite for non - uniqueness examples for a general doubly nonlinear equation ) . [ [ a - more - general - dissipation - potential . ] ] a more general dissipation potential . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as observed in remark [ more - general - alpha ] later on , in thm . [ teor4 ] we could consider a more general dissipation potential in . indeed , in place of subdifferential operator @xmath457}$ ] , we could allow for a general cyclical monotone operator @xmath458$. } \end{gathered}\ ] ] first , in section [ ss:3.1 ] we will approximate problem [ prob ] via time discretization . in fact , in the reversible case @xmath124 with @xmath459 , we will set up an _ implicit _ scheme ( cf . problems [ probk - rev ] and [ prob : rhoneq0 ] ) , whereas for the irreversible system @xmath133 with @xmath125 , we will employ the _ semi - implicit _ scheme of problem [ probk - irr ] . moreover , we will tackle separately the discretization of the isothermal irreversible system in problem [ probk - irr - iso ] . we refer to remarks [ rmk : comparison-1 ] and [ rem : irrev - discr ] for a thorough comparison between the various time - discretization procedures , and more comments . second , in sec . [ ss:3.2new ] we will prove existence results for problems [ probk - rev][probk - irr - iso ] . third , in sec . [ ss:3.2 ] we will perform suitable a priori estimates . [ not - alpha ] in what follows , also in view of the extension mentioned at the end of sec.[glob - irrev ] ( cf . [ more - general - alpha ] ) , we will use @xmath460 and @xmath461 as place - holders for @xmath462}$ ] and @xmath121}$ ] . we consider an equidistant partition of @xmath207 $ ] , with time - step @xmath463 and nodes @xmath464 , @xmath465 . in this framework , we approximate the data @xmath17 and @xmath48 by local means , i.e. setting for all @xmath466 @xmath467 [ probk - rev ] given @xmath468 find @xmath469 , with @xmath470 a.e . in @xmath32 , fulfilling @xmath471 where in the operator @xmath472 is defined by @xmath473 for the full reversible system with @xmath139 , we work under the stronger hypothesis ( viii ) and thus prescribe a suitable growth on the function @xmath310 , which is no longer bounded . therefore , in order to properly deal with the elliptic operator in the enthalpy equation on the time - discrete level , we need to truncate @xmath310 . we thus introduce the operator @xmath474 with @xmath475 observe that @xmath476 ( with @xmath477 from ) . accordingly , we will have to truncate the function @xmath359 , replacing it with @xmath478 [ prob : rhoneq0 ] starting from @xmath479 @xmath480 @xmath481 @xmath482 as in , find @xmath469 with @xmath470 a.e . in @xmath32 , fulfilling @xmath483 we now present the time discretization of the full irreversible system , postponing to remark [ rmk : comparison-1 ] a detailed comparison between problem [ probk - rev ] and the forthcoming problem [ probk - irr ] . let us only mention in advance that , in the irreversible case we will restrict to the particular choice @xmath484 . furthermore , in problem [ probk - irr ] instead of the time discretization of , we will consider the minimum problem , such that its euler equation is discretized . we resort to this approach in view of the passage to the limit argument as @xmath485 , mutuated from @xcite , which we will develop in the proof of thm . [ teor3 ] . finally , due to technical reasons related to the proof of the _ third a priori estimate _ in sec . [ ss:3.2 ] , we will also need to approximate the initial datum @xmath486 with a sequence @xmath487 with @xmath488 , cf . . we construct @xmath489 in such a way that , if @xmath490 complies with , then for every @xmath463 @xmath491 [ probk - irr ] starting from the data @xmath479 @xmath480 @xmath481 @xmath482 as in , with @xmath492 as in , find @xmath493 , such that for all @xmath494 there holds @xmath495 and @xmath496 fulfill @xmath497 and @xmath498 [ rmk : comparison-1 ] the main difference between systems and consists in the discretization of the coupling term @xmath330 in the temperature and in the phase parameter equations . indeed , in the reversible case @xmath330 on the l.h.s . of ( and accordingly its coupled term on the r.h.s . of , which will cancel out in the _ first a priori estimate _ of sec . [ ss:3.2 ] ) , is kept _ implicit_. only relying on this we can prove the positivity of the discrete enthalpy @xmath499 , which for system would not follow from other considerations . instead , in the time discretization we can allow for an _ explicit _ coupling term @xmath500 in and in . therein , the positivity of the discrete enthalpy will be proved by means of a suitable test of the discrete enthalpy equation , relying on the irreversibility . because of its implicit character , in lemma [ lemma : ex - discr - rev ] existence for system will be proved by resorting to fixed - point type existence results for elliptic systems featuring pseudo - monotone ( cf . e.g. ( * ? ? ? ii ) ) operators . instead , in the semi - implicit scheme of problem [ probk - irr ] , equations , , and are decoupled , hence we will proceed by tackling them separately , solving time - incremental minimization problems . such a procedure could be useful for the numerical analysis of the problem . that is why , in sec . [ ss:3.2new ] we will focus on the proof of lemma [ lemma : ex - discr ] and develop in detail the calculations for system , whereas we will only outline the argument for the existence of solutions to in lemma [ lemma : ex - discr - rev ] . in the time - discretization of the irreversible _ isothermal _ system , we approximate the given temperature profile @xmath331 ( cf . ) by local means as well , i.e. @xmath501 contrary to the temperature - dependent irreversible case , we may again address a general maximal monotone @xmath502 . however , in order to perform enhanced estimates on the discrete equation for @xmath2 ( cf . the _ seventh _ and _ eighth a priori estimates _ of sec . [ ss:3.2 ] ) , we will need to replace @xmath160 with its yosida regularization @xmath503 , namely the nondecreasing , lipschitz continuous derivative of the convex @xmath504 function @xmath505 , cf . e.g.@xcite . in problem [ probk - irr - iso ] below , we set the regularization parameter equal to the time - step , in view of passing to the limit simultaneously in the time discretization and in the yosida regularization as @xmath506 . furthermore , we will have to work with a suitable truncation of the coefficient @xmath112 in , cf . remark [ rem : irrev - discr ] below for further comments . [ probk - irr - iso ] starting from the triple of data @xmath507 defined as in and considering the discrete approximations @xmath508 of the given temperature profile @xmath331 , find @xmath509 , such that for all @xmath494 there holds @xmath510 [ rem : irrev - discr ] in problem [ probk - irr - iso ] we need to approximate @xmath160 by a lipschitz continuous function @xmath511 because , only with such a regularization can we test equation by the discrete difference @xmath512 ( cf . the following seventh a priori estimate ) . hence , we need to take the positive part of @xmath8 in because , replacing @xmath160 by its lipschitz regularization @xmath511 , we are no longer able to enforce the constraint that @xmath513 $ ] a.e . in @xmath32 . therefore , at the discrete level we loose all positivity information on the coefficient @xmath112 . the lack of the constraint @xmath514 $ ] also motivates the truncations @xmath515 in and @xmath516 in , mainly due to technical reasons ( cf . the _ first a priori estimate _ ) . first , we prove the existence of solutions to the semi - implicit schemes and . [ lemma : ex - discr ] let @xmath133 and @xmath125 . assume hypotheses ( i)(v ) , and on the data @xmath517 . then , problems [ probk - irr ] and [ probk - irr - iso ] admit at least one solution @xmath518 and @xmath519 , resp . furthermore , if holds , then any solution @xmath520 of problem [ probk - irr ] fulfills @xmath521 we treat problems [ probk - irr ] , and [ probk - irr - iso ] in a unified way , and proceed by induction on @xmath522 . thus , starting from a quadruple @xmath523 , we show that there exist functions @xmath524 and @xmath525 , resp . , solving for problem [ probk - irr ] , and for problem [ probk - irr - iso ] , resp . [ [ step-1-discrete - equation - for - chi . ] ] step @xmath11 : discrete equation for @xmath2 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the irreversible isothermal case ( i.e. for problem [ probk - irr - iso ] ) , in order to solve we start from the approximate equation @xmath526 where @xmath527 and @xmath528 is the yosida regularization of the operator @xmath461 . clearly , is the euler equation for the minimum problem @xmath529 where the function @xmath530 the latter admits a solution @xmath531 by the direct method of the calculus of variations ( also taking into account the fact that @xmath532 is bounded from below because @xmath85 is ) . we now want to pass to the limit in as @xmath533 . note that , a comparison in and the fact that @xmath534 is lipschitz continuous yield that @xmath535 . then , following ( * ? ? ? 3 ) ( to which we refer for all details ) , we multiply firstly by @xmath536 , and secondly by @xmath537 . to perform the latter estimate , we rely on the lipschitz continuity of @xmath538 and @xmath539 , as well as on the monotonicity of @xmath534 , yielding @xmath540 it follows from these tests that there exists a constant @xmath541 , depending on @xmath463 but not on @xmath527 , such that @xmath542 by comparison , @xmath543 . also in view of the regularity result , there exist @xmath544 for all @xmath545 such that , up to a subsequence , @xmath546 strongly converges in @xmath408 to @xmath547 as @xmath548 , and @xmath549 weakly converges in @xmath196 to @xmath550 as @xmath548 . therefore , @xmath551 so that @xmath552 thanks to @xcite . thus , passing to the limit as @xmath548 in for @xmath463 fixed , we conclude that the functions @xmath553 fulfill . clearly , the direct method of the calculus of variations also yields the existence of a solution to the minimum problem . [ [ step-2-discrete - equation - for - mathbfu . ] ] step @xmath554 : discrete equation for @xmath1 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + next , we solve , which can be rewritten @xmath555 in the form @xmath556 combining the fact that @xmath513 $ ] a.e . in @xmath32 with on @xmath8 and @xmath9 and , we conclude that ( the bilinear form associated with ) the operator on the left - hand side of the above equation is continuous and coercive . hence , by lax - milgram s theorem , equation admits a ( unique ) solution @xmath557 . since the right - hand side of is in @xmath558 , relying on the regularity results of , e.g. , @xcite , we conclude that in fact @xmath559 . the analysis of follows the very same lines . [ [ step-3-discrete - equation - for - w . ] ] step @xmath560 : discrete equation for @xmath165 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + finally , let us consider the functional @xmath561 @xmath562 now , @xmath563 is lower semicontinuous w.r.t . the topology of @xmath196 . furthermore , in view of and of the young inequality we have for a fixed @xmath362 @xmath564 choosing @xmath565 sufficiently small , we thus obtain that there exist two positive constants @xmath566 and @xmath567 such that @xmath568 this shows that the sublevels of @xmath563 are bounded in @xmath569 . hence , again by the direct method in the calculus of variations , we conclude that there exists @xmath570 , and @xmath499 satisfies the associated euler equation , namely . [ [ step-4-positivity . ] ] step @xmath447 : positivity . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us assume in addition that holds , and prove by induction on @xmath522 . preliminarily , we prove by induction on @xmath522 that @xmath571 clearly holds for @xmath572 thanks to . it remains to show that , if @xmath573 a.e . in @xmath32 , then @xmath574 a.e . in @xmath32 . indeed , let us test by @xmath575 . taking into account the definition of @xmath359 , we have that @xmath576 . combining this with the inequality @xmath577 and noting that @xmath578 a.e . in @xmath32 , also in view of we obtain @xmath579 yielding @xmath580 a.e . in @xmath32 , whence . now , to prove , we observe that holds for @xmath572 due to . suppose now that @xmath581 a.e . in @xmath32 : in order to prove that @xmath582 a.e . in @xmath32 , we test by @xmath583 . with analogous calculations as above we obtain @xmath584 @xmath585 where the last inequality is due to the fact that @xmath586 a.e . in @xmath32 , and that @xmath587 a.e . in @xmath32 by the previously proved and the irreversibility constraint . thus , we conclude . the existence result for problems [ probk - rev ] and [ prob : rhoneq0 ] reads : [ lemma : ex - discr - rev ] let @xmath124 . assume hypotheses ( i ) and ( iii)(v ) , and on the data @xmath517 . furthermore , if @xmath138 , assume hypothesis ( ii ) ; if @xmath139 , assume hypothesis ( viii ) and in addition that @xmath588 . then , problem [ probk - rev ] admits at least one solution @xmath589 . moreover , if @xmath590 a.e . in @xmath340 , and @xmath591 for a.a . @xmath64 , then any solution @xmath592 of problem [ probk - rev ] fulfills @xmath593 * step @xmath11 : existence of solutions . * our argument relies on existence results for elliptic systems from the theory of pseudo - monotone operators which can be found , e.g. , in ( * ? ? ? ii ) . indeed , we observe that system can be recast as @xmath594 denoting by @xmath595 the operator acting on the unknown @xmath496 and by @xmath596 the vector of the terms on the r.h.s . of the above equations , we can reformulate system in the abstract form @xmath597 in fact , mimicking for example the calculations in ( * ? ? ? * lemma 7.4 ) , it can be checked that @xmath595 is a pseudo - monotone operator ( according to ( * ? ? ? ii , def . 