Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
vortices play an important role in understanding a variety of problems in physics . in the 1970s , kosterlitz and thouless constructed a detailed and complete theory of 2d systems @xcite . the kt phase transition theory predicts that vortices pair unbinding will lead to a second - order transition in 2d systems , such as superfluid films , superconductors , and xy models @xcite . in a 2d xy model , there exist meta - stable states corresponding to vortices , which are closely bound in pairs below a critical temperature . above the critical temperature , the paired vortices unbind and become free . vortices in a 2d xy model disrupt the spin alignments even at large distances and correspond to singularities of the order - parameter field @xcite . in 2d xy model , vortices are topologically stable configurations . it is found that the high temperature disorder phase with an exponential correlation is a result of the formation of vortices . the critical temperature at which the kt transition occurs is , in fact , that at which vortex generation becomes thermodynamically favorable . at temperatures below this , the system has a power law correlation . in a low temperature phase , vortices appear in a small density of tightly bound dipole pairs . with an increase in temperature , vortex pairs dissociate and become free in the disorder phase . the evolution of vortices plays an important role in the kt transition of the 2d xy model . there has been progress on the study of the defects associated with an n - component vector order parameter field , @xmath4 . for the scalar case ( n=1 ) , the defects are domain walls , which are points of the spatial dimensionality d=1 , lines for d=2 , planes for d=3 , etc . more generally , for n = d , one has point defects . this leads to vortices for n = d=2 . it is interesting to consider the appropriate form for the point defect densities when expressed in terms of the vector order parameter field . this has been carried out by halperin @xcite , and exploited by liu and mazenko @xcite ; however , their analyses are incomplete @xcite . in a 2d system , a gauge field - theoretical formalism has been developed by kleinert @xcite . furthermore , the gauge theory of topological quantum melting in a @xmath5 dimensional bose system was developed by nussinov et al , and the superfluidity and superconductivity can arise in a strict quantum field - theoretical setting @xcite . a topological field theory for topological defects has been developed by duan et al @xcite . by using the @xmath6-mapping method and topological current theory , the evolution of the topological defect that relates to singularities of the order - parameter field , such as the vortex in bec @xcite and superconductivity @xcite , was studied . in this paper , we will discuss the topological quantization and evolution of vortex in two - dimensional xy model . we introduce a scalar field , @xmath0 , through the topological current theory . by solving the scalar field , the hamiltonian of the interaction of vortices in the 2d xy model is revisited . more recently , the dynamical behavior of the 2d xy model following a quench of the temperature to below the kt critical temperature has been studied @xcite . during the quench of a dynamic 2d xy model , aging phenomena and dynamic scaling behavior become extremely important @xcite . the assumption of dynamical scaling for the predicted asymptotic growth law the characteristic length , @xmath7 @xcite , which is also characteristic of the spacing between defects . however , scaling violations were reported . according to an expansion in @xmath8 using standard field - theoretic renormalization group theory , it can be shown that the growing length is given as @xmath9 for large t , where z is the critical exponent for equilibrium critical dynamics @xcite . this standard theoretical approach also shows that the result @xmath9 is independent of the initial conditions . however , this renormalization group method does not involve the effects of topological defects , such as vortices in the 2d xy model . in the recent work , it was shown that , for the specific case of the 2d xy model , @xmath10 if free vortices are present , while @xmath11 if there are no free vortices present in the initial state @xcite . through our topological current theory for the 2d xy model , the dynamic of vortices is studied , and the branch conditions for generating , annihilating , crossing , splitting and merging of vortices are given . during the growth or annihilation of vortices , the dynamical scaling law of relevant length in the 2d xy model , which is @xmath12 , can be obtained in the neighborhood of the limit point . this indicates that vortices in the 2d xy model are the source of the scaling violations . this dynamical scaling behavior is consistent with numerical simulations @xcite and experimental results @xcite . furthermore , we have also found that during the crossing , splitting , and merging of vortices , the dynamical scaling law of relevant length in the 2d xy model , which is @xmath3 . it has been shown that why the relevant length of vortices scales with time as @xmath13 for small values of times and then deviate toward a linear dependence in nematic liquid - crystal experimental observations @xcite . however , if vortices are at rest during splitting and merging , the dynamical scaling law of relevant length is @xmath14 . moreover , it is worthwhile to note that the dynamical scaling law of vortices , which was deduced from the topological current theory , only depends on topological properties of the order parameter field . the organization of this paper is as follows . in sec . ii , we describe the 2d xy model and the @xmath6-mapping method , and topological current theory is discussed . the hamiltonian of vortices in the 2d xy model is revisited in section iii . the evolution of vortices and dynamical scaling discussed in section iv . finally , in section v , we summarize our results . the 2d xy model is a system of spins confined to rotate in the plane of the lattice . the hamiltonian of the system is give as : @xmath15 where j is the strength of the nearest - neighbor interaction . @xmath16 denotes the angle of the spin on site i with respect to arbitrary polar direction in the 2d vector space containing spins . by using the continuum limit , we can approximate @xmath17 by the first two terms @xmath18 of the taylor expansion . we can express partial derivatives through @xmath19 for the two sites i and j. this leads to the continuum hamiltonian : @xmath20 where @xmath21 is the energy of the completely aligned ground state of @xmath22 spins . kosterlitz and thouless suggested that the disordering is caused by topological defects , such as vortices in two - dimensional xy model , which are characterized by a mapping from some loop @xmath23 in real space onto the order parameter space . for two - dimensional xy model , this implies @xmath24 where @xmath25 is the winding number of the vortex . by introducing a 2d xy model , the local order parameter is defined as : @xmath26 quantized vortices are topological objects associated with topological properties of the order parameter @xmath27 . it is worth noting that the phase of the order parameter is undefined at the vortex core . in other words , vortices correspond to singularities of the order - parameter field . we define the unit vector field @xmath28 as @xmath29 where @xmath30 . from the unit vector , @xmath31 , we can construct a topological current of the order parameter field in the 2d xy model , which carries the topological information of @xmath32 : @xmath33 we will see that this current does not vanish at the zero point of @xmath32 or the singularities of the unit field @xmath31 . by using the 2d topological current theorem , eq . ( [ eq : top ] ) can be rewritten in the compact form ; @xmath34 it is the important relation between the @xmath35like topological current of vortices and the order parameter , @xmath27 , where @xmath36 is the vector jacobian of @xmath37 : @xmath38 or @xmath39 according to the implicit function theorem , the jacobian s determinant can be given as : @xmath40 the solutions of the zero point of @xmath32 can be generally expressed as : @xmath41 which represent @xmath22 zero points , @xmath42 ( l=1,2, ,n ) , or a world line of @xmath22 vertices in space - time . with the @xmath43-function theory , @xmath44 , can be expanded as : @xmath45 where the positive integer , @xmath46 , is called the hopf index of map @xmath47 . the meaning of @xmath46 is that when the point @xmath48 covers the neighborhood of the zero , @xmath49 , once the vector field , @xmath37 , covers the corresponding region for @xmath46 times . using the implicit function theorem and the definition of the vector jacobian ( eq . [ eq : jacobian ] ) , we can find the velocity of the @xmath50-th defect , @xmath51 _ { \vec{z_l } } \nonumber\\ \vec{j } \left ( \frac{\phi}{x } \right ) & = & \left [ j^1\big(\phi / x\big ) , j^2 \left ( \phi / x \right ) \right]\end{aligned}\ ] ] the spatial and temporal components of the defect current , @xmath52 , can be written as the form of the current and the density of a system of @xmath22 classical point particles moving in a @xmath53-dimensional space - time , @xmath54 @xmath55 where @xmath56 is the brouwer degree , @xmath57 it can clearly be seen that eq . ( [ eq : current ] ) shows the movement of vertices . the topological charge of vertices in the 2d xy model are conserved : @xmath58 in addition , there is a constraint of charge neutrality : @xmath59 which indicates that vortices in the 2d xy model appear in pairs . in analogy to the velocity field in a superfluid , the distortion field , @xmath60 , carries the topological information of vortices in the 2d xy model . by using eq . ( [ eq : phi ] ) and ( [ eq : na ] ) , we can prove that @xmath61 and the vorticity is given as : @xmath62 where @xmath63 are the base vectors in the cartesian coordinate system . comparing eq . ( [ eq : vor ] ) and eq . ( [ eq : top ] ) , we conclude that @xmath64 therefore , in the 2d xy model , the vorticity of distortion field , @xmath65 , can be expressed in terms of the topological current of the order parameter field . by comparing the @xmath35like topological current eq . ( [ eq : jc ] ) and eq . ( [ eq : vc ] ) , we have the important relation between vorticity and the order parameter field in the 2d xy model : @xmath66 from eq . ( [ eq : vj ] ) , we see that the vorticity of distortion field , @xmath65 , does not vanish at the zero points of @xmath27 . the location and direction of the @xmath67th vortex are determined by the @xmath67th singular point , @xmath42 , and the vector jacobian , @xmath68 , on @xmath42 , respectively . in the absence of vorticity , there are no zero values of the order parameter field , @xmath44 is zero , and eq . ( [ eq : vj ] ) becomes : @xmath69 which is the condition of irrotationality . thus , eq . ( [ eq : vj ] ) describes both the vortex - state and the irrotationality - state . to describe a collection of vortices at locations @xmath42 , we substitute eq . ( [ eq : rho ] ) into eq . ( [ eq : vc ] ) , leading to : @xmath70 it can be see that eq . ( [ eq : ts ] ) represents n vortices that are charged with the topological charge , @xmath25 . @xmath25 is @xmath71 , which describes the inner topological structure of the vortex . to find the solution for eq . ( [ eq : ts ] ) , we can introduce a harmonious scalar field , @xmath0 defining by : @xmath72 in the absence of a vortex in the 2d xy model , we can see that the condition , @xmath73 , leads to @xmath74 . thus , the 2d distortion can be written as , @xmath75 , which contains two parts : the first part , @xmath76 , describe spin waves , while the second part is associated with vortices . by using 2d topological current theorem , it can be shown that : @xmath77 the scalar field , @xmath0 , behaves like the potential due to a set of charged particles . the solution of eq . ( [ eq : sd ] ) is given as : @xmath78 the continuum hamiltonian of the vortices in eq . ( [ eq : hami02 ] ) can be rewritten as : @xmath79 where @xmath80 is the core energy of the vortex . substituting eq . ( [ eq : sd ] ) and ( [ eq : ps ] ) into the hamiltonian ( [ eq : h2 ] ) , one can obtain : @xmath81 which is identical to the hamiltonian of a 2d coulomb gas with point charges of charge @xmath82 . the constraint given by eq . ( [ eq : cn ] ) is thus the constraint of charge neutrality . the zero point of the order parameter field , @xmath27 , which is associated with the locations of the core of vortices , plays an important role in describing the evolution of vortices in the 2d xy model . if the jacobian determinant , @xmath83 , we will have the isolated solution of the zeros of the order parameter field in @xmath5 dimensional space - time . but , when @xmath84 , the above results will change in some way and will lead to the branch process of vortices . we denote one of the vectors jacobians at the zero points as @xmath85 . according to the values of the vector jacobian at the zero points of the order parameter , there are usually two kinds of branch points , namely , the limit points and bifurcation points @xcite . each kind corresponds to different cases of branch processes . we explore what will happen to vortices at the limit point @xmath85 . the limit points are determined by @xmath86 considering the condition given by eq . ( [ eq : condition01 ] ) and making use of the implicit function theorem , the solution of the zero points of @xmath87 in the neighborhood of the point ( @xmath88 ) is given as : @xmath89 where @xmath90 . in this case , one can see that : @xmath91 or @xmath92 the taylor expansion of @xmath93 at the limit points @xmath85 is given as : @xmath94 which is a parabola in the x - t plane . from this equation , we can obtain two solutions @xmath95 and @xmath96 , which give two branch solutions ( world lines of the vortices ) . if @xmath97 we have the branch solutions for @xmath98 which are related to the origin of a dipole pair . otherwise , we have the branch solutions for @xmath99 , which related to the annihilation of a dipole pair ( fig . [ fig : b ] ) . -dimensional space - time . two world lines intersect with different directions at the bifurcation point ; i.e. , two vortices encounter at the bifurcation point.,scaledwidth=50.0% ] since the topological current is identically conserved , the topological charges of these two generated or annihilated vortices must be opposite of one another at the limit points , such as : @xmath100 this indicates that the vortices always generate and annihilate in pairs . from eq . ( [ eq : mv ] ) , one also obtains that the velocity of the vortices is infinite when they are annihilating , which agrees with the result introduced by bray @xcite . furthermore , a new result can be obtained that shows that the velocity of the vortices is infinite when they are generating , which is gained only from the topology of the order parameter field . for a limit point , it is required that , @xmath101 . a bifurcation point , on the other hand , must satisfy a more complex condition . this case will be discussed in the following . now , let us turn to consider the case in which the restrictions on the zero point @xmath85 are given by : @xmath102 which imply an important fact that the function relationship between t and x or y is not unique in the neighborhood of the bifurcation point @xmath85 . this fact is easily seen from @xmath103 which , under eq . ( [ eq : con ] ) , directly shows the indefiniteness of the direction of the integral curve of eq . ( [ eq : con2 ] ) at @xmath85 . for this reason , the point @xmath85 is called a bifurcation point of the orientation order parameter . as we know , at the bifurcation point , @xmath85 , the rank of the jacobian matrix , @xmath104 $ ] , is @xmath105 in the 2d vector order parameter . with the aim of finding the different directions of all branch curves at the bifurcation point , we assume : @xmath106 from the implicit function theorem , there is one function relation : @xmath107 according to the @xmath6-mapping theory , the taylor expansion of the solution of the zeros of the order parameter field in the neighborhood of @xmath85 can be expressed as : @xcite @xmath108 which leads to @xmath109 and @xmath110 where a , b , and c are constants determined by the order parameter . the solutions of eq . ( [ eq : two01 ] ) or eq . ( [ eq : two02 ] ) give different directions for the branch curves ( world line of the vortices ) at the bifurcation point . there are four possible cases which demonstrate the physical meaning of the bifurcation points . ) . ( a ) one world line resolves into three world lines ; i.e. , one vortex resolves into three vortices at the bifurcation point . ( b ) three world lines merge into one world line ; i.e. , three vortices merge into one vortex at the bifurcation point.,scaledwidth=45.0% ] case 1 @xmath111 : for @xmath112 , from eq . ( [ eq : two01 ] ) we get two different motion directions of the core of the vortex given as : @xmath113 which is shown in fig . [ fig : c ] , where two world lines of two vortices intersect with different directions at the bifurcation point . this shows that two vortices encounter and then depart at the bifurcation point . case 2 @xmath111 : for @xmath114 , form eq . ( [ eq : two01 ] ) , we obtain only one motion direction of the core of the vortex given as : @xmath115 which includes three important cases ( fig . [ fig : d ] ) . these three cases are : ( i ) two world lines tangentially contact ; i.e. , two vortices tangentially encounter at the bifurcation point . ( ii ) two world lines merge into one world line ; i.e. , two vortices merge into one vortex at the bifurcation point . ( iii ) one world line resolves into two world lines ; i.e. , one vortex splits into two vortices at the bifurcation point . case 3 @xmath116 : for @xmath114 , from eq.([eq : two01 ] ) we have : @xmath117 there are two important cases ( fig . [ fig : e ] ) : ( i ) one world line resolves into three world lines ; i.e. , one vortex splits into three vortices at the bifurcation point . ( ii ) three world line merge into one world line ; i.e. , three vortices merge into one vortex at the bifurcation point . case 4 ( a = c=0 ) : equations ( [ eq : two01 ] ) and ( [ eq : two02 ] ) give respectively : @xmath118 this case shows that two world lines intersect normally at the bifurcation point , which is similar to case 3 . it is no surprise that both parts of eq . ( [ eq : case4 ] ) are correct because they give the slope coefficients of two different curves at the same point @xmath85 . the remaining components of @xmath119 can be calculated from a function relation of eq . ( [ eq : im ] ) : @xmath120 the above solutions reveal the evolution of the vortices . the topological structure of the vortices is detailed in the neighborhood of the bifurcation points of the order parameter field . besides the encountering of vortices , i.e. , two vortices encountered at and then depart from the bifurcation point along different branch curves ( fig . [ fig : c ] and fig . [ fig : d](a ) ) , splitting and merging of vortices are also included . when multicharged vortices pass the bifurcation point , it may split into several vortices along different branch curves ( fig . [ fig : d](c ) , fig . [ fig : e](a ) ) . on the other hand , several vortices can merge into one vortex at the bifurcation point ( fig . [ fig : d](b ) , fig . [ fig : e](c ) ) . as before , since the topological current of the vortices is identically conserved , the sum of topological charges of the final vortices must be equal to that of the initial vortices at the bifurcation point , which is given as ( for fixed l ) : @xmath121 this indicates that vortices with a higher value of burgers vector can evolve to the lower value of burgers vector , or that vortices with a lower value of burgers vector can evolve to a the higher value of burgers vector through the bifurcation process . furthermore , we see that the generation , annihilation , and bifurcation of vortices are not gradual changes , but that they start at a critical value of a parameter , i.e. , a sudden change . it is important to note that further bifurcations are possible during the evolution of vortices besides the cases studied in this work . it is necessary need to assume that the terms in the taylor series considered above vanish and that the expansion of the field is dominated by higher order terms . in the neighborhood of the limit point , we denote the scale length @xmath122 . the growth velocity or annihilation velocity of vortices is @xmath123 @xcite . from eq . [ eq : mv ] ) , one can obtain the scaling law as : @xmath124 it can be seen that @xmath125 . in the 2d xy model , the growth or annihilation is parameterized in terms of a relevant characteristic length @xmath126 , which is also characteristic of the mean vortex - antivortex separation distance . from eq . ( [ scale2 ] ) , we find that the relevant length , @xmath126 , obeys the following : @xmath127 equation ( [ eq : scaling01 ] ) is the dynamic scaling law of the vortex - antivortex pairs . the low temperature equilibrium phase has essentially no vortex pairs , and @xmath126 is infinite . the relationship between the relevant length and the metastable vortex density below the critical temperature is given as : @xmath128 then the number of vortices that satisfies the power law @xmath129 . the dynamic scaling law of the vortex - antivortex pairs is consistent with renormalization group theory and results from recent numerical simulations using langevin equations and monte carlo methods @xcite . the specially prepared nematic liquid - crystal system @xcite , which is developed , exhibits 2d xy behavior and can be utilized to obtains the dynamical scaling behavior of vortices ( fig . [ fig : f ] ) . this result is consistent with the expansion in @xmath8 using standard field - theoretic renormalization group theory . this standard theory shows that the growing length , @xmath9 for large @xmath130 , where @xmath131 is the critical exponent for equilibrium critical dynamics @xcite . this standard theoretical approach also shows that the result @xmath9 is independent of the initial conditions . however , this renormalization group method does not involve the effects of topological defects , such as vortices in the 2d xy model . initially and is suddenly reduced to @xmath132 ( @xmath133 ) and @xmath134 ( @xmath135 ) . the experimental results , which are obtained by pargellis et al @xcite , were averaged over nine experimental runs ( @xmath136 ) . solid lines are the expected @xmath137 scaling . , scaledwidth=50.0% ] from the assumptions of dynamical scaling law , it has been predicted that the asymptotic growth law of the characteristic length is given by @xmath10 @xcite ; however , scaling violations were reported . in a recent work , it was shown that the specific case of the 2d xy model is given as @xmath7 if free vortices are present , while @xmath138 if there are no free vortices present in the initial state @xcite . since the topological current is identically conserved , there are no free vortices during the growth or annihilation process . in the neighborhood of the bifurcation point , we denote scale length @xmath139 . from eqs . ( [ eq : case1])-([eq : case3 ] ) , we can then obtain the growth or annihilation velocity of the vortices , which is given as @xmath140 the approximation asymptotic relation of is then @xmath141 it can be seen that this is the reason why the relevant length of the vortices scales with time as @xmath13 for small values of @xmath130 and then deviates towards a linear dependence in nematic liquid - crystal experimental observations @xcite . from eq . ( [ eq : case4 ] ) , one can obtain @xmath142 it can be seen that that vortices are relatively at rest when @xmath143 . during the growth or annihilation of vortices , the dynamical scaling law of the relevant length in the 2d xy model is given as @xmath12 . this indicates that the vortices in the 2d xy model are the source of the scaling violations . during the crossing , splitting and merging of vortices , the dynamical scaling law of relevant length in the 2d xy model is given as @xmath3 . however , if vortices are at rest during merging or splitting , the dynamical scaling law of relevant length is @xmath144 . moreover , it is worth noting that the dynamical scaling law of vortices depends only on topological properties of the order parameter field . in summary , we have studied the inner structure and evolution of vortices in the 2d xy model by making using of the @xmath6-mapping topological current theory . a scalar field , @xmath0 , is introduced through the topological current theory . by solving the scalar field , the interaction energy of vortices in the 2d xy model is revisited . by using the @xmath6-mapping current theory , the densities of vortices in terms of the order parameter field in the 2d xy model are obtained directly from the definition of the topological charges of the vortices . the inner topological structure of the charge of the vortices , which is characterized by the hopf index and the brouwer degree , was obtained . by using the 2d topological current theorem , the hamiltonian of the interaction vortices in the 2d xy model was revisited . this result is identical to the hamiltonian of a 2d coulomb gas with point charges of charge @xmath82 . furthermore , we have studied the evolution of vortices in the 2d xy model and concluded that there are crucial cases of branch processes in the evolution of vortices when @xmath84 and @xmath145 , i.e. , @xmath56 is indefinite . it is important to note that according to the landau - ginzburg approach , there is instability in the 2d xy model at the zero point of the order parameter fields in the low temperature phase . this indicates that vortices are unstable in the low temperature phase . in the high temperature phase , the zero point of the order parameter field is stable . due to the branch condition , the vortices generate or annihilate at the limit points and encounter , split , or merge at the bifurcation points of the order parameter field . this result also indicates that the velocity of the vortices is infinite when they are being annihilated or generated , which is obtained only from the topological properties of the order parameter field in the 2d xy model . the scaling law of relevant length , which is given as @xmath146 , can be obtained in the neighborhood of the bifurcation point during the growth or annihilation of vortices . during the crossing , splitting and merging of vortices , the dynamical scaling law of relevant length is @xmath3 . however , if vortices are at rest during merging or splitting , the dynamical scaling law of relevant length is @xmath147 . the dynamical scaling law of vortices only depends on topological properties of the order parameter field . we study a two - component vector order parameter , @xmath148 , over the base manifold , m ( in this paper @xmath149 ) : @xmath150 quantized defects are topological objects associated with topological properties of the order parameter @xmath27 . it is worth noting that the phase of the order parameter is undefined at the vortex core . in other words , vortices correspond to singularities of the order - parameter field . let us define the unit vector field , @xmath28 , as : @xmath151 where , @xmath30 . from the unit vector , @xmath31 , we can construct a topological current of the order parameter field in the 2d xy model , which carries the topological information of @xmath32 : @xmath152 it can be seen that this current does not vanish only at the zero point of @xmath32 , or the singularities of the unit field @xmath31 . obviously , the current given by eq.([eq : atop ] ) is identically conserved : @xmath153 suppose there is a vortex located at @xmath154 . the topological charge of the defect is defined by the gauss map , @xmath155 : @xmath156 using stokes theorem in the exterior differential form , one can deduce that : @xmath157 by using the @xmath6-mapping method , we can prove that the topological current given by eq . ( [ eq : atop ] ) is a @xmath43-like current : @xmath158 from the above formula , it can be seen that the topological current does not vanish only at the zero points of the vector field @xmath32 . substituting eq . ( [ eq : ana ] ) into eq . ( [ eq : atop ] ) and considering that : @xmath159 we have @xmath160 if we define the jacobian as : @xmath161 and , by virtue of the laplacian relation in @xmath6 space @xcite @xmath162 where @xmath163 is the two - dimensional laplacian operator in @xmath6 space , we obtain a @xmath43-function like current . this current is given by : @xmath164 from the above formula , one can see that the topological current does not vanish only at the zero points of the vector field @xmath32 . therefore , it is essential to investigate the solutions of @xmath165 . we would like to thank y. x. liu , t. zhu , and y. s. duan for their helpful discussion . y.c . was supported by the srf for rocs , sem , and by the fundamental research fund for physics and mathematics of lanzhou university . n. d. mermin , phys . rev * 176 * , 250 ( 1968 ) ; n. d. mermin and h. wagner , phys . rev . lett . * 22 * , 1133 ( 1966 ) ; v. l. berezinskii , sov jetp , * 32 * , 493 ( 1970 ) ; h. e. stanley and t. a. kaplan , phys . lett * 17 * , 913 ( 1966 ) ; j. m. kosterlitz and d. j. thouless , j. phys . c * 6 * , 1181 ( 1973 ) . k. j. strandburg , rev . phys . * 60 * , 161 ( 1988 ) . d. r. nelson , _ defects and geometry in condensed matter physics _ , ( cambridge university press , cambridge , 2002 ) . f. liu and g. f. mazenko , phys . b * 46 * , 5963 ( 1992 ) . y. s. duan , h. zhang , and l. b. fu , phys . e * 59 * , 528 ( 1999 ) . h. kleinert , _ gauge theory in condensed matter _ , ( world scientific , singapore , 1989 ) . j. zaanen , z. nussinov and s. i. mukhin , ann . phys(new york ) * 310 * , 181(2004 ) ; v. cvetkovic , z. nussinov , s. mukhin , and j. zaanen , europhys . lett * 81 * , 27001 ( 2007 ) . y. s. duan and s. l. zhang , int . * 29 * , 1593 ( 1991 ) ; g. h. yang and y. s. duan , int . j. theor . phys . * 37 * , 2371 ( 1998 ) ; y. s. duan and h. zhang , phys . e * 60 * , 2568 ( 1999 ) . y. s. duan and h. zhang , eur . j. d * 5 * , 47 ( 1999 ) . y. s. duan , h. zhang , and s. li , phys . rev . b * 58 * , 125 ( 1998 ) . a. j. bray , adv . phys . * 43 * , 357 ( 1994 ) . b. zheng , int . b * 12 * , 1419 ( 1998 ) . a. j. bary and a. d. rutenberg , phys . e * 49 * , r27 ( 1994 ) . h. k. janssen , b. schaub , and b. schmittmann , z. phys.b * 73 * , 539 ( 1989 ) . a. j. bray , a. j. briant , and d. k. jervis . lett * 84 * , 1503 ( 2000 ) . r. loft and t. a. degrand , phys . b * 35 * , 8528 ( 1987 ) . a. n. pargellis , s. green , and b. yurke , phys . e * 49 * , 4250 ( 1994 ) . a. n. pargellis , p. finn , j. w. goodby , p. panizza , b. yurke and p. e. cladis , phys . a * 46 * , 7765 ( 1992 ) . m. kubicek and m. marek , _ computational methods in bifurcation theory and dissipative structures _ , ( springer , new york , 1983 ) .	by using topological current theory we study the inner topological structure of vortices a two - dimensional ( 2d ) xy model and find the topological current relating to the order parameter field . a scalar field , @xmath0 , is introduced through the topological current theory . by solving the scalar field , the interaction energy of vortices in a 2d xy model is revisited . we study the dynamical evolution of vortices and present the branch conditions for generating , annihilating , crossing , splitting and merging of vortices . during the growth or annihilation of vortices , the dynamical scaling law of relevant length in a 2d xy model , @xmath1 , is obtained in the neighborhood of the limit point , given the dynamic exponent @xmath2 . this dynamical scaling behavior is consistent with renormalization group theory , numerical simulations , and experimental results . furthermore , it is found that during the crossing , splitting and merging of vortices , the dynamical scaling law of relevant length is @xmath3 . however , if vortices are at rest during splitting or merging , the dynamical scaling law of relevant length is a constat .
one of the most fascinating aspects of the physics of ultracold gases @xcite is the ability , in appropriate situations , to describe the interaction between non identical fermionic atoms in terms of a single parameter , namely the scattering length @xmath2 for two different atoms . for example in fermionic gases involving only atoms in the two lowest hyperfine states of a single element , such as @xmath1li or @xmath0k , the interaction is fully described by the scattering length between atoms belonging to these two different hyperfine states @xcite . moreover in the presence of a feshbach resonance this scattering length can be experimentally modified almost at will by merely changing the applied magnetic field . the case of wide feshbach resonances is particularly convenient since all the physics related to the origin of the resonance , namely the existence of a closed channel , is irrelevant for most purposes . this has allowed , in the case of two fermionic atomic species , the experimental realization of the bec - bcs crossover where the scattering length goes from small negative values to small positive values through the resonance , where it diverges . for small negative values of the scattering length two different atoms have a weak effective attraction , which for high enough density and small enough temperature , leads to the formation of cooper pairs analogous to the ones appearing in superconductors and corresponding superfluid properties . on the other hand for small positive values of the scattering length @xmath3 , two fermionic atoms of different species form a bound state resulting in the appearance of molecules , or dimers , with binding energy @xmath4 . these dimers behave statistically as bosons , and again for high enough density and small enough temperature a bose - einstein condensate forms with superfluid properties . the bec - bcs crossover occurs when one goes continuously from the bose condensate to the bcs superfluid by merely varying the scattering length through the change of magnetic field . the simplicity of the description of the interaction extends to the bec limit , when one considers dimer - dimer interaction , while this is a priori a more complex situation since dimers have an internal structure . nevertheless the bec limit is quite simple since the composite bosons one deals with are dilute . in this case again the interaction between these bosons is fully characterized by their scattering length @xmath5 . this is because in the low temperature regime corresponding to the existence of the bose - einstein condensate , the kinetic energies of the dimers will be quite small compared to their binding energy , and the structure of these composite bosons will be irrelevant . this scattering length @xmath5 has been obtained @xcite , in terms of the scattering length @xmath2 for fermions , as @xmath6 when the different fermions have equal masses . on the other hand when one goes away from the bec limit toward the unitarity limit , it is no longer true that the composite nature of the bosons is irrelevant . one might accordingly expect a more complex physics for this molecular gas . remarkably the @xmath7 equation of state on this bec side is nevertheless very well described @xcite in a wide domain of density by the lee - huang - yang @xcite equation of state for these composite bosons @xcite , making again merely use of the dimer - dimer scattering length @xmath5 between these bosons . this is as if these bosons were elementary . this is somewhat surprising . more precisely , assuming for example two fermions species with equal densities @xmath8 , a typical fermion wavevector is of order @xmath9 , and for @xmath10 the typical kinetic energy of a fermion is now of the same order as the dimer binding energy which is @xmath11 with @xmath12 being the reduced mass . in this case one has to consider the whole energy dependent scattering amplitude and not only its zero energy limit given by the scattering length . this is valid for dimer - dimer scattering as well as for single fermion - dimer scattering . in this last case these single fermions may arise because of thermal breaking of dimers , or because we consider unbalanced fermionic mixtures . actually for @xmath10 these single fermions arise also because we may no longer consider isolated dimers since they are too close , and we have actually to deal with a complicated many - body problem . however our point is that even the simplest minded description has to consider the energy dependence of scattering processes . actually it is possible to understand qualitatively the wide range of validity of the lee - huang - yang equation of state in the following way . one assumes first that only the s - wave scattering properties are relevant , since higher partial wave give a quite small contribution at low energy . then one supposes that the s - wave phase shift @xmath13 stays small enough so that its low energy expansion @xmath14 remains valid , where @xmath15 is the scattering length relevant for the considered process , dimer - dimer or fermion - dimer scattering . in this case the standard low energy expression for the scattering amplitude @xmath16 remains valid . this is what happens when the different fermions have equal masses . in this case the scattering amplitude has a fairly weak variation when @xmath17 is going from zero to the maximum value corresponding to dimer breaking , so the situation is not much different from evaluating the scattering amplitude at @xmath18 . the purpose of the present paper is to show that this simplicity is linked to the fact that the two atomic species have equal masses @xmath19 ( we follow the standard use of referring to the two different atomic species as @xmath20 and @xmath21 atoms , even if there is no relation between this notation and the physical spin of these atoms ) . it disappears when the masses are quite different . in this case , even though the elementary scattering process between different fermions is very simple to describe , nevertheless the mere complexity due to the existence of a dimer leads to complexity in the involved scattering processes . in the present paper we will consider only the simplest situation which displays such a feature . we will deal with the fermion - dimer scattering , not the dimer - dimer scattering . the lonely fermion has an @xmath20 spin and scatters with a @xmath20-@xmath21 dimer . we will furthermore assume that the available kinetic energy is small enough so the dimer is not broken by the collision . in other words we will deal only with elastic scattering . in a first part of the paper we will consider s - wave scattering , since it is the dominant one at very low energy and is anyway expected to be an essential component of the scattering process . this is the natural extension of the case where the scattering is fully described by the scattering length . actually , although we have performed numerical calculations for any value of the mass ratio @xmath22 , we deal extensively with the limit where this mass ratio is large . indeed in this case we will be able to obtain an analytical solution of the problem , which has the advantage of displaying quite explicitly the behaviour resulting from the mass difference between the two fermion species . naturally since we have already explored in a preceding work @xcite the behaviour of the scattering length in this limit , this paper can also be seen as a natural extension of our earlier work . but it has the major interest of displaying a strong structure in the s - wave scattering amplitude , which is actually essentially linked to the increase of the scattering length with the mass ratio @xcite . then in the second part of our paper we will turn to p - wave scattering . although it does not contribute in the zero energy limit , this partial wave turns out to be quite important . this is physically linked to the appearance of new bound states which appear at the level of the three - body problem when the mass ratio increases . these bound states have been studied in great details by kartavtsev and malykh @xcite , for the present situation of two identical fermions with mass @xmath23 and a different fermion with mass @xmath24 . they are not present in the s - wave channel and appear only for angular momenta @xmath25 . among them are the well - known efimov @xcite states , which are quite remarkable for their spectrum structure with binding energy going up to infinity when the range of the interaction goes to zero . for @xmath26 the efimov states appear only @xcite for a mass ratio @xmath27 . however there are already bound states for smaller mass ratios , which are not efimov states . a first bound state @xcite appears for mass ratio @xmath28 . actually very important contributions in the fermion - dimer scattering properties arise even for quite lower mass ratios . the basic reason is that , even before the mass ratio threshold for bound states is reached , virtual bound states are already present . for nonzero angular momenta , in particular @xmath29 , the centrifugal barrier acts to inhibit their decay and provide them with fairly long lifetime . hence they are not much different from real bound states . moreover , since their energy is positive , they give rise to resonances , which are strong when they are located at low energy . in particular the p - wave contribution becomes very strong , due to quasi - resonance , when the mass ratio approaches @xmath30 . however we find that these effects are already important for the mass ratio @xmath31 corresponding to the mixtures of @xmath1li and @xmath0k , which is probably presently the most investigated experimentally@xcite and which is of very high current interest @xcite . actually this point has already been noted , for the specific case of the @xmath1li-@xmath0k mixtures , by levinsen _ et al _ @xcite . this was mostly done for the case of narrow feshbach resonances , which are by far the most frequently found in these mixtures . they were also mostly interested in lower dimension physics . here we will consider only the case of wide feshbach resonance in 3d , because of its theoretical interest due to its simplicity . in the following we will assume that we are near such a feshbach resonance and ignore any stability problem . fortunately such a feshbach resonance , which is fairly broad and reasonably stable , has been identified experimentally near @xmath32 g @xcite in @xmath1li-@xmath0k mixtures . we will consider in detail the mass ratio dependence of the p - wave contribution . this is not only of theoretical interest since we will show that , by making use of optical lattices , it is possible to vary experimentally the effective mass ratio of @xmath1li-@xmath0k mixtures . this would allow to study experimentally the effect of the p - wave contribution when it is quite strong , or conversely to eliminate it to a large extent which could be in particular be quite useful if this contribution turned out to be experimentally detrimental . we note indeed that these virtual bound states could play a very important role in the three - body decay and in the stability of these mixtures , since they correspond to physical situations where three atoms stay close together for a fairly long time , increasing in this way the probability of three - body decay processes . in this way optical lattices could also be used to study and possibly remove instability problems . in this p - wave study we will consider mass ratios below the value @xmath30 , corresponding to the appearance of the first p - wave bound state , since clearly beyond this threshold the physical situation will get more complicated . this justifies also that we consider only the s - wave and p - wave components of the full scattering amplitude , since the higher angular momentum components are expected to give very small contributions to the total scattering in the regime we investigate . in practice the next section [ swave ] is devoted to the study of the s - wave contribution to the scattering amplitude . but in the first subsection we will present the basic general equations which will be used also for the p - wave contribution . then in the rest of this next section [ swave ] we will study analytically in details the case where the mass ratio @xmath33 is very large . we will conclude this section by considering numerically cases where this mass ratio corresponds to situations which can be reached experimentally . then in the last section [ pwav ] before our conclusion we study numerically the p - wave contribution . in the same way as in our earlier work @xcite we use the integral equation formulation of this dimer - fermion scattering problem first worked out by skorniakov and ter - martirosian@xcite . it is convenient to derive it rapidly in the case where the fermion masses are different @xcite by writing directly @xcite the integral equation satisfied by the full dimer - fermion scattering vertex @xmath34 . here @xmath35 is the ( conserved ) momentum - energy of the particles @xmath36 , with @xmath37 the total momentum and @xmath38 the total energy of incoming particles . @xmath39 is the momentum - energy of the incoming @xmath20 fermion and @xmath40 its outgoing value . since we want the dimer - fermion scattering amplitude , we can take the center of mass referential for which @xmath41 . if we call @xmath42 the outgoing momentum of the @xmath20 fermion , i.e. @xmath43 , the total energy @xmath38 is the sum of the binding energy of the dimer and of the kinetic energies of the outgoing particles : @xmath44 where @xmath45 is the dimer reduced mass , @xmath46 is the atom - dimer reduced mass @xmath47 . here @xmath48 is the mass of the dimer . the integral equation satisfied by @xmath34 is @xcite : @xmath49 where @xmath50 , with @xmath51 . here @xmath52 is the dimer propagator : @xmath53 while @xmath54 are the single fermion @xmath20 or @xmath21 propagators . just as in the equal mass case @xcite one can show that , in the right - hand side of eq.([3vertex ] ) , only the on - the - shell value of @xmath55 with respect to the variable @xmath56 is needed . hence we may restrict ourselves to consider only this on - the - shell value everywhere in eq.([3vertex ] ) . finally we have also to take the on - the - shell value for @xmath57 to obtain the scattering amplitude , which means @xmath58 . we are thus led to consider more specifically @xmath59 and the scattering amplitude will be obtained by taking @xmath60 . hence we introduce , instead of @xmath34 , a function @xmath61 which is directly related to the scattering amplitude , as we will see below @xcite : @xmath62 t_3 ( \{\textbf{k},k^2/2 m { _ { \uparrow}}\},\{\textbf{k}_0,k_0 ^ 2/2 m { _ { \uparrow}}\};\{{\bf 0},e\})\ ] ] in this part we are only interested in the s - wave component of the scattering amplitude . this component is obtained from the above @xmath63 by averaging it over the angle between the incoming momentum @xmath64 and the outgoing one @xmath42 , and we note @xmath65 the resulting angular average . noting merely @xmath66 the corresponding average on @xmath61 ( naturally @xmath66 still depends on @xmath67 ) , we see that it is simply given by eq.([defa3 ] ) where @xmath63 has been replaced by @xmath65 . obviously when we take the angular average of integral equation eq.([3vertex ] ) , we are left with an integral equation involving only @xmath65 , or equivalently @xmath68 . it is more convenient to write the resulting integral equation by making use of reduced units , obtained by setting @xmath69 , @xmath70 , @xmath71 and @xmath72 . we obtain : @xmath73 where we have used for the involved mass ratios the simpler notations @xmath74 , @xmath75 . in writing this equation we have used the fact that we consider only the case where the incoming dimer and fermion have kinetic energies low enough so that the dimer can not be broken in the scattering process . this implies that @xmath76 , or @xmath77 in our reduced units . this makes all the arguments in the logarithms positive . otherwise we should have included @xmath78 contributions in these arguments . in the case where @xmath79 one checks easily that this equation reduces to the one studied in @xcite , which gives the scattering length for the same process . it is clear that @xmath80 is a complex quantity . from eq.([eqa3bar ] ) one can write the coupled integral equations for its real and imaginary parts , which are convenient for the numerical solution of eq.([eqa3bar ] ) . in the case where the masses are equal @xmath81 one can check that the resulting equations are identical to the ones obtained by skorniakov and ter - martirosian@xcite for this specific case . finally it is convenient to consider the relation between @xmath61 and the wave function @xmath82 for the dimer - fermion scattering problem , as it is found in @xcite and @xcite . for large dimer - fermion distance @xmath83 it behaves as : @xmath84 where @xmath85 is the angle between the incoming fermion momentum @xmath64 and its outgoing value @xmath42 ( with @xmath86 ) , and @xmath87 is the scattering amplitude . its fourier transform reads : @xmath88 note that the second term has a sign opposite to the one found in @xcite , just because our definition for @xmath68 has an opposite sign . the scattering amplitude is obtained from @xmath89 ( with @xmath90 ) . when one takes the angular average of eq.([wavef ] ) , the resulting s - wave component @xmath91 of the wave function is given by : @xmath92 where @xmath13 is the s - wave phase shift , linked to s - wave component @xmath93 of the scattering amplitude by : @xmath94 let us go now to the interesting regime where the mass of the lonely fermion @xmath24 is very light ( or equivalently the mass @xmath23 of the two identical fermions is very heavy ) . in this case parameter @xmath95 is near zero and @xmath96 is near @xmath97 . in our preceding paper @xcite , where we had @xmath98 , we have carefully expanded eq.([eqa3bar ] ) near this limit . then by a fourier transform we have converted the resulting equation into a differential equation with respect to a variable which is called @xmath99 below . we have studied this last equation , matching in particular the behaviour for the small and the large values of the variable @xmath99 by appropriate consideration of the boundary conditions . however the important range , in order to obtain the scattering length , corresponds to the large @xmath99 values and the proper behaviour in this range can also be obtained from physical considerations , as we will see below . we can also proceed by continuity , requiring that in the present case the large variable @xmath100 behaviour reduces to the one we have found in the limit @xmath79 . more specifically we have seen in this preceding work that the careful expansion of the right - hand side of eq.([eqa3bar ] ) was useful only to satisfy the perfect matching with the boundary conditions for the small values of the variable @xmath99 . on the other hand only the expansion of the left - hand side of eq.([eqa3bar ] ) was necessary in order to obtain the large @xmath99 information necessary to obtain the scattering length . here we will take advantage of this fact to avoid the painfull task of expanding the right - hand side of eq.([eqa3bar ] ) and rather rely on the above arguments to obtain the required large variable @xmath99 behaviour . hence we will set @xmath95 and @xmath96 to their limiting value , namely @xmath101 and @xmath102 in the right - hand side . in the left - hand side , following our preceding work , we will also set @xmath103 because large values of @xmath104 are irrelevant in our problem . however we will naturally keep the overall factor @xmath95 in front of the left - hand side . this is the only difference with taking the full @xmath105 limit . we rewrite eq.([eqa3bar ] ) in this limit by making use , instead of @xmath106 , of the more convenient function @xmath107 defined by : @xmath108 we note that , since from eq.([eqa3bar ] ) @xmath106 is an even function of @xmath109 , @xmath107 is an odd function of @xmath109 . then eq.([eqa3bar ] ) becomes : @xmath110 where in the last equality we have made use of the odd parity of @xmath107 and , as in @xcite , we have set : @xmath111 we take now the fourier transform of this equation , introducing : @xmath112 and making use of : @xmath113 where we have restricted ourselves to the case @xmath114 . we obtain : @xmath115 since the @xmath116 solution : @xmath117 satisfies @xmath118 , it is convenient to set : @xmath119 to obtain the simpler equation : @xmath120 the fact that we find a second order differential equation is directly related to our lowest order expansion of eq.([eqa3bar ] ) in @xmath121 . going to higher order in this expansion would result in a higher order differential equation at this stage it is useful to notice the relation between @xmath122 and the scattering wavefunction eq.([waves ] ) . taking in eq.([tf ] ) the angular average over the direction of @xmath64 , performing the angular integration over @xmath83 , setting @xmath86 and making use of the definition eq.([deff ] ) , we obtain : @xmath123 -4\pi f(q)\end{aligned}\ ] ] where we have added in the right - hand side the term @xmath124 which is usely zero since @xmath125 and @xmath126 . however we have extended above the range of variation of @xmath56 to negative values and made use of the fact that the corresponding extension of @xmath127 is odd . the added term is necessary to satisfy in eq.([angav ] ) this parity property of @xmath127 . going to reduced variables and taking the fourier transform of eq.([angav ] ) for @xmath128 we obtain with eq.([f0 ] ) : @xmath129 hence from eq.([defg ] ) we have for large @xmath99 the very simple relation : @xmath130 this provides us with the boundary condition necessary to solve the second order differential equation eq.([eqg1 ] ) , which is accordingly an effective schrdinger equation for our scattering problem . except for the @xmath131 term this equation eq.([eqg1 ] ) is the same as the one we have found in our preceding work @xcite and we solve it by following the same procedure . performing the change of variable @xmath132 , implying @xmath133 , we find that @xmath134 satisfies : @xmath135 with @xmath136 . we notice again that , in the range of variable which is of interest , that is large @xmath137 , corresponding to small @xmath138 , @xmath139 is a quite slowly varying function of @xmath138 . this allows to treat it as a constant @xmath140 . as discussed in details in @xcite this approximate treatment is increasingly accurate when the mass ratio @xmath141 , i.e. @xmath142 , becomes increasingly small , which is precisely the limit we are interested in . we have found @xcite that the important range for matching the solution of eq.([eqg1 ] ) to the asymptotic behaviour , in the present case eq.([gasymp ] ) , is @xmath143 . hence we may take @xmath144 . then the further change of variable @xmath145 transforms eq.([eqgz ] ) into : @xmath146 for @xmath147 . the solution of this equation which matches , as we will see shortly , the asymptotic behaviour eq.([gasymp ] ) is @xcite the well known bessel functions @xmath148 ( or equivalently @xmath149 ) , which goes continuously to the solution @xmath150 that we have found in @xcite for the case @xmath151 . a convenient integral representation for this function is @xcite : @xmath152 from this representation one obtains the small @xmath153 behaviour : @xmath154\end{aligned}\ ] ] where @xmath155 is the euler function . from this result one recovers in particular , for @xmath156 , the known small @xmath153 asymptotic behaviour @xmath157 where @xmath158 is the euler constant . since @xmath159 we see that , for large @xmath99 , our solution has indeed the required asymptotic behaviour eq.([gasymp ] ) . this is naturally within an unimportant constant prefactor since the homogeneous differential equation eq.([eqgx ] ) does not allow to find this prefactor . but , as soon as @xmath160 is found , this prefactor is given explicitly by eq.([gasymp ] ) . comparing eq.([smalx ] ) with eq.([gasymp ] ) in order to identify the phase shift , we find easily : @xmath161=\bar{k}_0 \ln\left[\epsilon ^2 \ln ( 1/\epsilon ^2)\right]+{\rm arg}\left[i\gamma(2i\bar{k}_0)\right]\end{aligned}\ ] ] in particular in the limit @xmath162 we can make use of @xmath163 , implying @xmath164 \simeq -2\bar{k}_0c$ ] . comparing to the limiting behaviour @xmath165 we recover our result @xcite : @xmath166 for the fermion - dimer scattering length in the considered limit where the mass ratio @xmath167 is large . once the phase shift is known we obtain the scattering amplitude by merely making use of eq.([scattamp ] ) . this is obviously the simpler way . however let us briefly show that we recover this result by calculating directly @xmath168 from our solution @xmath148 of eq.([eqgx ] ) . from section [ base ] we have : @xmath169 the last equality resulting from eq.([deff ] ) . in the fourier transform of the decomposition eq.([defg ] ) we see that @xmath170 does not contribute to the right - hand side of eq.([f0 ] ) since , as we have seen in eq.([angav ] ) : @xmath171\end{aligned}\ ] ] so that , in this right - hand side , we are left only with the contribution of @xmath172 : @xmath173 we have seen that the solution of eq.([eqgx ] ) is proportional to @xmath148 . going back to the @xmath99 variable this translates into : @xmath174 where , as mentionned above , the prefactor @xmath175 can not be obtained from the homogeneous equation eq.([eqgx ] ) and has to be determined from the asymptotic behaviour eq.([gasymp ] ) for @xmath122 . from the asymptotic behaviour eq.([smalx ] ) we have : @xmath176 where @xmath160 is given by eq.([delta0 ] ) . comparing eq.([eqgr ] ) with eq.([gasymp ] ) we find : @xmath177 the fourier transform of @xmath122 is then : @xmath178 where we have made use of the odd parity of @xmath122 directly linked to the odd parity of @xmath107 . going back to the variable @xmath179 we have : @xmath180 where in the last step we have replaced the upper bound @xmath181 by @xmath182 , since in our case @xmath183 and , as it is clear from eq.([repr ] ) , @xmath148 decreases exponentially rapidly for large @xmath153 . the last integral can be performed analytically @xcite , leading to : @xmath184 since @xmath185 for @xmath186 , we find from eq.([f0bis ] ) and eq.([tffin ] ) : @xmath187\end{aligned}\ ] ] however we have from eq.([delta0 ] ) : @xmath188 taking eq.([alpha ] ) into account this leads to : @xmath189 which , when we go back to unreduced units , is identical to eq.([scattamp ] ) as it should be . this shows that our calculations and approximations in the very light mass limit are fully consistent . let us now consider the implications of our result eq.([delta0 ] ) for the phase shift and the corresponding scattering amplitude . although the euler @xmath190 function is easily obtained numerically , we have more insight if we make use of an excellent approximation . since for real @xmath153 we have : @xmath191={\rm arg}\left[ix\gamma(ix)\right]={\rm arg}\left[\gamma(1+ix)\right ] = { \rm im}\left[\ln\gamma(1+ix)\right]\end{aligned}\ ] ] we may use the excellent stirling formula @xmath192 which leads to : @xmath193 if we were to correct the small imperfection of the stirling formula for low @xmath138 by replacing @xmath194 by @xmath195 , this would allow eq.([delta01 ] ) to give the exact result @xmath196 with @xmath197 given by eq.([a3 ] ) , but the result would not be accurate for higher values of @xmath198 . our original result eq.([delta0 ] ) and its simplified form eq.([delta01 ] ) are displayed in fig.[fig1s ] for @xmath199 and @xmath200 . in this last case the two results are essentially undistinguishable in the figure . we can see that @xmath201 displays first a strong decrease from zero down to large negative values after which it increases back to small negative values , and even becomes positive in the case of large mass ratios . for large @xmath197 approximate analytical results may be obtained for the location of the minimum and its value . they both scale as @xmath202 . the essential qualitative feature of this result , namely the strong decrease of the phase shift , is directly linked to the large positive value of the scattering length . for example it is qualitatively fairly analogous to the one found for the @xmath203 phase shift of a square repulsive potential with strength @xmath204 and range @xmath95 , which is given by : @xmath205\end{aligned}\ ] ] where @xmath17 is related to the energy @xmath38 by @xmath206 and @xmath207 . this is plotted in fig.[fig2s ] for @xmath208 . we consider now the consequences of the above result for the s - wave scattering amplitude which is obtained from eq.([scattamp ] ) . we are mostly interested in its modulus @xmath209 . the result is plotted in fig . [ fig3s ] for @xmath210 . the very interesting feature is the existence of several zeros occuring when @xmath211 . the first few zeros are merely located at @xmath212 . physically these zeros imply that there is no contribution from s - wave scattering to the total cross section . we see also that the scattering amplitude is strongly peaked for zero energy , a feature which is even more important for the corresponding contribution @xmath213 to the cross section . this feature is clearly linked to the strong decrease of the phase shift when @xmath198 starts from zero . as mentionned above this is directly linked to the large value of the scattering length , and consequently this is clearly not an artefact of our approximate treatment . this behaviour implies that a rough model for the s - wave component of the effective fermion - dimer interaction could be a constant up to the energy @xmath214 followed by no interaction at all for higher relative kinetic energy . this is quite different from the simple picture of a constant interaction , whatever the energy , which is valid for example when the fermions have equal masses . these results we have obtained appear quite naturally in the small @xmath141 limit . their qualitative features are interesting enough to check that they are not just a curiosity of this simple , but not so physical , limit and that they rather stay valid qualitatively in the physical range for atomic mass ratios . we have checked that our result is in semiquantitative agreement with the exact numerical result from eq.([eqa3bar ] ) . this is displayed first for @xmath210 which is somewhat beyond realistic values . we see from fig . [ fig3s ] that our approximation is in fair agreement with the exact one for reasonably small energies ( the discrepancy at and near zero energy is just due to the difference between our approximate value for the scattering length and the exact one ) . there is however an increasing disagreement for higher values of the energy , which is not so surprising . our approximate phase shift is varying with energy somewhat less rapidly than the exact one . if we go now to the more realistic value corresponding to a hypothetical mixture of @xmath1li and @xmath215yb displaying a feshbach resonance , we obtain for this mass ratio @xmath216 the results reported in fig . [ fig4s ] . it is quite interesting to see that the results are not deeply modified compared to the low energy part of the preceding figure . both the approximate and the exact result display a single zero . just as in the preceding case , the exact zero is at a somewhat lower energy than what our approximate analytical expression gives . it is interesting that the relative location of this zero is in this order . indeed our figure shows that the approximate zero is near the dissociation threshold of the dimer , and that it is no longer present for slightly lower values of the mass ratio @xmath22 . on the other hand the exact zero is only halfway to the dissociation threshold . since the existence of this zero is a landmark of a strongly varying s - wave scattering amplitude , it is of high interest that it exists for even lower , and accordingly more realistic , mass ratio @xmath22 . we have found specifically that the exact zero is disappearing for @xmath217 . if we take the lighter fermion to be @xmath1li , this would imply for the heavier one a mass quite near the one of rubidium . unfortunately the two stable isotopes of rubidium are bosonic , but there are two fermionic ones , @xmath218rb and @xmath219rb , which have half - life of order of a month . finally if we take the case of the @xmath1li - @xmath0k mixture , which is displayed in fig . [ fig5s ] , we see that the scattering cross section ( obtained by squaring the amplitude ) near the dissociation threshold is only 5% of its zero energy value . finally it is worth noting that the disappearance of this zero in the scattering amplitude when the mass ratio is lowered is strikingly similar to the same disappearance @xcite in the function @xmath220 , which comes in the calculation of the scattering length @xmath68 . in this last case the disappearance occurs at a somewhat lower value of the mass ratio , and @xmath220 has no simple physical meaning , which makes the interpretation difficult . however one could go continuously from this function to the scattering amplitude by varying the last parameter from @xmath221 to @xmath64 . hence the two functions are closely related , and the simple physical explanation for the zero in the scattering amplitude ( it occurs when the phase shift reaches @xmath222 ) gives a corresponding understanding for @xmath220 . we turn now to the p - wave contribution to the scattering amplitude . the analysis is done with the skorniakov and ter - martirosian equations as detailed in section [ base ] . the only difference is that , instead of averaging eq.([defa3 ] ) to obtain the s - wave component , in order to obtain the p - wave component @xmath223 , where @xmath224 is the @xmath29 legendre polynomial , we have to project on this @xmath29 legendre polynomial . the resulting equation is quite similar to the s - wave one eq.([eqa3bar ] ) . this is seen quite clearly by introducing the notations ( making again use of reduced variables , but omitting systematically the bar over quantities in reduced units ) : @xmath225 and defining the kernels : @xmath226 and @xmath227 then the equations for the s - wave and p - wave components can be written : @xmath228 where @xmath229 used in section [ swave ] . since we are interested in this part in mass ratios smaller than in the preceding section [ swave ] , we provide first the results for the s - wave component for the specific mass ratios of interest . the results for @xmath230 in terms of @xmath231 are given in fig.[fig1 ] , for the mass ratios @xmath232 . we have also inserted the result for @xmath233 corresponding to the @xmath1li - @xmath0k mixture , already displayed in fig . [ fig5s ] . except for the fairly slow increase of the atom - dimer scattering length with the mass ratio found at @xmath234 , all the results are of order unity for @xmath235 and keep decreasing for increasing @xmath231 . we turn now to the p - wave component results . we first give in fig.[fig2 ] the results for mass ratios @xmath236 and @xmath237 , stopping at the value corresponding to the @xmath1li - @xmath0k mixture . for equal masses the p - wave component is always small , as it could be expected , and it can clearly be omitted when one deals with the atom - dimer scattering properties . hence the atom - dimer vertex can , to a large extent , be taken as a constant proportional to the atom - dimer scattering length . no complication is thus expected to arise from this side when many - body properties will be investigated . the situation is qualitatively similar for @xmath238 and even @xmath239 : the p - wave component is fairly small and can be neglected compared to the s - wave contribution . however , we see that this p - wave component rises steeply when the mass ratio goes from @xmath239 to @xmath237 . this last ratio is at the border of the resonant domain . this domain is now displayed in fig . [ fig3 ] , where the results for mass ratios @xmath240 and @xmath241 are plotted . we see that the hump , present for @xmath233 around @xmath242 , develops into a strong resonance at lower and lower energy when @xmath167 is increased . although we have sampled mass ratios which are very close to each other , we see that the resonance grows very rapidly with increasing mass ratio . the physical origin is clearly the development of a virtual bound state at positive energy . the lifetime of this state is directly related to the width of the resonance . since the resonance peak gets quickly narrower , the lifetime grows very rapidly with increasing mass . the link with the bound state which appears @xcite at zero energy for @xmath243 is confirmed if we look at the position of the resonance peak as a function of the mass ratio . this is shown in fig.[fig4 ] . we see that the position @xmath244 of the resonance peak extrapolates to zero for a mass ratio which is in close vicinity of @xmath30 . more precisely when the resonance peak reaches @xmath245 , it will be infinitely sharp , corresponding to an infinite value for @xmath246 . hence in this case @xmath247 will be infinitely large , which implies that the homogeneous eq.([inteq ] ) ( i.e. without the term @xmath248 in the right - hand side ) has a solution for zero energy @xmath234 . this is just stating the well - known result that bound states are solutions of the homogeneous integral equation corresponding to eq.([inteq ] ) . it is easy to find the lowest @xmath167 for which this homogeneous integral equation has a solution . we find that this occurs for @xmath249 in full agreement with kartavtsev and malykh @xcite . let us now come to the total scattering cross section @xmath250 , which is likely to be the easiest direct physical quantity to measure experimentally . since we neglect angular momenta higher than @xmath29 , we have for this cross section : @xmath251 the result is plotted in fig.[fig5 ] for @xmath236 and @xmath237 . we see that , while up to @xmath252 , the cross section is almost featureless , it display a strong resonance for @xmath233 which corresponds to the experimental value for @xmath1li - @xmath0k mixture . naturally when the mass ratio is further increased , this resonance becomes even stronger . this is shown in fig . [ fig5bis ] where we display also the case of @xmath253 . the resulting resonance dwarfs the preceding one . this strong resonance in the atom - dimer scattering cross section for @xmath1li - @xmath0k mixtures should imply a number of experimentally observable consequences . the more natural way to evidence it would be to measure the collision properties between a @xmath0k cloud and a @xmath0k -@xmath1li dimer cloud , in a way analogous to the recent experiments on @xmath1li for spin states @xcite . however this might not be so easy to perform . measuring the equation of state of a mixture is on the other hand a fundamental question which should be easier to answer . the resonance should affect in an important way the low temperature equation of state of a mixture of @xmath0k cloud and a @xmath0k -@xmath1li dimer , arising from an unbalanced mixture of @xmath0k and @xmath1li atoms on the bec side of the feshbach resonance . this will not appear at the mean field level since the angular average will cancel the effect because this is a p - wave resonance . but going to next order in @xmath0k , analogously to second order perturbation theory , there will be no cancellation . hence one should see a strong dependence on the @xmath0k density . one expects also a strong density dependence to arise when the fermi wavevector of the @xmath0k fermi sea reaches the wavevector corresponding to the resonance , since for lower wavevectors the scattering is essentially negligible . one should see something similar to a threshold effect in density . we note that the resonance should also have an important effect even on a balanced mixture of @xmath0k and @xmath1li , on the bec side , when the temperature is raised . indeed the dimers will be partially broken by thermal excitation which provides a natural source of free @xmath0k atoms , and effects analogous to the ones arising in the unbalanced mixture . similarly , going to high temperature , we expect this resonance to have a marked effect on the virial coefficients since these are systematically related to the three - body problem @xcite . naturally the resonance is also expected to affect strongly the transport properties . both the effects on the equation of state and on the transport properties should appear in the frequency and the damping of the collective modes @xcite . another way to understand the effect of this resonance on the @xmath1li - @xmath0k mixtures is to remark that the long lived virtual bound states responsible for it should behave in many respects in a way analogous to real bound states , as it has already been noted by levinsen _ et al _ @xcite . hence on the bec side the physical description should involve not only free fermions and dimers , but also the existence of trimers . obviously all the physical properties should be affected by the presence of this additional fermion species . a related point is that the existence of these trimers should affect the dimer - dimer scattering properties . in other words we have only explored the simpler fermion - dimer scattering properties , but we expect that the dimer - dimer scattering amplitude will display related energy structure , and they should not be so complicated to explore with the methods of our preceding work @xcite . finally it is worth stressing that the use of optical lattices should provide a very convenient and powerful way to explore the resonance we have pointed out . indeed the natural mass ratio of @xmath237 between @xmath0k and @xmath1li atoms happens to be just at the border of the strongly resonating mass ratio domain . a rather weak optical lattice could be used to slightly tune the effective mass of @xmath0k or @xmath1li ( by making an appropriate choice of the light frequency , only one atomic species is affected ) , producing strong modification of the resonance and hence of the physical properties of the mixture . in this way one could go into the strongly resonating domain , or even reach the threshold for the appearance of real bound states . or one could go in the other direction and basically get rid of the resonance , which would allow to prove that it is responsible for specific physical properties of the mixture . more specifically the 3d potential @xmath254 $ ] with @xmath255 provides , within second order perturbation theory , an effective @xmath256 given by @xmath257 . increasing the @xmath0k effective mass to reach the bound state ratio @xmath258 requires @xmath259 which corresponds to a fairly weak optical potential . this simple picture of effective mass modification works only if the size of the involved objects ( dimer , trimer ) is large compared to the optical wavelength @xmath260 . taking @xmath261 this should be a valid approximation in the vicinity of the feshbach resonance , where the scattering length becomes quite large . even if corrections to this simple picture are necessary , the qualitative physical trends should remain valid . in this paper we have studied the s - wave and p - wave contributions to the fermion - dimer scattering amplitude as a function of the mass ratio . when masses are equal the physical situation is quite simple . the p - wave contribution is completely negligible , and the modulus of the s - wave contribution is essentially constant . hence the energy dependence of the scattering properties is inessential , and the fermion - dimer scattering length @xmath68 is enough to fully characterize these properties . when the mass ratio is increased the situation becomes much more complex , both for s - wave and for p - wave . for the s - wave contribution we have obtained an analytical solution in the asymptotic limit of very large mass ratio . this allows to have explicitly the behaviour of the s - wave scattering amplitude . we find that it displays a large number of zeros . these are directly linked to the very rapid increase of the phase shift with energy , a zero appearing each time the phase shift crosses a multiple of @xmath222 . at low energy this rapid increase of the phase shift is directly linked to the known large value of the fermion - dimer scattering length for large mass ratio , although we find that at higher energy the phase shift saturates and goes back toward zero . in the range of mass ratio corresponding to possible experimental situations , the numerical solutions display a behaviour in qualitative agreement with the asymptotic limit , although the agreement is not quantitative which is not surprising . interestingly we find that a zero is still present in the s - wave scattering amplitude down to an experimentally reasonable mass ratio of @xmath262 . for the p - wave contribution the behaviour gets rapidly quite complex as soon as the mass ratio has a sizeable value . this is basically due to the appearance of a fermion - dimer bound state when the mass ratio reaches the value @xmath263 , and we have limited our study to mass ratios below this threshold . even below this threshold the existence of the bound state appears through the existence of virtual bound states at positive energy . these give rise to a resonance in the scattering amplitude . this resonance gets stronger and goes to lower energy when the mass ratio is increased toward the limiting value of @xmath263 , at which the resonance corresponds to a divergence at zero energy . it is very interesting that the mass ratio @xmath264 , corresponding to the @xmath0k - @xmath1li mixtures , much studied experimentally , is at the border where the p - wave resonance becomes important . roughly , below this mass ratio we find a fairly small bump in the modulus of the p - wave scattering amplitude , which is fairly unimportant in the overall scattering properties and may be omitted in a first approach . on the other hand above this mass ratio , the p - wave resonance grows very rapidly and becomes the dominant feature of the scattering . we have stressed explicitly that the use of optical lattices could allow to vary experimentally the mass ratio throughout this very interesting range of mass ratio . one could in this way disentangle the role of this p - wave resonance in the physical properties of these @xmath0k - @xmath1li mixtures , which are likely to be quite complex . in the same way one could check if these virtual bound states play an important role in the instabilities of these mixtures . for a recent review , see s. giorgini , l. p. pitaevskii and s. stringari , rev.mod.phys . * 80 * , 1215 ( 2008 ) . d. s. petrov , c. salomon , and g. v. shlyapnikov , phys . * 93 * , 090404 ( 2004 ) and phys . a * 71 * , 012708 ( 2005 ) . brodsky , a.v . klaptsov , m.yu . kagan , r. combescot and x. leyronas , j.e.t.p . letters * 82 * , 273 ( 2005 ) and phys . a * 73 * , 032724 ( 2006 ) . t. d. lee and c. n. yang , phys . rev . * 105 * , 1119 ( 1957 ) ; t. d. lee , k. huang and c. n. yang , phys . rev . * 106 * , 1135 ( 1957 ) . x. leyronas and r. combescot , phys . * 99 * , 170402 ( 2007 ) ; r. combescot and x. leyronas , phys . a * 78 * , 053621 ( 2008);f . alzetto and x. leyronas , phys . a * 81 * , 043604 ( 2010 ) . f. alzetto , r. combescot and x. leyronas , phys . a * 82 * , 062706 ( 2010 ) . m. iskin and c. a. r. s de melo , phys . a * 77 * , 013625 ( 2008 ) . m. iskin , phys . a * 81 * , 043634 ( 2010 ) . o. i. kartavtsev and a. v. malykh , j. phys . b * 40 * , 1429 ( 2007 ) and jetp letters , * 86 * , 625 ( 2007 ) . v. efimov , phys . b * 33 * , 563 ( 1970 ) ; e. braaten and h. w. hammer , phys . rep . * 428 * , 259 ( 2006 ) . v. efimov , nucl . a * 210 * , 157 ( 1973 ) . m. taglieber , a.c . voigt , t. aoki , t. w. hnsch , and k. dieckmann , phys . 100 * , 010401 ( 2008 ) . e. wille , f. m. spiegelhalder , g. kerner , d. naik , a. trenkwalder , g. hendl , f. schreck , r. grimm , t. g. tiecke , j. t. m. walraven , s. j. j. m. f. kokkelmans , e. tiesinga , and p. s. julienne , phys . lett . * 100 * , 053201 ( 2008 ) . a. ridinger , s. chaudhuri , t. salez , d. rio fernandes , n. bouloufa , o. dulieu , c. salomon and f. chevy , arxiv:1108.0618 s .- t . wu , c .- h . pao , and s .- k . yip , phys . rev . a * 74 * , 224504 ( 2006 ) ; c .- h . wu , and s .- k . yip , phys . a * 76 * , 053621 ( 2007 ) . m. m. parish , f. m.marchetti , a. lamacraft , and b. d. simons , phys . rev . lett . * 98 * , 160402 ( 2007 ) . j. von stecher , c. h. greene , and d. blume , phys . a * 76 * , 053613 ( 2007 ) ; d. blume , phys . a * 78 * , 013613 ( 2008 ) . m. a. baranov , c. lobo , and g. v. shlyapnikov , phys . a * 78 * , 033620 ( 2008 ) . g. orso , l. p. pitaevskii , and s. stringari , phys . a * 77 * , 033611 ( 2008 ) . i. bausmerth , a. recati , and s. stringari , phys . rev . a * 79 * , 043622 ( 2009 ) . r. b. diener and m. randeria , phys . rev . a * 81,033608 ( 2010 ) . a. gezerlis , s. gandolfi , k. e. schmidt , and j. carlson , phys . . lett . 103 * , 060403 ( 2009 ) . d. blume and k. m. daily , phys . 105 * , 170403 ( 2010 ) ; s. gandolfi and j. carlson , arxiv:1006.5186 j. levinsen , t. g. tiecke , j. t. m. walraven and d. s. petrov , phys . lett . * 103 * , 153202 ( 2009 ) ; j. levinsen and d. s. petrov , eur . j. d * 65 * , 67 ( 2011 ) . d. naik , a. trenkwalder , c. kohstall , f.m . spiegelhalder , m. zaccanti , g. hendl , f. schreck , r. grimm , t.m . hanna , and p.s . julienne , eur . j. d * 65 * , 55 ( 2011 ) . g. v. skorniakov and k. a. ter - martirosian , zh . . fiz . * 31 * , 775 ( 1956 ) [ sov . jetp * 4 * , 648 ( 1957 ) ] . this has been done also recently by iskin and s de melo @xcite l d. landau and e. m. lifshitz , _ quantum mechanics _ ( pergamon press , oxford , 1985 ) . i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series and products _ ( academic press , 1980 ) . a. sommer , m. ku , g. roati , and m. w. zwierlein , nature * 472 * , 201 ( 2011 ) . x. leyronas , phys . a * 84 * , 053633 ( 2011 ) .	we study near a feshbach resonance , as a function of the mass ratio , the fermion - dimer scattering amplitude in fermionic mixtures of two fermion species . when masses are equal the physical situation is known to be quite simple . we show that , when the mass ratio is increased , the situation becomes much more complex . for the s - wave contribution we obtain an analytical solution in the asymptotic limit of very large mass ratio . in this regime the s - wave scattering amplitude displays a large number of zeros , essentially linked to the known large value of the fermion - dimer scattering length in this regime . we find by an exact numerical calculation that a zero is still present for a mass ratio of 15 . for the p - wave contribution we make our study below the mass ratio of 8.17 , where a fermion - dimer bound state appears . we find that a strong p - wave resonance is present at low energy , due to a virtual bound state , in the fermion - dimer system , which is a forerunner of the real bound state . this resonance becomes prominent in the mass ratio range around the one corresponding to the @xmath0k - @xmath1li mixtures , much studied experimentally . this resonance should affect a number of physical properties . these include the equation of state of unbalanced mixtures at very low temperature but also the equation of state of balanced mixtures at moderate or high temperature . the frequency and the damping of collective modes should also provide a convenient way to evidence this resonance . finally it should be possible to modify the effective mass of one of the fermionic species by making use of an optical lattice . this would allow to study the strong dependence of the resonance as a function of the mass ratio of the two fermionic elements . in particular one could check if the virtual bound state is relevant for the instabilities of these mixtures .
phenomenological local potentials are conventionally used in nuclear physics for the description of @xmath1 interaction , nucleon - cluster or cluster - cluster interactions , etc . nevertheless , interaction between quantum particles is nonlocal in nature . the nonlocality arises from the composite structure of interacting particles , e.g. , from the quark structure of nucleons ; it appears as a result of relativistic effects or of the truncation of a model space , etc . the nonlocality can give rise to some peculiarities in scattering and/or in the structure of bound states . we shall show that a system with a nonlocal interaction can have a bound state of a special type that we shall refer to as isolated state ( is ) . is is a state with dying asymptotics like a usual bound state but is is not in correspondence with any of @xmath0-matrix poles . is can have negative or positive energy ; in the later case , is appears to be a bound state embedded in continuum ( bsec ) that have been studied both for local @xcite and nonlocal @xcite interactions . we shall present a model finite - rank separable interaction giving rise to is . exact solutions of the corresponding schrdinger equation can be easily found in the @xmath3-matrix formulation of the scattering theory @xcite@xcite . the @xmath3-matrix approach is based on the fourier - expansion of continuum spectra wave functions in terms of @xmath4 basis functions . in particular , the harmonic oscillator eigenfunctions as well as the so - called sturm - laguerre functions can be used as the basis functions in the method . the important point of this approach is the fact that the matrix of the kinetic energy operator @xmath5 is of a tridiagonal form on the oscillator or sturm - laguerre basis , i.e. , the kinetic energy matrix is a jacobi matrix ( or , shortly , @xmath3-matrix ) . the main advantage of this formalism in comparison with currently used formalisms in the scattering theory is the fact that it deals with algebraic rather than differential or integrodifferential equations . the radial wave function @xmath6 satisfies the schrdinger equation , @xmath7 where @xmath8 is the energy , @xmath9 is the angular momentum , @xmath5 is the kinetic energy operator , while an interaction in the system we describe by a separable nonlocal potential , @xmath10 of the rank @xmath11 with the harmonic oscillator formfactors , @xmath12 ^{\frac{1}{2 } } \left ( \frac{r}{r_{0 } } \right ) ^{l+1 } \exp \left ( - \frac{r^2}{2r^{2}_{0 } } \right ) l_{n}^{l+\frac{1}{2 } } \left ( \frac{r^2}{r_{0}^{2 } } \right ) . \label{(3)}\ ] ] here @xmath13 is the oscillator radius , and @xmath14 is a laguerre polynomial . in the @xmath3-matrix method , the wave function has a form of fourier series in terms of @xmath15 functions ( [ ( 3 ) ] ) , i.e. , @xmath16 the coefficients @xmath17 of the expansion ( [ ( 4 ) ] ) for @xmath18 are given by the formula , @xmath19 where @xmath20 is a momentum , and @xmath21 are eigenvectors of the infinite tridiagonal matrix of the kinetic energy @xmath22 . for calculation of @xmath23 and @xmath24 the following analytical expressions @xcite can be used , @xmath25 ^{\frac{1}{2 } } \frac{p^{l+1 } } { \gamma ( l+\frac{3}{2 } ) } \exp ( -\frac{p^2}{2 } ) \ : _ { 1}f_{1}(-n,\ , l+\frac{3}{2};\ , p^{2 } ) \ : , \label{(6a ) } \\\ ] ] @xmath26 ^{\frac{1}{2 } } \frac{(-1)^{l } } { p^{l } \gamma ( -l+\frac{1}{2 } ) } \exp ( -\frac{p^2}{2 } ) \nonumber \\ \times _ { 1}f_{1}(-n - l-\frac{1}{2},\ , -l+\frac{1}{2};\ , p^{2})\ : . \label{(6b)}\end{aligned}\ ] ] the coefficients @xmath27 for @xmath28 may be found by the formula , @xmath29 where @xmath30 @xmath31 is the transformation that diagonalizes the truncated hamiltonian matrix , @xmath32 , @xmath33 are eigenvectors and @xmath34 are the corresponding eigenvalues of this matrix , i.e. , @xmath35_{\mu\mu ' } = \delta _ { \mu\mu ' } \varepsilon _ { \mu}\;. \label{u}\ ] ] setting @xmath36 in eq.([(7 ) ] ) and using expression ( [ ( 5 ) ] ) both for @xmath37 and @xmath38 , we obtain the following formula for the phase shift @xmath39 in the partial wave characterized by the angular momentum @xmath9 , @xmath40 all the above expressions may be generalized on the coupled channel case @xcite , on the case of true few - body scattering ( @xmath41 ) @xcite , etc . the @xmath3-matrix approach allows one to analyze the wave functions and other scattering characteristics in a simple and visual manner . we shall demonstrate this feature with a simple model of is that will be proposed in the next section . let us consider the case when the hamiltonian matrix @xmath42 is of a block - diagonal form in the harmonic oscillator basis , i.e. , @xmath43 , where the submatrix @xmath44 acts in a finite - dimensional subspace of the basis functions ( [ ( 3 ) ] ) with @xmath45 . the infinite - dimensional submatrix @xmath46 acts in an orthogonal supplement to this subspace . bearing in mind the finite - rank structure of the potential ( [ ( 2 ) ] ) , we conclude that the hamiltonian matrix @xmath42 is of the form presented on fig . 1 . = 1.00 mm ( 72.34,74.33 ) ( 22.67,29.33)(29.67,30.00)[cc]@xmath46 ( 7.67,59.33)(15.00,15.00)[cc]@xmath44 ( 52.34,29.33)(1,-1)20.00 ( 48.34,29.33)(1,-1)20.33 ( 52.34,33.33)(1,-1)20.00 ( 14.01,46.66)(0,0)[cc]0 ( 37.34,67.33)(0,0)[cc]0 ( 37.34,17.33)(0,0)[cc]0 ( 66.67,42.66)(0,0)[cc]0 an infinite tridiagonal tail of the hamiltonian matrix represents the kinetic energy matrix elements , the remaining non - zero matrix elements are displayed graphically as the boxes representing submatrices @xmath44 and @xmath46 ( we include the infinite tridiagonal kinetic energy tail in the submatrix @xmath46 ) . obviously , the eigenvectors of the submatrix @xmath44 are eigenvectors of the total hamiltonian matrix @xmath47 , too . the corresponding wave functions are dying as @xmath48 similarly to the wave functions of bound states since they are linear combinations of a finite number of the functions ( [ ( 3 ) ] ) . nevertheless , the eigenstates of the submatrix @xmath44 are characterized by wave functions with unusual asymptotics @xmath49@xmath50 instead of standard ones @xmath49@xmath51 . submatrix @xmath44 may have positive or / and negative eigenvalues @xmath52 . if @xmath53 , the corresponding state appears to be bsec . if @xmath54 , the corresponding state appears to be a bound state of a special type . all eigenstates of the submatrix @xmath44 will be referred to as isolated states ( is ) , because they are isolated completely from the scattering states . the continuum spectrum states with oscillation type asymptotics of the wave function as well as the usual bound states with @xmath51type asymptotics of the wave functions , are expressed as infinite series ( [ ( 4 ) ] ) . they are , obliviously , eigenstates of the submatrix @xmath46 . the eigenvectors of the submatrix @xmath46 being the wave functions of scattering and usual bound states in the harmonic oscillator representation are , obviously , orthogonal to the eigenvectors of the submatrix @xmath44 . thus , the scattering and usual bound state wave functions have node(s ) at a small distance ( @xmath49@xmath55 ) in the presence of is(es ) and are similar to the wave function for the potential with forbidden states @xcite . the asymptotic behavior of scattering state wave functions is characterized completely by the @xmath0-matrix @xcite . so , the structure of the @xmath0-matrix is dictated by the infinite - dimensional submatrix @xmath46 and does not depend on the submatrix @xmath44 . the is energies @xmath52 , on the other hand , are dictated by the submatrix @xmath44 and do not depend on the submatrix @xmath46 . varying matrix elements @xmath56 of the submatrix @xmath44 one causes variation of the is energies @xmath52 without any variation of the @xmath0-matrix . thus , the energy of is @xmath52 is not in correspondence with the location of @xmath0-matrix poles . using symmetry properties of @xmath0-matrix as a function of complex momentum @xmath57 @xcite , it is easy to show @xcite that the energy of bsec is not in correspondence with any of @xmath0-matrix poles . an interesting new point , so far as we know never discussed in literature , is the appearance of the discrete spectrum states , i.e. , ises with negative energy @xmath58 , that are divorced from @xmath0-matrix poles . so , is being the state characterized by the asymptotically dying wave function and the energy at which the @xmath0-matrix has not a pole , can been treated as a generalization of bsec on the case of discrete spectrum states . we shall examine in more detail the formation of is in the spectrum of the quantum system with nonlocal interaction using as an example a simple analytically solvable model . the simplest realization of the situation depicted on the fig . 1 , corresponds to the case when @xmath44 is a 1@xmath591 matrix and the separable potential ( [ ( 2 ) ] ) is of the rank 2 , i.e. , @xmath60 . in this case , the appearance of is is connected with cancellation of the potential energy matrix elements @xmath61 with the kinetic energy matrix elements @xmath62 , i.e. , @xmath63 . the energy @xmath64 of the is is equal to the diagonal matrix element @xmath65 , @xmath66 . it should be mentioned , that @xmath65 can take an arbitrary value in our model . using eqs . ( [ ( 5)][(9 ) ] ) we obtain the following expression for the phase shift @xmath39 : @xmath67 ( t_{00 } - e ) + t^{2}_{01 } ( \varepsilon _ { 0 } - e ) \right\ } } { c_{0l}(p ) \left\ { \left [ v_{11}(\varepsilon _ { 0 } - e ) - \beta \right ] ( t_{00 } - e ) + t^{2}_{01 } ( \varepsilon _ { 0 } - e ) \right\ } - \frac{p[v_{11}(\varepsilon _ { 0 } -e ) - \beta ] } { \pi s_{01}(p ) } } , % \label{(tan)}\ ] ] where @xmath68 . suppose @xmath69 and @xmath70 . the evolution of the @xmath2-wave phase shift @xmath71 as @xmath72 tends to zero is represented at the fig . if @xmath73 there is a resonance at energy @xmath74 of the width @xmath75 that becomes smaller and smaller as @xmath76 tends to zero . at @xmath77 bsec arises as the resonance of zero width producing the jump of the height @xmath78 of the phase shift @xmath79 curve at the point @xmath80 . this spurious jump should be omitted resulting in the @xmath81 increase of extra @xmath78 . thus , applying the levinson theorem @xcite to the system pertaining is , one should treat it as a usual bound state . such behavior of the phase shift is typical for systems pertaining bsec that have been studied in various models @xcite@xcite . thus , our model represents an alternative simple analytical approach in the study of bsec . such behavior of @xmath82 is also typical for the so - called orthogonal scattering or for the scattering on the potential with forbidden state . as for the s - matrix , it is given by the following expression : @xmath83 ( t_{00 } - e ) + t^{2}_{01 } ( \varepsilon _ { 0 } - e ) \right\ } -\frac { p[v_{11}(\varepsilon _ { 0 } -e ) - \beta ] } { \pi s_{01}(p ) } } { c^{(+)}_{0l}(p ) \left\ { \left [ v_{11}(\varepsilon _ { 0 } - e ) - \beta \right ] ( t_{00 } - e ) + t^{2}_{01 } ( \varepsilon _ { 0 } - e ) \right\ } - \frac{p[v_{11}(\varepsilon _ { 0 } -e ) - \beta ] } { \pi s_{01}(p ) } } , \label{(10)}\ ] ] where @xmath84 . the single @xmath0-matrix pole in the non - physical sheet tends to the real energy point @xmath80 as @xmath76 tends to zero . it is inherent to situation with bsec . however , in the limit @xmath77 , the factor @xmath85 in the numerator of ( [ ( 10 ) ] ) cancels the same factor in the denominator and the singularity in the point @xmath86 in ( [ ( 10 ) ] ) disappears , i.e. , the @xmath0-matrix in the presence of bsec has no poles . nontrivial result is that if is belongs to the discrete spectrum ( i.e. , @xmath87 ) , it does nt generate @xmath0-matrix pole , too . the above results can be generalized as a following theorem . let all the eigenvalues of the truncated hamiltonian matrix @xmath88 be non - degenerate . then isolated states occur in the spectrum of the system with rank@xmath89 separable interaction ( [ ( 2 ) ] ) if and only if the truncated matrix @xmath88 and its principal minor @xmath90 of the rank @xmath91 have common eigenvalues . the number @xmath92 of the coinciding eigenvalues is equal to the number of isolated states . these eigenvalues and corresponding eigenfunctions are just the energies and the wave functions of the isolated states , respectively . it means that each is is a linear combination of the finite number of the harmonic oscillator functions @xmath93 @xmath94 and its asymptotical behavior @xmath95 is not connected with its energy . the equivalent formulation of the theorem is : the is with the energy @xmath96 exist if and only if some eigenvector @xmath97 of the matrix @xmath88 has the last component @xmath98 . the most interesting feature of the is is the fact that they are not in correspondence with @xmath0-matrix poles . this impressive property we use as a definition of is : is is the eigenstate of the hamiltonian with dying asymptotics of the wave function that do not generate @xmath0-matrix pole . the results given above can be strightforwardly generalized on the case of coupled channels problem , on the case of separable nonlocal potentials of finite rank with the sturm - laguerre formfactors , etc . as for physical applications , it should be noted that the is appears in variational calculations using the harmonic oscillator basis . it is well - known @xcite ( the dubovoy - flores theorem ) that the lowest eigenvalue @xmath99 of the matrix @xmath88 coincides with the lowest eigenvalue @xmath99 of the matrix @xmath90 if the parameter @xmath100 of the basis functions minimizes the eigenvalue @xmath101 . according to our theorem , it means that the corresponding eigenstate @xmath102 will be is . another physical application may concern the @xmath1 potential that should be nonlocal from the point of view of the quark model . a simple two - color `` toy '' model of @xmath1 interaction considered in ref . @xcite yields the separable nonlocal potential of the type described above with @xmath103 and @xmath104 while @xmath105 has a tendency to cancel @xmath106 . thus , the non - relativistic @xmath1 scattering problem arising in the quark resonating group model is close to the situation with bsec . if is is really inherent to the @xmath1-interaction , it may be used for explanation of the impressive success of the deep attractive potential with forbidden states description of @xmath1 scattering @xcite . to illustrate the usefulness of using potentials with is for the description of @xmath1 interaction , we make fit to @xmath1 scattering data by the potential of the type ( [ ( 2 ) ] ) . the singlet and triplet @xmath2-wave @xmath1 scattering phases are well reproduced by second - rank ( @xmath60 ) potential ( [ ( 2 ) ] ) with @xmath107 mev and @xmath108 mev , @xmath109 mev ; @xmath110 and @xmath111 are fixed by the condition @xmath112 . deuteron binding energy @xmath113 and deuteron rms radius @xmath114 are also reproduced by the potential . @xmath113 and the phase shifts do not depend on the parameters of interaction @xmath115 and @xmath116 . nevertheless , the triton binding energy @xmath117 depends on @xmath115 and @xmath116 . thus , varying @xmath115 and @xmath116 we managed to fit the potential simultaneously to the most important two- and three - body observables ( @xmath113 , @xmath114 , singlet and triplet @xmath0-wave @xmath1 scattering phases and @xmath117 ) without making use of three - body forces . 99 von neuman j.v . , wigner e. _ phys . z. _ * 30 * _ ( 1929 ) 465 . _ mulligan b. et.al . * c13 * _ ( 1976 ) 2131 . _ heller e.j . , yamani h.a . _ * a9 * _ ( 1974 ) 1201 . _ yamani h.a . , fishman l. _ j.math . * 16 * _ ( 1975 ) 410 . _ filippov g.f . _ yad . fiz . j. nucl . phys . ) _ * 33 * _ ( 1980 ) 928 . _ nechaev yu.i . , smirnov yu.f . _ j. nucl . phys . ) _ * 35 * _ ( 1982 ) 1385 . _ smirnov yu.f . , shirokov a.m. _ preprint itp-88 - 47p , kiev , 1988 . _ dorodnykh yu.l . , neudatchin v.g . , yudin n.p . _ j. nucl . phys . ) _ * 48 * _ ( 1988 ) 1796 . _ newton r.g . _ scattering theory of waves and particles , mc - graw - hill , new york ( 1966 ) . _ roriz l.a . , delfino a. _ canad . j. phys . _ * 67 * _ ( 1989 ) 37 . _ zakharievb.n . , suzko a.a . _ potentials and quantum scattering : direct and inverse problems , energoatomizdat , moscow ( 1985 ) . _ dubovoy m. , flores j. _ rev . * 17 * _ ( 1968 ) 289 . _ majling l. , rizek j. , pluhar z. _ bull . acad . sci . ussr , phys . ser . _ * 38 * _ ( 1974 ) 2141 . _ mazur a.i . , smirnov yu.f . , el - howary r.m . _ ukranian j. phys . _ * 32 * _ ( 1987 ) 1135 . _	we show that a quantum system with nonlocal interaction can have bound states of unusual type ( isolated states ( is ) ) . is is a bound state that do not generate a @xmath0-matrix pole . is can have positive as well as negative energy and can be treated as a generalization of bound states embedded in continuum on the case of discrete spectrum states . formation of is in the spectrum of quantum system is studied using a simple rank2 separable potential with harmonic oscillator formfactors . some physical applications are discussed , in particular , we propose separable @xmath1 potential that describes not only most important two - nucleon data ( deuteron binding energy and @xmath2-wave triplet and singlet scattering phases ) but also the trinucleon binding energy without making use of three - body forces .