2.1 ) ) on @xmath598 , coercive on that space . therefore , the leray - lions type existence result of ( * ? ? ? ii , thm . 2.6 ) applies , yielding the existence of a solution @xmath496 to . * step @xmath554 : non - negativity of @xmath499 . * let us assume in addition that @xmath590 a.e.in @xmath107 and @xmath599 a.e . in @xmath32 . then @xmath586 a.e . in @xmath32 . to prove , we proceed by induction on @xmath522 and show that , if @xmath573 a.e . in @xmath32 , then @xmath574 a.e . in @xmath32 . indeed , let us test by @xmath575 . taking into account the definition of @xmath359 , we have that @xmath600 ( here we have kept @xmath459 also to encompass the case with thermal expansion , cf . below ) . combining this with the inequality and noting that @xmath578 a.e . in @xmath32 , also in view of we obtain @xmath579 yielding @xmath580 a.e . in @xmath32 , whence . under the additional hypothesis ( viii ) ( which gives ) , an analogous proof of existence of solutions can be given for problem [ prob : rhoneq0 ] , hence we omit to give the details . [ [ notation - and - auxiliary - results . ] ] notation and auxiliary results . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + hereafter , for a given banach space @xmath186 and a @xmath601-tuple @xmath602 , we shall use the short - hand notation @xmath603 we recall the well - known _ discrete by - part integration _ formula @xmath604 we consider the left - continuous and right - continuous piecewise constant , and the piecewise linear interpolants of the values @xmath605 , namely the functions @xmath606$.}\ ] ] we also introduce the piecewise linear interpolant of the values @xmath607 ( namely , the values taken by the -piecewise constant- function @xmath608 ) , viz . @xmath609$.}\ ] ] note that @xmath610 for @xmath611 $ ] . in view of , , and , it is easy to check that the piecewise constant interpolants @xmath612 , @xmath613 , @xmath614 , and @xmath615 of the values @xmath616 , @xmath617 , and @xmath618 , fulfill as @xmath619 [ converg - interp ] @xmath620 finally , we shall denote by @xmath621 and by @xmath622 the left - continuous and right - continuous piecewise constant interpolants associated with the partition , i.e. @xmath623 if @xmath624 and @xmath625 if @xmath626 . clearly , for every @xmath209 $ ] we have @xmath627 and @xmath628 as @xmath629 . propositions [ prop : aprio ] and [ prop : aprio-2 ] collect in the cases @xmath138 and @xmath139 several a priori estimates on the approximate solutions , obtained by interpolation of the discrete solutions to problems [ probk - rev ] , [ probk - irr ] , [ probk - irr - iso ] , and problem [ prob : rhoneq0 ] , respectively . [ prop : aprio ] let @xmath125 . assume hypotheses ( i)(v ) and on the data @xmath517 . then , 1 . in the case @xmath318 there exist a constant @xmath630 such that for the interpolants of the solutions to problem [ probk - rev ] and to problem [ probk - irr ] there holds : @xmath631;w^{1,r'}(\omega)^ * ) } \leq s , \\ & \label{aprio8}\sup_{\tau>0 } \|\theta({\overline{{{w}}}_{\tau}})\|_{l^{2+\epsilon}(0,t;l^{2+\epsilon}(\omega ) ) } \leq s \quad \text{for any $ 0<\epsilon<\frac{\sigma ( d+2)}{d}-2$.}\end{aligned}\ ] ] 2 . if @xmath124 in addition there exists @xmath632 such that @xmath633 moreover , if @xmath146 also fulfills hypothesis ( vi ) , then @xmath634 3 . in the isothermal case with @xmath133 , if @xmath635 ( cf . ) and @xmath146 also fulfills hypothesis ( vi ) , estimates hold . moreover , there exists @xmath636 such that for ( the interpolants of ) the solutions to problem [ probk - irr - iso ] @xmath637 the constants in , , and , also depend on the parameters @xmath638 , @xmath639 , and @xmath23 , respectively . [ prop : aprio-2 ] let @xmath124 and @xmath139 . assume hypotheses ( i ) , ( iii)(v ) , and hypothesis ( viii ) ; suppose that the data @xmath517 comply with , and in addition that @xmath640 . then , for the interpolants of the solutions to problem [ prob : rhoneq0 ] estimates hold with a constant _ independent _ of @xmath641 , whereas estimates , and ( under the additional hypothesis ( vi ) ) hold for a constant _ depending _ on @xmath641 . moreover , there exists a constant @xmath642 such that @xmath643 we will treat the proofs of propositions [ prop : aprio ] and [ prop : aprio-2 ] in a unified way , developing a series of a priori estimates . [ [ proof - of - proposition - propaprio . ] ] proof of proposition [ prop : aprio ] . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + most of the calculations below will be detailed on the discretization scheme for the full _ reversible _ system , and whenever necessary we will outline the differences in comparison with the discrete systems of problems [ probk - irr ] and [ probk - irr - iso ] . furthermore , for each estimate we will specify the values of the parameters @xmath644 and @xmath95 for which it is valid and , to make the computations more readable , we will illustrate them first on the time - continuous level , i.e. referring to system . * first a priori estimate for @xmath318 , @xmath459 : * _ we test by @xmath343 by @xmath11 , by @xmath28 , add them and integrate in time . this is the so - called _ energy estimate_. _ we test by @xmath645 . note that @xmath646 for all @xmath494 . since @xmath513 $ ] a.e . in @xmath32 , thanks to we have that @xmath647 a.e . in @xmath32 , thus by we have @xmath648 on the other hand , using that @xmath649 and taking into account , we find @xmath650 we estimate the latter term by observing that @xmath651 altogether , collecting and summing over the index @xmath494 , we conclude @xmath652 we multiply by @xmath653 . with standard convexity inequalities , we obtain @xmath654 we then test by @xmath655 and add the resulting relation to and , summing over the index @xmath494 . the terms @xmath656 and @xmath657 cancel out . furthermore , we note that @xmath658 a.e . in @xmath32 since @xmath659 and @xmath660 a.e . in @xmath32 , and exploit the lipschitz continuity of the function @xmath539 , which enables us to estimate the last term on the left - hand side of . ultimately , we obtain @xmath661 where the last inequality follows from assumptions on the data , from the young inequality , and from estimating @xmath662 and dealing with the latter term like in . therefore , applying a discrete version of the gronwall lemma ( cf . , e.g. , ( * ? ? ? 2.2.1 ) ) , we conclude estimates , as well as estimate @xmath663 which in turn implies @xmath664 we can perform this energy estimate on problem [ probk - irr ] as well : calculations can be trivially adapted to , whereas derives from choosing in the minimum problem @xmath665 as a competitor . we again conclude as well as . in the case of problem [ probk - irr - iso ] , also features the term @xmath550 , whence the additional term @xmath666 on the left - hand side of . since @xmath667 , by monotonicity the latter term in nonnegative . taking into account this , replacing @xmath85 with @xmath532 in , and observing that the coefficient of @xmath668 on the right - hand side of is bounded , we may repeat the same calculations as in the above lines . the coercivity estimate goes through because @xmath669 , which is no longer guaranteed to be positive , is replaced by @xmath670 . furthermore , since @xmath671 a.e . in @xmath32 ( due to the irreversibility constraint ) , we have that @xmath672 $ ] a.e . in @xmath32 , thus we may again obtain . * second a priori estimate for @xmath318 , @xmath125 : * _ following @xcite ( see also @xcite ) , we test by @xmath673 and integrate in time . _ we test by @xmath674 . this gives rise to the following terms on the left - hand side : @xmath675 the latter inequality due to . moreover , always on the l.h.s . we have @xmath676 ( where @xmath677 stands for the vectorial laplace operator ) . on the right - hand side , we have @xmath678 where the latter inequality follows from . we now move the integral terms @xmath679 to the right - hand side . let us fix @xmath680 such that @xmath681 ( where @xmath67 is the exponent in ) . then , @xmath682 where the first and second inequalities respectively follow from the hlder and young inequalities , with @xmath683 as in , the third one from , and the last one taking into account estimates for @xmath684 , for @xmath685 , which in particular yields that @xmath686 a.e . in @xmath107 for some @xmath687 , and from choosing @xmath688 . furthermore , taking into account that @xmath689 , one easily checks that @xmath690 where the second inequality follows from the young inequality , from @xmath691 , and from estimating the latter term as in . analogously , again using that @xmath692 and that @xmath693 a.e . in @xmath32 , we have @xmath694 collecting and summing over the index @xmath494 , we obtain @xmath695 applying the discrete gronwall lemma once again , we conclude estimate , whence . it is immediate to check that calculations can also be performed on the discrete momentum equation in problem [ probk - irr - iso ] . [ remk : added - revision ] the calculations for the _ second a priori estimate _ carry over to the case the operator @xmath274 is replaced by the @xmath75-laplacian @xmath77 , provided that @xmath696 . indeed , this ensures the continuous embedding @xmath697 for some @xmath698 , which is crucial in the above calculations , cf .. * third a priori estimate for @xmath318 , @xmath125 : * _ boccardo&gallout - type estimate on . _ as in the proof of ( * ? ? ? 4.2 ) , we test equation by @xmath699 , where @xmath700 \text { is defined by } \pi({{w}})= 1-\frac1{(1+{{w}})^{\varsigma } } \quad \text { for some } \varsigma>0.\ ] ] note that @xmath699 is well - defined , since @xmath574 a.e . in @xmath32 , and it belongs to @xmath701 , as @xmath702 is lipschitz continuous . such a test function has been first proposed in @xcite , as a simplification of the technique by boccardo&gallout @xcite . we shall denote by @xmath703 the primitive of @xmath702 such that @xmath704 ( hence @xmath705 for @xmath706 ) . summing over @xmath494 , we obtain @xmath707 where the first inequality follows from @xmath708 , the fact that @xmath709 , and the second one from the convex analysis inequality @xmath710 and from the fact that , due to assumption , we have @xmath711 taking into account that @xmath712 for almost all @xmath64 and all @xmath713 , and relying on and on , we conclude that @xmath714 now , we argue in the very same way as in ( * ? ? ? * proof of prop . 4.2 ) . combining the hlder and gagliardo - nirenberg inequalities ( cf . ) with the previously proved estimate and with , we see that ( cf . * formula ( 4.35 ) ) ) @xmath715 where the restriction on the index @xmath638 in fact derives from the application of the gagliardo - nirenberg inequality . next , for a sufficiently small @xmath716 such that @xmath23 from fulfills @xmath717 , there holds @xmath718 ( where we have omitted to indicate the dependence of the constants on @xmath639 and @xmath23 ) . the first inequality follows from , the second one from the gagliardo - nirenberg inequality with @xmath719 and @xmath720 : in fact the constraints @xmath721 in accord with formula . finally , the last inequality in is due to the young inequality , with @xmath722 depending on the constant @xmath362 to be suitably specified , under the additional condition that @xmath723 fulfills @xmath724 combining with , we immediately obtain @xmath725 hence , we choose @xmath362 in such a way as to absorb the second term on the right - hand side into the left - hand side . therefore , on account of @xmath726 , which yields via and the poincar inequality . finally , estimate ensues from and . observe that , when performing this estimate on the semi - implicit equation , we will obtain on the r.h.s . of the term @xmath727 , and we can estimate @xmath728 thanks to . * fourth a priori estimate for @xmath318 , @xmath125 : * _ comparison in . _ it follows from estimates , , , and from the regularity result , that @xmath729 thus , for @xmath125 estimate follows from a comparison in . the same argument carries over to and to . * fifth a priori estimate for @xmath318 , @xmath125 : * _ comparison in . _ in view of estimates and of , a comparison argument in yields estimate . the same for . * sixth a priori estimate for @xmath124 , @xmath459 : * _ we test by @xmath730 and integrate in time . _ we test by @xmath731 . arguing as for via convexity inequalities and referring to notation for the symbol @xmath732 , we get @xmath733 where the last inequality follows from @xmath734 a.e . in @xmath32 , and the fact that @xmath539 is lipschitz continuous on @xmath86 $ ] . summing up the above inequality for @xmath494 and taking into account a priori estimates and , we conclude @xmath735 from this bound , exploiting the monotonicity of @xmath160 and applying ( * ? ? ? 2.17 ) , we deduce . in view of , from the estimate for @xmath736 we deduce . * seventh a priori estimate for @xmath133 , @xmath635 , and in the isothermal case : * _ we test by @xmath737 . _ since @xmath635 , we have that @xmath738 on @xmath32 . we test by @xmath739 . we observe that @xmath740 and , if holds , we have in addition @xmath741 furthermore , always by monotonicity , we get @xmath742 here , in order to perform a rigorous argument we should approximate the graph @xmath461 with a lipschitz continuous function @xmath743 as we have done in lemma [ lemma : ex - discr ] . however we prefer not to do it now in order not to overburden the calculations . combining and with the inequalities @xmath744 and summing over the index @xmath745 , we get ( cf . for the notation @xmath746 , with @xmath500 replaced by @xmath747 ) @xmath748{c}\begin{array}[t]{c}\underbrace{\sum_{k=1}^{k_\tau}\tau \int_{\omega}(h_\tau^{k-1 } -\gamma({\chi_{\tau}^{k}})){\mathrm{d}_{\tau,{k}}({\mathcal{b}}(\chi ) + \beta_\tau(\chi))}\dd x\,.}\vspace*{.6em}\end{array}\vspace*{-1em}\\_{\text{\footnotesize $ \doteq i_7$}}\vspace*{-.5em}\\_{\text{\footnotesize } } \end{array}}}\end{aligned}\ ] ] clearly , the first term on the right - hand side of is bounded thanks to . applying the discrete integration by part formula , we find @xmath749 now , by the lipschitz continuity of @xmath539 on @xmath86 $ ] , we have @xmath750 . furthermore , we find @xmath751 where the second inequality also follows from the fact that @xmath752 is constant . collecting and the above inequalities , we thus infer @xmath753 where we set @xmath754 . then , estimate ensues via the discrete gronwall lemma , taking into account that @xmath755 in view of , , and . ultimately , follows from and the regularity result . * eighth a priori estimate for @xmath133 and in the isothermal case : * _ comparison in . _ from a comparison argument in , we conclude that @xmath756 is estimated in @xmath757 . then , and follow from the fact that @xmath758 . [ [ proof - of - proposition - propaprio-2 . ] ] proof of proposition [ prop : aprio-2 ] . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the a priori bounds follow from the calculations developed for the _ first a priori estimate _ , which also yields and for a constant independent of @xmath759 . the boccardo&gallout - type _ third estimate _ is replaced by the following * ninth a priori estimate for @xmath124 , @xmath344 : * _ test by w. _ we test by @xmath760 . summing over @xmath494 and recalling we obtain @xmath761 for a suitably small constant @xmath762 , where we have used that the terms @xmath763 are uniformly bounded ( by a constant depending on @xmath759 ) . hence , estimate follows from using that @xmath640 , taking into account the previously proved bounds and , and applying the discrete gronwall lemma . estimate then ensues from a comparison in , in view of the previously proved estimates . finally , relying on , we are able to perform the analogue of the _ second a priori estimate _ on the momentum equation in the case @xmath139 as well , as the following calculations show . * tenth a priori estimate for @xmath124 , @xmath344 : * _ test by @xmath673 and integrate in time . _ we test by @xmath674 . every term can be dealt with like in the _ second estimate _ , in addition we need to estimate the term @xmath764 choosing @xmath765 sufficiently small in such a way as to absorb @xmath766 into , and estimating @xmath767 via ( observe that @xmath768 is lipschitz continuous ) , we re - obtain , for a constant _ depending _ on @xmath641 . moreover , estimate ensues from a comparison in . finally , estimates and can be obtained by repeating on equation the very same calculations developed for the _ sixth estimate _ : again , we get bounds depending on the truncation parameter @xmath641 . [ rmk : est - indep - delta ] a close perusal of the proof of proposition [ prop : aprio ] , and in particular of the calculations performed in the second and fourth a priori estimates , reveals that in fact estimates hold for constants _ independent _ of @xmath185 in both cases @xmath124 and @xmath133 . this will play a key role in the proof of theorem [ teor5 ] . we conclude this section by mentioning in advance , for the reader s convenience , that the relevant estimates for the proof of thm . [ teor1 ] are the _ first , second , third , fourth , fifth _ , and _ sixth a priori estimate _ ; for the proof of thm . [ teor1bis ] are the _ first , fourth , sixth , ninth _ , and _ tenth a priori estimate _ ; for the proof of thm . [ teor3 ] are the _ first , second , third , fourth _ , and _ fifth a priori estimate _ ; for the proof of thm . [ teor4 ] are the _ first , second , fourth , seventh _ , and the _ eighth a priori estimate_. preliminarily , we rewrite equations in terms of the interpolants @xmath769 @xmath770 @xmath771 @xmath772 @xmath773 @xmath774 @xmath775 @xmath776 @xmath777 and @xmath778 , namely @xmath779 , \end{aligned } \\ & \label{eq - u - interp } \partial_t{\widehat{{\mathbf{u}}}_{\tau}}(t)+ { \mathcal{v}\left({(a({\overline{\chi}_{\tau } } ( t ) ) + \delta)}{\partial_t { { { \mathbf{u}}}_{\tau}}(t)}\right ) } + { \mathcal{e}\left({b({\overline{\chi}_{\tau } } ( t))}{{\overline{{\mathbf{u}}}_{\tau}}(t)}\right ) } = { \overline{\mathbf{f}}_{\tau}}(t ) \quad { \text{a.e . in\,}}\ , \omega , \ { \text{for a.a.}}\ , t \in ( 0,t ) , \\ & \label{eq - chi - interp } \begin{aligned } \partial_t { { \chi}_{\tau}}(t ) + { \mathcal{b}}({\overline{\chi}_{\tau}}(t ) ) + { \overline{\xi}_{\tau}}(t)+\gamma({\overline{\chi}_{\tau}}(t ) ) & = - b'({\underline{\chi}_{\tau}}(t))\frac{\eps({\underline{{\mathbf{u}}}_{\tau}}(t))\mathrm{r}_e \eps({\underline{{\mathbf{u}}}_{\tau}}(t))}2 + \theta({\overline{{{w}}}_{\tau}}(t))\\ & \qquad \qquad { \text{a.e . in\,}}\ , \omega , \ { \text{for a.a.}}\ , t \in ( 0,t ) , \end{aligned}\end{aligned}\ ] ] where for later use in we have already integrated by parts in time , and @xmath780 is as in . in what follows we will take the limit of as @xmath619 by means of compactness arguments , combined with techniques from maximal monotone operator theory . _ step @xmath11 : compactness . _ first of all , we observe that due to estimates and , there holds @xmath781 therefore , , joint with and well - known weak and strong compactness results ( cf . @xcite ) , yield that there exist a vanishing sequence of time - steps @xmath782 and @xmath1 as in such that as @xmath783 @xmath784 $ , } \\ & \partial_t { \widehat{{\mathbf{u}}}_{\tau_k } } { \rightharpoonup}\partial_{tt}{\mathbf{u } } & \text { in $ l^2(0,t;l^{2}(\omega;\r^d ) ) $ , } \\ & \partial_t { { { \mathbf{u}}}_{\tau_k}}\to \partial_t { \mathbf{u}}&\text { in $ l^2(0,t;h^1(\omega;\r^d ) ) $ . } \end{array}\ ] ] a stability estimate analogous to the first of , the a priori bounds , and and the previously mentioned compactness arguments , imply that there exist @xmath785 and @xmath786 such that , along a not relabeled subsequence there hold as @xmath783 @xmath787 $ , } \end{array}\ ] ] as well as @xmath788 furthermore , if in addition @xmath146 complies with and , then , due to we also have the enhanced regularity , and the strong convergence @xmath789 as for @xmath790 , estimates and a generalization of the aubin - lions theorem to the case of time derivatives as measures ( see e.g. ( * ? ? ? * chap . 7 , cor . 7.9 ) ) yield that there exists @xmath165 as in such that , up to the extraction of a further subsequence , as @xmath783 there hold @xmath791 $ and $ 1\leq s<\infty$. } \end{array}\ ] ] furthermore , by an infinite - dimensional version of helly s selection principle ( cf . e.g. @xcite ) we have @xmath792 in @xmath793 for all @xmath209 $ ] . taking into account the a priori bound of @xmath794 in @xmath795 , we then conclude that @xmath796$.}\ ] ] clearly , the second of implies that @xmath797 a.e . in @xmath107 , hence by the continuity of @xmath359 we also have @xmath798 a.e . in @xmath107 . moreover , estimate guarantees that @xmath799 is uniformly integrable in @xmath432 . therefore , thanks e.g. to ( * ? ? ? iii.3.6 ) , we conclude that @xmath800 _ step @xmath554 : passage to the limit in . _ it follows from and that @xmath801 moreover , and easily yield that @xmath802 is bounded in @xmath803 . since @xmath804 taking into account , we can pass to the limit as @xmath805 in the third integral in the first line of . convergences , , as well as for @xmath806 , allow us to take the limit of as @xmath807 . hence we conclude that @xmath808 comply with . as for the passage to the limit in , we observe that , the compact embedding , and imply that @xmath809 and @xmath810 in @xmath811 . therefore , by the we immediately conclude that @xmath812 and @xmath813 in @xmath196 for almost all @xmath351 . also relying on the third of and on for @xmath814 , we find that @xmath384 fulfill . finally , combining with and taking into account the strong - weak closedness of the graphs of the operators @xmath815 , we immediately conclude that hold with @xmath816 we also observe that , , and yield @xmath817 therefore , relying on and the lipschitz continuity of @xmath539 on bounded intervals , we easily pass to the limit in and infer that @xmath818 fulfill . _ step @xmath560 : proof of total energy equality . _ we choose @xmath819 in , test by @xmath343 and integrate on time , and test by @xmath28 and integrate on time , then add the resulting relations . some terms cancel out , and to conclude we use the chain rule formula for the functional @xmath820 , as well as the fact that @xmath821 _ outline . _ first of all , relying on the a priori estimates of prop . [ prop : aprio-2 ] , we pass to the limit as @xmath485 in the time - discretization scheme : we thus obtain the existence of a triple @xmath822 solving the truncated version of system . secondly , we perform the passage to the limit as the truncation parameter @xmath823 tends to @xmath824 . in this step , we need to obtain for the functions @xmath825 a bound in the spaces specified in , _ independent _ of the parameter @xmath641 . in this direction , the key estimate consists in testing below by ( a truncation of ) @xmath826 , which is now an admissible test function for : it is indeed in view of performing this test , that we need to keep the two passages to the limit as @xmath485 and as @xmath827 distinct . in order to carry out the calculations related to such an estimate , we need to carefully tailor to the present truncated setting the _ formal _ computations outlined in remark [ rmk : afterthm2 ] . _ step @xmath11 : passage to the limit as @xmath485 , for @xmath759 fixed , in . _ the argument follows the very same lines as the one in the proof of thm . [ teor1 ] : it is even easier , due to the truncations of the functions @xmath310 and @xmath359 . therefore , we omit the details . let us just observe that passing to the limit as @xmath485 in leads to a solution @xmath828;l^2(\omega ) ) \cap h^1(0,t;h^1(\omega)')\ ] ] of the _ truncated _ enthalpy equation @xmath829 for all @xmath830 indeed , regularity follows from estimates , and in turn it allows for the _ stronger _ formulation ( in comparison with ) of the enthalpy equation . therefore , for every @xmath759 the ( cauchy problem for the ) truncated version of system , consisting of and of with @xmath359 replaced by @xmath768 , admits a solution @xmath831 , with the regularity for @xmath832 and for @xmath833 , further fulfilling ( with @xmath768 in place of @xmath359 ) . _ step @xmath554 : passage to the limit as @xmath827 . _ let @xmath834 be the family of solutions constructed in the previous step . since estimates and hold with a constant _ independent _ of @xmath641 , we conclude by lower semicontinuity that @xmath835 next , introduce the truncation operator @xmath836 and the sets @xmath837 hence , we test by @xmath838 and integrate on @xmath839 , @xmath351 : observing that @xmath840 we thus obtain @xmath841 where we have used the place - holder @xmath842 . now , arguing in the very same way as in rmk . [ rmk : afterthm2 ] we observe that @xmath843 for positive constants @xmath566 and @xmath567 _ independent _ of @xmath641 . let us now focus on the on the r.h.s . of : note that @xmath844 , whereas the second integral term can be estimated thanks to on @xmath48 . taking into account the growth of @xmath359 and , the third integral can be estimated by @xmath845 where we have used estimates . choosing @xmath362 sufficiently small , we can absorb the integral term on the r.h.s . of into the r.h.s . of . as in rmk . [ rmk : afterthm2 ] , we thus conclude that @xmath846 we now use in order to infer an analogous estimate for the family @xmath847 . to do so , we preliminarily observe that from the bound for @xmath848 we infer @xmath849 therefore , upon testing by @xmath832 , integrating in time , and repeating the same calculations as above ( also relying on ) , we end up with @xmath850 for some @xmath851 to be specified later , and we estimate @xmath852 where we have argued along the same lines as in rmk . [ rmk : afterthm2 ] . now , observe that for almost all @xmath351 @xmath853 where the second inequality is due to the growth of @xmath359 , and the last one to . observe that @xmath854 for @xmath855 thanks to , therefore @xmath856 as @xmath827 . for @xmath250 , in taking into account the sobolev embedding @xmath857 for every @xmath858 , we can replace @xmath859 with @xmath860 for any @xmath716 , and tune @xmath861 in such a way that the latter term will again converge to @xmath10 as @xmath827 . therefore the third term on the r.h.s . of is bounded . it remains to choose @xmath862 in such a way as to absorb the term @xmath863 into the l.h.s . of , and @xmath864 , @xmath865 so that @xmath866 is sufficiently small . also applying the gronwall lemma , we conclude that @xmath867 with easy calculations we also find @xmath868 with @xmath869 as in , the estimate for @xmath870 following from a comparison in . we are now in the position to obtain the further estimates @xmath871 where @xmath872 is the selection in @xmath873 fulfilling . indeed , can be proved relying on the previously obtained and , by performing on the truncated version of system , the time - continuous analogues of the _ sixth _ and _ tenth a priori estimates _ ( cf . [ ss:3.2 ] ) . hence , we can carry out the passage to the limit argument as @xmath874 . relying on the bounds and and on the compactness tools already exploited in the proof of thm . [ teor1 ] , we find that there exist @xmath332 and a ( not relabeled ) subsequence of @xmath875 such that ( the time - continuous analogues of ) convergences hold as @xmath827 . furthermore , estimates , the aforementioned generalization ( * ? ? ? 