a lot of experimental efforts have been devoted to searches for physics beyond the standard model ( sm ) . one of the evidences for the existence of the new physics has been found in the precise measurement of the anomalous magnetic moment of the muon ( muon @xmath0 ) . the experimental result of the muon @xmath0 reported by the brookhaven e821 experiment @xcite deviates from the sm prediction with the updates on the hadronic contributions @xcite : @xmath5 which corresponds to more than 3 sigma level deviation , where ref . @xcite is referred to for the hadronic light - by - light contribution . it is obviously too early to conclude that this is the signal of the new physics . nonetheless , it is interesting to investigate the new physics models which can explain the muon @xmath0 anomaly . the current deviation of the muon @xmath0 is comparable to the electroweak contribution of the sm . thus , the particles of the new physics should have large couplings to the muon if they are as heavy as @xmath6 , or they must be light enough to explain the anomaly if the interactions with the sm are suppressed to avoid experimental constraints . a simple model of the latter case is the hidden photon model @xcite . in presence of a u(1 ) symmetry in a hidden sector , the extra u(1 ) gauge boson , so - called the hidden photon , can couple to the sm particles through a kinetic mixing with the gauge boson of the sm hypercharge . the hidden photon contributes to the muon @xmath0 at the one - loop level similarly to the sm photon . this becomes effective if the hidden photon mass is @xmath7 and the mixing angle is @xmath8 . it is important to examine whether the hidden photon model satisfies the other experimental constraints . the severest upper bound on the hidden photon mass which can explain the muon @xmath0 anomaly is provided by searches for the hidden photon @xmath9 in the processes , @xmath10 , at the kloe experiment @xcite and @xmath11 at the apex experiment @xcite . on the other hand , the lower mass bounds have been obtained by the electron beam dump experiment at the fermilab e774 @xcite and the electron @xmath0 @xcite . in particular , the hidden photon contributes to the electron @xmath0 at the one - loop level similarly to the muon @xmath0 . recently , the sm prediction of the electron @xmath0 has been greatly improved . the tenth - order calculation of the qed contribution was completed @xcite , and the measurement of the fine structure constant was improved by the update on the measurement of the rubidium mass @xcite . it has been found that the experimental result @xcite agrees with the sm prediction @xcite : @xmath12 where the dominant uncertainties come from the experimental result of the electron @xmath0 and the determination of the fine structure constant . this constraint is improved by one order of magnitude after the analysis in ref . thus , the electron @xmath0 can provide a tighter constraint on the hidden photon parameter especially in a light mass region . in this paper , we will revisit the constraint on the hidden photon from the electron @xmath0 . are also studied in ref . @xcite . ] in addition to the direct contribution to the electron @xmath0 at the one - loop level , the sm prediction of the electron @xmath0 can be subject to the effect of the hidden photon through the determination of the fine structure constant . the fine structure constant which is determined independently of the electron @xmath0 is currently set by a precise measurement of the rubidium mass . the present best result is known to be @xcite @xmath13 , \label{eq : alpha}\ ] ] by combining several atomic parameters , where the leading uncertainty stems from the rubidium mass . since the precisions of each measurements are as good as or better than that of the electron @xmath0 , it is crucial to evaluate the hidden photon contribution to the fine structure constant for the purpose of deriving the electron @xmath0 constraint . in this paper , we will study the measurement of the fine structure constant and its correction from the hidden photon in detail . among the atomic parameters , transition frequencies of the hydrogen are one of the most precisely measured parameters . they are utilized to determine fundamental parameters such as the rydberg constant @xcite . the sm prediction of the transition frequencies has been studied well , which is compared to the experimental results at the extremely precise level . in the hidden u(1 ) model , the atomic energy levels are shifted by the hidden photon at the tree level . thus , the hidden photon parameter can be constrained tightly by the transition frequencies , independently of the electron @xmath0 @xcite . in this paper , we revisit the hidden photon contribution to the transition frequencies in addition to the lepton @xmath0 and the fine structure constant . a special care is taken to ensure that the input parameters are determined independently of the target observables such as the transition frequencies and the electron @xmath0 , in order to make the analysis self consistent . the lagrangian terms of the hidden photon model , which are relevant to the contributions to the lepton @xmath0 , are given by @xmath14 where the field / index , @xmath15 , represents the hidden photon , and @xmath16 is the field strength of the sm hypercharge . the second term in the right - handed side is a mass of the hidden photon . we assume that the hidden photon mass is induced by the higgs mechanism in the hidden sector . the last term corresponds to the kinetic mixing between the hidden and sm u(1 ) s , where the mixing @xmath17 is conventionally scaled by the weinberg angle , @xmath18 . in a low energy scale , by canonicalizing the kinetic terms and diagonalizing the mass matrix , the effective lagrangian of the hidden photon becomes @xmath19 where @xmath9 is the hidden photon in the mass eigenstate , @xmath20 is the electromagnetic current of the charged leptons , and @xmath21 is the weak neutral current . there are two types of the contributions to the lepton @xmath0 from the hidden photon . the first one is a radiative correction to the lepton photon vertex , which is evaluated at the one - loop level as @xcite @xmath22 \label{eq : lepton_g-2_exact } \\ & = & \frac{\alpha_0}{3\pi } \epsilon^2 \frac{m_\ell^2}{m_v^2 } \left [ 1 + \mathcal{o } \left ( \frac{m_\ell^2}{m_v^2 } \right ) + \mathcal{o } \left ( \frac{m_v^2}{m_z^2 } \right ) \right ] , \label{eq : lepton_g-2}\end{aligned}\ ] ] where @xmath23 is the fine structure constant of the lagrangian . the first term in eq . is the one - loop contribution of the electromagnetic current @xmath24 , and @xmath25 is from @xmath21 . in the last equation , @xmath26 appears by expanding the integrand of eq . . when the lepton is the electron , the expansion is valid for @xmath27 , while the contribution to the muon @xmath0 needs the evaluation without the expansion for @xmath28 . the second contribution to the lepton @xmath0 is through the determination of the fine structure constant . the one - loop radiative correction of the photon provides the leading contribution to the lepton @xmath0 as @xmath29 the higher order corrections are suppressed by orders of @xmath30 . here , the fine structure constant @xmath30 is defined as a parameter of the lagrangian . the determination of the numerical value of @xmath30 is based on the experimental measurements of the atomic parameters . in the literature , it is hypothesized that the sm is correct in the process of the @xmath30 determination , and the theoretical prediction of the electron @xmath0 is based on that sm hypothesis . however , it should be noted that this procedure can be affected by the hidden photon . then , the fine structure constant in the sm hypothesis , @xmath31 , does not coincide with the lagrangian parameter @xmath30 , but is corrected by the hidden photon effect as @xmath32 . consequently , the leading qed calculation of the photon radiative correction provides an extra contribution from the hidden photon apart from the sm prediction as @xmath33 in the next section , the determination of the fine structure constant will be investigated in detail . the fine structure constant , which is determined independently of the electron @xmath0 and used as an input for the theoretical calculation of the electron @xmath0 , is determined via the rydberg constant , @xmath34 where @xmath30 is the theoretically given fine structure constant , @xmath35 is the electron mass , @xmath36 is the speed of light , and @xmath37 is the planck constant . the most precise inferred value of the fine structure constant ( except for the electron @xmath0 ) is determined by the following relation @xcite @xmath38 namely , by the precise measurement of @xmath39 , @xmath40 , @xmath41 , and @xmath42 . here , @xmath43 is the relative atomic mass ( relative isotopic mass ) . in fact , the mass ratios are measured more precisely than their absolute values by experiments . on the other hand , the absolute mass has been measured very precisely for the rubidium atom . the progress of the mass measurement is the main part of the improvement of the recent result of the fine structure constant @xcite . within the sm , if these quantities are measured with an infinite precision , the right - hand side of eq . ( [ eq : alpha_exp ] ) gives rise to the theoretical value @xmath44 itself . however , in the presence of the hidden photon , the measurements of these quantities , which are implicitly based on the sm hypothesis , may lead to values which are different from their definitions : @xmath45 where @xmath46 , and @xmath42 . here , @xmath47 denotes the value of @xmath48 which is obtained when the determination is ( wrongly ) based on the sm only even in the presence of the hidden photon . in the following , we discuss effects of the hidden photon to these quantities in turn . the rydberg constant @xmath39 is determined by the precise measurements of the hydrogen transition frequencies , @xmath49 . the theoretical prediction of the atomic energy levels , which depend on the rydberg constant , has been known well in the sm , and the transition frequencies have been measured very precisely . the detailed procedure is found in ref . @xcite . in the presence of hidden photon , the energy levels of the states , @xmath50 , are represented by @xmath51 where @xmath52 denotes the principal quantum number , and the subindex , @xmath53 , is used to distinguish the atomic states . in the parenthesis , the first term is the leading contribution , which is obtained by solving the schrdinger equation of the hydrogen atom . the second term , @xmath54 , is sm corrections , which include relativistic corrections , recoil corrections , self energies , vacuum polarizations , two- and three - photon ( loop ) corrections , and finite size corrections of the nuclei . they are known well , and the uncertainties are controlled at the level of @xmath55 , where the leading error originates in the proton charge radius . see ref . @xcite for details of the theoretical evaluation . on the other hand , the hidden photon contributions are evaluated as @xmath56 where the leading term is from a tree - level correction to the static potential . the hidden photon is exchanged between the electron and the nuclei . since the correction is sufficiently small , this is evaluated by the first order perturbation theory of the hydrogen atom . higher order corrections are negligible . experimentally , the most precisely measured transition frequency is that of the @xmath57@xmath58 levels , @xmath59 , whose numerical result is found in table [ tab : numbers ] of appendix . this experimental value is compared to the theoretical prediction in eq . ( [ eq : ea_true ] ) , @xmath60 note that the prediction includes the hidden photon contributions . if it is assumed that the corrections are only from the sm , the rydberg constant is determined by @xmath61 the numerical value in the literature @xcite is based on the sm hypothesis . in the hidden photon model , @xmath62 deviates from the rydberg constant of the theory defined by eq . ( [ eq : rydberg ] ) as @xmath63 this is a source of the hidden photon contribution to the fine structure constant , which is additional to the sm evaluation of the constant in the literature . in this and next subsection we briefly review the measurements of the relative atomic masses , @xmath41 and @xmath40 , and discuss the possibility of hidden photon correction to the values . the relative atomic masses are measured by so - called the penning trap method , where charged particles are trapped in the electric and magnetic fields . masses of the charged particles are determined by using equations of motion . the relative atomic mass of a neutral atom , @xmath64 , is related with that of the @xmath52-th ionized atom , @xmath65 , by @xmath66 where @xmath67 is the total binding energy scaled by @xmath68 , namely in units of the relative atomic mass . as for the neutral atom with the atomic number @xmath69 , the total binding energy is a sum of the binding energy of the @xmath69 electrons , while it is that of the @xmath70 electrons for the @xmath52-th ionized atom , @xmath71 @xcite . the most precise inferred value of @xmath41 is found in ref . @xcite as @xmath72 . \label{eq : are}\ ] ] this is obtained from the @xmath73-factor of the electron bounded in a hydrogenic carbon ion , by using the following relation : @xmath74 where @xmath75 and @xmath76 are the electron s precession frequency and the cyclotron frequency of the ion , respectively . the ratio @xmath77 is precisely measured @xcite , which is listed in table [ tab : numbers ] . on the other hand , @xmath78 is derived from eq . ( [ eq : xandxnplus ] ) with @xmath79 by definition , where the relative binding energy is known with high precision by measurements of the carbon spectroscopy ( see table [ tab : numbers ] ) . the theoretical value of the @xmath73-factor is known as @xmath80 $ ] @xcite . with these numbers , one can reproduce the number in eq . ( [ eq : are ] ) . let us examine the hidden photon effects . first of all , the relation is independent of the hidden photon corrections as long as the binding energies are determined by experiments . next , eq . is derived by equations of motions of the cyclotron and the spin precession in the magnetic field of the penning trap . since the experiment is macroscopic , the hidden photon correction is negligible . however , the input parameter , @xmath81 , is set theoretically in eq . and can be affected by the hidden photon . the leading contribution is the same as @xmath82 in vacuum , @xmath83 the difference between @xmath84 and @xmath85 in vacuum is from the coulomb potential , which is of @xmath86 and unimportant here . thus , from eqs . ( [ eq : lepton_g-2 ] ) , ( [ eq : are_eq ] ) and ( [ eq : corr_g_e ] ) , compared to the sm hypothesis , the determination of @xmath41 is affected by the hidden photon as @xmath87 similarly to @xmath41 in the last subsection , @xmath40 is precisely measured by the penning trap . circulating a couple of ions in the same magnetic field , a ratio of the relative atomic masses is determined by measuring a cyclotron frequency ratio of the ions . from the equations of motions in the magnetic field , the relation is @xmath88 in particular , @xmath89 and @xmath90 have been measured very precisely @xcite , as listed in table [ tab : numbers ] . thus , @xmath91 is related to @xmath92 by @xmath93 the ion masses , @xmath91 and @xmath92 , are related to the neutral ones , @xmath40 and @xmath94 , respectively , by eq . ( [ eq : xandxnplus ] ) , in which the relative binding energies have been experimentally measured , as listed in table [ tab : numbers ] . on the other hand , @xmath94 is obtained by @xmath95 where the bond energy of co@xmath96 is also known ( see table [ tab : numbers ] ) . finally , @xmath97 is obtained by eq . ( [ eq : xandxnplus ] ) and precisely measured values of the cyclotron frequency ratios of @xmath98 and @xmath99 @xcite , which are summarized in table [ tab : numbers ] . combining all the numbers together with @xmath41 in eq . ( [ eq : are ] ) , we obtain @xmath100 , which is consistent with the referred value in @xcite , @xmath101 $ ] . in the above derivation , since the equations of motions are macroscopic , and the binding / bond energies are measured by the experiments , the hidden photon contributions are negligible . although the hidden photon can contribute to @xmath41 as discussed in the last subsection , the correction to @xmath40 is negligibly small , as estimated to be @xmath102 the last piece of eq . ( [ eq : alpha_exp ] ) is @xmath104 , which is a ratio of the plank constant to the mass of the rubidium . the measurement of the rubidium mass , @xmath105 , is responsible for the largest uncertainty of the fine structure constant , i.e. , the theoretical prediction of the electron @xmath0 . at present , the most precise result is based on an ingenious experiment using atomic interferometer @xcite . the basic idea is to measure the kinetic momentum / energy of the rubidium atom . the rubidium atoms are accelerated by the absorptions and the induced emissions of photons , or by so - called the bloch oscillation . then , the shift of the kinetic energy changes a frequency of the wave function of the rubidium atom . the phase shift due to the acceleration is measured by the ramsey - bord@xmath106 atomic interferometer : before the acceleration , the rubidium atoms are split into two packets by using the laser beams which are tuned to form so - called the @xmath107 pulse . by propagating these packets differently , the acceleration induces a difference of the phase of the wave functions . the relative phase , i.e. , the kinetic energy correction from the acceleration , is measured by making the two packets interfere with each other after merging them by hitting the @xmath107 pulses again . from the conservations of the energy and momentum , the relative frequency ( phase ) of the two packets , @xmath108 , is related to @xmath104 as @xmath109 where @xmath110 is a total contribution to the wave number of the rubidium packets by the bloch oscillations , and @xmath111 and @xmath112 are wave numbers of the latter @xmath107 pulse , which consists of the two laser beams with the wave number , @xmath111 and @xmath112 , in the opposite direction . from , the interference pattern of the atomic interferometer is obtained by sweeping @xmath111 and @xmath112 . since the frequencies of the laser beams are known very precisely , and the interference pattern is sensitive to the relative phase , @xmath104 is measured with high precision as @xmath113.\ ] ] the measurement of the rubidium mass is insensitive to the hidden photon effects . the relation ( [ eq : h / m ] ) is based only on the energy and momentum conservations , and the all the quantities except for @xmath104 are given in the experiment . although the precision of the ramsey - bord@xmath106 atomic interferometer relies on the ( meta- ) stability of the atoms during the propagation against spontaneous emissions , the hidden photon in interest is too weak to enhance the emission and to spoil the experiment . therefore , the measurement of @xmath104 in the hidden photon model coincides with that of the sm hypothesis as @xmath114 from eqs . ( [ eq : alpha_exp ] ) , ( [ eq : rydberg_measured ] ) , ( [ eq : are_corr ] ) , ( [ eq : arrb_corr ] ) , and ( [ eq : rb_corr ] ) , the contribution of the hidden photon to the fine structure constant becomes @xmath115 which leads to @xmath116 where eq . ( [ eq : lepton_g-2 ] ) was used in the second equality . the electron @xmath0 receives corrections from eqs . and . @xmath117 the correction through the determination of the fine structure constant is negligibly small compared to the loop contribution , and the dominant contribution is from the one - loop correction . in fig . [ fig : darkphoton ] , the constraints on the hidden photon model are displayed . the red solid line is the updated result of the bound from the electron @xmath0 , which excludes the region above the line at the 2@xmath118 level . this result is improved by one order of magnitude compared to the previous analysis , which is shown by the red dashed line in the figure @xcite . it is interesting to compare the result with the region favored by the muon @xmath0 . the 1 and 2@xmath118 regions are shown by the dark and light green bands , respectively , in the figure . the electron @xmath0 constraint now excludes a smaller mass region of the hidden photon more efficiently . in fig . [ fig : combined ] , the updated constraint is compared to the other experimental bounds @xcite . before the present analysis , the lower mass region has been excluded mainly by the beam dump experiment at e774 @xcite . as shown in the figure , the region which is favored by the muon @xmath0 becomes rejected by the electron @xmath0 . it is found that the allowed region of the muon @xmath0 can be accessed by various experiments , which are displayed by the dashed lines in fig . [ fig : combined ] @xcite . interestingly , even the test run of the hps experiment , which is planned in the near future @xcite , can cover most of the parameter space . and @xmath119 transitions are analyzed with the proton radius input by the electron proton experiment @xcite . the regions favored by the muon @xmath0 at the 1 and 2@xmath118 levels are shown by the dark and light green bands , respectively . ] is compared to the other experimental bounds @xcite . the gray regions have been already excluded by them . the green bands are favored by the muon @xmath0 ( see fig . [ fig : darkphoton ] ) . the regions enclosed by the colored dashed lines are suggested to be covered in future . see ref . @xcite for details of the experiments , where the excluded regions and future sensitivities are found . the recent update of the kloe bound @xcite is included ( see note added ) . ] among the atomic parameters , the transition frequencies are measured with extremely high precision . in fact , the accuracy is better than that of the electron @xmath0 . the transition frequencies can provide search for the hidden photon , which is independent of the electron @xmath0 . in order to compare the experimental result with the theoretical prediction , measurements of the input parameters should be independent of or at least insensitive to that of the transition frequencies . in this section , the transition frequencies are studied to search for the hidden photon in detail . a transition frequency is written by the rydberg constant and the corrections from the sm and the hidden photon as @xmath120 , \label{eq : transition}\ ] ] where the left - handed side is measured by experiments , while the quantities in the bracket are theoretically given . in particular , the sm corrections , @xmath121 , are well known for the hydrogen atom . they consist of relativistic corrections , recoil corrections , self energies , vacuum polarizations , two- and three - photon ( loop ) corrections , and finite size corrections of the proton . the detailed evaluations are summarized in @xcite . the relative uncertainty of the prediction is of @xmath55 . the leading part comes from the measurement of the proton charge radius , which will be mentioned later , and the next - to - leading one , which is of @xmath122 , is from ambiguity of the evaluation of the two - photon correction ( see discussions in @xcite ) . on the other hand , the hidden photon shifts the energy levels by @xmath123 , which is obtained in eq . . as we will discuss later , the other contributions are negligible . since the correction from the hidden photon @xmath123 is tiny , if eq . is used to constrain the hidden photon model , it is crucial to determine the rydberg constant @xmath39 with extremely high precision . although @xmath39 has been provided very precisely in codata @xcite , we should not refer to it in the current analysis . this is because the determination of the rydberg constant is based on the transition frequencies , and hence the analysis would become self inconsistent . is unclear in @xcite , in which they refer to the codata , despite that the hidden photon contributions to the transition frequencies are studied . ] on the other hand , if the rydberg constant is determined from the definition , , it is required to know the fine structure constant and the electron mass very precisely . if the lepton @xmath124 s are used to determine the fine structure constant , the hidden photon contribution to the transition frequency , @xmath123 , becomes obscure . also , the relation can not be used to determine the @xmath30 , because it depends on @xmath39 . thus , other methods are required to determine @xmath30 accurately . furthermore , it is very difficult to measure the electron mass with the required precision . the accuracy of @xmath41 is worse than that of eq . . in order to avoid the difficulties of the rydberg constant , let us consider a ratio of the transition frequencies , @xmath125 where the left - handed side is determined by experiments , and @xmath126 is defined by the limit of @xmath127 , @xmath128 it is noted that the rydberg constant disappears . the sm correction , @xmath121 , is precisely known @xcite . it is a function of @xmath129 , @xmath130 , @xmath131 , and the proton root - mean - square charge radius , @xmath132 scaled by the compton wavelength of electron , @xmath133 . among them , hidden photon contributions through @xmath130 , @xmath131 and @xmath134 can be safely neglected , since the corrections @xmath121 are less sensitive to these parameters . on the other hand , in the following analysis , the fine structure constant is determined by the rubidium atom based on the previous section . the contribution of the hidden photon gives arise mainly through the electron @xmath0 in the carbon atom . however , as we will be see later , it is orders of magnitude smaller than the contribution in the energy level shift , @xmath123 . the proton charge radius is currently determined by the transition frequencies as well as the electron proton scattering experiment @xcite . in our analysis , the recommended value by codata @xcite can not be applied , since @xmath132 is mainly determined by the spectroscopic data there . we will instead use the result from the electron proton scattering experiment , @xmath135 @xcite . its determination is considered to be insensitive to the hidden photon effects . since the scattering cross section is dominated by the qed as @xmath136 , the hidden photon contributions are at most @xmath137 . thus , in the parameter region where the muon @xmath0 is explained , the effect is blinded behind the experimental uncertainties . on the other hand , a more accurate proton radius has been obtained by measuring the lamb shift of the muon hydrogen between @xmath138 @xcite . unfortunately , the result is inconsistent with the codata recommended value as well as the electron proton scattering result . this problem may be solved by corrections to the sm estimation ( see , for instance , @xcite for a review ) . in the following analysis , we adopt the electron proton scattering result by following the argument by @xcite . among the transition frequencies , the levels between @xmath139 of the hydrogen has been measured with the best precision ( see table [ tab : numbers ] ) @xcite . it turns out that the severest constraint on @xmath123 is obtained from the ratio of the transition frequencies of @xmath139 and @xmath140 @xcite . the relation becomes @xmath141 where the sm corrections are calculated by following the analysis in ref . @xcite . from the experimental data and the theoretical calculations with the input parameters discussed above , we obtain @xmath142 where the largest uncertainty is from the experimental result of @xmath140 , while the theoretical error , which originates in the proton radius , is half of the experimental one . therefore , the constraint is obtained as @xmath143 at the 2@xmath118 level . we have checked that if the hidden photon contribution to the fine structure constant is taken into account , the numerical factor , @xmath144 , is corrected by @xmath145 , and thus , it is negligible . in fig . [ fig : darkphoton ] , the constraint from the transition frequency is plotted . above the blue solid line , the hidden photon model is excluded by the hydrogen spectroscopy . the current limit is weaker by about one order of magnitude than that of the electron @xmath0 . however , the constraint is independent of the electron @xmath0 , and the sensitivity is expected to be improved in future . the uncertainty of eq . is dominated by the experimental measurement of @xmath146 . it is expected to be reduced , since the experiment is in progress at the national physics laboratory @xcite . in addition , the theoretical error is mainly from the proton charge radius . if the inconsistency between the values of the electron proton scattering and the muon hydrogen will be resolved , the uncertainty in the latter result is smaller by one order of magnitude than that of the former , and the theoretical error associated to the proton radius will decrease greatly . the sub - leading contribution to the theoretical error is from the two - photon correction to the sm prediction . since this is of @xmath122 , the constraint from the transition frequency can be comparable to that of the electron @xmath0 in future . the hidden photon model is one of the simplest models which can explain the muon @xmath0 anomaly . in this paper , we have examined the constraints on the hidden photon model from the electron @xmath0 as well as from the precise measurements of atomic parameters . we have shown that the new constraint from the electron @xmath0 is more than one order of magnitude stronger than the previous analysis . the parameter space which can explain the muon @xmath0 anomaly becomes narrow , @xmath147100 mev and @xmath148@xmath149 . interestingly , this region can be covered by experiments in future @xcite . the theoretical prediction of the electron @xmath0 is based on a precise measurement of the fine structure constant , which is determined independently of the @xmath0 . we have also discussed the effect of the hidden photon on the determination of the fine structure constant . we have shown that , although the determination of the fine structure constant is affected by the hidden photon , its contribution to the electron @xmath0 is small compared with the direct loop contribution to the @xmath0 . we have also investigated a constraint on the hidden photon model from the precise measurements of hydrogen transition frequencies . since the atomic energy levels are affected by the hidden photon , the precisely measured transition frequencies can be used to constrain the hidden photon model . a careful analysis is essential , since the determination of the various quantities , such as the rydberg constant or transition frequencies , are often correlated . we have shown that , by taking a ratio of transition frequencies , the measurement becomes free from the uncertainty of the rydberg constant , and a severe constraint is obtained . the constraint on the hidden photon model is now about one order of magnitude weaker than the electron @xmath0 , but more precise measurements of the transition frequencies can improve the constraint , which may compete with the @xmath0 constraint . while completing this paper , we noticed that the version 2 of ref . @xcite was posted on arxiv , which also updated the bound on the hidden photon model from the electron @xmath0 . after this paper was posted , the kloe experiment updated the constraint at the 90% c.l . @xcite , which is shown in fig . [ fig : combined ] . this work was supported by grand - in - aid for scientific research from the ministry of education , science , sports , and culture ( mext ) , japan , no . 23740172 ( m.e . ) , no . 21740164 ( k.h . ) , and no . 22244021 ( k.h . ) . this work was supported by world premier international research center initiative ( wpi initiative ) , mext , japan . in this appendix , we list some values of precisely known quantities , which are used in sec . [ sec : alpha ] and sec . [ sec : new_constraint ] . in table [ tab : numbers ] , the hydrogen transition frequencies are shown for @xmath150@xmath151 and @xmath152@xmath150 transitions . the effects of hyper fine splitting is corrected in the table . the hidden photon affects the hyper fine splitting through the magnetic moments , but its contribution is smaller than the experimental uncertainties . the ratios of cyclotron frequencies ( and electron s precession frequency ) are also shown for the relevant ions . in the table , the relative binding energies of atoms and ions are also listed . we could not find the uncertainties for some of the binding energies , but we expect these uncertainties are close to the ones next to them . binding energies are measured in terms of @xmath153 @xcite , and conversion into units of ev or relative atomic mass unit may generate uncertainty . however , uncertainties coming from these conversion factors are negligibly small . .experimentally measured quantities which are used in the discussion in sec . [ sec : alpha ] and sec . [ sec : new_constraint ] . in the last two lines , @xmath154 denotes the bond energies . [ cols=">,<,^ , > , < , < , < , < " , ] 99 g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys . rev . d * 73 * , 072003 ( 2006 ) [ hep - ex/0602035 ] . k. hagiwara , a. d. martin , d. nomura and t. teubner , phys . b * 649 * , 173 ( 2007 ) [ hep - ph/0611102 ] , t. teubner , k. hagiwara , r. liao , a. d. martin and d. nomura , chin . c * 34 * , 728 ( 2010 ) [ arxiv:1001.5401 [ hep - ph ] ] , k. hagiwara , r. liao , a. d. martin , d. nomura and t. teubner , j. phys . g g * 38 * , 085003 ( 2011 ) [ arxiv:1105.3149 [ hep - ph ] ] ; m. davier , a. hoecker , g. lopez castro , b. malaescu , x. h. mo , g. toledo sanchez , p. wang and c. z. yuan _ et al . _ , eur . j. c * 66 * , 127 ( 2010 ) [ arxiv:0906.5443 [ hep - ph ] ] , m. davier , a. hoecker , b. malaescu , c. z. yuan and z. zhang , eur . j. c * 66 * , 1 ( 2010 ) [ arxiv:0908.4300 [ hep - ph ] ] , m. davier , a. hoecker , b. malaescu and z. zhang , eur . j. c * 71 * , 1515 ( 2011 ) [ erratum - ibid . c * 72 * , 1874 ( 2012 ) ] [ arxiv:1010.4180 [ hep - ph ] ] . j. prades , e. de rafael and a. vainshtein , ( advanced series on directions in high energy physics . 20 ) [ arxiv:0901.0306 [ hep - ph ] ] . p. fayet , phys . d * 75 * , 115017 ( 2007 ) [ hep - ph/0702176 ] . m. pospelov , phys . d * 80 * , 095002 ( 2009 ) [ arxiv:0811.1030 [ hep - ph ] ] . f. archilli , d. babusci , d. badoni , i. balwierz , g. bencivenni , c. bini , c. bloise and v. bocci _ et al . _ , phys . b * 706 * , 251 ( 2012 ) [ arxiv:1110.0411 [ hep - ex ] ] . s. abrahamyan _ et al . _ [ apex collaboration ] , phys . rev . lett . * 107 * ( 2011 ) 191804 [ arxiv:1108.2750 [ hep - ex ] ] . a. bross , m. crisler , s. h. pordes , j. volk , s. errede and j. wrbanek , phys . lett . * 67 * , 2942 ( 1991 ) . t. aoyama , m. hayakawa , t. kinoshita and m. nio , arxiv:1205.5368 [ hep - ph ] . r. bouchendira , p. clade , s. guellati - khelifa , f. nez , f. biraben , phys . rev . 106 * , 080801 ( 2011 ) [ arxiv:1012.3627 [ physics.atom-ph ] ] . d. hanneke , s. fogwell and g. gabrielse , phys . lett . * 100 * , 120801 ( 2008 ) [ arxiv:0801.1134 [ physics.atom-ph ] ] ; d. hanneke , s. f. hoogerheide and g. gabrielse , phys . a * 83 * , 052122 ( 2011 ) [ arxiv:1009.4831 [ physics.atom-ph ] ] . g. f. giudice , p. paradisi and m. passera , arxiv:1208.6583 [ hep - ph ] . p. j. mohr , b. n. taylor and d. b. newell , arxiv:1203.5425 [ physics.atom-ph ] . j. jaeckel and s. roy , phys . d * 82 * , 125020 ( 2010 ) [ arxiv:1008.3536 [ hep - ph ] ] . see also , v.popov , tr . j. of physics 23 , 943 , 1999 ; s. g. karshenboim , phys . lett . * 104 * , 220406 ( 2010 ) [ arxiv:1005.4859 [ hep - ph ] ] ; phys . d * 82 * , 073003 ( 2010 ) [ arxiv:1005.4872 [ hep - ph ] ] . t. beier _ et al . lett . * 88 * , 011603 ( 2001 ) . j. l. hewett , h. weerts , r. brock , j. n. butler , b. c. k. casey , j. collar , a. de govea and r. essig _ et al . _ , arxiv:1205.2671 [ hep - ex ] . a. hansson , talk presented at 8th patras workshop on axions , wimps and wisps , july 2012 , chicago and fermilab . j. c. bernauer _ et al . _ [ a1 collaboration ] , phys . rev . * 105 * , 242001 ( 2010 ) [ arxiv:1007.5076 [ nucl - ex ] ] . r. pohl , a. antognini , f. nez , f. d. amaro , f. biraben , j. m. r. cardoso , d. s. covita and a. dax _ et al . _ , nature * 466 * , 213 ( 2010 ) . h. davoudiasl , h. -s . lee and w. j. marciano , arxiv:1208.2973 [ hep - ph ] . j. flowers , p. baird , l. bougueroua , h. klein , h. margolis , ieee trans . instrum * 56 * ( 2 ) , 331 ( 2007 ) ; j. flowers , p. baird , h. klein , d. langham , h. margolis , b. walton , precision electromagnetic measurements digest , 2008 . cpem 2008 . c. e. moore , in nat . stand . ref . data ser . , nsrds - nbs 3 ( sect . 7 ) , 33 ( 1976 ) ; ibid . 8) , 31 ( 1979 ) ; ibid . 9 ) , 21 ( 1980 ) ; ibid . 10 ) 21 ( 1982 ) ; ibid . 11 ) , 33 ( 1985 ) ; w. c. martin , victor kaufman , arlene musgrove jpcrd 22(5 ) 1179 ( 1993 ) .	the hidden photon model is one of the simplest models which can explain the anomaly of the muon anomalous magnetic moment ( @xmath0 ) . the experimental constraints are studied in detail , which come from the electron @xmath0 and the hydrogen transition frequencies . the input parameters are set carefully in order to take dark photon contributions into account and to prevent the analysis from being self inconsistent . it is shown that the new analysis provides a constraint severer by more than one order of magnitude than the previous result . ut-12 - 31 + ipmu12 - 172 + * * constraints on hidden photon models + from electron @xmath0 and hydrogen spectroscopy .75 in motoi endo@xmath1 , koichi hamaguchi@xmath1 , go mishima@xmath2 0.25 in _ @xmath3 department of physics , the university of tokyo , tokyo 113 - 0033 , japan + @xmath4kavli institute for the physics and mathematics of the universe , the university of tokyo , kashiwa 277 - 8568 , japan _ .5 in
this paper summarises results from our hst / multiwavelength campaign on the 1996 outburst of gro j1655 - 40 , and near quiescent 1994 observations of gro j0422 + 32 . our primary goal is to probe and understand the mechanisms causing the dramatic luminosity evolution in these systems . [ v1 ] = 2.5 in the superluminal jet source gro j1655 - 40 has undergone repeated outbursts since its 1994 discovery . we mounted a coordinated hst and rxte campaign between 1996 may 14 and july 22 . fuller descriptions of these observations and their interpretation are given in hynes et al . ( 1998a , b ) . modelling of the @xmath0 feature yields e(b - v)=@xmath1 ( hynes et al . 1998a ) . figure 1 is the 1996 may 14 dereddened uv - optical spectrum . though the uv portion of the spectrum is consistent with the @xmath2 power - law predicted by the steady - state blackbody disk model , the optical ( @xmath3 ) spectrum rises to longer wavelengths in contrast to the predictions of the model . ignoring the @xmath4 data , a @xmath2 model can be fit to the uv data , implying a mass transfer rate of @xmath5m@xmath6yr@xmath7m@xmath6yr@xmath8 , if the compact object mass is @xmath9 as the quiescent observations imply ( orosz and bailyn 1997 , hereafter ob97 ) . the dominant source of uncertainty in this @xmath10 arises from the extinction correction . assuming an accretion efficiency of @xmath11 , the eddington rate is @xmath12m@xmath13yr@xmath8 , so near the peak of the outburst this interpretation of the uv spectrum implies @xmath14 . we need to invoke something other than a pure steady - state optically thick accretion disk to explain the optical light . the shape of the spectrum is qualitatively suggestive of an irradiated disk ; irradiation can alter the temperature profile of the outer disk producing a rise in flux towards longer wavelengths as illustrated in fig . the multiwavelength light curves for the outburst ( hynes et al . 1998a ) do not support a _ simple _ irradiation model : the optical and uv flux declines while the x - ray flux rises . nontheless correlated x - ray and optical / uv variability was detected ( see section 2.3 and fig . 3a ) indicating at least some of the optical / uv flux is due to reprocessing in the disk . while the optical fluxes fell by about a factor of three between our first and last visit , the colour temperature remained almost constant , dropping from 9800 k to 8700 k ; the effective emitting area dropped from @xmath15 to @xmath16 ( hynes et al . the well constrained system parameters for gro j1655 - 40 ( ob97 , hjellming and rupen 1995 ) imply the total available emitting area is @xmath17 , so it is possible to explain the optical emission at the peak of the outburst as thermal emission , but only if both the secondary star and the majority of the disk area have essentially the same temperature . in this scenario , the @xmath2 uv component can , however , still be attributed to the hot inner regions of the disk . while noting that the spectral shape is strongly dependent on the adopted reddening correction , our optical spectra appear more strongly peaked than a single temperature blackbody , so we considered non - thermal emission mechanisms ( hynes et al . 1998a , especially fig . 7 and 11 ) . self - absorbed synchrotron emission from a compact cloud of relativistic electrons produced good fits to the optical component . attributing substantial optical flux to this compact non - thermal source relieves the requirement for a large isothermal emitter in the system . but while the synchrotron models fit better than black bodies , they have an extra free parameter , and there is no external check comparable to that provided by the emitting area required for the black body interpretation . the intrinsic vri band linear polarisation ( @xmath18 ) detected in july 1996 ( scaltriti et al . 1997 ) seemed consistent with the hypothesis of optical synchrotron emission . subsequent phase - resolved polarisation measurements ( gliozzi et al . 1998 ) seem , however , to indicate an extended polarisation region , inconsistent with the possible compact non - thermal source . the lack of optical dips during near - total x - ray dips ( fig . 3b ) provides further evidence against a compact central source for the optical continuum . hence , on balance a thermal origin for the optical continuum seems most likely . indeed , the dim may _ require _ a large isothermal outer disk for a long - period system like gro j1655 - 40 . since the disk in such a system is large , the temperature for a steady state disk falls below the minimum temperature for the hot , high viscosity , state long before the outer disk is reached . even for an eddington mass transfer rate in gro j1655 - 40 the temperature drops below that of the hot state at a radius less than a quarter of the roche lobe radius . this means that there is no global steady - state solution for @xmath19 . it is not clear what will happen to the temperature distribution in such a case , but it is possible that in outburst much of the outer disk could be maintained just in the hot state . hence one might expect the outer disk to appear as an approximately constant temperature , shrinking area emitter as the decline proceeds . this hypothesis awaits tests with self - consistent numerical modeling of the dim in long period systems . [ pcyg ] * a ) * we found likely p - cygni profiles in the uv resonance lines ( fig 2b ) . the peak to trough separation is @xmath20 , slightly larger than seen in outbursting dwarf novae . line profiles produced by biconical accretion disk winds were calculated by shlosman and vitello ( 1993 ) who found ` classical ' p cygni profiles only for inclinations around 6070@xmath21 , in striking agreement with the inclination determined for gro j1655 - 40 ( ob97 ) . we conclude that there was likely a biconical accretion disk wind at the peak of the uv outburst , when @xmath22 . * a ) * correlated rapid variability between the optical / uv and x - ray emission occurred in simultaneous rxte and hst data from 1996 june 8 ( fig . the mean delay of the optical / uv was up to 25s ( hynes et al . hence , the correlations are likely due to reprocessing of the x - rays into optical and uv emission , the delay being due to the finite light travel time between the x - ray source and the reprocessing regions . the lag of up to 25s is consistent with reprocessing in the accretion disk . at the binary phase of the observations , lags of greater than 40s are expected for light echoes from the mass donor star . the lack of a reprocessing signal from the mass donor star may imply that the x - ray absorbing material in the accretion flow is sufficiently vertically extensive to effectively shield it from direct x - ray illumination . for the roche geometry deduced by ob97 , this implies @xmath23 is at least @xmath24 along the line of centres . * hst observations of this target covering the vacuum uv and the entire optical region were obtained on 1994 august 25 and 26 , approximately two years after the first observed outburst , and seven months after the last reported reflare ( callanan et al . 1995 ) . a full discussion of these data is given in hynes and haswell ( 1999 ) . figure 4a shows the r band light curve including the point deduced from our spectrum . the vertical marks above the light curves indicate the times of mini - outbursts according to the suggestion ( augusteijn , kuulkers , and shaham 1993 , chevalier and ilovaisky 1995 ) that they recur on a 120 day period . our observation clearly lies above the subsequent points , and it appears we saw the end of a previously unreported mini - outburst or of an extended plateau . figure 4b shows the spectrum after dereddening and subtracting the mass donor star flux , hence represents the intrinsic accretion spectrum . the spectral shape in fig . 4b is distinctive : there is a pronounced peak in the optical at @xmath25 ; and since the interstellar reddening towards this system is moderate , @xmath26 , it is unlikely that this is an artifact of an improper reddening correction . the best fitting black body shown in figure 4b is clearly less sharply peaked than the observed spectrum from gro j0422 + 32 , while a simple self - absorbed synchrotron spectrum can successfully reproduce the continuum shape . we stress , however , that this model is very simplistic and conclude simply that the continuum spectrum is highly suggestive of a self - absorbed synchrotron source , rather than black body emission . when compared with advective models , in which the quiescent optical continuum is dominated by synchrotron emission , the agreement between model and data is encouraging ( esin 1998 ) . the advective models for quiescent black hole transients predict that the inner disk is replaced by a hot , optically thin , quasi - spherical flow . figure 4b reveals strong evidence for non - thermal optical emission in gro j0422 + 32 near quiescence ; this does not necessarily confirm particular detailed advective models , but supports the general scenario . we plan to perform detailed comparisons of our gro j0422 + 32 spectrum with the latest models . we also considered the possibility of non - thermal emission in the optical light from gro j1655 - 40 during the decline from its 1996 outburst . in this case , however , it is alternatively possible to attribute the optical spectra to thermal emission , as the sharply peaked spectral shape is dependent on the exact details of the ( large ) reddening correction . the non - appearance of an optical modulation during near - total x - ray dips shown in fig 4b suggests that the x - ray source is far more compact than the source of the optical continuum , supporting the standard thermal interpretation of the optical continuum . at the peak of the optical / uv light curves in 1996 , the uv spectrum from gro j1655 - 40 was consistent with an eddington - rate mass transfer through the uv emitting parts of the disk . at this time , however , the x - ray luminosity was less than @xmath27 . this apparent discrepancy may be partly due to outflow from the inner regions of the disk : fig . 2b shows likely p - cygni profiles indicating a disk wind with velocity @xmath20 . in both sets of observations described here , the vacuum uv spectrum was of low signal to noise ratio . our single uv spectrum of gro j1655 - 40 proved informative , despite the effects of the large reddening . we have our multiwavelength network poised to observe newly discovered transients , and anticipate obtaining high quality uv / optical spectra through decline . with this program we should be able to address many of the outstanding issues regarding the temperature structure , morphology , irradiation behaviour , and outflow characteristics of these systems . * acknowledgments : * support for this work was provided by the nuffield foundation and by nasa through grant numbers go-6017.01 - 94a and go-4377.02 - 92a from the space telescope science institute , which is operated by aura under nasa contract nas5 - 26555 . we are grateful to our collaborators for their contributions to this work . fuller discussions of the work described herein is in hynes et al . ( 1998a , b ) and hynes and haswell ( 1999 ) . 1 . augusteijn , t. , kuulkers , e. , and shaham , j. , 1993 , a&a , 279 , l13 . 2 . callanan p. j. et al . , 1995 , apj , 441 , 786 3 . chevalier c. , ilovaisky s. a. , 1995 , a&a , 297 , 103 4 . esin , a. 1998 , private communication . garcia m. r. , callanan p. j. , mcclintock j. e. , zhao p. , 1996 , apj , 460 , 932 6 . gliozzi , m. , bodo , g. , ghisellini , g. , scaltriti , f. , trussoni , e. 1998 , a&a , 337 , l39 . 7 . hjellming r. m. , rupen m. p. , 1995 , nat , 375 , 464 8 . hynes , r.i , and haswell , c.a . , 1999 , mnras , in press , astro - ph/980941 . hynes , r.i , haswell , c.a . , shrader , c.r . , chen , w. , horne , k. , harlaftis , e.t . , obrien , k. , hellier , c. , and fender , r.p . 1998a , mnras , 300 , 64 . 10 . hynes , r.i , obrien , k.o . , horne , k. , chen , w. , and haswell , c.a . 1998b , mnras , 299 , l37 . 11 . orosz j. a. , bailyn c. d. , 1997 , apj , 477 , 876 ( ob97 ) 12 . scaltriti f. , bodo g. , ghisellini g. , gliozzi m. , and trussoni e. 1997 , a&a , 325 , 29 . shlosman i. , vitello p. , 1993 , ap.j , 409 , 372	results from an hst / multiwavelength campaign on the 1996 outburst of gro j1655 - 40 are reported . we find evidence for an approximately isothermal outer disk in gro j1655 - 40 during the decline from outburst , possibly indicating the outer disk being maintained in the hot state . hst observations of gro j0422 + 32 made 2 years after outburst reveal a sharply peaked optical / near uv spectrum , providing strong evidence for non - thermal emission .
in this article , we propose a formalism for cloning partial quantum information beyond the restrictions imposed by the aforementioned no - go theorems . we use a family of increasingly growing entangled states @xcite in order to preserve and propagate the information of an initial state . in particular , we are able to transmit more than just classical information to the forthcoming generation , i.e. , both the diagonal elements and coherences of the associated density matrix . finally , we analyze the feasibility of a possible experimental implementation with trapped ions . to introduce our protocol , let @xmath0 and @xmath1 be an arbitrary state and a blank state , respectively , and let @xmath2 be a hermitian operator . we define the cloning operation @xmath3 as @xmath4 we denote each subspace as an _ individual _ , see fig . [ dos ] . the expectation value of @xmath2 in the initial state is cloned into both subspaces of the final state . this is the cloning machine for observables introduced in ref . @xcite . here , we extend their results to an arbitrary dimension and show the existence of an additional operator @xmath5 , which does not commute with @xmath2 , and whose statistics is encoded in the global state of the system @xmath6 at this stage , it is convenient to point out the differences between both processes . on the one hand , eq . can be regarded as a replication of a characteristic from an individual into its progeny . in this sense , the global state is traced out and the only important feature is the information of the individual . on the other hand , eq . can be considered as a spread of the quantum coherences of the initial individuals to the forthcoming generations as a whole . our formalism combines the notions of cloning and preserving quantum information without contradicting the no - cloning and no - broadcasting theorems @xcite . this allows the propagation of the statistics of two non - commuting observables in a controllable way . the explanation of the cloning method requires a selection of the basis , provided here by an external environment . the dark state of an unknown environment dynamics is the blank qubit for the copying process , i.e. , the state that we define as the @xmath7 . this point will be relevant later in the discussion about the quantumness of the process . * cloning method . * let us work without loss of generality in a basis in which @xmath2 is diagonal . then , we define the cloning operation @xmath8 in terms of the generator of @xmath9-dimensional irreducible representation of the translation group @xmath10 and the projectors into each subspace @xmath11 . we clarify that @xmath12 , after all , does not explicitly depend on @xmath2 . @xmath13 for example , for @xmath14 , @xmath15 , and for @xmath16 , @xmath17 is given by @xmath18 we will demonstrate below that eq . fullfils eq . . the forthcoming theorems are proved in the supplementary information @xcite . let @xmath19 be a hilbert space of dimension @xmath9 , @xmath20 the unitary operation defined in eq . , @xmath21 , and @xmath22 @xmath23 bounded hermitian operators . then , the unitary @xmath24 satisfies eq . . we analyze now if the cloning unitary @xmath24 transmits information from the initial individual to the progeny as a whole , apart from cloning its expectation value independently . the process mimics the information transmission underlying collective behaviors present in many biological systems , such as self - organizing neurons . let @xmath20 be the unitary transformation defined in eq . . then , there exists a bounded antidiagonal operator @xmath5 , whose matrix elements are @xmath25 or @xmath26 , fulfilling eq . . notice that , in previous theorems , we have worked in a basis in which @xmath2 is diagonal . however , the cloning operation can be rewritten in any basis just by rotating the matrix with the proper unitary , @xmath27 , transforming @xmath5 into @xmath28 and @xmath24 into @xmath29 . when the cloning operation is sequentially reapplied , it propagates the information of the initial state , i.e. , it transmits the full statistics of the density matrix , tr@xmath30 $ ] , @xmath31 . we show here how two - qubit operations are extended into @xmath32-qubit states , @xmath33 where the subscripts of @xmath24 refer to the pair of qubits that is acted upon . the cloning for the subsequent generations , see fig . [ mas ] , is constructed through the product of pairwise cloning operations , @xmath34 . for instance , the density matrices of the first and second generations reads @xmath35 therefore , the mechanism is straightforwardly generalizable for obtaining sequential generations with the same information in each individual , in the spirit of a quantum genetic code . moreover , although eq . holds only for qubits , it is noteworthy to mention that an extension to higher dimensions is possible . this can be achieved using as building block the unitary gate @xmath12 defined in eq . . we will present now a counter - example showing that the cloning operation is not unique . when @xmath9 is not a prime number , @xmath36 , there are other cloning unitaries apart from @xmath37 . for instance , an additional @xmath38 is constructed via the solutions in each subspace , @xmath39 and @xmath40 , respectively , @xmath41 this result shows that it is possible to mix information among subspaces of different dimensions . an example for @xmath42 is depicted in fig . [ uno ] . + . * in these graphics , each color represents a matrix of the translation group . therefore , each of the three two - dimensional arrays groups the six matrices of the translation group for @xmath42 . the first array combines the @xmath43 operations of eq . , while the second and third show @xmath44 of eq . with @xmath45 and @xmath46 , respectively.,width=330 ] * quantumness and classicality . * in this section , we analyze the quantumness of the proposed biomimetic cloning protocol . in ref . @xcite , meznaric _ recently introduced a criterion to determine the nonclassicality of an operation @xmath47 . this method is based on the distance between the outcome of the operation and the pointer basis einselected by the environment . the quantum operation @xmath47 is composed with the completely dephasing channel @xmath48 , provided by the environment . the measurement of nonclassicality is obtained by maximizing over all states the relative entropy of both operations acting on an arbitrary quantum state . the completely dephasing channel plays the role of an external environment that einselects the pointer basis , in which @xmath47 is represented . effectively , as proposed in ref . @xcite , any operation is quantum whenever its column vectors are superpositions of the elements of the basis . on the contrary , the operation is classical if they are just permutations of the basis . it is natural to identify the basis of the ancillary qudits , @xmath49 , with the pointer states , since we assume that the system naturally provides blank qudits . by applying the classification formalism for qubits , the process of copying @xmath50 in the individual state and @xmath51 in the global state is classical , because the @xmath52 is written in terms of permutations of the pointer states . nevertheless , the complementary operation @xmath53 which clones @xmath51 in the individual state and @xmath50 in the global state , is quantum . another possibility is to consider that the classical pointer basis is defined by the @xmath47 operator itself . this means that we can construct the quantum channel @xmath54 which maps the initial state to any of the outcomes , considering the blank qudit and the other outcome as the environment . by construction , the unitary given by eq . leads to an injective channel , since the number of kraus operators is the same as the square of the matrix dimension . therefore , the only fixed point is the identity and the cloning operations are classical when written in the einselected basis @xcite . according to these results , the cloning formalism copies classical information but preserves quantum correlations , which makes the global operation quantum . the interpretation of this property is that the quantum part of the information is encoded in the global state . by analogy with biological systems , the environment plays a fundamental role in the kind of information that , similar to quantum darwinism @xcite , is preserved and cloned through a pure quantum mechanism . in our work , the quantumness is only revealed when considering collectively all outcomes of the copying process . * experimental implementation . * we consider that an experimental realization of our protocol in a quantum platform sets a significant step towards quantum artificial life . we propose an experimental setup of our formalism in trapped ions , arguably the most advanced quantum technologies in terms of coherence times and gate fidelities @xcite . current experimental resources would allow copying processes for qubits and qutrits of three and two generations , respectively . the number of logical gates and trapped ions required to perform the experiment is presented in table [ reo ] . we point out that the only limitations for performing trapped - ion experiments with higher dimensions and larger number of generations are decoherence times and gate errors . none of these are fundamental and near future improvements may allow to reach the implementation of higher dimensional individuals and many more generations with near perfect fidelities . sequential generation step & 0 & 1 & 2 & 3 + total gates ( 2 qubit gates ) & 0 & 19(2 ) & 57(6 ) & 133(14 ) + ions & 1 & 2 & 4 & 8 + + sequential generation step & 0 & 1 & 2 & - + total gates ( 2 qubit gates ) & 0 & 38(4 ) & 114(12 ) & - + ions & 2 & 4 & 8 & - + [ reo ] we encode the initial state in one of the ions , while the rest are initialized in the @xmath7 state . the cloning operation for two qubits is the cnot gate , which can be reproduced performing the m@xmath55lmer - s@xmath55rensen gate @xcite and a sequence of single qubit gates @xcite , @xmath56 . here , @xmath57 is the phase gate , @xmath58 is the hadamard gate , @xmath59 is the m@xmath55lmer - s@xmath55rensen gate , and the subindex denotes the ion number . we express the gates as products of carrier transitions @xmath60 and a phase factor , @xmath61 the first and second generations of cloned qubits are obtained as indicated in eq . . the next step in the copying process , i.e. , the third generation , is given by @xmath62 . + we consider now the implementation of the qutrit case in trapped ions . as there is no direct access to qutrit gates , we have to engineer a protocol with one and two - qubit gates . we suggest to add three ancillary levels that split the unitary operation into three subspaces of @xmath63 matrices . to achieve this , for the @xmath64 given in eq . , the modified @xmath65 is @xmath66 the first submatrix does not require any quantum gate . the second and third submatrices are the products of two cnot gates , for which the role of control and target ions is interchanged . hence , the implementation is reduced from qutrit to qubit operations . in summary , inspired by biological systems , we have brought concepts and applications into quantum information theory . for instance , our partial quantum cloning method makes use of global and local measurements in order to encode information of nonconmuting observables beyond the classical realm . moreover , we prove that the information transmission is purely quantum for a certain kind of operators . in parallel , we show that it is possible to implement our ideas in an ion - trap platform with current technology . replication is the most fundamental property that one may require from a biological system . we leave to forthcoming works the introduction of additional biological behaviors such as mutations , evolution and natural selection , englobed by the frame of quantum artificial life . this proposal should be considered as the first step towards mimicking biological behaviours in controllable quantum systems , a concept that we have called quantum biomimetics . 28 feynman , r. p. simulating physics with computers . phys . _ * 21 , * 467 - 488 ( 1982 ) . lloyd , s. universal quantum simulators . _ science _ * 273 , * 1073 - 1078 ( 1996 ) . wootters , w. k. & zurek , w. h. a single quantum can not be cloned . _ nature _ * 299 , * 802 - 803 ( 1982 ) . buzek , v. & hillery , m. quantum copying : beyond the no - cloning theorem . a _ * 54 , * 1844 - 1852 ( 1996 ) . scarani , v. , iblisdir , s. , gisin , n. & acn , a. quantum cloning _ rev . phys . _ * 77 , * 1225 - 1256 ( 2005 ) . delgado , y. , lamata , l. , len , j. , salgado , d. & solano , e. sequential quantum cloning _ phys . * 98 , * 150502 ( 2007 ) . gisin , n. & massar , s. optimal quantum cloning machines _ phys . lett . _ * 79 , * 2153 - 2156 ( 1997 ) . lamata , l. , len , j. , prez - garca , d. , salgado , d. & solano , e. sequential implementation of global quantum operations _ phys . lett . _ * 101 , * 180506 ( 2008 ) . saberi , h. ancilla - assisted sequential approximation of nonlocal unitary operations _ phys . a _ * 84 , * 032323 ( 2011 ) . ferraro , a. , galbiati , a. & paris , m. cloning of observables _ a _ * 39 , * l219-l228 ( 2006 ) . lindblad , g. a general no - cloning theorem _ lett . _ * 47 , * 189 - 196 ( 1999 ) . brougham , t. , andersson , e. & barnett , s. m. cloning and joint measurements of incompatible components of spin _ phys a. _ * 73 , * 062319 ( 2006 ) . vincent , j. f. v. , bogatyreva , o. a. , bogatyrev , n. r. , bowyer , a. & pahl , a. biomimetics : it s practice and theory _ j. r. soc . interface _ * 3 , * 471 - 482 ( 2006 ) . gruber , p. et al . _ biomimetics - materials , structures and processes _ , ( springer verlag , berlin heidelberg 2011 ) . parker , a. r. & townley , h. e. biomimetics of photonics nanostructures _ nature nanotechnology _ * 2 , * 347 - 353 ( 2007 ) . vilfan , m. et al . self - assembled artificial cilia _ proc . _ * 107 , * 1844 - 1847 ( 2010 ) . schrdinger , e. _ what is life ? _ , ( cambridge university press , cambridge , 1944 ) . d. abbot , p. c. w. davies , a. k. pati et al . _ quantum aspects of life _ , ( imperial college press , london , 2008 ) . gottesman , d. e - print quant - ph/9802007v1 wang , x. sanders , b. c. & berry , d. w. entangling power and operator entanglement in qudit systems _ phys . a. _ * 67 , * 042323 ( 2003 ) . barnum , h. , caves , c. m. , fuchs , c. a. , jozsa , r. & schumacher , b. noncommuting mixed states can not be broadcast _ phys . lett . _ * 76 , * 2818 - 2821 ( 1996 ) . see supplementary information for further explanations . meznaric , s. , clark , s. r. & datta , a. quantifying the nonclassicality of operations _ phys . * 110 , * 070502 ( 2013 ) . sanz , m. , prez - garca , d. , wolf , m. m. & cirac , j. i. a quantum version of wielandt s inequality _ ieee trans . inf . theory . _ * 56 , * 4668 - 4673 ( 2010 ) . zurek , w. h. quantum darwinism _ nature phys . _ * 5 , * 181 - 188 ( 2009 ) . leibfried , d. , blatt , r. , monroe , c. , & wineland , d. quantum dynamics of single trapped ions _ rev . * 75 , * 281 - 324 ( 2003 ) . s@xmath55rensen , a. & m@xmath55lmer , k. quantum computation with ions in thermal motion _ phys . _ * 82 , * 1971 - 1974 ( 1999 ) . cirac , j. i. & zoller , p. quantum computation with cold trapped ions _ * 74 , * 4091 - 4094 ( 1995 ) . the authors acknowledge funding from basque government bfi-2012 - 322 and it472 - 10 grants , spanish mineco fis2012 - 36673-c03 - 02 , ramn y cajal grant ryc-2012 - 11391 , upv / ehu ufi 11/55 , ccqed , promisce , and scaleqit european projects . u. a .- r . made the calculations while u. a .- r . , m. s. , l. l. and e. s. developed the model and wrote the manuscript . * competing financial interests : * the authors declare no competing financial interests . + + in this supplemental material we add mathematical calculations that complement the information of the article . @xmath67 here we give the proof of theorem 1 in the paper . @xmath68 \\ & = & \textrm{tr } [ \sum^{n}_{i , j=1 } s_i \rho s^{\dag}_j \theta \otimes x_i \rho_{e } x^{\dag}_{j}].\end{aligned}\ ] ] when looking at the second subspace , @xmath69 , only the terms with @xmath70 are left in the diagonal , therefore only those are relevant for the trace . we consider @xmath70 in the first subspace , @xmath71 . the remaining is the product between the diagonal terms of the density matrix @xmath72 and the observable @xmath2 , that is precisely . @xmath73=\langle \theta \rangle\ ] ] this shows that the mean value of @xmath2 is effectively cloned in the first individual by means of @xmath24 . the proof for the second individual is obtained in a similar way . @xmath67 we enunciate and prove a lemma that clarifies the properties of the cloning operation u. let @xmath74 be the unitary transformation given by the previous theorem . then , there exists a bounded operator @xmath75 such that eq . @xmath76 is fulfilled iff @xmath77 , with @xmath78 and @xmath79 arbitrary unitary matrices of dimensions @xmath80 and @xmath81 respectively , and @xmath82 a flip operator such that @xmath83 . let us assume that @xmath5 is hermitian , since otherwise @xmath84 must also be a solution . then , we can work without lost of generality in the basis in which @xmath5 is diagonal , i.e. @xmath85 , @xmath86 a diagonal matrix . therefore , eq . @xmath76 is transformed into the equation @xmath87 , with @xmath88 also unitary . the necessary condition for the existence of @xmath5 is @xmath89 , which trivially means that the eigenvalues of @xmath86 must be @xmath25 , @xmath26 or @xmath90 , with the same degeneration on both sides . if we call @xmath91 , @xmath92 and @xmath93 to the degeneration of the eigenvalues @xmath90 , @xmath25 and @xmath26 respectively , a simple counting with the condition @xmath94 leads to the equations @xmath95 whose only non trivial solution is @xmath96 and @xmath97 when @xmath80 is _ even_. in other words , @xmath5 must be a unitary rotation of @xmath98 . this provides the necessary condition for @xmath99 and @xmath100 to be connected by _ unitary but not for the case in which the unitary is already fixed . so let us determine for which unitary matrices such a @xmath5 exists . let @xmath82 be a flip operator which transforms @xmath83 . then , we have @xmath101 , with @xmath102 . this means that the most general possible @xmath103 , with @xmath104 , @xmath105 two unitary matrices of dimension @xmath81 and the result follows straightforwardly . @xmath67 now we prove theorem 2 in the paper . @xmath106\\ \nonumber & = & \textrm{tr}[\sum^{n}_{i , j=1}s_{i}\rho s^{\dag}_{j } \tau \otimes x_{i } \rho_{e } x^{\dag}_{j } \tau].\end{aligned}\ ] ] the second subspace , @xmath107 , will only contribute to the trace when it is diagonal . since @xmath5 is antidiagonal the only effective combination of @xmath108 is ( @xmath109 ) , ( @xmath110),@xmath111 when applying this restriction to the first subspace , @xmath112 , the global expresion turns out to be tr[@xmath113 , only if the matrix elements of @xmath5 are @xmath25 or @xmath26 . we have thus shown how @xmath5 is promoted into the global state of the system . @xmath67 we show here that the @xmath5 operator is also transmitted when cloning in any basis . when the unitary operation @xmath24 is rotated with the matrix @xmath59 it propagates the information of the rotated matrix @xmath114 . @xmath115=\text{tr}[u ' ( \rho \otimes \rho_e ) u'^\dag ( \tau\otimes\tau ) ] = \\ & & \text{tr}[(r^{\dag } \otimes r^{\dag } ) u ( r \otimes \mathbb{1 } ) ( \rho \otimes \rho_{e } ) ( r^{\dag } \otimes \mathbb{1 } ) u^{\dag } \\ & & ( r \otimes r ) ( r^\dag \tau r\otimes r^{\dag}\tau r)]=\\ & & \text{tr}[u ( r\otimes \mathbb{1 } ) ( \rho\otimes\rho_e ) ( r^\dag \otimes \mathbb{1 } ) u^\dag ( r\otimes r ) ( r^\dag \tau\otimes r^\dag \tau)]= \\ & & \text{tr}[u r\rho r^{\dag } u^{\dag } ( \tau\otimes \tau)]=\text{tr}[u \tilde{\rho } u^{\dag } ( \tau \otimes \tau)]=\text{tr}[\tilde{\rho}\tau].\end{aligned}\ ] ] @xmath116=\text{tr}[\rho r^{\dag}\tau r]=\text{tr}[r \rho r^{\dag } \tau]=\text{tr}[\tilde{\rho}\tau].\end{aligned}\ ] ]	we propose a bio - inspired sequential quantum protocol for the cloning and preservation of the statistics associated to quantum observables of a given system . it combines the cloning of a set of commuting observables , permitted by the no - cloning and no - broadcasting theorems , with a controllable propagation of the initial state coherences to the subsequent generations . the protocol mimics the scenario in which an individual in an unknown quantum state copies and propagates its quantum information into an environment of blank qubits . finally , we propose a realistic experimental implementation of this protocol in trapped ions . quantum information is a research field that studies how to perform computational tasks with physical platforms in the quantum regime . coping with complex quantum systems could give rise to an exponential gain in computational power and a new branch of possibilities as compared with classical computing @xcite . one of the turning points of quantum information is the no - cloning theorem @xcite , which expresses the impossibility of copying an unknown state . therefore , the notion of perfect universal quantum cloning was abandoned , and replaced by the cloning of restricted families of states or cloning with imperfect fidelities . a paradigmatic instance is the buzek and hillery universal quantum cloning machine @xcite , among other cases @xcite . another approach is _ partial quantum cloning _ , consisting in the copy of partial quantum information of a quantum state . in this sense , an interesting example is the cloning of the statistics associated with an observable @xcite . however , these methods are limited due to the classical character of the information one replicates , since it is impossible to clone two non - conmuting observables with the same unitary @xcite . for a long time , human beings mimicked nature to create or optimize devices and machines , as well as industrial processes and strategies . in particular , biomimetics is the branch of science which designs materials and machines inspired in the structure and function of biological systems @xcite . analogously , novel quantum protocols may be envisioned by mimicking macroscopic biological behaviors at the microscopic level , in what we call _ quantum biomimetics_. living entities are characterized by features such as self - reproduction , mutations , evolution or natural selection . among them , the ability to self replicate is the most basic one . in fact , even though they are allowed to perfectly replicate classical information , biological systems only reproduce part of this information in the following generations . a paradigmatic example is dna replication inside the nucleus of the cell , since only sequences of bases are copied , but not all of the information about the physical state of the molecule .
we introduce a very simple effective theory in which the standard model is supplemented by a spontaneously broken @xmath4 gauge symmetry , the massive gauge boson of which acts as a portal to the standard model ( sm ) by coupling to the top quark . this type of setup arises naturally in models of partial compositeness " in which the top quark acquires its large mass ( after electro - weak symmetry breaking ) through large mixing with composite states in a new strong sector , as in 4d duals to randall - sundrum ( rs ) models @xcite . the effective lagrangian contains @xcite , @xmath5 where @xmath6 is a dirac fermion which is a singlet under the sm gauge interactions but is charged under @xmath4 and turns out to play the role of a weakly interacting massive particle ( wimp ) . @xmath7 ( @xmath8 ) is the usual abelian field strength for the @xmath9 ( hyper - charge boson ) , @xmath10 is the @xmath9 coupling to right - handed top quarks , and @xmath11 is the coupling to right - handed wimps . @xmath12 is the wimp mass . one can easily include a coupling to the left - handed top ( and bottom ) . our choice to ignore such a coupling fits well with typical rs models @xcite , balancing the need for a large top yukawa interaction with control over corrections to precision electro - weak observables . the parameter @xmath13 encapsulates the strength of kinetic mixing between the @xmath2 and sm hyper - charge bosons . we have included hyper - charge-@xmath9 kinetic mixing through the term proportional to @xmath13 . such a term is consistent with the gauge symmetries , and even if absent in the uv , will be generated in the ir description by loops of top quarks can be engineered to vanish in the uv , for example , by embedding @xmath4 into a larger gauge group which breaks down at scales of order @xmath14 . ] . the kinetic mixing parameter @xmath13 generates an effective coupling of sm states to the @xmath9 , and through electro - weak symmetry breaking , mass mixing of the @xmath9 with the sm @xmath15 gauge boson resulting in a coupling of @xmath6 to the sm @xmath15 boson . in this approach , most of the standard model is fundamental , but with the wimp , higgs , and right - handed top largely composite . the higgs couples strongly to composite fields , and the amount of admixture in a given sm fermion determines its mass @xcite . in this picture , the @xmath2 is one of the higher resonances , built out of the same preons as the wimp and @xmath16 . rs theories provide a very motivated picture of the uv physics , but more generically , in any theory containing composite wimps and composite top quarks belonging to a common sector one would expect strong couplings between them as a residual of the strong force which binds them , and perhaps negligible coupling to the rest of the standard model . if the @xmath2 mixes with the electro - weak bosons , this results in strong constraints from precision data . we circumvent these constraints by considering a @xmath2 whose mixing with the @xmath15 is kinetic . at large @xmath2 masses this is not operationally different from the mass - mixing case , but it allows us to consider lower mass @xmath2s which are not ruled out by precision data . mediated production of @xmath17 and @xmath18 at clic.,scaledwidth=50.0% ] since the coupling of @xmath2 to light sm fermions is suppressed by the small kinetic mixing factor , the best probe of the dark sector is through the top portal . in particular , the @xmath2 can be produced by being radiated from top quarks , as illustrated in fig.[fig : diagram ] . there are very interesting signals at the lhc @xcite . in this note , we look at the prospects at @xmath0 colliders . in fig . [ fig : crosssection ] , we show the leading order cross section at clic for @xmath19 production , as calculated by calchep 2.5.4 @xcite . depending on the masses and couplings , the @xmath2 will predominantly decay into @xmath20 , @xmath21 , or into light fermions . decays into top quarks lead to four - top events with a very large cross section compared to the sm four - top rate , which leads to a characteristic same - sign di - lepton signature @xcite ( see also @xcite for studies of a @xmath22 final state at the lhc ) . the right - handed nature of the @xmath2 coupling to top quarks implies top polarisation also provides an interesting observable . when the @xmath2 decays into wimps , a @xmath23 missing energy final state results , which is particularly compelling at high energy @xmath0 colliders such as clic . we present preliminary results of an analysis aimed at characterising these signals in multi - tev @xmath0 collisions . signal events are generated using calchep , assuming @xmath24 , and subsequently hadronised with pythia 6.125 @xcite . we assume the beams to be unpolarised . events are processed through full detector simulation using the geant-4-based mokka @xcite program and reconstructed with marlin - based @xcite processors assuming a version of the ild detector @xcite , modified for physics at clic . the performances most relevant to the analysis of these events are a relative parton energy resolution at 1 tev @xmath25 0.08 - 0.10 , charged particle momentum resolution @xmath26 = 2@xmath2710@xmath28 gev@xmath29 , track extrapolation resolution of 5 ( @xmath30 m ) @xmath31 @xmath32 . we perform jet clustering forcing the number of jets to the number of top quarks expected in the signal final state of interest . jet reconstruction is performed with the durham algorithm @xcite . in this preliminary study we do not account for machine induced backgrounds and do not attempt to perform a detailed top reconstruction through the identification of its @xmath33 decay , which will need to be evaluated in a more refined analysis . we now discuss both signatures in turn . the standard model background for @xmath34 is dominated by @xmath35 , which , at 3 tev , has a cross section of 4.1 fb . by comparison , the signal cross section for @xmath36 gev , @xmath37 gev , @xmath38 is 16.5 fb ( note that this choice of parameters leads to the correct relic density for dark matter @xcite ) . for a higher mass @xmath39 gev , the cross section is 7.6 fb . the event selection requires 2 @xmath40 , where @xmath41 is the number of natural jets obtained with the durham algorithm for @xmath42 = 0.0025 , more than 50 particles with @xmath43 0.5 gev , total reconstructed energy in the detector larger than 500 gev , transverse energy larger than 250 gev , event thrust larger than 0.75 and sphericity lower than 0.30 . the missing energy in the event , @xmath44 , appears to be the best discriminating variable against the sm @xmath45 `` irreducible '' background . figure [ fig : t4 ] shows the @xmath46 invariant mass distribution for 1 ab@xmath29 of integrated luminosity at 3 tev . the signal is clearly visible over a negligible sm background . for @xmath47 = 360 gev there are 240 signal events in the mass peak around the generated @xmath2 mass . the @xmath2 mass can be determined with a statistical accuracy of @xmath48 3.5 gev . to conclude , @xmath3 and multi - top production are expected signatures for several processes at , and beyond , the tev frontier . they have the advantage that the irreducible sm background is small so that they can provide clean probes of new physics . they also carry experimental challenges from the reconstruction of a large number of partons , with missing energy and leptons . the prospects for studying these signatures at the lhc are promising @xcite . in this contribution , we started investigating these signatures at @xmath0 colliders using a template model but it is clear that this work can be useful to other models having similar final states . we acknowledge useful discussion with jean - jacques blaising , marcel vos and ambroise espargilire . we are grateful to the organisers of the lc09 workshop in perugia for their invitation to an interesting and well - organised meeting . the work of g.s is supported by the european research council starting grant cosmo@lhc . l. randall and r. sundrum , phys . lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] ; k. agashe , a. delgado and r. sundrum , annals phys . * 304 * , 145 ( 2003 ) [ arxiv : hep - ph/0212028 ] . c. b. jackson , g. servant , g. shaughnessy , t. m. p. tait and m. taoso , jcap * 1004 * ( 2010 ) 004 [ arxiv:0912.0004 [ hep - ph ] ] . r. contino , y. nomura and a. pomarol , nucl . b * 671 * ( 2003 ) 148 [ arxiv : hep - ph/0306259 ] . k. agashe , r. contino and a. pomarol , nucl . b * 719 * , 165 ( 2005 ) [ arxiv : hep - ph/0412089 ] ; k. agashe and r. contino , nucl . b * 742 * ( 2006 ) 59 [ arxiv : hep - ph/0510164 ] ; r. contino , t. kramer , m. son and r. sundrum , jhep * 0705 * ( 2007 ) 074 [ arxiv : hep - ph/0612180 ] . g. brooijmans _ et al . _ , arxiv:1005.1229 [ hep - ph ] . l. gauthier and g. servant . in preparation . a. pukhov , arxiv : hep - ph/0412191 . b. lillie , j. shu and t. m. p. tait , jhep * 0804 * , 087 ( 2008 ) [ arxiv:0712.3057 [ hep - ph ] ] ; a. pomarol and j. serra , phys . d * 78 * , 074026 ( 2008 ) [ arxiv:0806.3247 [ hep - ph ] ] ; k. kumar , t. m. p. tait and r. vega - morales , jhep * 0905 * , 022 ( 2009 ) [ arxiv:0901.3808 [ hep - ph ] ] . r. contino and g. servant , jhep * 0806 * ( 2008 ) 026 [ arxiv:0801.1679 [ hep - ph ] ] ; j. mrazek and a. wulzer , arxiv:0909.3977 [ hep - ph ] . t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * ( 2006 ) 026 [ arxiv : hep - ph/0603175 ] . mora de freitas , in proc . of the _ int . conf . on linear colliders ( lcws 04 ) _ , ed . de lecole polytechnique , paris , 2004 , vol .	four - top production and top pair production in association with missing energy at @xmath0 colliders are sensitive probes of beyond - the - standard - model physics . we consider standard model ( sm ) extensions containing a new @xmath1 which couples preferably to the most massive states of the sm such as the top quark or dark matter but has suppressed couplings to all the light states of the sm , as inspired by randall sundrum - like setups or theories of partial fermion compositeness . these simple models are poorly constrained by experimental data but lead to striking new signatures at colliders . in this note we consider @xmath2 production in association with a top quark pair in 3 tev @xmath0 collisions at clic , leading to interesting four - top final states and @xmath3 events . [ 1999/12/01 v1.4c il nuovo cimento ]
starburst ( sb ) galaxies in the nearby universe are the best sites to study dust . star formation in the sb galaxies should occur in localized regions , where massive ob stars are deeply embedded in gas clouds that contain significant amounts of dust on average . the bulk of the photons emitted from burst populations are absorbed and partly scattered by dust , and are re - radiated in at far infrared ( fir ) wavelengths . the intrinsic spectral energy distribution ( sed ) of the sb population is thus considerably modified in the uv . this applies not only to the nearby sb galaxies , but also to star - forming galaxies at high redshifts , since a timescale of heavy element synthesis in massive galaxies at an initial burst of star formation should be as short as @xmath4 yr ( e.g. , arimoto & yoshii 1987 ) , and thus galaxies at high redshifts should contain as much dust as the nearby galaxies do . in recent years , there have been many attempts to derive the history of the star formation rate ( sfr ) in the universe , e.g. , by converting the uv light densities to the sfrs of the cfrs ( canada france redshift survey ) galaxies at @xmath5 ( lilly et al . 1996 ) and the hdf ( hubble deep field ) galaxies at @xmath6 ( madau et al . 1996 ; connolly et al . however , without precise estimates of the dust extinction and stellar age , it could be misleading to use the observed uv flux density as an indicator of the sfr , because the uv light is most sensitive to dust obscuration ( meurer et al . 1997 ; rowan - robinson et al . 1997 ; calzetti 1998 ) . unfortunately , the uv - near infrared ( nir ) seds are degenerate in stellar age and dust attenuation ( gordon , calzetti , & witt 1997 ) . complicated analyses of the seds are required to solve this degeneracy by incorporating the dust effects into a model of galaxy spectral evolution . the nearby sb galaxies are most suitable to calibrate the dust effects in actively star forming environments . once the degeneracy is solved , an attempt can be made immediately to determine the ages of lyman - break galaxies ( steidel et al . 1996 ) at high redshifts which are obviously too red to be dust - free star - forming galaxies ( meurer et al . . it would also be possible to derive the age of primeval galaxy candidates in which the bulk of stars of present - day ellipticals formed in a relatively shortlived starburst for the first time ( hughes , dunlop , & rawlings 1997 ) . gordon et al . ( 1997 ) studied the uv - nir seds of 30 sb galaxies which manifest star - forming activity by the strong uv seds and intense line emissions . these are irregular and spiral galaxies with disturbed morphologies locating at a median distance 60 mpc ( hereafter h@xmath7=50 km s@xmath8 mpc@xmath8 ) . gordon et al . ( 1997 ) found that these galaxies show little evidence for a bump at @xmath9 , suggesting that the extinction curve similar to that of the small magellanic cloud ( smc ) is more suitable than that found for the milky way ( mw ) . the best geometry that describes the sb region was found to have the inner dust - free cavity having star clusters surrounded by the dust shell with local clumpy dust condensations . gordon et al . ( 1997 ) tried to get the ages for two sbs by using the uv - nir colors , but failed to determine them uniquely . this suggests that the sb ages can not be determined from the uv - nir seds alone . the aim of this work is to investigate the sb galaxies studied by gordon et al . ( 1997 ) with additional iras data and to derive the accurate age and the dust attenuation of the sb population . in 2 , we briefly describe our model , in 3 we describe our data sample , and in 4 we demonstrate that we are successful in disentangling the burst age and the dust attenuation in the uv - fir seds , and derive the burst parameters for 22 individual sb galaxies . in 5 , we discuss cosmological implications of our results and conclude this article in 6 . even in the sb galaxies , the burst population does not always dominate the galaxy s light , to which the underlying population contributes significantly , and sometimes exceeds the burst light if the sb is already aged . we , therefore , adopt a realistic model of star - forming spirals by superposing the burst population on the underlying galaxy . to model the seds of sb galaxies , both a stellar evolutionary synthesis model and a dust radiative transfer ( rt ) model are needed . we use the stellar evolutionary synthesis model of kodama & arimoto ( 1997 ) , the latest version of arimoto & yoshii ( 1986 ; 1987 ) . we calculate accurately the radiative transfer of the stellar photons through the dust by solving the transfer equations ( takagi , vanseviius , & arimoto 1998 ) . main ideas concerning the structure of sb galaxies are borrowed from witt , thronson & capuano ( 1992 ) and gordon et al . ( 1997 ) . we have adopted a two - component model for the sb galaxy : the galaxy consists of the sb region and the underlying population and is characterized with a free parameter @xmath10 , a mass ratio of the sb to the total galaxy . the radiative transfer and the evolutionary models are calculated for both components independently . we assume that the densities of star and dust in the underlying population follow the modified king s distribution : @xmath11 where @xmath12 means a core radius of the galaxy . the distribution of stars in the sb region is modeled by a similar but steeper function : @xmath13 where @xmath14 means the core radius of the sb region . for the underlying and sb regions , the ratio of the total radius , @xmath15 , to core radius , @xmath16 , is fixed to be @xmath17 . the distribution of dust in the sb is assumed to be homogenous for the entire region . both stellar and dust distributions are integrated up to the same limiting radius . as a test of validity of the geometry of our models , we have calculated a few models following the suggestion by gordon et al . they assumed the sb model as the dusty sphere with the inner dust - free cavity where the stellar population is residing . no significant differences are found for relatively small optical depths @xmath18 , where @xmath19 is integrated from the center to @xmath15 . the model calculations are carried out for @xmath20 in this paper , the main attention is drawn to the two cases @xmath21 and @xmath22 , which means @xmath23 and @xmath24 of galaxy mass are involved in the starburst . the cases with @xmath25 fail to cover the entire region of the observed colors of the sampled galaxies in terms of age and @xmath26 . for the underlying , we adopt typical parameters for an sc galaxy , i.e. , a mass of @xmath27 , and a radius of @xmath28 kpc . if we assume 30% sb mass to the underlying galaxy , it gives @xmath21 , the sb radius @xmath29 kpc , and the mass @xmath30 , respectively , if the the same average density is adopted in the sb region as in the underlying galaxy . the total mass of the galaxy is then @xmath31 . such a size of the sb region is in agreement with the area covered by the diaphragm for the observations presented in gordon et al . ( 1997 ) . the numerous tests with different geometrical parameters reveal only small differences in the modeling results . however , in cases with @xmath32 , the precise simultaneous determination of the age and dustiness of sb is model dependent , and have to be carefully calibrated . the radiative transfer in the galaxy and sb is calculated using the rt code developed by takagi et al . the code is based on an assumption of spherical symmetry with arbitrary radial distributions of stars and dust . direct integration along the rays is adopted . the steps of the mesh ( @xmath33 shells ) are determined automatically , depending on the stellar and dust distributions , to fulfill the energy conservation . the ray tracing is done in the way suggested by band & grindlay ( 1985 ) . the thermal emission from dust is calculated by assuming an equilibrium of dust particles in the radiation field and by taking into account self - absorption . up to six isotropic scattering terms are calculated , and losses of energy due to the scattering do not exceed @xmath34 . the total energy losses in the present models are less than @xmath35 . the isotropic scattering is assumed for simplicity without introducing significant errors , since we analyze optically thick cases . we should note that the thermal emission , which is crucial for the present study , is independent of the adopted phase function . the results were tested by comparing with those obtained by witt , thronson , & capuano ( 1992 ) and the output of rt benchmark by ivezi et al . ( 1997 ) , and were found to be consistent . also no significant discrepancies are found between our results and those obtained by dusty ( ivezi , nenkova & elitzur 1997 ) for the model of a star enshrouded in a dusty envelope with a central dust - free cavity . for the present investigation , @xmath19 is adopted as a free parameter , i.e. , we mimic the amount of dust by changing the optical depth of our models . that means the mass of dust depends on the details of model parameters . this is why we do not derive directly the mass of dust but refer only @xmath19 . any extinction law can be applied in the code , but for the present investigation we limit ourselves with the mw and smc extinction curves ( pei 1992 ) . the albedo is taken from pei ( 1992 ) for both of the extinction curves . we truncate the albedo curve for @xmath36 , without introducing any significant error , in the calculation of the scattered light . the evolutionary synthesis code developed by kodama & arimoto ( 1997 ) is used for the modeling of the luminosity and chemical evolution of the galaxy and sb . the code is based on the up to date ingredients , such as stellar isochrones , spectral libraries , and chemical yields of various elements , and has been successfully used in a numerous publications ( e.g. , kodama et al . 1998 ; kodama & arimoto 1998 ) . the code gives the evolution of synthesized spectra of galaxies ( the wavelength range @xmath37 ) in a consistent manner with the galaxy chemical evolution , i.e. , the effects of stellar metallicity are explicitly considered ( arimoto & yoshii 1986 ) . the underlying population of an sc galaxy is modeled by following the prescription given by arimoto , yoshii & takahara ( 1992 ) , i.e. , taking the values of all free parameters for an infall model ; a sfr @xmath38 gyr@xmath8 , an infall rate @xmath39 gyr@xmath8 , salpeter initial mass function ( imf ) with a slope @xmath40 , and lower and upper stellar masses , @xmath41 and @xmath42 , respectively . we adopt @xmath43 gyr as the age of galaxy because the observed sample consists of nearby galaxies . the sb population is modeled by assuming a closed box model with the sfr of @xmath44 gyr@xmath8 . the sfr adopted is equivalent to that of a giant elliptical galaxy ( arimoto & yoshii 1987 ) . the nebular continuum is neglected in the present models since we limit ourselves to ages @xmath45 myr when the amount of ionizing photons @xmath46 is negligible except for the very beginning of the bursts at @xmath47 myr ( leitherer & heckman 1995 ; fioc & rocca - volmerange 1997 ) . the imf for the sb is assumed to be salpeter - like with @xmath48 , and @xmath41 and @xmath42 , respectively . such a flat imf has been suggested for spheroidal galaxies ( e.g. , arimoto & yoshii 1987 ; kodama & arimoto 1997 ) and for the ob associations in the milky way ( massey , johnson , & degioia - eastwood 1995 ) . the metallicities given by calzetti , kinney , & storchi - bergmann ( 1994 ) for the sb galaxies are attributed to the ionized gas , and the averaged value for the sampled galaxies ( @xmath49 ) is very close to solar . nevertheless , we calculate our models with the _ initial _ metallicity , @xmath50 and @xmath51 , there is no significant differences . therefore , the _ initial _ metallicity of the sb region is adopted to be solar . the seds of the underlying galaxy and the sb region at different ages are shown in fig.1 . the thick solid line represents the underlying at @xmath43 gyr . for the sbs , all seds are plotted for @xmath52 . the optical depth for the underlying galaxy is fixed at @xmath53 throughout the paper . the observational data of sb galaxies are taken from gordon et al . they presented the homogeneous sample of aperture - matched observations of @xmath54 sb galaxies from the far - uv to nir . the size of the diaphragm of the observations is @xmath55 arcsec , which is equal to @xmath56 kpc at their sample s median distance of @xmath57 mpc . we have supplemented the nir colors for @xmath58 galaxies ( ngc1510 , ngc5236 , ngc5253 and ngc7552 ) by taking data from available multiaperture photometry ( de vaucouleurs & longo 1988 ) , which allows us to reduce the observed magnitudes to the effective aperture used by gordon et al.(1997 ) . the iras flux densities are taken from ned ( nasa / ipac extragalactic database ) and have been color corrected . two galaxies ( mrk 309 and mrk 542 ) in the sample have only upper limits for the @xmath59 flux density . it is important to note that the iras colors are derived for the total galaxy while our sample ( uv - nir ) is aperture limited . nevertheless , taking into account the conclusion drawn by calzetti et al.(1995 ) concerning the strong central concentration of fir in the galaxies of the present sample , we believe that the necessary aperture correction does not change significantly our main conclusions . finally we end up with a sample complete in b , h , and iras @xmath57 and @xmath33 @xmath60 fluxes for @xmath61 galaxies . the choice of present colors b , h , and fir is done after an analysis of all the observational data given by gordon et al.(1997 ) . _ only in the case of b - h and @xmath62 , the disentangling of the age and dustiness is successful . _ @xmath63 is the fir luminosity derived from @xmath57 and @xmath33 @xmath60 fluxes ( londsdale et al . 1985 ) . table 1 gives the observations and derived parameters of the sb galaxies : column ( 1 ) gives the name of galaxy , columns ( 2 ) and ( 3 ) indicate b - h and @xmath64 , respectively , and columns ( 4)-(11 ) give the derived age in myr and the optical depth ( see 4 ) for both the mw and smc extinction curves . the comparison of model results with the observed properties of sb galaxies is divided into two steps : a determination of the age and dustiness of sb galaxies , and a comparison of the observed seds with model predictions for the age and the opacity sequences . the calibration of sb galaxies with @xmath21 by the age and dustiness for the mw extinction curve is shown in fig.2a . the data points are taken from table 1 . the large circle indicates the underlying galaxy . an area indicated by a dashed curve represents the region occupied by coma spirals ( gavazzi , randone & branchini 1995 ) . the solid lines show the isochrones with increasing @xmath19 from the left ( @xmath53 ) to the right ( @xmath65 ) . the dot - dashed lines show the isoopaques with increasing age of the sb from the top ( @xmath66 myr ) to the bottom ( @xmath67 gyr ) . _ it is obvious that fir luminosity of galaxies helps to break the age - dustiness degeneracy , and we have the unique solution for the sbs with @xmath68_. the range of @xmath19 validity depends on @xmath10 assumed . fig.2b shows the same diagram as fig.2a but for @xmath69 , i.e. , the galaxy has 100% sb population . as we can see in fig.2b , the applicable range of @xmath19 is significantly larger , but the shape of the grid remains very similar to the case shown in fig.2a . therefore , we expect that the assumed value of the free parameter @xmath21 does not influence our conclusions significantly . the derived ages and optical depths @xmath19 for individual sb galaxies are given in table 1 for @xmath21 and @xmath70 . we note that the burst age and the opacity are nearly the same for different values of @xmath10 . the resulting age ranges from @xmath66 myr to @xmath71 myr , which agrees well with gordon et al.s ( 1997 ) estimate for the present sample of sbs . figs.2c , d are the same as figs.2a , b , respectively , but for the smc extinction curve . we note that the burst age and the opacity are nearly the same for different extinction laws . _ we therefore can conclude that the present method is robust and does not depend on the details of extinction curve and the burst strength . _ the parameters given in table 1 are used for grouping of the sb galaxies by the age and the opacity . two groups of galaxies are defined : ( a ) those with approximately the same ages ( @xmath72 myr ) but different optical depths ( @xmath73 ) , and ( b ) those with similar optical depths ( @xmath74 ) but different ages ( @xmath75 myr ) . the seds inside these groups are averaged and plotted in figs.3a - h . the model seds are superposed on the observational data without any fitting , but both are normalized at the h band to @xmath22 . figs.3a - d and figs.3e - h represent the two different extinction curves ( mw and smc , respectively ) . the seds in figs.3a , b , e , f represent @xmath26 sequences and in figs.3c , d , g , h age sequences . the influence of the extinction law is significant in both the uv and fir spectral regions . the behavior of the sed in uv can be predicted from the shape of the extinction curve only in the case of @xmath76 . a strong nonlinearity is observed for @xmath77 in case of @xmath21 , where the stronger influence of @xmath10 than of the extinction law on the uv part of sed is observed , since the significant obscuration in the uv hides sb population and the underlying galaxy , which is characterized by a mw extinction curve , becomes dominant . fir emission is affected more regularly by the extinction curve and the burst strength , i.e. , the fir part of sed becomes cooler for the smc extinction law and brighter for @xmath69 . as we can judge from the far - uv end of the seds , the extinction law is steeper than mw but flatter than smc . therefore , calzetti et al . ( 1994 ) and gordon et al . ( 1997 ) conclusions concerning the far - uv shape of the extinction curve in sb galaxies are confirmed . it is important to note that the _ observed _ difference in the uv part of sed is much smaller for the age sequence than for the @xmath26 sequence , but the trend is opposite in the fir region . along the @xmath26 sequence , the uv - nir part of sed becomes fainter , while the fir part becomes more luminous . on the other hand , both the uv - nir and fir parts of sed becomes fainter along the age sequence as the stellar population gets older . figs.3a - h show that b , h and fir colors in general give a reasonable estimation for the seds of sb galaxies from the uv to fir . nevertheless , we have to stress that figs.3a - h are given without any fitting of the parameters , i.e. , they represent a general behavior of our models used to determine the age and @xmath26 . moreover , the observational data are averaged for different age and dustiness groups . the fine tuning of model parameters for each particular galaxy is a subject for a forthcoming paper . gordon et al . ( 1997 ) suggested that the dust distribution in sb is best described by a clumpy media , which is not adopted in our modeling . however , we should note that our figs.2a - d would not change even if we take into account clumpy media , because they can be scaled in b - h and @xmath62 independently of the dust attenuation ; thus our arguments are safe so far as we do not discuss the mass of dust which is model dependent . moreover , witt & gordon ( 1996 ) suggested that the high conversion efficiency of uv / optical radiation into far - ir thermal dust emission observed in sb galaxies and mergers is due in large part to a change in the structure of the interstellar medium in such systems toward a _ more homogeneous widely extended distribution_. taking into account this suggestion and the curves corresponding to homogeneously extended distribution in their fig.19 ( the density ratio of interclump - to - clump media , @xmath78 ) , we can expect that our definition of @xmath26 for homogeneous distribution of dust is linearly related with their @xmath26 calculated for the clumpy media . figs.3a - h clearly reveal two problems : a discrepancy of the model seds and the observed ones in uv region , and a difference of the predicted and observed fluxes in mid - infrared region ( mir ) . the first point was discussed extensively by gordon et al . we have confirmed their conclusion that the extinction curve in the sb galaxies is steeper than in mw . therefore , from the comparison of the ages and @xmath26 s in table 1 , we conclude that the uncertainty of the extinction curve in the uv is not serious for the present study . however , the deviation of the mw and smc extinction curves adopted by us from the real one can not significantly change our results , because the main diagrams ( figs.2a - d ) are based on the bluest band b which is not influenced yet by the uv steepening of the extinction curve , and because the supplementary amount of the energy absorbed in far - uv does not increase significantly the fir flux ( trewhella , davies & disney 1997 ) . the second point , probably , can be explained by the shortcomings of our rt model , because the circumstellar dust and emission from very small grains are not included yet . the circumstellar dust mainly influences @xmath79 flux and is dependent on the number of stars with dusty shells in galaxy . the influence of small grains on the iras colors was discussed by boulanger et al . they concluded that iras @xmath80 and @xmath81 colors depend on the non - equilibrium emission from small grains while @xmath57 and @xmath59 colors depend on the equilibrium emission from large grains . extensive discussion on the applicability of so called classical grain model to fit the mir ( @xmath82 ) fluxes from the interstellar medium was given by helou , ryter & soifer ( 1991 ) . therefore , our model represents correctly the fir emission ( @xmath57 and @xmath59 ) from the grains in thermal equilibrium , which is calculated explicitly by our rt code . moreover , we are using only the fir colors for age - dustiness determination , and can neglect the discrepancies between the predicted seds and the observed colors in the mir region . figs.3a - h indicate that the observed fir flux ratios @xmath83 ( dust temperature indicator ) are smaller than the model predictions . this ratio depends on the dust temperature and on the emissivity law of dust grains . the dust temperature depends on the distribution of stellar density , which is rather approximate in our model , a flatter one would result in the cooler dust temperature . we assume the emissivity with the power @xmath84 ( @xmath85 , rowan- robinson , 1992 ) if @xmath86 is taken , the discrepancy of the flux ratio would reduce ( hughes et al . however , such a fine tuning of the model is beyond a scope of this article , since both effects would not change significantly the total fir flux used for breaking the age - dustiness degeneracy . we have demonstrated that the use of seds from far - uv to fir is quite effective in disentangling the age and opacity of the sb galaxies . for deriving the opacity , calzetti , kinney , & storchi - bergmann ( 1996 ) and calzetti ( 1997 ) assumed that the intrinsic line ratios are given by a simple model with the optically thick limit , the so - called case b of osterbrock ( 1989 ) , and estimated the opacity from the ratios of attenuated hydrogen emission lines . although the opacities derived by two methods are correlated , the opacity given by the line method is smaller than the fir method . this in turn should give systematically older ages for the line method . one should keep in mind that some of the sb galaxies assigned for ages younger than 20 myr could well be much younger if their colors are significantly reddened by possible nebular continuum ( fioc & rocca - volmerange 1997 ) . however , even if this is the case , the probability for galaxies younger than 10 myr to reside in this region should be very small , because of short timescale for that stage . additionally , if galaxies are uv - selected as in gordon et al . ( 1997 ) , it would be almost impossible to find such very young sb galaxies due to strong dust obscuration . before we conclude this article , we briefly discuss two cosmological implications of the present results . first , if we define the primeval galaxies as massive galaxies at the onset of the first episode of major star formation , presumably progenitors of the present - day ellipticals at high redshifts , these galaxies should look like the sb galaxies in which burst population dominates the galactic light . thus , it should be possible to determine the age and optical depth of the primeval galaxy candidates with a help of the rest - frame b - h _ vs. _ @xmath62 diagram . these galaxies should be spectacular in the rest - frame fir regime . at the highest redshifts , the fir @xmath57 and @xmath59 fluxes conveniently fall on into the region of submillimeter atmospheric windows at @xmath87 , @xmath88 , @xmath89 and @xmath90 . therefore , a search for the primeval galaxies with infrared satellites and submillimeter telescopes , such as iris ( infrared imaging surveyor ; murakami et al . 1998 ) and lmsa ( large millimeter submillimeter array ; ishiguro et al . 1994 ) , would be promising . indeed , the sensitivity of submillimeter arrays is expected to achieve the greatest advances in the immediate future ( hughes et al . 1997 ) . second , once the rest - frame b - h and @xmath62 are observed for high redshift galaxies including the lyman - break galaxies ( steidel et al . 1996 ) , radio galaxies ( dunlop et al . 1996 ) , and qso host galaxies , the ages and optical depths of these galaxies can be determined uniquely . then , the resulting age should play the role of a cosmological clock that gives the age - redshift relation of galaxies at high redshifts and will constrain the cosmological picture of the universe . the resulting @xmath2 should in turn give a correction factor for the uv light , which will make it possible to derive the correct sfr at high redshifts . in other words , it could be misleading to derive the sfr for high redshift galaxies from the uv light alone . in an attempt to break the age - dustiness degeneracy of the seds of starburst galaxies , we have built a new spectral evolution model of galaxy that included both stellar and dust effects . the evolutionary code of kodama & arimoto ( 1997 ) is used to model the spectral and chemical evolution of galaxies , and the radiative transfer is calculated using the rt code developed by takagi et al . ( 1998 ) . applying the model to 22 nearby sb galaxies sampled by gordon et al . ( 1997 ) , we have shown that the unique solution can be sought if the fir luminosity is used for the starbursts with @xmath91 . except for two opaque galaxies with @xmath92 , the resulting sb ages and opacities are in the range @xmath0 and @xmath1 , when the burst mass fraction @xmath21 is assumed . the result is robust and almost independent of the details of the dust distribution , the extinction curve , and the burst strength . even in a case of the sb dominant galaxies , the age and @xmath2 are uniquely assigned on the rest - frame b - h _ vs. _ @xmath62 diagram . for high redshift galaxies , these wavelength regions conveniently fall on into the nir , mir , and submillimeter regions . with the rapidly growing sensitivity of the submillimeter detectors , it should soon become possible to determine the age and @xmath2 of star - forming galaxies at redshifts @xmath3 and beyond . accurate estimates of @xmath2 for the high redshift galaxies might require a substantial revision of the broad picture of star formation history over the hubble time proposed by madau et al . ( 1996 ) . we are very grateful to the referee , k.d.gordon , for helping us to improve the paper significantly . wish to thank to t.shibazaki and r.hoshi for their continuous support and fruitful discussions . 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 work was financially supported in part by a grant - in - aid for the scientific research ( no.0940311 ) by the japanese ministry of education , culture , sports and science . arimoto , n. , & yoshii , y. 1986 , a&a , 164 , 260 arimoto , n. , & yoshii , y. 1987 , a&a , 173 , 23 arimoto , n. , yoshii , y. , & takahara , f. 1992 , a&a , 253 , 21 band , d.l . , & grindlay , j.e . 1985 , apj , 298 , 128 boulanger , f. , beichman , c. , desert , f.x . , helou , g. , perault , m. , & ryter , c. 1988 , apj , 332 , 328 calzetti , d. 1998 , astro - ph/9806083 calzetti , d. 1997 , aj , 113 , 162 calzetti , d . , bohlin , r.c . , kinney , a.l . , storchi - bergmann , t. , & heckman , t.m . 1995,apj , 443 , 136 calzetti , d . , kinney , a.l . , storchi - bergmann , t. 1994 , apj , 429 , 582 calzetti , d . , kinney , a.l . , storchi - bergmann , t. 1996 , apj , 458 , 132 connolly , a.j . , szalay , a.s . , dickinson , m. , subbarao , m.u . , & brunner , r.j . 1997 , apjl , 486 , l11 de vaucouleurs , a. , & longo , g. 1988 , catalogue of visual and infrared photometry of galaxies from 0.5 micrometer to 10 micrometer ( 1961 - 1985 ) , university of texas monographs in astronomy , austin : university of texas dunlop , j.s . , peacock , j. , spinrad , h. , dey , a. , jimenez , r. , stern , d. , & windhorst , r. 1996 , nature , 381 , 581 fioc , m. , & rocca - volmerange , b. 1997 , a&a , 326 , 950 gavazzi , g. , randone , i. , & branchini , e. 1995 , apj , 438 , 590 gordon , k.d . , witt , a.n . , carruthers , g.r . , christensen , s.a . , & dohne , b.c . 1994 , apj , 432 , 641 gordon , k.d . , calzetti , d. , witt , a.n . 1997 , apj , 487 , 625 helou , g. , ryter , c. , & soifer , b.t . 1991 , apj , 376 , 505 hughes , d.h . , dunlop , j.s . , & rawlings , s. 1997 , mnras , 289 , 766 ishiguro , m. , kawabe , r. , nakai , n. , morita , k .- i . , okumura , s.k . , & ohashi , n. 1994 , in astronomy with millimeter and submillimeter wave interferometry , iau cool . 140 , asp conf . ser . , 59 , 405 ivezi , . , groenewegen , m.a.t . , menshchikov , a. , & szczerba , r. 1997 , mnras , 291 , 121 ivezi , . , nenkova , m. , & elitzur , m. 1997 , user manual for dusty , university of kentucky kodama , t. , & arimoto , n. 1997 , a&a , 320 , 41 kodama , t. , & arimoto , n. 1998 , mnras , in press , astro - ph/9806029 kodama , t. , arimoto , n. , barger , a.j . , & aragn - salamanca , a. 1998 , a&a , 334 , 99 leitherer , c. , & heckman , t.m . 1995 , apjs , 96 , 9 londsdale , c.j . , helou , g. , good , j.c . , rice , w. 1985 , cataloged galaxies and quasars observed in the iras survey , jet propuls . lab . , pasadena , california lilly , s.j . , le fvre , o. , hammer , f. , & crampton , d. 1996 , apj , 460 , l1 madau , p. , ferguson , h.c . , dickinson , m.e . , giavalisco , m. , steidel , c.c . , & fruchter , a. 1996 , mnras , 283 , 1388 massey , p. , johnson , k.e . , & degioia - eastwood , k. 1995 , apj , 454 , 151 meurer , g.r . , heckman , t.m . , lehnert , m.d . , leitherer , c. , & lowenthal , j. 1997 , aj , 114 , 54 murakami , h. , et al . 1998 , isas research note no.651 osterbrock , d.e . 1989 , astrophysics of gaseous nebulae and active galactic nuclei , university science books , new york rowan - robinson , m. 1992 , mnras , 258 , 787 rowan - robinson , m. , et al . 1997 , mnras , 289 , 490 pei , y.c . 1992 , apj , 395 , 130 steidel , c.c . , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k.l . 1996 , apjl , 462 , l17 takagi , t. , vanseviius , v. , arimoto , n. 1998 , to be submitted to pasj . trewhella , m. , davies , j.i . , & disney , m.j . 1997 , proc . cold dust morphology , conf . at wits university in johannesburg , south africa , 22 - 26th january 1996 witt , a.n . , thronson , h.a . , & capuano , j.m . 1992 , apj , 393 , 611 witt , a.n . , & gordon , k.d . 1996 , apj , 463 , 681 ngc 1140 & 1.89 & 0.72 & 25 & 30 & 0.7 & 0.8 & & 30 & 40 & 0.5 & 0.6 ngc 1510 & 2.74 & 0.22 & 400 & 500 & 1.1 & 0.9 & & 400 & 500 & 1.1 & 0.8 ngc 1569 & 1.37 & 1.17 & 10 & 15 & 1.0 & 1.1 & & 10 & 15 & 0.8 & 0.9 ngc 1614 & 3.97 & 1.14 & 30 & 25 & @xmath93 & 4.3 & & 30 & 30 & 5.0 & 4.0ngc 4194 & 3.29 & 1.01 & 25 & 30 & 3.6 & 2.9 & & 30 & 30 & 3.3 & 2.6 ngc 4385 & 3.27 & 0.66 & 100 & 130 & 2.4 & 2.2 & & 100 & 200 & 2.2 & 2.1 ngc 5236 & 2.93 & 1.03 & 20 & 25 & 2.9 & 2.5 & & 25 & 30 & 2.6 & 2.2 ngc 5253 & 1.80 & 1.31 & 10 & 15 & 2.0 & 1.9 & & 10 & 15 & 1.8 & 1.6 ngc 5860 & 2.79 & 0.55 & 130 & 200 & 1.5 & 1.5 & & 200 & 300 & 1.4 & 1.4 ngc 6052 & 2.29 & 1.11 & 15 & 20 & 2.1 & 1.9 & & 20 & 25 & 1.9 & 1.7 ngc 6090 & 3.22 & 1.02 & 25 & 30 & 3.6 & 2.9 & & 25 & 30 & 3.3 & 2.6 ngc 6217 & 3.31 & 0.84 & 50 & 60 & 2.8 & 2.5 & & 50 & 80 & 2.6 & 2.4 ngc 7250 & 2.09 & 0.92 & 20 & 25 & 1.3 & 1.3 & & 25 & 30 & 1.1 & 1.1 ngc 7552 & 3.69 & 1.02 & 30 & 30 & 5.0 & 3.5 & & 30 & 30 & 4.5 & 3.1 ngc 7673 & 2.17 & 0.94 & 20 & 25 & 1.4 & 1.4 & & 25 & 30 & 1.2 & 1.2 ngc 7714 & 2.64 & 0.88 & 30 & 30 & 1.9 & 1.8 & & 30 & 40 & 1.7 & 1.6 mrk 309 & 2.76 & 1.43 & 10 : & 15 : & 5.0 : & 3.5 : & & 10 : & 15 : & 5.0 : & 3.1 : mrk 357 & 2.01 & 0.91 & 20 & 25 & 1.2 & 1.2 & & 25 & 30 & 1.0 & 1.0 mrk 542 & 2.79 & 0.40 & 400 : & 400 : & 1.5 : & 1.3 : & & 300 : & 400 : & 1.3 : & 1.2 : ic 214 & 4.18 & 0.92 & 100 & 50 & @xmath93 & 4.0 & & 100 & 80 & 5.0 & 3.6 ic 1586 & 2.49 & 0.59 & 80 & 100 & 1.1 & 1.2 & & 100 & 100 & 1.0 & 1.0 haro 15 & 2.10 & 0.74 & 30 & 30 & 0.9 & 1.0 & & 30 & 40 & 0.7 & 0.8	a spectral evolution model of galaxies that includes both stellar and dust effects is newly built . xapplying the model to 22 nearby starburst galaxies , we have shown that far infrared luminosity of galaxies helps to break the age - dustiness degeneracy . we have derived a unique solution of age and the dustiness for each starburst galaxy . the resulting starburst ages and optical depths are in the range @xmath0 and @xmath1 , respectively . the result is robust and is almost independent of model assumptions such as dust distributions , extinction curves , and burst strengths . with the rapidly growing sensitivity of submillimeter detectors , it should become possible in the near future to determine the age and @xmath2 of star - forming galaxies at redshifts @xmath3 and beyond . accurate estimates of @xmath2 for lyman - break galaxies and high - z galaxies might require a substantial revision of the previously claimed picture of star formation history over the hubble time .