7 , cor . 7.9 ) of the aubin - lions theorem to the case of @xmath127-functions applies , and convergences improve to @xmath876 $ and $ 1\leq \sigma<\infty$. } \end{array}\ ] ] with @xmath877;{w}^{2,{\mathfrak{s}}}(\omega)')$ ] and @xmath869 as in . relying on and on the lipschitz continuity of @xmath359 , it is not difficult to infer that @xmath878 in @xmath432 as @xmath783 . furthermore , combining and the last of we also have that @xmath879 in @xmath880 . therefore , we are able to pass to the limit in ( with test functions as in the statement of thm . [ teor1bis ] ) , and in the corresponding equations for @xmath1 and @xmath2 in the case @xmath139 , which concludes the proof . _ step @xmath10 : approximate equations . _ equations and can be rephrased in terms of the interpolants @xmath907 in a way analogous to . furthermore , taking into account that @xmath908}$ ] and that @xmath909 we observe that the minimum problem yields for every @xmath910 that @xmath911 ( recall the short - hand notation for @xmath732 ) . writing necessary optimality conditions for the above minimum problem , we infer @xmath912 \text { and all } \eta \in w^{1,p}(\omega ) \text { with } 0 \leq \eta \text { and } \eta \leq { \underline{\chi}_{\tau}}(t ) \ { \text{a.e . in\,}}\ , \omega , \end{aligned}\ ] ] where we have used the short - hand notation ( cf . ) @xmath913 letting @xmath914 in and dividing the resulting inequality by @xmath762 , we deduce that @xmath915 \\ & \text { and all } \varphi \in w^{1,p}(\omega ) \text { s.t.\ there exists $ \nu>0 $ with } 0 \leq \nu \varphi + { \overline{\chi}_{\tau}}(t ) \leq { \underline{\chi}_{\tau}}(t ) \ { \text{a.e . in\,}}\ , \omega . \end{aligned}\ ] ] choosing @xmath916 ( observe that it complies with the constraint above , upon taking @xmath917 ) , we therefore obtain @xmath918 therefore , upon summing over the index @xmath522 we deduce the _ discrete version _ of the energy inequality for all @xmath919 , viz . @xmath920 where we have estimated the last term on the right - hand side of using that @xmath921 , thanks to the lipschitz continuity of @xmath539 . _ step @xmath11 : compactness . _ in view of the a priori estimates from proposition [ prop : aprio ] , we infer that there exist a vanishing subsequence @xmath922 and limit functions @xmath332 such that convergences , , hold true as @xmath923 . observe that in particular yields that @xmath924 and @xmath106 a.e . in @xmath925 . arguing as in the proof of ( * ? ? ? * lemma 5.11 ) , we now prove that @xmath926 indeed , ( * ? ? ? * lemma 5.2 ) gives a sequence @xmath927 of test functions for , fulfilling @xmath928 observe that the first of and convergences yield in particular @xmath929 we have @xmath930 where the first inequality follows from and the second one from elementary algebraic manipulations . now , choosing @xmath931 in ( which we are allowed to do thanks to ) and integrating in time , we obtain @xmath932 due to the bounds and , and to and . we also have @xmath933 where the second inequality follows from and , and the last passage is due to . taking into account that @xmath934 in @xmath935 by , we also prove that @xmath936 as @xmath783 . in this way , from we conclude . observe that , combined with the bound then yields . _ step @xmath554 : passage to the limit . _ arguing in the very same way as for the proof of thm . [ teor1 ] , it is possible to prove that @xmath332 solve equations and . it now remains to prove the variational inequality , together with , and the energy inequality . as for the latter , it is sufficient to pass to the limit as @xmath783 in . for this , we use convergences , , , , , as well as , which in particular yields @xmath937 clearly , the last term on the right - hand side of tends to zero . since the argument for is perfectly analogous to the one developed in the proof of ( * ? ? ? 4.4 ) , we refer the reader to @xcite for all details and here just outline its main steps . passing to the limit in as @xmath938 with suitable test functions from ( * lemma 5.2 ) , we prove that for almost all @xmath351 @xmath939 from this , arguing as in the proof of ( * ? ? ? 4.4 ) we deduce that for almost all @xmath351 @xmath940 therefore , we take @xmath941 @xmath942 denoting the characteristic function of the set @xmath943 . from we deduce that , with this @xmath944 inequality holds . moreover , it is immediate to check that @xmath944 also complies with . _ step @xmath560 : strict positivity of the temperature . _ suppose that holds : the discrete strict positivity and convergences yield that , in the limit , @xmath945 for almost all @xmath412 . therefore , ensues . [ ss:5.2 ] _ step @xmath11 : compactness . _ for the interpolants @xmath946 of the solutions @xmath947 of the discrete problem [ probk - irr - iso ] , estimates and hold . therefore , standard strong and weak compactness results yield that there exist @xmath384 fulfilling and a subsequence @xmath948 such that convergences and hold . moreover , estimates also imply that @xmath2 has the enhanced regularity , and that @xmath949 furthermore , there exist @xmath950 and @xmath944 such that , possibly along a further subsequence , @xmath951 _ step @xmath554 : passage to the limit . _ relying on convergences , , and , we take the limit of the discrete momentum equation . as for , we observe that , thanks to estimate and the second of , there holds @xmath952 therefore , also taking into account we pass to the limit in and conclude @xmath453 fulfill , with @xmath330 replaced by @xmath331 . furthermore , combining with we have @xmath953 thanks to ( * ? ? ? 2.5 , p. 27 ) we conclude that @xmath454 a.e . in @xmath954 finally , testing equation by @xmath955 and integrating in time , with calculations analogous to we find for all @xmath209 $ ] @xmath956 ( here @xmath957 is as in , with @xmath958 in place of @xmath959 , and @xmath960 ) , where the last inequality is due to and , the mosco - convergence of @xmath961 to @xmath85 , the lipschitz continuity of @xmath539 , and . the last identity follows from equation . the aforementioned tokens of maximal monotone operator theory allow us to deduce from the fact that @xmath962 that @xmath963 a.e . in @xmath107 , which concludes the proof . [ more - general - alpha ] indeed , in the proof of theorem [ teor4 ] the fact that @xmath964}$ ] has never been specifically used , therefore thm . [ teor4 ] extends to a maximal monotone operator @xmath461 as in , observing that , up to perturbing @xmath965 with an affine function , it is not restrictive to suppose that @xmath966 let @xmath881 , @xmath375 , be two solution pairs like in the statement of theorem [ teor2 ] and set @xmath882 . taking into account that @xmath8 is constant ( cf . ) , hence @xmath883 for @xmath375 ( cf . ) , it is immediate to check that @xmath384 fulfill a.e . in @xmath884 @xmath885 now , we test by @xmath343 and integrate in time . recalling , it is not difficult to infer @xmath886 where we have @xmath887 whereas , the lipschitz continuity of @xmath9 on bounded intervals ( cf . ) and the hlder inequality yield @xmath888 where in the last inequality we have exploited the embeddings @xmath889 and ( * ? ? ? 16.4 , p. 102 ) , with @xmath762 a suitable constant to be chosen later and the constant @xmath567 also depending on @xmath890 . moreover , we get @xmath891 noting that @xmath892 we obtain from that @xmath893 next , we test by @xmath2 integrate the resulting equation in time . with elementary computations , also taking into account the lipschitz continuity of @xmath539 , the monotonicity of @xmath160 , and the crucial inequality , we get @xmath894 @xmath895 using and the fact that @xmath896 we get @xmath897 where the last inequality is obtained arguing as for . again exploiting the lipschitz continuity of @xmath752 on bounded intervals and the bound for @xmath898 in @xmath899 , we get @xmath900 where the last estimate also follows from the continuous embedding @xmath901 and the young inequality . collecting now , we arrive at @xmath902 summing up and and choosing @xmath903 , we conclude @xmath904 the application of the standard gronwall lemma gives immediately the desired continuous dependence estimate . [ slaplcd ] if we replace the @xmath67-laplacian with the linear @xmath75-laplacian in the equation for @xmath2 , the continuous dependence estimate of theorem [ teor2 ] can be performed without assuming @xmath8 to be constant ( cf . ) . indeed , in this case we would be able to deal with the additional term @xmath905 , which results from subtracting the equations fulfilled by solution pairs @xmath376 , @xmath375 . it would be possible to estimate it by means of the @xmath288-norm of @xmath906 , which would pop in on the left - hand side of . we now address the passage to the _ degenerate _ limit @xmath967 in the full system . for technical reasons which will be clarified in remark [ rmk : explanation ] later on , we focus on the _ irreversible _ case @xmath133 , and neglect the thermal expansion term in the momentum equation , i.e. take @xmath125 . furthermore , we confine the discussion to the case , in which , for @xmath968 the coefficients of _ both _ the elliptic operators in are truncated , cf . remark [ rmk : added ] below . in particular , we will take the functions @xmath8 and @xmath9 of the form @xmath969 [ trivial - case ] the choice @xmath970 and @xmath60 in and the truncation of both coefficients would lead to the momentum equation @xmath971 for which the asymptotic analysis @xmath137 would be less meaningful in the case of an _ irreversible _ evolution for @xmath2 . for , starting from an initial datum @xmath972 with @xmath973 , we would have @xmath974 for all @xmath975 $ ] . hence , the limit @xmath976 would not lead to elliptic degeneracy in . [ rmk : added ] it seems to us that _ both _ the coefficients @xmath8 and @xmath9 need to be truncated when taking the degenerate limit in the momentum equation . indeed , on the one hand the truncation of @xmath8 allows us to deal with the _ main part _ of the elliptic operator in . on the other hand , in order to pass to the limit in the quadratic term on the right - hand side of , we will also need to truncate @xmath9 , cf . later on . theorem [ teor3 ] guarantees that for every @xmath185 there exists a triple @xmath977 as in fulfilling the enthalpy equation with @xmath125 , the momentum equation @xmath978 ( where for later convenience we have dropped the operator notation ) , as well as @xmath979 for almost all @xmath980 ( changing sign in and recalling ) , and the energy inequality . as observed in section [ glob - irrev ] , the family @xmath164 then fulfills for all @xmath981 $ ] the _ energy inequality _ let us mention in advance that estimate for @xmath983 holds true only for the solutions @xmath164 obtained through the time - discretization procedure of section [ ss:3.2 ] . such solutions shall be referred to as _ approximable_. indeed , on the one hand , remark [ rmk : est - indep - delta ] ensures that the _ discrete _ estimates are valid with constants independent on @xmath169 : hence they are inherited by the approximable solutions @xmath983 , yielding estimate below . on the other hand , the calculations developed for the fourth a priori estimate in sec . [ ss:3.2 ] suggest that , in order to prove for _ all _ weak solutions @xmath984 to , it would be necessary to test by @xmath985 with @xmath702 as in . this is not an admissible choice due to the poor regularity of @xmath986 . since we do not dispose of a uniqueness result for the _ irreversible full _ system , we can not conclude for _ all _ weak solutions ( in the sense of def . [ def - weak - sol ] ) @xmath983 , and therefore we will restrict to _ approximable _ solutions . [ prop : indepe - delta ] assume hypotheses ( i ) , ( ii ) , and ( iv ) with @xmath152 , conditions on the data @xmath988 , and suppose that @xmath989 are given by . then , there exists a constant @xmath990 such that for all @xmath185 and for all @xmath164 _ ( approximable ) weak _ solutions to the irreversible full system , the following estimates hold @xmath991;w^{1,r'}(\omega)^ * ) } \leq \overline{s}.