compton scattering between photons and electrons is an important process in astrophysics . if the photons are not produced at the same place where the electrons are located , we call it `` global compton scattering '' . momentum and energy of photons and electrons can be exchanged in this process and these two aspects often play an important role in determining the dynamics of the gas flow . on the galactic scale , this so - called radiative feedback mechanism now is believed to be crucial for understanding agn feedback on galaxy formation and evolution ( e.g. , ciotti & ostriker 2001 , 2007 ; murray et al . 2005 ; hopkins et al . 2005 ) . on a smaller scale , the effect of compton scattering on the dynamics of gas flows surrounding a strongly radiating quasar has been investigated and outflow is produced as a consequence of compton heating ( e.g. , proga , ostriker & kurosawa 2008 ) . following the earlier work of krolik et al . ( 1981 ) , mathews & ferland ( 1987 ) considered the compton heating effect for the broad - line region of quasar . this effect is also important for the standard thin disk , if the disk is warped or irradiated by a source above the disk plane ( e.g. , shakura & sunyaev 1973 ; begelman , mckee & shields 1983 ; dubus et al . if the accretion flow is geometrically thick and optically thin , the photons can travel a large distance without being absorbed , therefore the global compton scattering effect is in principle important . this is the case for spherical accretion and hot accretion flows . the latter includes the advection - dominated accretion flow ( adaf ; narayan & yi 1994 ; 1995 ) and luminous hot accretion flow ( lhaf ; yuan 2001 ; 2003 ) , two types of hot accretion flows corresponding to low and high accretion rates , respectively . for spherical accretion , the interaction of momentum between photons and electrons sets up a largest possible luminosity the accretion flow could reach , namely the eddington luminosity @xmath12 ( but this limit does not apply when the accretion flow has a non - zero angular momentum ; see , e.g. , ohsuga & mineshige 2007 ) . the effect of the energy interaction between photons and electrons in a spherical accretion flow has been investigated by ostriker et al . it was found that when the luminosity is larger than a certain value the outward energetic photons could heat gas flow significantly so that the local sound speed is larger than the escape speed , or , in other words , the temperature is higher than the virial temperature , thus the accretion is suppressed ; and this effect due to energy input from the outgoing radiation field occurs for much lower luminosities than the momentum ( eddington ) limit . in almost all of the previous work on the dynamics of hot accretion flows , only the `` local '' compton scattering effect has been considered while the global compton effect has been neglected . here `` local '' means that photons are produced at the same region with where the electrons locate . this local compton scattering serves as the main cooling mechanism of electrons ( compton cooling ) and the main mechanism of producing x - ray emission ( thermal comptonization ) . the only works to our knowledge considering the global compton scattering are esin ( 1997 ) and park & ostriker ( 1999 ; 2001 ; 2007 ) . esin ( 1997 ) deals with a one - dimensional adaf and finds that the global compton heating / cooling is not important and can be neglected . park & ostriker ( 2001 ; 2007 ) deal with a two - dimensional flow and focus on the possible production of outflow in the polar region because of strong compton heating there . their conclusion is that compton heating effect is important in many cases , which is quite different from esin s result . they do not attempt to obtain the self - consistent solutions . all their work are based on the self - similar solution of adaf ( narayan & yi 1994 ; 1995 ) . the discrepancy between esin and park & ostriker is likely due to the additional but different assumptions adopted . while the self - similar approximation is quite successful in catching the main spirit of an adaf , it is not a good approximation when we want to calculate the radiation since order of magnitude error could be produced . this is because the self - similar approximation breaks down at the inner region of the adaf where most of the radiation comes from . when we consider the effect of global compton heating / cooling , obviously it is crucial to calculate the exact spectrum from the exact global solution of the accretion flow . this is the main motivation of the present paper . in addition , important theoretical progress on adaf solutions have been made since its discovery . the two most important ones are the presence of outflow ( e.g. , stone & pringle 2001 ) and significant direct electron heating by turbulent dissipation ( e.g. , quataert & gruzinov 1999 ) , which is much stronger than the heating by coulomb collisions between ions and electrons . both of them are important in determining the dynamics of adafs while they have not been properly taken into account in the above works . in the present paper we will focus on the one - dimensional case , given that the uncertainty of our understanding to two - dimensional structure of accretion flow is still large . of course , a full understanding to the two - dimensional case is obviously important and should be a topic of future research . consider a canonical hot accretion flow without taking into account the global compton heating or cooling . outflow is taken into account by adopting the following radius - dependent mass accretion rate ( e.g. , blandford & begelman 1999 ) : = -4r hv=_0()^s , where @xmath13 is the mass accretion rate at the outer boundary @xmath6 . the value of index @xmath14 describes the strength of the outflow and we use @xmath15 from the detailed modeling of sgr a * , the supermassive black hole in our galactic center ( yuan , quataert & narayan 2003 ) . the energy equations for ions and electrons are v ( - p_i ^2 ) = ( 1-)q^+-q_ie , v ( - p_e ^2 ) = q^++q_ie - q^- , where @xmath16 is the internal energy of electrons and ions per unit mass of the gas , @xmath17 is the coulomb energy exchange rate between electrons and ions , @xmath18 is the electron cooling rate , including synchrotron and bremsstrahlung emissions and their local comptonization , @xmath19 is the net turbulent heating rate , the value of @xmath20 describes the fraction of turbulent heating which directly heats electrons and we use @xmath21 again from the modeling of sgr a * ( see also sharma et al . 2007 ) . we first get the exact global solution of the hot accretion flow so that we know all the quantities such as density and temperature as a function of radius . this requires us to solve the set of equations describing the conservations of mass ( eq . 1 ) , energy ( eqs . 2 & 3 ) , and momentum ( ref . yuan , quataert & narayan 2003 ) . the global solution should satisfy the outer boundary condition at the outer boundary @xmath6 , the inner boundary condition at the horizon , and a sonic point condition at the sonic point . to calculate the rate of `` global '' compton heating / cooling of electrons at a given radius @xmath22 , we need to know the spectrum received at @xmath22 emitted by the whole flow . this requires us to solve the radiative transfer equations along the radial direction , which is complicated when scattering is important . for simplicity , here we deal with the scattering in a simple way and write the received spectrum at @xmath22 emitted by the flow inside of @xmath22 as , f_^in(r)=^r_r_se^- r here @xmath23 is the scattering optical depth from @xmath24 to @xmath22 , @xmath25 , and @xmath26 is the emitted monochromatic luminosity from a shell at @xmath24 with thickness @xmath27 and height @xmath28 . it includes synchrotron and bremsstrahlung emissions and their local comptonization . we approximate the calculation of the unscattered part of @xmath26 by solving the radiative transfer along the vertical direction of adafs adopting a two - stream approximation ( manmoto , mineshige & kusunose 1997 ) : dl_^un(r)= b_[1-exp(-2_^*)]rdr here @xmath29 denotes the planck spectrum , @xmath30 is the optical depth for absorption in the vertical direction with @xmath31 being the absorption coefficient on the equatorial plane . the free - free absorption and synchrotron self - absorption are included in this way . the strength of the magnetic field in the accretion flow is determined by a parameter @xmath32 defined as the ratio of the gas pressure to the magnetic pressure and we set @xmath33 . for the calculation of the compton scattered part of @xmath26 , we use the approach of coppi & blandford ( 1990 ; eq . the integration in eq . ( 4 ) begins from the black hole horizon @xmath34 . the spectrum received at @xmath22 emitted by the flow outside of @xmath22 is ( park & ostriker 2007 ) , f_^out(r)=^r_out_r dr the total spectrum received at @xmath22 is the sum of @xmath35 and @xmath36 . we assume that the electrons have a maxwell distribution with temperature @xmath37 ( @xmath38 ) and the energy of the photon before scattering is @xmath39 . since the electrons in the hot accretion flow are relativistic at the innermost region and the peak photon energy @xmath40 , to calculate the average energy of a scattered photon , we use the following exact form which is valid for any photon energy and electrons temperature ( guilbert 1986 ) : @xmath41 with @xmath42 and the cross - section for scattering ( , _ e)= ^+_-g_0(e^ ) e^2exp()d . here @xmath43 is a modified bessel function of second order and : g_n(y ) ^2_0 ( t(t-2)+1+ty+ ) . in the thompson limit , eqs . ( 7 ) & ( 8) are transformed into the familiar form of = + , and ( , _ e)=_t . the number of scattering in a region of the accretion flow with unit width in the radial direction and scattering optical depth @xmath44 is n=_es , with @xmath45 and @xmath46 are the temperature and number density of electrons in that region . the compton heating ( cooling ) rate in that region ( with unit radial length ) is then q_comp = n [ f_^in(r)+f_^out(r ) ] d. note that in the above equation we actually use the moment of intensity `` @xmath47 '' not @xmath48 . following park & ostriker ( 2007 ) , we formally define a `` radiation temperature '' ( or `` compton temperature '' ; see also levich & sunyaev 1970 ; krolik , mckee & tarter 1981 ) : _ x.under this definition , whether the compton scattering plays a heating or cooling effect roughly depends on whether the electron temperature @xmath45 is larger or smaller than @xmath49 . in the thompson limit , the compton heating / cooling rate is exactly proportional to @xmath50 . it is important to note here that the radiation temperature is obtained from the flux distribution by weighting with the factor @xmath51 . it physically represents the equilibrium between compton heating and cooling . for a typical quasar spectrum , it corresponds to @xmath52k or several kev where the bulk of the radiation is emitted in the uv or , in some cases , ir portions of the spectrum ( mathews & ferland 1987 ; sazonov , ostriker & sunyaev 2004 ) . as we will see below ( e.g. , ref . 3 ( c ) ) , the radiation temperature of an adaf spectrum is much higher because of its different spectrum . + the dominant heating term of electrons in eq . ( 3 ) is @xmath53 . we compare the rate of global compton heating / cooling with @xmath54 and the results are shown in fig . 1 for @xmath55 with @xmath56 , and @xmath57 ( @xmath58 ) . at large radii , @xmath3 , compton scattering heats electrons ; while at small radii , @xmath59 , it cools electrons . this is of course because the radiation temperature @xmath49 is lower ( higher ) than the electron temperature @xmath60 at the small ( large ) radii , as shown by the right panel of fig . 1 . + we can see from the left panel of fig . 1 that the compton effect is important when @xmath61 . in this case , its cooling effect can not be neglected . the corresponding accretion rate at the black hole horizon is @xmath62 and the corresponding luminosity is @xmath63 . the `` lowest '' value of @xmath13 above which compton heating effect is important is a function of @xmath6 . for @xmath64 , this value is @xmath65 and the corresponding luminosity is @xmath66 . when @xmath6 is larger , the value of critical @xmath13 is lower . in reality @xmath6 usually has a largest feasible value . if the adaf starts out from a transition from a standard thin disk , @xmath6 equals to the transition radius . if on the other hand the accretion flow starts out as an adaf such as in our galactic center , @xmath6 should be determined by the bondi radius . outside of bondi radius , the effect of compton heating is not so clear , because matching an adaf solution to one with proper boundary condition at infinity is an unsolved problem . our result that compton scattering heats electrons at large radii while cools electrons at small radii is qualitatively consistent with both esin ( 1997 ) and park & ostriker ( 2001 ; 2007 ) . however , esin ( 1997 ) found that compton heating rate is always even smaller than the coulomb collision heating rate therefore is negligible . this is different from our results and park & ostriker ( 2001 ; 2007 ) . the reason may comes from some over - simplifications and the different ( old ) adaf model adopted in esin ( 1997 ) . overall , we see that for extended solutions ( @xmath67 ) , both compton cooling in the inner parts and compton heating in the outer parts dramatically alter the solutions when @xmath68 . the above result indicates that we should take into account the effect of global compton heating / cooling when we calculate the global solution of the hot accretion flow when @xmath69 is relatively large . this has not been studied in esin ( 1997 ) and park & ostriker ( 1999 ; 2001 ; 2007 ) . we use an iteration method to achieve this . we first solve the global solution without considering the global compton effect , calculating the rate of compton heating / cooling at each radius as described above , @xmath70 . we then include this term in the energy equations of electrons : v ( - p_e ^2 ) = q^+ + q_ie - q^- + q_comp , and calculate the `` new '' global solution of the accretion flow based on this `` new '' equation . then we get a new compton heating / cooling rate . if the new rate is not equal to the guessed value we replace the guessed value with the new one and repeat this procedure until they are equal . however , we must emphasize that the solution obtained by the above approach is actually not exactly `` self - consistent '' . we use eq . ( 13 ) to calculate @xmath70 . but in eq . ( 13 ) , only the local compton scattering is considered when calculating @xmath71 , and the global scattering is difficult to include because we do nt know the spectrum emitted at other radii which again requires to consider global scattering . this is difficult to deal with even using the iteration approach . the best way to solve this problem is by monte carlo simulation combined with iteration method . this is beyond the scape of this paper and will be our next work . on the other hand , we believe our result should be a good zeroth - order approximation to the real solution . bearing this in mind , figs . 2&3 show the calculation results . figs . 2 ( a c ) are for a stellar mass black hole with black hole mass @xmath72 and @xmath73 . fig . 2(a ) shows the electron temperature of the global solution without ( solid line ) and with ( dashed line ) the global compton scattering effect included . because compton scattering plays a cooling role at small radii , we see that the electron temperature decreases after the compton effect is taken into account as we expect . 2(b ) shows the ratio of compton heating and local viscous heating of the electrons , @xmath74 , before ( solid line ) and after ( dashed line ) the global compton effect is included . it is interesting to note that at @xmath75 , the absolute value of this ratio is between @xmath76 . since we typically have @xmath77 , this implies that the right - hand - side of eq . ( 15 ) is negative in that region , i.e. , the viscous heating of electrons is smaller than its radiative cooling ( @xmath78 . in another words , the energy advection of electrons plays a heating role , just like the ions in the lhaf solution ( yuan 2001 ) . in the inner region of @xmath79 , where most of the radiation comes from , the absolute value of @xmath80 . we find in this case @xmath81 . so in the innermost region the viscous heating of electrons is equal to its radiative cooling ( @xmath78 . obviously , after taking into account the global compton cooling , the radiative efficiency will increase for a given accretion rate . but this does not mean that the highest luminosity @xmath82 a hot accretion flow can produce will increase . the main heating mechanism of electrons are viscous heating and compression work ( the second term in the right - hand side of eq . 15 ) while the main cooling comes from ( @xmath83 ) . the highest accretion rate beyond which a hot solution no longer exists is determined by the balance between heating and cooling . the heating term is roughly proportional to @xmath69 while the cooling term roughly to @xmath84 since compton scattering is a two - body collision process . this is why a hot accretion solution has a highest @xmath69 . obviously , when the global compton cooling @xmath70 is included , the cooling becomes stronger compared to the case of only local cooling @xmath18 , thus the balance between cooling and heating will occur at a lower @xmath69 . actually @xmath85 as shown in fig . 2 is almost the highest accretion rate at which we can get the self - consistent hot solution , which is a factor of 2 - 3 lower than the highest rate when the global compton effect is not taken into account . when @xmath13 is higher , we find that we are not able to get the self - consistent solution since the flow will collapse due to the strong radiative cooling . the decrease of the highest @xmath69 results in the decrease of @xmath82 . our calculation shows that @xmath82 decreases by a factor of @xmath86 2 , i.e. , we now have @xmath87 . we now check how different the spectrum produced by the self - consistent solution is compared to the spectrum produced by the `` old '' solution . 2(c ) shows the spectra from the hot accretion flow before ( solid line ) and after ( dashed line ) the compton effect is taken into account . we see that both the luminosity and the cutoff energy of the spectrum ( i.e. , the corresponding frequency of the peak of the spectrum ) decrease because of the global compton cooling . this is of course because the electron temperature of the self - consistent solution decreases compared to the `` old '' solution . we would like to emphasize again that only local seed photons are considered when we calculate the spectrum although we do consider the global scattering in calculating the dynamics . our calculation shows that @xmath88 , so we expect that when the global compton scattering is considered the luminosity of the `` exact '' self - consistent solution will be @xmath89 times higher than that shown by the dashed line in fig . but the slope and the cutoff energy will not change because they are irrelevant to the amount of seed photons . 3 ( a c ) are similar to figs . 2 ( a c ) , but are for a supermassive black hole with @xmath90 and accretion rate @xmath91 . we see from the figures that the electron temperature decreases after the global compton scattering effect is taken into account as we expect , because in most region the compton scattering will cool the electrons . correspondingly , the luminosity of the accretion flow also decreases by roughly a factor of 2 and the cutoff energy of the spectrum also becomes smaller . we have also calculated the average energy of the photons emitted by the self - consistent solution shown in fig . 3 ( c ) , @xmath92 , and the corresponding energy of the radiation temperature @xmath49 at @xmath93 , @xmath94 . the results are @xmath95 1 ev and 100 kev respectively , and they are shown by two arrows in fig . these values are much higher than that of a typical quasar spectrum where , e.g. , @xmath96 is only several kev ( mathews & ferland 1987 ; sazonov et al . 2004 ) . the spectrum shown in fig . 3 ( c ) extends to very high energy , @xmath97 mev . observationally , the @xmath98-folding energy of the average power - law x - ray spectrum observed by _ ginga _ , osse , and _ exsosat _ of radio - quiet seyfert 1s is @xmath99 mev ( zdziarski et al . 1995 ; gondek et al . 1996 ) , which is consistent with the model given the ( large ) error bar . better data is required to constrain the theoretical model . as we state in 2.2 , compton heating effect at large radii is another obstacle for us to obtain the self - consistent solution . for @xmath100 , if @xmath101 , we find that the compton heating effect around @xmath6 is so strong that the equilibrium temperature of electrons would be higher than the virial value defined as @xmath102 , which will in turn make the ion temperature also higher than the virial one due to the coulomb coupling between them . in this case , the gas is unbound thus can not be accreted . the corresponding highest luminosity in this case is @xmath103 . this value is similar to that obtained by ostriker et al . ( 1976 ) and park & ostriker ( 2001 ) . from figs . 1 & 3(b ) , we expect that @xmath13 and the critical @xmath6 ( signed as @xmath104 ) beyond which the equilibrium temperature is higher than the virial temperature are roughly anti - correlated , i.e. , a lower @xmath13 corresponds to a larger @xmath105 . exact estimation of the relation between @xmath13 and @xmath105 is not straightforward . this is because we need to know the radiation temperature @xmath49 as a function of @xmath13 which requires numerical calculations . note that from figs . 1 & 3(b ) the minimum value of @xmath105 should be larger than @xmath106 . although no steady self - consistent solution exists due to the strong compton heating at and beyond @xmath105 , an `` oscillation '' of the activity of the black hole is expected ( e.g. , cowie , ostriker , & stark 1978 ; ciotti & ostriker 2007 ) . when the accretion rate is high , only the gas inside of @xmath105 can be accreted . this active phase will last for a timescale of the accretion timescale at @xmath105 . then all the gas will be used up and the active phase stops . in this case , compton heating also stops so the gas outside of @xmath105 will be cooled by radiation and be accreted again and the cycle repeats . the time the non - active phase will last is determined by the radiative timescale of the gas at @xmath105 , since it is longer than the accretion timescale there . an alternative consequence of the strong compton heating at large radii is that the accretion can be self - regulated by irradiating the outer flow ( shakura & sunyaev 1973 ) . that is , the strong compton heating will not completely stop the accretion , but only decrease the accretion rate . this then reduces the energy release in the inner part , which in turn reduces the irradiation . a multi - dimensional numerical simulation is required to solve this issue and accurate time - dependence is needed as well since steady solutions may not be stable . for the massive black holes seen in the nuclei of most galaxies the compton heated interruption of the high luminosity states should be typical if a hot accretion flow exists there . we now know that the accretion flow in low - luminosity agns is of this type ( see yuan 2007 and ho 2008 for reviews ) . for luminous agns such as quasars , although people incline to think it is a standard thin disk which is optically thick , many problems remain for this model ( e.g. , shlosman , begelman , & frank 1990 ; koratkar & blaes 1999 ) . if the actual accretion flow is radially optically thin to compton scattering , similar to the hot accretion flow , our analysis applies . this kind of oscillation does not apply to stellar mass black holes in our galaxy . this is because the prerequisite for such oscillation is that the accretion rate is large and the hot accretion flow extends to large radii . for a stellar mass black hole , the accretion material comes from the companion star and it starts out as a standard thin disk . in the hard state the standard disk does not extend to the innermost stable circular orbit but is replaced by a hot accretion flow within a transition radius @xmath107 . when the accretion rate is high , @xmath107 is small ( yuan & narayan 2004 ) . so compton scattering cools rather than heats the hot accretion flow . but on the other hand , the photons emitted by the hot accretion flow will heat the cool electrons in the standard disk at @xmath107 and thus will change the dynamics of the transition . this effect has never been noted and could be a topic of future work . for a geometrically thick and optically thin hot accretion flow , the photons can travel for a long distance without being absorbed , and thus be able to heat or cool electrons via compton scattering . we investigate this global compton scattering effect and find that for an accretion rate described by @xmath0 the compton cooling effect will be important when @xmath108 ; while the compton heating effect will be important when @xmath109 and @xmath1 . specifically , the scattering heats electrons at @xmath110 while cools electrons at @xmath111 . if @xmath6 is larger , the critical @xmath13 above which the compton heating effect is important will become lower . we have successfully obtained the self - consistent steady solution with this effect included for @xmath112 and @xmath113 . but when @xmath114 and @xmath115 we fail because of the strong radiative cooling ( local plus global compton scattering ) . it is also difficult to get the self - consistent solution when @xmath114 ( @xmath116 ) and @xmath117 . this is because in this case the compton heating is so strong at and beyond @xmath6 that the equilibrium electron temperature there will be higher than the virial temperature . more generally we expect that the radius where the equilibrium temperature due to the compton heating is equal to the virial temperature , @xmath105 , is anti - correlated with @xmath118 . we argue that the black hole will manifest an oscillation of the activity in the case that we fail to get the steady solution . the period will be the radiative timescale of the gas at @xmath105 . all our discussions so far are for a one - dimensional ( but not spherical ) accretion flow . although big uncertainties exist for the vertical structure of accretion flow , we are certain that when @xmath69 is high , the scattering will be important , and consequently much of the luminosity will `` leak out '' perpendicular to the accretion flow as in the standard thin disk . this will have two effects . one is that the highest luminosity up to which we can get the self - consistent solution with the global compton effect included will be higher . in addition , the compton heating will be stronger in the vertical direction than in the equatorial plane of the flow . as a result strong wind will be launched as pointed out by park & ostriker ( 2001 ; 2007 ) and found by proga et al . all of the described effects are likely to become significant for agn accretion flows having @xmath119 and optically thin ( ref . 3 for discussion of this possibility in luminous agns ) , which , we know from recent applications of the soltan argument ( yu & tremaine 2002 ) , are the phases during which most massive black hole growth occurs . we thank ramesh narayan , myeong - gu park , eliot quataert , and the referee , andrzej zdziarski , for their useful comments on the work . this work was supported in part by the natural science foundation of china ( grants 10773024 , 10833002 , and 10825314 ) , one - hundred - talent program of chinese academy of sciences , and the national basic research program of china ( grant 2009cb824800 ) .	the hot accretion flow ( such as advection - dominated accretion flow ) is usually optically thin in the radial direction , therefore the photons produced at one radius can travel for a long distance without being absorbed . these photons thus can heat or cool electrons at other radii via compton scattering . this effect has been ignored in most previous works on hot accretion flows and is the focus of this paper . if the mass accretion rate is described by @xmath0 and @xmath1 , we find that the compton scattering will play a cooling and heating role at @xmath2 and @xmath3 , respectively . specifically , when @xmath4 , the compton cooling rate is larger than the local viscous heating rate at certain radius ; therefore the cooling effect is important . when @xmath5 , the heating effect at @xmath6 is important . we can obtain the self - consistent steady solution with the global compton effect included only if @xmath7 for @xmath8 , which corresponds to @xmath9 . above this rate the compton cooling is so strong at the inner region that hot solutions can not exist . on the other hand , for @xmath10 , we can only get the self - consistent solution when @xmath7 and @xmath11 . the value of this critical accretion rate is anti - correlated with the value of @xmath6 . above this accretion rate the equilibrium temperature of electrons at @xmath6 is higher than the virial temperature as a result of strong compton heating , so the accretion is suppressed . in this case the activity of the black hole will likely `` oscillate '' between an active and an inactive phases , with the oscillation timescale being the radiative timescale of the gas at @xmath6 .
transport in media with quenched or time - dependent disorder has received much interest due to the considerable fundamental and technological importance of the problem @xcite . transport processes underlay a variety of chromatographic separation techniques @xcite , extraction of oil from porous rocks @xcite , excitations quenching in amorphous solids and doped crystals @xcite , conductivity of complex media and permeability of disordered membranes @xcite . molecular diffusion measurements in disordered media often serve as a tool in characterizing the internal geometry over a range of molecular and macroscopic length scales @xcite . in addition , transport processes control the kinetics of chemical reactions in disordered media @xcite . following motion of individual particles in disordered media one most often encounters behavior characterized by a mean - square displacement @xmath4 growing as @xmath5 i.e. linearly in time with @xmath6 being the `` diffusion '' coefficient , which depends on the peculiar features of the disordered medium . such type of dynamics is ubiquitous due to its inherent relationship to the central - limit theorem , which makes it a paradigm of stochastic processes . one the other hand , there was a growing evidence for other laws of random transport , both slower and enhanced relative to the time dependence in eq.([dif ] ) . generally , one finds @xmath7 case with @xmath8 , known as the dispersive transport or subdiffusive regime , has been in the focus of many theoretical and experimental studies ( see , e.g. , refs.@xcite ) and has been attributed , for instance , to random walks on fractal structures @xcite or to diffusion in presence of temporal traps @xcite . a few stray examples of such a transport are time of flight measurements in amorphous semiconductors @xcite , defect diffusion in relaxation studies @xcite , excitation migration on the disordered array of donor centers ( see , e.g. , ref.@xcite ) , tracer diffusion in a linear array of convective rolls @xcite or in low - dimensional lattice gases @xcite , or the so - called sinai diffusion with extremely confined particle trajectories such that @xmath9 @xcite . enhanced diffusion regime , @xmath10 , is observed in turbulent diffusion ( @xmath11 ) , as well as in some other instances ( see , e.g. , ref.@xcite ) for some review ) . suppose now that in the disordered medium in question , in which , in absence of any fields acting on the species whose motion we follow ( say , tracer particles ) , we find that their mean displacement is zero , while the mean - square displacement @xmath4 obeys eq.([dif ] ) or the anomalous diffusion law in eq.([andif ] ) . suppose next that we manage to bias somehow the particle motion , say , by charging the tracer particles only and putting the whole system in a constant electric field @xmath12 . this situation is , of course , quite interesting in its own right and is appropriate to charge transfer or `` dynamic directed percolation '' in time - dependent inhomogeneous medium @xcite , such as , e.g. , certain biomembranes @xcite , solid protonic conductors @xcite , oil - continuous microemulsions @xcite or polymer electrolytes @xcite . in such a situation , the tracer particles will move preferentially in the direction of the field and will attain a non - zero mean displacement and a non - zero mean velocity @xmath13 . one may define then the time - dependent tracer particle mobility as @xmath14 while the `` diffusivity '' @xmath15 can be extracted from the behavior in @xmath16 of the field ( eqs.([dif ] ) or ( [ andif ] ) ) via : @xmath17 where @xmath18 is the spatial dimension of the system . now , the question is whether the mobility @xmath19 , calculated from the tracer particle mean displacement in the presence of an external electric field , and the diffusivity @xmath15 , eq.([di ] ) , deduced from the tracer particle mean - square displacement in the absence of the field , obey the generalized einstein relation of the form @xmath20 where @xmath21 denotes the reciprocal temperature ? the answer is trivially positive , of course , for isolated non - interacting particles performing conventional brownian motion . for particle diffusion in disordered medium the validity of the relation in eq.([einstein ] ) is much less evident . on the one hand , it has been found that eq.([einstein ] ) holds for the tracer particle motion in a one - dimensional ( 1d ) gas of hard spheres , whose whole dynamics consists of pairs of neighboring particles interchanging velocities at each collision and the initial velocity distribution being an equilibrium one @xcite . it holds also for the tracer particle diffusion in a 1d hard - core gas with diffusional dynamics when the 1d lattice is a ring of a finite length @xcite , as well as for infinite 1d lattices with non - conserved @xcite and conserved particles number @xcite . remarkably , in the latter case eq.([einstein ] ) holds for @xmath22 sufficiently large despite the fact that both @xmath19 and @xmath15 do not attain constant values as @xmath23 but rather are decreasing functions of time @xcite . in addition , the validity of the einstein relation has been corroborated for the charge carriers in semiconductors @xcite and for polymeric systems in the subdiffusive regime @xcite . on the other hand , it is well known that the einstein relation is violated in some physical situations ; for instance , it is not fulfilled for sinai diffusion @xcite describing brownian motion of a test particle in presence of a quenched random force , and for the scher - lax - montroll model of anomalous random walk @xcite ( see also refs.@xcite and @xcite for some other examples ) . apparently , the einstein relation is also violated for random walk in disordered lattices at their percolation threshold - the celebrated model of the `` ant in the labyrinth '' of de gennes @xcite . here , in absence of external force acting on the random walker , its mean - square displacement @xmath24 obeys eq.([andif ] ) with @xmath8 @xcite . in presence of external field , which produces a bias in the random walk making the walker more likely to step along the field than against it , the behavior appears to be rather complex . as a matter of fact , the bias has a dual effect - it induces drift in the direction of the field , but also creates temporal traps , such as in dead - end branches , to escape from which the walker has to move against the field . one then finds that the drift velocity is not a monotonic function of the bias @xcite and moreover , it has been argued that the drift velocity vanishes once the bias exceeds a certain threshold value @xcite ( see also @xcite ) . these theoretical arguments have been supported by exact calculation of the mean drift velocity on a random comb and randomly diluted bethe lattice @xcite ( see , however , ref.@xcite for some objections ) . therefore , a small value of external bias gives rise to a mean displacement in the direction of the field that increases linearly with time , and the mean velocity tends to a constant . this asymptotic value of the mean velocity is proportional to the field for small fields @xcite , which implies that the mobility tends to a constant as @xmath23 . contrary to that , since in the absence of the field the walker motion is subdiffusive @xcite , one has that here the diffusivity @xmath15 should vanish as @xmath23 . therefore , if the results of refs.@xcite are correct , they would imply that the einstein relation is not valid for diffusion in random lattices at percolation threshold . hence , in principle , it is not @xmath25 clear at all whether eq.([einstein ] ) should be valid in any case . in this presentation we overview some recent results on the generalized einstein relation in eq.([einstein ] ) for tracer particle ( tp ) diffusion in dynamical disordered environments , as exemplified here by non - interacting lattice gases of hard - core particles . the paper is structured as follows : in section 2 we introduce the model , basic notations , write down dynamic equations describing the time evolution of the system and discuss the way of their solution . in section 3 we first recall the properties of an unbiased tp diffusion in a 1d hard - core lattice gas . next , we turn to the case of a biased tp in an inert lattice gas and present explicit results describing the tp mean displacement and the mobility . we demonstrate then that the einstein relation in eq.([einstein ] ) holds exactly despite the fact that in this case diffusion is anomalous . further on , in section 4 we consider the case when the initial particle distribution on a 1d lattice is inhomogeneous and characterized by an @xmath26-shape , `` shock''-like density profile . we find that when the tp is attracted towards the particle phase with higher density , at a certain critical value of the constant attraction force the tp will not move , on average . assuming the validity of the einstein relation , we find the second moment of the tp displacement . this heuristic analytical result is confirmed numerically which signifies that the einstein relation holds apparently for such an inhomogeneous situation . next , in sections 5 and 6 we analyse the forms of the biased tp terminal velocity in one- , two- and three - dimensional hard - core lattice gases undergoing continuous particles exchanges with a reservoir . we determine the tp mobility and assuming the validity of the einstein relation , obtain the tp diffusion coefficient @xmath6 in the unbiased case . results for 1d systems are confirmed by numerical monte carlo simulations . we also show that in two- and three - dimensional systems in the conserved particle density limit ( when particle exchanges with the reservoir are forbidden ) our predictions for @xmath6 coincide with classical results of nakazato and kitahara @xcite , which are known to be in a very good agreement with the numerical data @xcite . in section 7 we present an exact solution for the biased tp mean displacement in a two - dimensional ( 2d ) lattice gas containing a single unoccupied site . on comparing our results with earlier results by brummelhuis and hilhorst @xcite obtained for the unbiased tp dynamics , we infer that also in this case , in which the tp diffusion is anomalously , logarithmically confined , the einstein relation does hold exactly . finally , we conclude in section 8 with a brief summary of results and discussion . consider a @xmath18-dimensional regular hypercubic lattice of spacing @xmath27 each site of which is brought in contact with a reservoir containing identic , electrically neutral particles - a vapor phase ( fig.1 ) , maintained at a constant pressure . we suppose next that the reservoir particles may be created ( adsorbed ) on any vacant site at a fixed rate @xmath28 , which rate depends on the reservoir pressure and the energy gain due to the `` creation '' event . further on , the particles may move randomly along the lattice by hopping at a rate @xmath29 to any of @xmath30 neighboring lattice sites , which process is constrained by hard - core exclusion preventing multiple occupancy of any of the sites . lastly , the particles may spontaneously disappear ( desorb ) from the lattice at rate @xmath31 , which is dependent on the barrier against desorption . both @xmath32 and @xmath33 are site and environment independent . _ two - dimensional realization of the model : adsorbed monolayer in contact with a vapor phase . gray spheres denote the monolayer ( vapor ) particles ; the smaller black sphere stands for the driven tracer particle . _ ] to describe the occupancy of lattice sites , we introduce a time - dependent variable @xmath34 , which may assume two values : @xmath35 note that the local variable @xmath34 can change its value due to creation of particles ( adsorbtion ) , desorption and random hopping events . note also that random hopping events do conserve the total number of adsorbed particles and hence , the average particle density @xmath36 . on the other hands , creation / desorption processes do change @xmath34 locally such that the total number of particles on the lattice is not explicitly conserved when such two processes are allowed . however , the mean particle density , @xmath37 , approaches as @xmath38 a constant value @xmath39 this relation is well known and represents the customary langmuir adsorption isotherm @xcite . we finally remark that in the analysis of the stationary - state behavior , we can always turn to the conserved particles number limit by setting @xmath32 and @xmath33 equal to zero and keeping their ratio fixed , i.e. supposing that @xmath40 . this limit will correspond to the model of biased tracer diffusion in a hard - core lattice gas with fixed particles density @xmath41 , and will allow us to check our analytical predictions against some already known results @xcite . now , we introduce a bias into the system . contrary to the well - studied situation in which all particles are subject to a constant force ( see , e.g. , refs.@xcite and references therein ) , we will focus here on essentially less studied case when only one of particles is subject to external bias while the rest have symmetric hopping probabilities . this particle may be thought of as a probe designated to measure the frictional properties of symmetric non - inteacting lattice gas . to this end , at @xmath42 we introduce at the lattice origin an extra hard - core particle , whose motion we would like to follow and which will be called in what follows as the tracer particle - the tp ; position of this particle at time @xmath22 is denoted as @xmath43 . we stipulate that the tp is different from other particles in two aspects : a ) it can not disappear from the lattice and b ) it is subject to some external driving force , which favors its jumps into a preferential direction . physically , ( a ) be realized , for instance , if this only particle is charged and the system is subject to a uniform electric field * e. we suppose here , for simplicity of exposition , that the external force @xmath44 is oriented according to the unit vector @xmath45 . the dynamics of the biased tp is defined in the usual fashion @xcite : we suppose that the tp , which occupies the site @xmath43 at time @xmath22 , waits an exponentially distributed time with mean @xmath46 , and then attempts to hop onto one of @xmath30 neighboring sites , @xmath47 , where @xmath48 are @xmath30 unit vectors of the hypercubic lattice . in what follows we adopt the notation @xmath49 , where @xmath50 will denote the direction of the external force @xmath44 . next , the jump direction is chosen according to the probablity @xmath51 , which obeys : @xmath52}{\sum_{\mu}\exp\big[\frac{\beta}{2}({\bf e \cdot e}_{\boldsymbol \mu})\big ] } , \label{defp}\ ] ] where @xmath21 is the reciprocal temperature , @xmath53 stands for the scalar product , the charge of the tp is set equal to unity and the sum with the subscript @xmath54 denotes summation over all possible orientations of the vector @xmath55 ; that is , @xmath56 . after the jump direction is chosen , the tp attempts to hop onto the target site . the hop is instantaneously fulfilled if the target site is vacant at this moment of time ; otherwise , i.e. , if the target site is occupied by any adsorbed particle , the jump is rejected and the tp remains at its position . to close this subsection , several comments on the relevance of the model to some physical situations are in order . as we have already remarked , in the conserved particles number limit the model under study reduces to the model of biased tracer diffusion in hard - core lattice gases , which is related to a wide variety of physical situations , including , for example , diffusion of interstitials in crystals . a detailed review can be found in ref.@xcite . in two - dimensions , in the non - conserved particles number case it describes tracer diffusion in adsorbed monolayers in contact with the vapor phase @xcite and allows to determine , for instance , the analog of the stokes formula for such monolayers and consequently , to elucidate their intrinsic frictional properties . on the other hand , the 2d case can be thought of as some simplified picture of the stagnant layers emerging in liquids that are in contact with a solid body . it is well - known ( see , e.g. ref.@xcite ) that liquids in the close vicinity of a solid interface - at distances within a few molecular diameters - do possess completely different physical properties compared to those of the bulk phase . in this stagnant region , in which an intrinsically disordered liquid phase is spanned by and interacts with the ordering potential of the solid , the liquid s viscosity is drastically enhanced and transport processes ( related , say , to biased diffusion of charged carriers in solutions ) are substantially hindered . thus our model can be viewed as a two - level approximate model of this challenging physical system , in which the reservoir mimics the bulk fluid phase with very rapid transport , while the adsorbed monolayer ( particles on a 2d lattice ) represents the stagnant layer emerging on the solid - liquid interface . lastly , the model under study represents a novel example @xcite of the so - called `` dynamic percolation '' models ( see , e.g. , refs.@xcite and references therein ) , invoked to describe transport processes in many situations with dynamical disorder . in this context , the lattice gas particles can be thought of as representing some fluctuating environment , which hinders the motion of an impure molecule - the tp , which might be , for example , a charge carrier . a salient feature of this model , which makes it different from the previously proposed models of dynamic percolative environments , is that we include the hard - core interactions between the `` environment '' particles and the tp , such that the latter may itself influence the dynamics of the environment . as we proceed to show , this circumstance turns out to be very important and will result in formation of inhomogeneous distribution of the environment particles . now , we derive the evolution equations in a general , @xmath18-dimensional case . we begin by introducing some auxiliary definitions . let @xmath57 denote the entire set of the occupation variables , which defines the instantaneous configuration of the lattice - gas particles at time moment @xmath22 . next , let @xmath58 stand for the joint probability of finding at time @xmath22 the tp at the site @xmath59 and all adsorbed particles in the configuration @xmath60 . then , denoting as @xmath61 a configuration obtained from @xmath60 by the kawasaki - type exchange of the occupation variables of two neighboring sites @xmath62 and @xmath63 , and as @xmath64 - a configuration obtained from the original @xmath60 by the replacement @xmath65 , which corresponds to the glauber - type flip of the occupation variable due to the creation / desorption events , we have that the time evolution of the configuration probability @xmath58 obeys the following master equation : @xmath66 the mean velocity @xmath13 of the tp can be obtained by multiplying both sides of eq.([eqmaitresse ] ) by @xmath67 and summing over all possible configurations @xmath68 . this results in the following exact equation determining the tp velocity : @xmath69 where @xmath70 is the probability of having at time t a lattice gas particle at position @xmath71 , defined in the frame of reference moving with the tp . in other words , @xmath72 can be thought of as being the particle density profile as seen from the moving tp . equation ( [ vitesse ] ) signifies that the velocity of the tp is dependent on the lattice gas particles density in the immediate vicinity of the tracer . if the lattice gas is perfectly stirred , or , in other words , if @xmath73 everywhere , ( which implies immediate decoupling of @xmath59 and @xmath60 ) , one would obtain from eq.([vitesse ] ) a trivial mean - field result @xmath74 which states that the only effect of the medium on the tp dynamics is that its jump time @xmath46 is merely renormalized by a factor @xmath75 ; @xmath76 defines simply the mean frequency of successful jump events . however , the situation appears to be more complicated and , as we proceed to show , @xmath72 is different from the equilibrium value @xmath41 everywhere , except for @xmath77 . this means that the tp strongly perturbs the particles distribution on the lattice - it is no longer uniform and some non - trivial density profiles emerge . now , in order to calculate the instantaneous mean velocity of the tp we have to determine the mean particles density at the neighboring to the tp sites @xmath78 , which requires , in turn , computation of the density profile @xmath72 for arbitrary @xmath79 . the latter can be found from the master equation ( [ eqmaitresse ] ) by multiplying both sides by @xmath80 and performing the summation over all configurations @xmath68 . in doing so , we find that these equations are not closed with respect to @xmath72 , but are coupled to the third - order correlations , @xmath81 in turn , if we proceed further to the third - order correlations , we find that these are coupled respectively to the fourth - order correlations . consequently , in order to compute @xmath13 , one faces the problem of solving an infinite hierarchy of coupled equations . we resort then to the simplest non - trivial closure of the hierarchy in terms of @xmath72 , which has been first proposed in ref.@xcite , and represent @xmath82 as @xmath83 some arguments justifying such an approximation can be found in ref.@xcite . using the approximation in eq.([decouplage ] ) , we obtain @xmath84 which holds for all @xmath71 , except for @xmath85 . one the other hand , for these special sites @xmath86 with @xmath87 we find @xmath88 where the coefficients @xmath89 are defined by @xmath90 note that eq.([systemek2 ] ) represents , from the mathematical point of view , the boundary conditions for the general evolution equation ( [ systemek1 ] ) , imposed on the sites in the immediate vicinity of the tp . equations ( [ systemek1 ] ) and ( [ systemek2 ] ) together with eq.([vitesse ] ) thus constitute a closed system of equations which suffice computation of all properties of interest , i.e. the time - dependent tracer particle velocity and the particle density profiles as seen from the tracer particle . the general approach to solution of coupled non - linear eqs.([vitesse]),([systemek1 ] ) and ( [ systemek2 ] ) has been discussed in detail in ref.@xcite . here we merely note that despite the fact that using the decoupling scheme in eq.([decouplage ] ) we effectively close the system of equations on the level of the pair correlations , solution of eqs.([systemek1 ] ) and ( [ systemek2 ] ) still poses serious technical difficulties : namely , these equations are non - linear with respect to the tp velocity , which enters the gradient term on the rhs of the evolution equations for the pair correlation , and does depend itself on the values of the monolayer particles densities in the immediate vicinity of the tp . solution of this system of non - linear equations for several particular case will be presented in the next sections . consider first biased tracer diffusion in the simplest case of a 1d hard - core lattice gas with conserved particles number ; that is , a lattice gas in which all particles were initially introduced onto the lattice and particle exchanges with the vapor phase are forbidden ( @xmath91 . hence , the particle density @xmath41 on the lattice is kept constant all the time . for simplicity , we also set here the spacing @xmath92 and @xmath93 . these parameters can be trivially restored in final results . in absence of external field acting on the tp , when its transition probabilities are symmetric , tracer dynamics has been studied first in refs.@xcite and the tp mean - square displacement @xmath94 has been determined exactly : @xmath95 which thus follows an anomalous diffusion law in eq.([andif ] ) with @xmath96 . consequently , the `` diffusivity '' d(t ) is given here by @xmath97 now , if we charge the tp and switch on an external electric field , how will the tp velocity and the mobility behave ? an early incomplete answer was presented in ref.@xcite which focused on the extreme case of an infinitely strong electric field @xmath98 such that the tp performs a totally directed random walk constrained by hard - core particles . by noticing that such a model is tantamount to the 1d stefan freezing problem or the problem of directional solidification @xcite , ( the role of the latent heat being played by the lattice gas particles ) , it was found that in this extreme case the tp mean velocity follows @xmath99 where @xmath100 is a function of @xmath41 defined implicitly by @xmath101 @xmath102 being the complementary error function . hence , the tp mean velocity in the biased case decays at exactly the same rate as the tp `` diffusivity '' in the unbiased case ! in view of such a coincidence , one is , of course , prompted to consider the general case @xmath103 and to check the prefactors in eq.([speed ] ) having in mind to verify the validity of eq.([einstein ] ) for anomalous tracer diffusion in such random dynamical environments @xcite . the general case of arbitrary @xmath104 has been studied in ref.@xcite within the framework of the approach outlined in section 2 and subsequently , in terms of a rigorous probabilistic approach in ref.@xcite . results of these two approaches coincide , which justifies the decoupling approximation in eq.([decouplage ] ) . it has been found that in this general case the tp mean velocity @xmath105 obeys at sufficiently large times the form of eq.([speed ] ) , i.e. @xmath106 , in which law the prefactor @xmath107 is determined by the following equation @xcite : @xmath108 where @xmath109\ ] ] in the limit @xmath110 the solution of the transcedental eq.([k ] ) has the form @xmath111 which implies that the time - dependent mobility in this case is given by @xmath112 finally , by comparing the prefactors in the latter equation and eq.([harris ] ) , we find that the generalized einstein relation in eq.([einstein ] ) holds exactly for such an anomalous diffusion ! to close this section , we make several comments on the particle density distribution as seen from the tracer particle . since the tp has a non - zero mean displacement , while the lattice gas particles have symmetric hopping probabilities and hence , a zero mean displacement , they tend to accumulate in front of the tp . consequently , a condensed , traffic jam - like region is formed in front of the tp , and a depleted by the lattice - gas particles region past the tp emerges . as a matter of fact , such a inhomogeneous density distribution is not stationary : the size of the traffic jam - like region , in which the particle density profile @xmath113 is nearly constant and larger than @xmath41 , @xmath114 , grows in proportion to the tp mean displacement . the size of the depleted region with a constant density @xmath115 also grows in proportion to @xmath116 . in this section we address the question of the validity of the generalized einstein relation in a somewhat exotic situation in which the initial particle distributions from both sides of the tp has an @xmath26-shape , shock - like profile characterized by two different mean densities @xcite . that is , as depicted in fig.2 , we suppose that the tp is initially at the origin , and the particles mean densities from the left and from the right of the origin , which we denote as @xmath117 and @xmath118 , respectively , are not equal to each other . the tp thus defines position of the `` shock front '' . without of lack of generality , we suppose that @xmath119 , and will call in what follows the particle phase which initially occupied the left half - line as the high - density phase ( hdp ) , while the phase initially occupying the right half - line will be referred to as the low - density phase ( ldp ) . _ initial `` shock''configuration of the lattice gas particles on a 1d lattice . @xmath117 is the initial mean density of particles at the half - line @xmath120 and @xmath118 is the mean density of particles at the half - line @xmath121 . all lattice gas particles , excluding the tp ( the `` shock '' front particle ) , have symmetric transition probabilities . the tp has asymmetric transition probabilities and experiences the bias in the negative direction . _ ] further on , after deposition on the lattice , particles start to perform @xmath122 , ( i.e. with equal probabilities for going to the left or to the right ) , hopping motion between the nearest lattice sites under the constraint that neither two particles can simultaneously occupy the same lattice site and can not pass through each other . next , we stipulate that the tp is subject to a constant force @xmath123 which favors its jumps in a preferential direction . thus for the tp the probabilities of going to the right ( @xmath124 ) and to the left ( @xmath125 ) will be different from each other . we will suppose that @xmath123 is directed towards the hdp . we adopt the convention that in this case @xmath123 is negative , @xmath126 ; the tp hopping probabilities @xmath124 and @xmath125 are related to the force and the reciprocal temperature @xmath21 through @xmath127 and @xmath128 . from the physical point of view , such a @xmath129 force can be understood as an effective boundary tension derived from the solid - on - solid - model hamiltonian of the phase - separating boundary @xcite and mimic , in a mean - field fashion , the presence of attractive interactions between the lattice - gas particles which are not explicitly included in the model . relation between @xmath124 , @xmath125 and the amplitude of attractive interparticle interactions has been discussed in refs.@xcite and @xcite . the tp dynamics in such a system has been analysed in refs.@xcite and @xcite and it has been found that the tp mean velocity obeys @xmath130 at sufficiently large times , where the prefactor @xmath131 is determined implicitly as the solution of the following equation : @xmath132 which generalizes the result of the previous section for the case of @xmath133 . equation ( [ j ] ) predicts that three different regimes can take place depending on the relation between @xmath134 and @xmath135 : \(1 ) when @xmath136 the parameter @xmath131 is finite and positive , which means that the hdp expands compressing the ldp . in the particular case @xmath137 and @xmath138 the parameter @xmath131 appears to be a positive , logarithmically growing with time function , which behavior agrees with the results of ref.@xcite . \(2 ) when @xmath139 the parameter @xmath131 is less than zero - the expanding ldp and the applied force effectively compress the hdp . \(3 ) when @xmath140 , the parameter @xmath141 . this relation between the system parameters when the hdp and the ldp are in equilibrium with each other and the tp mean displacement is zero , was found also in ref.@xcite from the analysis of the stationary behavior in a finite 1d lattice gas . `` shock front '' , or , in other words , the tp propagation in cases ( 1 ) and ( 2 ) has been discussed in detail in ref.@xcite . here , in view of our interest in the einstein relation , we will focus on the case ( 3 ) in which situation the high- and low - density phases coexist such that the tp mean displacement is equal to zero . on the other hand , it is clear that also in this case the tp will wander randomly around the equilibrium position and one may expect that its mean - square displacement will not be equal to zero . to determine the tp mean - square displacement @xmath142 we may try to extend the approach outlined in section 2 which will require some more cumbersome analysis . we will however pursue a different , heuristic approach and instead of evaluating @xmath142 in terms of the approach of section 2 we will resort to the einstein relation in eq.([einstein ] ) assuming that it also holds in the inhomogeneous situation under study . our result will be then checked by numeric simulations . to do this , we note first that contrary to the model of section 3 , here the tp does not move , on average , when the external force @xmath123 acting on this particle is not equal to zero , but rather to some critical value @xmath143 , such that @xmath144 hence , it seems reasonable to define the tp mobility in this inhomogeneous case as @xmath145 further on , we find from eq.([j ] ) that in the limit @xmath146 the prefactor @xmath131 follows @xmath147 which implies that @xmath19 obeys @xmath148 now , assuming that eq.([einstein ] ) holds , we get @xmath149 which reduces to the classical result in eq.([harr ] ) in the limit @xmath150 . _ mean - square displacement of the tp in the critical case @xmath151 . the solid lines show our analytical prediction from eq.([harrr ] ) and the noisy lines present the results of monte carlo simulations . the curves from top to bottom correspond to the following values of the parameters @xmath152 : the first two curves are the analytical result in eq.([harr ] ) and numerical data for the symmetric cases @xmath153 and @xmath154 . the lower curves correspond to @xmath155 , @xmath156 , @xmath157 and @xmath158 respectively . _ ] in fig.3 we present results of extensive monte carlo simulations which confirm the result in eq.([harrr ] ) and hence , signify that the einstein relation seemingly works in this inhomogeneous case . we turn now to discussion of the tp biased diffusion in situation in which the reservoir is present and particle exchanges with the reservoir are allowed ; that is , when both the parameters @xmath32 and @xmath33 are not equal to zero . in 1d systems with continuous particles exchanges with a reservoir , which situation is appropriate to adsorption on polymer chains @xcite , the biased tp will ultimately move with a constant velocity @xmath159 and particles distribution around the tp will be characterized by stationary density profiles @xmath160 @xcite . here , the general solution of eqs.([systemek1 ] ) and ( [ systemek2 ] ) has the following form : @xmath161 where the characteristic lengths @xmath162 obey @xmath163,\ ] ] while the amplitudes @xmath164 are given respectively by @xmath165 note that @xmath166 , and consequently , the local density past the tp approaches its non - perturbed value @xmath41 slower than in front of it ; this signifies that correlations between the tp position and particle distribution are stronger past the tp . next , @xmath167 is always positive , while @xmath168 ; this means that the density profile is a non - monotoneous function of @xmath169 and is characterized by a jammed region in front of the tp , in which the local density is higher than @xmath41 , and a depleted region past the tp in which the density is lower than @xmath41 . this is , of course , quite similar to the behavior observed in situations in which exchanges with the reservoir are not allowed ( sections 3 and 4 ) ; essential difference is , however , that here the inhomogeneous particles distribution around the tp attains a stationary form . _ terminal velocity of the tp as a function of the `` creation '' probability @xmath32 at different values of the parameter @xmath33 . the tp hopping probabilities are @xmath170 and @xmath171 . the solid lines give the analytical solution while the filled squares denote the results of monte - carlo simulations . upper curves correspond to @xmath172 , the intermediate - to @xmath173 and the lower - to @xmath174 , respectively . _ ] now , we are in position to obtain a system of two closed - form non - linear equations determining implicitly the unknown parameters @xmath175 and @xmath176 , which will allow us to compute the tp terminal velocity , related to @xmath177 through @xmath178 . substituting eq.([dprofiles ] ) into eq.([defa ] ) , we find @xmath179\ ] ] and @xmath180\ ] ] which system of equations defines the stationary velocity @xmath159 of the tp ( see fig.[vitfig ] ) as well as the density profiles ( see fig.[proffig ] ) . for arbitrary values of @xmath124 , @xmath32 and @xmath33 the parameters @xmath177 , defined by eqs.([o ] ) and ( [ b ] ) , and consequently , @xmath159 can be determined only numerically ( see figs.4 to 6 ) . however , @xmath159 can be found analytically in the explicit form in the limit of a vanishingly small force @xmath12 , @xmath110 . expanding @xmath177 in the taylor series in powers of @xmath12 and retaining only linear with @xmath12 terms , we find that the tp velocity follows @xmath181 which relation can be thought off as the analog of the stokes formula for driven motion in a 1d lattice gas undergoing continuous particles exchanges with the vapor phase . eq.([st ] ) signifies that the frictional force exerted on the tp by the lattice gas particles is @xmath182 . the friction coefficient , i.e. the proportionality factor in eq.([st ] ) , is given explicitly by @xmath183\ ] ] note that @xmath184 in eq.([dfriction ] ) can be written down as the sum , @xmath185 . the first term , @xmath186 is a typical mean - field result and corresponds to a perfertly homogeneous lattice gas ( see discussion following eq.([defk ] ) ) . the second term , @xmath187 has , however , a more complicated origin . namely , it reflects a cooperative behavior emerging in the lattice gas , associated with the formation of inhomogeneous density profiles ( see fig.[proffig ] ) - the formation of a `` traffic jam '' in front of the tp and a `` depleted '' region past the tp ( for more details , see ref.@xcite ) . the characteristic lengths of these two regions as well as the amplitudes @xmath164 depend on the magnitude of the tp velocity ; on the other hand , the tp velocity is itself dependent on the density profiles , in virtue of eq.([vitesse ] ) . this results in an intricate interplay between the jamming effect of the tp and smoothening of the created inhomogeneities by diffusive processes . note also that cooperative behavior becomes most prominent in the conserved particle number limit @xcite . setting @xmath188 , while keeping their ratio fixed ( which insures that @xmath41 stays constant ) , one notices that @xmath189 gets infinitely large . as a matter of fact , as we have shown already in section 3 , in such a situation no stationary density profiles around the tp exist . _ density profile around stationary moving tp for @xmath190 , @xmath174 and @xmath191 . the solid line is the plot of the analytical solution . filled squares are the results of monte - carlo simulations . _ ] in order to check our analytical predictions , we have performed numerical monte carlo simulations . results of these simulations , performed at different values of the parameters @xmath32 , @xmath33 , and @xmath192 , are also represented in figs.[vitfig ] and [ proffig ] . consider finally the situation with @xmath193 , in which case the terminal velocity vanishes and one expects conventional diffusive motion with the mean square displacement of the form of eq.([dif ] ) . heuristically , we can evaluate @xmath6 for the system under study if we assume the validity of the einstein relation in eq.([einstein ] ) . noticing that here the biased tp mobility @xmath54 is just @xmath194 , we find @xcite : @xmath195 monte carlo simulations ( see fig.6 ) evidently confirm our prediction for @xmath6 given by eq.([ddiffusion ] ) , and hence , confirm the validity of the einstein relation for the system under study . this is , of course , not an unexpected , but still a non - trivial result @xcite . _ self - diffusion coefficient of the tp as a function of the adsorption probability @xmath32 . notations and values of @xmath33 are the same as in figs.[vitfig ] . we turn now to the analysis of a biased tp diffusion in higher dimensional ( 2d and 3d ) lattice gases exposed to a vapor phase . here , the situation gets somewhat more difficult from the computational point of view ; we have now to solve the partial difference equations problem ( [ systemek1 ] ) , ( [ systemek2 ] ) rather than the mere difference equations arising in the 1d case . below we briefly outline such a solution for a 2d case . results for three - dimensions are obtained along essentially the same lines and here we display them without derivation . solution in two - dimensions can be found in a most convenient fashion if we introduce the generating function for the particle density profiles , defined as @xmath196 multiplying both sides of eqs . ( [ systemek1 ] ) and ( [ systemek2 ] ) by @xmath197 , and performing summations over @xmath198 and @xmath199 , we find that @xmath200 is given explicitly by @xmath201 where @xmath202 and @xmath203 equations ( [ adevelopper ] ) and ( [ defk ] ) determine the generation function for the density profiles exactly . before we proceed to the inversion of @xmath200 with respect to the variables @xmath204 and @xmath205 , we note that we can access interesting integral characteristics of the density profiles directly using the result in eqs.([adevelopper ] ) and ( [ defk ] ) . namely , as we have already remarked , the presence of the driven tp induces an inhomogeneous density distribution in the lattice . one can thus pose a natural question whether equilibrium between adsorption and desorption processes gets shifted due to such a perturbancy , i.e. whether the equilibrium density in the lattice gas is different from that given by eq.([rhostat ] ) . the answer is trivially `` no '' in the case when the particles number is explicitly conserved , but in the general case with arbitrary @xmath32 and @xmath33 this is not at all evident : similarly to the behavior in 1d system one expects that also in two - dimensions the density profiles are asymmetric as seen from the stationary moving tp and are characterized by a condensed , `` traffic - jam''-like region in front of and a depleted region past the tp . one anticipates then that the desorption events are favored in front of the tp , while the adsorption events are evidently suppressed by the excess density . on the other hand , past the tp desorption is diminished due to the particles depletion while adsorption may proceed more readily due to the same reason . it is thus not at all clear @xmath206 whether these two effects can compensate each other exactly , in view of a possible asymmetry of the density profiles , as it happens in the 1d model ( see fig.[proffig ] ) . for this purpose , we study the behavior of the integral deviation @xmath207 of the density from the equilibrium value @xmath208 , i.e. @xmath209 , which can be computed straightforwardly from eqs.([adevelopper ] ) and ( [ defk ] ) by setting both @xmath204 and @xmath205 equal to unity . noticing that @xmath210 , and that @xmath211 , we obtain then that @xmath207 is strictly equal to 0 . this implies , in turn , that the perturbancy of the density distribution in the lattice gas created by the driven tp does not shift the global balance between the adsorption and desorption events . inversion of the generating function with respect to @xmath204 and @xmath205 requires quite an involved mathematical analysis , which has been presented in detail in ref.@xcite . general solution for the density profiles reads : @xmath212 with @xmath213 where @xmath214 stands for the modified bessel function . we mention that @xmath215 has an interesting physical interpretation in terms of the generating function of a random walk of a single particle ( that is , in absence of the particles environment ) @xcite . now , the eqs.([2dsolh ] ) and ( [ repint ] ) display @xmath216 as a function of the coefficients @xmath217 that remain to be determined . as a matter of fact , these coefficients depend themselves on the local densities in the immediate vicinity of the tracer , i.e. on @xmath218 . this implies that we have to determine them from eqs.([2dsolh ] ) and ( [ repint ] ) in a self - consistent way @xcite . some analysis ( see ref.@xcite ) shows that @xmath217 are determined implicitly as the solution of the following system of three non - linear matrix equations @xmath219 where @xmath220 the matrix @xmath221 stands for the matrix obtained from @xmath222 by replacing the @xmath223-th column by the column - vector @xmath224 , @xmath225 while @xmath226 are expressed in terms of @xmath217 as @xmath227 lastly , we find that in 2d the tp terminal velocity obeys @xmath228 which represents the general force - velocity relation for the system under study , valid for arbitrary magnitude of the external bias and arbitrary values of other system s parameters , except apparently the situation when @xmath229 , @xmath230 and their ratio @xmath231 . this case corresponds to the percolation - type problem describing biased diffusion in a randomly diluted lattice - the model which has been discussed in the introduction . some analysis shows that @xmath159 defined by eq.([forcevitesse ] ) does not vanish for any value of external bias @xmath12 . this signifies that the decoupling approximation underlying the derivation of the result in eq.([forcevitesse ] ) is , of course , inappropriate for situations with @xmath232 disorder , but only for situations in which some mixing , either due to exchanges with the reservoir , or due to diffusional processes , is present . we turn now to the limit @xmath110 , in which case the problem simplifies considerably and allows to obtain explicit results for the local densities in the immediate vicinity of the tp and consequently , for the tp terminal velocity and its diffusivity . in this limit , we arrive again at a stokes - type formula of the form @xmath233 , where now @xmath234 with @xmath235 note that we again are able to single out two physically meaningful contributions to the friction coefficient @xmath184 . namely , the first term on the rhs of eq.([rrr ] ) is just the mean - field - type result corresponding to a perfectly stirred lattice gas , in which correlations between the tp and the lattice gas particles are discarded . the second term mirrors the cooperative behavior emerging in the lattice gas and is associated with the backflow effects . in contrast to the 1d case , however , the contribution to the overall friction coefficient stemming out of the cooperative effects remains finite in the conserved particles number limit . lastly , assuming _ a priori _ that the einstein relation holds for the system under study , we find the tp diffusion coefficient : @xmath236 it seems now interesting to compare our general result in eq.([2dautodiffgen ] ) against the classical result of nakazato and kitahara @xcite , which describes tp diffusion coefficient in a 2d lattice gas with conserved particles number . setting @xmath32 and @xmath33 equal to zero , while assuming that their ratio has a fixed value , @xmath237 , we find that the right - hand - side of eq.([2dautodiffgen ] ) attains the form @xmath238 which expression coincides with the earlier result obtained in refs.@xcite and @xcite within the framework of a different , compared to ours , analytical technique . the result in eq.([comparnaka ] ) is known to be exact in the limits @xmath239 and @xmath240 , and serves as a very good approximation for the self - diffusion coefficient in hard - core lattice gases of arbitrary density @xcite . in a similar fashion , we find that in 3d the tp terminal velocity obeys @xmath233 , where @xmath241 with @xmath242 and @xmath243 @xmath244 being the generating function , @xmath245 of the probability @xmath246 that a walker starting at the origin and performing a polya random walk on the sites of a 3d cubic lattice will arrive on the @xmath247-th step to the site with the lattice vector @xmath248 @xcite . assuming next that the einstein relation holds , we find that in 3d the tp diffusion coefficient reads : @xmath249 in the particular case of conserved particles number , when @xmath188 but their ratio @xmath250 is kept fixed , @xmath251 , the latter equation reduces to @xmath252 which coincides again with the classic result obtained earlier in refs.@xcite and @xcite . asymptotical behavior of the density profiles at large distances from the tp follows from the analysis of the analyticity properties of the complex function @xmath253 . it has been shown in ref.@xcite that in front of the tp , @xmath254 always decays exponentially with the distance : @xmath255 where the characteristic length @xmath256 obeys , in two - dimensions : @xmath257 note that @xmath256 stays finite for any values of the system parameters . on contrary , the behavior of the density profiles at large distances past the tracer qualitatively depends on the physical situation studied . in the general case when exchanges with the particles reservoir are allowed , the decay of the density profiles is still exponential : @xmath258 where @xmath259 note that in the general case the characteristic lenghts again , similarly to the 1d case , satisfy the inequality @xmath260 , which means that the correlations between the tp and the lattice particles are always stronger past the tp than in front of it . such correlations become extremely strong in the special case when the particles exchanges with the vapor phase are forbidden , i.e. , in the conserved particles number limit . in this case , we have that @xmath261 becomes infinitely large and , in the limit @xmath262 , the particle density follows @xmath263 remarkably enough , in this case the correlations between the tp position and the particles distribution vanish _ algebraically _ slow with the distance ! this implies , in turn , that in the conserved particles number case , the diffusive mixing of the lattice gas is not efficient enough to prevent the appearence of the quasi - long - range order and the medium `` remembers '' the passage of the tp on a long time and space scale , which signifies very strong memory effects . in three dimensions , an analogous result for the asymptotical behavior of the density profiles as seen from the stationary moving tp is : @xmath264 where @xmath265 is a constant @xcite . we turn finally to the extreme case of a very dense lattice gas in which all sites except one ( called in what follows `` a vacancy '' ) are filled with identical neutral , hard - core particles ( see , fig.7 ) . the particles move randomly by exchanging their positions with this single vacancy , subject to the hard - core exclusion constraint . the just described model , which represents , in fact , one of the simplest cases of the so - called `` slaved diffusion processes '' , has been studied over the years in various guises , such as , for instance , the `` constrained dynamics '' model of palmer @xcite . _ two - dimensional , infinite in both directions , square lattice in which all sites except one are filled with identical hard - core particles ( gray spheres ) . the black sphere denotes a single tracer particle , which is subject to external field @xmath266 , oriented in the positive @xmath267 direction , and thus has asymmetric hopping probabilities . the arrows of different size depict schematically the hopping probabilities ; a larger arrow near the tp indicates that it has a preference for moving in the direction of the applied field . _ ] brummelhuis and hilhorst @xcite were first to present an exact solution of this model in the lattice formulation . it has been shown that in the presence of a single vacancy the tp trajectories are remarkably confined ; the mean - square displacement shows an unbounded growth , but it grows only @xmath268 with time , @xmath269 which implies that the tp diffusivity @xmath270 obeys @xmath271 brummelhuis and hilhorst have also found @xcite that at sufficiently large times the probability @xmath272 of finding the tp at time moment @xmath273 at position @xmath274 converges to a limiting form as a function of the scaling variable @xmath275 . still striking , this limiting distribution is not a @xmath276 but a modified bessel function @xmath277 , which signifies that the successive steps of the tp , although separated by long time intervals , are effectively correlated . these results have been subsequently reproduced by means of different analytical techniques in refs.@xcite and @xcite . as in the previous sections , we focus here on tp dynamics in the biased case , when the tp is charged , while the lattice gas particles are neutral , and the system is subject to a constant electric field . we aim to determine exactly the tp mean - displacement @xmath278 , which will give us an access to the tp mobility and thus will allow to verify the validity of the einstein relation in this model with remarkably confined dynamics . a standard approach to define the properties of the tp random walk , as exposed in section 2 , would be to start with a master equation determining the evolution of the whole configuration of particles . in doing so , similarly to the analysis of the tracer diffusion on 2d lattices in the presence of a finite vacancy concentration ( see , e.g. , ref.@xcite ) , one obtains the evolution of the joint distribution @xmath279 of the tp position @xmath274 and of the vacancy position @xmath280 at time moment @xmath273 . the property of interest , i.e. the reduced distribution function of the tp alone will then be found from @xmath279 by performing lattice summation over all possible values of the variable @xmath280 . here we pursue , however , a different approach , which has been first put forward in the original work of brummelhuis and hilhorst @xcite ; that is , we construct the distribution function of the tp position at time @xmath273 directly in terms of the return probabilities of the random walk performed by the vacancy . the only complication , compared to the unbiased case considered by brummelhuis and hilhorst @xcite , is that in our case ten different return probabilities will be involved , instead of three different ones appearing in the unbiased case . hence , the analysis will be slightly more involved . we begin by introducing some basic notations . let @xmath272 be the probability that the tp , which starts its random walk at the origin , appears at the site @xmath274 at time moment @xmath273 , given that the vacancy is initially at site @xmath281 . * @xmath282 be the probability that the vacancy , which starts its random walk at the site @xmath281 , arrives at the origin @xmath283 for the first time at the time step @xmath273 . * @xmath284 be the conditional probability that the vacancy , which starts its random walk at the site @xmath281 , appears at the origin for the first time at the time step @xmath273 , being at time moment @xmath285 at the site @xmath48 . further on , for any time - dependent quantity @xmath286 we define the generating function : @xmath287 and for any space - dependent quantity @xmath288 the discrete fourier transform : @xmath289 where the sum with the subscript @xmath290 runs over all lattice sites . now , following brummelhuis and hilhorst @xcite , we write down directly the equation obeyed by the reduced probability distribution @xmath272 ( cf ref.@xcite for a study of the joint probability of the tp position and of the vacancy position in the unbiased case ) : @xmath291 next , using the definition of the generating functions and of the discrete fourier transforms , eqs.([l ] ) and ( [ y ] ) , we obtain the following matricial representation of the generating function of the tp probability distribution : @xmath292 in eq.([p ] ) the function @xmath293 stands for the determinant of the following @xmath294 matrix , @xmath295 where @xmath296 is a @xmath294 matrix of the form : @xmath297 and the coefficients @xmath298 , @xmath299 , defined as @xmath300 are the generating functions of the conditional probabilities for the first time visit of the origin by the vacancy . lastly , the matrix @xmath301 in eq.([p ] ) is given by @xmath302 the property of interest - the tp probability distribution function , will be then obtained by inverting @xmath303 with respect to the wave - vector @xmath304 and to the variable @xmath305 : @xmath306 where the contour of integration @xmath307 encircles the origin counterclockwise . after straightforward but rather tedious calculations ( see , ref.@xcite ) , we find that in the small-@xmath308 limit and @xmath309 , which defines the leading large-@xmath273 behavior of the probability distribution , the generating function @xmath303 obeys @xmath310 where the coefficients @xmath311 are all functions of the field strength @xmath12 and of the temperature only . now , the leading large-@xmath273 asymptotical behavior of the tp mean displacement can be obtained directly from eq.([v ] ) , since the generating function of the tp mean displacement : @xmath312 obeys ( see , e.g. , ref.@xcite ) : @xmath313 consequently , differentiating the expression on the right - hand - side side of eq.([v ] ) with respect to the components of the wave - vector @xmath308 , we find that the asymptotical behavior of the generating function of the tp mean displacement in the vicinity of @xmath314 follows @xmath315 further on , using the discrete tauberian theorem ( cf . ref.@xcite ) and eq.([alpha ] ) , we find the following force - velocity relation for the system under study @xmath316 which shows that the tp mean displacement grows @xmath268 with @xmath273 . in the limit @xmath110 , the coefficient @xmath317 obeys @xmath318 and hence , the mobility @xmath319 , defined in eq.([mob ] ) , follows @xmath320 comparing next the result in eq.([mob2 ] ) with that for the diffusivity @xmath270 , eq.([dii ] ) , derived by brummelhuis and hilhorst @xcite for the unbiased case , we infer that the tp mobility and diffusivity do obey , at least in the leading in @xmath273 order , the generalized einstein relation in eq.([einstein ] ) ! note , that this can not be , of course , an @xmath25 expected result , in view of an intricate nature of the random walks involved and anomalous , @xmath321 confinement of the random walk trajectories . we turn next to calculation of the asymptotic forms of the probability distribution @xmath272 . inverting @xmath322 with respect to @xmath308 , we find that in the large-@xmath273 and large-@xmath323 limits , in the general case when @xmath324 , @xmath272 obeys @xmath325 where @xmath326 is the modified bessel ( mcdonald ) function of zeroth order , while @xmath327 @xmath267 and @xmath328 being the components of the vector @xmath274 . the distribution in eq.([u ] ) reduces to the result obtained by brummelhuis and hilhorst @xcite in the limit @xmath193 . to conclude , in this presentation we have overviewed some recent results on biased tracer diffusion in disordered dynamical environments represented by hard - core lattice gases . we have considered several possible situations including 1d lattice gases with conserved and non - conserved particles density with initial homogeneous or `` shock''-like particle distributions , @xmath18-dimensional lattice gases undergoing continuous exchanges with a reservoir , as well as an extreme case of biased tracer diffusion in a 2d highly - packed lattice gas containing only a single vacancy . the main emphasis of our analysis here has been put on the question whether the einstein relation between the tracer particle mobility and the diffusivity is valid , despite the fact that in some cases the tracer diffusion is anomalously confined or , in case of regular diffusion , the mobility and the diffusivity are non - trivial functions of the lattice gas density and other pertinent parameters . for some situations we were able to furnish exact solutions , which show explicitly that this fundamental relation holds . in other instances , to calculate the tracer particle mobility we invoked an approximate approach based on decoupling of the third - order correlation functions . the tracer particle diffusivity has been then obtained by assuming that the einstein relations holds . extensive numerical simulations have been performed , which confirm our analytical predictions and thus the validity of the einstein relation . apart of it , we have shown that the lattice gas particles distribution as seen from moving tp is strongly inhomogeneous : the local particle density in front of the tp is higher than the average , which means that the lattice gas particles tend to accumulate in front of the driven tp , creating a sort of a `` traffic jam '' , which impedes its motion . the condensed , `` traffic jam''-like region in the case when the particles density is not explicitly conserved is characterized by a @xmath329 density profile which vanishes as an exponential function of the distance from the tracer . the characteristic length and the amplitude of the density relaxation function have been calculated explicitly . in one dimensional situations with conserved particles density the `` traffic jam '' region grows in size in proportion to the tp mean displacement and no @xmath329 profile exists . on the other hand , past the tp the local density is lower than the average . in one dimensional situations with conserved particle density this depleted by particles region also grows in size in proportion to the tp mean displacement . in higher dimensions , we have observed that depending on whether the number of particles is explicitly conserved or not , the local density past the tp may tend to the average value at large separations from the tp in a completely different fashion : in the non - conserved case the decay of the density is described by an exponential function , while for the conserved particles number case it shows an @xmath330 dependence on the distance , revealing in the latter case especially strong memory effects and strong correlations between the particle distribution in the lattice gas and the tracer position . further on , we have found that the terminal velocity @xmath159 of the tp depends explicitly on both the excess density in the `` jammed '' region in front of the tp , as well as on the density in the depleted region past the tracer . we realized that both densities are themselves dependent on the magnitude of the tracer velocity , applied external force , as well as on the rate of the adsorption / desorption processes and on the rate at which the particles can diffuse away of the tracer , which results in effective non - linear coupling between @xmath159 and @xmath12 . in consequence , in the general case ( for arbitrary adsorption / desorption rates and arbitrary external force ) , @xmath159 can be found only implicitly , as the solution of a transcendental equation relating @xmath159 to the system parameters . this equation simplifies considerably in the limit of a vanishingly small external bias ; in this case we recover a linear force - velocity relation , akin to the so - called stokes formula . this linear relation signifies that the frictional force exerted on the tracer particle by the host medium ( the lattice gas ) is viscous . the tp mobility , which is inverse of the stokes friction coefficient , thus results from an intricate cooperative behavior . s.havlin and d.ben-avraham , advan . * 36 * , 695 ( 1987 ) j.p.bouchaud , a.comtet , a.georges and p.le doussal , ann . ( ny ) * 201 * , 285 ( 1990 ) j.p.bouchaud and a.georges , phys * 195 * , 129 ( 1991 ) j.klafter , r.rubin and m.shlesinger , eds . , _ transport and relaxation in random media _ , ( world scientific , singapore , 1986 ) j.drake and j.klafter , _ dynamics in confined molecular systems _ , physics today * 43 * , 46 ( 1990 ) j.drake , j.klafter and p.levitz , _ chemical and biological microstructures as probed by dynamic processes _ , science * 251 * , 1574 ( 1991 ) ; j. m. drake , p. levitz , j. klafter , n.j . turro , k.s . nitsche and k.f . cassidy , phys . letters * 61 * , 865 ( 1988 ) j. m. drake , p. levitz and j. klafter , _ time resolved studies of optical probes in confining spaces _ , in : `` dynamical processes in condensed molecular systems '' eds . j. klafter , j. jortner and a. blumen ( world scientific , singapore , 1989 ) s.f.burlatsky , g.oshanin and a.i.chernoutsan , in : _ low dimensional conductors and superconductors _ , springer research reports in physics , eds . a.ovchinnikov and i.ukrainsky ( springer , berlin , 1991 ) , p.121 g.oshanin , m.moreau and s.f.burlatsky , adv . colloid and interface sci . * 49 * , 1 ( 1994 ) g.oshanin and s.f.burlatsky , j. stat . phys . * 65 * , 1109 , 1095 ( 1991 ) a.blumen , nuovo cimento b * 63 * , 50 ( 1981 ) a.blumen , j.klafter and g.zumofen , in : _ optical spectroscopy of glasses _ , ed . i.zschokke ( reidel , dordrecht , 1986 ) , p.199 s.f.burlatsky , g.oshanin and a.chernoutsan , phys . lett . a * 149 * , 47 ( 1990 ) m.f.shlesinger , h.scher and j.g.bendler , phys . today * 1 * , 26 ( 1991 ) h.scher and e.w.montroll , phys . rev . b * 12 * , 12 , 2455 ( 1975 ) g.oshanin , s.f.burlatsky , m.moreau and b.gaveau , chem . phys . * 177 * , 803 ( 1993 ) y.pomeau , a.pumir and w.young , phys . fluids a * 1 * , 462 ( 1989 ) t.e.harris , j. appl * 2 * , 323 ( 1965 ) d.g.levitt , phys . a * 8 * , 3050 ( 1973 ) j.k.percus , phys . rev . a * 9 * , 557 ( 1974 ) s.alexander and p.pincus , phys . b * 18 * , 2011 ( 1978 ) t.m.ligget , _ interacting particle systems _ , ( springer , new york , 1985 ) m.j.a.m . brummelhuis and h.j.hilhorst , j. stat . phys . * 53 * , 249 ( 1988 ) ; see also : marco brummelhuis , _ application of random walk : tracer diffusion , lattice covering , and damage spreading _ , ph . d. thesis , leiden university , the netherlands , 1991 ya.sinai , theory probab . * 27 * , 247 ( 1982 ) h.kesten , physica a * 138 * , 299 ( 1986 ) s.havlin , j. stat . phys . * 58 * , 653 ( 1990 ) j.klafter , g.zumofen and a.blumen , chem . phys . * 177 * , 821 ( 1993 ) ; m. f. shlesinger , j. klafter and g. zumofen , _ above , below and beyond brownian motion _ , american j. of phys . * 67 * , 1253 ( 1999 ) s.d.druger , a.nitzan , and m.a.ratner , j. chem . phys . * 79 * , 3133 ( 1983 ) ; s.d.druger , m.a.ratner , and a.nitzan , phys . b * 31 * , 3939 ( 1985 ) a.k.harrison and r.zwanzig , phys . rev . a * 32 * , 1072 ( 1985 ) r.granek and a.nitzan , j. chem . phys . * 90 * , 3784 ( 1989 ) o.bnichou , j.klafter , m.moreau and g.oshanin , phys . e * 62 * , 3327 ( 2000 ) e.e.polymeropoulos and j.brickmann , ann . * 14 * , 315 ( 1985 ) s.chandra , _ superionic solids , principles and applications _ , ( north - holland , amsterdam , 1981 ) s.bhattarcharya , j.p.stokes , m.w.kim , and j.s.huang , phys lett . * 55 * , 1884 ( 1985 ) g.s.grest , i.webman , s.a.safran , and a.l.r.bug , phys . rev . a * 33 * , 2842 ( 1986 ) a.l.r.bug and y.gefen , phys . a * 35 * , 1301 ( 1987 ) c.m.c.gambi , m.g.giri , m.carl , d.senatra , and a.chittofrati , phys . e * 56 * , 4356 ( 1997 ) h.cheradame , j.l.souquet and j.m.latour , mater . * 15 * , 1173 ( 1980 ) b.l.papke , r.dupon , m.a.ratner and d.f.shriver , solid state ion . * 5 * , 685 ( 1981 ) c.a.vincent , prog . solid state chem . * 17 * , 145 ( 1987 ) j.l.lebowitz and j.k.percus , phys . rev . * 155 * , 122 ( 1967 ) p.ferrari , s.goldstein and j.l.lebowitz , diffusion , mobility and the einstein relation , in : _ statistical physics and dynamical systems : j.fritz , a.jaffe and d.szasz ( birkhuser , boston , 1985 ) p.405 o.bnichou , a.m.cazabat , a.lemarchand , m.moreau and g.oshanin , j. stat . phys . * 97 * , 351 ( 1999 ) s.f.burlatsky , g.oshanin , m.moreau and w.p.reinhardt , phys . e * 54 * , 3165 ( 1996 ) g.oshanin , j.de coninck , m.moreau and s.f.burlatsky , in : _ instabilities and non - equilibrium structures vii _ , ed . e tirapegui , ( kluwer academic pub . , dordrecht ) , to be published ; cond - mat/9910243 c.landim , s.olla and s.b.volchan , commun . math . phys . * 192 * , 287 ( 1998 ) q.gu , e.a.schiff , s.grebner and r.schwartz , phys . * 76 * , 3196 ( 1996 ) f.amblard , a.c.maggs , b.yurke , a.n.pargellis and s.leibler , phys . rev . lett . * 77 * , 4470 ( 1996 ) e.barkai and j.klafter , phys . lett . * 81 * , 1134 ( 1998 ) e.barkai and v.n.fleurov , phys . e * 58 * , 1296 ( 1998 ) j.kurchan , j. phys . a * 31 * , 3719 ( 1998 ) l.f.cugliandolo , d.s.dean and j.kurchan , phys . lett . * 79 * , 2168 ( 1997 ) p.g.de gennes , recherche * 7 * , 919 ( 1976 ) y.gefen , s.alexander and a.aharony , phys . * 50 * , 77 ( 1983 ) h.bttger and v.v.bryskin , phys . status solidi b * 113 * , 9 ( 1982 ) m.barma and d.dhar , j. phys . c * 16 * , 1451 ( 1983 ) d.dhar and d.stauffer , int c * 9 * , 349 ( 1998 ) s.r.white and m.barma , j. phys . a * 17 * , 2995 ( 1984 ) y.gefen and i.goldhirsch , j. phys . a * 18 * , l1037 ( 1985 ) k.nakazato and k.kitahara , prog . . phys . * 64 * , 2261 ( 1980 ) k.w.kehr and k.binder , in : _ application of the monte carlo method in statistical physics _ , ed . k.binder , ( springer - verlag , berlin , 1987 ) and references therein . m.brummelhuis and h.hilhorst , physica a * 156 * , 575 ( 1989 ) m .- c.desjonqures and d.spanjaard , _ concepts in surface physics _ , ( springer verlag , berlin , 1996 ) r.a.tahir-kheli and r.j.elliott , phys . b * 27 * , 844 ( 1983 ) h.van beijeren and r.kutner , phys . lett . * 55 * , 238 ( 1985 ) s. katz , j.l . lebowitz and h. spohn , phys . b * 28 * , 1655 ( 1983 ) ; j. stat . phys . * 34 * , 497 ( 1984 ) s.n . majumdar and m. barma , phys . b * 44 * , 5306 ( 1991 ) s.n . majumdar and m. barma , physica a * 177 * , 366 ( 1991 ) y.levin , j.j.arenzon and m.sellitto , europhys . * 55 * , 767 ( 2001 ) o.bnichou , a.m.cazabat , j.de coninck , m.moreau and g.oshanin , phys . lett . * 84 * , 511 ( 2000 ) ; phys . b * 63 * , 235413 ( 2001 ) j.lyklema , _ solid - liquid interfaces _ , fundamentals of interface and colloid science , vol . * 2 * , ( academic - harcourt - brace , new york , 1995 ) s.f.burlatsky , g.oshanin , m.mogutov and m.moreau , phys . a * 166 * , 230 ( 1992 ) j.crank , _ free and moving boundary problems _ , ( oxford university press , oxford , 1987 ) the authors wish to thank professor j.l.lebowitz for posing us this question and encouraging our analysis . d.b.abraham , p.collet , j.de coninck and f.dunlop , phys . lett . * 65 * , 195 ( 1990 ) d.b.abraham , p.collet , j.de coninck and f.dunlop , j. stat . phys . * 61 * , 509 ( 1990 ) j.de coninck and f.dunlop , j. stat . phys . * 47 * , 827 ( 1987 ) g.oshanin , a.m.cazabat , j.de coninck and m.moreau , phys . e * 58 * , r20 ( 1998 ) g.oshanin , a.m.cazabat , j.de coninck and m.moreau , j. of mol . liquids * 76 * , 195 ( 1998 ) r.arratia , z. ann . probab . * 11 * , 362 ( 1983 ) r.f.steiner , j. chem . phys * 22 * , 1458 ( 1954 ) t.l . hill , j. chem . phys * 20 * , 1259 ( 1952 ) b.d.hughes , _ random walks and random environments _ , ( oxford science publ . , oxford , 1995 ) p.flajolet and a.m.odlyzko , siam journal on discrete math . * 3 * , 216 ( 1990 ) ajay and r.g.palmer , j. phys . a * 23 * , 2139 ( 1990 ) t.j.newman , phys . b * 59 * , 13754 ( 1999 ) z.toroczkai , int b11 * , 3343 ( 1997 ) r.k.p.zia and z.toroczkai , j. phys . a * 31 * , 9667 ( 1998 ) m.j.a.m . brummelhuis and h.j.hilhorst , physica a * 156 * , 575 ( 1989 ) o.bnichou and g.oshanin , phys . e * 66 * , 031101 ( 2002 )	in this presentation we overview some recent results on biased tracer diffusion in lattice gases . we consider both models in which the gas particles density is explicitly conserved and situations in which the lattice gas particles undergo continuous exchanges with a reservoir , which case is appropriate , e.g. , to adsorbed monolayers in contact with the vapor phase . for all these models we determine , in some cases exactly and in other ones - using a certain decoupling approximation , the mean displacement of a tracer particle ( tp ) driven by a constant external force in a dynamical background formed by the lattice gas particles whose transition rates are symmetric . evaluating the tp mean displacement explicitly we are able to define the tp mobility , which allows us to demonstrate that the einstein relation between the tp mobility and the diffusivity generally holds , despite the fact that in some cases diffusion is anomalous . for models treated within the framework of the decoupling approximation , our analytical results are confirmed by monte carlo simulations . perturbance of the lattice gas particles distribution due to the presence of a biased tp and the form of the particle density profiles are also discussed . * biased tracer diffusion in hard - core lattice gases : some notes on the validity of the einstein relation . * g.oshanin@xmath0 , o.bnichou@xmath1 , s.f.burlatsky@xmath2 and m.moreau@xmath0 @xmath0 _ laboratoire de physique thorique des liquides , + universit paris 6 , 4 place jussieu , 75252 paris , france _ _ @xmath1 laboratoire de physique de la matire condense , + collge de france , 11 place m.berthelot , 75252 paris cedex 05 , france _ @xmath3 _ united technologies research center , + united technologies corporation , + 411 silver lane , 129 - 21 east hartford , ct 06108 , usa _
geometrically frustrated quantum magnets above one dimension ( 1d ) have attracted strong interest in recent years @xcite . from an experimental point of view this is due in part to the discovery of materials like the two - dimensional ( 2d ) spin - dimer system srcu@xmath1(bo@xmath2)@xmath1 @xcite and the three - dimensional ( 3d ) tetrahedral compounds cu@xmath1te@xmath1o@xmath3x@xmath1 with x = cl , br @xcite which exhibit strong frustration of the magnetic exchange . from a theoretical point of view , the prime interest in such systems stems from the possibility of realizing quantum disordered states and spin liquid behavior . particularly promising candidates in this direction are systems which comprise tetrahedra coupled into 2d or 3d networks . indeed , experimental observations on pyrochlore materials , like csnicrf@xmath4 @xcite or tb@xmath1ti@xmath1o@xmath5 @xcite suggest the possibility of liquid - like behavior . in contrast to the latter systems , which are essentially classical due to ising anisotropy and spins larger than 1/2 , the @xmath6 invariant spin-1/2 heisenberg antiferromagnet on the 3d pyrochlore lattice is an open issue . experiments on the @xmath7 system y@xmath1ir@xmath1o@xmath5 @xcite seem to indicate the presence of quantum spin liquid state . moments are located on solid circles . crossed dimers at site @xmath8 ( bold solid lines ) are locally coupled via @xmath9 ( dashed lines ) and non - locally coupled to nearest neighbors dimer by @xmath9 and @xmath10 ( doted lines ) . on dimer exchange @xmath11 will be set to unity hereafter.,width=234 ] to reduce the complexity of the 3d system , a 2d projection , i.e. the planar pyrochlore , also known as checkerboard or crossed chain model ( ccm ) , which is shown in fig.[cdm - lattice ] for @xmath12 , has been considered as a simplified starting point . the ccm model and some of its anisotropic generalizations have been studied by employing a variety of techniques . analytic methods include semiclassical @xmath13 analysis @xcite , large - n limit of the sp(n ) model @xcite , series expansion ( se ) by unitary transformation @xcite and quantum field theory ( qft ) methods @xcite . besides , this model has been studied extensively by means of exact diagonalization ( ed ) techniques @xcite . these investigations have given a fairly complete picture of the phases of the system at different sectors of parameters space . at the checkerboard point ( @xmath12 ) a valence bond crystal ( vbc ) ground state with a spin gap has been found @xcite . for @xmath14 model of fig.[cdm - lattice ] tends to the 2d spin@xmath0 heisenberg model on the square lattice , which is known to have gapless magnon excitations and afm - lro . from this limit , by means of a phenomenological approach within the framework of the landau - ginzburg - wilson ( lgw ) theory , it has been suggested recently that the transition between the ( afm - lro ) phase and the vbc , as @xmath15 are increased , could happen within a coexistence region or by a first order transition @xcite but this point deserves more investigation . previous results from ed up to 36 spins for @xmath16 in ref.@xcite predict a finite gap at @xmath17 and an extrapolated closure of the gap at @xmath18 . furthermore , semiclassical approaches predict stability of the afm - lro phase , in the case of @xmath7 at @xmath16 for @xmath19 . in the opposite limit of weakly coupled spin@xmath0 chains , i.e. @xmath20 in fig.[cdm - lattice ] , there are controversial results . on the one hand , previous findings based on qft methods @xcite suggested that the model exhibits a 2d spin liquid ground state , being an example of the so called _ sliding luttinger liquid _ ( sll ) which is characterized by the absence of lro and the elementary excitations being massless , deconfined spinons . it has been suggested that the prime reason for this behavior stems from the momentum dependence of the completely frustrated inter chain coupling . the latter vanishes exactly at the afm nesting vector , suppressing a conventional magnetic instability . these results were corroborated subsequently by numerical studies @xcite , where a range @xmath21 $ ] for the extension of an effective 1d behavior was suggested . on the other hand , by employing bosonization and renormalization group methods , starykh et al . @xcite have recently suggested that instead of exhibiting a sll phase , the model of fig.[cdm - lattice ] enters a gapped _ crossed dimer _ phase for @xmath20 , with a spin gap @xmath22 which opens according to @xmath23 . in this situation it is desirable to obtain results from complementary methods , which will be the primary goal of this work . therefore , in the following analysis we will employ a se in terms of @xmath9 and @xmath10 as in fig.[cdm - lattice ] and will evaluate the ground state energy and the elementary excitations . our findings will support the existence of a crossed dimer phase , in agreement with the results of ref.@xcite . to perform a se , applicable in the vicinity of the anticipated sll phase we decompose the planar pyrochlore lattice into a set of crossed , _ dimerized _ chains as shown in fig.[cdm - lattice ] . i.e. the model represents a 2d array of afm coupled crossed spin@xmath0 dimers . we will set @xmath24 hereafter . the hamiltonian reads @xmath25 in which @xmath26 refers to the local crossed dimer unit at site @xmath8 , given by @xmath27 where all spins are at site @xmath8 and @xmath28 . + @xmath29 and @xmath30 in eq.([h ] ) represent the couplings between the crossed dimers at site @xmath8 with their next nearest neighbors via @xmath9 and @xmath10 respectively , i.e. @xmath31 ; \nonumber \\ v_d & = & j_d [ \mathbf{s}_{3 ; x , y } \cdot \mathbf{s}_{1 ; x+1,y+1 } + \mathbf{s}_{2 ; x , y } \cdot \mathbf{s}_{4 ; x-1,y+1 } ] . \nonumber \label{v}\end{aligned}\ ] ] table [ tableh0 ] depicts the eigenstates of @xmath26 . .eigenstates of a local crossed dimer unit @xmath26at site @xmath8 . each state is labelled by the quantum numbers : @xmath32 , @xmath33 and @xmath34 and the energy @xmath35 . [ cols="^,^,^,^,^",options="header " , ] from eq.([h0 ] ) it is clear that each state of @xmath26 can be labelled by the following quantum numbers : total spin @xmath32 , the spin of the dimer on the positive(negative ) diagonal , i.e. the @xmath36 and the energy @xmath35 . as one can observe , for @xmath37 the ground state is @xmath38 , i.e. both dimers are singlets . besides , for @xmath39 the first excited states are @xmath40 i.e. one of the dimers on the positive(negative ) diagonal is a triplet while the one on the other diagonal is a singlet . the other states are : a total spin singlet @xmath41 , a triplet @xmath42 and a quintet @xmath43 , all of them consisting of triplets on both diagonals . only the energies of the latter three states depend on @xmath9 . in this section we briefly describe the se expansion in terms of @xmath44 . to this end we rewrite eq.([h ] ) as @xmath45 @xmath46 has an equidistant ladder spectrum ( table [ tableh0 ] ) which we label with the _ total particle - number _ operator : @xmath47 . the _ vacuum _ corresponds to the @xmath48 sector : @xmath49 . _ one - particle _ states correspond to the @xmath50 sector , whose basis states @xmath51 are composed of a local triplet on the positive or negative diagonal excited from the vacuum : @xmath52 . the @xmath53 sector is of multi - particle nature . the @xmath54 operators in eq.([hrew ] ) refer to the local contributions proportional to @xmath9 in @xmath55 . finally , the third term in the same eq . represents the inter dimer coupling via @xmath44 . the @xmath56 and @xmath57 operators @xcite non - locally create(destroy ) @xmath58 energy quanta within the ladder spectrum of @xmath46 . note that only terms containing two dimers at different @xmath8 sites are present in @xmath44 . this imposes the constraint @xmath59 . models of type eq.([hrew ] ) allow for se using the flow equation method of wegner @xcite . there , by means of a continuous unitary transformation , one maps @xmath60 with @xmath61 in which the @xmath62 operators involve @xmath63 and @xmath64 products of type-@xmath56 and @xmath57 operators , respectively . we refer to @xcite for further details . the main advantage of the flow equation method is that unlike @xmath65 , the effective hamiltonian @xmath66 is constructed to _ conserve _ the total number of particles @xmath67 of @xmath46 at each order in the expansion . this allows for the se of several quantities from the bare eigenstates of @xmath46 as we show in the following sections . the ground state energy allows for an important consistency check of our calculation with respect to existing literature and other se results . as mentioned above , there is evidence for a _ quadrumer _ vbc phase of the ccm at the checkerboard point ( @xmath68 ) . this is based on ed @xcite and se starting from the quadrumer limit of the ccm @xcite . moreover , starykh et al . @xcite , have predicted a first order transition or a coexistence phase at some intermediate value of @xmath69 and @xmath70 between the vbc and a crossed dimer phase . therefore , in order for the present se to give results consistent with them for @xmath9 and @xmath10 in the vicinity of the checkerboard point , we expect the ground state energy of the crossed dimer state to be larger than that of the quadrumer vbc at @xmath68 . for @xmath37 the ground state of @xmath46 is the @xmath48 sector , i.e. @xmath71 . q - conservation of @xmath66 allows for the calculation of the evolution of this state as a se(@xmath72 ) by means of @xmath73 the actual calculation can only be done on finite clusters . to obtain results to @xmath74 , valid in the thermodynamic limit , i.e. for infinite sized systems , matrix elements given by eq.([egccm ] ) have to be evaluated on clusters with periodic boundary conditions ( pbc ) , sufficiently large to avoid a wrap - around at graph length @xmath75 @xcite . in our case , the size of the hilbert space , considering both perturbations @xmath9 and @xmath10 , restricts the calculation at @xmath76 due to memory limitations . we find this enough to quantitatively describe the combined effects of @xmath9 and @xmath10 in the range of interest . at @xmath77 the ground state energy per spin and pbc , @xmath78 , is given by @xmath79 as a check , note that @xmath80 reduces to the @xmath81 dimer expansion of the ground state energy of the dimerized spin@xmath0 chain @xcite . from bethe ansatz @xcite the ground state energy for the homogeneous spin@xmath0 chain is known : @xmath82 @xcite . comparing it with @xmath83 from eq.([egccm ] ) we obtain a relative error of @xmath84 , which gives a rough estimation of the convergence @xcite . fig.[eg - plot ] depicts the evolution of the ground state energy as given by eq.([egccm ] ) vs @xmath10 for different values of @xmath9 . as shown , @xmath78 monotonously decreases as @xmath9 is incremented from @xmath85 to @xmath86 , for fixed values of @xmath10 . the position @xmath87 of the maximum of @xmath78 vs @xmath10 shifts from @xmath88 to @xmath89 for @xmath9 taking fixed values from @xmath85 to @xmath86 , respectively . at the checkerboard point our se yields @xmath90 . note that our value is larger than the value of @xmath91 which has been obtained from the quadrumer se in ref.@xcite . while the expansion parameters and their respective orders can not be compared directly between the present se and that of ref.@xcite these findings are at least consistent with the stability of the quadrumer vbc in the vicinity of the checkerboard point . this , together with the fact that @xmath92 has a degenerated ground state ( see table 1 ) , suggests that eq.([egccm ] ) , at least at the checkerboard point , does not represent the ground state energy of the system anymore , which is @xmath91 . next we evaluate the dispersion of the one - triplet states . to this end it is necessary to diagonalize @xmath66 in the @xmath50 sector of @xmath93 , i.e. the subspace spanned by the vectors @xmath94 . first we switch to momentum space @xmath95,\ ] ] in which pbc is assumed and @xmath96 is the number of sites . because of the degeneracy of the q=1 sector , @xmath66 is a @xmath97 matrix in the basis given by eq.([fourier ] ) , whose entries @xmath98 are @xmath99,\end{aligned}\ ] ] where @xmath100 and @xmath101 are the hopping amplitudes from a @xmath102type of triplet on site @xmath103 to a @xmath104type of triplet on site @xmath8 . evaluation of the @xmath105 s is facilitated by symmetry which leaves only a subset of them to be independent . to obtain the hopping amplitudes valid in the thermodynamic limit , the @xmath106 s are calculated on finite clusters with open boundary conditions ( obc ) , and a cluster - topology suitable to embed all the paths of length @xmath75 connecting the site @xmath107 with @xmath108 at @xmath109 of the perturbation . from this , the triplet dispersion is obtained as @xmath110 where @xmath111 and the second term refers to the ground state energy which has to be evaluated on the identical cluster as @xmath112 . the triplet energy @xmath113 results from diagonalizing the @xmath114 matrix of eq.([matrix - trip ] ) , leading to two excitation branches . we have calculated the triplet dispersion @xmath115 up to @xmath77 @xcite . upon rotating @xmath116 along the positive and negative diagonals , by means of @xmath117 , the two branches of eq.([disp - eq ] ) have to reduce to the triplet dispersion of the dimerized spin@xmath0 chains along positive and negative diagonals for @xmath118 . for consistency , we have checked that this is satisfied up to the order of our se by comparing with existing se results on the single dimerized chain @xcite . using the results from the previous section we may now turn to one of the main objectives of this work , i.e. to study the evolution of the triplet gap under the combined effect of @xmath9 and @xmath10 . this will enable us to conclude on the stability of the dimerization along the chains and therefore to discriminate between the controversial proposals on the stability of the sll . the triplet gap : @xmath119 is obtained from the minimum in @xmath120-space of the two one - triplet branches . for @xmath121 we find the triplet gap to be at @xmath122 . it is given by @xmath123 we have analyzed @xmath124 both by its plain se and by using a biased - pad approximation@xcite . the latter consists of an extrapolation biased by the known analytical form of the gap - closure of the dimerized spin@xmath0 chain as a function of @xmath10 @xcite , i.e. @xmath125 . using this , @xmath10 is replaced in favor of @xmath126 in eq.([ccm - gap ] ) . then , standard pad approximation@xcite is performed around @xmath127 and finally @xmath128 is re - inserted , arriving at a transformed pad approximation in terms of the original variable @xmath10 . fig.[gap - jd ] depicts the evolution of @xmath124 as a function of @xmath10 for different values of @xmath9 . solid lines show the results from the plain series , i.e. eq.([ccm - gap ] ) and dashed lines exhibit the biased - pad(2,2 ) approximants . as the inset of fig.[gap - jd ] shows , the approximant describes the closure of the gap at @xmath129 very well . this justifies the use of this extrapolation technique and shows that our expansion can be employed to analyze the coupling of the spin@xmath0 chains via @xmath9 . as a first result from fig.[gap - jd ] , both the plain series and the biased - pad clearly show an _ opening _ of the gap at @xmath70 for increasing values of @xmath9 . as it is also obvious from these plots , this tendency persists for all @xmath130 . second , fig.[gap - jd ] shows , that the biased pad approximant leads to observable differences from the plain se only in the vicinity of the gap - closure . now , we consider the opening of the spin gap at @xmath131 for @xmath132 . to this end , we first shift the zero of energy in order to suppress the finite gap of the se at @xmath9=0 and @xmath10=1 , i.e. we consider @xmath133 . by this we account for the inability of the se to provide for a truly gapless spectrum in the limit of decoupled afm heisenberg chains . as it is obvious from fig.[gap - jd ] , this shift is very small for the biased pad , i.e. @xmath134 and @xmath135 for the plain se . the resulting spin gap versus @xmath9 is shown in fig.[gap - jp ] where the solid line represents the plain series and the dashed line corresponds to the biased - pad(2,2 ) approximant . these results may be compared with recent findings of starykh et al . @xcite . using qft methods , i.e. renormalization group and bosonization , they have obtained an analytic expression for the spin gap , which is quadratic in @xmath9 , namely @xmath136 where @xmath137 . this function is shown in the same fig.[gap - jp ] with a dash - dotted line . obviously , there is a very good agreement up to @xmath138 . this is remarkable , given the completely different nature of our approach and that of ref.@xcite . we note that the same agreement is found when we employ standard dlog - pad approximants instead of biased - pad . therefore , we are strongly tempted to claim that the stable phase of the ccm at finite @xmath9 and at @xmath131 is indeed gapped . this is inconsistent with a sll state . while fig.[gap - jp ] seems to suggest an ever - increasing spin gap , one has to remember , that at @xmath139 the nature of the first excited state of @xmath140 changes from the triplet @xmath141 to the singlet @xmath142 ( table 1 ) . this imposes an upper limit on the range of validity of the se . in summary , using series expansion , we have studied the zero temperature properties of the planar pyrochlore quantum magnet focussing on its behavior close to the crossed chain limit . in particular , we have considered the ground state energy , the elementary triplet dispersion and the spin gap . in relation with the ground state energy calculations our findings are consistent with previous results regarding the stability of the quadrumer vbc at the checkerboard point . our main result regards the existence of crossed dimer phase . our calculations univocally show an opening of the gap @xmath143 . moreover , our expansion ( eq.([ccm - gap ] ) ) , allows to route the physical origin of the crossed dimer phase by showing that it is adiabatically connected with decoupled crossed dimer units . on the other hand , the very good agreement with a recent field theory approach by starykh et.al . @xcite even for @xmath144 consolidates the hypothesis about the existence of a crossed dimer phase in the region of parameters space @xmath131 , @xmath145 and indicates the absence of luttinger liquid behavior , the so called sliding luttinger liquid phase , in this region of parameters space . the authors thank a. honecker for useful discussions . this research was supported in part through dfg grant no . br 1084/4 - 1 . the explicit expressions of the hopping elements @xmath105 and the triplet dispersion @xmath149 are too lengthy to be displayed in written form and will be made available electronically on request from the authors .	using series expansion based on the flow equation method we study the zero temperature properties of the spin@xmath0 planar pyrochlore antiferromagnet in the limit of strong diagonal coupling . starting from the limit of decoupled crossed dimers we analyze the evolution of the ground state energy and the elementary triplet excitations in terms of two coupling constants describing the inter dimer exchange . in the limit of weakly coupled spin@xmath0 chains we find that the fully frustrated inter chain coupling is critical , forcing a dimer phase which adiabatically connects to the state of isolated dimers . this result is consistent with findings by o. starykh , a. furusaki and l. balents ( phys . rev . b * 72 * , 094416 ( 2005 ) ) which is inconsistent with a two - dimensional sliding luttinger liquid phase at finite inter chain coupling .
the idea that we could live on a brane embedded in a higher - dimensional space dates back to refs . @xcite where it was proposed that the universe be modeled as a topological defect ( a domain wall or a string ) in a higher - dimensional space - time . the key feature of these models is the localization of matter on the brane , which makes the theory look four - dimensional at low energies . in the setup of ref . @xcite this was achieved by means of yukawa interaction between the domain wall and the fermion field . the latter has a zero mode localized on the brane , as well as the continuum modes separated from the zero mode by a gap . at low energies , only the zero mode remains in the spectrum and plays the role of a four - dimensional particle . it was suggested later @xcite that the gravitational attraction may also trap particles on the brane . this idea has been the subject of renewed interest since it has been shown @xcite that the graviton itself can be trapped on the brane by the gravitational force , leading to effectively four - dimensional gravity . considering a brane embedded in a 5d anti - de sitter ( ads@xmath4 ) space - time , the authors of ref . @xcite have found that in this `` warped '' geometry the 5d graviton has a zero - energy bound state , which , being localized on the brane , can be interpreted as the ordinary 4d graviton . even though there is no gap separating the zero mode of graviton from the continuum , the law of gravity on the brane is essentially newtonian , the corrections yielded by light continuum modes modifying this behavior only at very short distances . soon after , a model was proposed @xcite in which 4d newton s law is reproduced only at intermediate scales and gets modified at very large scales . the geometry considered consisted of a flat 5d space glued onto a portion of ads@xmath4 space . it has been shown that , even though there is no localized graviton , newton s gravity can still be recovered on the brane at intermediate scales . at short distances , the gravitational potential receives corrections similar to those encountered in the model of ref . @xcite . at very large scales , the extra dimensions `` open up '' and the gravitational potential receives `` five - dimensional '' contributions from the tensor modes , that is contributions scaling as @xmath5 , @xmath6 being the distance between sources . this behavior was related @xcite to the presence of a resonant mode a `` quasi - bound state '' at zero energy , with lifetime proportional to the long - distance scale of the model ( see also ref . @xcite itself ) . subsequent works @xcite made manifest the importance of the scalar sector ( and namely the radion mode ) in the description of the effective gravity on the brane in models with metastable gravitons . indeed , being constructed from the massive modes , the propagator of the effective 4d graviton is likely to have an incorrect tensor structure , leading to predictions inconsistent with the observations @xcite . coupling to the trace of the energy - momentum tensor , radion contributes to the graviton propagator and thus makes it possible to recover einstein gravity at the intermediate distances @xcite . model presented in @xcite has however a serious drawback : it requires the use of the negative tension branes or , in other words , violation of the condition of positivity of energy @xcite . it was shown @xcite that this forces radion to become a ghost @xcite and causes the effective gravity at very large distances to become scalar anti - gravity @xcite . while this behavior occurs at the experimentally inaccessible scales , presence of negative norm states certainly raises doubts concerning the consistency of the model . an alternative brane model producing an infrared modification of gravity was constructed in @xcite using a thin brane embedded in five - dimensional minkowski space . the action included a four - dimensional einstein term on the brane , induced by radiative corrections due to the matter fields localized there . this term was shown to dominate the gravitational interactions of the brane matter at intermediate scales , giving four - dimensional gravity at intermediate scales , while at large scales it became five - dimensional . this setup was subsequently extended to higher dimensions @xcite and its generalizations to smooth branes were also considered @xcite . recently , the idea of induced gravity was applied to the warped space - times @xcite . as yet , the setups with induced gravity are not entirely satisfactory : most of the models suffer from ghosts and/or strong gravity problems @xcite . nevertheless , the idea of infrared modification of gravity remains very appealing as it is thought @xcite that it could be a solution to the cosmological constant problem . the feature shared by these models is that at ultra - large scales gravity becomes higher - dimensional , that is the gravitational potential scales as @xmath7 , @xmath6 being the distance between sources and @xmath8 the number of extra dimensions ( at least as far as the tensor sector is concerned ) . it is natural to wonder whether this feature is inherent to the quasilocalized gravity on branes embedded in higher dimension space - times or if it could exhibit some different , more general , behaviors . more generally , one may ask what kinds of large - distance modifications of gravity are in principle possible in this setup and under what conditions they can be realized . in order to cast some light on this issue , we study in this paper tensor perturbations of the metric around a fairly generic five - dimensional warped background , in which the graviton is not fully localized on the brane ( becoming a metastable state ) . to some extent , our model can be considered as a toy model of quasilocalized gravity . of course , as shows the example of @xcite , studying only the tensor sector of metric perturbations is unlikely to give a full description of effective gravity on branes and we do not aspire to present a realistic solution for quasilocalization of gravity . the aim of this work is rather to signal a peculiar behavior of the potential yielded by the tensor perturbations in such kinds of backgrounds . for the sake of simplicity , throughout the paper we will loosely refer to the contribution of the tensor modes to the gravitational potential as to the potential itself , bearing in mind that in any concrete model one should account for the contribution from the scalar sector as well . same applies to our description of gravitational waves . we find that our model gives 4d gravitational potential at intermediate scales quite similarly to the situation encountered in the model of ref . this `` quasilocalization '' of gravity is a generic feature of the geometry which we consider whether or not the embedding space - time is asymptotically flat . at large distances , we observe a novel behavior : the potential gets modified and is of type @xmath1 , @xmath2 being a function of the parameters of the model . it becomes 5d ( @xmath3 ) only when the geometry is asymptotically flat . thus , the gravitational potential does not necessarily exhibit 5d behavior at very large distances . although throughout this paper we use the example and the language of gravity , our analysis is in fact quite universal , in that it is not constrained to the gravity sector , but can apply to the matter sector as well . indeed , our considerations are based on studying the properties of a shrdinger - like equation governing the behavior of the metric perturbations and , by consequence , the effective physics on the brane . actually , irrespective of the spin of the matter fields present in the bulk , the mass spectrum of the effective four - dimensional theory is determined by a schrdinger - like equation of some type ( or a system of thereof ) . using this fact , several authors have studied the localization of different matter fields in the ads@xmath4 background of ref . @xcite and its variations : it was shown that , similarly to gravitons , the spectrum of massless bulk scalars @xcite consists of a localized zero mode , followed by the continuum of arbitrarily light states . massless fermions @xcite can also be localized on the brane and mechanisms to localize gauge fields were proposed @xcite . localization of massive scalars and fermions was discussed in ref . @xcite ( see also ref . @xcite ) , where it was shown that while there are no truly localized states ( massive or massless ) , the setup allows for metastable massive particles on the brane , having a small , but finite , probability of tunneling into the bulk . in the scalar case , these resonant states were shown to give @xmath0 contribution to the effective static potential on the brane , while the light states of the continuum give a power - law behavior at large distances . our analysis completes the picture , revealing that , under certain conditions , in the absence of true bound states and massive resonances , it is possible to have an `` almost localized '' massless state , inducing a @xmath0 potential on the brane evolving towards a ( fractional ) power - law behavior at large distances . let us begin with the general setup . we consider 5d `` warped '' space - times preserving 4d poincar symmetry : @xmath9 where the `` warp factor '' @xmath10 in an even and non - decreasing function of @xmath11 . we suppose that the ordinary matter is localized on the brane centered at @xmath12 . in order to study the nature of the gravitational interactions experienced by this matter , we need to study perturbations around the background metric . we will consider exclusively the tensor perturbation @xmath13 , @xmath14 and @xmath15 being the four - dimensional tensor indices and @xmath16 metric determinant . the behavior of this perturbation is governed by the schrdinger - like equation @xmath17 where @xmath18 and the potential is given in terms of the warp factor : @xmath19 which is a form familiar from the supersymmetric quantum mechanics with the `` superpotential '' @xmath20 the factorization of the hamiltonian : @xmath21 ensures that the zero - energy mode @xmath22\ ] ] is the ground state of the system . in this way , studying the influence of the tensor modes on the effective gravity on the brane is reduced to studying the properties of the spectrum of the one - dimensional quantum mechanical system @xmath23 . the quantity of particular interest is the spectral density @xmath24 , as it enters the retarded green s function @xmath25 which determines the gravitational interactions on the brane and consequently the observable physical quantities , such as the induced gravitational potential and radiation of gravitational waves . let us remind that the ( five - dimensional ) retarded green s function @xmath25 of the tensor sector of the linearized einstein equations can be constructed from the full set of eigenmodes of eqn . @xmath23 via spectral decomposition : @xmath26 the purpose of this work is not to present any particular realistic setup for the quasilocalized gravity , but rather to study general properties of warped space - times in which this phenomenon can arise and determine what are the possible large distance behaviors of the gravitational potential ( induced by the tensor modes ) . with this aim in mind , we construct and study a toy model for a generic geometry allowing ( quasi)localization of gravity . concretely , we consider a class of potentials @xmath27 with `` volcano '' shape , possibly with tails for which we choose @xmath28 fall - off , familiar from the randall - sundrum model @xcite . knowing that the potential in question necessarily follows from a `` superpotential '' , we start our study by defining one of the following form : @xmath29 & z_0\leq z \leq z_c\ , \\ \displaystyle{\frac{c}{z - z_1 } } & z\geq z_c\ .\\ \end{cases}\ ] ] the motivation for choosing this particular form of superpotential is twofold : not only does it yield a simple potential for which an exact solution of the schrdinger equation can be determined , but also allows a unified description of the ( quasi)localization , naturally including regimes already discussed in the literature @xcite . our choice of superpotential corresponds in general to geometries which are not asymptotically flat except when the limit is considered . to ensure that the resulting potential does not have singularities stronger than finite jumps we demand that @xmath30 be continuous . this requirement entails relations between the constants : @xmath31=a\tan(az_0 ) } \ , \\ \displaystyle{b\tanh\left[b\left(z_2-z_c\right)\right]=\frac{c}{z_c - z_1 } } \ . \end{cases}\ ] ] the potential generated by the superpotential @xmath32 is symmetric and has the desired `` volcano - like '' shape . it is given by : @xmath33 the parameters of the potential must satisfy the condition following from the constraints @xmath34 by eliminating @xmath35 : @xmath36\right]=\left(z_c - z_1\right)\left[a\tan(az_0)-b\tanh\left[b\left(z_c - z_0\right)\right]\right ] \ .\ ] ] in order for @xmath34 to remain valid we must also have @xmath37 in spite of its simplicity , the potential @xmath27 presents a range of behaviors , depending on the choice of parameters , and thus generates different localization schemes for gravity . to start , in the limit @xmath38 , @xmath27 becomes a simple rectangular potential well , allowing a localized symmetric zero - mode of the form : @xmath39 where @xmath40 followed by the continuum spectrum starting at @xmath41 . for finite @xmath42 , the asymptotics of the potential makes the continuum descend towards @xmath43 . condition @xmath44 guarantees the absence of growing contribution to the @xmath43 wave function in the region @xmath45 , where it consequently falls off as @xmath46 . when @xmath47 , it is therefore still normalizable and , at the bottom of the continuum , we still have a localized zero - mode . the resulting gravity on the brane will then be of the type described in ref . @xcite . when @xmath48 , @xmath49 is no longer normalizable and the corresponding state can be qualified as `` quasi - bound '' . in the particular case @xmath50 , @xmath27 is identical to the `` volcano box '' potential used in @xcite in order to explain the quasilocalization of gravity in the asymptotically flat model @xcite . we claim that quasilocalization of gravity is present for the whole family of potentials with @xmath48 . let us from now on concentrate on this family . the solution for a symmetric continuum wave function has the form : @xmath51 } & z\geq z_c\ . \end{cases}\ ] ] where @xmath52 the coefficients in @xmath53 are determined , as usual , by the requirement of continuity of the wave function and its derivative at @xmath54 and @xmath55 : @xmath56n_\nu(\mu(z_c - z_1))-\mu f_{+}(\mu)\ , n_{\nu+1}(\mu(z_c - z_1))\right\ } \ , \\ b_2(\mu)&=-\frac{\pi}{2}\sqrt{z_c - z_1}\left\{\left[\kappa f_{-}(\mu)+\frac{2\nu+1}{2(z_c - z_1)}f_{+}(\mu)\right]j_\nu(\mu(z_c - z_1))-\mu f_{+}(\mu)\ , j_{\nu+1}(\mu(z_c - z_1))\right\}\ , \\ f_{+}(\mu)&= d_1(\mu)e^{-\kappa z_c } + d_2(\mu)e^{\kappa z_c}=\cos(kz_0)\cosh\left[\kappa(z_c - z_0)\right]-\frac{k}{\kappa}\sin(kz_0)\sinh\left[\kappa(z_c - z_0)\right]\ , \\ f_{-}(\mu)&= d_1(\mu)e^{-\kappa z_c } - d_2(\mu)e^{\kappa z_c}=-\cos(kz_0)\sinh\left[\kappa(z_c - z_0)\right]+\frac{k}{\kappa}\sin(kz_0)\cosh\left[\kappa(z_c - z_0)\right]\ .\end{aligned}\ ] ] the remaining overall constant @xmath57 , and consequently the spectral density @xmath58 can be determined from the normalization condition : @xmath59 which gives @xmath60 as we have already stressed in section [ setup ] , @xmath61 is the quantity relevant to the effective gravity on the brane and it is therefore useful to examine its behavior as a function of @xmath62 in some detail . in the following , we assume that @xmath63 , that is that the tunneling probability of the light modes through the potential barriers is very small . we suppose also that the constant @xmath64 is adjusted in such a way that the ratio @xmath65 . for @xmath66 , we can approximate @xmath67 and @xmath68 by their power series in @xmath62 : @xmath69 the condition of continuity of the superpotential @xmath44 forces @xmath70 to vanish and guarantees that @xmath71 and @xmath72 are small ( of order @xmath73 $ ] ) . as a consequence we have for @xmath61 : @xmath74 where @xmath75 stands for : @xmath76^{2 - 2\nu}e^{2b(z_c - z_0)}\ ] ] and @xmath77^{-2}\ , \qquad \alpha_1 ^ 2 = -\frac{\sin(2\nu\pi)}{2\pi b}\frac{1+bz_0}{z_c - z_1 } \ , \qquad \alpha_2 ^ 2 = \frac{1}{4\cos^2(\nu\pi)}\frac{\alpha_1 ^ 4}{\alpha_0 ^ 2 } \ .\ ] ] the equality @xmath78 shows that for small @xmath62 s the spectral density @xmath61 is a uniformly decreasing function of @xmath62 , strongly peaked at @xmath43 where it has a singularity of the form @xmath79 . the bulk of the weight of @xmath61 lies in the region between @xmath43 and @xmath80 where @xmath81 . we have therefore in our problem two characteristic mass scales , @xmath82 and @xmath83 . the assumption of small tunneling probability ensures that these scales are well separated , @xmath84 . let us remark that in the limit @xmath50 , we recover the results of @xcite and @xmath61 assumes the breit - wigner form : @xmath85 with the resonance width given by @xmath86 to conclude this chapter , let us discuss briefly the connection between the spectral density @xmath61 and the scattering matrix of our one - dimensional problem . the relevant eigenvalue of the @xmath87-matrix is : @xmath88 where the jost function @xmath89 is defined in such a way that tends to one in the non - interaction limit and is given by : @xmath90\ . \end{aligned}\ ] ] comparing the expression for the spectral density @xmath91 to @xmath92 , we see that @xmath61 can be written as @xmath93 it is known from the formal scattering theory @xcite that physical properties of a quantum system are closely related with the analytic properties of its scattering matrix . in particular , poles of the @xmath87-matrix in the complex @xmath62-plane correspond , depending on their location , to bound states , virtual states or resonances of the system . in our problem , @xmath87-matrix has in general a branch point singularity at @xmath62=0 , which it inherits from the bessel functions @xmath94 and @xmath95 entering the expression @xmath92 . the results of the next section will demonstrate that this type of singularity can also strongly influence the physics . let us now turn to the contribution of the tensor modes to the effective gravity on the brane . the static potential between two sources on the brane , separated by the distance @xmath96 receives yukawa - type contributions from all the modes and is given by : @xmath97 it is convenient to divide the integral into two parts : @xmath98 for distances @xmath99 the second integral is negligible and in the first one we can replace @xmath61 by its approximate form @xmath78 , which gives us : @xmath100 this integral is always saturated for @xmath81 , that is for @xmath101 . we can therefore extend the integration to infinity and evaluate the resulting integral in two regimes : @xmath102 and @xmath103 . in the first regime , that is for distances @xmath104 , the exponential in @xmath105 can be set to 1 and we obtain : @xmath106 where @xmath107 is defined by @xmath108 . therefore , in this ( large ) interval of distances , light modes of the continuum spectrum reproduce precisely the same @xmath0 potential we would get in the presence of a massless boson ( or , in other words , a normalizable zero mode ) . in the second regime , that is for large distances , @xmath103 , only @xmath109 will give significant contributions to the integral . we can therefore neglect @xmath75 and @xmath110 terms in the denominator of the integrand in @xmath105 , thus obtaining : @xmath111 hence , we have a modification of the @xmath0 law for large distances , with the potential of type @xmath112,where @xmath113 . thus , the gravity does not in general become five - dimensional , except in an asymptotically flat geometry ( @xmath50 ) , where the potential is : @xmath114 following @xcite , we can also study the propagation of gravitational waves generated by a periodic point - like source on the brane , @xmath115 ( we omit the four - dimensional indices ) . the field induced on the brane is given by the convolution of the source with the green s function @xmath116 @xmath117 which after inserting @xmath116 and simplifying becomes @xmath118 where again @xmath96 and @xmath119 when @xmath120 and @xmath121 when @xmath122 . only modes with @xmath120 are actually radiated ; the other ones exponentially fall off from the source . being interested in the propagation of waves , we can integrate in @xmath123 only up to @xmath124 . considering the range of frequencies @xmath125 allows us to replace @xmath61 by @xmath78 and approximate @xmath126 by @xmath127 : @xmath128 where we have extended the integration to infinity , using the fact that the integral is saturated for @xmath129 . for distances @xmath130 the phase factor in @xmath131 varies very slowly and can be set to 1 . we then obtain the usual @xmath0 dependence of wave amplitude on the distance to the source : @xmath132 for distances @xmath130 we can neglect @xmath75 and @xmath110 terms in the denominator of the integrand in @xmath131 , obtaining @xmath133 that is , the amplitude is proportional to @xmath134 , where @xmath135 . this behavior indicates dissipation of the gravitational waves into the extra dimension . in this paper we have studied a toy model for quasilocalization of gravity on a brane embedded in an asymptotically warped space - time , restricting our attention to the tensor sector of metric perturbations . while , admittedly , our model is only of limited use as a model for gravity , it allows to uncover quite peculiar behavior of the gravitational potential ( and related quantities ) induced by the tensor modes . in the absence of a normalized zero - mode , the effective potential is of the 4d form @xmath0 , regardless of whether or not the embedding space - time is asymptotically flat . at large scales , the presence of the extra dimensions becomes significant and the potential gets modified . the nature of this modification depends on the asymptotic geometry of the space - time and potential does not necessarily assume the 5d form @xmath136 . although we did not present a concrete model which would possess asymptotically non - flat solutions with quasilocalized gravity , some features inherent to such models can be derived . by einstein equations , the warped geometry ( [ metric ] ) corresponds to the matter energy - momentum tensor which obeys @xmath137 where @xmath138 and @xmath139 are the energy density and pressure in the @xmath11 direction , respectively . making use of eq . ( [ w ] ) at large @xmath11 we find that our asymptotically non - flat ansatz requires @xmath140 and thus violates the weak energy condition . in this respect , our model is quite similar to ( and may share the problems of ) that of ref . it remains yet to be seen whether these problems can be overcome . this is related to an important question , namely the impact of the scalar sector of metric perturbations on the effective gravity in a theory where the embedding space - time is not asymptotically flat and the graviton is quasi - localized . it is legitimate to suspect that it could be as important as in the model of ref . @xcite and this matter should be examined on a concrete brane setup . finally , it is worth pointing out that eq . ( [ schro ] ) which is the starting point of our analysis is quite universal , in that it describes with a suitably adjusted potential the localization of matter fields as well . therefore , our conclusions are directly applicable to the case of matter fields whenever the corresponding potential can be cast in a `` supersymmetric '' form @xmath141 . the latter condition is equivalent to the requirement that the four - dimensional particle is massless ( the massive case was treated in ref . @xcite ) . in the case of matter , the potential in eq . ( [ schro ] ) is no longer related to the metric of space - time through the relation ( [ pot ] ) and the problem of violation of the weak energy condition does not need to arise . 99 v. a. rubakov , m. shaposhnikov , phys . b * 125 * ( 1983 ) 136 k. akama , in : _ gauge theory and gravitation _ , proceedings of the international symposium , nara 1982 , ( springer - verlag , lect . notes phys . * 176 * , 1983 , p.267 ) m. visser , phys . b * 159 * ( 1985 ) 22 e. j. squires , phys . b * 167 * ( 1986 ) 286 l. randall , r. sundrum , phys . * 83 * ( 1999 ) 4690 r. gregory , v. a. rubakov , s. m. sibiryakov , phys . * 84 * ( 2000 ) 5928 g. dvali , g. gabadadze , m. porrati , phys . b * 484 * ( 2000 ) 129 r. gregory , v. a. rubakov , s. m. sibiryakov , phys . b * 489 * ( 2000 ) 203 l. pilo , r. rattazzi , a. zaffaroni , j. high energy phys . * 0007 * ( 2003 ) 056 g. dvali , g. gabadadze , m. porrati , phys . b * 485 * ( 2000 ) 208 g. dvali , g. gabadadze , phys . d * 63 * ( 2001 ) 065007 g. dvali , g. gabadadze , x. r. hou , e. sefusatti , phys . d * 67 * ( 2003 ) 044019 m. kolanovi , phys . d * 67 * ( 2003 ) 106002 s. dubovsky , m. libanov , j. high energy phys . * 11 * ( 2003 ) 038 a. padilla , hep - th/0402079 m. luty , m. porrati , r. rattazzi , j. high energy phys . * 0309 * ( 2003 ) 029 v. a. rubakov , hep - th/0303125	we present a toy model of a generic five - dimensional warped geometry in which the 4d graviton is not fully localized on the brane . studying the tensor sector of metric perturbation around this background , we find that its contribution to the effective gravitational potential is of 4d type ( @xmath0 ) at the intermediate scales and that at the large scales it becomes @xmath1 , @xmath2 being a function of the parameters of the model ( @xmath3 corresponds to the asymptotically flat geometry ) . large - distance behavior of the potential is therefore not necessarily five - dimensional . our analysis applies also to the case of quasilocalized massless particles other than graviton .
hadronic light - cone distribution amplitudes ( das ) have been known to play an essential role in the qcd description of hard exclusive processes via the factorization theorem @xcite . taking the transverse separation to zero , the factorization takes the form of a convolution of a perturbatively calculable hard - scattering amplitude and the process - independent nonperturbative das . these nonperturbative das motivated many theoretical studies to calculate meson das using nonperturbative methods such as the qcd sum rule @xcite , the chiral - quark model from the instanton vacuum @xcite , the dyson - schwinger equations ( dse ) approach @xcite , and the light - front quark model ( lfqm ) @xcite . among them , the lfqm appears to be one of the most efficient and effective tools in studying hadron physics as it takes advantage of the distinguished features of the light - front dynamics ( lfd ) @xcite . specifically , the rational energy - momentum dispersion relation of lfd , namely @xmath4 , yields the sign correlation between the light - front ( lf ) energy @xmath5 and the lf longitudinal momentum @xmath6 and leads to the suppression of vacuum fluctuations in lfd . this simplification is a remarkable advantage in lfd and facilitates the partonic interpretation of the amplitudes . based on the advantage of lfd , the lfqm has been quite successful in describing various static and non - static properties of hadrons such as meson mass spectra @xcite , the decay constants @xcite , electromagnetic and weak transition form factors @xcite and generalized parton distribution amplitudes ( gpds ) @xcite . in ref . @xcite , we have analyzed twist-2 das of pseudoscalar ( @xmath7 ) and vector ( @xmath8 ) mesons using the lfqm @xcite . in more recent works @xcite , we have extended our lfqm to analyze twist-3 pseudoscalar ( @xmath9 ) das of pseudoscalar mesons @xcite and chirality - even twist-3 ( @xmath10 ) das of vector mesons @xcite and discussed the link between the chiral symmetry of qcd and the numerical results of the lfqm . in particular , we have discussed a wave function dependence of the lf zero - mode @xcite contributions to @xmath9 and @xmath10 not only for the exactly solvable manifestly covariant bethe - salpeter ( bs ) model but also for the more phenomenologically accessible realistic lfqm @xcite using the standard lf ( slf ) approach . we also linked the covariant bs model to the standard lfqm providing the correspondence relation @xcite between the two models . effectively , we prescribed a consistent substitution for the lf vertex function in the covariant bs model with the more phenomenologically accessible gaussian wave function provided by the lfqm analysis of meson mass @xcite . the remarkable finding is that the zero - mode contribution as well as the instantaneous contribution revealed in the covariant bs model become absent in the lfqm . without involving the zero - mode and instantaneous contributions , our lfqm result of twist-3 das @xmath9 and @xmath10 provides the consistency with the chiral symmetry anticipated from qcd s conformal limit @xcite . the purpose of this work is to extend our previous work @xcite to analyze the twist-3 pseudotensor da @xmath1 of a pion within the lfqm . the paper is organized as follows : in sec . [ sec : ii ] , we compute the twist-3 pseudotensor da @xmath1 in an exactly solvable model based on the covariant bs model of ( 3 + 1)-dimensional fermion field theory . we then linked the covariant bs model to the standard lfqm following the correspondence relation @xcite between the covariant bs and lfqm models and present the form of @xmath1 as well as @xmath0 in our lfqm . in sec . [ sec : iii ] , we present our numerical results . summary and discussion follow in sec . [ sec : iv ] . the @xmath11 and @xmath12 are defined in terms of the following matrix elements of gauge invariant nonlocal operators in the light - cone gauge @xcite : [ deq:3.1 ] ^p_3;m(x ) & = & ^_- e^-i(p ) 0\|\|q()i_5 q(-)\|m(p ) , + ^_3;m(x ) & = & - ^_- ^x_0 dx e^-i(p ) 0\|\|q ( ) i(/p /- p)_5 q(-)\|m(p ) , respectively , where @xmath13 and @xmath14 is the four - momentum of the meson ( @xmath15 ) and @xmath16 corresponds to the longitudinal momentum fraction carried by the quark and @xmath17 . the normalization parameter @xmath18 results from quark condensate . for the pion , @xmath19 from the gell - mann - oakes - renner relation @xcite . the nonlocal matrix elements @xmath20 for pseudoscalar ( @xmath21 ) and pseudotensor ( @xmath22 ) channels are given by the following momentum integral in two - point function of the manifestly covariant bs model [ deq:4 ] m_= n_c e^-i k e^-i ( k - p ) h_0 , where @xmath23 denotes the number of colors . the quark propagators of mass @xmath24 and @xmath25 carry the internal four - momenta @xmath26 and @xmath27 , respectively . in order to regularize the covariant loop , we use the usual multipole ansatz @xcite for the @xmath28 bound - state vertex function @xmath29 of a meson : @xmath30 , where @xmath31 and @xmath32 are constant parameters . after a little manipulation , we obtain for the pion ( @xmath33 ) [ deq:5.1 ] ^p_3;(x ) & = & m^2_0 , + ^_3;(x ) & = & - ^x_0 dx ( * k^2_+ m^2_q ) , where @xmath34[x ( m_m^2 - m^2_{\lambda})]^2}$ ] and @xmath35 . in the standard lfqm , the wave function of a ground state pseudoscalar meson as a @xmath36 bound state is given by @xmath37 , where @xmath38 is the radial wave function and the spin - orbit wave function @xmath39 with the helicity @xmath40 of a quark(antiquark ) that is obtained by the interaction - independent melosh transformation @xcite from the ordinary spin - orbit wave function assigned by the quantum numbers @xmath41 . for the radial wave function @xmath38 , we use both the gaussian or harmonic oscillator ( ho ) wave function @xmath42 and the power - law ( pl ) type wave function @xmath43 @xcite as follows [ qm2 ] _ e^-k^2/2 ^ 2 , _ pl(x,k_)= , where @xmath44 and @xmath45 and @xmath46 is the variational parameter fixed by the analysis of meson mass spectra @xcite . the normalization of @xmath38 is given by @xmath47 in bs model with our lfqm wave function @xmath38 as follows ( see eq . ( 35 ) in @xcite ) [ deq:12 ] . the correspondence in eq . ( [ deq:12 ] ) is valid again in this analysis of a pseudotensor twist-3 da @xmath48 . we now apply this correspondence to both pseudoscalar da @xmath49 in eq . ( [ deq:5.1 ] ) and pseudotensor da @xmath48 in eq . ( 5 ) to obtain them in our lfqm as follows : [ deq:13 ] ^p_3;(x ) & = & m^2_0 , + ^_3;(x ) & = & ^x_0 dx ( k*^2_+ m^2_q ) , respectively . in the numerical computations , we use @xmath50 gev and @xmath51 gev for the linear ( ho ) confining potential model parameters for the gaussian wave function , which were obtained from the calculation of meson mass spectra using the variational principle in our lfqm @xcite . for the pl wave function , we use @xmath52 gev and @xmath53 gev adopted from @xcite . we first compute the second transverse moments @xmath54 for both pseudoscalar ( @xmath14 ) and pseudotensor ( @xmath55 ) channels , and the results obtained from the linear [ ho ] parameters are given by @xmath56 $ ] and @xmath57 $ ] , respectively . the twist-3 das @xmath0 ( left panel ) and @xmath1 ( right panel ) of pion.,title="fig:",width=264,height=264 ] the twist-3 das @xmath0 ( left panel ) and @xmath1 ( right panel ) of pion.,title="fig:",width=264,height=264 ] fig . [ fig1 ] shows the two - particle twist-3 pion @xmath0 ( left panel ) and @xmath1 ( right panel ) obtained from the nonzero constituent quark masses using gaussian wave functions with ho ( solid lines ) model parameters and pl wave functions ( dashed lines ) . we also plot our results in the chiral symmetry @xmath58 limit for both gaussian ( dotted lines ) and pl ( dot - dashed lines ) wave functions and compare them with the chiral - limit prediction of dse approach employing the dynamical chiral symmetry breaking - improved kernels @xcite ( double - dot - dashed line ) as well as the asymptotic result @xmath59 ( circled data ) for the case @xmath1 . our results for both @xmath0 and @xmath1 are normalized without the momentum cutoff ( i.e. @xmath60 ) . for @xmath0 case , our results with nonzero constituent quark masses show rather convex shapes for both gaussian and pl wave functions but they show quite different end point behaviors , i.e. the end points are more enhanced for the pl wave function than the gaussian wave function . the difference between the two wave functions are more drastic in the chiral symmetry limit , where the result of gaussian wave function reproduces the result @xmath61 anticipated from the qcd s conformal limit @xcite but the result of pl wave function shows the concave shape similar to the result of dse approach @xcite , in which the following asymptotic form @xmath62 was obtained . while the authors in @xcite explained that the difference , i.e. @xmath63 term in chiral symmetry limit may come from the mixing effect between the two- and three - particle twist-3 amplitudes , we observe that this difference may come from the cutoff scale of the transverse momentum scale associated with the different choice of lf wave functions . especially , we find that the end point behaviors of @xmath0 also affect the asymptotic form in the chiral symmetry limit . the cutoff dependent behaviors of @xmath0 obtained from both gaussian and pl wave functions are also presented in @xcite , where the concave shape for the gaussian wave function can also be seen with the cutoff scale @xmath64 gev or less being taken but the cutoff dependence was shown to be more sensitive for the pl wave function than the gaussian one . for @xmath1 case , our results with nonzero constituent quark masses for both gaussian ( solid line ) and pl ( dashed line ) show again different end point behaviors , i.e. the end points are more enhanced for the pl wave function than the gaussian wave function . however , in the chiral symmetry limit , gaussian ( dotted line ) and pl ( dot - dashed line ) wave functions show very similar shapes each other . furthermore , the result from gaussian wave function reproduces exactly the asymptotic form @xmath59 . the same chiral - limit behavior was also obtained from the dse approach @xcite . as one can see from fig . [ fig3 ] , the twist-3 pseudoscalar @xmath0 is more sensitive to the shape of the model wave functions ( gaussian vs. pl ) than the twist-3 pseudotensor @xmath1 . it is quite interesting to note in the chiral symmetry limit that while @xmath0 is sensitive to the shapes of model wave functions , @xmath1 is insensitive to them . the twist-3 pseudoscalar da @xmath9 and pseudotensor da @xmath65 are usually expanded in terms of the gegenbauer polynomials @xmath66 and @xmath67 , respectively , as follows @xcite : @xmath68 , and @xmath69 . the coefficients @xmath70 are called the gegenbauer moments , which describe how much the das deviate from the asymptotic one . in addition to the gegenbauer moments , one can also define the expectation value of the longitudinal momentum , so - called @xmath71-moments , as follows : @xmath72 . the gegenbauer- and @xmath71- moments of the pseudoscalar twist-3 das @xmath0 and twist-2 da @xmath73 can be found in our previous works @xcite . @lcccccc models & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 + ho & -0.1155 & -0.0268 & -0.0046 & 0.1604 & 0.0565 & 0.0263 + linear & -0.0803 & -0.0256 & -0.0082 & 0.1725 & 0.0647 & 0.0318 + pl & -0.0375 & -0.0092 & -0.0031 & 0.1871 & 0.0762 & 0.0406 + sr @xcite & 0.0979 & -0.0016 & -0.0011 & 0.2325 & 0.1075 & 0.0624 + dse @xcite & @xmath80 & @xmath80 & @xmath80 & 0.20 & 0.085 & 0.047 + @xmath81qm @xcite & -0.0984 & -0.0192 & -0.0037 & 0.1663 & 0.0612 & -0.0015 + @xmath59 & @xmath80 & @xmath80 & @xmath80 & 0.20 & 0.086 & 0.048 + in table [ t1 ] , we list the calculated gegenbauer- and @xmath71- moments of pseudotensor twist-3 pion da @xmath1 obtained from the gaussian wave functions with linear and ho potential models and pl wave function . we also compare our results with other model predictions , e.g. qcd sum rules ( sr ) @xcite , dse approach @xcite and the chiral quark model ( @xmath81qm ) @xcite . as expected from the isospin symmetry , all odd gegenbauer and @xmath71 moments are all zero . it is interesting to note that the sign of @xmath74 is negative from our lfqm and @xmath81qm predictions but is positive for qcdsr prediction . larger positive value of @xmath74 leads to more flat shape of da but the larger negative value leads to more narrower shape of da . knowing our lfqm results from the ho model are exact to the asymptotic result in the chiral - symmetry limit , i.e. @xmath82_{\rm ho}=[\la\xi^2\ra^{\sigma}_\pi]_{\rm asy}$ ] in @xmath83 limit , one can see that the @xmath71-moments are reduced when the chiral symmetry is broken . we also should note for the same reason that our lfqm results are in good agreement with dse results in the chiral - symmetry limit of @xmath1 . we analyzed the two twist-3 das of pion , i.e. pseudoscalar @xmath0 and pseudotensor @xmath1 , within the lfqm . we also investigated the discrepancy of the asymptotic forms of @xmath0 between dse approach @xcite and qcd s conformal limit expression @xcite from the perspective of dependence of da on the form of lf wave functions , e.g. gaussian vs. pl wave functions . in order to compute the twist-3 pseudotensor da @xmath1 , we utilized the same manifestly covariant bs model used in @xcite and then substituted the lf vertex function in the covariant bs model with the more phenomenologically accessible gaussian and pl wave functions . linking the covariant bs model to the standard lfqm , we used the same correspondence relation eq . ( [ deq:12 ] ) between the two models as the one found in @xcite . the remarkable finding in linking the covariant bs model to the standard lfqm is that the treacherous points such as the zero - mode contributions and the instantaneous ones existed in the covariant bs model become absent in the lfqm with the gaussian or pl wave function . our lfqm descriptions for both twist-3 @xmath2 and @xmath3 satisfy the fundamental constraint(i.e . symmetric form with respect to @xmath16 ) anticipated from the isospin symmetry . for the @xmath0 case , our results with nonzero constituent quark masses show rather convex shapes for both gaussian and pl wave functions but they show quite different end point behaviors , i.e. the end points are more enhanced for the pl wave function than the gaussian wave function . the difference between the two wave functions are more drastic in the chiral symmetry limit , where the result of gaussian wave function reproduces the result @xmath61 anticipated from the qcd s conformal limit @xcite but the result of pl wave function shows the concave shape similar to the result of dse approach @xcite . for the @xmath1 case , our results from both gaussian and pl wave functions in the chiral symmetry limit , show very similar shapes each other . especially , the result from gaussian wave function reproduces exactly the asymptotic form @xmath59 anticipated from qcd s conformal limit . the same chiral - limit behavior was also obtained from the dse approach @xcite . the remarkable thing is that our predictions for the two twist-3 das of @xmath84 and chirality - even twist-2 da @xmath85 and twist-3 da @xmath86 of @xmath87 @xcite obtained from the gaussian wave function in the chiral limit exactly reproduce the forms anticipated from qcd s conformal limit . we summarize in table [ t2 ] the asymptotic forms for the two twist-3 das of @xmath84 and chirality - even twist-2 and twist-3 das of rho meson compared with dse approach @xcite and qcd s conformal limit expression @xcite . lllll model in @xmath83 & @xmath2 & @xmath3 & @xmath85 & @xmath86 + lfqm ( ho model ) @xcite & 1 & @xmath59 & @xmath59 & @xmath88 $ ] + dse approach @xcite & @xmath89 & @xmath59 & @xmath90 & @xmath90 + qcd s conformal limit @xcite & 1 & @xmath59 & @xmath59 & @xmath88 $ ] + the idea of our lfqm is to provide the nonperturbative wave functions at the momentum scale consistent with the use of constituent quark mass . the das determined from this nonperturbative wave functions can be fed into the qcd evolution equation to provide the shorter distance information of the corresponding hadrons . the das obtained without the cutoff should not be regarded as the fully evolved das but still be nonperturbative as they just mean that the cutoff dependence becomes marginal beyond a certain nonperturbative cutoff scale . choi , h .- m . , ji , c .- r . : self - consistent covariant description of vector meson decay constants and chirality - even quark - antiquark distribution amplitudes up to twist 3 in the light - front quark model . . choi , h .- , c .- r . : consistency of the light - front quark model with chiral symmetry in the pseudoscalar meson analysis . . choi , h .- m . , ji , c .- r . : mixing angles and electromagnetic properties of ground state pseudoscalar and vector meson nonets in the light - cone quark model . . choi , h .- m . : decay constants and radiative decays of heavy mesons in light - front quark model . .	we analyzed two twist-3 distribution amplitudes of pion , i.e. pseudoscalar @xmath0 and pseudotensor @xmath1 , within the lfqm . our lfqm descriptions both for twist-3 @xmath2 and @xmath3 not only satisfy the fundamental constraint required from the isospin symmetry , but also reproduce exactly the asymptotic forms anticipated from qcd s conformal limit .
mond @xcite attributes the mass discrepancies in galactic systems not to dark matter but to a departure from the standard dynamics at low accelerations . its basic tenets are : ( i ) galactic systems showing large mass discrepancies are governed by new dynamics that are invariant under space - time scaling @xmath11 . there , newton s @xmath12 is replaced by a scale - invariant constant @xmath13 . ( ii ) the boundary between the standard and the scale - invariant regime is marked by the constant @xmath14 , which is an acceleration . thus , much below @xmath15 ` the deep - mond limit ' dynamics are scale invariant , and much above it standard dynamics are approached . references @xcite are recent reviews of mond . in contradistinction from the dark - matter paradigm , mond decrees , as a law of physics , that baryons determine the full dynamics of a system . so , mond predicts that various ` dynamical ' galaxy properties deduced from the measured accelerations are tightly correlated with ` baryonic ' properties those deduced directly from the distribution of baryonic mass ( see , e.g. , the recent ref . an example of such a ` mond law ' is the mass - asymptotic - speed relation ( massr ) @xcite , which is arguably the most famous prediction of an exact , functional relation . it had predicted a specific version of the ` baryonic tully - fisher relation ' . here , i deal with a new mond law : _ a functional relation _ between a ` baryonic ' and a ` dynamical ' property of pure disc galaxies . the one is the central surface ( baryonic ) density of the disc , @xmath1 , the other is the total , ` dynamically ' measured central surface density , @xmath0 . the impetus to look for such a mond central - surface - densities relation ( csdr ) has come from the recent finding of a correlation between two similar quantities in a large sample of disc galaxies in ref . @xcite . this mond csdr is quite different from other mond laws of galactic dynamics such as the massr , or the discrepancy - acceleration relation and thus broadens the scope of the mond codex ( see sec . [ discussion ] ) . preliminary aspects of the mond csdr have been discussed in ref . @xcite in an approximate way , in terms of the mean density and characteristic size of the galaxy . in particular , it was shown that mond predicts that , for @xmath16 , we have @xmath17 , which implies that @xmath18 . for @xmath10 , mond predicts that @xmath0 scales as @xmath19 , with the coefficient estimated approximately , and being somewhat system dependent . being independent of @xmath1 in this limit . ] these predictions are born out by the findings of ref . @xcite . here , i go rather far beyond these preliminary predictions , and show in sec . [ derivation ] that for pure discs , the mond csdr , @xmath1-@xmath0 relation is functional , and i derive @xmath20 analytically for the full @xmath1 range . the predicted relation is compared with the data of ref . @xcite in sec . section [ discussion ] is a discussion . a true , or ` baryonic ' , mass distribution @xmath21 ( of a galaxy , say ) produces a newtonian acceleration field @xmath22 that is determined from the poisson equation . in modified - gravity formulations of mond , which i shall use here , the acceleration field , @xmath23 , is determined from another equation . one may , on occasion , want to interpret the mond field in newtonian terms and define the ` dynamical ' density of the galaxy as that for which @xmath24 is the newtonian field : _ d-()^-1 . this density is made of the baryonic contribution plus a phantom density that a newtonist would attribute to dark matter . @xmath25 is determined , or constrained , from the observed dynamics of the system ( rotation curves , light bending , velocity dispersions , etc . ) consider , as a special case , an axisymmetric , thin , disc galaxy , reflection - symmetric about its ( @xmath26 ) midplane . the dynamical , face - on , central surface density is _ -^^0_d(z)dz , where we work in cylindrical coordinates with the @xmath27-axis along the axisymmetry axis , and @xmath28 . the assumed symmetries will enter through their implications that along the @xmath27 axis the newtonian and mond accelerations have only a @xmath27 component , and that they are each equal in magnitude and opposite in sign at @xmath27 and @xmath29 . i only use , directly , the assumption that the system is ` thin ' along the @xmath27 axis ; i.e. , all the matter along this axis is concentrated in a thin disc at the origin , where the baryonic surface - density is @xmath1 . this still holds for a disc that flares , or is not otherwise thin , in the outer parts . but overall flatness may also enter some aspects indirectly ( see below ) . the baryonic @xmath1 is related to @xmath30 , the absolute value of the newtonian acceleration just outside the disc , at the origin : @xmath31 . because we are dealing with a thin disc , the expression for @xmath0 in eq . ( [ ii ] ) has two contributions : one from the disc ( baryonic plus phantom ) which is given by @xmath32 , where @xmath33 is the absolute value of the mond acceleration just outside the disc , at the origin . the reflection symmetry is invoked here . the other contribution is ( out)=2_0^+^^0_d dz=-_0^+^ dz , where the integration is from just outside the disc . in qumond ( quasilinear mond ) @xcite , @xmath34 is determined from the field equation = , where the interpolating function @xmath6 has the limits @xmath35 , which follows from scale invariance , and @xmath36 for correspondence with newtonian dynamics . applying gauss s theorem to get the jump condition at the origin , we have @xmath37 . thus ( disc)=y_0(y_0 ) , where @xmath38 , and @xmath39 the mond surface density @xcite : = 138(/1.210 ^ -8)^-2 . to calculate @xmath40 , note that along the @xmath27-axis ( outside the disc , where @xmath41 ) we have from eq.([qumon ] ) , @xmath42/dz$ ] , where @xmath43 . i assumed that along the @xmath27-axis @xmath44 always points to the origin . this is obvious for a thin disc . it is true for a much larger class of galaxies , but not for any ` thick disc ' . substituting in eq.([iii ] ) , changing variables to @xmath45 , and integrating by parts , one gets ( out)=. the first term cancels the disc contribution from eq.([iv ] ) , and we are left with the desired result = ( / ) ; ( y)_0^y(y ) dy . this is a surprising result , because the dynamics is not local . the value of @xmath25 at a given point , and its distribution along the @xmath27-axis , depend on the full mass distribution of the galaxy . yet , the integral that is @xmath0 turns out to depend only on @xmath1 ! the high-@xmath1 asymptote ( newtonian limit ) is gotten by taking @xmath46 , which gives @xmath47 . the opposite asymptote ( the deep mond limit ) is gotten from the scale - invariant limit @xmath48 , and gives : @xmath49 . in this limit , @xmath50 and @xmath40 contribute equally , @xmath19 each . the two asymptotes meet at @xmath51 . since everywhere @xmath52 , the mond value of @xmath0 is always above the two asymptotes . for example , for the limiting form : @xmath53 , we have from eq.([vi ] ) @xmath54 for @xmath55 , and @xmath56 for @xmath57 . for @xmath58/2 $ ] , which is widely used in mond rotation curve analysis , eq.([vi ] ) gives ( y)=y/2+y^1/2(1+y/4)^1/2 + 2sinh^-1(y^1/2/2 ) . in the nonlinear poisson formulation of mond @xcite , the mond acceleration field is determined by the field equation = -= , with @xmath59 , @xmath60 . defining @xmath61 , @xmath45 has to be a monotonic function of @xmath62 @xcite ; so the relation can be inverted . define then @xmath6 , such that @xmath63 . the interpolating function @xmath6 is the equivalent of the above @xmath6 in qumond , and has the same large- and small-@xmath45 limits . the relation between @xmath33 and @xmath30 is the same as for qumond . thus , eq.([iv ] ) for @xmath50 holds here as well . to calculate the integrand in eq.([iii ] ) for @xmath40 , note that along the @xmath27-axis , and outside the disc , we have from the field equation ( [ poissona ] ) ( @xmath64 ) ( x)-g d / dz=0 , where @xmath65 . i assumed that along the @xmath27-axis , @xmath24 always points to the origin ( so @xmath66 for @xmath67 ) . this can be proven @xcite for a thin disc , but is also true for a much larger class , but not for any ` thick disc ' . differentiating the relation @xmath61 with respect to @xmath45 , we have : @xmath68 . noting also that @xmath69 , extracting from these the expression for @xmath70 along the @xmath27 axis , substituting in the integrand , and integrating by parts , one finds that expression ( [ vi ] ) for @xmath0 holds here as well . as a result , all the corollaries listed for qumond carry to the nonlinear poisson theory . what can be said about the general case of an arbitrary mond theory , arbitrary dynamical and baryonic surface densities , and any galactic system ? confine ourselves first to a two - parameter family of self - similar systems such as all exponential discs , with constant thickness - to - size ratio with a baryonic density distribution @xmath71 , with the same @xmath72 . for this family , any definition of a baryonic surface density local , such as our @xmath1 , or global , such as the total mass over some area is given by @xmath73 up to a dimensionless constant . if @xmath74 is some ` dynamical ' surface density such as our @xmath0 , or @xmath75 defined below in sec . [ data ] then in any nonrelativistic mond theory , we have on dimensional grounds , @xmath76 , where @xmath77 is a dimensionless function , constructed from dimensionless combinations of its variables . the only such variable is readily seen to be @xmath78 ( or some function of it ) . we thus have for the family @xmath79 . in the newtonian limit , where @xmath15 is not available , we must have @xmath80 ; the dimensionless constant @xmath81 depends on the family and choice of surface densities ( but not the exact theory ) . this is just the newtonian result . in the opposite , deep - mond limit , scale invariance dictates that @xmath82 , or @xmath83 . this is because under space - time scaling , a dynamical surface density , which is derived as @xmath84 , scales as @xmath85 , while @xmath86 , defined as @xmath87 , scales as @xmath88 ( and the two sides of any dml relation have to scale in the same way ) . the dimensionless constant @xmath89 may depend on the family , the choice of surface densities , and the exact mond theory . but , since mond is assumed not to involve large dimensionless constants , @xmath81 and @xmath89 should be of order unity , if @xmath74 and @xmath86 are bulk properties ( and , thus , do not introduce small dimensionless quantities in themselves ) . thus for all system families taken together , any mond theory would predict a strong @xmath90 correlation , with the two universal asymptotes . the surprising and helpful aspect of our analysis here is that the mond csdr for pure discs is universal , and can be derived in the two full - fledged mond theories practiced at present ( they give @xmath91 , @xmath92 ) . figure [ fig1 ] shows the data points of ref . for the data , our @xmath1 is represented by a proxy : the _ stellar _ central surface mass density , @xmath93 , converted from the central surface brightness using a universal mass - to - light ratio in the 3.6 micron photometric band , @xmath94 solar units . as a proxy for our @xmath0 , ref . @xcite used an expression by toomre @xcite : @xmath95 , shown in ref . @xcite to give @xmath0 for thin - disc , flat galaxies [ @xmath96 is the rotation curve].^{1/n}$ ] . they are sensitive to different regions of the rotation curve for different @xmath97 . @xmath98 . ] reference @xcite discusses at length why @xmath93 , and @xmath99 while not quite the same as @xmath1 and @xmath0 are good proxies . also shown in fig . [ fig1 ] is the best fit that ref . @xcite gives for some 3-parameter fit function , not motivated by theory . the mond predictions ( for @xmath100 ) are plotted in fig . [ fig1 ] as the @xmath1-@xmath0 plane . shown are the asymptotes and the full relation for the widely used choice @xmath6 that corresponds to @xmath101 , with @xmath4 from eq . ( [ eta ] ) . and they all give comparable results , hardly distinguishable within the data spread . they all nearly coincide with the deep - mond asymptote up to @xmath39 , and with the ` newtonian ' asymptote above @xmath102 ; in between , the prediction lies somewhat above the two asymptotes . ] the @xmath1-@xmath0 relation . the blue ( thicker ) line ( full and dashed ) is the equality line ( the newtonian asymptote of the mond prediction ) . the thinner , red line ( full and dashed ) is the predicted , deep - mond asymptote . the thinnest , black line is the full mond relation , eq.([eta ] ) , for the commonly used @xmath101 . the data points are from ref . for the data , the toomre surface density , @xmath75 , is taken as a proxy for @xmath0 , and the proxy for @xmath1 is the central , stellar surface density , @xmath103 . the dotted line is the best - fit to the data in ref . @xcite , with some 3-parameter formula ( not theoretically motivated ) . no fitting is involved in the mond curves . the values of the mond surface density , @xmath39 is marked . including the contribution of gas to @xmath1 would move some of the low-@xmath1 data points somewhat to the right , and correcting for the fact that @xmath104 would move them slightly upward ( see comments in subsec . [ commdata ] ) . ] it is seen that the mond csdr , which involves no free parameters , agrees very well with the proxy data : the normalization , the asymptotic slopes , the position of the break , and the general behavior . @xmath103 does not include the contribution from the gas , which is said to be small in most cases . however , some low-@xmath1 galaxies are exceptions , and it would have been instructive to include the gas contribution , so as to get the true , baryonic , central surface density . departures of @xmath105 from the universal value used by ref . @xcite will contribute to scatter around the predicted relation . and , possible small systematic variations of @xmath105 , for example with @xmath1 , could lead to a small change in the slope . even for a pure baryonic disc , @xmath75 , which is exact for thin discs , somewhat underestimates mond s @xmath0 , because mond s @xmath25 has a ` thick ' phantom component [ from which @xmath40 is made ] . for example , it is readily seen that for a spherical system , @xmath106 . in the high-@xmath1 regime , this should not matter , as the ` phantom ' component is negligible . in the deep - mond , low-@xmath1 regime , we saw that the ` phantom disc ' and ` phantom halo ' contribute equally to @xmath0 , according to mond . the disc component is accounted for correctly by the corresponding contribution to @xmath75 . for the ` phantom halo ' component , @xmath75 is between @xmath107 and 1 of the contribution to @xmath0 . in all , we thus expect @xmath75 to underestimate @xmath0 by a factor between @xmath108 and @xmath109 not much of a correction compared with the data spread . as an example , i calculated the relevant quantities for a deep - mond , thin , kuzmin disc , for which the mond rotation curve is known analytically : @xmath110 , with @xmath111 , and @xmath112 the kuzmin scale length @xcite . i found that @xmath113 , indeed between 0.75 and 1 . the mond csdr differs from other mond laws ( discussed , e.g. , in ref . @xcite ) in several important regards : a. it is a relation between a ` local ' baryonic attribute , @xmath1 , defined at the center , and a ` global ' dynamical one : the column dynamical density along the symmetry axis . b. it encompasses the full gamut of accelerations . in comparison , the massr relates a global baryonic attribute the total mass with the asymptotic rotational speeds , and it involves only the deep - mond regime , and only the outskirts of systems . the mond ` mass - discrepancy - acceleration relation ' for disc galaxies , relates the baryonic and dynamical accelerations at the same radius , in the plane of disc galaxies . and , the mond mass - velocity - dispersion relation ( an analog of the virial theorem ) relates two global attributes , and holds in the low - acceleration regime . with good enough data , one can , in principle determine the mond interpolating function by differentiating the observed @xmath4 , from eq . ( [ vi ] ) . this can also be done in an independent way from the discrepancy - acceleration relation . but this will have to await better data to constrain @xmath4 . the mond csdr falls very near the 3-parameter best - fit of ref . the scatter around this best fit was shown in ref . @xcite to be @xmath114 dex , consistent with observational and procedural errors , i.e. , with no intrinsic scatter . furthermore , the residuals do not correlate with various galaxy properties , such as size , gas fraction , or stellar mass . this is exactly what mond predicts , while , as also stressed in ref . @xcite , in the dark - matter paradigm , there is no reason why @xmath0 , which would stand for the central column density of baryon plus dark matter , should be so well correlated with the local @xmath1 . especially , the correlation in the low @xmath1 region where the discrepancy between @xmath0 and @xmath1 is large goes quite against the grain of the cold - dark - matter paradigm : even with schemes , such as ` feedback ' , ` abundance matching ' , and the like put in by hand to save this paradigm from various embarrassments one would expect large scatter in any relation between the ` dynamical ' and baryonic properties , which , to boot , one would expect to depend on galaxy properties ( see , e.g. , the relevant discussion in ref . @xcite ) . m. milgrom , astrophys . j. 270 , 365 ( 1983 ) . m. milgrom , astrophys . j. 270 , 371 ( 1983 ) . b. famaey & s. mcgaugh , liv . 15 , 10 ( 2012 ) . m. milgrom , scholarpedia , 9(6 ) , 31410 ( 2014 ) . m. milgrom , mnras 437 , 2531 ( 2014 ) . f. lelli , s.s . mcgaugh , j.m . schombert , & m.s . pawlowski , astrophys . 827 , l19 ( 2016 ) . m. milgrom , mnras 398 , 1023 ( 2009 ) . f. donato , & al . , mnras 397 , 1169 ( 2009 ) . m. milgrom , mnras 403 , 886 ( 2010 ) . j. bekenstein & m. milgrom , astrophys . j. 286 , 7 ( 1984 ) . m. milgrom , j. phys . a 35 , 1437 ( 2002 ) . a. toomre , astrophys . j. 138 , 385 ( 1963 ) . r. brada & m. milgrom , mnras 276 , 453 ( 1995 ) . k.a . oman , & al . , mnras 452 , 3650 ( 2015 ) .	i derive a new mond relation for pure - disc galaxies : the ` dynamical ' central surface density , @xmath0 , deduced from the measured velocities , is a _ universal function _ of only the true , ` baryonic ' central surface density , @xmath1 : @xmath2 , where @xmath3 is the mond surface density constant . this surprising result is shown to hold in both existing , nonrelativistic mond theories . @xmath4 is derived : @xmath5 , with @xmath6 the interpolating function of the theory . the relation aymptotes to @xmath7 for @xmath8 , and to @xmath9 for @xmath10 . this study was prompted by the recent finding of a correlation between related attributes of disc galaxies by lelli et al . ( 2016 ) . the mond central - surface - densities relation agrees very well with these results .