\end{aligned}\ ] ] estimates are straightforward consequences of the energy inequality , taking into account that @xmath992 for a constant independent of @xmath209 $ ] , estimating @xmath993 and applying the gronwall lemma . hence , follows from a comparison in , in view of . finally , can be proved by observing that the discrete estimates are in fact _ independent _ of the parameter @xmath185 , hence they carry over to the _ approximable _ solutions @xmath984 . as pointed out in @xcite ( see also @xcite ) , estimates suggest that for the analysis @xmath137 it is meaningful to work with the quantities @xmath994 in terms of which rewrites as @xmath995 for later purposes , we also observe that , in the setting of and with notation , the total energy inequality for the triple @xmath977 can be reformulated as @xmath996 the following result shows that the limit @xmath137 preserves the structure of the momentum equation , as well as the enthalpy equation . the weak formulation of the equation for @xmath2 is generalized by , cf . [ rmk : comparison - with - hk ] . [ teor5 ] assume hypotheses ( i ) , ( ii ) , ( iv ) with @xmath152 , and ( v ) with @xmath146 fulfilling . assume conditions on the data @xmath988 , and suppose that @xmath989 are given by . then , there exist @xmath165 as in , and @xmath997 such that @xmath998 fulfilling the weak enthalpy equation with @xmath125 , the _ weak _ momentum equation @xmath999 as well as @xmath1000 and the _ total energy inequality _ for almost all @xmath1001 $ ] @xmath1002 ( with @xmath1003 a suitable subsequence of @xmath1004 from ) , and for all @xmath1005 there holds @xmath1006 [ rmk : comparison - with - hk ] let us briefly compare the concept of weak solution ( to the _ degenerating _ irreversible full system ) arising from , with the notion of weak solution ( to the _ non - degenerating _ irreversible full system ) given in definition [ def - weak - sol ] , in the case in which @xmath1007 . suppose that the functions @xmath384 in and have further regularity properties , and that @xmath1008 a.e . in @xmath340 . then , holds a.e . in @xmath340 , hence it is immediate to realize that coincides with . furthermore , subtracting from the weak enthalpy equation tested by @xmath11 , we obtain a generalized form of the energy inequality for almost all @xmath154 $ ] and for @xmath155 . it follows from estimates and the same compactness arguments as in the proofs of thms . [ teor1 ] and [ teor3 ] that there exist a vanishing sequence @xmath1009 and functions @xmath165 as in and @xmath1010 as in such that as @xmath783 @xmath1011 $ and all $ 1\leq s<\infty$ , } \\ & \label{convergences - degen - u-1 } { \mathbf{u}}_{\delta_k } { { \rightharpoonup^*}}{\mathbf{u}}\text { in $ w^{1,\infty}(0,t;l^2(\omega;\r^d ) ) \cap h^{2}(0,t;h^{-1}(\omega;\r^d ) ) $ , } \\ & \label{convergences - degen - mu } { { \mbox{\boldmath$\mu$}}}_{\delta_k } { \rightharpoonup}{{\mbox{\boldmath$\mu$}}}\text { in $ l^2(0,t;l^2(\omega;\r^{d\times d}))$ , } \\ & \label{convergences - degen - eta } { { \mbox{\boldmath$\eta$}}}_{\delta_k } { { \rightharpoonup^*}}{{\mbox{\boldmath$\eta$}}}\text { in $ l^\infty(0,t;l^2(\omega;\r^{d\times d}))$ , } \\ & \label{convergences - degen - chi-1 } \chi_{\delta_k } { { \rightharpoonup^*}}\chi \text { in $ l^\infty(0,t;w^{1,p}(\omega ) ) \cap h^1 ( 0,t;l^2(\omega))$ , } \\ & \label{convergences - degen - chi-2 } \chi_{\delta_k } \to \chi \text { in $ \mathrm{c}^0([0,t];\mathrm{c}^0(\overline\omega))$ , } \end{aligned}\ ] ] the latter convergence due to the compactness results in @xcite and the compact embedding @xmath1012 . observe that and respectively yield @xmath1013;l^2 ( \omega;\r^d))$ } , \qquad \chi_{\delta_k } \to \chi \text { in $ \mathrm{c}^0_{\mathrm{weak}}([0,t];w^{1,p } ( \omega))$.}\ ] ] from , exploiting we deduce that @xmath1014 thus , we are in the position of passing to the limit as @xmath1009 in for the functions @xmath1015 , and conclude for @xmath808 . exploiting and the fact that @xmath1016 is nonincreasing for all @xmath1017 with the very same argument as in the proof of ( * ? ? ? 4.3 ) it is possible to prove that @xmath170 and @xmath171 have the form . in order to do that we can use the boundedness of @xmath1018 and of @xmath1019 in @xmath1020 for any compact cylinder @xmath310 of the form @xmath1021 $ ] on which @xmath1008 . notice that on these cylinders @xmath1022 for some @xmath1023 . hence , exploiting convergence , we infer that there exists @xmath1024 such that , for any @xmath1025 , we have @xmath1026 for all @xmath1027 . thus also @xmath1028 for all @xmath1029 $ ] because @xmath1016 is nonincreasing for all @xmath1030 . then we can identify at the limit @xmath170 and @xmath171 and cover @xmath1031 in by cylinders of the form @xmath310 above . hence , relying on it is immediate to pass to the limit in and conclude . next , we prove that @xmath1032 for this , we repeat the arguments from step @xmath554 in the proof of thm . [ teor3 ] , based on ( * ? ? ? * lemma 5.11 ) . namely , we apply ( * ? ? ? * lemma 5.2 ) , which gives a sequence @xmath1033 , such that @xmath1034 in @xmath935 and @xmath1035 a.e . in @xmath954 relying on assumption , with the same calculations as in we then have @xmath1036 now , choosing @xmath1037 as a test function for and integrating in time , we obtain @xmath1038 where @xmath1039 the second inequality due to , and the third one to the fact that @xmath1040 a.e . in @xmath107 . calculations completely analogous to the ones developed in the proof of thm . [ teor3 ] yield that @xmath1041 as @xmath1042 . in this way , from we conclude . we are now in the position to pass to the limit in ( the time - integrated version of ) and conclude . to this aim , we observe that , for any fixed test function @xmath1043 as in , @xmath1044 is a compact subset of @xmath1045 $ ] . hence there exists @xmath1046 such that @xmath1047 for all @xmath1048 , and , by , there exists @xmath1049 such that for @xmath1050 @xmath1051 therefore , @xmath1052 since @xmath1053 by . also exploiting , we succeed in taking the limit of the left - hand side of . as for the right - hand side , we use and argue in the following way @xmath1054 where we have used that , thanks to and , @xmath1055 uniformly on @xmath1044 , thus the last inequality e.g. follows from the lower semicontinuity result of @xcite . finally , follows from taking the limit as @xmath1056 of the total energy inequality , written on the interval @xmath839 for _ any _ @xmath154 $ ] . observe that , by ( cf . also the arguments in the proofs of ( * ? ? ? 4.3 ) ) , the map @xmath1057 has ( uniformly ) bounded variation . therefore , by helly s theorem up to a subsequence there exists @xmath1058 such that @xmath1059 for all @xmath209 $ ] . to identify @xmath1058 , we take the @xmath1060 as @xmath1056 of the first , second , fourth , and fifth term in @xmath1061 , exploiting convergences , , , , as well as , and relying on lower semicontinuity arguments . therefore we conclude that holds . finally , inequality follows combining the following facts : on the one hand , since @xmath1062 is uniformly bounded due to estimates , the dominated convergence theorem ensures @xmath1063 on the other hand , on account convergences , by weak lower semicontinuity arguments we have that @xmath1064 is greater or equal than the right - hand side of . this concludes the proof . these are the reasons why we have restricted the analysis of the degenerate limit to the irreversible system . within this setting , we further need to assume @xmath125 . indeed , because of the lack of estimates on @xmath1068 for @xmath137 , we would not be able to the limit in the term @xmath1069 in as @xmath1070 . we also point out that , seemingly , the total energy inequality can not be improved to an inequality holding on any subinterval @xmath1071 . indeed , for the sequence @xmath1072 only the weak convergence is available , which does not allow us to take the limit of the right - hand side of but for @xmath155 . finally , observe that the proof of thm . [ teor5 ] simplifies if the operator @xmath1073 is given by the nonlocal @xmath75-laplacian operator @xmath77 . in this case , in order to pass to the limit in it is no longer necessary to prove the strong convergence for @xmath1074 . in fact , the term @xmath1075 in is replaced by @xmath1076 , which can be dealt with by weak convergence arguments due to the linearity of the operator @xmath77 . the authors would like to thank all the referees for their careful reading of the paper , and in particular one of them for a very useful suggestion on how to improve thm . [ teor5 ] . they are also grateful to christiane kraus and christian heinemann for fruitful discussions on some topics related to this paper . h. brezis : oprateurs maximaux monotones et smi - groupes de contractions dans les espaces de hilbert , north - holland mathematics studies , no . 5 . , north - holland publishing co. , amsterdam - london ; american elsevier publishing co. , inc . , new york , 1973 . p. krej , e. rocca , j. sprekels : liquid - solid phase transitions in a deformable container , contribution to the book `` continuous media with microstructure '' on the occasion of krzysztof wilmanski s 70th birthday , springer ( 2010 ) , 285300 .
in this paper , we analyze a pde system arising in the modeling of phase transition and damage phenomena in thermoviscoelastic materials . the resulting evolution equations in the unknowns @xmath0 ( absolute temperature ) , @xmath1 ( displacement ) , and @xmath2 ( phase / damage parameter ) are strongly nonlinearly coupled . moreover , the momentum equation for @xmath1 contains @xmath2-dependent elliptic operators , which degenerate at the _ pure phases _ ( corresponding to the values @xmath3 and @xmath4 ) , making the _ whole _ system degenerate . that is why , we have to resort to a suitable weak solvability notion for the analysis of the problem : it consists of the weak formulations of the heat and momentum equation , and , for the phase / damage parameter @xmath2 , of a generalization of the principle of virtual powers , partially mutuated from the theory of rate - independent damage processes . to prove an existence result for this weak formulation , an approximating problem is introduced , where the elliptic degeneracy of the displacement equation is ruled out : in the framework of damage models , this corresponds to allowing for _ partial damage _ only . for such an approximate system , global - in - time existence and well - posedness results are established in various cases . then , the passage to the limit to the degenerate system is performed via suitable variational techniques . * key words:*phase transitions , damage phenomena , thermoviscoelastic materials , elliptic degenerate operators , nonlocal operators , global existence of weak solutions , continuous dependence . * ams ( mos ) subject classification:*35k65 , 35k92 , 35r11 , 80a17 , 74a45 .
recently , a great deal of theoretical attention has been devoted to the accurate determination of the shear - viscosity - to - entropy - density ratio ( @xmath0 ) of the quark - gluon plasma ( qgp ) and the hadron resonance gas ( hrg ) , and how its temperature dependence around the transition point between these two phases might be probed in heavy - ion collisions . in particular , a recent study @xcite suggested the possibility of discriminating experimentally between different parametrizations of the temperature dependence of @xmath0 using systematic analyses of the anisotropic flows @xmath1 and @xmath2 as functions of transverse momentum @xmath3 and pseudorapidity @xmath4 . in this paper , we present a complementary analysis of the hbt radii for au+au collisions at rhic energies , to argue that the hbt radii can be used profitably to help constrain viable parametrizations of @xmath5 . the hbt radii are physical observables which are well defined on an event - by - event basis . as parameters that describe the size and shape of the particle emitting source , they fluctuate from event to event . experimental data so far @xcite determine only the average source radii characterizing a large ensemble of events @xcite . elsewhere @xcite we discuss new ideas of how to complement such measurements of the ensemble - averaged hbt radii by a determination of their variances , i.e. , the width parameters characterizing their event - by - event fluctuations . in that work @xcite it is also shown that the experimentally accessible observables characterizing the mean source radii and their variances clearly track the algebraic mean and variance of the source radii that are calculated theoretically from the individual emission functions of a large set of dynamically simulated heavy - ion collisions with fluctuating initial conditions . we here show the transverse pair momentum ( @xmath6 ) dependencies of these mean radii and their variances for the same four parametrizations of the specific shear viscosity @xmath0 studied in @xcite . we also compare the mean radii to those one would obtain from a single average source by dynamically evolving a single initial