an interesting topic in recent years both in the general relativity community and in the string theory community is that the bh area ( or entropy ) product formula of all horizons independent of the adm ( arnowitt - deser - misner ) mass of the background space - time @xcite . for example , the area product formula for a kerr bh @xcite depends only on the angular momentum parameter : @xmath6 where @xmath7 and @xmath8 are area of the inner and outer horizons . whenever , we have taken the perturbed space - time with a spinning bh in some non - trivial environment e.g. a bh surrounded by a ring of matter or a multiple bh space - time the same formula holds . hence , the area product formula of outer horizon or event horizon ( @xmath9 ) and inner horizon or cauchy horizons ( @xmath10 ) for kerr bh is of an _ universal _ quantity : it holds independently of the environment of the bh . on the other hand , if we incorporate the bps ( bogomolni - prasad - sommerfeld ) states , the area product formula of @xmath11 should read @xcite @xmath12 where @xmath13 is the planck length . this indicates the area product should be quantized . alternatively , the area products independent of mass implies that there should be an important role of the cauchy horizon in the bh thermodynamics as well as in bh physics . now the relevant question is that the mass independent product formula is generic ? it has been shown explicitly by visser @xcitethat by incorporating the cosmological constant , the area product of all physical horizons is not mass independent . but typically , some complicated function of inner and outer horizon area is indeed mass independent . previous studies have not made use of the extended phase space formalism . thus in this work , we wish to examine the thermodynamic product formula in _ extended phase space_. where the adm mass of an ads bh could be treated as the enthalpy of the space - time and the cosmological constant should be treated as the thermodynamic pressure @xcite . therefore there must exists a conjugate quantity which is a thermodynamic volume associated with the bh space - time . besides area ( or entropy ) products , it needs to be evaluated whether other thermodynamic products @xcite like bh temperature products , specific heat products , komar energy products etc . are provide any universal characterization or not , and here we first introduced the _ thermodynamic volume products _ when one must considered the extended phase space thermodynamics . does it independent of the adm mass parameter ? we will investigate this issues in the present work . so , when the cosmological constant treated as a thermodynamic pressure and its conjugate variable as a thermodynamic volume what happens the _ smarr mass formula _ , _ smarr - gibbs - duhem _ relation and bh _ equations of state _ in the extended phase space . additionally , we find the mass independent volume products relation in parallel with the entropy product relations . bh thermodynamic properties have been investigated for many decades and still it is going on . in the present study , the main motivation comes from the seminal work of hawking and page @xcite where the thermodynamic properties of bhs in schwarzschild - ads space has been explicitly studied . the author discussed the phase transition ( between small and large bhs for schwarzschild - ads bh ) which is called famous hawking - page phase transition . the special interest is due to the application of ads space - time in gauge / gravity duality via dual conformal field theory ( cft ) through ads / cft correspondence @xcite . several exotic phenomena occurs in the ads space - time . first example of course be hawking - page phase transition in schwarzschild - ads spacetime . the second one is that in charged ads spacetime , the gravitational analogue of the liquid / gas phase transition has been observed in the phase diagram which was explicitly investigated by several authors @xcite and the fact that for charged ads bh the notion of thermodynamic equilibrium is a straightforward concept . the third one is that kerr - ads spacetime admits reentrant phase transition and showing a tri - critical point in their phase diagram @xcite . the current interest is involved due to the variation of negative cosmological constant and also it is proportional to the thermodynamic pressure @xcite . the thermodynamic products especially area ( or entropy ) products in charged ads bhs were calculated in @xcite but the author has not been considered there the extended phase space . here we shall compute the volume products by considering the thermodynamic pressure ( @xmath14 ) is equal to the negative cosmological constant ( @xmath15 ) divided by @xmath16 ( where @xmath17=1 ) i.e. @xmath18 and the corresponding thermodynamic volume can be defined as @xmath19 this volume for charged - ads bh should read as @xmath20 where @xmath21 is the corresponding horizon radius and @xmath22 . it has been shown that the _ reverse isoperimetric inequality _ is satisfied for event horizon @xcite . here we conjecture that this inequality is valid for all the horizons i.e. @xmath23 it should be noted that a class of bhs with non - compact event horizons do not satisfy this inequality @xcite . the structure of the paper is as follows . in sec.([tnt ] ) , we have described the thermodynamic properties of rn - ads bh . the sec.([quint ] ) describes the thermodynamic properties of the rn - ads bh surrounded by quintessence . in sec . ( [ fr ] ) , we have given the thermodynamic properties of the @xmath0 gravity . finally , we conclude in sec.([dis ] ) . in appendix - a , we have examined the thermodynamic volume products for axisymmetric space - time and in appendix - b , we have investigated the @xmath1 criticality of inner horizon for rn - ads bh . let us begin with the charged - ads space - time metric can be written as in terms of schwarzschild like coordinates@xcite : @xmath24 where , @xmath25 and @xmath26 is the metric on the unit sphere in two dimensions . the electromagnetic potential one form for the space - time ( [ metric ] ) is @xmath27 the bh horizons determined by the condition @xmath28 i.e. @xmath29 in terms of thermodynamic pressure , this could be rewritten as @xmath30 to finding the roots we apply the vieta s rule , we get @xmath31 the entropy of the bh can be defined as @xmath32 where the area of the bh is @xmath33 and now the bh temperature reads as @xmath34 the electric potential could be defined as @xmath35 we should be noted that in the extended phase space the adm mass can be treated as the total gravitational enthalpy of the system i.e. @xmath36 . where @xmath37 is the thermal energy of the system@xcite . then the first law of bh thermodynamics in the extended phase space becomes @xmath38 and the corresponding smarr - gibbs - duhem relation becomes @xmath39 now we compute the mass(or enthalpy ) independent entropy sum and entropy product formula using eqs.([eq1],[eq2],[eq4],[eq5],[eq6 ] ) : @xmath40 in terms of two horizons , the mass - independent entropy product formula should read @xmath41 although it is a complicated function of two horizons but it is explicitly mass independent function of inner horizon and outer horizons . now we turn into another important relations that is the _ volume sum _ and _ volume product _ which are mass independent : @xmath42 again in terms of two horizons , the mass independent volume product formula should be @xmath43 & = & \frac{3}{8\pi p } ~. \label{eqq2}\end{aligned}\ ] ] these are explicitly mass - independent relations in the extended phase space . finally , the equation of state in the extended phase space should read @xmath44 where @xmath45 . now in terms of specific volume @xmath46 the above equation could be re - written as @xmath47 the critical point can be obtained from the following conditions : @xmath48 the critical values are explicitly computed in @xcite . defining further @xmath49 , @xmath50 and @xmath51 , the law of corresponding states become @xmath52 . , @xmath53 and @xmath54 in the extended phase space the gibbs free energy could be defined as @xmath55 and the helmholtz free energy is given by @xmath56 which is very important to determine the behavior of the critical exponents . , @xmath57 , @xmath58 and @xmath59 . the numerical values for charged - ads bh are calculated in@xcite as @xmath60 , @xmath61 , @xmath62 and @xmath63 . ] one may compute the entropy via the relation : @xmath64 which is exactly same as in eq . ( [ eq5 ] ) . there are two types of specific heat . the specific heat at constant thermodynamic volume and the specific heat at constant pressure . they are defined as @xmath65 and @xmath66 from eq . [ eq23 ] , we can easily see that the entropy @xmath67 is independent of @xmath68 therefore we get @xmath69 and we find @xmath70 the specific heat @xmath71 diverges at @xmath72 or i.e. at @xmath73 which implies a second order phase transition occurs at that point . in this section , we will show how the quintessence dark energy matter does affect on the thermodynamic product relation in the extended phase space . the metric function of eq . ( [ metric ] ) for rn - ads bh surrounded by quintessence can be written as @xcite @xmath74 where @xmath75 is the state parameter and @xmath76 is the normalization factor related to the density of quintessence . the ranges for quintessence dark energy is @xmath77 and for phantom dark energy : @xmath78 . in terms of @xmath76 , the density of quintessence can be defined as @xmath79 in the extended phase space the function can be written as @xmath80 now the horizon eq . can be written as @xmath81 using vieta s theorem , we find @xmath82 it should be mentioned that @xmath83 is an integer quantity . the entropy @xmath67 and electric potential @xmath84 are same as in rn - ads case . now the mass of the bh could be expressed in terms of the horizon radius and dynamic pressure : @xmath85 hence the first law of thermodynamics becomes @xmath86 where @xmath87 is defined to be a physical quantity conjugate to the state parameter @xcite . the corresponding smarr relation reads @xmath88 again the mass(or enthalpy ) independent entropy sum and entropy product relations are @xmath89 similarly , the mass independent volume sum and volume product relations are @xmath90 these are explicitly mass - independent relations for rn - ads bh surrounded by quintessence . it follows from the above analysis that the entropy product and volume product relations are strictly dependent on _ quintessence dark energy matter_. it is quite interesting to mentioned that the entropy product is mass - independent but there has been effect of quintessence dark energy matter on that thermodynamic product relations . now the bh temperature for all the horizons could be defined as @xmath91 and the bh equation of state should read @xmath92 where @xmath45 . again in terms of specific volume @xmath46 the above equation should be rewritten as @xmath93 the critical values are explicitly calculated in @xcite . so we do not written here . the gibbs free energy for all the horizons could be written as @xmath94 again we compute the specific heat at constant thermodynamic pressure to study the local stability of the bh given by @xmath95 it should be noted that the specific heat diverges at @xmath96 which signals a second order phase transition . this section is dedicated to study the thermodynamic properties of a static , spherically symmetric ads bh in @xmath0 gravity . it is a kind of modified gravity and it is very useful tool for explaining the current and future state of the accelerating universe . it is also helpful for explaining the inflation and structure formation in the early universe . the metric @xcite function for this kind of gravity can be written as @xmath97 where @xmath98 . the quantities @xmath99 and @xmath100 are related to the @xmath101(adm mass ) and @xmath102(electric charge ) in this gravity becomes @xmath103 as is the thermodynamic pressure in @xmath0 gravity can be written as @xmath104 and the constant scalar curvature as @xmath105 . the corresponding thermodynamic volume can be defined as @xmath106 . therefore the horizon equation should read @xmath107 to finding the roots we again apply the vieta s rule , we have @xmath108 the entropy can be defined as @xmath109 and the bh temperature@xcite should be @xmath110 again the electric potential in @xmath0 gravity could be defined as @xmath111 now the mass(or enthalpy ) independent entropy sum and entropy product formula in @xmath0 gravity should read : @xmath112 in terms of two horizons , the mass - independent entropy product formula reads as @xmath113 again the mass independent _ volume sum _ and _ volume product _ becomes @xmath114 again in terms of two horizons , the mass independent volume product formula reads @xmath115 & = & \frac{3}{8\pi p } ~. \label{eqq4}\end{aligned}\ ] ] once again these are explicitly mass - independent relation in the extended phase space in @xmath0 gravity . it should be noted that in the limit @xmath116 , one obtains the result of rn - ads bh in extended phase space . for our record we should be noted that the equation of state in @xmath0 gravity@xcite : @xmath117 in terms of specific volume @xmath46 , the above eq . could be rewritten as @xmath118 from the equation of state we can easily derived the critical constants by applying the appropriate condition .. gravity explicitly computed in @xcite are @xmath119 , @xmath120 and @xmath121 finally , the gibbs free energy @xcite should read @xmath122 in this case , the specific heat at constant pressure is found to be @xmath123 the specific heat diverges at @xmath124 or i.e. at @xmath125 again it signals a second order phase transition . the present study demonstrated that the thermodynamic properties of spherically symmetric charged - ads black hole , charged ads bh surrounded by quintessence and charged ads bh in @xmath0 gravity in the extended phase - space . the extended phase space means where the cosmological constant should be treated as thermodynamic pressure and its conjugate variable as a thermodynamic volume . we derived various thermodynamic products particularly entropy products and thernodynamic volume products . in all the three cases , it has been shown that the mass(or enthalpy ) independent properties turn out to be an universal like quantities . it should be noted that the presence of the quintessence matter does affect on the expression of entropy product and thermodynamic volume products . the first law of bh thermodynamics and smarr formula have been studied for all the horizons . the bh equation of state has been derived for all the horizons . the divergence of the specific heat indicates that the second order phase transition should occur at a certain condition . in summary , the thermodynamic relations that we derived provide some universal characterization of the bh which _ may _ provide insight into the origin of bh entropy both _ inner and outer_. [ [ section ] ] in the main work , we have considered the thermodynamic volume for different spherically symmetric cases where it is related to the entropy via a proportionality of the type @xmath126 and we proved that for each cases , the thermodynamic product relations are _ independent of mass _ thus the relations are _ universal _ in nature . now here we shall show what happens in case of axisymmetric cases where the thermodynamic volume is _ not _ directly proportional to the above mentioned relation ? for example , we have taken the kerr bh . the thermodynamic quantities for all the horizons @xmath127 @xcite are @xmath128 and the thermodynamic volume @xcite for @xmath9 is @xmath129 ~.\label{ap1}\end{aligned}\ ] ] and we claim that the thermodynamic volume for @xmath10 is defined to be @xmath130 ~.\label{vi}\end{aligned}\ ] ] we find an interesting relation between volume products and area products for an axisymmetric spacetime having two physical horizons namely @xmath9 and @xmath10 : @xmath131 the main potential point of interest here is that the product of thermodynamic volume of @xmath127 and it is found to be for kerr bh : @xmath132 it indicates that the thermodynamic volume product does _ depend _ on the mass parameter and the universality that we have found in spherically symmetric cases breaks down for axisymmetric cases . thus the conclusion is that although the area ( or entropy product ) of @xmath127 is universal for simple kerr bh but the volume product is _ not universal_. the result should be valid for kn bh as well and the volume product is found to be @xmath133 therefore , it can be easily extend to kerr - ads bh and kn - ads bh also . [ [ section-1 ] ] in this section , we shortly introduce @xmath1 criticality of cauchy horizon for charged ads bh . does the inner horizon obey an equation of state ? what happens to the inner horizons during phase transition has been discussed in @xcite . here we are interested to show what happens the bh equation of state in case of cauchy horizon ? what are the values of critical constant for this horizon does it same as is for event horizon . this is the main interest here . since we have taken charged ads space - time and we have assumed that the bh should have at least two physical horizons . the outer horizon @xmath134 and inner horizon @xmath135 . then we find the relevant thermodynamic quantities for @xmath9 @xcite : @xmath136 @xmath137 @xmath138 @xmath139 the critical point can be obtained at the point of inflection of @xmath9 : @xmath140 and the critical values explicitly derived in @xcite are @xmath141 now we have derived the thermodynamic quantities for @xmath10 : @xmath142 @xmath143 @xmath144 @xmath145 similarly , we can compute the critical values for @xmath10 by applying the condition at the point of inflection : @xmath146 and the critical values we find @xmath147 the only difference in critical values between two horizons is @xmath148 the critical temperature for @xmath10 is negative but other values are same . it is also true that the @xmath1 diagram is qualitatively different for @xmath9 and @xmath10 . the critical ratio for @xmath10 is calculated to be @xmath149 where as for @xmath9 , @xmath150 . thus , we conclude that @xmath151 finally , using the properties of symmetry in nature of @xmath152 , one obtains the following thermodynamic quantities at @xmath10 : @xmath153 i would like to thank the editor prof . j. p. s. lemos and the anonymous referee for their useful suggestions for which the manuscript is enhanced substantially . i am specially thanks to the referee for some interesting comments .	we have examined the thermodynamic properties for a variety of spherically symmetric charged - ads black hole ( bh ) solutions , including the charged ads bh surrounded by quintessence dark energy and charged ads bh in @xmath0 gravity in _ extended phase - space_. this framework involves treating the cosmological constant as thermodynamic variable ( for example : thermodynamic pressure and thermodynamic volume ) . then they should behave as an analog of van der waal ( vdw ) like systems . in the extended phase space we have calculated the _ entropy product _ and _ thermodynamic volume product _ of all horizons . the mass ( or enthalpy ) independent nature of the said product signals they are _ universal _ quantities . the divergence of the specific heat indicates that the second order phase transition occurs under certain condition . in the appendix - a , we have studied the thermodynamic volume products for axisymmetric spacetime and it is shown to be _ not universal _ in nature . finally , in appendix - b , we have studied the @xmath1 criticality of cauchy horizon for charged - ads bh and found to be an universal relation of critical values between two horizons as @xmath2 , @xmath3 , @xmath4 , @xmath5 . the symbols are defined in the main work .
optical lattices ( ol s ) offer the ultimate control of atoms trapped by them . this leads to the formation of correlated phases of matter @xcite , this being useful for quantum simulation purposes @xcite and quantum information processing ( qip ) applications . the degree of precision achieved with them so far , has allowed to achieved self - consistent light - matter states in a bose - einstein condensate ( bec ) inside an optical cavity @xcite . using the dynamical properties of the light @xcite the structural dicke phase transition was achieved forming a state with supersolid features @xcite . however , the study of the full quantum regime of the system has been limited to few atoms @xcite . as the light matter coupling is strongly enhanced in a high finesse optical cavity in a preferred wavelength , the atoms re - emit light comparable with the lasers used in the trapping process . as a consequence , an effective long - range ( nonlocal ) interaction emerges driven by the cavity field . it is now experimentally possible to access the regime where light - matter coupling is strong enough and the cavity parameters allow to study the formation of quantum many - body phases with cavity decay rates of mhz @xcite and khz @xcite . the light inside the cavity can be used to control the formation of many - body phases of matter even in a single cavity mode @xcite . this leads to several effects yet to be observed due to the dynamical properties of light @xcite . moreover , it has been shown that multimode atomic density patterns can emerge , even their coherences can become structured and light - matter quantum correlations can control the formation of correlated phases . thus , a plethora of novel quantum phases due to the imprinting of structure by design in the effective light - induced interaction occurs @xcite . in addition to light - scattering @xcite , homogenous quantum many - body phases can be measured by matter wave scattering @xcite and dynamical structure factors can be obtained via homodyne detection @xcite . recently , density ordering has been achieved with classical atoms @xcite . further , multimode cavities extend the range of quantum phases even further @xcite . therefore , by carefully tuning system parameters and the spatial structure of light , one can design with plenty of freedom the quantum many - body phases that emerge . the quantum nature of the potential seen by the atoms changes the landscape of correlated quantum many - body phases beyond classical optical lattice setups . very recently an optical lattice in an optical cavity has been realised @xcite and self - oganized mott - insultator phases have been achieved @xcite moreover , the interplay between short range processes , such as on - site interactions and tunneling , and long - range cavity induced interactions can change significantly the properties of the light in the system . as these processes compete to optimise the energy in the system , the back - action of the matter affects the light generating nonclassical features @xcite . we show how such nonclassical effects of the light inside the cavity arise due to the emergence of structured quantum phases of matter . this can be traced back to the particular structure of the full light - matter state , which we construct beyond the limit where the light can be integrated out ( adiabatically eliminated ) @xcite . the formulation of the explicit form of the light - matter state of the system and deriving the effective matter hamiltonian incorporating the effect of light at the quantum level is a difficult problem . we provide an alternative to those methods that allows for the construction of the effective hamiltonian , where the effect of local processes ( regular atomic tunneling and on site interaction ) is both considered in the properties of the light . the series of terms that arise , compose different hierarchies of light - induced interactions in addition to the adiabatic limit . this leads to a new effective hamiltonian where the effect of local processes , such as tunneling and on - site interactions , and the global structure imprinted by the light is relevant . we use the technique of canonical transformations constructing the set of unitary operators to remove the non - diagonal terms due to the light @xcite . the underlying symmetries broken by design by pumping light into the system modify the structure of both matter and light and the competition between global and local processes are the origin of nonclassical features . we find the effective matter hamiltonian and that the full light - matter state is a superposition of squeezed coherent states . these depend on the emergent quantum many - body phases of matter the system supports and their structural properties . we demonstrate how the quantum ( quadratures ) and classical ( amplitude ) properties of the light encode information about the strongly correlated phases of matter . as a corollary of our results , we find the conditions to optimise quadrature light squeezing in the system and the effect of the structure induced to the matter . thus , our work will foster the design of this kind of states and their possible application towards quantum multimode systems in the analogous interdisciplinary field of optomechanics @xcite . towards possible applications , there is an active interest in achieving large light squeezing in optomechanical systems where relevant achievements have already been made @xcite . recently , using trapped ions @xcite superpositions of squeezed states have been achieved as proposed by @xcite . additionally , stationary entanglement of photons and atoms in a cavity has been studied @xcite , seeding patters via the cavity field @xcite and quantum control projection @xcite , opening the venue for applications on qip . beyond the quantum properties of light and matter , we find the effective master equation that describes the evolution of the system . this enables the possibility to study the effect of measurement back - action and its direct interplay with local processes . additionally , this can be used for state preparation using state projection via measurement back - action @xcite , while engineering of non - trivial correlated quantum states is possible @xcite , and opens the possibility to optimise nonclassical properties of light . the system consists of atoms trapped in an ol inside single - mode cavity with the mode frequency @xmath0 and decay rate @xmath1 in off - resonant scattering . the pump light has amplitude @xmath2 ( in units of the rabi frequency ) and frequency @xmath3 ( @xmath4 ) . the system is illuminated in a plane transverse to the cavity axis ( not necessarily at @xmath5 ) . the cavity mode couples with the atoms via the effective coupling strength @xmath6 , with @xmath7 the light - matter coupling coefficient and @xmath8 is the detuning between the light and atomic resonance @xcite . this can be described by the hamiltonian @xmath9 , where @xmath10 is the regular bose - hubbard ( bh ) hamiltonian @xcite , @xmath11 with @xmath12 the nearest neighbour tunneling amplitude , @xmath13 the on - site interaction and @xmath14 the chemical potential . the light is described by @xmath15 and the light - atom interaction is @xcite : @xmath16 with @xmath17 . @xmath18 is the density coupling of light to the atoms , @xmath19 is due to the inter - site densities reflecting matter - field interference , or bonds @xcite . the sums go over illuminated sites @xmath20 , and nearest neighbour pairs @xmath21 . the operators @xmath22 ( @xmath23 ) create ( annihilate ) bosonic atoms at site @xmath24 , @xmath25 ( @xmath26 ) photons in the cavity , while the number operator of atoms per site is given by @xmath27 . @xmath28 is the relevant contribution to the quantum potential seen by atoms on top of classical ol described by the bh model , where the on - site interaction @xmath13 and hopping amplitude @xmath12 are short - range local processes . the effective parameters of the bose - hubbard hamiltonian with the cavity field can be calculated from the wannier functions and are given by @xmath29 where @xmath30 are the cavity and pump mode functions and @xmath31 are the wannier functions . the classical optical lattice potential is given by @xmath32 . the classical optical lattice defining the regular bose - hubbard hamiltonian is weakly dependent of the cavity parameters . the atoms are mainly trapped by the strong classical lattice , which is created inside a cavity by external laser beams . this external potential is insensitive to the quantum state of atoms . the light scattered into the cavity constitutes a quantum perturbation of the strong classical potential . this perturbation strongly depends on the many - body atomic state . additional nonlinear dependence on the bose hubbard parameters leading to semi - classical effects can be incorporated as in @xcite . further , the classical optical lattice and cavity light can be detuned from each other . the light is pumped from the side of the main axis of the high q cavity , at an angle not necessarily at @xmath5 . the system is depicted in fig.[fqp ] , where the effect on the scattered light is shown and will be explained through the paper . moreover , it is useful to exploit the spatial structure of light as a natural basis to define atomic modes , as the coupling coefficients @xmath33 can periodically repeat in space @xcite . all atoms equally coupled to light belong to the same mode , while the ones coupled differently belong to different modes @xmath34 . then we have for the atomic operators , @xmath35 where the light induced density " @xmath36 and bond " @xmath37 mode operators , such that : @xmath38 with @xmath39 corresponding to the posible values of @xmath40 and @xmath41 corresponding to @xmath33 where the pair @xmath21 are nearest neighbours . for example , when illuminating in the diffraction minima @xmath42 and @xmath43 where @xmath44 are some constants . these encompass the different sets of values taken by the wannier overlap integrals @xmath33 @xcite . in addition , it is useful to define new emergent mode structures @xmath45 corresponding to @xmath46 , with @xmath21 nearest neighbours . in this case in diffraction minima , note that these emergent bond terms are absent for structureless light , that is , scattering as it happens in the diffraction maxima of light ( @xmath48 ) . in order to describe the physics of the system , one can construct an effective matter hamiltonian from @xmath49 . however , difficulty arises because the operators @xmath50 and the bh hamiltonian do not commute in general . eliminating the light in the adiabatic limit , one can construct the effective hamiltonian by different methods @xcite . however , the description is only accurate as long as the magnitude of the detuning @xmath51 is very large compared with any other energy scale . beyond the adiabatic limit , our method relies on a series of canonical transformations constructed to eliminate the non - diagonal terms from the light - matter hamiltonian . in our method , the cavity decay rate @xmath1 has been introduced phenomenologically to reproduce the limit of adiabatic elimination of the light field . we find the additional corrections due to the non - commutativity the light - matter interaction with the local processes of the matter part of the hamiltonian . essentially , we perform a sequence of rotations on the hilbert space using the formula , @xmath52_n}{n ! } \label{formula}\ ] ] with @xmath53_n$ ] the @xmath54-th order commutator with respect to @xmath55 . a sequence of rotation operators @xmath55 is performed , where these are chosen to remove the non - diagonal part of the light field in the light - matter interaction via the commutator expansion after their action . the particular structures needed and the number of rotations depend on the underlying structure of the matter part hamiltonian in the light - matter interaction and the bh model . their interplay with the by - products of each rotation determines the consecutive rotation operator to be constructed . thus , one provides an ansatz for each @xmath55 operator and eliminates according to the by - product of the next transformation @xcite . the result of our method is a theory that incorporates in a perturbative operator expansion the interplay of the non - commutative character between local processes ( tunnelling and on - site interactions ) and the long - range ( nonlocal ) light induced effective interactions . the first transformation recovers the limit when light can be adiabatically eliminated , explicitly this is : @xmath56 therefore , @xmath57 is analogous to the displacement operator from quantum optics @xcite , @xmath58 is the cavity purcell factor . thus using ( [ formula ] ) , @xmath59 \label{adialim } \\ & + & \big[{\mathcal{h}}^b,\hat f\big]c\hat a^\dagger-\big[{\mathcal{h}}^b,\hat f^\dagger\big]c^\hat a \label{dyneff } \\ & + & \frac{1}{2}\big[\big[{\mathcal{h}}^b,\hat f\big]c\hat a^\dagger-\big[{\mathcal{h}}^b,\hat f^\dagger\big]c^\hat a,\hat f\big]c\hat a^\dagger \label{sq1 } \\ & -&\frac{1}{2}\big[\big[{\mathcal{h}}^b,\hat f\big]c\hat a^\dagger-\big[{\mathcal{h}}^b,\hat f^\dagger\big]c^\hat a,\hat f^\dagger\big]c^\hat a+\dots \label{sq2}\end{aligned}\ ] ] to remove the additional non - diagonal light terms beyond the adiabatic limit ( [ adialim ] ) we need to perform additional rotations as we state below . dynamical effects occur due to ( [ dyneff ] ) and light squeezing originates from ( [ sq1 ] ) and ( [ sq2 ] ) , while the dots refer to higher order photon processes due to the commutator expansion that will not be considered in what follows . a subsequent series of rotations is employed to eliminate term by term the emergence of non diagonal terms in the light sector due to the interplay with short range processes . the sequence of rotations is , @xmath60 where each rotation is given by , @xmath61 with @xmath62 , where , @xmath63 , such that @xmath64 additionally , we cast our results using the natural choice of dimensionless expansion parameters , which are defined as : @xmath65 and @xmath66 , while @xmath67 where we have used collective weighted bond " operators @xmath68 and weighted bond current " operators @xmath69 corresponding to the light induced modes @xmath34 . these are defined as , @xmath70 with @xmath71 . thus these collective operators are spatially modulated by the difference in density between nearest neighbour pairs @xmath72 . the above operators can be traced back to the modulations to the densities and currents that arise due to the fact that short range tunneling and/or on - site interactions do not commute with light - matter interaction components . the particular binomial structure of @xmath73 and @xmath74 arises as each rotation operator needed to diagonalize the hamiltonian generates higher order operator polynomials terms recursively due to the commutator expansion . these operators correspond to the emergence of particle - hole excitations and matter self - interactions in analogy to the feynman diagram expansion in momentum space . considering @xmath75 and @xmath76 as expansion parameters it is enough to consider the first few terms in the expansion of order ( @xmath77 ) linear in @xmath75 , @xmath76 and their product . thus @xmath78 and @xmath79 are the leading terms , however the full expansion can be used depending on the coupling strengths and partial re - summation of families of terms can be employed as in the standard techniques of many - body physics @xcite . the final transformation to obtain the effective matter hamiltonian of the system is achieved by , @xmath80=s(\hat \xi)\ ] ] which is the analogous squeezing operator from quantum optics @xcite . the squeezing amplitude operator is given by , @xmath81 where it is useful to define density fluctuations " operators @xmath82 and bond current fluctuations " @xmath83 , @xmath84 these operators of quantum fluctuations of matter arise due to higher order light processes entangling the matter and the light . the transformation is used to eliminated non - diagonal terms due to higher order photon processes . we have restricted the expansion to two photon processes . therefore this contains the first nontrivial correction . the effect of additional higher order photon processes could be included in principle performing additional rotations , but we will not pursue this in what follows . after straight forward algebraic procedure applying each transformation and computing the relevant commutators using ( [ formula ] ) , we get the effective atomic matter hamiltonian : @xmath85 where , @xmath86 , which is the light part after the rotation to the pump frame of reference @xcite . the effective atomic hamiltonian after the rotations is @xmath87 with , @xmath88 the result in the adiabatic limit of light @xcite with @xmath89 , @xmath90 , @xmath91 and @xmath92 . here we have neglected the non - abelian shift in ( [ adialim ] ) , but in general it can be non - zero depending on the light mode functions for complex values i.e. travelling wave configurations where via either pump or cavity modes light in - between sites can be focused . as it has been shown @xcite , this leads to the formation of structures of density and bond modes that can be nearly independent from each other . it is possible to generate in a single mode cavity spatial multimode structures of @xmath93 density modes @xcite and @xmath94 bond modes @xcite by carefully choosing how the light is pumped into the system @xcite . in the adiabatic limit ( [ ad ] ) , the structure of matter is controlled by the interplay between the bh processes , regular atomic tunneling and on - site interaction , and the light induced interaction proportional to @xmath95 . the ground state of ( [ ad ] ) will be achieved whenever atoms scatter light maximally for @xmath96 or minimally @xmath97 @xcite . as we will show below , the additional terms @xmath98 are related to light squeezing and @xmath99 arises due to the dynamical corrections from the light induced processes and their interplay with the short - range bh processes due to the structure imprinted on the matter . these will modify the landscape of quantum phases the system can access , as well as , the properties of light beyond being a superposition of structured coherent states @xcite . using the light induced mode decomposition , then we can write , @xmath100 , \label{mdecomp}\end{aligned}\ ] ] with @xmath101 , where we have used the light induced density " @xmath36 and bond " @xmath37 mode operators . the additional contributions in @xmath102 are the first order corrections in @xmath103 and @xmath76 originated by the non - commutative nature between local processes and the global structure introduced due to the light induced modes . the terms due to light induced dynamics are @xmath104 these dynamical terms can have a significant effect on the effective hamiltonian renormalizing the effective light induced interaction terms . additionally , they can aid the formation of structured ground - states due to their dependency on the atom number difference between nearest neighbour sites . in general , these terms tend to induce structure in the atomic density as @xmath105 increases due to light - matter quantum correlations @xmath106 and due to semiclassical effects for @xmath107 @xcite . in principle , going beyond the perturbation character of the expansion could be handled via renomalization . close to a structural phase transition , where @xmath108 , for example , from a structured ground - state ( with dw order ) to a homogenous ground - state ( a normal superfluid ) , the leading behaviour shows an instability for @xmath109 as , @xmath110 the formation of this instability means a structure ground state can be an energetically favourable solution depending on the coupling constants strengths and the competition with other processes , from the adiabatic limit and the bh model . this provides an amazing potential for manipulation with the purpose of quantum simulation , as one can select the inhibition or enhancement of the interplay with local processes . one can design this using the structure constants @xmath33 and @xmath95 , thus controlling the light induced mode formation and changing the onsite interactions via feshbach resonances or even via the classical optical lattice potential . the expansion could be further manipulated by the use of diagrammatic tools , including well know partial resumations techniques ( i.e. feynman diagrams ) but we will not pursue this here , as we are interested in the regime where @xmath75 and @xmath76 are perturbation parameters . the terms due to the interplay with short - range processes @xmath99 and @xmath98 contain the effect of higher order correlated processes , as they contain in addition to 2-point correlations , @xmath54-point correlations with @xmath111 . beyond the adiabatic limit and dynamically generated processes , the additional terms in @xmath102 modify the energy due to light squeezing are : @xmath112 @xmath113 and @xmath114 are strongly smeared out in the limit where the effect of the light is classical . this occurs because the atoms maximise light scattering to reach the ground - state of the effective hamiltonian . however , these are relevant for the case where the strong classical signal is suppressed and the effect of light - matter quantum correlations is significant @xcite . in general , they have a suppression effect upon fluctuations for @xmath115 while they promote an instability for @xmath116 , as @xmath117 . in addition , they introduce a direct modification to tunneling amplitudes that are controlled by the pattern of emergent bond modes via the different possible couplings @xmath45 . importantly , as we will show their origin has a non - trivial effect on the properties of light . as shown previously , the effective hamiltonian @xmath102 is diagonal in the light sector and first order quantum corrections have been included . it follows that the full solution to the light - matter state can be written as : @xmath118 where the subscript @xmath119 " ( @xmath120 " ) corresponds to the light ( matter ) part ; @xmath121 , and @xmath122 . the light components are squeezed coherent states @xmath123 , with the squeezing operator @xmath124 $ ] and the displacement operator @xmath125 @xcite . the ground state of the effective hamiltonian is @xmath126 . the light amplitudes due to the projection of the matter structure are @xmath127 , @xmath128 , with @xmath129 . the weights due to the dynamical character of the light are @xmath130 with @xmath131 . in addition , the squeezing parameter amplitudes corresponding to the projection onto the matter sector are @xmath132 with , the squeezing amplitude operator @xmath133 . therefore , the structure of the strongly correlated matter gets imprinted in the quantum properties of light via the squeezing parameter projections @xmath134 . this generates a non - trivial superposition of squeezed coherent states entangled with the strongly correlated matter . in the above , we have neglected next - nearest neighbour and higher processes , considered all the first order corrections @xmath135 and terms order @xmath136 while constraining to see the effect of two photon processes . however , this is not a limitation in our method , since additional @xmath54-neighbour processes or additional @xmath54-photon processes can be incorporated straightforward if relevant . the number of photons can be written as , @xmath137 where @xmath138 and @xmath139 the weights corresponding to the matter component projections ( the probabilities ) . @xmath140 are the coherent state components and @xmath141 the corresponding squeezing parameters depending on the projections on the matter states , and @xmath142 the number of light induced components . therefore the quantum properties of the matter are accessible at the level of the photon number . moreover the light - amplitude is sensitive to the particular structure that emerges due to the correlated phases of matter @xcite . in the limit of large detunning @xmath143 the above reduces to , @xmath144 which is equivalent to the adiabatic limit @xcite . ( red and blue ) , the quadrature is centred around zero ( purple ) . the quadrature widths are given by @xmath145e^{-2r})^{1/2}/2 $ ] and @xmath146e^{2r})^{1/2}/2 $ ] , with @xmath147 the squeezing parameter , @xmath148 and @xmath149 , with @xmath89 . the system exhibits squeezing in the quadrature @xmath150 . each coherent state component is squeezed at the same angle . ( b ) quadrature squeezing at @xmath150 in db , @xmath151 , dashed lines correspond to squeezing projecting to one component , solid lines to the full state . ( c ) optimal squeezing as a function of the cavity decay rate , optimal squeezing is achieved whenever @xmath152 solid line . dashes correspond to projecting to one component with two light induced modes , dots corresponds to projecting to a single component with 4 light induced modes . parameters are : @xmath153 ( b ) and ( c).,scaledwidth=95.0% ] the squeezing in the quadratures can be written as , @xmath154 with @xmath155 , @xmath156 , and @xmath157 . this can be rewritten as , @xmath158 in particular when @xmath159 and @xmath160 so that @xmath161 is hermitian , then : @xmath162 for two light induced modes with density coupling ( @xmath163,@xmath164 ) in mean - field approximation , the above reduces to : @xmath165 where @xmath166 , and @xmath149 . the additional terms to regular squeezing are due to atomic fluctuations in each light induced component . we have used the identities , @xmath167 and that in mean - field approximation @xmath168 @xcite . beyond mean - field approximation , as the system goes in the normal sf state ( @xmath169 ) additional corrections due to additional coherent amplitudes that depend on their atomic fluctuations will increase the super - poissonian character of the light quadratures . the relation between quadrature components is shown in fig.[qpl]a . in particular , the quadratures at @xmath150 can be squeezed several db in the case where the cavity detunning is of the order of the recoil energy @xmath170 , as in @xcite with @xmath171 and @xmath172 . choosing @xmath173 and @xmath174 with @xmath175 in the sf state where fluctuations are maximal , one can easily achieve @xmath176 and about 10db of squeezing improving the ratio @xmath177 by changing the detunning or improving the cavity , see fig.[qpl]b . currently , squeezing with microwave fields has reached 10db @xcite . minimising ( [ sqe ] ) , optimal squeezing is found when @xmath178 . optimal squeezing at @xmath150 is given by , @xmath179 therefore , optimal squeezing is limited strongly by how small is the ratio @xmath180 . note that each coherent state component is squeezed stronger by a factor of @xmath181 , squeezing in each component @xmath182 more with respect to the total state , see dashed lines in fig.[qpl](b ) and ( c ) . thus the projection to a single component improves optimal squeezing . this projection is even natural , being a consequence of spontaneous symmetry breaking in the system without optical lattice @xcite . in the case of @xmath93 density modes one has in general for @xmath183 , @xmath184 with @xmath185 corresponding number operator of the light - induced density mode component per site . in general , projecting to a single component of @xmath93 modes produces an enhancement factor on squeezing of @xmath181 . for example , projecting to a single component with four light induced modes gives a enhancement factor of @xmath186 , see fig.[qpl](c ) dotted line . therefore , by incrementing the number of light - induced modes one can optimise squeezing in a single component even though the cavity decay ratio @xmath180 is not that small . the general structure of the light - matter properties for arbitrary number of bond and density modes is rather involved as it contains information regarding the correlated phases of matter that emerge . it is instructive to see the effect on the squeezing parameter @xmath147 for some cases , as we will show in what follows . when atoms scatter light homogeneously ( @xmath187 , @xmath188 , @xmath189 ) , local density imbalance is suppressed . as it has been shown @xcite , in the adiabatic limit sf and mi scatter light differently depending on the properties of the quantum many - body state . as other energy scales become relevant , the additional terms amount to renormalization of the induced interaction in the matter wave coherences , the terms in @xmath190 , such that @xmath191 for @xmath192 . this allows to enhance the effect due to the matter wave coherences via the on - site interaction in the effective hamiltonian . the squeezing parameter using mean - field approximation is @xmath193 , with atom number per site @xmath194 and the sf order parameter @xmath195 . therefore , for a deep classical ol ( @xmath196 ) there will be no squeezing due to @xmath147 . as the classical ol becomes shallower ( @xmath197 ) , the squeezing parameter is maximal for mi and smoothly decreases as we reach the sf state , see fig . [ transitions]a . this is correlated with the fact that light scattering while illuminating in between density maxima ( at the bonds ) is maximal in the mi while decreasing as the sf grows @xcite . as the number of light induced modes in the matter increases , the induced structures play a substantial role on light squeezing . for 2 light induced modes , such that their amplitudes alternate sign every other site ( @xmath198 or @xmath199 and @xmath200 ) , we find that the matter induces structure to the squeezing parameter . as it has been shown @xcite besides from sf , mi the system supports gapped superfluid states , dimer phases , supersolid ( ss ) and density waves ( dw ) . without bond ordering ( @xmath163 , @xmath198 ) , the squeezing parameter is different for ss , dw , sf and mi phases . in mean field theory , the squeezing parameter is proportional to the product of sf order parameters in each light induced component , @xmath201 where @xmath202 correspond to each light induced mode component . thus , for an insulating state ( dw or mi ) @xmath203 while for a sf @xmath204 and in the ss state @xmath205 . indeed , as the onsite interaction increases i.e. for half integer fillings when atoms scatter light maximally to reach the ground state of @xmath102 ( @xmath96 ) , light will be squeezed maximally in sf , while as ss emerges , it will diminish until reaching the dw state where no squeezing is possible , see fig . [ transitions]b . the total coherent state amplitude is @xmath206 when dw order is present , while @xmath207 in the sf or mi . moreover , when atoms scatter light minimally to reach the ground - state ( @xmath97 ) , the squeezing parameter is different for sf and gapped sf states . the squeezing parameter for a quantum superposition ( qs ) state @xcite is @xmath208 for incommensurate fillings @xmath209 with @xmath210 positive integer . thus , for a gapped sf @xmath211 while for a normal sf state @xmath212 . ) , as the system goes from mi to sf the squeezing parameter diminishes while the change becomes more visible as density increases @xmath213 . ( b ) two component system for maximum light scattering ( @xmath214 ) at @xmath175 filling . the system goes from dw to ss and to sf as @xmath215 increases , dashed line is the difference in density @xmath216 and solid line is the squeezing parameter @xmath147 . ( b ) inset , mi to sf transition ( @xmath214 ) at @xmath217 . ( c ) squeezing parameter @xmath147 for the supersolid dimer ( ssd ) to sf transition ( @xmath218 ) at @xmath219 . for @xmath220 the system is in ssd while for @xmath221 is sf . parameters:(a ) @xmath222 , @xmath223 , @xmath224 ( b ) @xmath225 , @xmath226 , @xmath227 ) ; ( c ) @xmath228 , @xmath229 , @xmath230 . @xmath231 , @xmath232 . , scaledwidth=95.0% ] in addition , dynamical terms can induce bond ordering due to the emergent coupling @xmath45 as @xmath233 increases . emergent bond ordering due to density coupling occurs because products of weighted bond and bond current operators modify the effective hamiltonian via @xmath99 . these terms arise because on - site interaction and tunneling do not commute in general with the light - induced long - range interaction . the new terms that appear in the effective hamiltonian favour density imbalance as @xmath234 increases and modify the coupling of matter wave coherences with it . explicitly , we have to order @xmath235 , @xmath236\ ] ] the current operators @xmath237 are structureless but the weighted bond operators @xmath238 induce a staggered field between bonds as density varies between every other site and the difference between atom populations can alternate sign . essentially , the density variation acts as an additional dynamical diffraction element that affects the interference of the matter waves in between density maxima . the matter waves in other to compensate the staggered field and optimize the energy in the effective hamiltonian acquire a phase pattern between adjacent sites . this translates in the formation of dimer states . the difference in phase of the matter waves @xmath239 . thus , for @xmath96 , bond ordering will occur and dimer physics @xcite will emerge even in a deep optical lattice . as a consequence , 4 bond light induced modes will form leading to a superposition of 4 light - matter correlated squeezed coherent states . the squeezing parameter can be cast as @xmath240/\delta_c^2 $ ] with @xmath241 , @xmath242 and @xmath243 the order parameter of each effective induced mode . therefore , the squeezing parameter of light inherits the structure due to bond - ordering even in a deep optical lattice . therefore , the interplay between short - range processes and the long - range cavity induced interaction leads to the emergence of physics absent in the classical optical lattice and the adiabatic light limit in this configuration . this implies a new alternative for the design of hamiltonians containing dimer physics in analogy with spin - liquid hamiltonians @xcite . in the case with only off - diagonal light - matter coupling ( @xmath244 , @xmath199 , @xmath189 ) for maximal light scattering ( @xmath96 ) , the squeezing parameter is different between sf , superfluid dimer ( sfd ) , supersolid dimer ( ssd ) and ss states . the effective interaction strength in the adiabatic limit gets renormalized by the term @xmath245 to order @xmath246 . the squeezing parameter in mean - field approximation and using the typical statistical properties of the states @xcite can be estimated as @xmath247 , where the populations for each dimer are @xmath248 and @xmath249 . in contrast to diagonal coupling ( @xmath250 ) , the normal sf state @xmath147 is minimal ( @xmath251 ) , as @xmath252 , @xmath253 and @xmath254 . in sfd , bond ordering occurs , thus @xmath255 , @xmath256 , @xmath257 , @xmath239 with @xmath258 . typically , dimer states have @xmath259 , thus @xmath260 depending on the parameters chosen for the system . for ssd , bond ordering and density modulation occurs , then @xmath261 , @xmath262 , @xmath263 , @xmath239 . thus , @xmath147 is maximal as @xmath264 , see fig . [ transitions]c . therefore , as bond ordering occurs and dw order emerges @xmath147 is different from zero . for minimal light scattering ( @xmath97 ) one has direct information regarding ss order solely due light - matter quantum correlations . the squeezing parameter has considerably simpler structure with respect to dimer phases since , @xmath265 with @xmath266 the number of atoms in each light induced mode per site . the squeezing parameter tracks directly the emergence of dw order . therefore , for ss @xmath258 and for homogenous sf @xmath203 , while the coherent state amplitudes are @xmath267 but the number of photons is @xmath268 . beyond the @xmath269 limit , we use the methods of quantum optics @xcite and we find the effective master equation for the system as @xmath270+\frac{g_{\mathrm{eff}}\kappa}{\delta_c}\big(2\hat g^\dagger\tilde{\rho}\hat g+[\hat g^\dagger\hat g,\tilde\rho]_+\big ) \\ \tilde\rho&=&\sum_{\varphi_q,\varphi_l}p_{q , l}\beta_{\varphi_q}\beta_{\varphi_l}\|\tilde\alpha_{\varphi_q},\xi_{\varphi_q}\rangle_a\|\varphi_q\rangle_b \;_b\langle \varphi_l\|_a\langle \xi_{\varphi_l } , \tilde\alpha_{\varphi_l } \| \nonumber\end{aligned}\ ] ] where @xmath271 is the density matrix , @xmath272_+$ ] is the anti - commutator and @xmath273 are the matter coefficients ( probabilities ) that can describe either a pure or mixed state . the second term in the master equation is the effective liouvillian which includes dissipation . measurement back - action beyond the @xmath89 , @xmath90 and @xmath91 limits can be devised by using @xmath161 as the effective jump operators for quantum trajectories . the operator @xmath161 is related to the coherent state amplitudes of light via @xmath274 . in the above the markov approximation is implied in the limit when @xmath275 @xcite . this allows to consider the effect of measurement back - action , the role of local processes and their interplay due to light - induced non - local interactions simultaneously . this opens a new venue for exploration regarding the design of global structured dissipation channels and measurement induced projection and state design of non - trivial quantum correlated states @xcite , as well as , control @xcite and the transition to classicality @xcite . as it has been shown , this can greatly enhance and optimise the desired quantum properties of light by design . we have shown that quantum optical lattices offer a new tool to engineer a generalised class of states that are a non trivial superposition of structured squeezed coherent states of light entangled with matter . these states are entangled with the matter at the fundamental level due to the structure of quantum many - body matter states . we have demonstrated that breaking symmetries by design one can induce structure to the parameters that control the nonclassical features of light . this has been shown to be accessible via quantities such as , the photon number and the quadratures of light . we have shown how the quantum properties of light contain the information of matter - field coherences , density patterns of matter and light - matter quantum correlations . thus the properties of strongly correlated phases of matter get imprinted on the quantum properties of light . moreover , we have found that the interplay between induced long - range processes and ordinary short - range atomic processes lead to the modification of the effective hamiltonian of the system . we have obtained that the effect of local processes can be used to generate delocalized dimer phases due to the dynamical properties of light even in deep optical lattices . moreover , one can optimise these nonclassical features depending on cavity parameters and the structure imprinted to the matter that gets transferred to the light - induced mode structure . additionally , the non - trivial light - matter correlated states that arise can be used to design dissipation channels via the effective master equation and measurement back - action . a pathway to study the behaviour we describe is to combine several recent experimental breakthroughs : detection of light scattered from ultracold atoms in ol was performed , but without a cavity @xcite and bec was trapped in a cavity , but without a lattice @xcite . as our treatment of the system is based on off - resonant scattering , this is not sensitive to a detailed atomic level structure . therefore , our treatment applies to analogous arrays of natural or artificial quantum objects such as : spins , fermions , molecules ( including biological ones ) @xcite , ions @xcite , atoms in multiple cavities @xcite , semiconductor @xcite or superconducting qubits @xcite . the work was supported by the epsrc ( ep / i004394/1 ) . 99 lewenstein m , sampera a , and ahufinger v ( 2012 ) _ ultracold atoms in optical lattices : simulating quantum many - body systems _ oxford university press . bloch i , dalibard j , and sylvain j ( 2012 ) _ nat . * 8 * 267 - 276 .	quantum trapping potentials for ultracold gases change the landscape of classical properties of scattered light and matter . the atoms in a quantum many - body correlated phase of matter change the properties of light and vice versa . the properties of both light and matter can be tuned by design and depend on the interplay between long - range ( nonlocal ) interactions mediated by an optical cavity and short - range processes of the atoms . moreover , the quantum properties of light get significantly altered by this interplay , leading the light to have nonclassical features . further , these nonclassical features can be designed and optimised .
cliques are highly - interconnected subgraphs ( complete graphs ) , and appear dominantly in networks which describe wide - ranging complex systems occurring from the level of cells to society . and , the cliques are actively investigated in recent years because of provisions of important insights to information processing , hierarchical modularity , and community structures . for instance , in gene regulatory networks , small cliques correspond to the feed - forward loop which is one of the network motifs @xcite . the motifs play an important role in gene regulation @xcite , and are regarded as building blocks of life . furthermore , the cliques are a representation for clusters , communities , and groups @xcite because there are edges among persons as nodes if there are friendships , partnerships , and _ etc . _ among the persons in social networks . therefore , the cliques help to detect community structures @xcite in social networks . again , in protein - protein interaction networks , the cliques are powerful tools for understanding evolution of proteins and functional predictions of proteins having unknown function @xcite because proteins which have same functions tend to interact . motivated by these breakthroughs , recent efforts have taken place to analytically evaluate the abundance of subgraphs , including cliques , based on statistical mechanics @xcite , providing excellent knowledge about the local interaction patterns @xcite and the time evolution of the abundance of subgraphs including cliques @xcite . these previous works focus on the local information such as the subgraph and clique abundance , and the size of the giant components led by percolation via a class of subgraphs such as the subgraph percolation @xcite , the @xmath2percolation @xcite , and the clique percolations @xcite . in recent years , however , it has been revealed that real - world networks are constructed by overlapping subgraphs including cliques @xcite ; thus it is important to elucidate global structures in networks consisting of cliques . for example , dynamics of a high order emerge by the combined network motifs in gene regulatory networks @xcite . in particular , the several power - law statistical properties have been empirically found in real - world complex networks . one of the properties is scale - free connectivity @xcite which is characterized by a power - law degree distribution @xmath3 with @xmath4 empirically found @xcite . the scale - free connectivity means that a few nodes ( hubs ) integrate a great number of nodes and most of the remaining nodes do not . another of the properties is hierarchical modularity which is characterized by a power - law clustering spectrum @xmath5 with @xmath6 empirically found , and this property suggests a hierarchical structure of the cliques @xcite . a clustering spectrum is defined as an average clustering coefficient of nodes with degree @xmath7 , where the clustering coefficient means the density of edges among neighbors of a node . since these properties reflect a global structure of a network , it is significant to clarify relationships between these properties and the global structures of the combined cliques . in this paper , we propose the @xmath0-clique network as a powerful tool for understanding global structures of combined highly - interconnected subgraphs . furthermore , we provide the theoretical predictions for well - known statistical properties of @xmath0-clique networks embedded in a complex network using the degree distribution and the clustering spectrum , and evaluate our theoretical predictions with numerical simulations . the theoretical predictions are established by applying the statistical method in @xcite . moreover , we discuss relationships of statistical properties which are observed between several real - world networks and their @xmath0-clique networks . @xmath0-clique networks are represented as sets of nodes and edges which are contained in @xmath0-node cliques , corresponding to @xmath0-node complete graphs , embedded in an original network . figure [ fig : c_net ] shows a schematic diagram of @xmath0-clique networks . the original network [ fig . [ fig : c_net ] ( a ) ] has two clique networks [ figs . [ fig : c_net ] ( b ) and ( c ) ] , and the clique networks are expressed as the circled black nodes with black edges . the gray nodes and edges are eliminated because the nodes and edges are affiliated with no cliques . following a procedure , @xmath0-clique networks are extracted from an original network . in addition , original networks are equivalent to @xmath8-clique networks in the absence of isolated nodes corresponding to nodes which have no edges . in this paper , we assume that the original networks have no isolated nodes . we utilize the algorithm based on the network motif detection @xcite to find the cliques although finding clique abundance is computationally intractable ( np - hard ) , enumeration of @xmath0-cliques in a given network can be done in polynomial time if @xmath0 is a constant @xcite . -clique networks embedded in the original network ( a ) . the @xmath0-clique networks [ ( b ) and ( c ) ] are expressed as the circled black nodes with black edges . ] -clique networks embedded in the ba network with @xmath9 and @xmath10 ( shifted for clarity ) . @xmath11 means the average degree . the symbols correspond to the numerical results , and the dashed lines are theoretical predictions given by eq . ( [ eq : p_n(k ) ] ) . the solid lines show @xmath12 . ] we consider degree distributions from @xmath0-clique networks @xmath13 . the degree distribution is defined as the existence probability of nodes with degree @xmath14 which is the number of edges at a node in a @xmath0-clique network . in addition , @xmath15 denotes the degree distribution @xmath16 from an original network because @xmath17 . in order to establish a theoretical prediction on the degree distribution of @xmath0-clique networks , we propose an approximation method based on the statistical method in @xcite . we assume that the clustering spectrum @xmath18 corresponds to the probability that two neighbors of a node with degree @xmath7 ( @xmath19 ) are linked . first , we consider the probability @xmath20 that an edge on a node with degree @xmath7 is eliminated due to the extraction of @xmath0-clique networks from an original network . for simplicity , we assume that the probability of an edge to be eliminated from a node is independent from the probability of another edge to be eliminated from the same node and the probability of the same edge to be eliminated from a neighbor . this assumption is a suitable approximation in the case of random graphs @xcite because the probability that there is an edge between two nodes is constant . we show that the approximation is also suitable in the case of arbitrary large - scale graphs ( networks ) for large @xmath7 with numerical simulations . here , we focus on a subset which consists a node with degree @xmath7 , neighboring nodes and edges among these nodes . then , the edge on the node with degree @xmath7 belongs to @xmath21 @xmath0-cliques which are formed with the probability @xmath22 , where @xmath23 . that is , the probability that the edge is not contained in one of @xmath21 @xmath0-cliques is @xmath24 $ ] . since the edge is eliminated if the edge is contained in no @xmath0-cliques , from the assimptation of independence , the probability @xmath20 can be written as @xmath25 next , we characterize the conditional probability that the degree shifts from @xmath7 to @xmath14 due to the extraction of @xmath0-clique networks using the probability @xmath20 . the conditional probability can be expressed using the bimodal formula , and we have @xmath26^{k^{(n)}}\phi_n(k)^{(k - k^{(n)})}. \label{eq : phi(k\|j)}\ ] ] the degree distribution from an @xmath0-clique network @xmath13 is proportional to the sum of @xmath27 for @xmath28 . therefore , the degree distribution is finally described as @xmath29 where @xmath30 and @xmath31 correspond to the total number of nodes in an original network and in a @xmath0-clique network , respectively . using @xmath16 and @xmath18 , the total number of nodes in the @xmath0-clique network can be estimated by @xmath32 . \label{eq : n_n}\ ] ] in order to confirm the theoretical predictions , we performed numerical simulations for the barabsi - albert ( ba ) network @xcite , which provides power - law degree distribution ; @xmath3 with the degree exponent @xmath33 . figure [ fig : deg_ba ] shows the degree distributions of @xmath0-clique networks embedded in the ba network . as shown in fig . [ fig : deg_ba ] , our theoretical predictions are in good agreement with the numerical results , indicating that the approximation is suitable . in addition , the different degree distributions are observed between the @xmath0-clique networks and the original network . the degree at a node shifts due to the extraction @xmath0-clique networks from an original network . here , we consider the theoretical predictions for the shifts with the statistical properties from an original network . using the probability @xmath20 [ eq . ( [ eq : phi(k ) ] ) ] that an edge is eliminated due to the extraction of @xmath0-clique networks , the expectation value of the degree at a node in a @xmath0-clique network can be written as @xmath34 . \label{eq : k'}\ ] ] the probability @xmath20 is dependent on the clustering spectrum @xmath18 as shown in eq . ( [ eq : phi(k ) ] ) . since it is empirically found that the spectrum follows the power law in most complex networks @xcite , we assume the power - law spectrum ; hence @xmath35 . moreover , we use the feature of napier s number , @xmath36 for large @xmath7 , to rewrite the probability @xmath20 [ eq . ( [ eq : phi(k ) ] ) ] . in doing such we have @xmath37 , \label{eq : phi(k)_2}\ ] ] where @xmath38 in particular , the probability @xmath20 is independent of the degree @xmath7 when @xmath39 , and the proportional relationship between @xmath14 and @xmath7 is satisfied . in order to confirm the theoretical prediction , we performed numerical simulations for the ba network . figure [ fig : chg_deg ] shows the shift of the degree at a node due to the extraction of the @xmath0-clique networks . as shown in fig . [ fig : chg_deg ] , our theoretical prediction is in good agreement with the numerical results . figure [ fig : phi_k ] shows the probability @xmath20 which is obtained from the extraction of @xmath0-clique networks . assume that @xmath35 , @xmath40 and @xmath41 are about 0.02 and 0.1 with least - square method , respectively . we give the theoretical prediction with these values . as shown in eq . ( [ eq : phi(k)_2 ] ) , @xmath42 declines exponentially with @xmath7 , indicating that a degree of a high - degree node tends to stay . the prediction is in agreement with the numerical results . -clique networks from the ba network with @xmath9 and @xmath10 . the symbols correspond to the numerical results , and the dashed lines are given by eq . ( [ eq : k ] ) . the solid lines show @xmath43 . ] is obtained from the extraction of @xmath0-clique networks from the ba network with @xmath9 and @xmath10 . the symbols correspond to the numerical results , and the dashed lines are given by eq . ( [ eq : phi(k)_2 ] ) . ] in the case of @xmath44 , however , the agreements are weak in fig . [ fig : chg_deg ] and [ fig : phi_k ] . there are two reasons . one is the assumption of independence . in scale - free network , low - degree nodes tend to connect to high - degree nodes . as shown in fig . [ fig : phi_k ] , the probability that an edge on the high - degree node is eliminated is very small . for this reason , real @xmath20 for small @xmath7 tends to be smaller than eq . ( [ eq : phi(k)_2 ] ) . therefore , real @xmath14 tends to be larger than our theoretical prediction . another is fluctuation in clustering spectra @xmath18 . in the case of scale - free networks , the fluctuation is large for small @xmath7 , and is contrary small for large @xmath7 because of heterogeneous connectivity . and , the probability that a @xmath0-clique is formed is described as @xmath22 . that is , the error increases with @xmath45 . therefore , our theoretical prediction tends to be in weak agreement in the case of large @xmath0 and small @xmath7 . the clustering spectrum of the ba network is independent of the degree @xmath7 @xcite . that is , @xmath46 . according to eq . ( [ eq : d_n ] ) , we predict that the shifts of the degree follow the nonlinear relationship because of the nonzero @xmath47 ; for example , @xmath48 and @xmath49 . as shown in fig . [ fig : chg_deg ] , our prediction is in agreement with the numerical results . .network sizes , average degrees , and characteristic exponents of the investigated real - world networks and the ba network . the exponents @xmath50 and @xmath41 are extracted using the maximum likelihood estimation @xcite and the analytical approximation @xcite ; thus @xmath51 , respectively . [ cols="<,^,^,^,^,^,^",options="header " , ] -clique networks embedded in the investigated networks ( shifted for clarity ) . the solid lines show @xmath52 in the each main panel . the exponents @xmath50 are provided from table [ table : real_net ] , respectively . the each inset shows the shift of the degree due to the extraction of @xmath1-clique network . in the each inset , the solid lines correspond to @xmath53 . ( a ) internet ( as level ) , ( b ) metabolic network of _ e. coli _ , and ( c ) protein - protein interaction network of yeast . ] we discuss statistical properties of @xmath0-clique networks embedded in a network with power - law statistical properties . here , we focus on the scale - free connectivity which is one of the well - known power - law statistical properties and is defined as a power - law degree distribution , @xmath3 . in networks with scale - free connectivity , we predict that forms of the degree distributions are invariant between the 3-clique and the original network when @xmath39 . this is because the proportional relationship between the degrees at nodes in the original network and in the @xmath0-clique networks , is satisfied under this condition . in order to verify our prediction , we investigate the degree distributions from @xmath0-clique networks embedded in several real - world networks with scale - free connectivity : the autonomous system representation of the internet @xcite , the metabolic network of _ escherichia coli _ @xcite , and the protein - protein interaction network of yeast @xcite . these real - world networks have hierarchical modularity , indicating the power - law clustering spectra ; hence , @xmath5 with @xmath6 @xcite . in addition , we also consider the ba network , which does not have hierarchical modularity , for comparison . we summarize the networks size , the average degrees , and the exponents characterizing each network in table [ table : real_net ] . the exponents @xmath41 from the real - world networks with hierarchical modularity are almost one @xcite ( see also table [ table : real_net ] ) . therefore , we expect that the forms of the degree distributions are invariant between the 3-clique and the original network because @xmath54 . figure [ fig : real_deg ] shows the degree distributions of @xmath0-clique networks embedded in the real - world networks . as expected , the forms of the degree distributions are invariant between the original and the 3-clique network because of the proportional relationship between @xmath7 and @xmath55 ( see the insets in fig . [ fig : real_deg ] ) . in contrast , the exponent @xmath41 from the ba network is equivalent to zero @xcite ( see also table [ table : real_net ] ) because of there is no hierarchical modularity . therefore we predict that the power - law degree distribution from an original network is variant due to the extraction of the @xmath1-clique network ( because @xmath48 ) . figure [ fig : deg_ba ] shows the degree distributions of @xmath0-clique networks embedded in ba networks . as expected , the form of the degree distribution is variant between the 3-clique network and the original network because of the nonlinear relationship between @xmath7 and @xmath55 ( fig . [ fig : chg_deg ] ) . in this paper , we have provided theoretical predictions using the approximation method for the degree distribution of a @xmath0-clique network and the shifts of the degree due to the extraction of the @xmath0-clique network . moreover , we performed numerical simulations and show that the numerical results are in good agreement with our theoretical predictions , indicating that the approximation method is suitable . furthermore , we have found that the power - law degree distributions are identical between the 3-clique and the original networks in the scale - free networks with hierarchical modularity using our theoretical predictions . we have only focused on the power - law degree distributions in this paper . however , because of the proportional relationship between @xmath7 and @xmath55 , the converse holds for the other power - law statistical properties which are observed in real - world networks : the hierarchical modularity @xcite and the assortativity @xcite . we have confirmed that the power - law statistical properties are invariant between the 3-clique networks and the original networks , although there is no space for the showing of the data . the invariance of the statistical properties implies that structural properties are identical between @xmath1-clique and original networks . in addition , from these results , we expect that the @xmath1-clique networks are constructed by the same mechanisms as the original networks with hierarchical modularity . in contrast , we have found that the 3-clique network embedded in the ba network which does not have hierarchical modularity has different statistical properties from the original network . that is , the structural properties are different between @xmath1-clique and original networks in the ba network . we believe that these results provide new insights into global structures of combined network motifs , community structures @xcite in social and biological networks . in this paper , expressly , we found structural properties are identical between 3-clique networks and original networks . this lets us expect that 3-clique networks are constructed by the same design principles as the original networks with hierarchical modularity , and it implies that the clique networks help to understand design principles and global structures of combined significant subgraphs which reflect community and functional modules in networks . for example , it is believed that most real - world networks are constructed by the preferential attachment @xcite . because of a structural identity between 3-clique networks and original networks , we expect that the clique networks are also constructed by the same preferential attachment as the original networks . this mechanism suggests the preferential attachment of cliques @xcite . actually , it is reported that there is a preferential attachment of community in social networks @xcite . in biological networks , furthermore , cliques correspond to functional modules such as network motifs . in particular , 3-node clique , which denotes the network motifs such as the feedforward loop and so on , appears frequently . from our result , we expect that a network which consists of network motifs only is constructed by the same preferential attachment as an original network . if so , the motifs may concentrate on hubs . actually , the concentration of motifs has been found by the network analysis @xcite . in this manner , we believe that we can find new structural properties and new insights into design principles of networks via an analysis of clique networks . and , our theoretical predictions may help the analysis and its interpretation . in biological networks , especially , since it is difficult to discuss network formation processes because of no ancestral networks , we believe that the analysis help to understand design principles of networks . in addition , we may establish more realistic growing network models via the analysis . this work was partially supported by grant - iazn - aid no.18740237 from mext ( japan ) .	we propose the @xmath0-clique network as a powerful tool for understanding global structures of combined highly - interconnected subgraphs , and provide theoretical predictions for statistical properties of the @xmath0-clique networks embedded in a complex network using the degree distribution and the clustering spectrum . furthermore , using our theoretical predictions , we find that the statistical properties are invariant between @xmath1-clique networks and original networks for several observable real - world networks with the scale - free connectivity and the hierarchical modularity . the result implies that structural properties are identical between the @xmath1-clique networks and the original networks . cliques , scale - free networks , hierarchical modularity , real - world networks 89.75.hc , 89.75.da
existence of a non - zero neutrino magnetic moment has long been a concern of great interest , since it can have an observable laboratory effect such as neutrino - charge lepton elastic scattering , @xmath4 , and also some important astrophysics effect , such as cooling of sn 1987a , cooling of helium stars , etc . it is likely that neutrinos may have a small but nonvanishing mass ; for various bounds on magnetic moments and masses , see @xcite and @xcite . within the framework of the standard model , a nonzero neutrino mass usually imply a nonzero magnetic moment . it has been shown that @xcite , for a massive neutrino , @xmath5 where @xmath6 is the bohr magneton . in models beyond the standard model , right - handed neutrinos are often included in interactions ( for a review e.g. @xcite ) , so that we need not depend on a nonzero neutrino mass to generate a nonzero magnetic moment . in this article , we consider the possibility of using leptoquark interactions to generate a nonzero neutrino magnetic moment . in many unification models , such as su(5 ) , s0(10 ) , etc . , one often put quarks and leptons into the same multiplet , so that leptoquarks arise naturally for connecting different components within the same multiplet . what makes a leptoquark unique and interesting is that it couples simultaneously to both a lepton and a quark . this may help generating a nonzero neutrino magnetic interaction . specifically , when a top quark involves in the loop diagram , its mass provide a large enhancement for the neutrino magnetic moment . ( in such diagram , a massless neutrino needs some massive internal fermion to flip its chirality , giving rise to some nonzero magnetic moment . ) we add right - handed neutrinos in the general renormalizable lagrangian of leptoquarks . owing to existence of lepton numbers which recognize the generation , we distinguish leptoquarks by its generation quantum number , but this may induce four - fermion interactions which may enhance some helicity suppressed process such as @xmath7 to the extent that leptonic universality may even be violated . this usually gives a tight constraint on the leptoquark @xcite . for non - chiral vector leptoquarks of electromagnetic strength coupling , this corresponds to having a mass heavier than 50 tev for the first generation leptoquark . with such a heavy leptoquark , we still find a nonzero neutrino magnetic moment @xmath8 up to @xmath9 . for the second and third generation leptoquarks , their masses are not severely constrained by the above process . assuming their mass lying somewhere between @xmath1 tev , we obtain @xmath8 of order @xmath2 ( from the second generation leptoquark ) and @xmath3 ( from the third generation leptoquark ) , respectively . such predictions may already have some observable effects such as those mentioned earlier . leptoquarks arise naturally in many unification models which attempt to put quarks and leptons in the same multiplet . there are scalar and vector leptoquarks which may couple to left- and right - handed neutrinos at the same time , but only vector leptoquarks can couple to the upper component of the quark su(2 ) doublet . the heaviness of the top quark may enhance the neutrino magnetic moment once we use the vector leptoquark to connect to the quark doublet . of course , there are subtleties regarding renormalization of the vector leptoquark which may be treated in a way similar to gauge bosons . in our calculation , we adopt feynman rules in the @xmath10-gauge and take @xmath11 at the end of the calculation while neglecting all unphysical particles in the @xmath12-gauge . ( this is a step which has often been employed in nonabelian gauge theories . ) we begin our analysis by constructing a general renormalizable lagrangian for the quark - lepton - leptoquark coupling . following @xcite , we demand such action to be an su(3)@xmath13su(2)@xmath13u(1 ) invariant which conserves the baryon and lepton numbers but , in addition to @xcite we add terms which couple to right - handed neutrinos . for leptoquarks with the fermion number @xmath14 , @xmath15 and , for @xmath16 , @xmath17 the notation adopted above is self explanatory ; for example , @xmath18 denotes scalar and vector leptoquarks respectively , the superscript is its average electric charge or the hypercharge @xmath19 , and the subscript of a leptoquark denotes which su(2 ) multiplet it is in , and the generation index is suppressed . from here it is clear that among those leptoquarks that couple to neutrinos of both chiralities , a radiative @xmath20 diagram with the exchange of a virtual @xmath21-type quark can proceed only when accompanied by a vector leptoquark , namely @xmath22 in @xmath23 or @xmath24 in @xmath25 ; on the other hand , the exchange of a virtual @xmath26-type quark can proceed only with the scalar leptoquark , namely @xmath27 in @xmath23 or @xmath28 in @xmath25 . note that we do not consider mixing between different leptoquarks due to higgs interactions , which will introduce additional parameters . the diagram in question is shown explicitly in fig given these couplings , it is straightforward to calculate induced neutrino magnetic moments via one loop diagrams . to see that the heavy top quark mass can enhance the prediction , we calculate the @xmath20 diagram with the exchange of up - type quark and @xmath22 i.e. the first term in @xmath23 . as one of the standard methods to treat loop diagrams involving massive vecter particles , we use feynman rules in the @xmath12-gauge and take @xmath29 at the end of calculation while neglecting any unphysicical particle . in addition to minimum substitution , we add the term @xmath30 in the lagrangian , such that the whole @xmath31 coupling is in a form similar to the non - abelian @xmath32 type coupling , and the procedure results in a finite limit under @xmath33 . we obtain , with all couplings chosen to be real , @xmath34 where @xmath35 , @xmath36 , and @xmath37 , while @xmath38 and @xmath39 are given by @xmath40 note that one obtains the desirable chiral structure for the magnetic moment interaction by two different ways : the first is to have an odd number of mass insertions of the quark mass term , giving rise to the first term in @xmath41 ; the other way is by the neutrino mass term , resulting in the second term of @xmath41 . there are two advantages with the first scenario . first of all , one can obtain a nonzero magnetic moment without restricted by the very light neutrino mass . second , one may have a prediction enhanced considerably by the heavy top quark mass . before working out numerical predictions , we need to consider the constrains arising from the leptonic decays of the pseudoscalar meson , such as @xmath42@xcite . intergrating out @xmath43 and performing fierz reordering , we obtain @xmath44 relevant to the leptonic decay of a pseudoscalar meson , @xmath45 we consider the universality constraint arising from the @xmath0 leptonic decay , and neglect the neutrino mass contribution . define @xmath46 . the first and third terms of @xmath44 have interference with the standard model fermi interaction . this is an order of @xmath47 correction to @xmath48 , while the other term is a correction of order @xmath49 . furthermore , the first term which is scalar coupling is enhanced by a factor of @xmath50 so this is the dominent term to constrain the mass of the leptoquark . we assume @xmath51 which is a natural assumption for the vecter leptoquark . we obtain @xmath52 where experimental average @xmath53@xcite , and standard model calculation @xmath54 @xcite . this correspond to @xmath55 for a coupling of the electromagnetic strength , this correspond to having the vector leptoquark with a mass greater than 50 tev for the first generation . this constraint is in fact more servere than what we may obtain from the atomic parity vialation experiment , which we shall ignore in this paper . for the second and third generations leptoquarks , there is no direct restriction from the universality of the @xmath56 leptonic decay , nor from the atomic parity violation experiment . nevertheless , one can find various lower bounds for the leptoquark mass @xcite , from direct searches at the hera @xmath57 collider , the tevatron @xmath58 collider , and at the lep @xmath59 collider . typical bounds from direct searches is about few hundreds gev , while the bounds form indirect searches are given in @xcite . we shall consider a leptoquark mass in the general range of tev s . for the reason of comparisons , let us recall briefly some of the upper limit obtained from the leptonic scattering such as elastic @xmath60(@xmath61 ) with @xmath62(@xmath63 ) , @xmath64 , etc . , and also from the astrophysical processes such as cooling of helium stars , red giant luminosity and so on @xcite . as a reference point , we recall the standard model formula on the neutrino magnetic moment arising from a nonzero neutrino mass@xcite , @xmath65 ( referred to as `` the extended standard electroweak theory '' ) . accordingly , the upper limit of @xmath8 for the first generation neutrino is @xmath66 with @xmath67 ev . the upper limit may also be obtained from leptonic scatterings , which is typically @xmath68 , or from astrophysics studies with a more stringent upper limit of @xmath69 . our numerical results for the first generation are summarized in fig . 2 , where the neutrino magnetic moment @xmath70 in units of @xmath6 is shown as a function of the leptoquark mass . we note that , for the leptoquark mass @xmath71 $ ] of 50 to 100 tev , @xmath70 is of order @xmath72 , a value compatible with the extended standard electroweak theory . the upper limit of @xmath8 for the second generation neutrino is @xmath73 ( with @xmath74 mev ) in the extended standard electroweak theory @xcite , or in the range of @xmath68 from leptonic scatterings , while from astrophysics the typical value is @xmath69 . in fig . 3 , we describe our prediction on the neutrino magnetic moment @xmath70 in units of @xmath6 as a function of the leptoquark mass of 1 to 100 tev . we obtain @xmath70 around @xmath75 , a value very close to being observable . the upper limit of @xmath8 for the third generation neutrino is @xmath76 ( with @xmath77 mev ) in the extended standard electroweak theory @xcite , or in the range of @xmath78 from leptonic scatterings , while from astrophysics studies the upper limit is @xmath79 . in fig . 4 , we plot the third generation neutrino magnetic moment @xmath70 in units @xmath6 as a function of the leptoquark mass in the range of 1 to 100 tev . we find that @xmath70 is of order @xmath80 . vector leptoquarks in the tev mass range , when couple to both left- and right - handed neutrinos , offer an alternative mechanism for generating a nonvanishing neutrino magnetic moment , which in some cases is by no means negligible . this alternative mechanism ( which does not require a nonzero neutrino mass ) makes use of the special feature that leptoquarks couple simultaneously to leptons and quarks . for the third generation neutrino , there is a potential enhancement from the very large top quark mass making the corresponding predicted neutrino magnetic moments fairly sizable . we would like to acknowledge dr . chan for valuable discussions . this work was supported in part by a grant from national science council of republic of china ( nsc88 - 2112-m002 - 001y ) . particle data group : c. caso _ et al . j. * c3 * , 1 ( 1998 ) , and references therein . y. fukuda _ ( super - kamiokande collaboration ) , phys . lett . * 81 * , 1562 ( 1998 ) ; phys . lett . * b436 * , 33 ( 1998 ) . for a recent review of various hints of neutrino mass , see b. kayser , hep - ph/9810513 . r.n . mohapatra and p.b . pal , in `` massive neutrinos in physics and astrophysics '' , world scientific , singapoer , 1991 . k. fujikawa and r.e . shrock , phys . . lett . * 45 * , 963 ( 1980 ) . o. shanker , nucl . b204 * , 375 ( 1982 ) ; w. buchmller and d. wyler , phys . lett . * 117b * , 377 ( 1986 ) . w. buchmller , r. rckl and d. wyler , phys . lett . * 191b * , 442 ( 1987 ) . marciano and a. sirlin , phys . lett . * 71 * , 3629 ( 1993 ) . s. davison , d. bailey , and b.a . campbell , z. phys . * c61,613 ( 1994 ) ; m. leurer , phys . rev . d50 * , 536 ( 1994 ) , and * d49 * , 333 ( 1994 ) .	allowing leptoquarks to interact with both right - handed and left - handed neutrinos ( i.e. , `` non - chiral '' leptoquarks ) , we show that a non - zero neutrino magnetic moment can arise naturally . although the mass of the non - chiral vector leptoquark that couples to the first generation fermions is constrained severely by universality of the @xmath0 leptonic decays and is found to be greater than 50 tev , the masses of the second and third generation non - chiral vector leptoquarks may evade such constraint and may in general be in the range of @xmath1 tev . with reasonable input mass and coupling values , we find that the neutrino magnetic moment due to the second generation leptoquarks is of the order of @xmath2 while that caused by the third generation leptoquarks , being enhanced significantly by the large top quark mass , is in the range of @xmath3 . epsf.tex ( # 1 width # 2)=#2