profile , constructed as the ensemble average of the fluctuating initial conditions of the full ensemble . to save time and effort , we employ for this pioneering study the shortcut of calculating the hbt radii from the source variances of the emission function @xcite for directly emitted pions only , instead of performing a 3-d gaussian fit to the correlation function including all resonance decays @xcite . this approximation , which holds exactly only for gaussian sources @xcite , is known to be sufficient to discern qualitative features of the hbt radii and their @xmath7 dependences , although it is not accurate enough for quantitative comparisons with experimental data @xcite . in this paper , we compute the hbt radii , as outlined in the previous section , for central ( 0 - 10% ) au+au collisions at 200@xmath8gev , based on charged pion correlations . we use the iebe - vishnu package @xcite in modeling the evolution of heavy - ion collisions . for the ( hydrodynamical ) initial conditions of the fireball , we employ the mc - glauber model @xcite with p+p multiplicity fluctuations @xcite to generate the initial entropy density distribution in the transverse plane . the subsequent dynamical evolution is performed on an event - by - event basis , meaning that each set of initial conditions that we generate initializes one of the @xmath9 hydrodynamically evolved events in our ensemble . in this paper , we take @xmath10 . we perform the hydrodynamical evolution of the initial conditions with the vish2 + 1 code @xcite , using the s95p - pce165-v0 equation of state @xcite . we choose this evolution to begin at an initial proper time @xmath12fm/@xmath13 , where @xmath14 corresponds to the instant at which the two nuclei collide . also , although necessary for quantitatively precise comparisons with experimental data @xcite , we do not incorporate here any initial transverse flow into the hydrodynamical evolution . for our study , we use for @xmath5 one of the four parametrizations given in @xcite : * lh - lq , in which @xmath15 for all @xmath16 ; * lh - hq , in which @xmath15 for @xmath17 and @xmath18 * hh - lq , in which @xmath15 for @xmath19 and @xmath20 * hh - hq , for which we use @xmath21 for @xmath17 and @xmath22 for @xmath19 . here , @xmath23mev @xcite represents the transition temperature between the qgp and hrg phases . the four parametrizations are shown in fig . [ f1 ] . used in this work , taken from @xcite . [ f1 ] ] as also pointed out in @xcite , for identical initial conditions viscous entropy production generates different final multiplicities for different choices of @xmath24 . this effect is especially severe for the parametrizations lh - hq and hh - hq where at early times , when the longitudinal - transverse velocity shear is biggest , @xmath24 is very large . to compensate for this effect , we rescale our initial entropy profiles by a constant factor @xmath25 , which depends on the chosen parametrization of @xmath24 . for a given choice of @xmath24 , @xmath25 is chosen so that an ensemble - averaged initial entropy profile always reproduces the same final thermal @xmath26 multiplicity . is a fixed fraction of the total multiplicity @xmath27 . therefore , this normalization ensures that , on average , all events have the same final charged multiplicity . ] normalizing all of the averaged profiles to the ideal fluid case ( which thus corresponds to @xmath28 ) we find * lh - lq : @xmath29 * lh - hq : @xmath30 * hh - lq : @xmath31 * hh - hq : @xmath32 as expected , parametrizations with a large plasma viscosity require a significant suppression of the initial entropy density profile in order to compensate for viscous heating and to reproduce the measured final multiplicities in au+au collisions at rhic energies . finally , we terminate the hydrodynamical evolution on a decoupling surface ( the `` freeze - out surface '' ) of constant temperature @xmath33 mev . on the freeze - out surface for each event , the pion emission function is defined by the cooper - frye integral @xcite @xmath34 here , @xmath35 is the first - order viscous corrections to the ideal distribution function @xmath36 @xcite for which we assume a quadratic dependence on @xmath37 . @xmath38 and @xmath39 are the shear stress and flow velocity profiles along the freeze - out surface @xmath40 , respectively . @xmath41 , @xmath16 , @xmath42 , and @xmath43 are the chemical potential , decoupling temperature , energy density , and pressure , respectively , which are all constant on @xmath40 by construction . @xmath44 is the outward pointing normal vector on @xmath40 at point @xmath45 . theoretically , the hbt radii may be defined for either a single event or an ensemble containing many events . due to statistical limitations arising from the finite number of particles emitted in a single event , precise measurements of the full set of hbt radii and their @xmath7 dependences for a single event are not possible @xcite . three - dimensional experimental hbt analyses , therefore , exclusively report measurements based on large collections of events , rather than measurements of event - by - event hbt radii . in order to obtain an apples - to - apples comparison with experimental data , theoretical hbt calculations must therefore , at some level , average over all events in the ensemble . this procedure , known as _ ensemble averaging _ , can be performed in several ways , two of which we consider here . according to one prescription , the initial conditions for all events in the ensemble are averaged _ before _ their hydrodynamical evolution . we refer to this method of ensemble averaging as `` single - shot hydrodynamics '' ( ssh ) @xcite . for a sufficiently large ensemble , averaging over many , individually fluctuating and bumpy initial conditions results in a _ smooth _ initial transverse entropy density profile , which in turn may be evolved hydrodynamically as a single , averaged event . this prescription eliminates any sensitivity to event - by - event fluctuations in the initial state . we will here denote the hbt radii derived from single - shot hydrodynamics by @xmath46 . an alternative prescription for computing the ensemble - averaged hbt radii evolves each fluctuating event _ independently _ and averages over the entire ensemble only _ after _ the hbt radii have been computed for each event from its individual hydrodynamic emission function @xmath47 . this leads to the following definition of ensemble - averaged hbt radii : @xmath48 we denote by @xmath49 the hbt radii of the @xmath50th event , and we label the mean hbt radii computed according to this `` direct ensemble average '' ( dea ) by @xmath51 . the basic formalism describing our calculation of the hbt radii was presented in @xcite . once the emission function for a particular event has been obtained from the hydrodynamic output using eq . , the hbt radii corresponding to that emission function may be defined by @xcite @xmath52 where @xmath53 , @xmath54 , @xmath55 , and the @xmath56 dependence of @xmath57 arises from the @xmath56 dependence of the emission function in the definition of the source average : @xmath58 @xmath59 is the average pair momentum in the lab frame , @xmath60 . in this paper , we consider only pairs of identical particles at mid - rapidity ( @xmath61 ) . the dependence of the hbt radii on the pair emission angle @xmath62 also allows them to be expanded in a fourier series ( we suppress the additional dependence of all hbt parameters on @xmath7 and @xmath63 ) , @xmath64 , \label{r2ij_ft_defn}\end{aligned}\ ] ] where the flow - plane angle @xmath65 is defined by the complex phase of the @xmath66th - order anisotropic flow coefficient @xmath67 @xcite : @xmath68 in this work , we average @xmath69 , @xmath70 , and @xmath71 over @xmath62 , but we report the 2nd @xmath62-harmonic @xmath72 of the out - side cross - term @xmath73 , since the mean value of its @xmath62-average @xmath74 vanishes for symmetric nucleus - nucleus collisions at midrapidity @xcite . for single events with fluctuating initial conditions , @xmath72 receives non - zero contributions from both @xmath75 and @xmath76 on an event - by - event basis , but the former contribution averages to zero for a sufficiently large ensemble . we therefore consider only the mean and variance of the event - wise fluctuations of @xmath77 for the remainder of this paper . for notational simplicity , we will drop the 0 subscript from the azimuthally averaged radii , and write simply @xmath78 . at @xmath61 , eq . ( [ svhbt_defn ] ) gives @xmath79 @xmath69 and @xmath71 thus depend only on the geometric aspects of the emission function . by contrast , both @xmath70 and @xmath80 involve a combination of spatial and temporal source variances : @xmath81 in @xcite we describe a method for experimentally measuring , in addition to the mean , also the variance of an event - by - event distribution of the hbt radii . as a prediction for such measurements , we here also compute the variances of the hbt radii for our ensemble of @xmath82 events , defined by makes @xmath83 an unbiased estimator of the true variance of the event - by - event distribution of hbt radii @xcite . ] @xmath84 in this paper , we will use and plot @xmath85 as a measure for the widths of the hbt radii distributions . for the quantity @xmath86 , we define the corresponding variance to be @xmath87 in this subsection , we present the mean hbt radii computed according to the two averaging prescriptions described above ( single - shot hydrodynamics ( ssh ) and the direct ensemble average ( dea ) ) , and study their sensitivity to the temperature dependence of the specific shear viscosity @xmath24 . vs. @xmath7 from single - shot hydrodynamics , for ideal and viscous fluids with the temperature dependent shear viscosities @xmath24 shown in fig . note the almost perfect overlap of the green ( dashed ) and blue ( dotted ) curves in panels ( b , c ) . [ f2 ] ] , but leaving out in eq . ( [ cooper_frye_defn2 ] ) the viscous correction @xmath35 at freeze - out . comparison of figs . [ f2]a , d and [ f3]a , d shows that the sensitivity of @xmath88 and @xmath89to the temperature dependence of @xmath24 below @xmath90 is almost entirely due to the theoretically poorly controlled viscous correction @xmath35 at freeze - out , and that any sensitivity to the temperature dependence of @xmath24 during the dynamical evolution , i.e. _ before freeze - out _ , is strongly weighted at high temperatures , i.e. early times . the sensitivity to @xmath35 at freeze - out of the hbt radii probing the outward and longitudinal dimensions of the emission function ( panels ( b , c ) ) is negligible . [ f3 ] ] first we consider in fig . [ f2 ] the radii @xmath91 extracted from single - shot hydrodynamics which averages over event - by - event fluctuations in the initial state . panels ( a ) and ( d ) show that the sideward radius @xmath92 and the out - side cross term @xmath93 exhibit sensitivity to the temperature dependence of @xmath24 both below and above @xmath90 . however , @xmath24 affects hadronic observables in two distinct ways . first , there is the cumulative dynamical effect of shear viscosity on the development of flow in the fireball ; flow probes the entire temperature history of @xmath24 between the initial and decoupling temperatures . second , the emission function is affected by the viscous correction @xmath35 to the distribution function at freeze - out , eq . ( [ cooper_frye_defn2 ] ) , which is controlled by the value of @xmath24 on the freeze - out surface . this gives a contribution to observables such as the hbt radii that depends exclusively on the behavior of @xmath24 at @xmath94 and is basically independent of its prior history ( except for the accumulated effect of @xmath24 on the flow pattern whose associated shear tensor also affects the shear stress at freeze - out ) . to separate the two effects , we plot in fig . [ f3 ] the hbt radii computed without the @xmath35 correction , so that only the cumulative dynamical effects of @xmath24 remain . figures [ f3]a , d show that deleting @xmath35 removes all sensitivity of the sideward and out - side hbt radii on the shear viscosity in the hadronic phase , leaving only a weak sensitivity on the behavior of @xmath24 in the high - temperature qgp phase . on the other hand , comparison of panels ( b , c ) in figs . [ f2 ] and [ f3 ] reveals that the outward and longitudinal hbt radii are almost completely unaffected by the behavior of @xmath24 below @xmath90 and , in particular , receive no significant contribution from the viscous correction @xmath35 at freeze - out . this finding is at variance with the results reported in teaney s pioneering analysis @xcite of @xmath35-effects on the hbt radii . we note , however , that teaney s analysis was based on a blast - wave parametrization of the hydrodynamic flow profile at freeze - out , rather than the profile from a genuine hydrodynamic simulation of the flow . it was noted before in appendix e of ref . @xcite that @xmath35 is very sensitive to the details of the velocity shear tensor at freeze - out , and even its sign is fragile . as a function of pair momentum @xmath7 , for the same choices of @xmath24 as in figs . [ f2 ] and [ f3 ] . this figure should be compared with fig . [ f2 ] . [ f4 ] ] furthermore , there are well - known conceptual uncertainties about the correct form of @xmath35 , especially its @xmath3 dependence @xcite . these uncertainties render suspect any observable that strongly depends on @xmath35 at freeze - out .