a relatively new branch of physics combining hadronic physics and astrophysics is called astro - hadron physics . " i would like to describe some recent efforts to understand such extremely compact astrophysical objects as neutron stars starting from data obtained in laboratories , in particular from heavy - ion experiments that have been performed in such laboratories as gsi of germany , cern ( sps ) of switzerland and elsewhere . the central thesis that i will develop is that there are what may look like alternative " descriptions for the same process in different languages ranging from bare hadrons to quasiparticle hadrons to quasiparticle quarks characterized by what i would call cheshire cat mechanism . " this duality " nature seems to be operative from elementary hadrons to superdense matter perhaps existing in compact stars , suggesting a unity of particle , nuclear and astro physics . as in all works at an embryonic stage , there are false starts , pitfalls , wrong tracks and controversies . what i shall describe therefore are not necessarily well - established facts . much work will be needed to validate or invalidate some or all of them . even so , what we have at this stage is so exciting that it deserves much more attention than presently paid , particularly from the asia pacific physics community where some of the early significant works have been done by young theorists working on phd theses . i will start the discussion with the proton , the constituent of the nucleus , go to the simplest nucleus , i.e. , two - nucleon system including the deuteron , then to nuclear matter and finally to dense hadronic matter relevant to the interior of neutron stars . the topics are selected to highlight work done in korea by thesis students or young theorists . clearly part of the connections are incomplete or faulty , needing more solid structure but the logic appears sound and worth pursuing for possible breakthrough . i start with the nucleon , specifically the proton which is the lowest baryonic state . in the fundamental theory of strong interactions , qcd , this state is a bound state of three quarks in the color - singlet state confined in as yet poorly understood way within a region wherein the gluons play a crucial role . to describe this from the qcd lagrangian using its microscopic variables is at present practically impossible but the most tantalizing fact is that it can be given various different ( albeit approximate , yet qualitatively correct ) descriptions indicating some sort of dualities " are in action . one apt way of seeing this is to use cheshire cat principle " ( for a general introduction , see @xcite ) . consider three colored quarks @xmath0 of the quantum numbers of a proton confined in a spherical bag " of radius @xmath1 surrounded by a cloud of goldstone pions , the latter being indispensable to the system in order to be consistent with the spontaneously broken chiral symmetry . with a suitable set of boundary conditions consistent with the symmetries of qcd that mediate the communication between the inside microscopic qcd variables and the outside macroscopic hadronic variables , nearly all low - energy observables of the proton can be understood independently of the size of the confining bag @xcite . indeed for a large bag with @xmath2 fm , the quark - gluon degrees of freedom dominate while for a small bag with @xmath3 fm , it is the goldstone degrees of freedom that govern the dynamics of the system . this feature is known as the cheshire cat phenomenon @xcite . when the bag " is shrunk to a point @xmath4 , which is allowed by cheshire cat principle , one gets the celebrated skyrmion @xcite , a description of the baryon that is the more accurate the larger the number of colors @xmath5 . it seems that in nature , @xmath6 is already quite large , so the skyrmion picture is qualitatively correct . one conspicuous exception to the success of the cheshire cat phenomenon has , up to date , been the flavor singlet axial charge ( which will be referred in short to as fsac ) of the proton which is often associated with the so - called proton - spin problem . " it turns out that there is no obstacle to the cheshire cat manifesting in this quantity , that is , there is no proton - spin crisis . " the resolution lies in subtle quantum anomalies which are now understood . we do not yet have a complete answer to the issue which is the phd thesis subject of hee - jung lee at seoul national university but i will briefly describe how this cheshire cat property can be recovered in the fsac when chiral symmetry and chiral anomaly are judiciously taken into account @xcite . the interplay between the boundary conditions and casimir effects is found to play a crucial role . since the flavor - singlet axial current is not conserved because of the anomaly , the color can not be confined inside the bag unless a suitable boundary condition is put at the surface that cancels the outflow of the color as discovered by h.b . nielsen et al @xcite . the boundary term that does this is proportional to the chern - simons current on the surface , i.e. , the chern - simons flux ( which is invariant under neither small nor large gauge transformation ) . this influences nontrivially the fsac of the proton . in a nut - shell , what happens is that the fsac contributed by the matter fields ( quarks inside the bag and @xmath7 outside the bag ) and the fsac contributed by the gauge field ( gluons inside the bag ) more or less ( or possibly exactly if treated rigorously ) cancel , leaving behind only the small contribution from the ( gauge field ) vacuum fluctuation which is effectively a casimir effect caused by the boundary with its color - anomalous boundary condition . the cancellation and the remnant small fsac are shown in fig . [ fsac ] . the small fsac of the proton that is left over which is independent of the size @xmath1 provides another evidence that the proton could be equivalently understood _ both _ in terms of quarks / gluons and in terms of macroscopic hadronic variables . when @xmath1 is taken to be big , it is the qcd variables that figure significantly , e.g. , the mit bag . when the size @xmath1 is shrunk to a point , the proton is a skyrmion . thus we have the equivalence of the skyrmionic proton and the quark - gluonic ( qcd ) proton , that is , the cheshire cat . since there is no way that one can exactly bosonize four - dimensional qcd , the equivalence can be only approximate . thus we simply have an approximate equivalence which can be made more precise by doing more work . in what follows , i will develop the thesis that it is this feature observed in the proton that underlies many - body dynamics going all the way to neutron star matter . let me now turn to nuclei . for this , consider two nucleon interactions at very low energy . the two - nucleon system is the nuclear system that is the simplest and the only nuclear system that can be treated accurately and systematically . at present it is only at the probed energy - momentum much less than the pion mass @xmath8 mev that is amenable to an accurate computation . at very low energy , according to weinberg s theorem , " the content of qcd can be phrased in terms of the nucleons and pions treated as effective fields . in fact if we are probing a scale much less than the pion mass , we can even ignore the pions and work with the nucleons only . the corresponding framework is an effective field theory ( eft ) . what this means is that we can shrink following the notion of the cheshire cat the bag of the nucleon and work in the skyrmion limit , namely , zero - radius bag . thus a nucleus will be a collection of point - like objects interacting strongly subject to the constraint of qcd symmetries . there has recently been an intense activity on this eft for two - nucleon systems @xcite . some of the developments are summarized in a series of workshops devoted to the issue @xcite . up to date , there are two successful approaches to eft in nuclear physics . one is the original weinberg approach @xcite where a systematic power counting is made only to the `` irreducible graphs , '' for which chiral perturbation theory ( with pions figuring prominently ) becomes applicable in organizing the expansion of the series and the reducible graphs are summed to all orders with the irreducible graphs entering as vertices . this scheme used in ref.@xcite which in spirit is close to the original wilsonian eft but incurs possible errors in the power counting involves a chiral - symmetry scale @xmath9 ( where @xmath8 mev is the pion mass ) as the counting is applied only to the irreducible terms . i will call this the @xmath10 counting . the other approach @xcite motivated to account more explicitly for the anomalously large s - wave scattering lengths in the two - nucleon scattering allows one to do a systematic counting for the s - matrix as a whole . this approach renders a more systematic accounting of the powers of @xmath11 where @xmath12 ( where @xmath13 is the deuteron binding energy and @xmath14 is the proton mass ) , @xmath15 ( probe momentum ) as well as @xmath16 ( pion mass ) but at the expense of certain predictivity . this is referred to as @xmath17-counting scheme . in essence , both approaches , though somewhat different in strategy , are equally consistent with the tenet of eft and more or less equivalent in their predictive power . in what follows , i will simply focus on the @xmath10-counting approach which works stunningly well for two - body systems . actually in the processes that i will consider , the @xmath10 counting is found to be more readily adaptable to and predictive in treating nuclear physics problems . one non - trivial advantage of the @xmath10 scheme is that it allows one to calculate precisely defined corrections to what can be obtained from so - called realistic potential models ( pm in short ) that have been developed by nuclear theorists since a long time , thus giving the realistic potential models ( pm ) a first - principle justification . it allows us to study processes involving not only few - body but also many - body systems . for instance , it is possible to calculate the hep " process in the sun @xmath18 ( which is currently an exciting issue after the recent surperkamiokande neutrino data ) with an accuracy that can be controlled systematically . such a calculation in the @xmath17-counting scheme is most likely to be a hard task . in a series of recent papers , tae - sun park ( currently a post - doc at triumf in vancouver ) and his collaborators @xcite have shown in a finite - cutoff regularization that the eft results of the leading order terms in _ all _ two - body observables at low energy @xmath19 are precisely reproduced by the potential model results . this is the case not only for scattering amplitudes but also all electroweak response functions . what eft can do that the potential models ( pm ) can not is that the corrections to the leading order results given accurately by the pm are calculable systematically . this is the _ power _ of eft as a _ theory_. for low - energy processes , this privileged role of the pm in eft can be understood by the fact that the tail of the wave functions is a physical quantity and the realistic potential models which are fit to experiments have the _ correct _ asymptotic properties in the wave function @xcite . considered to order @xmath20 where @xmath21 is the order in the @xmath10 counting ( which i will consider relative to the leading order term in the expansion of the irreducible graphs ) , the s - wave scattering amplitudes are accurately postdicted by t .- park et al @xcite and further improved by chang - ho hyun ( a graduate student of seoul national university ) et al @xcite up to @xmath22 for @xmath23 and a cutoff appropriate to the number of pions exchanged ( one or two ) in the irreducible graphs . all deuteron properties are also well understood within the same scheme @xcite . as an important spin - off , the scheme allowed the calculations to order @xmath24 and @xmath25 of the proton fusion process in the sun @xcite p+pd+e^+ + _ e[pp ] and of the threshold np capture @xcite with polarized projectile and target nucleons + d+.[np ] the process ( [ pp ] ) crucial for the solar neutrino problem is given in the scheme to an accuracy of @xmath26 percents ( the uncertainty here is due to the exchange current that appears at order @xmath25 ) . the unpolarized cross section for ( [ np ] ) has been computed to the accuracy of 1 percent in complete agreement with the experiment . more significantly , the polarization observables @xmath27 ( circular polarization ) and @xmath28 ( anisotropy ) have been predicted _ parameter - free _ in ref.@xcite . since there are no experimental data available yet , this is a genuine prediction involving matrix elements that are suppressed relative to the allowed term by @xmath29 3 orders of magnitude . these quantities are currently being measured in several laboratories and will soon be available . the outcome will be an exciting check of the prediction of the theory , perhaps the first of the kind in nuclear physics . in all these postdictions and predictions , there is very little @xmath10 dependence as required by the tenet of eft , assuring that the scheme is fully consistent . one can go up in the momentum range by doing higher order calculations . phillips and cohen @xcite discuss how the two - body em form factors can be described in the @xmath10 scheme . pushing somewhat the validity of the scheme , one can calculate even the process e+de+n+p involving large momentum transfers @xmath30 gev . in fact this process measured in 1980 s at als of saclay and elsewhere is considered to be the _ unambiguous _ confirmation of meson - exchange currents in nuclei ( see @xcite ) . in sum , the dilute nuclear systems such as the two - nucleon bound and scattering states can be quite accurately described in terms of certain effective degrees of freedom connected to qcd via a cheshire cat mechanism . i shall now jump directly to the infinite - body problem , namely , nuclear matter , although the extension of eft to three- or more - body problems is not fully worked out yet . for more than a few nucleons and many - nucleon systems like nuclear matter and denser matter , the eft described above can not be straightforwardly applied . in fact a systematic approach of the type does not yet exist . the reason is rather simple to understand . first of all when several nucleons are involved the relevant kinematics is not always one to which the low - energy / momentum expansion with a manageable number of terms is applicable and further a systematic expansion would involve many terms whose coefficients are not fully determined from either theory or experiments . this means that strictly speaking , no parameter - free calculation that does not invoke some ad hoc assumptions can be done . thus some clever intuition is needed to overcome this technical difficulty . one economic way to short - cut the formidable - looking obstacles is the br scaling introduced by brown and rho @xcite . clearly this can not be the only way but it has not yet met with contradictions while having a variety of success . the basic idea is that nuclear matter at its equilibrium density represents the ground state of the matter and that at that point , the nucleons and mesons are quasiparticles as they are at zero density . we know from migdal s work that nucleons in nuclear matter are quasiparticles in the sense of landau fermi liquid theory @xcite . in modern language , this means that the nucleon mass and quasiparticle interactions are fixed - point quantities . we go one step further and assume that mesons are also quasiparticles at the equilibrium point of nuclear matter . this may sound absurd as one knows that mesons interact strongly with many inelastic channels open , so the notion of quasiparticles with well - defined _ effective _ mass for them does not sound right . but then before the advent of shell model one used to say the same thing about the nucleons and yet nucleons in nuclear matter are bona - fide quasiparticles . skeptics will then argue that the quasiparticle notion works for the nucleons because of pauli exclusion principle but for bosons , there is no such thing . this is a valid objection to which no one can at present offer satisfactory answers . so assuming that bosons in medium can be treated in the tree approximation with a point - like structure is an assumption yet to be tested . i shall simply proceed to use the notion in a variety of processes until it meets contradictions . so far there is no evidence that this notion conflicts with nature . the next step then is to construct an effective lagrangian theory which preserves the known symmetries of qcd . in principle , this lagrangian should be usable for a systematic calculation of the type described above for two - nucleon interactions and indeed for particles near on - shell " in medium , such higher - order calculations must be done to describe the needed amplitudes . i shall focus , however , on processes which are off - shell and hence can be treated in the tree order . what are then the relevant degrees of freedom ? we assume that goldstone theorem is applicable in medium , that is , there are zero - mass goldstone ( in the chiral limit ) or light - mass pseudo - goldstone ( in nature ) particles @xmath31 , massive nucleons @xmath32 appearing as matter fields , vector bosons @xmath33 and possibly scalars @xmath34 etc . we assume that such classic low - energy theorems " as goldberger - treiman , adler - weisberger , gell - mann - oakes - renner , kawarabayashi - suzuki - ryazuddin - fyyazuddin ... relations hold in the medium but with masses given by ( ) ^n.[br ] here the star denotes in - medium quantity and @xmath35 stands for the quark condensate in the vacuum ( the starred quantity being the same in the in - medium vacuum . " ) the index @xmath21 depends on models . empirically it is close to @xmath36 . the quark condensate is believed to vanish ( in the chiral limit ) at some critical density @xmath37 ( this is more or less supported by models but lattice calculations are not yet available ) corresponding to chiral phase transition , so one may think of the mass as an indicator for chiral properties of dense matter . this is currently a hot topic with considerable controversy . the first question one must answer for the viability of the theory is : can this theory describe nuclear matter correctly ? if the answer were no , then the theory should be abandoned . the answer comes out to be yes : the theory that gives the correct properties of nuclear matter is a walecka mean - field - type theory with the nucleon @xmath32 , isoscalar - vector @xmath38 , and scalar @xmath34 fields that are coupled linearly . the lagrangian is of the form of a linear walecka model @xcite with , however , the masses of the particles scaling a la br ( [ br ] ) . it has been thought for some time that such a theory with the br scaling would not give a stable system , not to mention a correct binding energy , saturation density @xmath39 and compression modulus @xmath40 . but this ( thinking ) turns out not to be correct . in a seoul national university phd thesis work that succeeds to resolve some of the long - standing problems , chaejun song ( presently a post - doc at suny , stony brook ) with help from his senior collaborators @xcite has shown that to correctly interpret the theory , it is essential to express the density dependence of the masses in terms of certain chiral - invariant fermion bilinears . the so - called rearrangement terms " do come out correctly for the resulting equation of state . all properties of nuclear matter are found to be satisfactorily reproduced while maintaining consistency with all thermodynamic properties . this gives the assurance that one should be able to make small fluctuations around the ground state using the effective lagrangian , the equilibrium minimum representing the fixed point . there are some experimental data already available in the literature that we can use to test certain aspect of the br scaling in fluctuations around the equilibrium density . a number of cases are available but i shall pick two here for illustration . * the first is the anomalous " orbital gyromagnetic ratio @xmath41 in heavy nuclei . the orbital gyromagnetic ratio @xmath42 is the coefficient figuring in the convection current for a nucleon sitting on the fermi surface responding to slowly varying electromagnetic field : = g_l ( e / m_n).[j ] because of the many - body interactions , @xmath42 has an anomalous term @xmath43 , g_l=+g_l.[gl ] note that the current ( [ j ] ) carries the bare " mass @xmath44 , not the effective quasiparticle ( or landau ) mass @xmath45 . thus the first term of ( [ gl ] ) correctly describes charge conservation . this is the analog to kohn s theorem in electronic systems @xcite and makes the calculation of @xmath41 a highly constrained one providing a stringent consistency condition . indeed the mapping " of the br scaling lagrangian theory to landau fermi - liquid fixed point theory gives a highly non - trivial result as shown in @xcite : g_l=49 [ ^-1 - 1 - 12 _ 1^]_3 where @xmath46 is as given in ( [ br ] ) and @xmath47 is the pion contribution to the landau @xmath48 parameter which is completely determined by chiral lagrangian . at the nuclear matter density @xmath49 , both @xmath46 and @xmath47 are known numerically , ( _ 0)= 0.78 , _ 1^(_0)=-0.459 . since heavy nuclei must have @xmath50 , the prediction in heavy nuclei is g_l 0.23_3 . this prediction is consistent with the experimental data @xmath51 extracted from giant resonances in the lead region . it is also consistent with magnetic moments in the lead region . * another case that provides support for the scheme is the axial charge transitions in heavy nuclei @xmath52 with change of one unit of isospin . as shown in the phd thesis ( at seoul national university ) of tae - sun park @xcite , this particular transition is highly enhanced in nuclei ( by a factor of @xmath29 2 with respect to the single - particle strength ) due to one soft pion being exchanged between two nucleons that are involved in the response to the axial charge operator . in heavy nuclei , this is further enhanced because of the scaling @xmath53 at nuclear matter density @xcite which can be seen as one goes up in mass number . experiments are available in medium and heavy nuclei where this enhancement has been seen and confirmed unambiguously @xcite . up to here , i have indicated how the br scaling works for the nucleon mass and , indirectly , for the pion decay constant . so far the scaling for the meson masses has not been tested although it figures indirectly in both cases discussed above . heavy - ion collisions could provide a qualitative test of the behavior of mesons in dense medium . heavy - ion collisions produce hot and dense matter . the physics of the process must therefore be able to probe the behavior of the relevant degrees of freedom which are mainly mesonic at high temperature and/or high density . this is a big area and much debate has been going on . a recent review can be found in @xcite . in a recent phd thesis work , youngman kim of hanyang university ( who is currently a post - doc at suny , stony brook and university of south carolina , columbia ) has shown that both scalar and vector mesons in hot and dense medium do indeed scale a la br scaling if trace anomaly and hidden gauge symmetry of qcd are properly taken into account @xcite . without going into the details of the dynamics involved which goes out of scope for this note , it is difficult to be precise about what we are dealing with . so let me be glib about it and just show the result and give some ( perhaps biased ) remarks . in fig.([ceres ] ) is given the cern ( ceres ) experiment for @xmath54 gev per nucleon central collision of @xmath55 on @xmath56 producing dileptons measured at cern . the differential cross section is plotted vs. invariant mass @xmath14 . now if one takes the masses of the particles involved in the collision to be those of free - space particles , workers in the field more or less agree on the predicted cross sections . despite the large error bars for the experiments , one can see that the free - mass description largely under - estimates the cross section for around @xmath57 mev . as shown by li , ko and brown @xcite , the data can be explained quite economically with the br - scaling masses , the primary agent for this being the dropping mass of the @xmath58 meson which plays the principal role in the dilepton process . a similar fit is obtained in the @xmath59-on-@xmath56 process . unfortunately , this simple picture is blurred by a controversy on the precise cause for the shift of the peak in the dilepton data . it appears at present that this br - scaling mechanism is not the only one that could explain the data . there are various other ( alternative ? ) explanations such as the increased width of the vector meson @xcite thereby possibly invalidating the quasiparticle interpretation or nonperturbative quark - gluon plasma effect @xcite etc . whether or not all these alternatives represent different physical phenomena is not known . in any event , it would be premature to conclude that the quasiparticle description a la br is invalidated . i would say that whether or not a quasiparticle picture for hadrons is applicable in dense medium is an open question that can not be settled by a few - order calculation in a strong - coupling situation . in fact in condensed matter physics , there are cases where low - order treatments fail to give the correct fermi liquid structure , the latter resulting only when all - order calculations are performed . for an example , see @xcite . let me now turn to fluctuations around the ground state of nuclear matter in other flavor directions than the up- and down - quark flavors , say , in the strangeness flavor direction . for this we need to extend the flavor space to @xmath60 . we know how to do this . a kaon propagating in dense medium interacts with the background with its mass and coupling modified by the medium . the coupling with the matter field ( nucleons ) is given by @xmath61 where @xmath62 is the in - medium pseudo - goldstone boson ( pion to the leading order ) decay constant which scales with density . by ( [ br ] ) , the scaling coupling can be related to the scaling in the masses of the mesons that are exchanged between kaon and nucleon in a description where heavy mesons are explicitly accounted for . the masses for @xmath63 so predicted are plotted in fig.([kaonmass ] ) . actually what is calculated here is the kaon dispersion relation with inputs from experiments which is equivalent , to leading order , to using a br - scaled chiral lagrangian in the tree order . this and related matters are discussed in the phd thesis work @xcite of chang - hwan lee at seoul national university ( presently at suny , stony brook ) . this prediction has been beautifully checked by experiments @xcite as shown in fig.([exp - th ] ) . an important consequence of the dropping @xmath64 mass is its effect on the equation of state for dense neutron matter and consequently on neutron stars . this is the link between hadron physics and astrophysics that has been emphasized by gerald e. brown and chang - hwan lee . the basic idea is quite simple . in dense medium , the @xmath64 mass drops continuously as density increases . but as density increases , the electron chemical potential @xmath65 in neutron - star matter increases . when the effective kaon mass crosses the electron chemical potential , the electrons can turn into kaons which by nature of their bosonic character can condense as first suggested by kaplan and nelson @xcite . this is shown schematically in fig.([kcon ] ) . the onset of kaon condensation is expected to soften the equation of state of the matter . the intricate interplay between this phenomenon , hadronic interactions and gravitational interactions has been worked out by li , lee and brown @xcite . their finding is that the maximum mass of neutron stars can be lowered by about 0.4@xmath66 , once kaon condensation as constrained by the dropping kaon mass reflected in the empirical dispersion relation is introduced . the authors point to the growing interplay between hadron physics , relativistic heavy - ion physics and the physics of compact objects in astrophysics . " this may provide a natural explanation ( hopefully without fine - tuning ) for the observation that the well - measured neutron - star masses fall within a narrow window @xmath67 ( see fig.[nstarmass ] @xcite ) . ) the last topic i would like to discuss is infinite nuclear matter at large density , a density much greater that considered above . this may be relevant to neutron star cooling although at present it is not clear whether at the density appropriate for neutron star matter other processes can not compete with it @xcite . the old lore that at an asymptotic density , the matter can be described by perturbative qcd with weakly interacting boring quarks is now widely recognized to be simply wrong . what may be happening at super - high density is something a lot more intriguing and exciting than previously thought . this explains a flurry of activity throughout the world , including the present intense activity at korea institute for advanced study kias ) in which i have been participating . what is most surprising and in some sense unexpected is that at high density the cheshire cat picture re - emerges ! in fact , at high density , there ceases to be any real distinction between quarks and hadrons . this can be best seen in terms of a quark soliton analogous to the qualiton kaplan @xcite introduced as a model for the constituent quark . the mechanism i will discuss exploits that at high density diquarks condense giving rise to color superconductivity as proposed some years ago @xcite and recently revived @xcite . since the resulting qualiton is formed from a color superconducting ground state , it seems proper to call it superqualiton @xcite . it has been argued on general symmetry and dynamical grounds @xcite that at high density , hadronic matter of flavor @xmath60 is characterized by the condensate = - = ^ij^abi_i [ 1 ] where @xmath68 is some constant , @xmath69 are @xmath70 indices , @xmath71 are color indices , and @xmath72 are flavor indices . equation ( [ 1 ] ) holds for parity - even states . such a condensate locks color and flavor so that global color and chiral symmetry are broken to the diagonal subgroup @xmath73 . the consequence of this is that there is an invariant @xmath74 subgroup that contains a twisted " photon , measured with which all excitations carry integer charges reminiscent of the han - nambu quarks and have quantum numbers that correspond to those of the mesons and baryons present at zero density . there is then a continuity between the excitations at high density in terms of quarks and gluons and hadronic excitations at low density in terms of baryons and mesons . this clearly is a case of cheshire cat . now to see that this is the cheshire cat in the sense formulated in terms of the chiral bag @xcite , consider the excitation of a quark on top of the diquark - condensed vacuum . " in @xcite , such a quark is argued to behave like a baryon . now i claim that this quark is a quark soliton , i.e. , superqualiton @xcite . to describe the low - energy dynamics of the color - flavor locking phase , introduce a field @xmath75 which maps space - time to the coset space , @xmath76 . one can take it to be @xmath77 where @xmath78 is the anomalous dimension of the diquark field of order @xmath79 and @xmath80 denotes the quark field with momentum close to a fermi momentum @xmath81 . the pairing involves quarks near the opposite edge of the fermi surface . similarly , we introduce a right - handed field @xmath82 , also a map from space - time to @xmath83 , to describe the excitations of the right - handed diquark condensate . if this field takes a vacuum expectation value as a consequence of the diquark condensation which will , owing to ( [ 1 ] ) , have the form = -= _ a , then 16 nambu - goldstone bosons will get excited . eight of them will get eaten up by the gluons to give masses to the gluons . the massive gluons then turn into massive vector mesons whose quantum numbers are those of the light - quark vector mesons present at zero density . the remaining eight ( pseudoscalar ) nambu - goldstone bosons are the equivalents of the ones present at zero density and are represented by the interpolating field ( [ u ] ) . in analogy to the usual skyrmion at zero density , this field supports a soliton which is a fermion , the quantum numbers of which are identical to those of the usual baryon . the effective lagrangian that gives rise to this soliton should in principle be derived from qcd . at the moment such an effective lagrangian is not known . however one can venture to make a few interesting conjectures . viewed as a superqualiton whose mass is given by the soliton mass , there is nothing that requires that the soliton mass be equal to or near the superconductivity gap @xmath84 ( which is dictated by the condensate ) . in fact there is nothing which would prevent the mass from being much less than the gap . thus one could imagine that light fermions are excited _ within _ the gap . correlations between light superqualitons could rearrange the ground state into a different form from that of the standard superconductivity . there could be other modes of similar nature such as particle - hole excitations from the opposite ends of the fermi sea ( much like the cooper pair but involving particles and holes ) which could give rise to a crystalline structure etc . @xcite . for this and other reasons the phenomenon of color superconductivity in qcd at high density could be completely different from the usual bcs superconductivity . the issue of how this matter could influence the structure of compact stars ( e.g. , cooling , equation of state etc . ) is an open one actively studied presently in kias @xcite . the most important outcome of the recent development of eft in nuclear physics is that the highly successful approach to nuclear structure using realistic nuclear potentials ( pm ) is rendered a first - principle interpretation in that it represents the leading term in the eft expansion with the corrections thereof systematically calculable . this confers the power of modern field theory techniques to the standard nuclear physics approach that has been practiced with success since a long time . this bridging " comes about thanks to a possible duality that i refer to as cheshire cat principle between qcd variables and macroscopic ( color - singlet ) variables . this also provides a potential link between the physics of the elementary nucleon , nuclei , hadronic matter and compact - star matter . in the case of high density , the picture becomes even more intriguing . there we see emerging the symbolic ( approximate ) equality quark"qualiton"baryon " . it is amusing that the notion of the cheshire cat which was conceived by the need to reconcile the traditional meson - exchange description with the modern qcd description for nuclear processes @xcite ( i.e. , the little bag " with pion cloud , chiral bag etc ) at low density re - emerges at high density where one would have expected the bona - fide qcd to be uniquely applicable . most significant of all , the interplay between hadronic physics , relativistic heavy - ion physics and the physics of compact stars in astrophysics highlights the unity of physics , an endeavor that could make a mainstream of physics research in asia pacific research centers like apctp and in korean institutes like kias . such a potential seems only natural given that some of the most significant contributions have been , and are being , made by young korean theorists graduate students and post - docs actively working in the field and not less significantly that several experimental collaborations between korean experimenters and the alice ( cern ) and rhic ( brookhaven ) project teams purporting to probe the hot and dense matter relevant to early universe and compact stars are in the process of being formed . i am very grateful to chang - hwan lee for his invaluable help in preparing this note and for useful comments . i would like to thank also all those who participated this year in the working group on various aspects of dense hadronic matter , qcd and astrophysical consequences at kias , in particular , deog ki hong , steve hsu , kuniharu kubodera , kurt langfeld , hyun kyu lee , dong - pil min , byung - yoon park , maciek nowak , sang - jin sin , vicente vento and ismail zahed . 99 ed . by g.e . brown , c .- h . lee , h.k . lee and d .- min ( world scientific , singapore , 1999 ) m.a . nowak , m. rho and i. zahed , _ chiral nuclear dynamics_(world scientific , singapore , 1996 ) a. hosaka and h. toki , phys . 277 * , 65 ( 1996 ) . s. nadkarni , h.b . nielsen and i. zahed , nucl . phys . * b253 * , 308 ( 1984 ) ; m. rho , phys . rep . * 240 * , 1 ( 1994 ) . skyrme , nucl . phys . * 31 * , 556 ( 1961 ) e. witten , nucl . phys . * b223 * , 422 , 433 ( 1983 ) h.b . nielsen , m. rho , a. wirzba and i. zahed , phys . b269 * , 389 ( 1991 ) ; phys . lett . * 281 * , 345 ( 1992 ) . min , b .- y . park , m. rho and v. vento , the proton spin in the chiral bag model : casimir contribution and cheshire cat principle , " nucl . phys . * a * , in press , hep - ph/9810539 s. weinberg , ; ; m. rho , ; t .- s . park , d .- min and m. rho , ; ; c. ordonez and u. van kolck , ; c. ordonez , l. ray and u. van kolck , ; ; u. van kolck , d.b . kaplan , m. savage and m. wise , ; ; nucl - th/9801034 ; nucl - th/9802075 ; nucl - th/9804032 t .- s . park , k. kubodera , d .- min and m. rho , ; 1998 and 1999 joint caltech - int workshops : _ nuclear physics with effective field theory_. m. rho , nucl - th/9812012 . lepage , nucl - th/9706029 d.r . phillips and t.d . cohen , nucl - th/9906091 . hyun , d .- min and t .- s . park , nucl - th/9908073 . park , k. kubodera , d .- min and m. rho , astrophys . j. * 507 * , 443 ( 1998 ) . park , k. kubodera , d .- p . min and m. rho , nucl - th/9904053 , nucl - th/9906005 . w . chen , g. rupak and m.j . savage , nucl - th/9905002 . m. rho , ann . nucl . part . sci . * 34 * , 531 ( 1984 ) b. frois and j.f . mathiot , comments on nucl . phys . * 18 * , 307 ( 1989 ) . brown and m. rho , . migdal , _ theory of finite fermi systems and applications to atomic nuclei _ ( interscience , new york , 1967 ) . j.d . walecka , ann . phys . * 83 * , 491 ( 1974 ) . c. song , d .- min and m. rho , phys . b424 * , 226 ( 1998 ) . w. kohn , phys . rev . * 123 * , 1242 ( 1961 ) . b. friman and m. rho , nucle . phys . * a606 * , 303 ( 1996 ) . park , d .- min and m. rho , phys . rep . * 233 * , 341 ( 1993 ) . k. kubodera and m. rho , phys . lett . * 67 * , 3479 ( 1991 ) . warburton , phys . * 66 * , 1823 ( 1991 ) . r. rapp and j. wambach , hep - ph/9909229 . y. kim , hanyang university thesis ( august 1999 ) ; y. kim , h.k . lee and m. rho , phys . rev . * c52 * , r1184 ( 1995 ) ; y. kim and h.k . lee , phys . rev . * c55 * , 3100 ( 1997 ) ; y. kim and h.k . lee , hep - ph/9905268 . g. agakichiev et al . ( ceres collaboration ) , phys . lett . * 75 * , 1272 ( 1995 ) . ko and g.e . brown , nucl . phys . * a606 * , 568 ( 1996 ) ; g.q . brown and h. sorge , nucl phys . * a611 * , 539 ( 1996 ) . lee , h. yamagishi and i. zahed , phys . rev . * c58 * , 2899 ( 1998 ) ; c .- h . lee , j. wirrstam , i. zahed and t.h . hansson , phys . lett . * b448 * , 168 ( 1999 ) . e. arrigoni , con - mat/9903431 , phys . , in press . lee , phys . rep . * 275 * , 255 ( 1996 ) . li , c .- h . lee and g.e . brown , phys . * 79 * , 5214 ( 1997 ) ; nucl . a625 , 372 ( 1997 ) . d.b kaplan and a.e . nelson , phys . lett . * b175 * , 57 ( 1986 ) . r. barth et al , phys . lett . * 778 * , 4027 ( 1997 ) . s.e . thorsett and d. chakrabarty , astro . j. * 512 * , 288 ( 1999 ) . k. langfeld and m. rho , hep - ph/9811227 b .- y . park , m. rho , a. wirzba and i. zahed , dense qcd : overhauser or bcs pairing ? " to appear in hep - ph , kias - p99047 . kaplan , phys . * b235 * , 163 ( 1990 ) ; nucl . phys . * b351 * , 137 ( 1991 ) . d. bailin and a. love , phys . rep . * 107 * , 325 ( 1984 ) . recently brought to limelight , in particular , by f. wilczek , panic 99 , uppsala , june 1999 , hep - ph/9908480 d.k . hong , m. rho and i. zahed , qualitons at high density , " hep - ph/9906551 , to appear in phys . b. t. schfer and f. wilczek , phys . 82 * , 3956 ( 1999 ) , hep - ph/9811473 ; hep - ph/9903503 . hong , h.k . lee , m.a . nowak and m. rho , neutrino interactions in dense color - flavor - locked matter , " to appear . brown and m. rho , phys . lett . * b82 * , 177 ( 1979 ) .	the effort currently in vogue in some small circle of physicists to go from a proton to nuclei to compact stars involves various aspects of particle and nuclear physics that require input from laboratory experiments , transcending narrow specialization in diverse sub - fields . several topics on this matter are discussed in this note . the notion of cheshire cat principle is introduced for nucleons , nuclei and dense hadronic matter and is confronted with experimental data on the proton , two - nucleon systems and heavy - ion experiments , with a leaping extrapolation to the structure of neutron stars . the matter discussed here illustrates that a close contact with experiments , indispensable in the present case , is essential for significant progress in _ any _ field of physics . i suggest that this is a field that has a tremendous potential for breakthrough in the asia pacific countries , particularly in korea , where some of the seminal works in this area have been done by young theorists , both pre- and post - graduate . -1 cm .25 in # 1#2#3#4 # 1 * # 2 * , # 3 ( # 4 ) # 1 # 1\| # 1 \| # 1 0.4 cm * selected topics in astro - hadron physics : * 0.2 cm * going from a proton to nuclei to neutron stars * 1.5 cm mannque rho 0.5 cm _ service de physique thorique , ce saclay , 91191 gif - sur - yvette , france _ e - mail : [email protected] and _ school of physics , korea institute for advanced study , seoul 130 - 012 , korea _ 0.5 cm
after the discovery of a new boson with a mass around 125 gev at the large hadron collider ( lhc ) @xcite , there have been numerous studies attempting to pin down its properties , namely , its spin - parity and its couplings to standard model ( sm ) particles @xcite . the bosonic decay modes of this particle have been analyzed with greater precision than the fermionic modes by both the atlas and cms collaborations , since the latter requires much more statistics and possibly the application of new search strategies . the signal - strengths reported in various channels by the experiments are broadly consistent with the sm predictions within about two standard deviations , and a preliminary analysis of spin - parity using the @xmath1 channel suggests that a cp - even scalar hypothesis is favoured over other alternatives @xcite . therefore , the accumulating evidence is in favour of an sm - like ( @xmath2 ) higgs boson , and we are going to assume so henceforth . global fits of the higgs data have been used by both experimentalists and theorists to derive bounds on possible deviations from the sm . such deviations in the higgs couplings can be parametrized either by including a multiplicative ( or additive ) constant to the sm coupling , or by including new lorentz structures not present in the renormalizable sm lagrangian . in the framework of the sm as an effective field theory valid below a cut - off scale @xmath3 , higher - dimensional operators involving the sm fields and invariant under the sm gauge group can be used to capture possible new physics effects . a complete list of such operators has been written down in ref . @xcite , while a minimal basis has been obtained rather recently in ref . such an approach is valid as long as there is no new light degree of freedom coupled to the sm sector below the scale @xmath3 . null results in lhc searches for new particles provide some motivation for this approach , although the presence of new particles charged under the sm gauge group is still viable even with masses around the electroweak symmetry breaking scale . henceforth , we assume that there is no such new state , and work with the sm lagrangian supplemented with dimension-6 operators involving sm gauge bosons and the sm higgs doublet . as is well - known , although there exists one possible dimension-5 operator , it plays a role only in the generation of neutrino masses . electroweak precision measurements constrain the overall strength of the operators involving sm electroweak gauge bosons @xcite . however , such constraints come from one - loop contributions of these operators to the self - energy diagrams of the gauge - bosons , parametrized in terms of the so - called oblique corrections @xcite . in contrast , the higgs boson couplings to w , z or photon pairs can be affected at the tree level itself by a class of such operators , and therefore , it is possible to impose stronger constraints on their co - efficients using the already accumulated lhc higgs data . this fact has been observed in several studies performing global fits to the higgs data , and deriving limits on the operator co - efficients @xcite . however , in most cases , these studies make an important assumption , namely , that the efficiencies of experimental cuts used for various final states are the same as the corresponding efficiencies in the sm case . to understand where the efficiency of experimental cuts enter the global fits , let us recall that the global fits are performed by comparing the experimentally observed signal strength ( @xmath4 ) in a channel @xmath5 with the corresponding signal strength predicted by a particular framework beyond the standard model ( bsm ) , @xmath6 , which is defined as @xmath7_{\rm bsm}}{\left[\sigma(pp \rightarrow h)\times { \rm br } ( h \rightarrow x \bar{x } ) \times \epsilon_{x \bar{x}}\right]_{\rm sm } } , \label{mu}\ ] ] where , @xmath8 denotes the efficiency of the experimental cuts applied to select a particular final state . although the assumption that @xmath9 can be justified if the higgs couplings only receive a multiplicative modification to the sm one , it is not _ a priori _ clear whether such an assumption holds after the inclusion of dimension-6 operators . this is because these operators bring in new lorentz structures to the higgs - gauge boson couplings , which in turn modify the distributions of kinematic variables on which these cuts are imposed . some of these distributions have been used in earlier studies , with special emphasis on the spin - parity determination of the newly discovered particle @xcite . in this paper , we assume @xmath2 for this particle , and investigate how additional interaction terms with gauge boson pairs , gauge invariant and of higher dimension , affect higgs phenomenology . with this in view , we subject the contributions of the additional operators to the cuts used on specific final states . thus we demonstrate through rigorous monte carlo simulations how much the efficiencies can get modified , and to what extent they alter the bounds on the operator co - efficients . we use lhc higgs search studies in the @xmath10 and @xmath11 channels as examples , implement the cuts used by the atlas collaboration in our toy detector simulation , and determine the modified efficiencies for two such dimension-6 operators channel , we refer the readers to ref . @xcite . ] . we also simultaneously allow the modification of the sm coupling to the weak gauge bosons by a multiplicative constant , keeping the custodial @xmath12 symmetry intact . it should be mentioned that generically more than one higher - dimensional operator can be present in the effective low - energy theory with different coupling strengths . in that sense , our study with one operator considered at a time is illustrative , and focuses on the important effect of new lorentz structures in the cut efficiencies . moreover , the method developed here is of general utility in studying all possible higher - dimensional operators . this paper is organized as follows . in section [ sec-2 ] , we describe the higher dimensional operators considered , the modified higgs - gauge boson couplings that they lead to , and constraints on them from electroweak precision tests . in section [ sec-3 ] , we describe the set - up of our monte - carlo simulation , including its validation against the atlas higgs search studies in the @xmath10 channel . the modified decay widths , cross - sections and efficiencies are presented in section [ sec-4 ] , including simple parametrizations of each of these . section [ sec-5 ] is devoted to the re - evaluation of constraints on these operators using a fit to the higgs data in bosonic channels , while in section [ sec-6 ] we study the modified efficiencies in the associated production of higgs . in section [ sec-7 ] we show the modifications to certain kinematic observables in presence of the new operators . we summarize our findings in section [ sec-8 ] . to illustrate the modification of experimental cut - efficiencies on including new lorentz structures in higgs - gauge boson interactions ( henceforth called @xmath13 interactions ) we take the following dimension-6 operators as examples : @xmath14 where , the field strength tensors for the @xmath15 and @xmath16 gauge groups are @xmath17 here , @xmath18 denotes the sm higgs doublet , @xmath19 are the pauli matrices , and @xmath20 , @xmath21 are the @xmath15 , @xmath16 gauge couplings respectively . as mentioned in the introduction , the sm @xmath22 and @xmath23 couplings can also get modified by a multiplicative factor in a general setting , where , for example , the sm higgs doublet is part of a larger scalar sector . we therefore include the possibility of having the @xmath22 and @xmath23 couplings modified by the same factor @xmath24 , assuming custodial invariance . since generically , in presence of an arbitrary number of extra scalar singlets or doublets , @xmath25 @xcite , we scan the range @xmath26 in our analysis . one of the goals of this study is to determine to what extent @xmath24 can be different from its sm value of @xmath27 , while including new dimension-6 operators . taking these two modifications into account , the lagrangian in the higgs sector becomes @xmath28 where , the operators @xmath29 and @xmath30 are given by equation [ ops ] , and the @xmath24 dependence of the sm part comes from the following terms @xmath31 here , @xmath32 denotes the vacuum expectation value of the higgs field , and @xmath33 denotes the higgs boson . the higgs couplings in @xmath34 to fermions and gluons are not modified . since the new higgs couplings generated by @xmath29 and @xmath30 , in particular the new lorentz structures , are crucial to our discussion , we note down the additional @xmath13 interactions generated by the operators in equation [ ops ] @xcite : @xmath35 where , @xmath36 , for @xmath37 , and , @xmath38 we have used the shorthand @xmath39 and @xmath40 , @xmath41 being the weinberg angle . the interaction terms involving the derivatives of the gauge fields bring in momentum dependent vertices , which are responsible for the modified kinematics in higgs boson production via weak - boson fusion and associated production with a @xmath42 or a @xmath43 , as well as the decay of the higgs particle to electroweak gauge boson final states . the kinematics is affected most when the new interactions appear in both the production and decay processes , an example of which we shall discuss in sections [ sec-3 ] and [ sec-4 ] . we should remark here that in @xmath44 , only the new tree - level terms generated due to @xmath29 and @xmath30 have been considered . there will be additional contributions coming from the @xmath42 boson loop ( apart from the sm contribution modified by the inclusion of @xmath24 , which we take into account ) , since now the @xmath22 coupling also involves momentum - dependent terms . however , on naive power - counting in the number of loops and derivatives , these contributions are sub - leading , and the new divergences arising from this loop diagram will be cancelled by the next higher - order terms in the derivative expansion . with the current level of precision in the data , such terms can be safely neglected . finally , though we have noted the new contribution to the @xmath45 vertex for completeness , there is no data in this channel so far , and therefore , the effect of the modification to this channel is sub - dominant . the operators @xmath29 and @xmath30 contribute to the so - called peskin - takeuchi @xmath46 parameters @xcite , and are therefore constrained by electroweak precision data @xcite . following ref . @xcite , the bounds at @xmath47 c.l . , taking one operator at a time , are given by @xmath48 these bounds can change once we include the factor @xmath24 in equation [ sm - beta ] . however , as we shall see , the higgs data puts much stronger constraints on these two operators compared to the precision data , and therefore , we do not consider modifications to the precision constraints in this study . we note in passing that since the inclusion of the higher - dimensional operators in equation [ lag ] modifies the @xmath22 and @xmath23 vertices from their corresponding sm values , this will spoil the unitarity of @xmath49 ( @xmath50 ) scattering amplitudes at high energies . for the values of operator coefficients allowed by the current higgs data , the violation of unitarity appears at energies of a few tev , with the exact value depending upon the specific choice of operators @xcite . since the higher - dimensional operators themselves arise from integrating out heavy fields of mass @xmath51 ( for weakly coupled ultra - violet completions ) , one expects in general that the presence of these new degrees of freedom in the uv completion will eventually restore the unitarity of the full theory at high energies . as noted in the previous section , we expect in general significant modifications to the kinematics in processes where the new lorentz structures appear both in the production and decay vertices . an example of such a process is the production of higgs boson via vbf and its subsequent decay to @xmath10 . the atlas collaboration has presented a detailed analysis of such a scenario in the @xmath53 channel ( @xmath54 ) in ref . @xcite using the @xmath55 tev , @xmath56 data . the response of sm - type interactions to all the individual cuts can be readily checked from this analysis , which thus provides a much needed calibration for the simulation with additional operators . therefore , we use this channel to set - up and validate our monte carlo as well as our detector simulation code . we have used feynrules @xcite to extract the feynman rules from the lagrangian in equation [ lag ] , madgraph-5 @xcite to generate the parton level events , pythia-6 @xcite for parton shower and hadronization , and our own detector simulation code for analyzing the hadron - level events . jet formation and underlying events have been simulated within the pythia framework . .__comparison of the efficiencies of experimental cuts on the signal cross - section in the @xmath57 channel , for the @xmath58 category , demanding different flavour leptons ( @xmath59 ) in the final state . the signal cross - section here refers to the sum of vbf and vh processes . the atlas numbers have been taken from ref . @xcite , for which we show the @xmath60 range ( the central value is written within brackets ) . _ _ [ cols="<,^,^",options="header " , ] as in equation [ eff - def ] , we define the efficiency for this channel as a function of @xmath24 and @xmath61 ( @xmath62 or @xmath63 ) as @xmath64_{\rm after ~cuts}}{\left[\sigma(pp \rightarrow h)_{\rm vh , \rm v \rightarrow j j } \times { \rm br } ( h \rightarrow \gamma \gamma ) \right]_{\rm before ~cuts}}. \label{eff - def2}\end{aligned}\ ] ] by performing a scan over the ( @xmath24,@xmath65 ) parameter space , we obtain the combined efficiencies ( for @xmath30 and @xmath29 ) of the isolation cuts and the atlas cuts in table [ tab : gaga ] and they are well fit by the following functions : @xmath66 @xmath67 in equations [ eff - bb - gaga ] and [ eff - ww - gaga ] , the denominator and the numerator represent @xmath68 , before and after all the cuts respectively , and some common numerical factors between the two , like the total higgs decay width have been cancelled out . in figures [ fig : eff - vs - f - bb ] and [ fig : eff - vs - f - ww ] , we show the variation of @xmath69 as a function of @xmath63 and @xmath62 respectively for different values of @xmath24 . the red ( solid ) , green ( dashed ) and black ( dot - dashed ) curves correspond to @xmath70 respectively . for @xmath71 or @xmath72 we recover the sm efficiency ( @xmath73 ) in both the cases for all values of @xmath24 . within the range of @xmath63 shown in figure [ fig : eff - vs - f - bb ] , the efficiency for @xmath74 can change from its sm value by upto @xmath75 , while for @xmath76 , it can increase by upto @xmath77 . on the other hand , for the range of @xmath62 shown in figure [ fig : eff - vs - f - ww ] , for @xmath78 ( @xmath76 ) , the efficiency can increase from its sm value by upto @xmath79 @xmath80 . thus our overall conclusion is that the modification of cut efficiencies in the vh production mode is in general less pronounced than in the vbf higgs production with @xmath81 . since the modifications of cut efficiencies discussed so far originate from changes in kinematic distributions , in this section , we explore some of these distributions in the presence of higher dimensional operators . study of differential distributions is the next step in experimental analysis of the higgs sector , and preliminary results with the current data have already been presented in ref @xcite . as an example we choose the diphoton channel in the vbf category @xcite , and consider the operator @xmath29 for illustration . all the distributions are shown after applying the standard trigger and isolation cuts for jets and photons . the kinematic variables considered are : * @xmath82 , where @xmath83 and @xmath84 are the two tagged jets ordered in terms of their transverse momenta . * @xmath85 . * the invariant mass of the two tagged jets , @xmath86 . for this as well as the distributions listed below , the cut @xmath87 ( see ref . @xcite ) is imposed . * @xmath88 , where @xmath89 is the transverse thrust , @xmath90 , @xmath91 are the transverse momenta of the two isolated photons and @xmath92 is the transverse momentum of the diphoton system @xcite . this and the subsequent distributions are subjected to the cuts @xmath93 gev and @xmath94 , where @xmath95 is the azimuthal angle separation between the diphoton system and the system of the two tagged jets . the criterion of no hadronic activity in the rapidity gap between the two tagged jets is also imposed @xcite . * @xmath96 , where @xmath97 is the pseudorapidity of the diphoton system . * @xmath98 is the minimal @xmath99 between a photon and a tagged jet . + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) , @xmath106 ( solid red line ) and @xmath107 ( solid green line ) . the cut - off scale chosen is @xmath108 tev.__,title="fig:",width=321,height=321 ] + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] + ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] ( left ) and @xmath100 ( right ) ; middle row : @xmath101 ( left ) and @xmath102 ( right ) ; bottom row : @xmath103 ( left ) and @xmath104 ( right ) , for the parameter points @xmath105 ( shaded blue region ) and @xmath109 ( solid red line ) . the cut - off scale chosen is @xmath108 tev . both the parameter points are allowed by the current data at 95% c.l.__,title="fig:",width=325,height=325 ] + + the last five kinematic variables form a subset of the inputs for the boosted decision tree ( bdt ) employed by atlas for studying this channel @xcite . figure [ fig : dist - b1-f0pm10 ] shows the normalised distributions in the above six variables for @xmath110 and @xmath111 at the @xmath55 tev lhc . the values @xmath112 are chosen for illustrative purpose only , since , as seen in section [ sec-5 ] , although they are allowed by the lhc data in the @xmath113 channel , the current measurement in the @xmath11 channel restricts @xmath62 to smaller values . therefore , in figure [ fig : dist - b1pt8-f0p3 ] we show the aforementioned distributions in the sm , and for the parameters @xmath114 the latter being within the @xmath115 allowed range of the global fit ( see figure [ margin ] ) . for both the above figures , the cut - off scale has been chosen as @xmath108 tev . we note that the distributions of @xmath100 have two peaks . the peak at @xmath116 is due to vh contamination . moreover , the relative heights of the two peaks change on introducing higher dimensional operators . we have considered some illustrative dimension-6 operators for @xmath13 interactions , and their potential contributions to the higgs data , in conjunction with the sm - like operators . parametrizing the strength of the additional interactions by @xmath61 ( @xmath117 ) , and the simultaneous modification to the sm - like couplings to @xmath42 and @xmath43 bosons by @xmath24 , we show , after a detailed cut - based monte carlo analysis , how the efficiencies of different acceptance cuts are altered for various values of @xmath61 and @xmath24 . we find that in general there can be substantial modification of this kind , which underscores the importance of a detailed study of the effect of all such additional operators on the kinematics of various final states . when one further imposes the constraints on the ( @xmath118 ) space as resulting from a global fit of the lhc data available till date , the @xmath61-parameters are in general restricted to rather modest values while @xmath24 is restricted to be in the range @xmath119 $ ] . thus the effects of cuts in the diboson channels may not be drastically different , unless there is ground for relaxing their constraints . in general , the @xmath120 channel is more sensitive to the higher - dimensional operators than the gluon - fusion and @xmath121 production modes . we also present several kinematic distributions , some of which are used in bdt analyses , which can potentially bring out signatures of the new operators , even with moderate strength . it should be remembered here that our analysis is purely phenomenological and data - driven ; the assumption of any specific ultraviolet ( uv ) completion is deliberately avoided . in a specific uv completion scheme , more than one higher - dimensional operator relevant at the lhc scale may be generated , which can affect some of our conclusions . for example , with additional operators present , a situation as restrictive as indicated by figure [ fig : gaga - ww - total ] may not arise due to the accidental cancellation of different contributions . however , studying one operator at a time gives us valuable insight on how it typically affects various observables in the higgs sector an insight that is lost in the introduction of all operators of a given dimension simultaneously . in this spirit , we have explored two operators which can most significantly modify the interaction of the higgs with a pair of electroweak gauge bosons . we thank atri bhattacharya , sanjoy biswas , ushoshi maitra and tanumoy mandal for discussions and technical help . useful comments from bruce mellado are also acknowledged . the work of sm is supported by world premier international research center initiative ( wpi initiative ) , mext , japan . sb and bm acknowledge the funding available from the department of atomic energy , government of india , for the regional centre for accelerator - based particle physics , harish - chandra research institute . computational work for this study was partially carried out at the cluster computing facility at the harish - chandra research institute ( http://cluster.hri.res.in ) . g. aad _ et al . _ [ atlas collaboration ] , phys . b * 716 * ( 2012 ) 1 . s. chatrchyan _ et al . _ [ cms collaboration ] , phys . b * 716 * ( 2012 ) 30 . for similar studies in this direction before the higgs discovery , see , + f. bonnet , m. b. gavela , t. ota and w. winter , phys . d * 85 * , 035016 ( 2012 ) ; j. r. espinosa , c. grojean , m. muhlleitner and m. trott , jhep * 1205 * , 097 ( 2012 ) p. p. giardino , k. kannike , m. raidal and a. strumia , jhep * 1206 * , 117 ( 2012 ) ; t. li , x. wan , y. -k . wang and s. -h . zhu , jhep * 1209 * , 086 ( 2012 ) ; m. rauch , arxiv:1203.6826 [ hep - ph ] ; j. r. espinosa , m. muhlleitner , c. grojean and m. trott , jhep * 1209 * , 126 ( 2012 ) ; j. ellis and t. you , jhep * 1206 * ( 2012 ) 140 ; d. carmi , a. falkowski , e. kuflik and t. volansky , jhep * 1207 * , 136 ( 2012 ) ; m. duhrssen , s. heinemeyer , h. logan , d. rainwater , g. weiglein and d. zeppenfeld , phys . d * 70 * ( 2004 ) 113009 ; r. lafaye , t. plehn , m. rauch , d. zerwas and m. duhrssen , jhep * 0908 * ( 2009 ) 009 ; m. klute , r. lafaye , t. plehn , m. rauch and d. zerwas , phys . rev . lett . * 109 * , 101801 ( 2012 ) ; a. azatov , r. contino , d. del re , j. galloway , m. grassi and s. rahatlou , jhep * 1206 * , 134 ( 2012 ) . i. low , j. lykken and g. shaughnessy , phys . d * 86 * , 093012 ( 2012 ) ; t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , phys . d * 86 * , 075013 ( 2012 ) ; p. p. giardino , k. kannike , m. raidal and a. strumia , phys . lett . b * 718 * , 469 ( 2012 ) ; j. baglio , a. djouadi and r. m. godbole , phys . b * 716 * , 203 ( 2012 ) ; j. ellis and t. you , jhep * 1209 * , 123 ( 2012 ) ; m. montull and f. riva , jhep * 1211 * , 018 ( 2012 ) ; j. r. espinosa , c. grojean , m. muhlleitner and m. trott , jhep * 1212 * , 045 ( 2012 ) ; d. carmi , a. falkowski , e. kuflik , t. volansky and j. zupan , jhep * 1210 * ( 2012 ) 196 ; s. banerjee , s. mukhopadhyay and b. mukhopadhyaya , jhep * 1210 * , 062 ( 2012 ) ; f. bonnet , t. ota , m. rauch and w. winter , phys . rev . d * 86 * , 093014 ( 2012 ) ; t. plehn and m. rauch , europhys . lett . * 100 * , 11002 ( 2012 ) ; a. djouadi , arxiv:1208.3436 [ hep - ph ] ; b. batell , s. gori and l. -t . wang , jhep * 1301 * ( 2013 ) 139 ; g. moreau , phys . rev . d * 87 * , 015027 ( 2013 ) ; g. bhattacharyya , d. das and p. b. pal , phys . d * 87 * , 011702 ( 2013 ) ; d. choudhury , r. islam and a. kundu , arxiv:1212.4652 [ hep - ph ] . g. belanger , b. dumont , u. ellwanger , j. f. gunion and s. kraml , jhep * 1302 * , 053 ( 2013 ) . m. klute , r. lafaye , t. plehn , m. rauch and d. zerwas , europhys . * 101 * ( 2013 ) 51001 ; c. grojean , e. e. jenkins , a. v. manohar and m. trott , jhep * 1304 * , 016 ( 2013 ) ; k. cheung , j. s. lee and p. -y . tseng , jhep * 1305 * , 134 ( 2013 ) ; j. elias - mir , j. r. espinosa , e. masso and a. pomarol , jhep * 1308 * , 033 ( 2013 ) ; j. ellis , v. sanz and t. you , arxiv:1303.0208 [ hep - ph ] ; p. p. giardino , k. kannike , i. masina , m. raidal and a. strumia , arxiv:1303.3570 [ hep - ph ] ; r. contino , m. ghezzi , c. grojean , m. muhlleitner and m. spira , jhep * 1307 * , 035 ( 2013 ) ; j. ellis and t. you , jhep * 1306 * , 103 ( 2013 ) ; a. djouadi and g. moreau , arxiv:1303.6591 [ hep - ph ] ; w. -f . chang , w. -p . pan and f. xu , phys . d * 88 * , 033004 ( 2013 ) ; t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , arxiv:1304.1151 [ hep - ph ] ; b. dumont , s. fichet and g. von gersdorff , jhep * 1307 * , 065 ( 2013 ) ; j. elias - miro , j. r. espinosa , e. masso and a. pomarol , arxiv:1308.1879 [ hep - ph ] ; m. b. einhorn and j. wudka , arxiv:1308.2255 [ hep - ph ] ; a. pomarol and f. riva , arxiv:1308.2803 [ hep - ph ] . talk given by michael duehrssen on behalf of the atlas collaboration , https://indico.cern.ch/getfile.py/access?contribid=231&sessionid=15&resid=0&materialid=slides&confid=218030 ; g. aad _ et al . _ [ atlas collaboration ] , arxiv:1307.1432 [ hep - ex ] . g. aad _ et al . _ [ atlas collaboration ] , arxiv:1307.1427 [ hep - ex ] . talk given by josh bendavid on behalf of the cms collaboration , + https://indico.cern.ch/getfile.py/access?contribid=430&sessionid=15&resid=0&materialid=slides&confid=218030 ; w. buchmuller and d. wyler , nucl . b * 268 * , 621 ( 1986 ) . b. grzadkowski , m. iskrzynski , m. misiak and j. rosiek , jhep * 1010 * , 085 ( 2010 ) . k. hagiwara , r. szalapski and d. zeppenfeld , phys . b * 318 * , 155 ( 1993 ) . m. c. gonzalez - garcia , int . j. mod . a * 14 * , 3121 ( 1999 ) . m. e. peskin and t. takeuchi , phys . lett . * 65 * , 964 ( 1990 ) ; m. e. peskin and t. takeuchi , phys . d * 46 * , 381 ( 1992 ) . g. altarelli , r. barbieri and f. caravaglios , nucl . b * 405 * , 3 ( 1993 ) . e. masso and v. sanz , phys . d * 87 * , no . 3 , 033001 ( 2013 ) . t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , phys . d * 87 * , 015022 ( 2013 ) ; t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , arxiv:1306.0006 [ hep - ph ] . a. falkowski , f. riva and a. urbano , arxiv:1303.1812 [ hep - ph ] . t. plehn , d. l. rainwater and d. zeppenfeld , phys . lett . * 88 * , 051801 ( 2002 ) . a. djouadi , r. m. godbole , b. mellado and k. mohan , phys . b * 723 * , 307 ( 2013 ) . y. gao , a. v. gritsan , z. guo , k. melnikov , m. schulze and n. v. tran , phys . d * 81 * , 075022 ( 2010 ) ; p. avery , d. bourilkov , m. chen , t. cheng , a. drozdetskiy , j. s. gainer , a. korytov and k. t. matchev _ et al . _ , phys . d * 87 * , 055006 ( 2013 ) . j. s. gainer , j. lykken , k. t. matchev , s. mrenna and m. park , phys . lett . * 111 * , 041801 ( 2013 ) . j. f. gunion , h. e. haber and j. wudka , phys . rev . d * 43 * , 904 ( 1991 ) . d. choudhury , r. islam and a. kundu , phys . d * 88 * , 013014 ( 2013 ) ; m. dahiya , s. dutta and r. islam , arxiv:1311.4523 [ hep - ph ] . n. d. christensen and c. duhr , comput . commun . * 180 * , 1614 ( 2009 ) . j. alwall , m. herquet , f. maltoni , o. mattelaer and t. stelzer , jhep * 1106 * , 128 ( 2011 ) . t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) . a. djouadi , phys . rept . * 457 * , 1 ( 2008 ) . lhc higgs working group , + https://twiki.cern.ch/twiki/bin/view/lhcphysics/crosssections	the inclusion of higher - dimensional gauge invariant operators induces new lorentz structures in higgs couplings with electroweak gauge boson pairs . this in principle affects the kinematics of higgs production and decay , thereby modifying the efficiencies of the experimental cuts compared to what simulations based on the standard model interactions yield . taking some sample cases , we perform a rigorous analysis of how the efficiencies differ for various strengths of the additional operator vis - a - vis the standard model interactions , scanning over the values of both of them . while the response to cuts can be markedly different in some regions , we find that the sensitivity to new operator structures is relatively limited , so long as we remain confined to the 2@xmath0 regions around the best fit signal strengths measured at the large hadron collider . we also show modifications to certain kinematical distributions including the new operators in the diphoton final state . ipmu13 - 0160 + recapp - hri-2013 - 018 * higher dimensional operators and lhc higgs data : + the role of modified kinematics * + shankha banerjee , satyanarayan mukhopadhyay and biswarup mukhopadhyaya + + 1 true cm
in 1989 witten derived knot invariants such as the jones polynomial from quantum field theory based on the chern - simon lagrangian @xcite . inspired by witten s work in this paper we shall derive knot invariants from a gauge model of quantum electrodynamics ( qed ) and its nonabelian generalization . from our approach we shall first derive the jones polynomial and then we derive new knot and link invariants which extend the jones polynomial and can give a complete classification of knots and links . this paper is organized as follows . in section [ sec2 ] we give a brief description of a gauge model of qed and its nonabelian extention . in this paper we shall consider a nonabelian extension with a @xmath3 gauge symmetry . with this quantum field model we introduce the partition function and the correlation of a wilson loop which will be a knot invariant of the trivial knot ( also called the unknot ) . this correlation of wilson loop will later be generalized to be knot invariants of nontrivial knots and links . to investigate the properties of these partition function and correlations in section [ sec5 ] we derive a chiral symmetry from the gauge transformation of this new quantum field model . from this chiral symmetry in section [ sec6 ] , section [ sec7 ] we derive a conformal field theory which contains topics such as the affine kac - moody algebra and the knizhnik - zamolodchikov equation . this kz equation is an equation of correlations from which in section [ sec9 ] we can derive the skein relation of the jones polynomial . a main point of our theory on the kz equation is that we can derive two kz equations which are dual to each other . from these two kz equations we derive a quantum group structure for the @xmath4 matrices from which a wilson loop is formed . then from the correlation of these @xmath4 matrices in section [ sec10 ] and section [ sec11 ] we derive new knot invariants which extend the jones polynomial and gives a complete classification of knots . in section [ sec12 ] we extend the new invariants to the case of links and we compute these new link invariants for some examples of links . then in section [ sec13 ] with the new knot invariants we give a classification table of knots which is formed by using the power index @xmath0 which comes from these new knot invariants of the form @xmath5 where @xmath6 denotes a wilson loop and @xmath2 is the braiding matrix for the quantum group structure and is the monodromy of the two kz equations . to begin our derivation of knot and link invariants let us first describe a quantum field model . similar to the wiener measure for the brownian motion which is constructed from the integral @xmath7 we construct a measure for qed from the following energy integral : @xmath8ds \label{1.1}\ ] ] where @xmath9 denotes the proper time in relativity ; @xmath10 denotes the electric charge and the complex variable @xmath11 , real variables @xmath12 , @xmath13 represent one electron and two photons respectively . by extending @xmath14 to @xmath15 we get a quantum theory of qed where @xmath16 denotes the planck constant and @xmath17 is a constant related to absolute temperature . the integral ( [ 1.1 ] ) has the following gauge symmetry : @xmath18 where @xmath19 is a real valued function . we remark that a main feature of ( [ 1.1 ] ) is that it is not formulated with the four - dimensional space - time but is formulated with the one dimensional proper time . we refer to @xcite for the physical motivation of this new qed theory . we can generalize the above qed model with @xmath20 gauge symmetry to qcd type models with nonabelian gauge symmetry . as an illustration let us consider @xmath3 gauge symmetry . similar to ( [ 1.1 ] ) we consider the following energy integral : @xmath21ds \label{n1}\ ] ] where @xmath22 is a two dimensional complex vector ; @xmath23 @xmath24 where @xmath25 denotes a real component of a gauge field @xmath26 ; @xmath27 denotes a generator of @xmath3 ; and @xmath28 @xmath24 where @xmath29 denotes the charge of interaction . from ( [ n1 ] ) we can develop a qcd type model as similar to that for the qed model . we have that ( [ n1 ] ) is invariant under the following gauge transformation : @xmath30 where @xmath31 and @xmath32 . we shall mainly consider the case that @xmath33 is a function of the form @xmath34 where @xmath35 and @xmath36 are analytic functions . since ( [ n1 ] ) is not formulated with the space - time , as analogous to the approach of witten on knot invariants @xcite the following partition function will be shown to be a topological invariant for knots : @xmath37 where @xmath38 which may be called a wilson loop as analogous to the usual wilson loop @xcite where @xmath39 denotes a closed curve of the following form @xmath40 where @xmath36 is an analytic function such that @xmath41 . this closed curve @xmath39 is in a two dimensional phase plane @xmath42 which is dual to @xmath43 . we let this closed curve @xmath39 represents the projection of a knot in this two dimensional space . as usual the notation @xmath44 in the definition of @xmath45 denotes a path - ordered product and @xmath2 denotes a representation of @xmath3 @xcite@xcite . we remark that we also extend the definition of @xmath45 to the case that @xmath39 is not a closed curve with @xmath46 . we shall show that ( [ n3 ] ) is a topological invariant for a trivial knot . this means that in ( [ n3 ] ) the closed curve @xmath39 represents a trivial knot . we shall extend ( [ n3 ] ) to let @xmath39 represent nontrivial knot . our aim is to compute the above knot invariant and its generalization to knot invariants of nontrivial knots which will be defined . for a given curve @xmath47 which may not be a closed curve we define @xmath48 by ( [ n4 ] ) where @xmath49 and @xmath50 . then under an analytic gauge transformation we have the following chiral symmetry : @xmath51 where @xmath52 denotes an analytic function . this chiral symmetry is analogous to the chiral symmetry of the usual nonabelian guage theory where @xmath53 denotes an element of @xmath3 @xcite . we may extend ( [ n5 ] ) by extending @xmath36 to complex variable @xmath11 to have the following chiral symmetry : @xmath54 this analytic continuation corresponds to the complex transformation @xmath55 for describing quantum physics . from this chiral symmetry we have the following formulas for the variations @xmath56 and @xmath57 with respect to the chiral symmetry : @xmath58 and @xmath59 where @xmath11 and @xmath60 are independent variables and @xmath61 when @xmath62 . in ( [ k1 ] ) the variation is with respect to the @xmath11 variable while in ( [ k2 ] ) the variation is with respect to the @xmath60 variable . this two - side - variations is possible when @xmath63 . let us define @xmath64 where @xmath65 is a constant . as analogous to the wzw model @xcite@xcite @xmath66 is a generator of the chiral symmetry for ( [ k1 ] ) . let us consider the following correlation @xmath67 by taking a gauge transformation on this correlation and by the gauge invariance of ( [ n1 ] ) we can derive a ward identity from which we have the following relation : @xmath68 where @xmath69 denotes the variation of the field @xmath70 with respect to the chiral symmetry and the closed line integral @xmath71 is with center @xmath11 and we let the generator @xmath66 be given by ( [ n6 ] ) . we remark that our approach here is analogous to the wzw model in conformal field theory @xcite . from ( [ n5 ] ) and ( [ n6 ] ) we have that the variation @xmath72 of the generator @xmath66 of the chiral symmetry is given by @xcite@xcite : @xmath73 from ( [ n8b ] ) and ( [ n8c ] ) we have that @xmath66 satisfies the following relation of current algebra @xcite@xcite@xcite : @xmath74 where we write @xmath75 then from ( [ n8d ] ) we can show that @xmath76 satisfy the following affine kac - moody algebra @xcite@xcite@xcite : @xmath77 = if_{abc}j_{m+n}^c + km\delta_{ab}\delta_{m+n , 0 } \label{n8}\ ] ] where the constant @xmath78 is called the central extension or the leval of the kac - moody algebra . let us consider another generator of the chiral symmetry for ( [ k2 ] ) given by @xmath79 similar to @xmath66 by the following correlation : @xmath80 we have the following formula for @xmath81 : @xmath82 where @xmath83 denotes an integral with clockwise direction while @xmath71 denotes an integral with counterclockwise direction . we remark that this two - side variation from ( [ n8a ] ) and ( [ n8aa ] ) is important for deriving the two kz equations which are dual to each other . then similar to ( [ n8c ] ) we also we have @xmath84 then from ( [ n8b1 ] ) and ( [ n8c1 ] ) we can derive the current algebra and the kac - moody algebra for @xmath81 which are of the same form of ( [ n8d ] ) and ( [ n8 ] ) . let us first consider ( [ k1 ] ) . from ( [ n8b ] ) and ( [ k1 ] ) we have @xmath85 let us define an energy - momentum tensor @xmath86 by @xmath87 where @xmath29 is the dual coxter number . in ( [ k4 ] ) the symbol @xmath88 denotes normal ordering . this is the sugawara construction of energy - momentum tensor where the appearing of @xmath29 is from a renormalization of quantum effect by requiring the operator product expansion of @xmath89 with itself to be of the following form @xcite @xcite @xcite : @xmath90 for some constant @xmath91 where @xmath92 denotes the dimension of @xmath3 . then we have the following @xmath93 operator product : @xmath94 where @xmath95 and @xmath96 where @xmath97 is for @xmath98 . from ( [ k4 ] ) and ( [ k6 ] ) we have the following equation @xcite@xcite : @xmath99 then form ( [ k3 ] ) we have @xmath100 by ( [ k7 ] ) and ( [ k8 ] ) we have @xmath101 from this equation and by the @xmath102 operator product ( [ k3 ] ) we have the following knizhnik - zamolodchikov equation @xcite @xcite : @xmath103 we remark that in deriving ( [ n9 ] ) we have used line integral expression of operators with counterclockwise direction @xcite@xcite . it is interesting and important that we also have another knizhnik - zamolodchikov equation which will be called the dual equation of ( [ n9 ] ) . the derivation of this dual equation is dual to the above derivation in that the line integral for this derivation of dual equation is with clockwise direction in contrast to counterclockwise direction in the above derivation and that the operator products and their corresponing variables are with reverse order to that in the above derivation . from ( [ k2 ] ) and ( [ n8b1 ] ) we have a @xmath104 operator product given by @xmath105 similar to the above derivation of the kz equation from ( [ d2 ] ) we can then derive the following knizhnik - zamolodchikov equation which is dual to ( [ n9 ] ) : @xmath106 following the idea of witten @xcite , if we cut a knot we get two pieces of curves crossing ( or not crossing ) each other once . this gives two primary fields @xmath107 and @xmath108 where @xmath107 corresponds to a piece of curve with end points parametrized by @xmath109 and @xmath110 and @xmath108 corresponds to the other piece of curve with end points parametrized by @xmath111 and @xmath112 . let us write @xmath113 for @xmath114 and @xmath115 and for some @xmath116 with @xmath117 and @xmath118 . these two pieces of curves then correspond to the following four - point correlation function : @xmath119 ( in the notation @xmath120 we have supressed the @xmath60 variables for simplicity ) . then we have @xcite@xcite : @xmath121^{-2\delta}g(x ) \label{n12}\ ] ] where @xmath122 and @xmath123 and from the kz equation @xmath124 satisfies the following equation : @xmath125 g \label{n13}\ ] ] where @xmath126 this equation has two independent conformal block solutions forming a vector space of dimension 2 . let @xmath127 be a vector in this space and let @xmath128 denotes the braid operation . then following witten @xcite we have @xmath129 where @xmath130 and @xmath131 . then following witten @xcite from ( [ n15 ] ) we can derive the following skein relation for the jones polynomial : @xmath132 where @xmath133 , @xmath134 and @xmath135 are the jones polynomials for undercrossing , overcrossing and zero crossing respectively . let us consider again the correlation @xmath136 in ( [ n11 ] ) which also have the following form : @xmath137 from it in this section we shall present a method which is different from the above section to derive new knot invariants . these new knot invariants will extend the jones polynomial and they will be defined by generalizing ( [ n3 ] ) . we have that @xmath138 satisfies the kz equation for the variables @xmath109 , @xmath111 and satisfies the dual kz equation for the variables @xmath110 and @xmath112 . by solving the kz equation we have that @xmath138 is of the form @xmath139 where @xmath140 and @xmath141 denotes a constant matrix which is independent of the variable @xmath142 . similarly by solving the dual kz equation we have that @xmath138 is of the form @xmath143 where @xmath144 denotes a constant matrix which is independent of the variable @xmath145 . from ( [ m2 ] ) , ( [ m3 ] ) and we let @xmath146 , @xmath147 where @xmath70 is a constant matrix we have that @xmath138 is given by @xmath148 now let @xmath149 . then as @xmath150 we have @xmath151 where @xmath152 is the monodromy of the the kz equation @xcite . we remark that ( [ m5 ] ) is a multivalued function . from ( [ m5 ] ) we have the following relation between the partition function @xmath153 and the matrix @xmath70 : @xmath154 where @xmath155 denotes the identity matrix . now let @xmath39 be a closed curve in the complex plane with initial and final end points @xmath109 . then the following correlation function @xmath156 which is the definition ( [ n3 ] ) defined along the curve @xmath39 , with @xmath157 , can be regarded as a knot invariant of the trivial knot in the three dimensional space whose porjection in the complex plane is the curve @xmath39 . indeed , from ( [ m1 ] ) and ( [ m5 ] ) we can compute ( [ m6 ] ) which is given by : @xmath158 from ( [ m6a ] ) we see that ( [ m6 ] ) is independent of the closed curve @xmath39 which represents the projection of a trivial knot and thus can be regarded as a knot invariant for the trivial knot . in the following let us extend the definition ( [ m6 ] ) to knot invariants for nontrivial knots . since @xmath159 is the monodromy of the kz equation , we have a branch cut such that @xmath160 where @xmath109 and @xmath111 denote two points on a closed curve such that along the direction of the curve the point @xmath109 is before the point @xmath111 . from ( [ m7 ] ) we have @xmath161 similarly for the dual kz equation we have @xmath162 and @xmath163 where @xmath110 before @xmath112 . from ( [ m7a ] ) and ( [ m8a ] ) we have @xmath164 where @xmath109 and @xmath110 denote the end points of a curve which is before a curve with end points @xmath111 and @xmath112 . from ( [ m9 ] ) we see that the algebraic structure of these @xmath4 matrices is analogous to the quasi - triangular quantum group @xcite@xcite . now we let @xmath165 represent a piece of curve with initial end point @xmath166 and final end point @xmath167 . then we let @xmath168 represent two pieces of uncrossing curve . then by interchanging @xmath109 and @xmath111 we have @xmath169 represent the curve specified by @xmath170 upcrossing the curve specified by @xmath171 at @xmath11 . similarly by interchanging @xmath110 and @xmath112 we have @xmath172 represent the curve specified by @xmath170 undercrossing the curve specified by @xmath171 at @xmath11 . now for a closed curve we may cut it into a sum of parts which are formed by two pieces of curve crossing or not crossing each other . each of these parts is represented by ( [ m11 ] ) , ( [ m12 ] ) or ( [ m13 ] ) . then we may define a correlation for a knot whose projection is this closed curve by the following form : @xmath173 where we use ( [ m12 ] ) as an example to represent the state of the two pieces of curve specified by @xmath170 and @xmath171 . the @xmath174 means multiplications of a sequence of parts represented by ( [ m11 ] ) , ( [ m12 ] ) or ( [ m13 ] ) according to the state of each part . the order of the sequence in ( [ m14 ] ) follows the order of the parts given by the direction of the knot . we shall show that ( [ m14 ] ) is a knot invariant for a given knot . in the following let us consider some examples to illustrate the way to define ( [ m14 ] ) and to show that ( [ m14 ] ) is a knot invariant . we shall also derive the three reidemeister moves for the equivalence of knots . let us first consider a knot in fig . 1 . for this knot we have that ( [ m14 ] ) is given by @xmath175 where the product of @xmath4 is from the definition ( [ m12 ] ) . in applying ( [ m12 ] ) we let @xmath109 as the starting and the final point . we remark that the @xmath4 matrices must be put together to follow the definition ( [ m12 ] ) and they are not separated to follow the direction of the knot . fig.1 then we have that ( [ m15a ] ) is equal to @xmath176 where we have used ( [ m9 ] ) . we see that ( [ m16 ] ) is just the knot invariant ( [ m6 ] ) of a trivial knot . thus the knot in fig.1 is with the same knot invariant of a trivial knot and this agrees with the fact that this knot is topologically equivalent to a trivial knot . then let us derive the reidemeister move 1 . consider the diagram in fig.2 . we have that by ( [ m12 ] ) the definition ( [ m14 ] ) for this diagram is given by : @xmath177 where@xmath178 represent a piece of curve with initial end point @xmath109 and final end point @xmath111 which has no crossing . when fig.2 is a part of a knot we can also derive a result similar to ( [ m18 ] ) which is for the reidemeister move 1 . this shows that the reidemeister move 1 holds . fig.2 then let us derive reidemeister move 2 . by ( [ m12 ] ) we have that the definition ( [ m14 ] ) for the two pieces of curve in fig.3a is given by @xmath179 where the two products of @xmath4 separated by the @xmath180 are for the two crossings in fig.3a . we have that ( [ m19 ] ) is equal to @xmath181 where we have used ( [ m9 ] ) . this shows that the diagram in fig.3a is equivalent to two uncrossing curves . when fig.3a is a part of a knot we can also derive a result similar to ( [ m20 ] ) for the reidemeister move 2 . this shows that the reidemeister move 2 holds . fig.3a fig.3b as an illustration let us consider the knot in fig . 3b which is related to the reidemeister move 2 . by ( [ m12 ] ) we have that the definition ( [ m14 ] ) for this knot is given by @xmath182 where we let the curve be with @xmath110 as the initial and final end point and we have used ( [ m7a ] ) and ( [ m8a ] ) . this shows that the knot in fig.3b is with the same knot invariant of a trivial knot . this agrees with the fact that this knot is equivalent to the trivial knot . let us then consider a trefoil knot in fig.4a . by ( [ m12 ] ) and similar to the above examples we have that the definition ( [ m14 ] ) for this knot is given by : @xmath183 where we have repeatly used ( [ m9 ] ) . then we have that ( [ m21 ] ) is equal to : @xmath184 where we have used ( [ m7a ] ) and ( [ m9 ] ) . we see that ( [ m22 ] ) is a knot invariant for the trefoil knot in fig.4a . fig.4a fig.4b then let us consider the trefoil knot in fig . 4b which is the mirror image of the trefoil knot in fig.4a . the definition ( [ m14 ] ) for this knot is given by : @xmath185 where similar to ( [ m21 ] ) we have repeatly used ( [ m9 ] ) . then we have that ( [ m25 ] ) is equal to : @xmath186 where we have used ( [ m8a ] ) and ( [ m9 ] ) . we see that this is a knot invariant for the trefoil knot in fig.4b . we notice that the knot invariants for the two trefoil knots are different . this shows that these two trefoil knots are not topologically equivalent . then let us derive the reidemeister move 3 . we have that the definition ( [ m14 ] ) for the diagram in fig.5a is given by @xmath187 where we let the the curve with end points @xmath188 , @xmath189 starts first , then the curve with end points @xmath109 , @xmath111 starts second . similar to the derivation of the above invariants by ( [ m7a ] ) , ( [ m8a ] ) , ( [ m9 ] ) we have that ( [ m29 ] ) is equivalent to the following correlation : @xmath190 fig.5a fig.5b on the other hand the definition ( [ m14 ] ) for the diagram fig.5b is given by @xmath191 where the ordering of the three curves is the same as that in fig.5a . by ( [ m7a ] ) , ( [ m8a ] ) , ( [ m9 ] ) we have that ( [ m31 ] ) is equal to @xmath192 then by reversing the ordering of the curves with end points @xmath109 , @xmath111 and with end points @xmath112 , @xmath193 respectively we have that both ( [ m30 ] ) and ( [ m31 ] ) are equal to @xmath194 this shows that fig.5a is equivalent to fig.5b and this gives the reidemeister move 3 . from the above examples we see that knots can be classified by the number @xmath0 of product of @xmath2 and @xmath195 matrices . more calculations and examples of the above knot invariants will be given elsewhere . with the knot invariants in the above section we can now give a classification of knots . let @xmath196 and @xmath197 be two knots . since the two w - products @xmath198 and @xmath199 faithfully represent two oriented pieces of curves which are crossing or not crossing to each other we have that from the orientation of a knot @xmath200 we have that the product of sequence of w - matrices ( we may call this product as a generalized wilson loop ) which are formed according to the orientation of @xmath200 in the correlation ( [ m14 ] ) of defining an invariant of the knot @xmath200 faithfully represents the knot @xmath200 . from this we have that @xmath196 and @xmath197 are topological equivalent if and only if their generalized wilson loops can be transformed to each other by using ( [ m7a ] ) , ( [ m8a ] ) and ( [ m9 ] ) . we note that in the above section we can derive the reidemeister moves by using ( [ m7a ] ) , ( [ m8a ] ) and ( [ m9 ] ) . thus this equivalence of @xmath196 and @xmath197 represented by their generalized wilson loops agrees with the fact that @xmath196 and @xmath197 are topologically equivalent if and only if they can be transformed to each other by the reidemeister moves . then since each knot can be changed to a trivial knot by applying braiding operation @xcite which is equivalent to ( [ m7a ] ) , ( [ m8a ] ) and ( [ m9 ] ) it follows that these new knot invariants can be equivalently transformed to the form @xmath201 where @xmath0 is an integer and @xmath202 is the knot invariant for the trivial knot . this form can also be shown by a direct computation which is similar to the computation of the invariant of the trivial knot . we notice that since this new knot invariant is of the form @xmath203 we have that knots can be completely classified by the power index @xmath0 of @xmath2 . similar to the case of knots we have that the generalized wilson loop for a link can faithfully represents this link and that the correlation ( [ m14 ] ) of the generalized wilson loop of this link is an invariant which can completely classifies links . for the case of link as similar to the case of knot the ordering of the crossings @xmath198 can be given by following the orientation of each component of a link . when a component of a link has been traced for one loop the crossings @xmath198 related to this component can be ordered and the pieces @xmath204 related to these crossings ( which may come from other components of the link ) have also been required being ordered . in the following section we give some examples to illustrate the formation and computations of this new link invariant . let us first consider the link in fig.6a . we may let the two knots of this link be with @xmath109 and @xmath112 as the initial and final end point respectively . we let the ordering of these two knots be such that when the @xmath11 parameter goes one loop on one knot then the @xmath11 parameter for another knot also goes one loop . the correlation ( [ m14 ] ) for this link is given by : @xmath205 we let the ordering of the @xmath4-matrices in ( [ l1a ] ) be such that @xmath170 and @xmath206 start first . then next @xmath207 and @xmath171 follows . form this ordering we have that ( [ l1a ] ) is equal to : @xmath208 where we have used ( [ m7a ] ) and ( [ m8a ] ) . since by definition ( [ m14 ] ) we have that @xmath209 is the knot invariant for two unlinking trivial knots , equation ( [ l1 ] ) shows that the link in fig.6a is topologically equivalent to two unlinking trivial knots . similarly we can show that the link in fig.6b is topologically equivalent to two unlinking trivial knots . fig.6a fig.6b fig.7a fig.7b let us then consider the hopf link in fig.7a . the correlation ( [ m14 ] ) for this link is given by : @xmath210 the ordering of the @xmath4-matrices in ( [ l3 ] ) is such that @xmath170 starts first and @xmath171 follows it . then next we let @xmath207 starts first and @xmath206 follows it . the ordering is such that when the @xmath11 parameter goes one loop in one knot of the link we have that the @xmath11 parameter also goes one loop on the other knot . from the ordering we have that ( [ l3 ] ) is equal to : @xmath211 then let us consider the following correlation : @xmath212 we let the ordering of the @xmath4-matrices in ( [ l5 ] ) be such that @xmath170 starts first and @xmath206 follows it . then next @xmath207 starts first and @xmath171 follows it . from the ordering we have that ( [ l5 ] ) is equal to : @xmath213 on the other hand from the ordering of ( [ l5 ] ) we have that ( [ l5 ] ) is equal to : @xmath214 where we have repeatly used ( [ m9 ] ) . from ( [ l4 ] ) , ( [ l6 ] ) and ( [ l7 ] ) we have that the knot invariant for the hopf link in fig.7a is given by : @xmath215 then let us consider the hopf link in fig.7b . the correlation for this link is given by @xmath216 by a derivation which is dual to the above derivation for the hopf link in fig.7a we have that ( [ l9 ] ) is equal to @xmath217 we see that the invariants for the above two hopf links are different . this agrees with the fact that these two links are not topologically equivalent . we can extend the above computations to other links . as examples let us consider the linking of two trivial knot with linking number @xmath218 . similar to the above computations we have that this link which analogous to the hopf link in fig.7a is with an invariant equals to @xmath219 . also for this link which analogous to to the hopf link in fig.7b is with an invariant equals to @xmath220 . more calculations and examples of these new link invariants will be given elsewhere . in this section let us give some arguments to determine the new knot invariants of prime knots and nonprime knots . we have shown that this new invariant of each knot is of the form @xmath221 where @xmath0 is an integer . this power index @xmath0 can be regarded as a measure of the complexity of a knot . let us determine @xmath0 for prime and nonprime knots . we need only to determine @xmath0 for knots with positive @xmath0 since the corresponding mirror image will have negative @xmath0 if the mirror image is not equivalent to the corresponding knot . we have shown that the invariants for links of two trivial knots with linking number @xmath218 as in fig.7a are with the product of @xmath2 of the form @xmath222 with @xmath223 . this @xmath0 is an even number . now if we insert these links into the knot table of prime knots we see that the prime knots must with @xmath0 being an odd number ( we may refer to the knot table in @xcite . we may have nonprime knots which are not in this knot table having the same @xmath0 as these links . in this case our new link invariants still can distinguish them because the corresponding link invariants are of two - loop form while the corresponding knot invariants are of one - loop form ) . then we expect that for prime knots @xmath0 is an odd prime number . computations show that for the knot @xmath224 we have @xmath225 , for the knot @xmath226 we have @xmath227 . then from some arguments on the effect of @xmath2 we should have that for the knot @xmath228 we have @xmath229 and for the knot @xmath230 we have @xmath231 . then how about the knot @xmath232 ? we have that the numbers from @xmath233 to @xmath234 are ocupied by knots and links of two trivial knots . then @xmath235 is ocupied by the link of two trivial knots with linking number @xmath236 . thus for the knot @xmath232 we should have @xmath237 or @xmath238 . from some argument on the effect of @xmath2 we should have @xmath238 for @xmath232 . then is there a knot with @xmath237 ? let us first consider the granny knot ( or the square knot ) which is a nonprime knot composed with the knot @xmath224 and its mirror image . this square knot has @xmath239 crossings and @xmath240 alternating crossings and thus its complexity which is measured by the power index @xmath0 of @xmath2 is less than that of @xmath228 which is with @xmath236 alternating crossings . thus this granny knot is with @xmath241 ( @xmath227 has been occupied by @xmath226 ) . let us denote this granny knot by @xmath242 where @xmath243 denotes the connected sum of two knots such that the resulting total number of alternating crossings is equal to the total number of alternating crossings of the two knots minus @xmath244 . then let us consider the reef knot which is a nonprime knot composed with two identical knots @xmath224 . this knot has @xmath239 alternating crossings which is equal to the total number of crossings as that of @xmath232 . since this knot is nonprime its complexity is less than that of @xmath232 where the complexity may be measured by the power index @xmath0 of @xmath2 . thus this reef knot is with power index @xmath0 less than @xmath245 . then if we also regard the total number of alternating crossings of a knot as a way to measure the complexity of a knot we have that the power index @xmath0 of this granny knot is greater than @xmath236 since @xmath228 is with @xmath236 alternating crossings and with @xmath229 . let us denote this reef knot by @xmath246 where @xmath247 denotes the connected sum for two knots such that the resulting total number of alternating crossings is equal to the total number of alternating crossings of the two knots . now let us look for knots with @xmath236 or @xmath239 alternating crossings . let us consider the nonprime knot @xmath248 composed with a knot @xmath224 and a knot @xmath226 with @xmath249 crossings and @xmath236 alternating crossings . in this case we have that the power index @xmath0 of @xmath248 should be greater than that of @xmath228 which is exactly with @xmath236 alternating crossings since @xmath248 in addtion has @xmath249 crossings . then the power index @xmath0 of @xmath248 should be less than that of @xmath230 which is also with @xmath236 alternating crossings but these crossings are arranged in a more complicated way which is an effect of @xmath250 such that @xmath230 is with @xmath231 . thus @xmath248 is with @xmath251 . then we consider the nonprime knot @xmath252 with @xmath253 crossings and @xmath236 alternating crossings . the power index of this knot should be less than that of @xmath254 since it is with @xmath239 alternating crossings . then the power index of @xmath254 should be greater than that of @xmath230 since it has in addition @xmath253 crossings which would be enough for a greater power index @xmath0 . then we have that @xmath252 is with @xmath255 since it has @xmath253 crossings and @xmath236 alternating crossings and thus is with the same complexity as @xmath230 . then since @xmath230 can not have @xmath255 we thus have that @xmath252 is with @xmath255 . then we consider the nonprime knot @xmath256 with @xmath234 crossings and @xmath239 alternating crossings . the power index @xmath0 of this knot should be greater than that of @xmath254 which has exactly @xmath239 alternating crossings . then the power index @xmath0 of this knot should be less than that of @xmath232 which also has @xmath239 alternating crossings but these crossings are arranged in a more complicated way with an effect of @xmath257 from that of @xmath258 . thus @xmath256 is with @xmath259 and finally we have that @xmath254 is with power index @xmath237 . thus for @xmath0 from @xmath233 to @xmath245 we have fill in a suitable knot with power index @xmath0 ( except the case @xmath260 which is filled in with the hopf link ) such that odd prime numbers are filled with prime knots . in a similar way we may determine the power index @xmath0 of other prime and nonprime knots . we list the results up to @xmath261 in a form of table . @xmath262 from this table we see that comparable nonprime knots ( in a sense from the table ) are grouped in each of the intervals between two prime numbers . it is interesting that in each interval nonprime numbers are one - to - one filled with the comparable nonprime knots while prime numbers are filled with prime knots . this grouping property of classification reflects that the new knot invariants ( and hence the power index @xmath0 ) give a classification of knots . let us find out some rules for the whole classification table . we have shown that even numbers can not be filled with prime knots . thus even numbers ( except @xmath244 ) can only be filled with nonprime knots . on the other hand each odd nonprime number is between two even numbers which are power indexes of nonprime knots . thus the knot corresponding to this odd number is in the same group of these two nonprime knots and is comparable with these two nonprime knots and thus must also be a nonprime knot . from this we have that nonprime numbers are power indexes of nonprime knot . similarly by this grouping property we have that odd prime numbers are power indexes of prime knots . it is interesting to note that from the above knot table we see that in the @xmath243 product the knot @xmath263 plays the role of the number @xmath244 in the usual multiplication of numbers . thus the @xmath243 product ( or the connected sum ) is a kind of multiplication corresponding to the usual multiplication of numbers . however the general rule for this multiplication is rather complicated . this reflects the fact that the numbers @xmath0 are the power indexes of @xmath2 which are with simple rule for the addition ( and not for multipication ) . from this generalized multiplication ( which is the connected sum ) of knots which corresponds to the usual multiplication of the power indexes @xmath0 we also have that prime knots are with odd prime numbers as power indexes and nonprime knots are with nonprime numbers as power indexes . thus we have a classification table of knots such that each prime knot corresponds to a prime power index @xmath0 and each nonprime knot corresponds to a nonprime power index @xmath0 . more computations to verify this knot table shall be given alsewhere . in this paper from a new quantum field model we have derived a conformal field theory and a quantum group structure for generalized wilson loops from which we can derive the jones polynomial and new knot and link invariants which extend the jones polynomial . we show that these new invariants can completely classify knots and links . these new invariants are in terms of the monodromy @xmath2 of two knizhnik - zamolochikov equations which are dual to each other . in the case of knots these new invariants can be written in the form @xmath203 from which we may classify knots with the power index @xmath0 of @xmath2 . a classification table of knots can then be formed where prime knots are classified with odd prime numbers @xmath0 and nonprime knots are classified with nonprime numbers @xmath0 .	we propose a gauge model of quantum electrodynamics ( qed ) and its nonabelian generalization from which we derive knot invariants such as the jones polynomial . our approach is inspired by the work of witten who derived knot invariants from quantum field theory based on the chern - simon lagrangian . from our approach we can derive new knot and link invariants which extend the jones polynomial and give a complete classification of knots and links . from these new knot invariants we have that knots can be completely classified by the power index @xmath0 of @xmath1 where @xmath2 denotes the @xmath2-matrix for braiding and is the monodromy of the knizhnik - zamolodchikov equation . a classification table of knots can then be formed where prime knots are classified by prime integer @xmath0 and nonprime knots are classified by nonprime integer @xmath0 . * pacs codes : * 02.40-k , 02-40.re , 11.15.-q , 11.25.hf * keywords : * new quantum field models , new quantum invariants of knots and links .
this work was supported by the federal agency for atomic energy of russian federation and by grants nsh-843.2006.2 , dfg-436 rus 113/820/0 - 1(r ) , rffi-05 - 02 - 04012-nnioa , and ptdc / fis/70843/2006-fisica . 99 s. godfrey and n. isgur , phys . * d32 * , 189 ( 1985 ) ; m. kroll , r. ricken , d. merten , b. metsch , and h. petry , eur . j. * a9 * , 73 ( 2000 ) ; a.m. badalyan and b. l. g. bakker , phys . rev . * d66 * , 034025 ( 2002 ) ; a.m. badalian , phys . atom . nucl . * 66 * , 1342 ( 2003 ) . jaffe , phys . rev . * d15 * , 267 , 281 ( 1977 ) . achasov , s.a . devyanin , and g.n . shestakov , phys . lett . * b96 * , 168 ( 1980 ) ; d. black , a.h . fariborz , f. sannino , and j. schechter , phys . rev . * d59 * , 074026 ( 1999 ) ; m. alford and r.l . jaffe , nucl . b578 * , 367 ( 2000 ) ; l. maiani , f. piccinini , a.d . polosa , and v. riquer , phys . * 93 * , 212002 ( 2004 ) . j. weinstein and n. isgur , phys . rev . * d27 * , 588 ( 1979 ) . d. lohse , j.w . durso , k. holinde , and j.speth , phys . b234 * , 235 ( 1990 ) ; g. janssen , b.c . pearce , k. holinde , and j. speth , phys . rev * d52 * , 2690 ( 1995 ) . oller and e. oset , nucl . phys . * a620 * , 438 ( 1997 ) ; ( e ) nucl . phys . * a652 * , 407 ( 1999 ) . markushin , eur . j. * a8 * , 389 ( 2000 ) . c. hanhart , presentation at this conference ; arxiv:0711.0578 . n.n . achasov and v.n . ivanchenko , nucl . phys . * b315 * , 465 ( 1989 ) . et al . _ , b440 * , 442 ( 1998 ) ; m.n . et al . _ , phys . lett . * b485 * , 349 ( 2000 ) . et al . _ [ cmd-2 collab . ] , phys . * b462 * , 380 ( 1999 ) . a. aloisio _ [ kloe collab . ] , phys . lett . * b536 * , 209 ( 2002 ) ; a. aloisio _ et al . _ , phys . b537 * , 21 ( 2002 ) . j. haidenbauer , c. hanhart , yu.s . kalashnikova , a.e . kudryavtsev , and a.v . nefediev , eur . j. * a24 * , 437 ( 2005 ) . t. barnes , phys . b165 * , 434 ( 1985 ) . s. krewald , r.h . lemmer , and f. sassen , phys . rev . * d69 * , 016003 ( 2004 ) . achasov , a.v . kiselev , phys . rev . * d76 * 077501 , ( 2007 ) , arxiv : hep - ph/0606268 . j. haidenbauer , c. hanhart , yu.s . kalashnikova , a.e . kudryavtsev , and a.v . nefediev , arxiv : hep - ph/0608191 ; arxiv:0711.2902 . c. hanhart , kalashnikova , a.e . kudryavtsev , and a.v . nefediev , phys . * d75 * , 074015 ( 2007 ) . j. haidenbauer , c. hanhart , yu.s . kalashnikova , a.e . kudryavtsev , and a.v . nefediev , phys . rev . * c73 * , 045203 ( 2006 ) . smith , ann . phys . * 53 * , 521 ( 1969 ) . f. close , n. isgur , and s. kumano , nucl . phys . * b389 * , 513 ( 1993 ) w. kwong and j.l . rosner , phys . * d38 * , 279 ( 1988 ) . r. barbieri , r. gatto , and r. kogerler , phys . b60 * , 183 ( 1976 ) . et al . _ [ belle collab . ] , phys . rev . * d75 * , 051101 ( 2007 ) .	it is argued that radiative decays of scalars @xmath0 can serve as a decisive tool in establishing the nature of the latter . in particular , predictions for the widths of the radiative decays @xmath1 ( @xmath2 , @xmath3 ) are given in the framework of the molecule model of the scalars . finite range corrections are discussed in detail for the two - gamma decays of hadronic molecules , with a special attention payed to the interplay of various scales involved in the problem and to the gauge invariance of the amplitude . the results are applied to the two - photon decay of the @xmath4 , and the existing experimental data on this decay are argued to support the molecule assignment for the scalar @xmath4 . = 11.6pt the problem of the structure of light scalar mesons is of a fundamental importance for understanding the properties of the entire scalar sector , that is , the sector of states with the quantum numbers of the vacuum , including purely gluonic excitations . in particular , the identification of the @xmath5 and @xmath4 mesons , together with the experimental studies of the lightest scalars ( @xmath6 and @xmath7 ) , will allow one to establish the structure of multiplets of scalars and to find the signature of the scalar glueball in the spectrum of physical states . there are several models for the @xmath5 and @xmath4 . the latter can be considered as @xmath8 quark antiquark states @xcite strongly coupled to the mesonic continuum and thus strongly distorted with the unitarisation process . however , due to the proximity of the @xmath9 threshold , it is natural to assume a considerable admixture of the four quark component in the wave functions of these mesons , either as a compact four quark with hidden strangeness @xcite , or as a @xmath9 molecule . these might be t - channel exchanges to be responsible for the formation of such a molecule @xcite . it is therefore important to establish a test which would allow one to distinguish between these models and thus to reveal the actual nature of these scalars ( in particular , efficient methods to discriminate between the molecule and compact states are strongly needed for the recent progress see @xcite ) . since years , radiative decays of the @xmath10 , @xmath11 , have been considered as such an experimental tool @xcite . indeed , these decays point to a large @xmath9 component in the scalars wave function @xcite . still a number of shortcomings of this approach should be mentioned . first of all , the radiative decays of the @xmath12 do not allow one to probe the nonstrange component of the scalars , and the contribution of the quark loops is strongly suppressed as compared to the contribution of the meson loops . finally , the phasespace available in the final state of these decays is limited to a large extend . in the meantime , another class of radiative decays involving scalars is known the radiative decays of the scalars themselves : @xmath13 , where the vector in the final state is either massive ( @xmath14 or @xmath15 ) or massless , that is one deals with a two photon decay in the latter case . whatever model of scalars is used , gauge invariance imposes strong constraints on the decay amplitude : @xmath16,\quad a=\frac{m^2_v}{m^2},\quad b=\frac{m^2_s}{m^2 } , \label{w1}\ ] ] where @xmath17 is the kaon mass and @xmath18 are the four momenta of the vector particles . below we shall evaluate the widths of the radiative decays involving the scalars @xmath0 in the molecule assignment for the latter . [ cols="^,^,^ " , ] notice that another approach to two photon decays of molecules is known in the literature , namely the approach based on the formula @xmath19 which is written in analogy with that for the positronium two photon decay . although this approach appears quite successful in qed , it has to fail in hadronic physics . first of all , the w.f . of the kaon molecule is simply not known , so one has to rely on models . moreover , since @xmath20 is very sensitive to the details of the bound state formation , the predictions of this approach may vary drastically ( the predictions found in the literature vary by an order of magnitude , from 0.6 kev in @xcite to 6 kev in @xcite ) . furthermore , any attempt to evaluate corrections to this leading term results in either gauge invariance or energy conservation law breaking . indeed , the decay amplitude in this approach is usually given as an overlap integral between the molecule w.f . and the amplitude of the process @xmath21 : @xmath22 . \label{m1}\ ] ] if the amplitude @xmath23 is taken off - shell , then it obviously fails to be gauge invariant . on the contrary , for the on shell gauge invariant amplitude @xmath23 , kaons carry the energy @xmath24 , rather than @xmath25 , so that energy conservation law is violated . the last , but not the least , argument against using the @xmath20-based approach to hadronic processes is that the hierarchy of scales has to be different , namely @xmath26 , in order to validate the given formula with @xmath20 . finally , we estimate the finite range corrections to the point - like predictions quoted in tab.[t1 ] we are interested in the potentially large corrections of order @xmath27 . if the vector in the final state is massive , then the photon in the final state is soft ( @xmath28 ) , and the kinematics of the loop becomes nonrelativistic @xcite . inclusion of the finite - range effects amounts to the substitution @xmath29 , with a suitable form of the vertex @xmath30 . gauge invariance requires then that the extra momentum dependence coming from the vertex argument should be gauged , that is @xmath31 and the derivative term gives rise to an extra contact diagram with the photon emission from the scalar vertex . gauge invariance is therefore preserved to order @xmath32 . notice however that one should be extremely careful when treating the loop integrals entering the decay amplitude . indeed , although the full amplitude is finite , every individual integral is divergent . if a cut - off is introduced then to make them finite , gauge invariance may be badly broken and , as a result , wrong conclusions may be deduced ( see , for example , @xcite and explanations in @xcite ) . finally one arrives at the conclusion that no corrections of order @xmath27 appear for the point - like predictions @xcite . the same conclusion holds for the two photon decays of scalars , though it is not straightforward to arrive at this conclusion and one needs to develop a selfconsistent gauge - invariant approach to this decay . it was suggested in @xcite to use an effective kaon interaction lagrangian ( for the neutral - particle exchange , generalisation to the charged - particle exchange being trivial ) written to order @xmath33 : @xmath34 ^ 2 , \label{lint}\ ] ] with the coupling constants @xmath35 being of the same order of magnitude . this lagrangian is subject to renormalisation to order @xmath33 see @xcite for the details . in the renormalised theory , the kaon propagator and the photon emission vertex , which are the dressed quantities , built as solutions of the corresponding field theoretical equations , read : @xmath36 where the ellipsis denotes terms which do not contribute to the decays under consideration . the renormalisation constants for the kaon propagator and for the photon emission vertex coincide due to gauge invariance . the kaon mass @xmath17 is the renormalised physical mass . finally , the most important ingredient the scalar vertex beyond the point - like limit comes as a solution of the homogenious bethe salpeter equation @xcite : @xmath37 this vertex is to be normalised @xcite , which gives for the scalar coupling the formula @xmath38 which coincides with the point - like result ( [ gs0 ] ) as @xmath39 . with the scalar vertex , the dressed kaon propagator , and the dressed photon emission vertex in hand we are in a position to evaluate the width of the scalar two - photon decay up to the order @xmath33 . the amplitude of the process is given by the set of diagrams formally coinciding with those for the point - like vertex , depicted at fig.[3d ] . notice , however , an important difference : all ingredients are dressed now , and this is a necessary condition to preserve gauge invariance beyond the point - like limit . the only quantity which should not be dressed is the @xmath40 vertex in the third diagram . indeed , the scalar vertex @xmath41 obeys the bethe salpeter equation and thus absorbs all dressing diagrams . in view of the fact that we deal with an explicitly gauge invariant amplitude , we use the trick suggested in @xcite and , in order to extract the amplitude , we read - off the coefficient at the structure @xmath42 in the transition matrix element @xmath43 \epsilon_{1\mu}^\epsilon_{2\nu}^,\quad p = q_1+q_2 , \label{vertex}\ ] ] which is a particular case of ( [ w1 ] ) adapted for the two photon case . then @xmath44 and , by an explicit calculation , one can find that @xmath45 . therefore , no large corrections of order @xmath27 appear for the point - like result ( [ beauty ] ) . we conclude therefore , that finite range effects give only moderate corrections to the point - like predictions ( of order @xmath46 in the amplitude ) , provided they are included in a self - consistent and gauge - invariant way @xcite . we refer to the point - like results presented in tab.1 as to the molecule model predictions for the radiative decays involving scalars . for the sake of comparison , we quote in tab.[t1 ] the results of calculations in the quark antiquark assignment for the scalars , which can be obtained with the help of the results of @xcite . from tab.1 one can conclude that experimental data are well described in the molecule assignment for the scalars ( a recent result by belle @xcite @xmath47 kev gives an even better coincidence with the point - like prediction ) . furthermore , predictions for the radiative decays of scalars with massive vectors in the final state demonstrate a clear hierarchy , depending on the assignment prescribed to the scalar mesons . this makes these decays an extremely promising tool in establishing the nature of the @xmath0 .
in solving many mathematical and physical problems by means of numerical methods one is often challenged to seek derivatives of various functions given in discrete points . in such cases , when it is difficult or impossible to take derivative of a function analytically one resorts to numerical differentiation . it should be noted that there exists a great deal of formulae and techniques of numerical differentiation ( see , for instance , ref . @xcite ) . as a rule , the function in question @xmath0 is replaced with the easy - to - calculate function @xmath1 and then it is approximately supposed that @xmath2 . the derivatives of higher orders are computed similarly . therefore , in order to obtain numerical value of the derivative of the considered function it is necessary to indicate correctly the interpolating function @xmath3 . if the values of the function @xmath0 are known in @xmath4 discrete points , the function @xmath3 is usually taken as the polynomial of @xmath5th power . to find the derivative of functions having the intervals both quick and slow variation quasi - uniform nets are used ( see ref . this method has an advantage since constant small mesh width is unfavorable in this case , because it leads to the strong enhancement of the function values table . the problem of the numerical differentiation accuracy is also of interest . the numerical differentiations formulae , taking into account the values of the considered function both at @xmath6 and @xmath7 ( @xmath8 is a point where the derivative is computed ) , are called central differentiation formulae . for instance , the formulae based on stirling interpolating polynomial can be included in this class . such formulae are known to have higher accuracy compared to the formulae , using unilateral values of a function , i.e. , for instance , at @xmath6 . the range of numerical differentiation formulae based on different interpolating polynomials is limited , as a rule , to finite points of interpolation . all available formulae known at the present moment are obtained for a certain concrete limited number of interpolation points ( see refs . it can be explained by the fact that the procedure of the finding of the interpolating polynomial coefficients in the case of the arbitrary number of interpolation points is quite awkward and requires formidable calculations . it is worth mentioning that the procedure of the numerical differentiation is incorrect . indeed , in ref . @xcite it was shown that it is possible to select such decreasing error of the function in question which results in the unlimited growth of the error in its first derivative . some recent publications devoted to the numerical differentiation problem should be mentioned ( see , e.g. , ref . @xcite ) . in this work the finite difference formulae for real functions on one dimensional grids with arbitrary spacing were considered . the formulae of central differentiation for the finding of the first and the second derivatives of the functions given in @xmath9 discrete points are derived in this paper . the number of interpolation points is taken to be arbitrary . the obtained formulae for the derivatives calculation do not require direct construction of the interpolating polynomial . as an example of the use of the developed method we calculate the first derivative of the function @xmath10 . the obtained result is studied in the limiting case @xmath11 . we examine the spectral characteristics of the weight coefficients sequence of the numerical differentiation formulae for the different number of the interpolation points . the performed analysis can be applied to the studying of the accuracy of the numerical differentiation technique developed in this work . it is found that the derived formulae of numerical differentiation have a high accuracy in a very wide range of spatial frequencies . without the restriction of generality we suppose that the derivative is taken in the zero point , i.e. @xmath12 . let us consider the function @xmath0 given in equidistant points @xmath13 , where @xmath14 and @xmath15 is the constant value . we can pass the interpolating polynomial of the @xmath16th power through these points @xmath17 the values of the function in points of interpolation @xmath18 coinciding with the values of the interpolating polynomial in these points : @xmath19 . let us define as @xmath20 the differences of the values of the function @xmath0 in diametrically opposite points @xmath21 and @xmath22 , i.e. @xmath23 . we can present @xmath20 in the form @xmath24 to find the coefficients @xmath25 , @xmath26 , we have gotten the system of inhomogeneous linear equations with the given free terms @xmath20 . it will be shown below that this system has the single solution . we will seek the solution of the system [ eq . ] in the following way @xmath27 where @xmath28 are the undetermined coefficients satisfying the condition @xmath29 thus , the system of equations [ eq . ] is reduced to the equivalent , but more simple system [ eq . ] , in which for each fixed number @xmath30 it is necessary to find the coefficients @xmath31 . let us resolve the system of equations [ eq . ] according to the cramer s rule : @xmath32 where @xmath33 @xmath34 in eq . we used the formula for the calculation of the vandermonde determinant . from eq . it follows that the determinant of the system of equations [ eq . ] is not equal to zero , i.e. the system of equations [ eq . ] has the single solution . the most simple expression for @xmath35 is obtained in the case of @xmath36 that corresponds to a calculation of the first - order derivative @xmath37 from eq . as well as taking into account eqs . and we get the expression for the coefficients @xmath38 @xmath39 where @xmath40 it should be noted that one can similarly get the expression for the coefficients @xmath41 which is presented in the following way @xmath42 taking into account eqs . - we finally get the formula for the first derivative of the function @xmath0 @xmath43 the algorithm for the computation of the coefficients @xmath38 is presented in the appendix [ append ] and the results for the certain concrete number of the interpolation points @xmath9 are given in tab . [ alphatab ] . .[alphatab ] values of the coefficients @xmath38 . [ cols="<,^,^,^,^,^,^",options="header " , ] note that the expression for the first derivative obtained by this method coincides with the value presented in the refs . @xcite for @xmath44 that corresponds to three and five points of interpolation . however , technique developed in this article allows one to calculate the coefficients @xmath38 , and hence the first derivative , for any value of @xmath45 . similar formula can be obtained for the calculation of the second derivative . we give without proof corresponding expression @xmath46 where @xmath47 and the product @xmath48 is introduced in eq . . as an example of the use of the obtained central differentiation formulae we will compute the first derivative of the function @xmath10 at @xmath49 . let us set the value of the mesh width @xmath15 equal to @xmath50 . notice that , as a rule , the less the mesh width @xmath15 the more exact result numerical differentiation gives . we have chosen rather big value of @xmath15 . the eq . for this case takes the form @xmath51 in eq . we take that @xmath52 if @xmath53 is an even number , and @xmath54 if @xmath53 is an odd number . let us study the obtained result in the limiting case @xmath11 . first , it is necessary to calculate the value of the product @xmath48 within the limit @xmath11 @xmath55 here we used the known value of infinite product @xmath56 using eqs . and , we find that the expression for the coefficients @xmath38 within the limit @xmath11 is represented in the following way @xmath57 now it is easy to complete the studying of eq . . substituting the result from eq . to eq . and using the known value of infinite series we get that @xmath58 thus , it is shown that the method of the derivatives finding , developed in this paper , gives for the function @xmath10 the value of the first derivative which coincides with the exact analytical one even at rather crude mesh width . in the section [ ffderiv ] of the present work we derived the formula for the finding of the first derivative of the function @xmath0 at @xmath49 . this result can be easily generalized for the case of the arbitrary point @xmath59 . if we set that @xmath60 , and moreover supposing that @xmath61 for @xmath62 and @xmath63 ( see tab . [ alphatab ] ) , then in the considered case eq . reads as follows @xmath64 here the summing is taken over all range of the function involved : @xmath65 . for instance , if the values of the function are set on the limited equidistant collection of elements @xmath66 , then eq . can be rewritten in the form @xmath67 it is worth noticing that in eq . we used the periodicity condition of the weight coefficients @xmath68 fig . [ alphan1 ] presents the example of the weight coefficients of the differentiating sequence @xmath69 . . ] thus the first derivative computation of the function @xmath0 at the points @xmath59 , @xmath70 , is reduced to the procedure of the calculation of the mutual correlation function between the finite sequences @xmath38 and @xmath71 . it is known ( see , e.g. , ref . @xcite ) that if a function satisfies the dirichlet conditions in the interval @xmath72 , then it can be expanded into the fourier series @xmath73 where the expansion coefficients are presented in the way @xmath74 if the first derivative @xmath75 satisfies the analogous conditions as the function @xmath0 , then the following expression will be valid @xmath76 therefore , form eqs . and it follows that the differentiation procedure is the linear filter with the frequency characteristic : @xmath77 @xcite . similarly we receive for the second derivative @xmath78 in this case the the frequency characteristic of the corresponding filter has the form : @xmath79 . according to wiener - khinchin theorem ( see , e.g. , ref . @xcite ) the mutual correlation function between the two finite sequences can be calculated with the help of the inverse fourier transform of the mutual spectrum of the considered sequences . thus , if we define that @xmath80 is the complex spectrum of the differentiating sequence @xmath38 , and @xmath81 is the spectrum of the function @xmath0 , then it follows from eq . that @xmath82 where @xmath83 is the complex conjugated quantity with respect to @xmath84 . comparing eqs . and we obtain that the accuracy of the numerical differentiation performed with the use of the various types of the sequences @xmath38 is characterized by the closeness of imaginary parts of their spectra to the linearly growing sequence @xmath85 . the spectra of the sequences @xmath38 are depicted in fig . [ spectra111211 ] for the various values of @xmath45 at @xmath86 . at @xmath86 . ] it can be seen from this figure that for @xmath87 , i.e. for the sequence shown in fig . [ alphan1 ] , the imaginary part of the spectrum is the branch of the function @xmath88 . the linearity condition is satisfied only in the vicinity of zero and @xmath89 . however , at @xmath90 the spectrum practically does not differ from the linear one up to @xmath91 . the more close to linear one is the spectrum of the sequence @xmath92 . the difference between the imaginary parts of the spectra of the sequences @xmath38 and the linearly growing sequence @xmath85 are presented in fig . [ log101 ] . versus @xmath93 for different @xmath45 . ] the computations have been performed with the accuracy up to @xmath94 , thus the reliable results at @xmath90 have been obtained for @xmath95 , and at @xmath96 for @xmath97 . the presented results demonstrate the high accuracy of the numerical differentiation carried out with the help of the sequences @xmath38 in the wide range of the spatial frequencies . now let us briefly consider the sequences for the calculation of the second derivative @xmath98 , which are given in eq . . their spectral properties can be obtained in the similar manner as we have done it for the case of the sequences @xmath38 and therefore we just present the final results . the spectra of the sequences @xmath98 are shown in fig . [ spectra111212 ] . at @xmath86 . ] it follows from this figure that the closeness of the corresponding spectrum to the parabola @xmath99 in the case of @xmath87 exists only in the vicinity of zero . the spectra at @xmath90 and @xmath96 are close to function @xmath100 in a wider range of @xmath93 ( @xmath101 and @xmath102 respectively ) . the difference between the real parts of the spectra of the sequences @xmath98 and the parabola @xmath103 are depicted in fig . [ log102 ] in the logarithmic scale . versus @xmath93 for different @xmath45 . ] this figure again demonstrates the high accuracy of the second derivative computation with the use of the sequences @xmath98 . in conclusion we note that the method of central differentiation formulae finding has been developed in this article . the elaborated technique does not require direct construction of the interpolating polynomial . we have derived simple and convenient expressions for the first and the second derivatives [ eqs . and ] of the function given in @xmath9 discrete points . the number @xmath45 was taken to be arbitrary . in contrast to the results of the ref . @xcite , where the recursion relations for the calculation of the weight coefficient being used in numerical differentiation formulae were considered , in the present work the expressions for the considered weight coefficients have been derived in the explicit form for the arbitrary number of interpolation points . as an example of the use of the developed method we have calculated the first derivative of the function @xmath10 . the obtained result has been studied in the limiting case @xmath11 . we have examined the spectral characteristics of the weight coefficients sequence of the numerical differentiation formulae for the different number of the interpolation points . the performed analysis has allowed one to study the accuracy of the numerical differentiation carried out with the help of the developed method . it has been found that the derived formulae of numerical differentiation posses the high accuracy in a rather wide range of the spatial frequencies . as it has been shown in this paper , the formulae for the derivatives finding gave correct results in the case of large number of interpolation points . thus , the developed method can be useful in lattice simulation of quantum fields @xcite . to get the exact results at calculations on lattices one has to use nets with the big number of points . derivatives which one encounters in theories of quantum fields , as a rule , do not exceed the second order . therefore , the formulae obtained in this article could be of use in carrying out mentioned above research . this research was supported by grant of russian science support foundation . the author is indebted to sergey i. dvornikov for helpful discussions . in this appendix we present the algorithm for the computation of the coefficients @xmath38 on the matlab 6.5 programming language .	we derived the formulae of central differentiation for the finding of the first and second derivatives of functions given in discrete points , with the number of points being arbitrary . the obtained formulae for the derivative calculation do not require direct construction of the interpolating polynomial . as an example of the use of the developed method we calculated the first derivative of the function having known analytical value of the derivative . the result was examined in the limiting case of infinite number of points . we studied the spectral characteristics of the weight coefficients sequence of the numerical differentiation formulae . the performed investigation enabled one to analyze the accuracy of the numerical differentiation carried out with the use of the developed technique . mathematics subject classification : primary 65d25 ; secondary 65t50
the past few years have seen a large increase in the interest for modeling dynamic interactions between individuals . while many real world data contain continuous - time information on the interactions , as e.g. email exchanges between employees in a company @xcite or face - to - face contact between individuals measured through sensors @xcite , most models are discrete in time . commonly , data are aggregated on predefined time intervals to obtain a sequence of snapshots of interaction random graphs . besides the loss of information induced by data aggregation , the specific choice of the time intervals has a direct impact on the results , which is most often overlooked . thus , developing models of interaction that exploit the continuous - time aspect of the data either called _ longitudinal networks , interaction event data , link streams _ or _ temporal networks _ is an important research issue . statistical methods for the analysis of longitudinal networks form a huge corpus , especially in social sciences and we do not pretend to provide an exhaustive bibliography on this topic . we refer to the very nice and recent review by @xcite for a more complete view on temporal networks . a natural way of modeling temporal event data is based on stochastic point processes . an important line of research involves continuous - time markov processes with seminal works on dyad - independent models @xcite up to the development of so - called stochastic actor oriented models ( e.g. * ? ? ? * ; * ? ? ? * ) . in these works observations consist in a series of time intervals of interaction and interactions are assumed to last during the whole corresponding time interval . here , we focus on a rather different setup where each interaction is identified with a time point . furthermore , we consider a model that allows for dependencies of the processes modeling the interactions of pairs of individuals . the analysis of event data is an old and important area in statistics ( see e.g. * ? ? ? generally a multivariate counting process @xmath0 is considered , that counts the number of interactions of each pair @xmath1 of individuals up to time @xmath2 . in @xcite counting processes have been introduced in the context of _ action _ data , which are a set of time - stamped directed interactions between individuals that , in addition , are marked by a label ( representing a behavioral event ) . the model may be viewed as an instance of cox s multiplicative hazard model with time - dependent covariates and constant baseline function . in the same vein , @xcite propose a general regression - based modeling of the intensity of non recurrent interaction events . they consider two different frameworks : cox s multiplicative and aalen s additive hazard rates ( see e. g. * ? ? ? @xcite propose another variant of cox s multiplicative intensity model for recurrent interaction events where the baseline function is specific to each individual . in the above mentioned works a set of statistics is chosen by the user as potential candidates that modulate the interactions . as in any regression framework , the choice of these statistics might raise some issues : increasing their number potentially leads to a high - dimensional problem , and interpretation of the results might be blurred by the correlation between these statistics . the approaches by @xcite , @xcite , @xcite and others are based on conditional poisson processes characterized by random intensities , also known as doubly stochastic poisson processes or cox processes . a particular instance of the conditional poisson process is the hawkes process , which is a collection of point processes with some background rate , where each event adds a nonnegative impulse to the intensity of all other processes . @xcite develop a model for spatial - temporal networks with missing information , based on such self - exciting point processes for temporal dynamics combined with a gaussian mixture for the spatial dynamics . similarly , @xcite combine temporal hawkes processes with latent distance models for implicit networks that can not be observed directly . clustering individuals based on interaction data represents a well - established technique for taking into account the intrinsic heterogeneity and summarizing information . in the context of dynamic random graphs , where a discrete - time sequence of graphs is observed , recent approaches propose to generalize the so - called stochastic block model to a dynamic context @xcite . stochastic block models posit that each individual belongs to a latent group and interactions between two individuals are conditionally independent of the interactions of any other pair , given the latent groups of the interacting individuals . another attempt to use stochastic block models in the context of interaction events appears in @xcite generalizing the approach of @xcite by adding discrete latent variables on the individuals . in this work a semiparametric stochastic block model for recurrent interaction events in continuous time is introduced , to which we refer as the poisson process stochastic block model . this is a stochastic block model where interactions are modeled by conditional inhomogeneous poisson processes , whose intensities only depend on the latent groups of the interacting individuals . in contrast to many other works , we do not rely on a parametric model where intensities are modulated by predefined network statistics , but intensities are modeled and estimated in a nonparametric way . the model is shown to be identifiable . our estimation and clustering approach is a semiparametric version of the variational expectation - maximization algorithm , where the maximization step is replaced by nonparametric estimators of the intensities . semiparametric generalizations of the classical expectation - maximization ( ` em ` ) algorithm have been proposed in many different contexts ( see e.g. @xcite for semiparametric mixtures or @xcite for a semiparametric hidden markov model ) . however , we are not aware of other attempts to incorporate nonparametric estimates in a variational approximation of ` em ` . two versions are developed for the nonparametric part of the model : a histogram approach based on the work of @xcite and a kernel estimator based on @xcite . for the histogram approach , an integrated classification likelihood criterion is proposed to select the number of latent groups adaptively . synthetic experiments enlighten both the clustering capacities of our method as well as the performance of the nonparametric estimation of the different intensities . moreover , the analysis of several real datasets illustrates the strengths and weaknesses of our approach . the supplementary material , whose references appear as s.xx , provides the proofs of all theoretical results , technical details on the algorithm and more detailed results of the analysis of the real data examples . we are interested in the pairwise interactions of @xmath3 individuals during some time interval @xmath4 $ ] . for notational convenience , we choose to restrict our attention to directed interactions without self - interactions . the undirected case as well as self - interactions are treated similarly and simulations as well as a real data example from section [ sec : real_data ] use the undirected setup . the set of all possible pairs of individuals , which is also the set of all possible dyads in the graph , is denoted by @xmath5 the cardinality of @xmath6 is @xmath7 . the observations @xmath8 are the interactions occurring in time interval @xmath4 $ ] , that is @xmath9 where @xmath10\times \mathcal r$ ] corresponds to the event that the individuals with indices @xmath11 and @xmath12 interact at time @xmath13 . the number of events in time interval @xmath4 $ ] is @xmath14 . we assume that @xmath15 , i.e. there is at most one event at a time . to model the distribution of these observations , every individual is assumed to belong to one out of @xmath16 groups , and the relation between two individuals , that is the way they interact with another , is driven by their group membership . more precisely , let @xmath17 be independent and identically distributed ( latent ) random variables taking values in @xmath18 with non zero probabilities @xmath19 for the moment , @xmath16 is considered to be fixed and known . when no confusion occurs , we also use the notation @xmath20 with @xmath21 such that @xmath22 has multinomial distribution @xmath23 with @xmath24 . now , our poisson process stochastic block model ( ppsbm ) is defined as follows . for every @xmath25 , the interactions of individuals @xmath26 and @xmath27 , conditional on the latent groups @xmath22 and @xmath28 , are modeled by a conditional inhomogeneous poisson process @xmath29 on @xmath4 $ ] with intensity depending only on the latent groups @xmath22 and @xmath28 . we consider nonnegative intensity functions @xmath30 with @xmath31 such that the conditional intensity of process @xmath29 , given that @xmath32 and @xmath33 , is @xmath34 for any @xmath25 . the corresponding cumulative intensities are denoted by @xmath35.\ ] ] the set of observations @xmath8 is a realization of the multivariate counting process @xmath36 with conditional intensity process @xmath37 . the process @xmath38 is not a poisson process , but a counting process with intensity @xmath39 . we denote @xmath40 the infinite - dimensional parameter of a poisson process stochastic block model . the distribution of the multivariate counting process @xmath41 under parameter value @xmath42 is denoted @xmath43 . concerning the identifiability of parameter @xmath42 from the distribution of the multivariate counting process @xmath41 , it is clear that at best the poisson process stochastic block model is identifiable up to label switching , as defined below . furthermore , as the functions @xmath30 are intensities , they are only identifiable almost everywhere on @xmath4 $ ] . we denote @xmath44 the set of permutations of @xmath45 . the parameter @xmath46 of a poisson process stochastic block model is identifiable on @xmath4 $ ] up to label switching if for all @xmath42 and @xmath47 such that @xmath48 , there exists a permutation @xmath49 such that @xmath50 , \qquad ( q , l=1,\dots , q).\ ] ] the following assumption ensures identifiability up to label switching in a very general setting . [ hyp : ident ] the set of intensities @xmath51 contains exactly @xmath52 distinct functions . the intensities @xmath30 may take identical values at some points or on subsets of @xmath4 $ ] , but should not be equal almost everywhere . [ prop : ident ] under assumption [ hyp : ident ] , the parameter @xmath53 is identifiable on @xmath4 $ ] , up to label switching , from the poisson process stochastic block model distribution of the multivariate counting process @xmath41 on the same interval , as soon as @xmath54 . this result strongly relies on the only available identifiability result for weighted stochastic block models , namely theorem 12 in @xcite , which can be applied under assumption [ hyp : ident ] . one may wonder whether the necessary condition that any two rows ( or any two columns ) of the parameter matrix @xmath55 are distinct is , in fact , a sufficient condition for identifiability . however , to our knowledge such a result has never been established even in the simple binary case . in the binary stochastic block model , the results in @xcite establish _ generic _ identifiability , which means identifiability except on a subset of parameters with lebesgue measure zero , without specifying the exceptional subset . for the directed and binary stochastic block model , @xcite establish identifiability under the assumption that the product vector @xmath56 ( or @xmath57 ) has distinct coordinates . this condition is slightly stronger than the one previously mentioned . another partial identifiability result appears in @xcite for some block models . these last two approaches are specifically adapted to the discrete setup ( maybe even to the binary one ) and can not be generalized to the continuous case . proposition [ prop : ident ] does not cover the affiliation case , where only two intensities @xmath58 and @xmath59 are considered such that for all @xmath60 @xmath61 [ prop : ident_affil ] if the intensities @xmath62 and @xmath63 are distinct functions on @xmath4 $ ] , then both @xmath62 and @xmath63 are identifiable on @xmath4 $ ] from the affiliation poisson process stochastic block model distribution of the multivariate counting process @xmath41 on the same interval , as soon as @xmath54 . moreover , for any @xmath64 , the proportions @xmath65 are also identifiable , up to a permutation , from the same distribution . in this section we introduce processes and notation that will be used throughout the manuscript . first , for any @xmath66 , we consider the ( unobserved ) number of dyads @xmath67 with latent groups @xmath60 @xmath68 the ( unobserved ) counting process @xmath69 has conditional intensity @xmath70 and falls in the class of aalen s multiplicative intensity models . this is a central property on which our work often relies . we also define @xmath71 as the ( unobserved ) binary indicator of observation @xmath72 belonging to groups @xmath60 by @xmath73 as these quantities are unobserved , our work relies on proxies . we consider the set @xmath74 of candidate proxies for the unobserved latent groups @xmath75 given by @xmath76 , \sum_{q=1}^q\tau^{i , q}=1 ~~\text { for } i=1,\dots , n , q=1,\dots , q \right\}.\ ] ] while the latent variables @xmath75 are indicators , their counterparts @xmath77 are weights representing the probability that node @xmath26 belongs to group @xmath78 . now , for every @xmath79 , replacing all latent variables @xmath75 in by @xmath77 , we define @xmath80 , @xmath81 and @xmath82 which are estimators of @xmath83 , @xmath84 and @xmath85 , respectively . the complete - data likelihood of observation @xmath8 and latent variables @xmath86 is @xmath87 the likelihood of the observed data @xmath88 is obtained by summing the complete - data likelihood over the set of all possible configurations of the latent variables @xmath89 . this set is so huge that the likelihood of the observed data is intractable for direct maximization . hence , an expectation - maximization ( ` em ` ) algorithm @xcite is used , which is an iterative procedure especially adapted to cope with latent variables . the ` em ` algorithm consists of an ` e`-step and an ` m`-step that are iterated until convergence . in our model two different issues arise . first , as already observed for the standard stochastic block model @xcite , the ` e`-step requires the computation of the conditional distribution of @xmath90 given the observations @xmath8 , which is not tractable . therefore , we use a variational approximation @xcite of the latent variables conditional distribution to perform the ` e`-step . we refer , for instance , to @xcite for a general description of the variational ` em ` algorithm and its links to ` em ` in stochastic block models . second , part of our parameter is infinite dimensional so that the ` m`-step is partly replaced by a nonparametric estimation procedure , giving rise to a semiparametric ` em ` algorithm . our complete algorithm is summarized in section [ sec : vem ] , algorithm [ algo : vem ] . the standard ` e`-step consists in computing the expectation of the complete log - likelihood given the observations at some current parameter value @xmath42 . this requires the knowledge of the conditional latent variables distribution @xmath91 , which is not tractable , mainly because the latent variables @xmath22 are not conditionally independent . the idea is to perform a variational approximation of the conditional latent variables distribution @xmath92 by a simpler distribution . more precisely , using the class of parameters @xmath93 defined in , we consider for every @xmath79 the conditional factorized distribution @xmath94 of @xmath90 given @xmath8 defined by @xmath95 with corresponding expectation @xmath96 . then , we search for the parameter @xmath97 that yields the best approximation @xmath94 of @xmath92 . more precisely , @xmath98 where @xmath99 denotes the kullback - leibler divergence . the variational ` e`-step is completed by the computation of the current expected complete data log - likelihood @xmath100 from a practical point of view , it can be shown that the solution @xmath101 of is also the solution of a fixed point equation , which in practice is found by successively updating the variational parameters @xmath77 via the following equation until convergence . [ prop : e_step ] the solution @xmath101 to the minimization problem given in satisfies the following fixed - point equation @xmath102,\qquad ( i=1,\dots , n,~q=1,\dots , q),\ ] ] where @xmath103 means proportional to and @xmath104 with @xmath105 the indicator function of set @xmath106 . in a parametric context , the ` m`-step consists in the maximization of @xmath107 with respect to @xmath53 . considering only the finite - dimensional part @xmath108 of the parameter , we easily obtain that the maximizer @xmath109 of @xmath110 with respect to @xmath108 is @xmath111 concerning the infinite - dimensional parameter @xmath55 , we replace the maximization of @xmath110 with respect to @xmath55 by a nonparametric estimation step . in the following , we develop two different approaches for updating @xmath55 : a histogram and a kernel method . in both cases estimation would be straightforward using the process @xmath112 defined by , which unfortunately is not observed . it is thus natural to use its ( current ) variational approximation , namely the weighted cumulative process @xmath113 defined in section [ sec : proc_notation ] . in this part the intensities @xmath30 are estimated by piecewise constant functions and we propose a data - driven choice of the partition of the time interval @xmath4 $ ] . the procedure is based on a least - squares penalized criterion following the work of @xcite . the detailed construction is provided in the supplementary material . for @xmath114 , where @xmath115 is to be chosen , we denote by @xmath116 the regular partition of @xmath4 $ ] into @xmath117 intervals with length @xmath118 , namely @xmath119 one may also use regular dyadic partitions , where @xmath120 , defining nested models that have some practical advantages concerning the implementation of the algorithm . in the following let @xmath121 be fixed . for @xmath122 the estimated mean number of observed interactions between individuals @xmath72 with latent groups @xmath60 occurring in time interval @xmath123 is @xmath124 for any fixed value of @xmath117 , a projection estimator @xmath125 on the space of piecewise constant functions on @xmath116 is given by @xmath126 where @xmath127 is the length of interval @xmath128 . now , adaptive estimation consists in choosing the best estimator among the collection of estimators @xmath129 . so we introduce an estimator of the partition through @xmath130 that minimizes a penalized least - squares criterion , that simplifies to @xmath131 the selected partition size @xmath130 depends on the groups @xmath60 and may be different for different values of @xmath60 . finally , the adaptive estimator of intensity @xmath30 is @xmath132 @xcite develops her approach in the aalen multiplicative intensity model , which is slightly different from our context . moreover , our setup does not satisfy the assumptions of theorem 1 in @xcite , since the number of jumps of the processes @xmath38 is not bounded by a known positive number , because here the @xmath38 are counting processes . nevertheless , in our simulations this procedure successfully estimates the intensities @xmath30 ( see section [ sec : simus ] ) . we refer to @xcite for a theoretical study of an adaptive nonparametric estimation of the intensity of a poisson process . @xcite also studies other penalized least squares estimators ( for e.g. using fourier bases ) , which might be used here similarly . an alternative way for nonparametric intensity estimation is based on kernel estimators , that are explored in the following section . kernel methods are suited to estimate smooth functions . in this part kernel estimators of the intensities @xmath30 are provided . a similar procedure has been proposed for a non variational version of the ` em ` algorithm in @xcite . if the variational parameters @xmath77 are good approximations of the latent variables @xmath75 , then the intensity of process @xmath81 defined in section [ sec : proc_notation ] is approximately @xmath133 , where @xmath80 is the variational mean number of dyads with latent groups @xmath60 . following @xcite and considering a nonnegative kernel function @xmath134 with support within @xmath135 $ ] together with some bandwidth @xmath136 , the intensity @xmath30 is estimated by @xmath137 if @xmath138 and @xmath139 otherwise , where @xmath82 is defined in section [ sec : proc_notation ] . the bandwidth @xmath140 can be chosen adaptively from the data following the procedure proposed by @xcite . kernel methods are not always suited to infer a function on a bounded interval as boundary effects may deteriorate their quality . however , it is out of the scope of this work to investigate refinements of this kind . during the implementation of the algorithm , two issues arise : convergence and initialization . as our algorithm is an iterative procedure , one has to test for convergence . a stopping criterion can be defined based on the current expected complete data log - likelihood @xmath141}}(\theta^{[s]})$ ] . concerning initialization the algorithm be may run several times with different starting values . one can choose them randomly or by some k - means method . see the supplementary material for details . algorithm [ algo : vem ] provides a full description of the procedure . @xmath142 initialize @xmath143}$ ] output @xmath144},\alpha^{[s]})$ ] to choose the best number of groups @xmath16 , we propose an integrated classification likelihood criterion that performs data - driven model selection . roughly , this criterion is based on the complete data log - likelihood penalized by the number of parameters . it has been introduced in the mixture context in @xcite and adapted to the stochastic block model in @xcite . the issue here is that our model contains a nonparametric part , so that the parameter is infinite dimensional . however , in the case of histogram estimators , once the partition is selected , there is only a finite number of parameters to estimate , which can be used to build our integrated classification likelihood criterion . more precisely , for any @xmath16 let @xmath145 be the estimated parameter with @xmath16 groups and @xmath146 the corresponding maximum a posteriori classification at @xmath145 obtained by our variational ` em ` algorithm . the parameter @xmath147 has two components : the first one @xmath148 is a vector of dimension @xmath149 , while the second has dimension @xmath150 , where @xmath151 denotes the size of the partition used in the histogram estimator @xmath152 . in the adaptation of the integrated classification likelihood criterion to the stochastic block model these components are treated differently : the first one , that concerns the @xmath3 individuals , is penalized by a @xmath153 term , while the second one concerning the dyads is penalized by a @xmath154 term . we refer to @xcite for more details . in our case , the integrated classification likelihood criterion is @xmath155 after fixing an upper bound @xmath156 we select the number of groups @xmath157 in this section we investigate the numerical performance of our method for clustering individuals and estimating the intensities of the inhomogeneous poisson processes . we also study the performance of the integrated classification likelihood criterion for recovering the true number of latent groups . the following two scenarios are used in our simulations in the undirected setup where @xmath158 for any @xmath66 . 1 . we consider the affiliation model with @xmath159 latent groups and equal group probabilities @xmath160 . to evaluate the classification performance , the intensities are sinusoids with varying shifting parameter @xmath161 . clustering is supposed to be more difficult for small values of @xmath161 . the intensities are shown in figure [ fig_intensities_q2 ] in the supplementary material . the number of individuals @xmath3 varies in @xmath162 . 2 . to evaluate the intensity estimators , we consider a poisson process stochastic block model with @xmath163 groups with equal probabilities @xmath164 . the six intensity functions have rather different shapes and amplitudes ( see figure [ estimated_intensities_q3_n20 ] ) . the number of individuals @xmath3 varies in @xmath165 . for every setting , @xmath166 datasets are simulated under the corresponding poisson process stochastic block model and the variational ` em ` algorithm is applied . the histogram estimator is applied with a regular partition and @xmath167 , while the kernel estimator uses the epanechnikov kernel . . left : @xmath168 , right : @xmath169.,title="fig:",scaledwidth=40.0%,height=151 ] . left : @xmath168 , right : @xmath169.,title="fig:",scaledwidth=40.0%,height=151 ] to assess the clustering performance , we use the adjusted rand index @xcite that evaluates the agreement between the estimated and the true latent structure . for two classifications that are identical ( up to label switching ) , this index equals @xmath170 , otherwise the adjusted rand index is smaller than @xmath170 and negative values are possible . figure [ boxplot_affiliation ] shows the boxplots of the adjusted rand index obtained with the histogram and the kernel versions of our method in scenario 1 . for small values of the shifting parameter ( @xmath171 ) , the intensities are so close that the classification is very difficult , especially when @xmath172 is small . the classification improves when the shift between the intensities and/or the number of observations increase , achieving ( almost ) perfect classification for large values of @xmath161 and/or @xmath3 . we also observe that the kernel version of our method gives better classification results than the histogram method , which might be due to the choice of actually continuous intensities . concerning the recovery of the intensities in scenario 2 , the quadratic risk is used to measure the distance between the true intensity @xmath30 and its estimate @xmath173 defined by @xmath174 table [ risk_q3 ] gives the mean value of the risk @xmath175 and its standard deviation estimated over @xmath166 repetitions for both the histogram and the kernel version of our method . table [ risk_q3 ] also reports the mean number of observations @xmath72 with latent groups @xmath60 , namely @xmath176 . moreover , figure [ estimated_intensities_q3_n20 ] shows for each pair of groups @xmath60 the true and the estimated intensities for one dataset with @xmath177 . . mean number of events with latent groups @xmath60 and mean quadratic risk ( with standard deviation ) for the histogram and the kernel estimators of the intensities in scenario 2 averaged over @xmath178 repetitions . all values associated with the risk are multiplied by 100 . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] . true intensities ( black continuous ) , histogram estimator ( red dashed ) and kernel estimator ( blue dotted ) for each pair of groups @xmath60.,height=264 ] as expected , when the true intensity is piecewise - constant , the histogram version of our method outperforms the kernel estimator . conversely , when the true intensity is smooth , the kernel estimator is more appropriate to recover the shape of the intensity . in some cases , as e.g. for the intensity with latent groups @xmath179 , the estimators achieve comparable results . a well - known drawback of the kernel estimator is that it suffers from boundary effects . this is observed here for the intensities with groups @xmath180 and @xmath181 , but is less crucial for the other intensities that appear to be null at the interval boundaries . finally , we use scenario 2 to illustrate the performance of the integrated classification likelihood criterion to select the number @xmath16 of latent groups from the data . for each of the 1000 simulated datasets , the maximizer @xmath182 of the integrated classification likelihood criterion defined in with @xmath183 is computed . results are reported in figure [ qbest-20 ] in the supplementary material . for @xmath177 the correct number of groups is recovered in @xmath184 of the cases . moreover , when the criterion does not select the correct number @xmath16 , the adjusted rand index of the classification with three groups is rather low . this indicates that in those cases the classification obtained with three groups is not the correct one , so that rather the variational em algorithm is to blame for bad results than the integrated classification likelihood criterion . for @xmath185 our procedure selects the correct number of groups for each simulated datasets . here , we use the cycle hire usage data from the bike sharing system of the city of london from 2012 to 2015 @xcite . these data are also analyzed in @xcite with a different perspective . we focus on two randomly chosen weekdays , which are february 1st , 2012 ( day 1 ) and february 2nd , 2012 ( day 2 ) . data consist in pairs of stations associated with a single hiring / journey ( departure station , ending station ) and corresponding time stamp ( hire time with second precision ) . the datasets have been pre - processed to remove journeys that either correspond to loops , last less than 1 minute or more than 3 hours or do not have an ending station ( lost or stolen bikes ) . the datasets contain @xmath186 and @xmath187 stations on day 1 and day 2 with @xmath188 and @xmath189 hire events respectively . with more than 170,000 oriented pairs of stations the number of processes @xmath38 is huge , but only a very small fraction around 7% of these point processes are non null ( i.e. contain at least one hiring event between these stations ) . this is to be expected as bike sharing systems are mostly used for short trips and stations far one from another are unlikely to be connected . as data correspond to origin / destination flows , it is natural to work with a directed setup and we applied the histograms version of our algorithm on a dyadic partition with maximum size @xmath190 . the integrated classification likelihood criterion achieves its maximum with @xmath191 latent groups for day 2 and @xmath192 on day 1 . in order to compare results across the two datasets and keep interpretation simple , we focus on the classification obtained with @xmath193 clusters . geographic locations of the bike stations and the clusters are represented on a city map ( thanks to the openstreetmap project ) , see figure [ fig : cycles_day2 ] in the supplementary material for day 2 . clusters for day 1 are very similar , so that in the following we only concentrate on day 2 . we observe that our procedure globally recovers geographic clusters , as stations are expected to be mainly linked through geographic proximity in the datasets . a closer look at the clusters then reveals more information . there is one cluster containing only four bike stations ( cluster number 5 , appearing as light blue diamonds in figure [ fig : cycles_day2 ] ) , while all other clusters contain between 38 and 125 stations . this small cluster contains one bike station at kings cross railway station and three stations next to waterloo railway station . indeed , two of these bike stations are among those with the highest activities ( for both departures and arrivals ) in comparison to all other stations , while this is not the case of the other two stations in the cluster . thus it is very unlikely that a snapshot approach , where events are aggregated over a predetermined time window , would have yield the same cluster . consequently , the explanation for this clustering is the similarity of the temporal profiles of these four stations . indeed , figure [ fig : cycles_temp_2 ] shows that these four stations are outgoing stations in the morning with much more departures than arrivals around 8 a.m. and incoming stations at the end of the day , with more arrivals than departures between 5p.m and 7 p.m. looking at the temporal profiles of stations close to the two other main railway stations in london ( victoria and liverpool street stations ) , this pattern is not observed and the stations are clustered differently . thus , this small cluster is characterized by stations used by people living in the suburbs and working in the city center . this result highlights the specificity of our model that is able to find clusters from similar temporal profiles in sharp contrast with aggregated data approaches . ) and incoming ( @xmath194 ) processes from the 4 stations @xmath26 in the smallest cluster : representation of volumes of connections to all other stations during day 2.,scaledwidth=80.0%,height=377 ] we then used a kernel estimator of the intensities per ( directed ) groups pairs @xmath195 ( see figure [ fig : cycles_intens ] in the supplementary material ) . cluster number 5 ( which is the small cluster mentioned above ) has high ( directed ) intensities of connections with cluster number 3 ( shown in green plus sign @xmath196 in figure [ fig : cycles_day2 ] ) . this last cluster groups the stations belonging to the business city center of london . we observe a large intensity of connections from cluster 5 ( king s cross and waterloo railway stations ) to cluster 3 ( city business center ) in the morning and in the other direction ( from cluster 3 to cluster 5 ) at the end of the day . moreover , cluster 5 also appears to have ( a smaller amount of ) connections with clusters 1,4 and itself . connections with cluster 1 ( black circles @xmath197 in figure [ fig : cycles_day2 ] ) and cluster 4 ( blue cross @xmath198 in figure [ fig : cycles_day2 ] ) are similar ( from cluster 5 in the morning and back to cluster 5 in the evening ) to those with cluster 3 but at a smaller scale . those results support and extend the previous description of the role of cluster 5 . to conclude this section we mention that @xcite use a completely different approach , relying on poisson mixture models on the same origin / destination flows . this approach does not take into account the network structure of the data ( where e.g. two flows from the same station are related ) . as a consequence , clusters are obtained on pairs of stations from which interpretation is completely different and in a way less natural . to understand contacts between children at school and to quantify the transmission opportunities of respiratory infections , data on face - to - face interactions in a french primary school were collected . the dataset is presented in detail in @xcite and available online @xcite . children are aged from 6 to 12 years and the school is composed of five grades , each of them comprising two classes , for a total of 10 classes ( denoted by @xmath199 ) . each class has an assigned teacher and an assigned room . the school day runs from 8.30am to 4.30pm , with a lunch break from 12 pm to 2 pm and two breaks of 20 - 25 min around 10.30am and 3.30pm . lunch is served in a common canteen and a shared playground is located outside the main building . as the playground and the canteen do not have enough capacity to host all pupils at a time , only two or three classes have breaks together , and lunch is served in two turns . the dataset contains @xmath200 face to face contacts among @xmath201 individuals ( @xmath202 children and @xmath203 teachers ) observed during two days . we applied our procedure in the undirected setup with histograms based on a dyadic partition with maximum size @xmath204 . for @xmath205 , figure [ fig_school_tau ] shows the clustering of the @xmath3 individuals into @xmath16 groups , where children from different classes are represented in different colors . when q is small ( @xmath206 ) , our procedure gathers all pupils from one class and their corresponding teacher in the same cluster . for larger values of @xmath16 , our procedure makes a sharper clustering according to the behavior of the children . for example for @xmath207 , the procedure separates children from the same class : either to isolate a few of them in a group ( 3 children of class @xmath208 are put together in one group ) , or to put together children of different classes ( one group is made of children of classes @xmath209 and @xmath210 ) . teachers never form a particular group apart , but they are in the cluster of their assigned class , suggesting that contacts among teachers are sparse and that in this dataset clustering is mainly driven by communities ( _ i.e. _ groups of highly connected individuals , with few inter - groups interactions ) . the model selection criterion for choosing the best number of groups @xmath16 does not provide a reasonably small number of clusters that could be used for interpretation of the data . it has already been observed by other authors that this may happen for large datasets ( see * ? ? ? * and the references therein ) . thus , we choose to further analyse the data for @xmath207 groups . individuals ( represented by different colours ) into @xmath211 groups . for each picture the vertical bars represent the @xmath16 clusters . colours indicate the grades and the teachers , plain and hatching distinguish the two classes in the same grade . , height=377 ] we observe that the intensities representing the most activity are the intra - group intensities . as clusters mainly correspond to classes , this highlights that most contacts involve children of the same class and that the dataset is structured into communities . moreover , peaks of interactions are observed during the two breaks around 10.30am and 3.30pm . at lunch time interactions between children vary from the first to the second day and are less important than during the breaks where they play together . concerning inter - group connections , most of the estimated intensities for groups @xmath60 with @xmath212 can be considered as null , except for some that we discuss now . first , as our procedure splits some children of the same class into separate groups , the inter - group interactions associated with these clusters correspond in fact to intra - class interactions . second , intensities between groups made of children of the same grade are significant , suggesting that children mostly interact with children of the same age . third , the children of class @xmath213 are partitioned into two clusters , and the intensity of one of these clusters drops to zero during lunch time , whereas children of the other group interact a lot during lunch . it seems that our procedure has recognized two subgroups in class @xmath213 : children having lunch at school and those going home for lunch . fourth , class @xmath208 is split into three groups with 20 , 3 and 2 pupils , respectively . the estimated intensities suggest a particular behaviour of some of the children : there is no contact between the two children in the smallest cluster , but they have very strong interaction with the three pupils in the other cluster . as a conclusion , we recover many results of @xcite . in particular , we detect subgroups of pupils with a specific behavior at some period of the day ( leaving school for lunch ) . we think that this is mainly due to the fact that the poisson process stochastic block model takes into account the information provided by the timestamps of the events . without this temporal information , it is hard to imagine to obtain similar results . in particular , aggregating data on a day scale to construct discrete time networks would not provide such a refined analysis . we analyse the email exchanges between @xmath214 persons working at enron , mostly in the senior management , covering the period of the affair that led to the bankruptcy of the company @xcite . the dataset @xcite contains @xmath215 emails exchanged among these 147 persons between june 14 , 2000 and june 13 , 2002 , for which the sender , the recipient and the time when the email was sent are known . as data are obviously directed , we use the directed version of our method to analyse them . here again , the integrated classification likelihood criterion does not provide a reasonably small enough number of clusters that could be used for interpretation . we choose to analyse the data with @xmath216 clusters . there is one group which is rather silent with very little activity and intensities close to zero . the other groups are characterized by substantial intra - group communication . it is instructive to compare the temporal profiles of the estimated intensities . first , we observe that communication is not symmetric . for example , people in cluster 3 regularly send emails to cluster 2 over the whole observation period , but the latter only respond rather late during the second half of 2001 , see figure [ fig_enron_intra ] . second , the intra - group communication differs a lot from one group to the other , see figure [ fig_enron_inter ] . cluster 3 has a rather constant communication intensity over the whole period , while the intra - group intensity of cluster 1 is increasing with a peak at the end of 2002 , which seems to be the consequence of the beginning of the investigations . in contrast , cluster 4 has an important intra - group activity until june 2001 , then the intensity drops down and achieves another peak just before the inquiry , when it is known that number of individuals acted to hide sensitive information on the scandal . thus , one may suspect people from cluster 4 to hold relevant information for the investigators . finally , we compare these results with those obtained using a classical stochastic block model . indeed , taking @xmath217 in our approach amounts to forget the timestamps of the emails , as the algorithm then only considers email counts over the whole observation period . in other words , using @xmath217 boils down to a classical stochastic block model with poisson emission distribution and mean parameter @xmath218 ( see for instance * ? ? ? * ) . we compare the classifications obtained by the two procedures for @xmath216 clusters . the associated adjusted rand index is @xmath219 indicating that part of the individuals are clustered in the same way and some are treated very differently in the two models . indeed , both methods find a large common cluster with people that do not communicate a lot , while most hesitation is about persons in cluster 3 in the poisson process stochastic block model , which are mainly split into two groups in the classical stochastic block model and partly mixed with people from cluster 1 . in fact , cluster 1 and 3 in the poisson process stochastic block model have significant intra - group communication but very different temporal profiles as mentioned above . intuitively , it may be difficult to distinguish these groups when we only look at the total count data as in classical stochastic block model . this means that taking into account the time information of the events may be very useful to improve the classification of the individuals compared to classical stochastic block model . we would like to thank agathe guilloux for pointing out valuable references , nathalie eisenbaum for her help on doubly stochastic counting processes and pierre latouche for sharing information on datasets . 43 natexlab#1#1 allman , e. , matias , c. & rhodes , j. ( 2009 ) . identifiability of parameters in latent structure models with many observed variables . statist . _ * 37 * , 30993132 . allman , e. , matias , c. & rhodes , j. ( 2011 ) . parameters identifiability in a class of random graph mixture models . _ j. stat . plan . inference _ * 141 * , 17191736 . andersen , p. k. , borgan , . , gill , r. d. & keiding , n. ( 1993 ) . _ statistical models based on counting processes_. springer series in statistics . springer - verlag , new york . baraud , y. & birg , l. ( 2009 ) . estimating the intensity of a random measure by histogram type estimators . theory related fields _ * 143 * , 239284 . bickel , p. j. , chen , a. & levina , e. ( 2011 ) . the method of moments and degree distributions for network models . statist . _ * 39 * , 22802301 . biernacki , c. , celeux , g. & govaert , g. ( 2000 ) . assessing a mixture model for clustering with the integrated completed likelihood . _ ieee trans . pattern anal . machine intel . _ * 22 * , 719725 . bhning , d. ( 1995 ) . a review of reliable maximum likelihood algorithms for semiparametric mixture models . _ j. stat . plan . inference _ * 47 * , 5 28 . bordes , l. , chauveau , d. & vandekerkhove , p. ( 2007 ) . a stochastic em algorithm for a semiparametric mixture model . data anal . _ * 51 * , 5429 5443 . butts , c. t. ( 2008 ) . a relational event framework for social action . methodol . _ * 38 * , 155200 . calo project ( 2015 ) . http://www.cs.cmu.edu/ enron/. celisse , a. , daudin , j .- j . & pierre , l. ( 2012 ) . consistency of maximum - likelihood and variational estimators in the stochastic block model . _ electron . j. statist . _ * 6 * , 18471899 . cho , y .- s . , galstyan , a. , brantingham , p. j. & tita , g. ( 2014 ) . latent self - exciting point process model for spatial - temporal networks . _ discrete continuous dyn . b _ * 19 * , 13351354 . corneli , m. , latouche , p. & rossi , f. ( 2016 ) . exact icl maximization in a non - stationary temporal extension of the stochastic block model for dynamic networks . _ neurocomputing _ * 192 * , 81 91 . dannemann , j. ( 2012 ) . semiparametric hidden markov models . _ j. comput . graph . statist . _ * 21 * , 677692 . daudin , j .- , picard , f. & robin , s. ( 2008 ) . a mixture model for random graphs . _ statist . comput . _ * 18 * , 173183 . dempster , a. p. , laird , n. m. & rubin , d. b. ( 1977 ) . maximum likelihood from incomplete data via the em algorithm . _ j. roy . ser . b _ * 39 * , 138 . dubois , c. , butts , c. t. & smyth , p. ( 2013 ) . stochastic blockmodeling of relational event dynamics . in _ aistats _ , vol jmlr workshop and conference proceedings . grgoire , g. ( 1993 ) . least squares cross - validation for counting process intensities . j. statist . _ * 20 * , pp . 343360 . guigours , r. , boull , m. & rossi , f. ( 2015 ) . discovering patterns in time - varying graphs : a triclustering approach . data anal . classif . _ , 128 . holme , p. ( 2015 ) . modern temporal network theory : a colloquium . j. b _ * 88 * , 234 . hubert , l. & arabie , p. ( 1985 ) . comparing partitions . _ j. classif . _ * 2 * , 193218 . jordan , m. , ghahramani , z. , jaakkola , t. & saul , l. ( 1999 ) . an introduction to variational methods for graphical models . * 37 * , 183233 . klimt , b. & yang , y. ( 2004 ) . the enron corpus : a new dataset for email classification research . in _ machine learning : ecml 2004 _ , j .- f . boulicaut , f. esposito , f. giannotti & d. pedreschi , eds . lecture notes in computer science_. springer berlin heidelberg . linderman , s. & adams , r. ( 2014 ) . discovering latent network structure in point process data . in _ proceedings of the 31st international conference on machine learning ( icml-14 ) _ , e. p. xing & t. jebara , eds . mariadassou , m. , robin , s. & vacher , c. ( 2010 ) . uncovering latent structure in valued graphs : a variational approach . stat . _ * 4 * , 71542 . martinussen , t. & scheike , t. h. ( 2006 ) . _ dynamic regression models for survival data_. statistics for biology and health . springer , new york . matias , c. & miele , v. ( to appear ) . statistical clustering of temporal networks through a dynamic stochastic block model . _ j. r. stat . _ available at arxiv:1506.07464 . matias , c. & robin , s. ( 2014 ) . modeling heterogeneity in random graphs through latent space models : a selective review . _ esaim proc . & surveys _ * 47 * , 5574 . perry , p. o. & wolfe , p. j. ( 2013 ) . point process modelling for directed interaction networks . _ j. r. stat . . b. stat . methodol . _ * 75 * , 821849 . ramlau - hansen , h. ( 1983 ) . smoothing counting process intensities by means of kernel functions . _ * 11 * , pp . 453466 . randriamanamihaga , a. n. , cme , e. , oukhellou , l. & govaert , g. ( 2014 ) . clustering the vlib dynamic origin / destination flows using a family of poisson mixture models . _ neurocomputing _ * 141 * , 124 138 . reynaud - bouret , p. ( 2006 ) . penalized projection estimators of the aalen multiplicative intensity . _ bernoulli _ * 12 * , 633661 . robin , s. , bar - hen , a. , daudin , j .- j . & pierre , l. ( 2007 ) . a semi - parametric approach for mixture models : application to local false discovery rate estimation . _ data anal . _ * 51 * , 5483 5493 . snijders , t. & van duijn , m. ( 1997 ) . simulation for statistical inference in dynamic network models . in _ simulating social phenomena _ , r. conte , r. hegselmann & p. terna , eds . 456 of _ lecture notes in economics and mathematical systems_. springer berlin heidelberg , pp . 493512 . snijders , t. a. b. , koskinen , j. & schweinberger , m. ( 2010 ) . maximum likelihood estimation for social network dynamics . stat . _ * 4 * , 567588 . sociopatterns ( 2015 ) . http://www.sociopatterns.org/. stehl , j. , voirin , n. , barrat , a. , cattuto , c. , isella , l. , pinton , j .- f . & et al . ( high - resolution measurements of face - to - face contact patterns in a primary school . _ plos one _ * 6 * , e23176 . transport for london ( 2016 ) . cycle hire usage data 2012 - 2015 . vu , d. q. , hunter , d. , smyth , p. & asuncion , a. u. ( 2011 ) . continuous - time regression models for longitudinal networks . in _ adv neural inf process syst 24 _ , j. shawe - taylor , r. zemel , p. bartlett , f. pereira & k. weinberger , eds . curran associates , inc . , pp . 24922500 . wasserman , s. ( 1980 ) . analyzing social networks as stochastic processes . _ jasa _ * 75 * , 280294 . wasserman , s. ( 1980 ) . a stochastic model for directed graphs with transition rates determined by reciprocity . methodol . _ * 11 * , pp . 392412 . xu , k. & hero , a. ( 2014 ) . dynamic stochastic blockmodels for time - evolving social networks . _ ieee j. sel . topics signal process . _ * 8 * , 552562 . yang , t. , chi , y. , zhu , s. , gong , y. & jin , r. ( 2011 ) . detecting communities and their evolutions in dynamic social networks a bayesian approach . _ * 82 * , 157189 . supplementary material for : a semiparametric extension of the stochastic block model for longitudinal networks + _ all the references are from the main manuscript , except for those appearing as s - xx that are within this file . _ for notational convenience , the proofs are presented in the undirected setup , where the set of intensities is @xmath220 . the directed case can be treated in the same way . we start by considering the distribution of one marginal process @xmath38 , which is a cox process directed by the random measure @xmath221 such that @xmath222 ( here , for any @xmath223 , we use the notation @xmath224 for the measure on @xmath4 $ ] defined by @xmath225 for all measurable @xmath226 $ ] . we also recall that @xmath227 is the dirac mass at point @xmath228 ) . it is known that the mapping of probability laws of random measures into laws of cox processes directed by them is a bijection ( see for example proposition 6.2.ii in * ? ? ? in other words , here the distribution of @xmath38 uniquely determines the finite measure ( on the set of measures on @xmath4 $ ] ) @xmath229 . then , under assumption 1 that the intensities @xmath30 are distinct , the corresponding measures @xmath224 are all different and we may recover from the distribution of our counting process @xmath38 the set of values @xmath230 or equivalently the set @xmath231 . in particular , we recover the functions @xmath30 almost everywhere on @xmath232 $ ] , up to a permutation of these @xmath233 values . however , to recover those values up to a permutation in @xmath44 , it is necessary to consider higher - order marginals . we now fix three distinct indices @xmath234 and consider the trivariate counting process @xmath235 . in the same way , these are cox processes directed by the triplet of random measures @xmath236 such that @xmath237 we write this distribution in such a way that distinct components appear only once @xmath238 \nonumber\\ & + \sum_{\substack{q , l , m \\ \|\{q , l , m\}\|=3 } } \pi_q\pi_l\pi_m \delta_{(a^{(q , l ) } , a^{(q , m ) } , a^{(l , m ) } ) } . \end{aligned}\ ] ] using the same reasoning , we identify the triplets of values @xmath239 up to a permutation on the triplets @xmath240 . among these , the only values with three identical components are @xmath241 and thus the measures @xmath242 are identifiable , up to a permutation in @xmath44 . going back to and looking for the dirac terms at points that have two identical components ( of the form @xmath243 and two other with permuted components ) , we can now identify the set of measures @xmath244 this is equivalent to saying that we identify the measures @xmath245 up to a permutation in @xmath44 . obviously , this also identifies the corresponding intensities @xmath246 almost everywhere on @xmath4 $ ] , up to a permutation in @xmath44 . to finish the proof , we need to identify the proportions @xmath247 . note that as we identified the components @xmath248 , we recover from the set of values @xmath249 up to the same permutation as on the @xmath250 s . this concludes the proof . we follow some of the arguments already appearing in the proof of proposition 1 . let @xmath251 ( resp . @xmath252 ) denote the measure whose intensity is @xmath58 ( resp . @xmath59 ) the univariate process @xmath38 is a cox process directed by the random measure @xmath221 that is now distributed as @xmath253 thus the measures @xmath251 and @xmath252 are identifiable from the distribution of @xmath38 , but only up to a permutation . once again , we rather consider the trivariate cox processes @xmath235 directed by the random measures @xmath236 whose distribution in the affiliation case has now five atoms @xmath254 as previously , these five components are identifiable , up to a permutation on @xmath255 . now it is easy to identify the three components for which two marginals have same parameters and the third one has a different parameter . thus , we recover exactly the measures @xmath251 and @xmath252 . this also identifies the corresponding intensities @xmath58 and @xmath59 almost everywhere on @xmath4 $ ] . now , the identification of the proportions @xmath256 follows an argument already used in the proof of theorem 13 in @xcite that we recall here for completeness . from the trivariate distribution of @xmath235 and the already recovered values @xmath251 and @xmath252 , we identify the proportion @xmath257 . similarly , for any @xmath258 , by considering the multivariate distribution of @xmath259 , we can identify the dirac mass at point @xmath260 and thus its weight , which is equal to @xmath261 . by the newton identities , the values @xmath262 determine the values of elementary symmetric polynomials @xmath263 . these , in turn , are ( up to sign ) the coefficients of the monic polynomial whose roots ( with multiplicities ) are precisely @xmath264 . thus , the proportion parameters are recovered up to a permutation . for the kullback - leibler divergence we compute @xmath265 according to , the complete - data log - likelihood @xmath266 is @xmath267 where @xmath83 and @xmath85 have been introduced in equations and , respectively . now , note that @xmath268 = \p_\tau(z^{i , q}=1\|\mathcal o ) = \p_\tau(z_i = q\|\mathcal o)=\tau^{i , q}$ ] . moreover , by the factorization property , for every @xmath269 we have @xmath270={e}_\tau[z^{i , q}\|\mathcal o]{e}_\tau[z^{j , l}\|\mathcal o]=\tau^{i , q}\tau^{j , l } .\ ] ] the quantity @xmath80 is thus equal to @xmath271 $ ] , namely the variational approximation of the mean number of dyads with latent groups @xmath60 . similarly , @xmath82 equals @xmath272 $ ] , the variational approximation of the probability that observation @xmath273 corresponds to a dyad with latent groups @xmath60 . it follows that @xmath274 where @xmath275 is @xmath276 the variational ` e`-step consists in maximizing @xmath277 with respect to the @xmath77 s which are constrained to satisfy @xmath278 for all @xmath26 . in other words , we maximize @xmath279 with lagrange multipliers @xmath280 . the partial derivatives are @xmath281 the partial derivatives are null iff @xmath282 and the @xmath77 s satisfy the fixed point equations , with @xmath283 being the normalizing constant . in this part , each intensity @xmath30 is estimated by a piecewise constant function and we propose a data - driven choice of the partition of the time interval @xmath4 $ ] . in the following @xmath60 is fixed and we start by considering a fixed partition @xmath284 of @xmath4 $ ] with partition size @xmath285 . denote @xmath286 the space of piecewise constant functions on @xmath287 . note that the total number of dyads @xmath288 is an upper bound for @xmath80 ( the variational mean number of dyads in group @xmath60 ) . following @xcite , we consider the projection estimator of @xmath30 on @xmath286 defined as @xmath289 where the least - squares contrast is defined ( relatively to the counting process @xmath81 ) for all @xmath290,dt)$ ] by @xmath291 the ( variational ) mean number of observations @xmath72 with group membership @xmath60 occurring in time interval @xmath292 for @xmath293 is @xmath294 denote @xmath295 the length of interval @xmath292 . then the estimator @xmath296 is given by @xmath297 we remark that with a fixed partition @xmath298 , the setup is purely parametric with a finite number of parameters @xmath299 that determine the piecewise constant function @xmath300 this means that with this point of view , a classical ` m`-step can be performed with some objective function @xmath301 to be maximized with respect to @xmath108 and @xmath302 . interestingly , it turns out that the solution of this ` m`-step is exactly the same as the projection estimators given by , that minimize the contrasts @xmath303 on @xmath286 . note that as the estimators of @xmath30 are computed separately , the approach allows to choose different partitions for different groups @xmath60 . now we turn to the choice of the partition and provide an adaptive model selection method , that is applied to every function @xmath30 separately . let @xmath304 be a finite collection of partitions of @xmath4 $ ] considered for the estimation of @xmath30 with fixed @xmath60 . adaptive estimation consists in choosing the best estimator among the collection of estimators @xmath305 with @xmath296 defined by . the choice is based on a penalized least - squares criterion of the form @xmath306 for some penalty function @xmath307 that penalizes large partitions . following @xcite we take for @xmath304 either the collection of regular partitions @xmath308 of @xmath4 $ ] with @xmath309 intervals each of length @xmath310 for @xmath311 , or the collection of dyadic partitions @xmath312 of @xmath4 $ ] with @xmath313 intervals of length @xmath314 for @xmath315 ( where @xmath115 and @xmath316 are to be chosen ) . furthermore , the penalty function is given by @xmath317 where @xmath318 denotes the finest partition in the collection @xmath319 , that is @xmath320 in the regular case and @xmath321 in the dyadic case , and @xmath322 denotes the @xmath323-th interval of partition @xmath324 . denote by @xmath325 the partition that minimizes @xmath326 over @xmath327 . let @xmath328 be the size of partition @xmath329 . then the adaptive estimator of intensity @xmath30 is given by @xmath330 that writes @xmath331 , \quad \hat { \alpha}^{(q , l)}_{\text{hist } } ( t ) = \hat \alpha_{\hat { \mathcal{e}}^{(q , l)}}^{(q , l)}(t ) = \frac1{t\bar{y}^{(q , l)}}\sum_{k=1}^{\hat d^{(q , l ) } } \hat d^{(q , l)}n^{(q , l)}(e_k^{\hat{\mathcal e } } ) \1_{e_k^{\hat{\mathcal e}}}(t).\ ] ] a natural stopping criterion of the variational ` em ` algorithm is based on function @xmath277 defined in . indeed , @xmath332 , where @xmath333 denotes the entropy of the distribution @xmath334 defined as @xmath335 . as our estimation procedure aims at maximizing @xmath277 , the algorithm may be stopped at iteration @xmath336 if the increase of @xmath277 is less than a given threshold @xmath337 , that is when @xmath338},\tau^{[s+1]})-j(\theta^{[s]},\tau^{[s]})}{j(\theta^{[s]},\tau^{[s]})}\right\|<\varepsilon.\ ] ] we use several initializations of the algorithm , relying on different aggregated datasets ( on the whole time interval or on sub - intervals ) and applying a k - means algorithm on the rows of the adjacency matrix of these aggregated datasets . ( bold line ) and the inter - group intensity @xmath30 for @xmath212 ( dotted line ) with different shifting parameter @xmath339 . , height=340 ] d. j. daley and d. vere - jones . _ an introduction to the theory of point processes . i_. probability and its applications ( new york ) . springer - verlag , new york , second edition , 2003 . elementary theory and methods .	to model recurrent interaction events in continuous time , we propose an extension of the stochastic block model where each individual belongs to a latent group and interactions between two individuals follow a conditional inhomogeneous poisson process whose intensity is driven by the individuals latent groups . the model is shown to be identifiable and an estimation procedure is proposed based on a semiparametric variational expectation - maximization algorithm . two versions of the method are developed , using either a nonparametric histogram approach ( with an adaptive choice of the partition size ) or kernel intensity estimators . the number of latent groups can be selected by an integrated classification likelihood criterion . finally , we demonstrate the performance of our procedure on synthetic experiments and the analysis of several real datasets illustrates the utility of our approach . * keywords : * dynamic interactions ; expectation - maximization algorithm ; integrated classification likelihood ; link streams ; longitudinal network ; semiparametric model ; stochastic block model ; variational approximation .
spatial indexing of geographic data has always been an important component of database systems . since the wide - spread adoption of social media and social networks , the size of data to be indexed has grown by multiple orders of magnitude , making even more demand on the efficiency of indexing algorithms and index structures . certain fields of natural sciences also face similar problems : astronomers , for example , have to perform spatial searches in databases of billions of observations where spatial criteria can be arbitrarily complex spherical regions . inspired by astronomical problems , szalay et al . @xcite came up with a solution to index spatial data stored in microsoft sql server years before native support of geographic indexing appeared in the product . their indexing scheme is called hierarchical triangular mesh ( htm ) and uses an iteratively refined triangulation of the surface of the sphere to build a quad - tree index . in the present case study , we demonstrate how we applied htm to index real gps coordinates collected from the open data streams of twitter , and performed a huge spatial join on the data to classify the coordinates by the administrative regions of the world . our results show that pre - filtering capabilites of htm are significantly better than that of the built - in spatial index of sql server and that htm renders spatial joins originally thought impossible to be able to be computed in a reasonable time . source code and full queries are available at the following url : http://www.vo.elte.hu/htmpaper . our data set consists of short messages ( tweets ) collected over a period of two years from the publicly available `` sprinkler '' data stream of twitter . more than half of the tweets , over a billion , are geo - tagged . we built a microsoft sql server database of the tweets @xcite and wanted to classify the geo - tagged messages by political administrative regions to investigate the geographic embedding of the social network of twitter users . microsoft sql server supports spatial indexing since version 2008 via a library implemented on top of the integrated .net framework runtime ( sql clr ) . the library works by projecting the hemisphere onto a quadrilateral pyramid , then projecting the pyramid onto a square , and tessellating the square using four levels of fix - sized rectangular grids to construct a spatial index . the number of grid cells can be set between @xmath0 and @xmath1 providing a maximal index depth of 32 bits . the index structure itself is materialized as a hidden , but otherwise normal database table containing one row for each cell touched by the geography object being indexed . the table uses 11 bytes for the spatial index which is complemented by the primary key of the indexed object , in our case an additional 10 bytes . the final index size can be controlled by limiting the number of cells stored for each geography object resulting in less effective pre - filtering of spatial matches when the index size is kept small . the hard limit on the number of index entries per geography object is 8192 . similarly to the built - in spatial index of sql server , htm indexing is also implemented in .net . by default , htm calculates a 40 bit deep hash , the so called htm i d , from the coordinates . the 40 bit hash length corresponds to an almost uniform resolution of 10 meters on the surface of the earth . htm tessellates two dimensional shapes with small spherical triangles , called trixels . trixels are represented by integer intervals ; all coordinates with an htm i d falling into the interval are guaranteed to be covered by the trixel . as htm is a custom library , we had full control over the index structures , therefore we simply stored the 8 byte htm identifiers in the same table where the tweets were , and built an auxiliary index on the table ordered by the htm identifier . together with the primary key , the auxiliary index size was 18 bytes per row , exactly one row per coordinate pair . we classified tweets using the maps from gadm.org , an open database of global administrative regions of every country in the world . the maps were loaded into the database as geography objects using the built - in geography type . for indexing , however , we decided to use htm . this raised a problem because the htm library was built for astronomical applications where regions on the sphere are better represented as unions of convex shapes contoured by great _ or _ small circles than by vertices of polygons connected by great circles . this union - of - convexes representation is not appropriate for highly detailed complex maps as shapes would need to be decomposed into convexes first , a process that significantly increases the size of the data structures . unfortunately , no code exists to directly compute the htm tessellation of maps in the polygon representation , so we had to use another approach . by combining the htm library and the built - in geographic library of sql server , we determined the approximate htm tessellation of regions up to a given precision by iteratively refining htm trixels at the boundaries . our solution , see algorithm [ lst : alg ] , goes as follows . we construct a coarse tessellation of the region based on the bounding circle , then we intersect each trixel with the region using the built - in functions of sql server . if a trixel is completely inside the region it is added to the result set . similarly , a completely disjoint trixel is discarded . trixels intersecting with the boundary are refined into four sub - trixels and the algorithm is called recursively . passing only the intersection of the original map with the trixel to the recursive call reduces the total runtime of the tessellation significantly . the algorithm uses the maximum depth of htm trixels as a stop condition to limit the resolution of the tessellation but can be easily modified to use an upper limit on the number of trixels instead . also , instead of trying to keep the index tables small , we store every trixel of the tessellation . trixels on the deepest level which intersect with the boundary of the geography object are flagged as `` partial '' . figure [ fig : cali ] illustrates the results of the level 9 htm covering of california with partial trixels in green . retlist @xmath2 t.partial @xmath3 false retlist.add(t ) region2 @xmath3 region.stintersection(t ) t.partial @xmath3 true retlist.add(t ) tlist2 @xmath3 t.refine(t.level+1 ) retlist.addrange ( ) retlist .... create table tweet ( i d bigint primary key , htmid bigint , coord geography ) create index ix_tweet_htm on tweet ( htmid ) create spatial index ix_tweet_spatial on tweet ( coord ) create table region ( i d int primary key , geo geography ) create spatial index ix_region_spatial on region ( geo ) create table regionhtm ( i d int , --foreign key to region.id start bigint , end bigint , partial bit ) .... in order to explain the internals of htm indexing , we create the schema of our database with query [ query : schema ] . the htm - based pre - filtering of a spatial join between a table containing gps coordinates and another containing the tessellations of complex regions requires an inner join with a ` between ` operator in the join constraint . query [ query : htmfilter ] is a simplified example of such pre - filtering query . we will refer to these types of queries as _ range joins_. as range joins are highly optimized in the database engine , we expect excellent pre - filtering performance . the ` loop join ` hint is added to suggest a query plan that consist of scan of the ` tweet ` table , while doing index seeks on the much smaller ` regionhtm ` table . this is optimal as long as the htm index of the regions can be kept in memory . using the built - in spatial index of sql server , pre - filtering of a spatial join can be done with query [ query : sqlfilter ] . it translates into a rather complex execution plan that uses the spatial index for table ` tweet ` only , while calculating the tessellation of the geography objects in table ` region ` during query execution , or vice versa . by specifying query hints , one can tell the server which spatial index to use , but it seems impossible to use the spatial indices on _ both _ tables at the same time . this behavior has a tremendous impact on the performance of spatial joins when using the built - in indices . in case of the sql server geography index , exact containment testing can be done by simply replacing the function call to ` filter ` with ` stcontains ` , as in query [ query : sqlcontains ] . when using the htm index , points in full trixels are already accurately classified with query [ query : htmfilter ] , only points in partial trixels need further processing to filter out false positive matches . this is done in query [ query : htmcontains ] which again relies on the spatial functions of sql server . also note , that query [ query : htmcontains ] , by referencing the column ` region.geo ` , uses the entire region for testing point containment . in case of computing the spatial join of billions of coordinates with a limited number of complex regions , it is well worth to pre - compute the intersections of partial trixels and regions first and use them for containment testing instead of the whole regions . due to publicational constraints we omit the query but all performance metrics quoted in the paper are measured using the pre - computed intersections of the regions and partial trixels for exact containment testing . we measured the index generation time for the 50 continental states ( plus washington d.c . ) of the united states using two different depths ( @xmath4 and @xmath1 grids ) of the sql server geography index and three different depths ( level 12 , 14 and 16 ) of htm . for comparison , the @xmath4 resolution of the sql server index roughly corresponds to a level 12 htm index and the @xmath1 grid resolution corresponds to a level 16 htm index . the benchmarks were run on a 16-core database server with 96 gb memory . as our dataset fits into memory , queries are basically cpu - limited . index generation times are summarized in table [ tab : idxgen ] . note , that sql server executed the geography index generation on two threads , while htm ranges were generated on a single thread only . while we have no control over the internals of geography indices , the iterative refining of htm tessellation could be replaced with a smarter , multi - threaded one . also , the size of the geography indices is internally limited to 8192 entries per region , while the htm indices were calculated without pruning , ultimately resulting in much larger index sizes . .... select t.id , h.id , h.partial from tweet t inner loop join regionhtm h on t.htmid between h.start and h.end .... .... select t.id , r.id from tweet t inner join region r on r.geo.filter(t.coord ) = 1 .... [ query : sqlfilter ] .... select tweet.id , region.id from tweet t inner join region r on r.geo.stcontains(t.coord ) = 1 .... .... select tweet.id , regionhtm.id from tweet t inner loop join regionhtm h on t.htmid between h.start and h.end inner join region r on r.id = h.id where h.partial = 0 or r.geo.stcontains(t.coord ) = 1 .... it is also rather instructive to compare the two indexing schemes by the false positive rates of pre - filtering . the results are listed in table [ tab : fp ] . false positives rates of the htm index are significantly lower in all cases , especially for higher index depths . in the case of the sql server spatial index , increasing the resolution does not help , but the opposite : it just makes things worse , as the number of index rows ( and the resolution of the tessellation ) is limited to a maximum of 8192 cells , insufficiently small for complex maps . strictly limiting index size only helps when the number of shapes to be indexed is large and the shapes are relatively small and simple . when spatial indices fit into the memory , or at least can be read quickly from the disk , pre - filtering using range joins is expected to be significantly faster , even for indices with millions of rows , rather than exact containment testing against complex shapes . to test the performance of point classification , we prepared three samples having approximately 300 thousand , 1 million , 5 million points in each , uniformly sampled from the original database . some tests were also run using the entire data set of more than one billion tweets . the coordinates covered the entire world but the majority of them were within the continental united states . the geographical distribution of the samples is realistic and follows the population density weighted by the local twitter usage rate . to evaluate index performance , we computed a spatial join between a the samples of gps coordinates with different cardinality and the 51 regions , first with pre - filtering only , then with exact containment test . results of pre - filtering are listed in table [ tab : prefilter ] , while exact containment metrics are shown in table [ tab : total ] . all queries were executed using cold buffers , thus include i / o and cpu times . the spatial join performance of the htm index turned out to be about a hundred times better than the performance of the built - in geography index . pre - filtering itself is about a thousand times faster than the built - in index , and usually could be done in a few seconds for the smaller samples . such short query times are hard to be measured correctly , and values show a significant scatter when the queries are repeated . the two main reasons behind the significantly better performance of htm are : 1 ) htm - based pre - filtering could benefit from the spatial index on both tables , whereas sql server s geography library only used the index for one of the tables and calculated the tessellation for the other table on the fly . 2 ) the extensive pruning of index entries resulted in a very high rate of false positives in case of sql server s geography index . because of the pruning , increasing the index resolution could not actually increase the resolution of the tessellation in case of the rather complex maps . by using a significantly larger , but still manageable index table , and by intersecting the trixels of the tessellations with the regions to reduce the complexity of exact containment testing , htm indexing could reduce the cost of spatial joins to a minimum . based on these results , it is clear that running the point classification using only the built - in geography index of sql server index is not a viable solution for any task similar to ours , namely , when the number of points is in the billions range . .index generation time and number of index rows of the regions . [ cols="<,>,>",options="header " , ] in this paper , we investigated the feasibility of efficient classification of gps coordinates of twitter messages by geographic regions using a relational database management system , microsoft sql server 2012 . we evaluated the performance of the built - in spatial indexing technology side by side with a customized solution based on hierarchical triangular mesh ( htm ) indexing . the built - in spatial index was found to be inadequate to perform spatial joins between large sets of gps coordinates ( on the scale of billions ) and complex geographic regions . we showed that our solution , a heuristic combination of existing techniques for handling spatial data in a relational database environment , can easily be a hundred times faster and makes the computation of the aforementioned spatial join available in reasonable time . we pointed out that the strength of htm indexing is the great control the database programmer has on both the index structure and query plans ( via hints ) . we also demonstrated that aggressive pruning of spatial indices is not a good idea when indexing of very complex regions is a requirement , as range - join - based pre - filtering is significantly faster than exact containment testing , even in case of millions of index entries . to make exact containment testing even faster , we pre - computed the intersections of the complex geographic regions and partial htm trixels and use these much smaller shapes to filter out false positives . the authors thank the partial support of the european union and the european social fund through project futurict.hu ( grant no . : tamop-4.2.2.c-11/1/konv-2012 - 0013 ) , and the otka 103244 grants . eitkic 12 - 1 - 2012 - 0001 project was partially supported by the hungarian government , managed by the national development agency , and financed by the research and technology innovation fund and the makog foundation .	we present a case study about the spatial indexing and regional classification of billions of geographic coordinates from geo - tagged social network data using hierarchical triangular mesh ( htm ) implemented for microsoft sql server . due to the lack of certain features of the htm library , we use it in conjunction with the gis functions of sql server to significantly increase the efficiency of pre - filtering of spatial filter and join queries . for example , we implemented a new algorithm to compute the htm tessellation of complex geographic regions and precomputed the intersections of htm triangles and geographic regions for faster false - positive filtering . with full control over the index structure , htm - based pre - filtering of simple containment searches outperforms sql server spatial indices by a factor of ten and htm - based spatial joins run about a hundred times faster .