- related uncertainties may also affect some of the conclusions in ref . @xcite about the dependence of charged hadron elliptic and quadrangular flow , @xmath95 and @xmath96 , on the behavior of @xmath24 in the hadronic phase . for example , we found that , when calculated without the @xmath35 correction , almost all sensitivity of @xmath97 of directly emitted ( `` thermal '' ) pions to the temperature dependence of @xmath24 in the hadronic phase disappears , leaving only a ( significantly weaker , but robust ) dependence of this observable to the temperature dependence of @xmath24 in the qgp phase above @xmath90 . ] therefore , we caution the reader not to trust the apparent sensitivity in fig . [ f2]a , d of @xmath88 and @xmath89 to the behavior of @xmath24 in the hadronic phase . on the other hand , the sensitivity of @xmath98 and @xmath99 exclusively to the temperature dependence of @xmath24 in the qgp phase _ above @xmath90 appears to be robust and unaffected by freeze - out uncertainties related to @xmath35 . these high - temperature effects of @xmath24 on @xmath100 and @xmath101 are not huge , but they can reach 20% ( for the squared hbt radii ) for a range of @xmath7 values below 1.5gev . studies such as @xcite have explored the sensitivity of the hbt radii to shear viscosity using constant ( @xmath16-independent ) @xmath0 and varying that constant . different from our approach here , when changing @xmath0 they also changed other hydrodynamic parameters to ensure that not only the normalization but also the slope of the @xmath3-spectra was held fixed . additionally , both of these studies extract the hbt radii from a gaussian fit to the full three - dimensional correlation function instead of using the short - cut through the source variances . these differences make a direct comparison of our work with their results difficult . our findings are qualitatively supported by the work @xcite ( whose authors , like us , used the source variances method ) where it was found that an increase in the knudsen number ( or , equivalently , @xmath0 ) resulted in a decrease in @xmath70 . the purpose of the present work is not a realistic comparison with experimental data which , as done in @xcite , would require a simultaneous tuning of several additional parameters ( see also ref . our goal is to check systematically the sensitivity of the hbt radii to a possible @xmath16-dependence of @xmath0 , in particular in the context of event - by - event source fluctuations . it is to these fluctuations that we turn our attention next . we now investigate how the mean hbt radii are affected by event - by - event fluctuations in the initial conditions . to do so we evolve 1000 central ( 0 - 10% centrality ) au+au events at @xmath102gev with fluctuating initial conditions and average their correspondingly fluctuating hbt radii after freeze - out , as described in sec . [ ssec2d ] . the results are presented in fig . [ f4 ] ( including the full distribution function ( [ cooper_frye_defn2 ] ) at freeze - out ) and fig . [ f5 ] ( without the viscous @xmath35 correction in ( [ cooper_frye_defn2 ] ) at freeze - out ) . compared to figs . [ f2 ] and [ f3 ] , we see no qualitative differences . again , robust ( although not very strong ) sensitivities to the temperature dependence of @xmath24 _ above _ @xmath90 are seen in all radii , but for @xmath103 these are buried by a stronger sensitivity to the viscous @xmath35 correction at freeze - out ( which , we reiterate , is theoretically not well controlled ) . as was the case for the @xmath104 from single - shot hydrodynamics , the @xmath35 corrections to the @xmath105 are negligible when @xmath106 . , but without the @xmath35 correction . this figure should be compared with fig . [ f5 ] ] in fig . [ f6 ] we plot the fractional change @xmath107 of the mean squared hbt radii arising from event - by - event fluctuations , as a function of pair momentum . generically , event - by - event fluctuations are seen to boost the @xmath62-averaged means of the fluctuating radii by a few percent above the corresponding radii obtained from single - shot hydrodynamic evolution of a smooth averaged initial condition . exceptions to this rule are the sideward radius for the hh - lq and hh - hq parametrizations ( i.e. , for large shear viscosities at freeze - out ) in the range @xmath108gev , driven by effects from the @xmath35 correction at freeze - out , and the longitudinal radius at small pair momentum @xmath109gev . the systematic effects of event - by - event fluctuations on the mean @xmath62-oscillation amplitude @xmath110 are less unambiguous , with fluctuations increasing or decreasing @xmath110 relative to @xmath93 in different @xmath7-ranges depending on the specific temperature dependence selected for @xmath24 ; the differences can become substantial at small @xmath7 where the oscillation amplitude is small , with @xmath110 even changing sign relative to @xmath93 in the lh - hq and hh - hq parametrizations . generically the differences between the mean fluctuating hbt radii and their single - shot hydrodynamic analogs are largest for ideal fluid evolution and somewhat smaller for viscous evolution . this is consistent with the idea that viscosity attenuates fluctuation - driven effects . of the squared hbt radii due to event - by - event fluctuations , as a function of pair momentum @xmath7 . see text for discussion . [ f6 ] ] [ cols="^ " , ] for a better understanding of how different temperature dependences of @xmath24 affect the magnitudes and @xmath7 dependences of the radius parameters @xmath91 and @xmath51 we can look directly at the emission functions . these are shown in fig . [ f7]a for a randomly selected single fluctuating event , and in fig . [ f7]b for the single - shot hydrodynamic event with smooth ensemble - averaged initial conditions . having seen in figs . [ f2]-[f5 ] that at rhic energies the hbt radii discriminate mostly between @xmath24 parametrizations that differ in the value of @xmath0 at the earliest times and highest temperatures , we focus in fig . [ f7]b on the hh - lq and hh - hq parametrizations ( a comparison of lh - lq with lh - hq would lead to similar conclusions ) . as is well known ( see , e.g. , ref . @xcite ) , the average radial flow velocity increases approximately linearly with @xmath111 as one moves from the top of the freeze - out surface in fig . [ f7]b ( at @xmath112 ) to its vertical part ( where @xmath113 takes its maximum value ) . for the cases studied in fig . [ f7 ] , this is shown in fig . [ f8 ] . this observation explains why , for increasing @xmath7 , the region of maximal emissivity ( colored red ) moves from around @xmath112 for small momentum pairs to @xmath114 for large momentum pairs . [ f7]a illustrates how this phenomenon manifests itself in an individual fluctuating event : while the regions of highest emissivity roughly follow the average tendencies seen in fig . [ f7]b , these tendencies are strongly modulated by fluctuations in the shape of the freeze - out surface and of the radial flow along this surface , giving rise to peaks and valleys of emissivity as one moves around the freeze - out surface at constant values of @xmath111 or @xmath115 . [ f7]a also illustrates that , in an individual event , the regions of maximal emissivity have a strong dependence on the azimuthal direction @xmath62 of pair emission , and that therefore , for an individual fluctuating event , one should expect strong @xmath62-dependence of the hbt radius parameters . comparing the freeze - out surfaces and regions of maximal emissivity shown in fig . [ f7]b for the two ensemble - averaged sources that were evolved with shear viscosities @xmath24 differing only in their behavior at high temperatures , we can make several observations : + ( 1 ) the larger specific shear viscosity of the hh - hq parametrization drives stronger radial flow which causes the fireball to expand to larger radii and complete its freeze - out at earlier times than for the hh - lq parametrization . this agrees with similar observations made in ref . @xcite . + ( 2 ) for longitudinally boost - invariant expansion with longitudinal velocity component @xmath116 , emissivity regions centered at later times see a smaller longitudinal flow gradient and thus a larger longitudinal region of homogeneity , reflected in a larger value for @xmath117 @xcite . [ f7]b shows that maximal emissivity for small and intermediate @xmath7 pairs is shifted to later times for the hh - lq parametrization than for hh - hq , while for large @xmath118gev the pairs are emitted at roughly the same times for both parametrization . this explains why in figs . [ f2]c ( and also in fig . [ f4]c ) the longitudinal hbt radii are larger for hh - lq than for hh - hq at small and intermediate @xmath7 but nearly identical at large @xmath7 . + ( 3 ) for small @xmath7 , the regions of highest emissivity are geometrically smaller in the outward and sideward direction and look at its horizontal width when frontally viewed . ] directions for the hh - lq case than for hh - hq , but the opposite is true at intermediate and high pair momentum . this explains the behavior of the sideward radius @xmath103 for these two parametrizations shown in figs . [ f2]a . the most important aspect of the freeze - out surfaces shown in figs . [ f7]b is the earlier freeze - out associated with larger shear viscosities at early times ( high temperatures ) . it reduces the emission duration and increases the longitudinal flow gradient at freeze - out , which reduces the @xmath119 and @xmath120 contributions to @xmath117 and @xmath121 in eqs . ( [ r2sl_defn ] ) , ( [ r2o_defn ] ) and is the root cause for the robustly smaller @xmath121 and @xmath117 values observed in figs . [ f2]b - c and [ f4]b - c for the hh - hq parametrization of @xmath24 compared to the hh - lq case . its effect on the cross - terms variances @xmath122 , @xmath123 and @xmath124 ( and thus the behavior of the cross - term radius @xmath125 ) is less intuitively obvious . last but not least , it is well established that the rate at which @xmath103 decreases with increasing pair momentum @xmath7 is a measure of the radial velocity gradient along the freeze - out surface @xcite . comparison of figs . [ f8]b , e , h ( middle column ) to figs . [ f8]c , f , i ( right column ) shows that evolution with @xmath126 leads to a larger radial flow gradient along the freeze - out surface than evolution with @xmath127 ( which has smaller shear viscosity at early times ) . this correlates with the steeper @xmath7 dependence for @xmath103 in figs . [ f2]a and [ f4]a for the hh - hq parametrization ( green dashed curves ) than for the hh - lq one ( black dash - dotted curves ) . finally , we consider the variances of the event - by - event distributions of the hbt radii . we present them in the form of _ relative _ widths , normalized to the directly ensemble averaged squared radii , in figures [ f9 ] ( full freeze - out distribution function ) and [ f10 ] ( without the viscous @xmath35 correction on the freeze - out surface ) . for the @xmath62-averaged radius parameters the relative widths are seen to be of order @xmath128 , with little dependence on either @xmath7 or the specific shear viscosity used in the hydrodynamic evolution . the event - by - event distribution of the out - side oscillation amplitude @xmath129 shows a somewhat smaller relative width , especially at small @xmath7 where the mean amplitude is small . at larger @xmath7 , shear viscosity appears to reduce the relative width of the event - by - event distribution of this oscillation amplitude compared to ideal fluid evolution . , for ideal fluid evolution and viscous dynamics with the four parametrizations shown in fig . [ f1 ] . [ f9 ] ] focusing on the @xmath62-averaged radii ( panels ( a - c ) in figs . [ f9 ] and [ f10 ] ) we observe that , just as was the case for their ensemble averages , all sensitivity of their variances to the temperature dependence of @xmath0 in the hadronic phase appears to come from the viscous @xmath35 correction at freeze - out and thus not to be of dynamical origin . the sensitivity of the hbt variances to dynamical evolution effects caused by varying the temperature dependence of @xmath0 at temperatures above freeze - out can be seen in fig . [ f10 ] where the @xmath35 contribution at freeze - out is removed . the results for viscous evolution split into two bands , one for high , the other for low shear viscosity at early times . the relative widths for ideal fluid dynamics fall in between these two bands . the two bands are characterized by the behavior of @xmath0 at _ high _ temperature , not at low @xmath16 : higher shear viscosity at early times ( i.e. at high temperatures ) leads to hbt variances that are 10 - 20% larger than those for evolution with lower shear viscosity at early times . the behavior of the shear viscosity at late times in the hadronic phase has no visible dynamical effect on the normalized widths of the hbt radii . we have presented a first analysis of the sensitivity of the hbt radii to the temperature dependence of the specific shear viscosity in relativistic heavy ion collisions at top rhic energies , taking into account that the hbt radii fluctuate from event to event and that a consistent comparison of experimental data with theoretical predictions should thus be based on event - by - event evolution of fluctuating initial conditions on the theory side . we explored the effect of event - by - event fluctuations on the mean hbt radii ( and their deviation from the values obtained from single - shot hydrodynamics where the initial - state fluctuations are ignored by averaging over them before hydrodynamic evolution ) and their relative widths . we found that fluctuations tend to increase the mean value of the fluctuating hbt radii by a few percent above the value obtained by the traditional process of averaging over the fluctuations already in the initial state , and found that the event - by - event distributions of the hbt radii are characterized by relative widths of order 10% . comparing the hbt radii from hydrodynamic evolutions with five different assumptions for the specific shear viscosity and its temperature dependence we showed that both the mean squared hbt radii and their relative widths are affected by a possible increase of the specific shear viscosity by a factor 10 between @xmath90 and @xmath130 at the level of 10 - 20% , with larger shear viscosities leading to smaller mean hbt radii and larger relative widths of their event - by - event distributions . there is little to no effect on either from variations in the behavior of @xmath24 in the hadronic phase below @xmath90 , except for the sideward and out - side radius parameters @xmath103 and @xmath72 and the sideward variance @xmath131 which all show significant sensitivity to the viscous @xmath35 correction at freeze - out . the latter , in turn , depends on the value of @xmath0 at the freeze - out temperature but not on its prior dynamical evolution in the hadronic phase . since the @xmath35 correction is not theoretically well constrained at the present moment , this sensitivity of @xmath69 and @xmath72 to @xmath0 at freeze - out should be viewed with some caution . , but without the viscous correction @xmath35 to the freeze - out distribution function . one sees that for the normalized width of the distribution of sideward radii @xmath103 , deleting @xmath35 removes all sensitivity to the temperature dependence of @xmath0 in the hadronic phase , while all other normalized widths are only weakly affected by @xmath35 . [ f10 ] ] the levels of sensitivity of the hbt radii and their variances to the temperature dependence of @xmath0 observed here are comparable to those observed in ref . @xcite for the elliptic and quadrangular flows . this indicates that precise 3-dimensional hbt measurements , in particular new measurements of their variances @xcite , can play a valuable supporting role in constraining the temperature dependence of the qgp shear viscosity . the authors would like to thank chun shen for a careful reading of the manuscript which exposed an error in its original version . this work was supported by the u.s . department of energy , office of science , office of nuclear physics under awards no . de - sc0004286 and ( within the framework of the jet collaboration ) de - sc0004104 . c. shen , z. qiu , h. song , j. bernhard , s. bass and u. heinz , arxiv:1409.8164 [ nucl - th ] . b. alver , m. baker , c. loizides and p. steinberg , arxiv:0805.4411 [ nucl - ex ] . h. song and u. heinz , phys . c * 77 * , 064901 ( 2008 ) . c. shen , u. heinz , p. huovinen and h. song , phys . c * 82 * , 054904 ( 2010 ) . s. pratt , phys . lett . * 102 * , 232301 ( 2009 ) ; and nucl . a * 830 * , 51c ( 2009 ) . f. cooper and g. frye , phys . d * 10 * , 186 ( 1974 ) . b. r. schlei , u. ornik , m. plumer and r. m. weiner , phys . b * 293 * , 275 ( 1992 ) . u. heinz , a. hummel , m. a. lisa and u. a. wiedemann , phys . c * 66 * , 044903 ( 2002 ) . a. tamhane and d. dunlop , _ statistics and data analysis : from elementary to intermediate _ ( pearson / prentice - hall , 2000 ) .
recent work has shown that a temperature dependence of the shear viscosity to entropy ratio , @xmath0 , influences the collective flow pattern in heavy - ion collisions in characteristic ways that can be measured by studying hadron transverse momentum spectra and their anisotropies . here we point out that it also affects the pair momentum dependence of the hanbury - brown twiss ( hbt ) radii ( the source size parameters extracted from two - particle intensity interferometry ) and the variance of their event - by - event fluctuations . this observation establishes interferometric signatures as useful observables to complement the constraining power of single - particle spectra on the temperature dependence of @xmath0 .
it is well - known that exact inference in _ tree - structured _ graphical models can be accomplished efficiently by message - passing operations following a simple protocol making use of the distributive law @xcite . it is also well - known that exact inference in _ arbitrary _ graphical models can be solved by the junction - tree algorithm ; its efficiency is determined by the size of the maximal cliques after triangulation , a quantity related to the treewidth of the graph . figure [ fig : examples_intro ] illustrates an attempt to apply the junction - tree algorithm to some graphical models containing cycles . if the graphs are not chordal ( ( a ) and ( b ) ) , they need to be triangulated , or made chordal ( red edges in ( c ) and ( d ) ) . their clique - graphs are then guaranteed to be _ junction - trees _ , and the distributive law can be applied with the same protocol used for trees ; see @xcite for a beautiful tutorial on exact inference in arbitrary graphs . although the models in this example contain only pairwise factors , triangulation has increased the size of their maximal cliques , making exact inference substantially more expensive . hence approximate solutions in the original graph ( such as loopy belief - propagation , or inference in a loopy factor - graph ) are often preferred over an exact solution via the junction - tree algorithm . even when the model s factors are the same size as its maximal cliques , neither exact nor approximate inference algorithms take advantage of the fact that many factors consist only of _ latent _ variables . in many models , those factors that are conditioned upon the observation contain fewer latent variables than the purely latent cliques . examples are shown in figure [ fig : examps ] . this encompasses a wide variety of models , including grid - structured models for optical flow and stereo disparity as well as chain and tree - structured models for text or speech . simple analysis reveals that the probability of choosing a permutation that does not contain a value inside a square of size @xmath1 is @xmath2 this is precisely @xmath3 , where @xmath4 is the cumulative density function of @xmath5 . it is immediately clear that @xmath6 , which defines the best and worst - case performance of algorithm [ alg1 ] . the case where we are sampling from multiple permutations simultaneously ( i.e. , algorithm [ alg : ext ] ) is analogous . we consider @xmath9 permutations embedded in a @xmath10-dimensional hypercube , and we wish to find the width of the smallest shaded hypercube that includes exactly one element of the permutations ( i.e. , @xmath11 , \ldots , p_{k-1}[i]$ ] ) . this is represented in figure [ fig : perms](c ) for @xmath12 . note carefully that @xmath10 is the number of _ lists _ in ( eq . [ eq : hatk ] ) ; if we have @xmath10 lists , we require @xmath9 permutations to define a correspondence between them . unfortunately , the probability that there is no non - zero entry in a cube of size @xmath13 is not trivial to compute . it is possible to write down an expression that generalizes ( eq . [ eq : factprob ] ) , such as @xmath14 ( in which we simply enumerate over all possible permutations and ` count ' which of them do not fall within a hypercube of size @xmath15 ) , and therefore state that @xmath16 however , it is very hard to draw any conclusions from ( eq . [ eq : pkm ] ) , and in fact it is intractable even to evaluate it for large values of @xmath17 and @xmath10 . hence we shall instead focus our attention on finding an upper - bound on ( eq . [ eq : runtimeexact ] ) . finding more computationally convenient expressions for ( eq . [ eq : pkm ] ) and ( eq . [ eq : runtimeexact ] ) remains as future work . although ( eq . [ eq : runtimek1 ] ) and ( eq . [ eq : runtimeexact ] ) precisely define the running times of algorithm [ alg1 ] and algorithm [ alg : ext ] , it is not easy to ascertain the speed improvements they achieve , as the values to which the summations converge for large @xmath17 are not obvious . here , we shall try to obtain an upper - bound on their performance , which we assessed experimentally in section [ sec : experiments ] . in doing so we shall prove theorems [ the : alg1 ] and [ the : algext ] . ( see algorithm [ alg1 ] ) consider the shaded region in figure [ fig : perms](d ) . this region has a width of @xmath19 , and its height @xmath1 is chosen such that it contains precisely one non - zero entry . let @xmath20 be a random variable representing the height of the grey region needed in order to include a non - zero entry . we note that @xmath21 our aim is to find the smallest @xmath19 such that @xmath22 . the probability that none of the first @xmath1 samples appear in the shaded region is @xmath23 next we observe that if the entries in our @xmath24 grid do not define a permutation , but we instead choose a _ random _ entry in each row , then the probability ( now for @xmath25 ) becomes @xmath26 ( for simplicity we allow @xmath1 to take arbitrarily large values ) . we certainly have that @xmath27 , meaning that @xmath28 is an upper - bound on @xmath29 , and therefore on @xmath30 . thus we compute the expected value @xmath31 this is just a geometric progression , which sums to @xmath32 . thus we need to find @xmath19 such that @xmath33 clearly @xmath34 will do . thus we conclude that @xmath35 ( see algorithm [ alg : ext ] ) we would like to apply the same reasoning in the case of multiple permutations in order to compute a bound on @xmath18 . that is , we would like to consider @xmath9 _ random _ samples of the digits from @xmath36 to @xmath17 , rather than @xmath9 permutations , as random samples are easier to work with in practice . to do so , we begin with some simple corollaries regarding our previous results . we have shown that in a permutation of length @xmath17 , we expect to see a value less than or equal to @xmath37 after @xmath38 steps . there are now @xmath39 other values that are less than or equal to @xmath37 amongst the remaining @xmath40 values ; we note that @xmath41 hence we expect to see the _ next _ value less than or equal to @xmath37 in the next @xmath38 steps also . a consequence of this fact is that we not only expect to see the _ first _ value less than or equal to @xmath37 earlier in a permutation than in a random sample , but that when we sample @xmath1 elements , we expect _ more _ of them to be less than or equal to @xmath37 in a permutation than in a random sample . furthermore , when considering the _ maximum _ of @xmath9 permutations , we expect the first @xmath1 elements to contain more values less than or equal to @xmath37 than the maximum of @xmath9 random samples . ( eq . [ eq : pkm ] ) is concerned with precisely this problem . therefore , when working in a @xmath10-dimensional hypercube , we can consider @xmath9 random samples rather than @xmath9 permutations in order to obtain an upper - bound on ( eq . [ eq : runtimeexact ] ) . thus we define @xmath25 as in ( eq . [ eq : replace ] ) , and conclude that @xmath42 thus the expected value of @xmath25 is again a geometric progression , which this time sums to @xmath43 . thus we need to find @xmath44 such that @xmath45 clearly @xmath46 will do . as mentioned , each step takes @xmath47 , so the final running time is @xmath48 . to summarize , for problems decomposable into @xmath49 groups , we will need to find the index that chooses the maximal product amongst @xmath10 lists ; we have shown an upper - bound on the expected number of steps this takes , namely @xmath50
_ maximum a posteriori _ inference in graphical models is often solved via message - passing algorithms , such as the junction - tree algorithm , or loopy belief - propagation . the exact solution to this problem is well known to be exponential in the size of the model s maximal cliques after it is triangulated , while approximate inference is typically exponential in the size of the model s factors . in this paper , we take advantage of the fact that many models have maximal cliques that are larger than their constituent factors , and also of the fact that many factors consist entirely of latent variables ( i.e. , they do not depend on an observation ) . this is a common case in a wide variety of applications , including grids , trees , and ring - structured models . in such cases , we are able to decrease the exponent of complexity for message - passing by @xmath0 for both exact _ and _ approximate inference .