the neutrinos , in the standard model of elementary particle physics , are assumed to be massless . by the discovery of the neutrino oscillations , this assumption became invalidated . the neutrinos are massive , even though they carry , as compared to the electron , only very small masses . then the _ photon _ is left as the only known massless and free elementary particle . the gluons do not qualify in this context since they are confined and can not exist as free particles under normal circumstances . consequently , the photon is the only particle that is directly related to the light cone @xmath4 and that can be used for an operational definition of the light cone ; here @xmath5 is the metric of spacetime , @xmath6 a coordinate differential , and @xmath7 . we are back as the name light cone suggests anyway to an electromagnetic view of the light cone . speaking in the framework of classical optics , the _ light ray _ would then be the elementary object with the help of which one can span the light cone . we take `` light ray '' as synonymous for radar signals , laser beams , or electromagnetic rays of other wavelengths . it is understood that classical optics is a limiting case , for short wavelengths , of classical maxwell - lorentz electrodynamics . in other words , if we assume the framework of maxwell - lorentz electrodynamics , we can derive , in the geometric optics limit , light rays and thus the light cone , see perlick @xcite and the literature given there . however , the formalism of maxwell - lorentz electrodynamics is interwoven with the riemannian metric @xmath5 in a nontrivial way . accordingly , in the way sketched , one can never hope to find a real derivation of the light cone . therefore , we start from the _ premetric _ form of electrodynamics , that is , a metric of spacetime is not assumed . nevertheless , we can derive the generally covariant maxwell equations , expressed in terms of the excitation @xmath0 and the field strength @xmath1 , from the conservation laws of electric charge and magnetic flux . we assume a local and linear spacetime relation between @xmath2 and @xmath3 . then we can solve the maxwell equations . in particular , we can study the propagation of electromagnetic waves , and we can consider the geometrical optics limit . in this way , we derive the light rays that are spanning the light cone . in general , we find a _ quartic _ wave covector surface ( similar as in a crystal ) that only reduces to the pseudo - riemannian light cone of general relativity if we forbid _ birefringence _ ( double refraction ) in vacuum . hence , in the framework of premetric electrodynamics , the local and linear spacetime relation , together with a ban on birefringence in vacuum , yields the pseudo - riemannian light cone of general relativity . accordingly , the geometrical structure of a riemannian spacetime is derived from purely electromagnetic data . we consider that as our contribution to the einstein year 2005 , and we hope that going beyond the geometrical optics limit will yield better insight into the geometry of spacetime . the axiomatic scheme that we are going to present here is already contained in our book @xcite where also references to earlier work and more details can be found . in the meantime we learnt from the literature that appeared since 2003 ( see , e.g. , delphenich @xcite , itin @xcite , kaiser @xcite , kiehn @xcite , and lindell & sihvola @xcite ) and improved our derivation of the light cone , simplified it , made it more transparent ( see , e.g. , @xcite ) . the formalism and the conventions we take from @xcite . in our approach , we start from a 4-dimensional spacetime manifold that is just a continuum which can be decomposed locally in ( 1-dimensional ) time and ( 3-dimensional ) space . it carries no metric and it carries no ( linear or affine ) connection . as such it is inhomogeneous . it does nt make sense to assume that a vector field is constant in this continuum . only the constancy of a scalar field is uniquely defined . also a measurement of temporal or spatial intervals is still not defined since a metric is not yet available . in technical terms , the spacetime is a 4-dimensional connected , hausdorff , paracompact , and oriented differential manifold . on such a manifold , we assume the existence of a foliation : the spacetime can be decomposed locally into three - dimensional folios labeled consecutively by a monotonically increasing `` prototime '' parameter @xmath8 , see fig . [ fig:1 ] . a vector field @xmath9 , transverse to the foliation , is normalized by @xmath10 . accordingly , we find for the dimensions @xmath11=[\sigma]^{-1}=t^{-1}$ ] , where @xmath12 denotes the dimension of time . we can decompose any exterior form @xmath13 in `` time '' and `` space '' pieces . the part _ longitudinal _ to the vector @xmath9 reads @xmath14 the part _ transversal _ to the vector @xmath9 @xmath15 putting these two parts together , we have the space - time decomposition @xmath16 with the absolute dimensions @xmath17=[\psi]\,t^{-1}$ ] and @xmath18=[\psi]$ ] . the 3-dimensional exterior derivative is defined by @xmath19 . we can use the notion of the _ lie derivative _ of a @xmath20-form @xmath13 along a vector field @xmath21 , i.e. , @xmath22 , and can introduce the derivative of a transversal field @xmath23 with respect to prototime as @xmath24 we assume that spacetime is `` populated '' with classical matter , either described by fields and/or by fluids . in between the agglomerations of matter , there may also exist vacuum . matter is divided into electrically charged and neutral matter . turning to the physics of the former , we assume that on the spatial folios of the manifold we can determine an electric charge @xmath25 as a 3-dimensional integral over a charge density and a magnetic flux @xmath26 as a 2-dimensional integral over a flux density . this is at the bottom of classical electrodynamics : spacetime is filled with matter that is characterized by charge @xmath25 and by magnetic flux @xmath26 . for neutral matter both vanish . the absolute dimension of charge will be denoted by @xmath27 , that of magnetic flux by @xmath28 = [ action / charge ] = @xmath29 , with @xmath30 as the dimension of action . one can catch single electrons and single protons in traps and can _ count _ them individually . thus , the electric charge conservation is a fundamental law of nature , valid in macro- as well as in micro - physics . accordingly , it is justified to introduce the absolute dimension of charge @xmath27 as a new and independent concept . let us define , in 4-dimensional spacetime , the electrical current 3-form @xmath31 , with dimension @xmath32=q$ ] . its integral over an arbitrary 3-dimensional spacetime domain yields the total charge contained therein : @xmath33 . accordingly , the local form of charge conservation ( axiom 1 ) reads : @xmath34 this law is metric - independent since it is based on a _ counting _ procedure for the elementary charges . using a foliation of spacetime , we can decompose the current @xmath31 into the 2-form of the electric current density @xmath35 and the 3-form @xmath36 of the electric charge density : @xmath37 then ( [ dj1 ] ) can be rewritten as the continuity equation : @xmath38 both versions of charge conservation , eqs.([dj1 ] ) and , can also be formulated in an integral form . electric charge was postulated to be conserved in all regions of spacetime . if spacetime is topologically sufficiently trivial , we find , as consequence of , that @xmath31 has to be exact : @xmath39 this is the inhomogeneous maxwell equation in its premetric form . the _ electromagnetic excitation _ 2-form @xmath2 , with @xmath40=[j]=q$ ] , is measurable with the help of ideal conductors and superconductors and thus has a direct _ operational _ significance . by decomposing @xmath2 into time and space , we obtain the electric excitation 2-form @xmath41 ( historical name : dielectric displacement " ) and the magnetic excitation 1-form @xmath42 ( magnetic field " ) : @xmath43 substituting ( [ hhd ] ) into ( [ jdh ] ) , we recover the pair of the 3-dimensional inhomogeneous maxwell equations @xmath44 with the derivation of the inhomogeneous maxwell equations the information contained in axiom 1 is exhausted . as is evident from the coulomb - gauss law @xmath45 , it is the active character of @xmath36 that plays a role in this inhomogeneous maxwell equation : the charge density @xmath36 is the source of @xmath46 ( and , analogously , the current density @xmath35 that of @xmath47 ) . since we search for new input , it is near at hand to turn to the _ passive _ character of charge , that is , to wonder what happens when a test charge is put in an electromagnetic field . in the purely electric case with a test charge @xmath48 , we have @xmath49 with @xmath50 and @xmath51 as components of the covectors of force and electric field strength , respectively . the simplest relativistic generalization for defining the electromagnetic field is then of the type @xmath52 is a twisted 2-form , the magnetic excitation @xmath53 a twisted 1-form . the electric field strength @xmath54 is a 1-form and the magnetic field strength @xmath55 a 2-form , both without twist.,height=226 ] accordingly , with the force density covector ( or 1-form ) @xmath56 , we can formulate axiom 2 as @xmath57 here @xmath58 is a local frame , with @xmath59 . axiom 2 provides an operational definition of the electromagnetic field strength 2-form @xmath3 , the absolute dimension of which turns out to be @xmath60=h / q $ ] . its @xmath61 decomposition @xmath62 introduces the electric field strength 1-form @xmath54 and the magnetic field strength 2-form @xmath55 , see fig.[fig:3 ] . if we substitute and into , we recover , for @xmath63 , the lorentz force density . the field strength @xmath3 , as a 2-form , can be integrated over a 2-dimensional area @xmath64 in 4-dimensional spacetime . this yields the total magnetic flux @xmath26 piercing through this area : @xmath65 . in close analogy to electric charge conservation , we assume that also the flux is conserved . then , in local form , magnetic flux conservation ( axiom 3 ) reads . in premetric electrodynamics this has been done by edelen @xcite , kaiser @xcite , and by us @xcite . however , then one has to change axiom 2 , too , and the lorentz force density picks up an additional term @xmath66 . this destroys axiom 2 as an operational procedure for defining @xmath3 . moreover , magnetic charges have never been found . ] @xmath67 the faraday induction law and the sourcelessness of @xmath55 are the two consequences of @xmath68 . in this sense , axiom 3 has a firm experimental underpinning . ... is meant to be the `` naked '' or `` featureless '' spacetime manifold , without metric and without connection , together with the maxwell equations @xmath69 , the lorentz force formula , and the electromagnetic energy - momentum current to be discussed below , see . we stress that the poincar group and special relativity have nothing to do with the foundations of electrodynamics as understood here in the sense of the decisive importance of the underlying _ generally covariant _ conservation laws of charge ( axiom 1 ) and flux ( axiom 3 ) . historically , special relativity emerged in the context of an analysis of the electrodynamics of moving bodies @xcite , but within the last 100 years classical electrodynamics had a development of its own and its structure is now much better understood than it was 100 years ago . _ diffeomorphism invariance _ was recognized to be of overwhelming importance . poincar invariance turned out to play a secondary role only . of course , premetric electrodynamics so far does not represent a complete physical theory . the excitation @xmath2 does not yet communicate with the field strength @xmath3 . only by specifying a `` spacetime '' relation between @xmath2 and @xmath3 ( the constitutive law of the spacetime manifold ) , only thereby we recover under suitable conditions our normal riemannian or minkowskian spacetime which we seem to live in . in this sense , a realistic spacetime and thus an appropriate _ geometry _ thereof emerges only by specifying additionally a suitable spacetime relation on the featureless spacetime . as explained , axiom 1 , axiom 2 , axiom 3 , together with axiom 4 on energy - momentum , constitute premetric electrodynamics . let us display the first three axions here again , but now axiom 1 and axiom 3 in in the more general integral version . for any submanifolds @xmath70 and @xmath71 that are closed , i.e. , @xmath72 and @xmath73 , the axioms read @xmath74 by de rham s theorem we find the corresponding differential versions @xmath75 the physical interpretation of the quantities involved is revealed via their @xmath76-decompositions , , , and @xmath77 , see fig.[fig:4 ] . constitute the electromagnetic field . the excitations @xmath78 are _ extensive _ quantities ( how much ? ) , the field strengths @xmath79 _ intensive _ quantities ( how strong ? ) . , height=226 ] let us now turn to the energy - momentum question . using the properties of the exterior differential , we can rewrite the lorentz force density as @xmath80 here the _ kinematic energy - momentum _ 3-form of the electromagnetic field , a central result in the premetric electrodynamics , reads ( axiom 4 ) @xmath81\,.\label{simax}\ ] ] the remaining force density 4-form turns out to be @xmath82 the absolute dimension of @xmath83 and of @xmath84 is @xmath85 , where @xmath86 denotes the dimension of length . [ provided , additionally , a linear connection is given with the covariant differential @xmath87 , then @xmath88 with the new supplementary force density @xmath89 which contains the covariant lie derivative . in general relativity theory , @xmath90 eventually vanishes for the standard maxwell - lorentz electrodynamics . ] the system of the maxwell equations @xmath91 is apparently underdetermined . it gets predictive power only when we supplement it with a spacetime ( or constitutive ) relation between the excitation and the field strength . as axiom 5 , we postulate a general local and linear spacetime relation @xmath92 here excitation and field strength decompose according to @xmath93 and @xmath94 , respectively . the constitutive tensor @xmath95 , as 4th rank tensor with 36 independent components , has to be space and time dependent since constant components would not have a generally covariant meaning on the naked spacetime manifold we consider . let us decompose @xmath96 into irreducible pieces . in the premetric framework we can only perform a _ contraction_. a first contraction yields @xmath97 a second one @xmath98 then , introducing the traceless piece @xmath99 we can rewrite the original constitutive tensor as @xmath100}^{l ] } + { \frac 1 6}\,\kappa\,\delta_{[i}^k\delta_{j]}^l.\label{kap - dec}\end{aligned}\ ] ] the _ skewon _ and the _ axion _ fields are conventionally defined by @xmath101 substituting ( [ kap - dec ] ) into ( [ hchif ] ) and using , we obtain the spacetime relation explicitly : @xmath102k } + \alpha\,f_{ij}\,.\ ] ] the _ principal _ ( or the metric - dilaton ) part @xmath103 of the constitutive tensor with 20 independent components will eventually be expressed in terms of the metric ( thereby cutting the 20 components in half ) . [ in standard maxwell - lorentz electrodynamics @xmath104\label{ml}\ ] ] the principal part @xmath103 must be non - vanishing in order to allow for electromagnetic wave propagation in the geometrical optics limit , see the next section . the skewon part @xmath105 with its 15 components was proposed by us . we put forward the hypothesis that such a field exists in nature . finally , the axion part @xmath106 had already been introduced in elementary particle physics in a different connection but with the same result for electrodynamics , see , e.g. , wilczek s axion electrodynamics @xcite and the references given there . the spacetime relation we are discussing here is the constitutive relation for spacetime , i.e. , for the vacuum . however , one has analogous structures for a _ medium _ described by a local and linear constitutive law . the _ skewon _ piece in this framework corresponds to chiral properties of the medium inducing optical activity , see lindell et al . @xcite , whereas the concept of an _ axion _ piece has been introduced by tellegen @xcite for a general medium , by dzyaloshinskii @xcite specifically for cr@xmath107o@xmath108 , and by lindell & sihvola @xcite in the form of the so - called perfect electromagnetic conductor ( pemc ) . recently , lindell @xcite discussed the properties of a _ skewon - axion _ medium . the following alternative representation of the constitutive tensor is useful in many derivations and for a comparison with literature , see post @xcite , @xmath109 with @xmath110 } -\epsilon^{klm[i}\!\not\!s_m{}^{j]}}_{15,\ , \text{skewon}}+\underbrace{\epsilon^{ijkl}\ , \alpha\;}_{1,\ , \text{axion}}\,.\ ] ] it is convenient to consider the excitation @xmath2 and the field strength @xmath3 as 6-vectors , each comprising a pair of two 3-vectors . the spacetime relation then reads @xmath111 accordingly , the constitutive tensors are represented by the @xmath112 matrices @xmath113 the @xmath114 matrices @xmath115 are defined by @xmath116 or explicitly , recalling the irreducible decomposition ( [ result1 ] ) , @xmath117 the constituents of the principal part are the _ permittivity _ tensor @xmath118 , the _ impermeability _ tensor @xmath119 , and the magnetoelectric cross - term @xmath120 , with @xmath121 ( fresnel - fizeau effect ) . the skewon @xmath122 and the axion @xmath106 describe electric and magnetic faraday effects and ( in the last two relations ) optical activity . if we substitute , , into , we find a 3-dimensional explicit form of our axiom 5 formulated in : @xmath123 after the spacetime relation ( axiom 5 ) has been formulated , we have a complete set of equations describing the electromagnetic field . we can now study the propagation of electromagnetic waves la hadamard . the sourceless maxwell equations read @xmath124 in the geometric optics approximation ( equivalently , in the hadamard approach ) an electromagnetic wave is described by the propagation of a discontinuity of the electromagnetic field . the surface of discontinuity @xmath125 is defined locally by a function @xmath26 such that @xmath126 on @xmath125 . the jumps @xmath127 $ ] of the electromagnetic quantities across @xmath125 and the wave covector @xmath128 then satisfy the geometric hadamard conditions : @xmath129 = 0\,,\;[dh]=q\wedge h = 0\;\qquad \rightarrow \qquad h = q\wedge c\,,\\ & & \,[f ] = 0\,,\;[df]=q\wedge f = 0\;\qquad \rightarrow \qquad f = q\wedge a\,.\end{aligned}\ ] ] here @xmath130 and @xmath131 are arbitrary 1-forms . we use the spacetime relation and find for the jumps of the field derivatives @xmath132 with @xmath133 . accordingly , ( see post @xcite , eq.(9.40 ) for @xmath134}=0 $ ] ) . ] @xmath135 this equation is a 3-form with 4 components . we have to solve it with respect to @xmath131 . as a first step , we have to remove the gauge freedom @xmath136 present in ( [ qwh ] ) . we choose the gauge @xmath137 . after some heavy algebra , we find ( see @xcite for details , @xmath138 ) @xmath139 these are 3 equations for three @xmath140 nontrivial solutions exist provided @xmath141 we can rewrite the latter equation in a manifestly 4-dimensional covariant form ( @xmath142 , @xmath143 ) , @xmath144 with @xmath145 . the 4-dimensional tensorial transformation behavior is obvious . we define 4th - order tamm rubilar ( tr ) tensor density of weight @xmath146 , @xmath147 it is totally symmetric @xmath148 . thus , it has 35 independent components . because @xmath149 , the total antisymmetry of @xmath150 yields @xmath151 an explicit calculation shows that @xmath152 summarizing , we find that the wave propagation is governed by the extended fresnel equation that is generally covariant in 4 dimensions : @xmath153 the wave covectors @xmath27 lie on a _ quartic fresnel wave surface _ , not exactly what we are observing in vacuum at the present epoch of our universe . some properties of the tr - tensor , see @xcite , were discussed recently by beig @xcite . recalling the ` 6-vector ' form of the spacetime relation ( [ cr ] ) with the @xmath114 constitutive matrices ( [ ab - matrix0 ] ) and ( [ cd - matrix0 ] ) , we can decompose the tr - tensor into time and space pieces : @xmath154 then the fresnel equation ( [ fresnel ] ) reads @xmath155 or @xmath156 with @xmath157 @xmath158\nonumber\\ & & + ( { \mathfrak c}^d{}_c + { \mathfrak d}_c{}^d)({\mathfrak a}^{c(a } { \mathfrak c}^{b)}{}_d + { \mathfrak d}_d{}^{(a}{\mathfrak a}^{b)c } ) - { \mathfrak c}^d{}_d { \mathfrak a}^{c(a}{\mathfrak c}^{b)}{}_c\nonumber\\ & & - { \mathfrak d}_c{}^{(a}{\mathfrak a}^{b)c}{\mathfrak d}_d{}^d -{\mathfrak a}^{dc}{\mathfrak c}^{(a}{}_c { \mathfrak d}_d{}^{b)}\nonumber\\ & & + \left({\mathfrak a}^{(ab)}{\mathfrak a}^{dc}- { \mathfrak a}^{d(a}{\mathfrak a}^{b)c}\right){\mathfrak b}_{dc}\,,\label{ma2}\\ m^{abc } & = & \epsilon^{de(c\|}\left[{\mathfrak b}_{df } ( { \mathfrak a}^{ab ) } \,{\mathfrak d}_e^{\ f } - { \mathfrak d}_e^{\ a}{\mathfrak a}^{b)f}\ , ) + { \mathfrak b}_{fd}({\mathfrak a}^{ab)}\,{\mathfrak c}_{\ e}^f - { \mathfrak a}^{f\|a}{\mathfrak c}_{\ e}^{b ) } ) \right . \nonumber \\ & & \left . + { \mathfrak c}^{a}_{\ f}\,{\mathfrak d}_e^{\ b ) } \,{\mathfrak d}_d^{\ f } + { \mathfrak d}_f^{\ a}\,{\mathfrak c}^{b)}_{\ e } \,{\mathfrak c}^{f}_{\ d } \right ] \,,\label{ma3}\\ m^{abcd } & = & \epsilon^{ef(c}\epsilon^{\|gh\|d}\,{\mathfrak b}_{hf } \left[\frac{1}{2 } \,{\mathfrak a}^{ab)}\,{\mathfrak b}_{ge } - { \mathfrak c}^{a}_{\ e}\,{\mathfrak d}_g^{\ b)}\right ] \,.\label{ma4}\end{aligned}\ ] ] and trivial impermeability @xmath159 . the skewon field vanishes . there are two branches , the outer part of the surface is cut into half in order to show the inner branch . we use the dimensionless variables @xmath160.,height=302 ] let us look at some fresnel wave surfaces in order to get some feeling for the physics involved . divide by @xmath161 ( here @xmath162 is the frequency of the wave ) and introduce the dimensionless variables ( @xmath130 = velocity of light in special relativity ) @xmath163 then we have @xmath164 we can draw these _ quartic _ surfaces in the dimensionless variables @xmath165 , @xmath166 , @xmath167 , provided the @xmath168 s are given . according to - , the @xmath168 s can be expressed in terms of the @xmath114 matrices @xmath169 . these matrices are specified in - in terms of the permittivity etc .. a comparison with the spacetime relations in the form of , is particularly instructive . and trivial impermeability @xmath159 with a skewon field of electric faraday type @xmath170 ( all other components vanish ) . the surface has the form of a toroid ( depicted with two cuts ) . we use the dimensionless variables @xmath171.,height=264 ] let us start with a simple example . we assume that the permittivity is anisotropic but still diagonal , @xmath172 , whereas the impermeability is trivial @xmath159 . no skewon field is assumed to exist . whether an axion field is present or not does nt matter since the axion does not influence the light propagation in the geometrical optics limit . with mathematica programs written by tertychniy @xcite , we can construct for any values of @xmath173 the fresnel wave surface ; an example is displayed in fig.[anisotropic ] . more complicated cases are trivial permittivity @xmath174 and trivial impermeability @xmath159 , but a nontrivial skewon field . we can take a skewon field of electric faraday type @xmath175 , for example , see fig.[skewonelectricfaraday ] , or of magnetoelectric optical activity type @xmath176 , see fig.[skewonoptical ] . in both figures and in the subsequent one @xmath177 is the admittance of free space . the characteristic feature of the skewon field is the emergence of specific _ holes in the fresnel surfaces _ that correspond to the directions in space along which the wave propagation is damped out completely @xcite . this effect is in agreement with our earlier conclusion on the dissipative nature of the skewon field . and trivial impermeability @xmath159 with a skewon field of the magneto - electric optical activity type @xmath178 ( all other components vanish ) . it has two intersecting toroidal branches . we use the dimensionless variables @xmath179.,height=264 ] now we can combine anisotropic permittivity with the presence of a skewon field . then we expect to find some kind of fig.[anisotropic ] `` enriched '' with holes induced by the skewon field . this time we choose a spatially isotropic skewon field with @xmath180 . the outcome is depicted in fig.[isotropicskewon ] . the four holes confirm our expectation . and trivial impermeability @xmath159 with a spatially isotropic skewon field @xmath181 ( all other components vanish ) . we use the dimensionless variables @xmath182.,height=264 ] the propagation of light in _ local _ and _ linear _ premetric vacuum electrodynamics is characterized by the extended fresnel equation ( [ fresnel ] ) or ( [ fren ] ) . we can solve the fresnel equation with respect to the frequency @xmath162 , keeping the 3covector @xmath183 fixed . with the help of mathematica , we found the following four solutions @xcite : @xmath184 we introduced the abbreviations @xmath185 with @xmath186 now , let us demand the absence of birefringence ( also called double refraction ) . in technical terms this means , see the solutions ( [ freqk1 + ] ) , ( [ freqk1- ] ) , that @xmath187 and @xmath188 . then we have the degenerate solution @xmath189 the condition @xmath190 yields directly @xmath191 , and , using this , we find @xmath192 thus , @xmath193 accordingly , the quartic wave surface ( [ fren ] ) in this case reduces to @xmath194 ^ 2=0\,.\ ] ] multiplication yields @xmath195 if we substitute @xmath196 as defined in ( [ fresnel ] ) , we have explicitly ( @xmath7 ) @xmath197 this equation is quadratic in the 4-dimensional wave covector @xmath198 . therefore we recover the conventional _ light cone _ of general relativity at each point of spacetime , see fig . [ fig:5 ] . thereby the _ causal structure _ of spacetime is determined . thus , up to a scalar factor , we derived the riemannian metric of general relativity . moreover , as we have shown @xcite , we find the correct _ lorentzian signature_. the lorentzian ( also known as minkowskian ) signature can be traced back to the lenz rule , which determines the sign of the @xmath199 term in the induction law . and this sign is different from the one in the corresponding @xmath200 term in the oersted - ampre - maxwell law . in other words , the lorentz signature is encoded in the decomposition formulas and . neither is the minus sign in a convention nor the plus sign of the @xmath201 term in . since the lenz rule is related to the positivity of the electromagnetic energy , the same is true for the lorentzian signature . this derivation of the signature of the metric of spacetime from electrodynamics provides new insight into the structures underlying special as well as general relativity . at first the skewon and the axion emerged at the premetric level in our theory and only subsequently the metric . consequently , the axion and the skewon should be regarded as more fundamental fields ( if they exist ) than the metric . in the meantime , we phased out the skewon field since we insisted , in sec.[vanishing ] , on vanishing birefringence in vacuum . as to the metric , we recognize that multiplication of the metric by an arbitrary function @xmath202 was left open in the derivation of the last section , see : @xmath203 thus , only the conformally invariant part of the metric is determined . in other words , we have actually constructed the _ conformal _ ( or the light cone ) _ structure _ on the spacetime manifold , see , e.g. , weyl @xcite , schouten @xcite , and pirani & schild @xcite . it is known from _ special relativity _ that the light cone ( with lorentzian signature ) is invariant under the 15-parameter conformal group , see barut & raczka @xcite and blagojevi @xcite . the latter , in minkowskian coordinates @xmath204 , is generated by the following four sets of spacetime transformations : @xmath205 here @xmath206 are the 15 constant parameters , and @xmath207 . the poincar subgroup ( [ transl ] ) , ( [ lor ] ) ( for a modern presentation of it , see giulini @xcite ) leaves the spacetime interval @xmath208 invariant , whereas the dilatations ( [ dil ] ) and the proper conformal transformations ( [ proper ] ) change the spacetime interval by a scaling factor @xmath209 and @xmath210 , respectively ( with @xmath211 ) . in all cases the light cone @xmath212 is left invariant . the weyl subgroup , which is generated by the transformations ( [ transl])-([dil ] ) , and its corresponding noether currents were discussed by , e.g. , kopczyski et al . @xcite . for massless particles , instead of the poincar group , the conformal or the weyl group come under consideration , since massless particles move on the light cone . even though the light cone stays invariant under all transformations - , two reference frames that are linked to each other by a proper conformal transformation do nt stay inertial frames since their relative velocity is not constant . if one wants to uphold the _ inertial _ character of the reference frames , one has to turn to the weyl transformation , that is , one has to specialize to @xmath213 . the conformal group in minkowski space illustrates the importance of the light cone structure on a _ flat _ manifold . this is suggestive for the light cone on an arbitrarily curved manifold , even though there is no direct relation between - and the light cone structure we derived in the last section . the light cone metric @xmath214 introduces the hodge star @xmath215 operator . we then can straightforwardly verify that the principal part of the spacetime relation is determined as @xmath216 , where the coefficient of proportionality can be an arbitrary scalar function @xmath217 of the spacetime coordinates . this function is naturally identified with the _ dilaton field _ , see brans @xcite and fujii & maeda @xcite . introducing the ( levi - civita ) dual of the excitation , @xmath218 , we can then finally rewrite the spacetime relation for vanishing birefringence in vacuum as @xmath219\,f_{kl}\,,\label{nobirefr}\ ] ] that is , we are left with the constitutive fields dilaton @xmath220 , metric @xmath214 , and axion @xmath106 . the combination @xmath221j}(x)$ ] is conformally invariant , in complete agreement with the above analysis . the conformal structure of spacetime is laid down in . hence only 9 of the 10 independent components of the pseudo - riemannian metric @xmath5 are specified . we need , in addition to the conformal structure , a _ volume measure _ for arriving at a unique riemannian metric . this can be achieved by postulating a time or length standard . in exterior calculus , ( [ nobirefr ] ) reads @xmath222 the axion has not been found so far , so we can put @xmath223 . moreover , under normal cicumstances , the dilaton seems to be a constant field and thereby sets a certain scale , i.e. , @xmath224 , where @xmath225 is the admittance of free space the value of which is , in si - units , @xmath226 . ( the exact implementation of this assumption will have to be worked out in future . ) accordingly , we are left with the spacetime relation of conventional maxwell - lorentz electrodynamics @xmath227 weyl @xcite , in 1921 , proved a theorem that the projective and the conformal structures of a metrical space determine its metric uniquely . as a consequence weyl @xcite argued that _ ... the metric of the world can be determined merely by observing the `` natural '' motion of material particles and the propagation of action , in particular that of light ; measuring rods and clocks are not required for that . _ here we find the two elementary notions for the determination of the metric : the paths of a freely falling point particles , yielding the projective structure , and light rays , yielding the conformal structure of spacetime . later , in 1966 , pirani and schild @xcite , amongst others , deepened the insight into the conformal structure and the weyl tensor . in 1972 , on the basis of weyl s two primitive elements , ehlers , pirani , and schild ( eps ) @xcite proposed an axiomatic framework in which weyl s concepts of free particles and of light rays were taken as elementary notions that are linked to each other by plausible axioms . requiring compatibility between the emerging projective and conformal structures , they ended up with a weyl spacetime(riemannian metric with an additional weyl covector ) . they set a scale [ as we did in the last section , too ] and arrived at the pseudo - riemannian metric of general relativity . in this sense , eps were able to reconstruct the metric of general relativity . subsequently , many authors improved and discussed the eps - axiomatics . access to the corresponding literature can be found via the book of majer and schmidt @xcite or the work of perlick @xcite and lmmerzahl @xcite , e.g .. for a general review one should compare schelb @xcite and for a new axiomatic scheme schrter @xcite . as stated , the point particles and light rays were primary elements that were assumed to exist and no link to mechanics nor to electrodynamics was specified . the particle concept within the eps - axiomatics lost credibility when during the emergence of gauge theories of gravity ( which started in 1956 with utiyama @xcite even before the eps - framework had been set up in 1972 ) the first quantized wave function @xmath13 for matter entered the scene as an elementary and `` irreducible '' concept in gravity theory . when _ interference in an external gravitational field was discovered experimentally in 1975 by collella , overhauser , and werner ( cow ) @xcite , see also @xcite , sec.7 , it was clear that the point particle concept in the eps - framework became untenable from a physical point of view . for completeness let us mention some more recent experiments on matter waves in the gravitational field or in a noninertial frame : * the werner , staudenmann , and colella experiment @xcite in 1979 on the phase shift of neutron waves induced by the rotation of the earth ( sagnac - type effect ) , * the bonse & wroblewski experiment @xcite in 1984 on neutron interferometry in a noninertial frame ( verifying , together with the cow experiment , the equivalence principle for neutron waves ) , * the kasevich & chu interferometric experiment @xcite in 1991 with laser - cooled wave packets of sodium _ atoms _ in the gravitational field , * the mewes et al . experiment @xcite in 1997 with interfering freely falling bose - einstein condensed sodium atoms , see ketterle @xcite , fig.14 and the corresponding text , * the nesvizhevsky et al . experiment @xcite in 2002 on the quantum states of neutrons in the earth s gravitational field , and * the fray , hnsch , et al . @xcite experiment in 2004 with a matter wave interferometer based on the diffraction of atoms from effective absorption gratings of light . this interferometer was used for two stable isotopes of the rubidium atom in the gravitational field of the earth . thereby the equivalence principle was tested successfully on the atomic level . clearly , without the schrdinger equation in an external gravitational field or in a noninertial frame all these experiments can not be described . still , in most textbooks on gravity , these experiments are not even mentioned ! in the 1980 s , as a reaction to the cow - experiment , audretsch and lmmerzahl , for a review see @xcite , started to develop an axiomatic scheme for spacetime in which the point particle was substituted by a matter wave function and the light ray be a wave equation for electromagnetic disturbances . in this way , they could also include projective structures with an asymmetric connection ( i.e. , with torsion ) , which was excluded in the eps approach a priori . turning to the conformal structure , which is in the center of our interest here , lmmerzahl et al . @xcite , see also @xcite , reconsidered the audretsch - lmmerzahl scheme and derived the inhomogeneous maxwell equation from the following requirements : a well - posed cauchy problem , the superposition principle , a finite propagation speed , and the absence of birefringence in vacuum . the homogeneous maxwell equation they got by a suitable definition of the electromagnetic field strength . with a geometric optics approximation , compare our sec.[propagation ] , they recover the light ray in lowest order . and this is the message of this type of axiomatics : within the axiomatic system of audretsch and lmmerzahl et al . , the light ray , which is elementary in the eps - approach , can be derived from reasonable axioms about the propagation of electromagnetic disturbances . as with the substitution of the mass point by a matter wave , this inquiry into the physical nature of the light ray and the corresponding reshaping of the eps - scheme seems to lead to a better understanding of the metric of spacetime . and this is exactly where our framework fits in : we also build up the maxwell equations in an axiomatic way and are even led to the signature of the metric , an achievement that needs still to be evaluated in all details . let us then summarize our findings : we outlined our axiomatic approach to electrodynamics and to the derivation of the light cone . in particular , with the help of a local and linear spacetime relation , * we found the skewon field @xmath228 ( 15 components ) and the axion field @xmath106 ( 1 component ) , * we found a _ quartic _ fresnel wave surface for light propagation . * in the case of vanishing _ birefringence , _ the fresnel wave surface degenerates and we recovered the light cone ( determining 9 components of the metric tensor ) and , together with it , the conformal and causal structure of spacetime and the hodge star @xmath229 operator . * if additionally the dilaton @xmath220 ( 1 component ) is put to a constant , namely to the admittance of free space @xmath230 [ @xmath226 in si - units ] , and the axion @xmath106 removed , we recover the conventional maxwell - lorentz spacetime relation @xmath231 . thus , in our framework , the conformal part of the _ metric _ emerges from the local and linear spacetime relation as an _ electromagnetic construct . _ in this sense , the light cone is a derived concept . one of us is grateful to claus lmmerzahl and jrgen ehlers for the invitation to the potsdam seminar . many thanks go to sergey tertychniy ( moscow ) who provided the mathematica programs for the drawing of the fresnel surfaces . financial support from the dfg , bonn ( he-528/20 - 1 ) and from intas , brussels , is gratefully acknowledged . j. ehlers , f.a.e . pirani , and a. schild , _ the geometry of free fall and light propagation , _ in : _ general relativity , papers in honour of j.l . synge _ , l. oraifeartaigh , ed . ( clarendon press , oxford , 1972 ) , pp . 6384 . s. fray , c. alvarez diez , t. w. hnsch and m. weitz , _ atomic interferometer with amplitude gratings of light and its applications to atom based tests of the equivalence principle , _ _ phys . lett . _ * 93 * ( 2004 ) 240404 ( 4 pages ) ; arxiv.org/physics/0411052 . f.w . hehl and yu.n . obukhov , _ electric / magnetic reciprocity in premetric electrodynamics with and without magnetic charge , and the complex electromagnetic field , _ _ phys . lett . _ * a323 * ( 2004 ) 169175 ; arxiv.org/physics/0401083 . c. lmmerzahl , a. macas , and h. mller , _ lorentz invariance violation and charge ( non)conservation : a general theoretical frame for extensions of the maxwell equations , _ _ phys . * d71 * ( 2005 ) 025007 ( 15 pages ) . ni , _ equivalence principles and precision experiments . _ in _ precision measurement and fundamental constants ii , _ b.n . taylor , w.d . phillips , eds . bur . stand . ( us ) spec . publ . 617 ( us government printing office , washington , dc , 1984 ) pp . 647651 . f.a.e . pirani and a. schild , _ conformal geometry and the interpretation of the weyl tensor _ , in : _ perspectives in geometry and relativity . _ essays in honor of v. hlavat . b. hoffmann , editor ( indiana university press , bloomington , 1966 ) pp . 291309 . j. schrter , _ a new formulation of general relativity . _ part i. pre - radar charts as generating functions for metric and velocity . pre - radar charts as generating functions in arbitrary space - times . gtr as scalar field theory ( altogether 58 pages ) . preprint , univ . paderborn ( july 2005 ) . tellegen , _ the gyrator , an electric network element , _ in : _ philips technical review _ * 18 * ( 1956/57 ) 120124 . reprinted in h.b.g . casimir and s. gradstein ( eds . ) _ an anthology of philips research . _ ( philips gloeilampenfabrieken , eindhoven , 1966 ) pp . 186190 . tertychniy [ national research institute for physical , technical , and radio - technical measurements ( vniiftri ) , 141570 mendeleevo , russia . email : [email protected] ] provided the mathematica programs for constructing the figures of the quartic fresnel surface . h. weyl , _ zur infinitesimalgeometrie : einordnung der projektiven und konformen auffassung , _ _ nachr . knigl . gesellschaft wiss . gttingen , math .- phys . klasse , _ pp . 99112 ( 1921 ) ; also in k. chandrasekharan ( ed . ) , _ hermann weyl , gesammelte abhandlungen _ vol.ii , 195207 ( springer , berlin , 1968 ) . h. weyl , _ raum , zeit , materie _ , vorlesungen ber allgemeine relativittstheorie , reprint of the 5th ed . of 1923 ( wissenschaftliche buchges . , darmstadt , 1961 ) . translation of the 4th ed . : _ space time matter _ ( dover publ . , new york , 1952 ) .	we consider _ spacetime _ to be a 4-dimensional differentiable manifold that can be split locally into time and space . no metric , no linear connection are assumed . _ matter _ is described by classical fields / fluids . we distinguish electrically charged from neutral matter . _ electric charge _ and _ magnetic flux _ are postulated to be conserved . as a consequence , the inhomogeneous and the homogeneous maxwell equations emerge expressed in terms of the excitation @xmath0 and the field strength @xmath1 , respectively . @xmath2 and @xmath3 are assumed to fulfill a _ local and linear _ `` spacetime relation '' with 36 constitutive functions . the propagation of electromagnetic waves is considered under such circumstances in the geometric optics limit . we forbid _ birefringence _ in vacuum and find the light cone including its lorentzian signature . thus the conformally invariant part of the metric is recovered . if one sets a scale , one finds the pseudo - riemannian metric of spacetime .
the efficient and reliable detection of the quantum mechanical state of a nanoscale system is a key component of all present designs of quantum circuits.@xcite one nondestructive readout scheme currently in use for the important class of superconducting flux qubits is based on a heterodyne detection of the dynamic response of a dc superconducting quantum interference device ( dc - squid ) detector which is inductively coupled to the qubit . @xcite thereby , the dc - squid is operated in its linear regime as a shunted variable inductor in a resonant circuit . in this set - up , its resonance frequency depends on the magnetic flux generated by the qubit being in the ground or excited state . hence , measuring the impedance of the resonant circuit as a function of an externally applied bias current yields two characteristic lorentzian resonances at two different resonance frequencies , which depend on the two qubit states . this detection scheme , hence , allows us to infer the state of the qubit from the resonant response of the detector in the nanocircuit . in order that a reliable discrimination of the two qubit states becomes possible in this continuous type of readout design , the probability distributions for the readout values have to be only weakly overlapping . due to thermal and quantum fluctuations , the readout naturally is a random process,@xcite and the noise properties of the nanocircuit around the detector resonances determine the discrimination power of the set - up . an alternative readout scheme is the josephson bifurcation amplifier.@xcite it is based on a classical driven nonlinear resonator and exploits the classical bifurcation point of the dynamically induced bistability with a small- and a large - oscillation state.@xcite the response ( or output ) of the nonlinear resonator around the bifurcation point is very sensitive to small changes in the circuit parameters . this is an ideal prerequisite for a sensitive detector . depending on the state of the qubit to be sensed , the resonator bifurcation point is shifted to a different frequency , allowing for large discrimination powers between the large- and small - oscillation detector state of up to @xmath0.@xcite nevertheless , since the detector is a classical macroscopic device , it introduces considerable dephasing and relaxation to the qubit state , yielding a reduced contrast of the qubit rabi oscillations of less than @xmath1.@xcite this implies that the thermal noise properties of the nonlinear detector ( together with semiclassical corrections due to quantum fluctuations ) around the classical bifurcation point determine the discrimination power between the two states close to the classical bifurcation point . @xcite hence , it would be desirable to combine the advantage of a large discrimination power of a nonlinear detector with the reduced noise sensitivity of a nanocircuit operated close to the quantum regime . in this paper , we introduce a combination of both strategies and propose a nonlinear detector scheme in the form of a nonlinear resonator with an amplitude modulated drive in its few - photon deep quantum regime . in particular , in this regime , we shall exploit sharp multiphoton resonances in the nonlinear resonator , @xcite which are induced by the external driving field close to the fundamental resonator frequency . they can be used for the detection of the states of the qubit and offer the advantage of being rather sharp and externally tunable by varying the parameters of the external drive . the concept is an extension of the case of a linear resonator , where the fundamental resonance frequency is shifted depending on the qubit state . however , the multiphoton resonances in the nonlinear detector close to the detector s fundamental frequency show very small line widths . the width of the @xmath2-photon resonance is determined by the corresponding @xmath2-photon rabi frequency , which decreases with increasing photon number . the sharp resonance lines , in turn , offer the advantage that only a few measurement cycles are necessary to ensure a large discrimination power . to understand the back action of the nonlinear multiphoton detector on the qubit state , we determine the relaxation rate of the qubit due to the coupling to the driven dissipative nonlinear oscillator around a multiphoton resonance . notably , the back action of the resonator on the qubit is sufficiently weak , yielding to a good qubit - state measurement fidelity . furthermore , we show that the discrimination power of the set - up is rather large and beyond @xmath0 for our choice of realistic parameters of a flux qubit circuit . in fact , it gives rise to an enhanced measurement fidelity as compared to the linear parametric oscillator . furthermore , we show that the nonlinear multiphoton detector does not have a worse measurement efficiency as compared to the linear detector scheme . we determine the measurement efficiency of the set - up via the ratio of the time it takes to collect enough information on the qubit state ( measurement time ) and the relaxation time . it turns out that the measurement efficiency does not considerably decrease as compared to the linear case . hence , the detection scheme indeed has the advantage of an overall reduced back action in combination with an enhanced discrimination power , together with a sufficiently large measurement efficiency . an experimental realization of a driven nonlinear resonator in its few - photon quantum regime is in principle possible with present set - ups and technology . in a recent experiment,@xcite a nanoscale superconducting microwave resonator has been driven to its nonlinear regime by fast frequency - chirped voltage pulses . at low enough temperature , the regime of quantum noise has been reached . in this experiment , the applied driving strength has been rather large , which corresponds to a large photon number transferred to the resonator . no particular few - photon resonances have been revealed and the nonlinear response is similar to previous schemes on classical bifurcation detectors using a time - dependent driving frequency.@xcite . however , the route to the few - photon regime seems to be clear . the paper is organized as follows . in sec . [ sec : model ] , we start from a typical experimental setup for the flux qubit and its squid detector and we derive the hamiltonian model . this serves to motivate an experimental realization of our proposed detection scheme . moreover , we discuss the regime of validity of the model . in accordance with the approximation made in sec.[sec : model ] , we continue the study of the coherent dynamics in sec . [ sec : floquet ] in the rotating wave approximation . dissipative coupling to the environment is included on the level of a born - markov master equation in the rotating frame in sec.[sec : diss ] . in sec . [ sec : detdyn ] , we analyze the multiphoton transitions in the nonlinear response of the duffing oscillator and show that their resonance frequency depends on the qubit state . then , in sec.[sec : backqubit ] we determine the back action of the driven dissipative detector on the qubit dynamics by analyzing the population difference of the qubit states at the multiphoton transitions in the detector . furthermore , we determine the measurement efficiency in sec . [ sec : queff ] . in order to relate the theoretical approach in the following to realistic devices , we start by deriving the model from standard set - ups already realized in experiments . for this , we use a typical architecture of a persistent current qubit which is inductively coupled to a driven squid . we consider the experimental set - up used in ref . for the qubit , consisting of a superconducting loop interrupted by three josephson junctions , two of which have equal josephson energies , while the coupling energy of the third is smaller , in order to yield a double - well potential configuration . in this low - inductance circuit , the flux through the loop remains close to the externally applied value @xmath3 . when the latter is close to @xmath4 , where @xmath5 is an integer and @xmath6 is the flux quantum , the device is described by the hamiltonian in terms of the pauli matrices @xmath7 as ( @xmath8 ) @xmath9 with the two eigenstates @xmath10 and @xmath11 of @xmath12 corresponding to the two persistent current states @xmath13 . the minimal energy level splitting @xmath14 and the current @xmath15 are determined by the charging and josephson energies of the josephson junctions . the asymmetry is given by @xmath16 . in the energy eigenbasis , the hamiltonian follows as @xmath17 , with @xmath18 , and @xmath19 is the corresponding pauli matrix with @xmath20 . the detection of the qubit state essentially involves the measurement of the magnetic flux produced by the persistent current states . to this end , one can use the driven squid as a sensitive magnetometer , @xcite operating in its nonlinear region . below , we will restrict to the few - photon deep quantum regime . we consider the standard setup of a dc - squid formed by two josephson junctions in a superconducting loop , but subject to a time - dependent external bias current.@xcite moreover , we assume a negligible ring inductance @xmath21 of the squid ( low - inductance approximation ) . @xcite in this configuration , the superconducting phase differences at each junction , @xmath22 and @xmath23 , play the role of dynamical variables with a constraint given by the flux quantization , i.e. , @xmath24 , where @xmath25 is the external magnetic flux piercing the superconducting loop and @xmath26 . note that within the low - inductance approximation , @xmath27 with the critical current @xmath28 of the squid . thus , the system is described by the generalized coordinate @xmath29 , with the effective lagrangian @xcite @xmath30 where we have assumed a symmetric loop , with @xmath31 as the josephson energy , and @xmath32 as the capacitance of each junction . moreover , we include a time - periodic ac current @xmath33 with frequency @xmath34 and amplitude @xmath35 injected `` into '' the loop . the above lagrangian describes an effective superconducting loop ( with a negligible ring inductance ) with a single josephson junction @xcite with a tunable josephson energy @xmath36 , critical current @xmath37 , cross - junction phase difference @xmath38 , and capacitance @xmath39 . in order to tune the resonance frequency , the squid is shunted @xcite with a capacitance @xmath40 . next , we shall establish the optimal working point of the qubit - detector system , where the dissipative influence entering via the detector is minimal . the qubit and the squid are coupled by means of their mutual inductance @xmath41.@xcite thereby , the squid induces the flux @xmath42 in the qubit loop , where @xmath43 is the circulating current in the squid . the latter can be determined by using current conservation in the loop and the josephson relations for the two junctions in the squid . for the symmetric squid,@xcite it follows that @xmath44 . thus , the total magnetic flux in the qubit is affected by its coupling with the squid , and it is composed of the external flux and the squid - generated contribution , i.e. , @xmath45 . this implies that the energy bias of the qubit acquires a contribution that depends on the circulating current in the squid , leading to the effective asymmetry @xmath46 , where @xmath47 . therefore , two sources of noise can affect the qubit dynamics , i.e. , the fluctuations from the external flux @xmath3 and from the bias current @xmath48 in the squid,@xcite which is related to @xmath38 by the josephson equation @xmath49 . by tuning the bias current to the critical value @xmath50 characterized by @xmath51 , the influence from current fluctuations in the squid can be minimized @xcite and the optimal working point is reached . for a nonsymmetric squid , the lowest - order contribution is linear in @xmath52,@xcite while in the symmetric case this lowest - order contribution vanishes , which implies that around the optimal working point the phase @xmath38 is very small , @xmath53 . in the following , we consider a setup close to the optimal point , where we can expand the expression for @xmath43 up to second order in @xmath38 , yielding the interaction term @xmath54 with the coupling constant @xmath55 . as we operate the detector in its nonlinear regime , we expand the potential term @xmath56 in eq . ( [ lagrangian ] ) around the optimal point up to fourth order in @xmath38 , where @xmath57 is the effective mass , @xmath58 the corresponding frequency , and @xmath59 the strength of the nonlinearity . we switch to a description in terms of the creation and annihilation operators @xmath60 and @xmath61 , defined by @xmath62 with @xmath63 the zero - point energy of the phase @xmath38 . adding the time - dependent driving term yields us to the hamiltonian of the driven squid described by the quantum duffing oscillator model @xmath64 with nonlinearity and driving strength given by @xmath65 , and @xmath66 , respectively . similarly , the interaction hamiltonian in terms of ladder operators reads as @xmath67 with @xmath68 . notice that @xmath69 and @xmath70 depend on the external flux @xmath71 , i.e. , they are tunable in a limited regime with respect to the desired oscillator frequency @xmath72 , where the coupling term is considered as a perturbation to the squid ( @xmath73 ) , in order to keep the dynamics of the oscillator to dominate . the dependence of the dimensionless ratios @xmath74 and @xmath75 is shown in fig . [ fig0 ] . and @xmath75 on the external flux @xmath76 in the squid . the parameters of the squid are chosen as @xmath77 pf , @xmath78 na , @xmath79 na , and @xmath80 ph . @xcite [ fig0],width=264 ] we restrict to parameters of the external magnetic flux in the squid loop , which generate a weak nonlinearity and a weak qubit - detector coupling strength , @xmath81 , i.e. , for @xmath82 . a typical dependence of both parameters for typical experimental parameters is shown in fig . both cases of @xmath83 and @xmath84 can be achieved . for our purpose of a qubit - detector setup , the qubit - resonator coupling typically will be required as small enough in order to ensure a minimal back action . on the other hand , the qubit - detector coupling should be large enough so that an efficient detection of the qubit state becomes possible . as is shown in fig . [ fig0 ] and will be quantitatively discussed in the sequel of this paper , this can indeed be achieved for realistic parameters . moreover , the choice of the parameter regime also justifies us to restrict the influence of the resonator coupling on the effective qubit bias to lowest order in @xmath38 only . eventually , the total system is described by the hamiltonian @xmath85 . before we address the dynamics of the detection scheme based on the nonlinear response of the duffing oscillator to the applied periodic driving in the stationary regime , we discuss the coherent dynamics generated by @xmath86 , which is periodic in time . here , we are interested in exploiting few - photon transitions in the detector around the fundamental detector frequency @xmath72 . hence , higher harmonics have a small amplitude and can effectively be neglected . furthermore , we focus on the regime of weak nonlinearity , weak driving , and weak qubit - detector coupling as characterized by @xmath87 . the proposed mechanism of detection is most conveniently discussed in the simplest case , when the dynamics occurs close to the fundamental oscillator resonance @xmath88 . then , the rotating - wave approximation ( rwa ) can be invoked in order to obtain a simple interpretation in terms of few - photon transitions . in passing , we note that we have also performed a complete analysis in terms of full floquet theory , thereby avoiding the rwa . for all cases shown below , both approaches yield coinciding results . we switch to the rotating reference frame by the transformation @xmath89 . then , the rwa eliminates the fast oscillating terms from the transformed hamiltonian @xmath90 and the time - independent schrdinger equation in the rotating frame @xmath91 follows , with the rwa hamiltonian given by @xmath92 with @xmath93 the detuning frequencies follow as @xmath94 and @xmath95 , and @xmath96 . the quasienergies @xmath97 and the rwa eigenstates @xmath98 result from a straightforward numerical diagonalization of @xmath99 . in the static frame , an orthogonal ( at equal times ) set @xmath100 of approximated solution of the schrdinger equation follows as @xmath101 here , the quasienergy states @xmath102 are time periodic with period @xmath103 and form a complete basis that will be used below for the description of the dissipative dynamics . we note that an analytic expression for the multi - photon resonances would follow from a van - vleck perturbative approach in a similar manner as for the pure quantum duffing oscillator.@xcite however , the resulting expression will be cumbersome and not further illuminating for the present purpose . we note , furthermore , that the qubit - detector interaction occurs via a parametric coupling @xmath104 , and via a two - photon coupling @xmath105 . the electronic nanocircuit is embedded in a dissipative environment . in particular , the squid is shunted with an ohmic resistor , which yields dissipative fluctuations @xmath106.@xcite we focus to the case of an underdamped squid , where the shunt resistance is large , @xcite and use the standard harmonic bath in order to model the fluctuations , which are rooted in current fluctuations and can be encoded in the ohmic spectral density @xmath107.@xcite they couple to the resonator s dipole operator , i.e. , @xmath108 . we note that , in the same way , the direct coupling of the qubit to the electromagnetic fluctuations could be included . however , we have checked @xcite that for a related set - up of a flux qubit coupled to a harmonic oscillator , such a direct dissipation of the qubit yields only minor quantitative corrections , which should be included in a quantitative description of an experiment,@xcite but do not add qualitatively new physics . the time evolution of the reduced density operator @xmath109 is described in terms of a standard markovian master equation projected onto the basis of the quasienergy states @xmath110 @xmath111 where @xmath112 . the dissipative transition rates are given by @xcite @xmath113 with @xmath114 and @xmath115 being the fourier components according to @xmath116 . furthermore , we have used the planck numbers @xmath117 , where @xmath118 $ ] with @xmath119 , temperature @xmath120 and @xmath121 being the heaviside function . since within the rotating wave approximation , @xmath122 , the only non - zero fourier components are @xmath123 , and @xmath124 and the master equation ( [ fme ] ) considerably simplifies as it involves only single step transitions , i.e. , one - photon emission ( for @xmath125 ) into and absorption ( for @xmath126 ) processes from the bath . we note that neglecting also the quasienergy dependence of the planck numbers would yield the well - known lindblad master equation . in order to measure the dynamic response of the resonator to the external drive at asymptotically long times , a heterodyne detection scheme such as in ref . can be used,@xcite where the coupled qubit - oscillator system approaches the steady state @xmath127 . we determine the stationary solution characterized by @xmath128 numerically . for this , we solve the corresponding eigenvalue problem and @xmath129 follows as eigenvector to the eigenvalue zero . with this , we compute the nonlinear response of the detector , characterized by the mean value @xmath130 at asymptotic times . as we restrict the discussion to the regime close to the first harmonic ( small detuning ) , higher harmonics can be neglected and we immediately obtain @xmath131 as the system is driven with frequency @xmath132 , @xmath133 also oscillates with time . its amplitude is given by @xmath134 correspondingly , we evaluate the population difference @xmath135 of the qubit states and obtain @xmath136 where @xmath137 . the population difference oscillates , with a maximal value given by @xmath138 of the nonlinear response of the decoupled quantum duffing detector ( @xmath139 ) as a function of the external driving frequency @xmath132 . ( b ) the corresponding quasienergy spectrum @xmath97 . the labels @xmath2 denote the corresponding @xmath2-photon ( anti-)resonance . the parameters are @xmath140 , @xmath141 , @xmath142 , and @xmath143 . [ fig1],width=302 ] before turning to the quantum detection scheme , we discuss the dynamical properties of the isolated detector , which is the quantum duffing oscillator . a key property is its nonlinearity which generates multiphoton transitions at frequencies @xmath34 close to the fundamental frequency @xmath72 . in order to see this , one can consider first the undriven nonlinear oscillator with @xmath144 and identify degenerate states , such as @xmath145 and @xmath146 ( for @xmath147 ) , when @xmath148.@xcite for finite driving @xmath149 , the degeneracy is lifted and avoided quasienergy level crossings form , which is a signature of discrete multiphoton transitions in the detector . as a consequence , the amplitude @xmath150 of the nonlinear response signal exhibits peaks and dips , which depend on whether a large or a small oscillation state is predominantly populated.@xcite the formation of peaks and dips goes along with jumps in the phase of the oscillation , leading to oscillations in or out of phase with the driving . a typical example of the nonlinear response of the quantum duffing oscillator in the deep quantum regime containing few - photon ( anti-)resonances is shown in fig . [ fig1](a ) ( decoupled from the qubit ) , together with the corresponding quasienergy spectrum [ fig . [ fig1](b ) ] . we show the multiphoton resonances up to a photon number @xmath151 . the resonances get sharper for increasing photon number , since their widths are determined by the rabi frequency , which is given by the minimal splitting at the corresponding avoided quasienergy level crossing . performing a perturbative treatment with respect to the driving strength @xmath152 , one can get the minimal energy splitting at the avoided quasienergy level crossing @xmath153 as @xcite @xmath154 because the nonlinearity @xmath70 is typically fixed by the design of the squid , the rabi frequency can be easily tuned by tuning the driving strength @xmath152 . next , we consider a finite coupling of the detector to the qubit whose state is to be sensed , i.e. , @xmath155 . the coupling inevitably induces relaxation and decoherence in the qubit , characterized by the relaxation and dephasing rate , @xmath156 and @xmath157 , respectively . typically , the detector couples only weakly to the system , i.e. , @xmath158 . then , the associated relaxation and dephasing times ( @xmath159 and @xmath160 , respectively ) are still much larger than the corresponding relaxation time scale for the detector given by @xmath161 . in passing , we note that the corresponding relaxation time around a _ resonant _ multiphoton transition ( in the underdamped case ) has been shown in refs . to be comparable to @xmath162 . moreover , we bias the qubit with a large asymmetry , @xmath163 in order to `` gauge '' the detector response . for a rough evaluation of the order of magnitude of the involved time scales , we may neglect the nonlinearity of the detector ( @xmath164 ) for the moment and estimate the effective relaxation rate for the qubit coupled to an ohmically damped harmonic oscillator.@xcite this model can be mapped to a qubit coupled to a structured harmonic environment with an effective ( dimensionless ) coupling constant @xmath165 . for the realistic parameters used in fig . [ fig1 ] and @xmath166 , we find that @xmath167 , giving rise to an estimated relaxation rate@xcite @xmath168 ( evaluated at low temperature ) . hence , this illustrates that we can easily achieve the situation where @xmath169 required for this detection scheme . then , for a waiting time ( after which we start the measurement ) much longer than the relaxation time @xmath170 of the nonlinear oscillator , but still smaller than @xmath171 , the oscillator is able to reliably detect the qubit state . of the detector as a function of the external driving frequency @xmath34 in the presence of a finite coupling @xmath172 to the qubit ( black solid line ) . the blue dashed line indicates the response of the isolated detector . the parameters are the same as in fig . [ fig1 ] and @xmath173 and @xmath174 , in correspondence to realistic experimental parameters @xcite . [ fig2],width=302 ] in fact , under these conditions , the state of the qubit , apart from the inevitable dephasing , remains unaffected in a time window before it reaches its global stationary state and an effective shift of the oscillator s eigenfrequency arises due to the parametric coupling term @xmath175 in eq . ( [ rwahamiltonian ] ) . treating the qubit - detector interaction term in eq . ( [ squidh ] ) perturbatively to lowest order in @xmath69 , the eigenfrequency shift follows straightforwardly as @xmath176 thus , the nonlinear response is shifted by @xmath177 ( @xmath178 ) if the qubit is prepared in the state @xmath179 ( @xmath180 ) . this is illustrated in fig . [ fig2 ] , in which we show the nonlinear response of the resonator for the uncoupled ( blue dashed line ) and the coupled ( black solid line ) case . of the detector coupled to the qubit prepared in its ground state @xmath181 ( orange solid line ) and in its excited state @xmath182 ( black dashed line ) for the same parameters as in fig . the quadratic qubit - detector coupling induces a global frequency shift of the response by @xmath183 . ( b ) discrimination power @xmath184 of the detector coupled to the qubit for the same parameters as in a ) . [ fig3],width=302 ] for a fixed value of @xmath69 , the shift between the two cases of the opposite qubit states is given by the frequency gap @xmath185 . figure [ fig3 ] ( a ) shows the nonlinear response of the detector for the two cases when the qubit is prepared in one of its eigenstates : @xmath182 ( orange solid line ) and @xmath181 ( black dashed line ) . an important feature of a detection scheme is that it is efficient in discriminating the states to be detected . this can be quantified by the discrimination power of the detector , which can be defined for our case as @xmath186 the result for @xmath184 is shown in fig . [ fig3 ] ( b ) . the discrimination power shows a rich structure of local maxima and minima , which indicates that it can be tuned directly by tuning the driving frequency . it is moreover important to realize that the discrimination power can be optimized by tuning @xmath69 . in the optimized case , a local maximum of the multiphoton resonance for one qubit state can be made to coincide with a local minimum of the response for the opposite qubit state yielding to a maximal discrimination power . an example where the discrimination power has been optimized with respect to the three - photon resonance is shown in fig . [ fig3 ] ( b ) . another important prerequisite for a useful detection scheme is that the coupling of the qubit to the detector around a multiphoton resonance does not generate a destructive back action on the qubit dynamics . in this section , we show that the back action in this design is surprisingly small for a realistic choice of parameters . the back action of the detector on the qubit arises in the form of two contributions from the coupling . first , this coupling has a parametric component @xmath187 , which commutes with the hamiltonian . thus , in the presence of a coupling of the oscillator to the bath , this term only produces dephasing and no relaxation , as it is , for instance , required for a quantum non - demolition measurement . this part guarantees an efficient detection of the qubit state . the second component @xmath188 in the coupling term yields transitions in the qubit when two - photon processes are induced in the detector by the external driving and/or by dissipative transition . since , at low temperature , dissipation is dominated by photon leaking and the driving is very weak , the decay rate of the qubit from its excited state to its ground state accompanied by the emission of two oscillators photons , largely exceeds the excitation rate from the ground state to the excited state accompanied by the absorption of two photons originally coming from the bath or the driving . on the other hand , when the effective oscillator frequency is close to a multiphoton resonance , photon absorption in the coupled system is enhanced and thus the asymptotic qubit population might be reduced . thus , for a large asymmetry @xmath189 , peaks and dips in the qubit population difference @xmath190 are expected when multiphoton transitions in the detector are induced . this is what is shown in fig . [ fig4](a ) , where @xmath190 is shown for several values of @xmath152 . for an easier orientation , we show in addition the corresponding stationary nonlinear response of the detector in fig . [ fig4](b ) . for increasing driving , the deviation from the expected value @xmath191 becomes more pronounced for larger photon numbers @xmath2 and larger driving @xmath152 . the reason is that , for increasing driving , a larger rabi frequency for the corresponding transition results [ see eq . ( [ rabi ] ) ] . from fig . [ fig4 ] , it follows that when the qubit is prepared in its ground state @xmath182 ( we consider @xmath163 ) the back action is very small . the impact is less than @xmath192 for the considered realistic parameters , yielding to a readout contrast of more than @xmath0 . this has to be compared with presently achievable readout contrasts of less than @xmath1 , @xcite which results from an architecture with a classical josephson bifurcation amplifier . in passing , we note that the detector response can also be calculated from the stationary solution of the master equation ( [ fme ] ) , but for the parameters considered here ( in particular because of the large qubit bias ) , this coincides with the shifted one . moreover , we note that the components @xmath193 and @xmath194 can be tuned by @xmath195 and @xmath14 . therefore , @xmath196 can in principle be eliminated by setting @xmath197 , which would imply that the measurement scheme keeps the state of the qubit without any relaxation but only pure dephasing ( ideal quantum nondemolition measurement ) . however , turning off the splitting implies a major change in the experimental design of the sample , since this parameter is determined by the josephson energy in the junctions of the superconducting flux qubit and , thus , may not be easy to be realized . of the qubit states , and ( b ) the corresponding detector response @xmath150 as a function of the external frequency @xmath198 for the same parameters as in fig . [ fig4],width=302 ] the back action of the detector on the qubit should be small not only when the qubit is in its ground state but also when it is in its excited state . we therefore address next the relaxation rate of the qubit . energy relaxation in the qubit induced by the measurement process will be proportional to the fluctuations of the square of the phase operator @xmath38 induced by the detector s environment.@xcite this relaxation process is characterized by the transition rate @xcite @xmath199,\ ] ] which has been computed perturbatively to lowest order in @xmath200 . here , @xmath201&=&\frac{\omega_{\rm ex}}{4\pi } \int_0^{2\pi/\omega_{\rm ex } } \!\!\!\!dt \int_{-\infty}^{+\infty } \!\!\ ! d\tau e^{i\omega \tau } \langle \ { \chi^2_+(\tau+t),\chi^2_+(t)\ } \rangle \nonumber \\ & & \end{aligned}\ ] ] is the symmetrized power spectrum of @xmath202 averaged over the period of the external driving ( see appendix for details ) , with @xmath203 indicating the anticommutator . the fact that information on the qubit state is acquired in the detector via the same channel by which dissipation is introduced is nicely reflected in the expression of the relaxation rate in eq.([relaxrate ] ) . in fig . [ fig5](a ) , the relaxation rate @xmath156 is shown for a large negative asymmetry in the qubit . the relaxation rate is strongly peaked around the multiphoton transitions . there , the noise from the detector absorbs more energy from the qubit around the multiphoton transition @xmath153 since the parametric component @xmath193 of the coupling becomes negligible , leading to a dominant relaxation process induced by @xmath194 . we emphasize that although the relaxation is maximally enhanced at a multiphoton resonance , the absolute value of @xmath156 is still very small in comparison to the damping constant , e.g. , @xmath204 . thus , we can infer the qubit state with sufficient precision by operating the detector in its steady state regime as it has been assumed in section [ nobackaction ] . the measurement of the qubit state requires a coupling to the outer world , which clearly introduces noise to the qubit . in turn , the noisy detector yields measurement results , which are statistically distributed . this implies that several measurements have to be performed to obtain a reliable statistics . hence , the relaxation time of the qubit state should not only exceed the typical relaxation time of the detector but also the time it takes to acquire sufficient information to infer the qubit state ( the measurement time @xmath205 ) . hence , for a good measurement fidelity , @xmath206 should be smaller than the characteristic time @xmath171 given by eq . , or , @xmath207 . the measurement time can be formalized @xcite as the ratio of the symmetrized power spectrum @xmath208 of the phase operator @xmath38 ( evaluated at zero frequency ) and the square of the difference between the two expectation values of @xmath38 when the qubit is in the two opposite states , i.e. , with eq . ( [ domex ] ) , @xmath209 ^ 2 } % ( \langle \chi_+ \rangle_\uparrow - % \langle \chi_+ \rangle_\downarrow)^2 } \,\ , .\ ] ] the result for @xmath205 as a function of @xmath34 is shown in fig . [ fig5 ] b ) for the parameter set used above , for which the discrimination power @xmath184 around the 3-photon resonance has been maximized . in correspondence with this is the relative minimum of @xmath205 around the 3-photon resonance , see fig . [ fig5 ] b ) . interestingly enough , the time scale of the measurement time around this resonance is @xmath210 . considering realistic numbers of a typical experimental set - up @xcite , where @xmath72 is in the regime of a few ghz , we obtain a time scale of @xmath211 ps for the nonlinear quantum detection scheme . this should be contrasted to the measurement time of @xmath212 ns obtained in ref . . in between the multiphoton resonances , the dependence of @xmath205 on @xmath34 shows a rich structure including several singularities , which are simply due to the several crossings of the two nonlinear response curves shown in fig . [ fig3 ] a ) , where @xmath184 becomes zero , implying insufficient discrimination of the two qubit states . with this , we can evaluate the measurement efficiency , defined by the ratio @xmath213 , with @xmath214 . this quantity sets the probability to infer the qubit state , based on the nonlinear response of the detector . we show the result for the efficiency of the measurement in fig . [ fig5](c ) . related to the multiphoton resonances in the detector , the efficiency also shows local maxima . for the discrimination power being optimized around the three - photon resonance , the measurement efficiency displays a clear local maximum [ see fig . [ fig5](c ) ] . due to the small size of the relaxation rate @xmath156 of the detector , the overall measurement efficiency is rather large in comparison to the detection set - up with a linear resonator , @xcite ensuring @xmath207 . of the nonlinear quantum detector , ( b ) the measurement time @xmath205 , and ( c ) the measurement efficiency @xmath215 as a function of the external frequency @xmath132 . the parameters are the same as in fig . [ fig5],width=321 ] to conclude , we have introduced a scheme for quantum state detection on the basis of a nonlinear detector which is operated in the regime of resonant few - photon transitions . discrete multiphoton resonances in the detector can be used to infer the state of the parametrically coupled qubit via a state - dependent frequency shift of the detector s nonlinear response function . the multiphoton resonances are well separated in the spectrum and sharp enough to allow for a good resolution of the qubit state . by analyzing key quantities of the detector , we have shown that the nonlinear few - photon detector can be operated efficiently , reliably , and with sufficiently weak back action . in fact , it can be efficiently tuned by tuning the amplitude of the ac bias current of the squid . furthermore , we have shown that the sharpness of the multiphoton resonances can be used to obtain an increased discrimination power as compared to the linear parametric detection scheme . clearly , the relaxation rate at a multiphoton resonance for the qubit becomes maximal , but in general remains very small . the measurement time around a multiphoton resonance can be tuned such that it becomes minimal . for realistic experimental parameters , we find surprisingly small measurement times , allowing in principle for fast measurements . moreover , the efficiency of the measurement , which takes the time to acquire enough information to infer the qubit state into account , also assumes large values , thus allowing for a reliable and highly efficient measurement of the qubit state . we have chosen realistic values for the involved model parameters such that an experimental realization of this quantum measurement scheme should become possible in the near future . the nonlinear detection scheme in the deep few - photon quantum regime offers thus the advantage of an increased discrimination power of more than @xmath0 ( for our choice of realistic parameters ) , as compared to previous classical detection schemes based on the josephson bifurcation amplifier . a possible setup in order to realize the nonlinear few - photon detector could be the architecture used in a recent experiment.@xcite the low - temperature regime , where quantum noise effects are important , has already been reached . in order to operate in the regime of only few photons in the resonator , the sensitivity and stability of the devices might have still to be further increased . however , no principle obstacles are apparent . this work was supported by the daad ( german academic research service ) research grant no . ref : a/08/73659 . v. p. was supported by the nsf ( grant no . emt / qis 082985 ) . we thank p. nalbach for valuable discussions . to calculate the power spectrum of a driven out - off - equilibrium system is a non trivial task , since the time reversal symmetry , which simplifies the calculation in equilibrium systems , is not given anymore . an elegant way to compute this in terms of correlation functions is presented in ref . . the correlation function between two operators @xmath150 and @xmath216 is described as a mean value @xmath217 here , the trace is over the whole system - plus - bath , with the `` density '' operator @xmath218 u^\dagger ( t+\tau , t)$ ] and @xmath219 , with @xmath220 being the hamiltonian of the whole system - plus - bath , @xmath221 the density operator of the total system , and @xmath222 the time ordering operator . furthermore , @xmath150 and @xmath216 are in the heisenberg representation . in the superoperator notation , @xmath224 ( liouville superoperator ) is represented by a @xmath225 supermatrix @xmath226 , where @xmath227 is the number of effective states in the system . in the same way , the density operator @xmath228 formally is a @xmath229 dimensional column vector @xmath230 . the solution of the master equation is reduced to an eigenvalue problem of the matrix @xmath231 , as @xmath232 where @xmath233 and @xmath234 are the left and right eigenvectors , respectively , with eigenvalue @xmath235 . in the superoperator notation , the master equation is expressed as @xmath236 , and its solution is given by @xmath237 in the regime of the rwa , and at low temperature the master equation conserves the trace and the positivity of the density operator , i.e. , it assumes lindblad form . therefore , we can expand the solution of the master equation in terms of the right - eigenvector @xmath233 @xmath238 with @xmath239 . here , we have used the orthogonality property @xmath240 . the corresponding expression follows for the operator @xmath241 , but with a different initial condition . it is easily understandable in the operator notation @xmath242 where @xmath243 , and @xmath244 ( @xmath245 ) is the operator representation of @xmath233 ( @xmath234 ) in the quasienergy states . considering an initial time in the stationary regime , i.e. , @xmath246 , and after averaging the initial time @xmath247 over the period @xmath248 , the correlation function reads @xmath249 with @xmath250 where @xmath251 and @xmath252 are the coefficients of the fourier expansion of @xmath253 and @xmath254 , respectively . the power spectrum is obtained directly from the fourier transform of eq . ( [ corrf ] ) . l. s. bishop , j. m. chow , j. koch , a. a. houck , m. h. devoret , e. thuneberg , s. m. girvin , and r. j. schoelkopf , nature phys . * 5 * , 105 ( 2009 ) . d. m. larsen and n. bloembergen , opt * 17 * , 254 ( 1976 ) m. i. dykman and m. v. fistul , phys . b * 71 * , 140508 ( 2005 ) .	we introduce a detection scheme for the state of a qubit , which is based on resonant few - photon transitions in a driven nonlinear resonator . the latter is parametrically coupled to the qubit and is used as its detector . close to the fundamental resonator frequency , the nonlinear resonator shows sharp resonant few - photon transitions . depending on the qubit state , these few - photon resonances are shifted to different driving frequencies . we show that this detection scheme offers the advantage of small back action , a large discrimination power with an enhanced read - out fidelity , and a sufficiently large measurement efficiency . a realization of this scheme in the form of a persistent current qubit inductively coupled to a driven squid detector in its nonlinear regime is discussed .
the study of the evolution of self - gravitating systems ( even in the spherically symmetric case ) requires the use of numerical procedures and/or the introduction of simplifying assumptions . the former usually lead to model dependent conclusions and the later are frequently too restrictive and/or deprived of physical justification . an alternative path to this question consists in perturbing the system , compeling it to withdraw from equilibrium state . evaluating it after its departure from equilibrium , it is possible to study the tendency of the evolution of the object . this is usually done following a first order perturbative method which neglects cuadratic and higher terms in the perturbed quantities . if the relevant processes occuring in the self - gravitating object take place on time scales which are of the order of , or smaller than , hydrostatic time scale , then the quasistatic approximation fails ( e.g. during the quick collapse phase preceding neutron star formation ) . in this case it is necessary to evaluate the system immediately after its departure from equilibrium , where _ immediately _ means on a time scale of the order of relaxation times . this approach has proven to be useful in the dissipationless case @xcite . nevertheless , it has been recently found @xcite that for non - viscous dissipative case this approach can not always be applied . in particular , the goodness of the first order perturbation method can be examined by means of the value of the local parameter @xmath0 where @xmath1 is the thermal conductivity , @xmath2 is the temperature , @xmath3 is the relaxation time for thermal signals and @xmath4 and @xmath5 are the energy density and the radial pressure respectively . if @xmath6 then it has been shown that @xcite the effective inertial mass density of a fluid element vanishes and becomes negative if @xmath7 . the point for which the system reaches condition @xmath6 is called the _ critical point_. this strange behaviour of matter , at , and beyond the critical point might suggest that first order perturbation method fails under such conditions . in some cases , systems with a value of @xmath8 close to , or beyond , the critical point are forbidden by causality conditions , but this is not always true . the existence of this critical point seems to take a special relevance for the last ones . efectively , if causality conditions do not forbid the critical point , then there can exist systems that can not be studied using a first order perturbation method . furthermore , causality conditions @xcite have been found by means of a perturbative method up to first order in the perturbed quantities . thus , for such systems , causality conditions must be taken with care . the aim of this paper is to elucidate the existence of a similar critical point in dissipative viscous systems . to do this we shall assume that , initially , the system is either static or slowly evolving along a sequence of states in which it is not only in hydrostatic equilibrium , but also thermally adjusted . then , we shall perturb the dissipative flows and radial velocity ( as seen by a minkowskian observer ) . as it has been mentioned above , the system must be evaluated on a time scale of the order of relaxation times after the perturbation takes place . thus , the properties of the system are still the same , and only time derivatives of perturbed quantities have changed appreciably , but not the quantities themselves . in order to find the critical point we shall use the transport equations for the dissipative flows and the radial momentum conservation equation . the eckart - landau transport equations @xcite imply a vanishing relaxation time for dissipative flows . the adoption of these equations is not advisable here for two reasons : first , they predict an infinite speed for thermal and viscous signals propagation and unstable equilibrium states @xcite . second , we are evaluating the system immediately after perturbation ( in the sense described above ) . thus , to be consisten with this choice we must use transport equations with non vanishing relaxation times . in this work , transport equations are introduced using the mller - israel - stewart second order phenomenological theory for dissipative fluids @xcite . after the condition for the critical point is found we contrast stability and causality conditions with this one . we shall see that , in some cases , the critical point may be reached without violating causality conditions and , as it has been mentioned above , causality conditions must be used with caution . finally , we show that neutrino trapping during gravitational collapse @xcite can lead to values of @xmath8 beyond the critical point . the paper is organized as follows . in the next section the field equations , the conventions , and other useful formulae are introduced . in section 3 we briefly present transport equations . in section 4 the full system of equations is evaluated at the time when the object starts to depart from equilibrium finding the expression for the critical point . finally , a discussion of the reliability of stability and causality conditions is given in the last section . we adopt metrics of signature -2 and geometrised units @xmath9 throughout the text ( except in the example preesented in last section ) . the quantities subscripted with @xmath10 denote that they are evaluated at the surface of the sphere . we consider spherically symmetric distributions of collapsing viscous fluid , undergoing dissipation in the form of heat flow , and bounded by a spherical surface @xmath11 . in bondi coordinates @xcite the line element takes the form @xmath12 -r^2\left ( d\theta ^2+sin^2\theta d\varphi ^2\right ) , \label{1}\ ] ] where @xmath13 is a timelike coordinate ( @xmath14 ) , @xmath15 is a null coordinate ( @xmath16 ) , and @xmath17 and @xmath18 are the usual angle coordinates . the @xmath19-coordinate is the retarded time in the flat space - time and therefore , @xmath20constant surfaces , are null cones open to the future . @xmath21 and @xmath22 are functions of @xmath19 and @xmath23 , and the mass function , @xmath24 , can be defined as @xcite @xmath25 bondi and schwarzschild coordinates ( @xmath2 , @xmath26 , @xmath27 , @xmath28 ) are related by means of the expressions @xmath29 and @xmath30 on the other hand , local minkowskian coordinates ( @xmath31 , @xmath32 , @xmath33 @xmath34 ) are related to bondi s radiation coordinates by @xmath35 @xmath36 @xmath37 @xmath38 for a local minkowskian observer comoving with the fluid , the stress - energy tensor splits into three terms . first , an anisotropic material part , @xmath39 where , @xmath40 , @xmath41 , @xmath42 denotes the material energy density , @xmath43 refers to the material pressure for this observer , and @xmath44 is the material part of the tangential pressure . therefore , @xmath45 refers to the material anisotropy . the second one is the radiation term , which in this lagrangian frame reads @xcite @xmath46 where @xmath47 denotes the radiation energy density , @xmath48 the heat flow , @xmath49 the radiation pressure , and @xmath50 . finally , the viscous part can be written as @xmath51 where @xmath52 is the spatial projection tensor , and @xmath53 is the bulk viscous pressure . the traceless viscous pressure tensor @xmath54 takes , for this observer , the form @xmath55 where @xmath56 is the shear viscous pressure . thus , for a local observer comoving with the fluid the stress - energy tensor in local minkowskian coordinates is @xmath57 where @xmath58 @xmath59 is the total energy density , @xmath60 is the radial pressure , @xmath61 is the non - viscous radial pressure , and @xmath62 is the tangential pressure . these physical variables are obtained as measured by this minkowskian observer , and the effects of gravitation are introduced by means of the local coordinate transformation ( @xmath63 ) between minkowski coordinates and the bondi ones ( [ local1]-[local4 ] ) . the dynamics of the system ( the radial velocity of a fluid element as measured by a minkowskian observer at rest in bondi coordinates _ , _ @xmath64 ) can be studied applying a lorentz boost , @xmath65 , in the radial direction to @xmath66 thus , the stress - energy tensor as measured by an observer using bondi coordinates , with a radial velocity , with respect to the matter configuration @xmath67 , is given by the expression @xcite @xmath68 _ i.e. _ @xmath69 with @xmath70 @xmath71 and @xmath72 note that @xmath73 the traceless viscous tensor is , for this observer @xmath74 thus , einstein equations for the line element ( [ 1 ] ) , read @xmath75 = \frac{% 1\ } { 1-w^2}\left ( \rho + 2wq+pw^2\right ) , \label{9}\ ] ] @xmath76 @xmath77 and @xmath78 \right ) = -8\pi p_{\perp } , \label{12}\ ] ] where subscripts @xmath79 and @xmath80 denote partial derivative with respect to @xmath19 and @xmath23 coordinates respectively . from ( [ 9 ] ) and ( [ 10 ] ) , it follows @xmath81 = \frac{1\ } { 1-w^2}\left ( \rho w+q\left ( 1+w^2\right ) + pw\right ) . \label{15}\ ] ] next , from the conservation equation @xmath82 , we obtain after long but simple calculations @xmath83 -\frac 2r\left ( p_{\perp } -p\right ) -\frac 2r\left ( p-\widetilde{p}\right ) = 0 , \label{16}\ ] ] where @xmath84 , \label{17}\ ] ] and @xmath85 . \label{18}\ ] ] finally , taking the @xmath19-derivative of ( [ 11 ] ) and using ( [ 2 ] ) , ( [ 13 ] ) , ( [ 14 ] ) , ( [ 9 ] ) , ( [ 10 ] ) and ( [ 15 ] ) @xmath86 , \label{19}\ ] ] where @xmath87 and from ( [ 9 ] ) , ( [ 2 ] ) , and ( [ 10 ] ) , we obtain @xmath88 \left ( \rho w+q(1+w^2)+pw\right ) . \label{20}\ ] ] all these equations will be used in section 4 . outside fluid distribution the metric is the vaidya one @xcite , a particular case of the bondi metric with @xmath89 , and @xmath90 . the continuity of the first and second fundamental forms across @xmath11 leads to the well - known result @xcite @xmath91 or equivalently ( see @xcite for details ) @xmath92 as we mentioned before we shall use the mller - israel - stewart second order phenomenological theory for dissipative fluids @xcite . although it may be not reasonable in some situations , we shall assume here for simplicity , that there is not viscous / heat coupling ( _ i.e. _ @xmath93 in @xcite ) . thus , transport equations read @xcite @xmath94 @xmath95 and @xmath96 where @xmath97 is the projector onto the three space orthogonal to @xmath98 , @xmath99}$ ] is the vorticity , @xmath100 is the expansion scalar , and @xmath101 , @xmath102 , and @xmath103 denote the thermal conductivity , and the bulk and shear viscous coefficients respectively . also , @xmath2 , @xmath104 , @xmath105 , and @xmath106 denote temperature and relaxation times respectively . overdot denotes @xmath107 , and the shear tensor is given by @xmath108 the traceless viscous pressure tensor @xmath109 is given by the expression ( [ 25 ] ) . observe that , due to the symmetry of the problem , equations ( [ 21 ] ) , and ( [ 23 ] ) only have one independent component @xcite . let us now write the expressions for different terms in ( [ 21 ] ) , they are @xmath110 + \frac{\widetilde{m}_{,1}}r-\frac{\widetilde{m}}{r^2}+% \frac{\widetilde{m}_{,0}}v(1-w)\right ) \label{26}\end{aligned}\ ] ] @xmath111 @xmath112 @xmath113 \dot{u}_r , \label{29}\ ] ] @xmath114 @xmath115 , \label{30}\ ] ] and @xmath116}q_\nu = \tau _ \kappa h^{r\alpha } u_{[\alpha ; \beta ] } q^\beta = 0 , \label{nueva2}\ ] ] where @xmath117 @xmath118 and the expansion scalar is given by @xmath119 @xmath120 . \label{36}\ ] ] for ( [ 22 ] ) , we have @xmath121 , \label{34}\ ] ] and @xmath122 @xmath123 . \label{35}\ ] ] finally , for the different terms in ( [ 23 ] ) we get @xmath124 @xmath125 , \label{38}\ ] ] @xmath126 , \label{39}\ ] ] @xmath127 @xmath128 , \label{40}\ ] ] and @xmath129 note that for this observer the shear scalar and vorticity scalar are given by @xmath130 and @xmath131}u_{[\gamma ; \delta ] } = 0,\ ] ] respectively . transport equations ( [ 21]-[23 ] ) , together with ( [ 16 ] ) will be evaluated after the system departs from equilibrium , neglecting terms of order @xmath132 and higher . we assume that , before perturbation , the system is slowly evolving along a sequence of states in which it is close to hydrostatic equilibrium and thermally adjusted - the so called _ complete equilibrium _ . thus , the radial velocity , as seen by a minkowskian observer , is small . this means that cuadratic and higher terms in @xmath64 may be neglected in a first order perturbation theory . a system is thermally adjusted if it changes its properties considerabily only within a time scale @xmath133 that is large as compared with the kelvin - helmholtz time scale @xmath134 . thus , before perturbation we can assume that the @xmath19-derivatives of the perturbed quantities can be neglected up to first order , and consequently @xmath135 on the other hand , the hydrostatic equilibrium can be justified in terms of the characteristic times : if the hydrostatic time scale @xmath136 is much shorter than the kelvin - helmholtz time scale @xmath137 then inertial terms in the equation of motion @xmath138 can be ignored . this condition will be accomplished for small values of luminosity @xmath139 , and consequently for small values of @xmath140 thus , before perturbation , we can assume @xmath141 in the whole system . it seems also reasonable to assume that in such system , bulk viscous pressure and shear viscous pressure must be also small ( _ i.e. _ @xmath142 ) . note that from ( [ 15 ] ) and ( [ 20 ] ) @xmath143 , @xmath144 and @xmath145 are of order @xmath146 therefore , their products and second time derivatives may be neglected ( @xmath147 ; @xmath148 ) ( an invariant characterization of slow evolution may be found in @xcite ) . we shall evaluate the system immediately after perturbation ( in the sense described in the introduction ) . physically , this implies that the perturbed quantities ( @xmath149 @xmath150 @xmath56 and @xmath53 ) are still much less than unity . nevertheless , the system is departing form hydrostatic equilibrium and thermal adjustment . thus , the @xmath19-derivatives of the perturbed quantities are small but different from zero ( _ i.e. _ @xmath151 ) . thus , our initially slowly evolving system is characterized by : 1 . before perturbation @xmath152 2 . after perturbation @xmath153 in both cases we have @xmath154 @xmath155 the initially static case can also be considered . this system is characterized by : 1 . before perturbation @xmath156 2 . after perturbation @xmath157 and @xmath158 let us now start by evaluating ( [ 21 ] ) . in the static case we obtain before perturbation @xmath159 and immediately after perturbation , neglecting terms of order @xmath160 and higher , @xmath161 for bulk viscous pressure equation , we obtain in the initially static case after perturbation @xmath162 expressions ( [ 47 ] ) and ( [ 48 ] ) are also obtained for the initially slowly evolving case applying ( [ 41 ] ) and ( [ 42 ] ) . the evaluation of the equation for the shear viscous pressure ( [ 23 ] ) yields , after perturbation , for both possible initial configurations @xmath163 thus , the three transport equations ( [ 21]-[23 ] ) , evaluated after perturbation , lead to expressions which are the same for the two initial configurations . finally , let us evaluate conservation equation @xmath82 ( [ 16 ] ) after perturbation . in the initially static case , we have before perturbation condition ( [ 44bis ] ) . therefore , equation ( [ 16 ] ) becomes @xmath164 -\frac 2r\left ( p_{\perp } -p\right ) = 0 , \label{54}\ ] ] which is the equation of hydrostatic equilibrium for anisotropic fluids , and @xmath165 denotes the total outward force acting on a given fluid element . after perturbation we obtain , using ( [ 19 ] ) , ( [ 20 ] ) , and ( [ 47]-[49 ] ) @xmath166 \times w_{,0 } , \label{54b}\ ] ] where @xmath167 or equivalently @xmath168 \frac{(1 - 2m / r)}{(\rho + p)}\times \left [ 1-\alpha \right ] ^{-1}. \label{55}\ ] ] assuming the second initial case ( slowly evolving ) , conservation equation @xmath169 with ( [ 19 ] ) , ( [ 20 ] ) , and conditions ( [ 47]-[49 ] ) lead , before perturbation , to expression @xmath170 + % \widetilde{r}-\frac 2r\left ( p-\widetilde{p}\right ) , \label{56}\ ] ] where @xmath171 -\frac 2r\left ( p_{\perp } -p\right ) , \label{57}\ ] ] and @xmath172 ( as well as @xmath165 in the previous case ) may be easily interpreted as the total outward force acting on a given fluid element . after perturbation we obtain from ( [ 19 ] ) and ( [ 20 ] ) @xmath173 @xmath174 . \label{58}\ ] ] using this last expression together with ( [ 47]-[49 ] ) , we obtain for equation ( [ 16 ] ) after perturbation @xmath175 \times w_{,0 } , \label{59}\ ] ] or @xmath176 \frac{(1 - 2\widetilde{m}/r)}{(\rho + p)}\times \left [ 1-\alpha \right ] ^{-1}. \label{60}\ ] ] equations ( [ 55 ] ) and ( [ 60 ] ) may be compared with the newtonian form force @xmath177 mass @xmath178 acceleration , where here the term @xmath179 , \ ] ] stands for the _ effective inertial mass _ . this one vanishes for @xmath180 , implying the vanishing of @xmath172 , even though the time derivative of the radial velocity is different from zero . as it has been mentioned in the introduction , @xmath6 corresponds to the _ critical point . _ this one coincides with the given in @xcite if @xmath102 and @xmath103 are zero . note that the effective inertial mass decreases as @xmath8 grows . as @xmath181 approaches to unity the system seems to be more unstable , and for @xmath182 a vanishingly small radial force leads to non zero values of @xmath183 this fact contradicts the assumption that the hydrostatic equilibrium corresponds to a vanishing total radial force , and consequently the reliability of a perturbative approach is in question under such condition . this approach also predicts an anomalous behaviour beyond the critical point . if @xmath184 then an outward force ( @xmath185 ) implies an inward acceleration ( @xmath186 ) . thus , we may conclude that we can neglect cuadratic and higher terms in the perturbed variables only if @xmath8 is not close to , or beyond , the critical point . causality and stability conditions @xcite have been found using a perturbative method up to first order . therefore , it seems interesting to answer to the following question . are systems with @xmath187 , or @xmath188 always forbidden by causality conditions ? if the answer is no , then the reliability of causality conditions is uncertain for such systems . in the next section we shall try to answer this question . in order to contrast the @xmath189 and @xmath7 conditions with stability and causality conditions , it is convenient to write ( [ 61 ] ) , using the notation adopted in @xcite . thus , @xmath190 according to linear perturbation theory @xcite , causality and stability requires @xmath191 @xmath192 and @xmath193 where @xmath194 the adiabatic contribution to the speed of sound is denoted by @xmath195 , @xmath196 is the particle number density , and @xmath197 , @xmath198 , and @xmath199 denote specific heat at constant volume , isothermal compressibility , and thermal expansion coefficient respectively . as usually , they are defined by @xmath200 @xmath201 and @xmath202 as we already mentioned , if the two viscosity coefficients vanishes , we recover the result found in @xcite . in this case it can be shown that the critical point is very close to the point where ( [ 63]-[65 ] ) break down @xcite for small values of @xmath203 . let us now consider the case where there is only bulk viscosity ( @xmath204 ) . in this case @xmath205 @xcite , and the critical point is overtaken if @xmath206 is satisfied , whereas causality requires @xmath207 therefore , it appears that the critical point is forbidden by causality and stability requirements . in the pure shear viscosity case ( @xmath208 ) , the critical point is overtaken if @xmath209 whereas the most restrictive causality condition , is given ( in this case ) by ( [ 63 ] ) @xmath210 again , the critical point is beyond the point where causality is violated . let us now consider the general case ( @xmath211 ) . we may write ( [ 63 ] ) as @xmath212 we are going to find conditions for which the point where causality is violated is beyond the critical point . assuming that our system is at , or beyond , the critical point ( @xmath213 ) , then we should demand by virtue of ( [ 62 ] ) and ( [ 74 ] ) @xmath214 or , after some elementary algebra @xmath215 \left ( \frac 2{1+c_s^2}\right ) , \label{76}\ ] ] and combining it with ( [ 65 ] ) @xmath216 \left ( \frac 2{1+c_s^2}\right ) , \label{77}\ ] ] implying @xmath217 \left ( \frac 2{1+c_s^2}\right ) > 0 , \label{78}\ ] ] which is equivalent to @xmath218 . \label{79}\ ] ] thus , if @xmath219 , it is , in principle , possible to attain the critical point without violating causality conditions ( [ 63 ] ) and ( [ 65 ] ) . note that condition ( [ 64 ] ) has not been used in this calculus , so it should be demanded in addition to condition @xmath219 . therefore , from ( [ 64 ] ) , ( [ 66 ] ) , and ( [ 79 ] ) , the critical point can be overtaken without violating causality conditions if the following conditions are satisfied @xmath220 note that conditions ( [ 79b ] ) and ( [ 79c ] ) do not imply necessarily @xmath221 , but if causality conditions and @xmath7 are accomplished , then conditions ( [ 79b ] ) and ( [ 79c ] ) must be fulfilled . an example of this situation is an ultrarelativistic monoatomic ideal gas ( @xmath222 , @xmath223 , @xmath224 ) . for this fluid @xmath225 , @xmath226 , @xmath227 , @xmath228 , @xmath229 @xmath230 and @xmath231 then ( [ 79b ] ) and ( [ 79c ] ) are accomplished , but @xmath232 . on the other hand if @xmath233 , then the critical point is less restrictive than ( [ 65 ] ) , and causality and stability conditions can be used freely . thus , we face the following alternatives : 1 . in the case of pure bulk or shear viscosity ( without heat conduction ) , the critical point is well beyond the point where causality breaks down . therefore , the system should not reach the critical point in those cases , and linear approximation can be applied to find stability and causality conditions . 2 . in the non - viscous case , the critical point is very close to the point where causality is violated for small values of the sound speed . since the linear approximation is not reliable close to the critical point , then it might be possible for a given system to attain the critical point . 3 . in the general case it may happen than causality breaks down beyond the critical point . thus , it appear that there exist situations where a given physical system may attain the critical point and even go beyond it . it is worth noticing that condition @xmath7 can be accomplished in non very exotic systems . one of them is an interacting mixture of matter and neutrinos , which is a well - known scenario during the formation of a neutron star in a supernova explosion . in this case the heat conductivity coefficient is given by @xcite @xmath234 where @xmath3 is the mean collision time , @xmath235 , @xmath236 is the number in neutrino flavors and @xmath10 is the radiation constant . assuming that two viscosity coefficients vanishes , and @xmath237 then @xmath238 using usual units , the critical point is overtaken if @xmath239{\frac{6\rho c^3}{7n_\nu a}}\sim 4.29\times 10 ^ 8\rho ^{1/4 } , \label{79f}\ ] ] where we have adopted @xmath240 , @xmath241 , @xmath2 is in kelvin and @xmath4 is given in g @xmath242 the values of temperature , for which @xmath6 , are presented in figure 1 as a function of energy density . these ones are similar to the expected temperature that can be reached during hot collapse in a supernova explosion ( * ? ? ? * @xmath243 18.6 ) . we would like to conclude with the following comment : for degenerate matter , when thermal conductivity is dominated by electrons , thermal relaxation time may be of the order of milliseconds ( or even larger ) , due to larger mean free path of electrons @xcite , but this is of the same order of magnitude as the time scale of the quick phase preceding neutron star formation . therefore for this last scenario ( at least ) , the basic assumption of our approach is justified . this work has been partially supported by the spanish ministry of education under grant no . pb94 - 0718	in the context of the mller - israel - stewart second order phenomenological theory for dissipative fluids , we analyze the effects of thermal conduction and viscosity in a relativistic fluid , just after its departure from hydrostatic equilibrium , on a time scale of the order of relaxation times . stability and causality conditions are contrasted with conditions for which the effective inertial mass vanishes .
gravitational microlensing has been a powerful probe of massive dark objects in the galactic halo ( paczyski 1986 ; griest 1991 ; alcock et al . if the lens is a binary , a pair of caustic crossings may be observed . the caustics of the lens system are curves on which the magnification is formally infinite , and quite large in practice . in addition to caustic crossing events , binary lenses can give smaller perturbations to a standard microlensing lightcurve . such perturbations are in principle detectable ( mao & di stefano 1995 ; di stefano & perna 1997 ) . binary events yield more information about the source lens system than single lens events , and it is even possible to image the surface of the source star due to the large magnification . caustic crossing events have been seen towards the galactic bulge and the small and large magellanic clouds ( alcock et al . 1997 ; 2000a ; afonso et al . 1998 ; albrow et al . 1995 ; 1999 ; udalski et al . 2000 ; afonso et al . mao & paczyski ( 1991 ) studied caustic crossing events in classical microlensing . they came to the conclusion that approximately 7% of microlensing events towards the galactic bulge would exhibit caustic crossings , with the assumption of perfect time sampling . the pixel technique extends the feasibility of microlensing experiments by removing the requirement that individual stars be resolved ( crotts 1992 , baillon et al . 1993 ) . the possibility of detecting caustic crossings due to binary lenses in pixel events is intriguing , as it allows us to study stellar populations at great distances . several groups have successfully used the pixel , or difference image analysis technique to detect candidate microlensing events towards m31 ( crotts & tomaney 1996 ; ansari et al . the macho collaboration has also used this technique to identify candidates towards the lmc , in addition to the classical events ( alcock et al . this technique has a great future as more telescope resources are brought to bear , both on the ground and in space . space based surveys will allow the reach of microlensing surveys to extend at least as far as the virgo cluster , at a distance of about 16 mpc . in this paper , we extend the mao & paczyski ( 1991 ) analysis to pixel microlensing surveys of stellar systems . in pixel microlensing , detected single lens events tend to have high magnifications , so their rate is suppressed by a factor of @xmath0 , where @xmath1 is the magnification required for detection . however , caustic crossing events always exhibit large magnifications at the caustic crossings . this enhances the rate of caustic crossing events relative to single lens events in pixel microlensing . on the other hand , finite source effects will limit the maximum magnification during a caustic crossing , rendering some classically detectable caustic crossing events undetectable in pixel microlensing . in this paper we show that the larger region of high magnification is the stronger effect . as a result , caustic crossing events in pixel microlensing represent fractions of 10 - 15% of microlensing events , compared with the 2 - 3% fraction we find for observable caustic crossing events with resolved sources . finer time sampling would increase both fractions , and the latter would then agree with the mao & paczyski ( 1991 ) result . our primary goal with this work is to show that microlensing events exhibiting caustic crossings are detectable in pixel microlensing surveys . to this end , we estimate the rate of caustic crossing events given some basic assumptions . less important to our purposes are considerations of accurately determining the properties of the source lens system , which we leave to future work . we begin with a theoretical discussion of the two point mass gravitational lens , the appropriate model of a binary star system . this model has been studied extensively ( schneider & wei 1986 ; erdl & schneider 1993 ) . we quote some relevant results . we take the masses of the lenses to be @xmath2 and @xmath3 , with @xmath4 , and we define dimensionless masses @xmath5 such that @xmath6 . furthermore , we define @xmath7 , with @xmath8 without loss of generality . we take the standard notation of distances to the source @xmath9 , deflector @xmath10 , and deflector source distance @xmath11 . we define the einstein radius and the einstein angle for the system as ( einstein 1936 ) @xmath12 we adopt a complex parameterization ( witt 1990 ) of the lens system . we introduce complex angular coordinates on the plane of the sky , @xmath13 . given two lenses at angular positions @xmath14 and @xmath15 , a source at @xmath16 will have images at the solutions @xmath17 of the lens equation @xmath18 here @xmath19 is the deflection angle , and the bars indicate complex conjugation . we will call the plane described by @xmath16 the source plane , and the plane described by @xmath17 the image plane . the variables @xmath16 and @xmath17 are angles expressed in units of the einstein angle @xmath20 . a source at angular position @xmath16 is at a transverse distance @xmath21 from the line of sight through the origin of the coordinate system . likewise , a lens at angular position @xmath17 is at a distance @xmath22 . in this formalism , the jacobian of @xmath23 in eq . ( [ eq : zeta ] ) is @xmath24 the magnification of an image at @xmath17 is given by @xmath25 , with the sign of @xmath26 giving the parity of the image . the total magnification is the sum of the individual magnifications of the images . the points @xmath17 where the jacobian vanishes , @xmath27 , trace out the critical curves . the inverse images @xmath23 of the critical curves are the caustic curves . when the source is on a caustic , and so its images are on the critical curve , the magnification is formally infinite . let @xmath28 be the angular distance between the two lenses in units of the einstein angle @xmath20 . according to the value of @xmath28 , there are three regimes in a binary lens system : the so called close , intermediate , and wide binaries . a close binary exhibits three caustics , one with four cusps on the line of the lenses and two with three cusps off the line . the intermediate binary exhibits one six cusped caustic , and the wide binary exhibits two four cusped caustics on the lens line . the separation points @xmath29 and @xmath30 between close and intermediate and between intermediate and wide binaries are as follows ( erdl & schneider 1993 ; rhie & bennett 1999 ) , @xmath31 in the equal mass case , @xmath32 , we have @xmath33 and @xmath34 . as @xmath35 , the intermediate binary vanishes , and the transition between close and wide is at @xmath36 . the critical curves are simply parameterized by ( witt 1990 ) @xmath37 at each @xmath38 , there are four roots of the resulting quartic equation in @xmath17 . each root @xmath39 ( @xmath40 ) describes a portion of the critical curve . extending the parameter range to @xmath41 , these portions can be patched together so that they follow each other continuously for every disjoint piece of the caustic . in this way , we achieve one `` continuous '' parameterization of the caustic , @xmath42 . the continuous parameterization is explicitly given by @xmath43,\ ] ] with @xmath44{r\pm\sqrt{s^3+r^2}},\\ s & = & -\left(\frac{d^2-e^{i\phi}}{3}\right)^2,\\ r&=&\left(\frac{d^2-e^{i\phi}}{3}\right)^3+\frac{1}{2}d^2e^{2i\phi}(\mu^2 - 1).\end{aligned}\ ] ] here , @xmath45 gives the degree of asymmetry of the lens , @xmath46 is something like a dipole moment about the origin , and the other parameters have no obvious simple interpretation . the parameterization of the caustics , @xmath47 , follows trivially from that of the critical curves , @xmath42 , using the lens equation , eq . ( [ eq : zeta ] ) . lastly , we expand the parameterizations of the critical curves and caustics in the limits where @xmath48 . defining @xmath49 with @xmath50 , we find as @xmath51 : @xmath52 with @xmath53 errors and as @xmath54 : @xmath55 with @xmath56 errors . in computing the rate of single lens events , the einstein angle is used as a `` cross section '' for microlensing . any source which passes inside the einstein ring is said to be microlensed . griest ( 1991 ) has shown that the rate of single lens events depends linearly on the threshold minimum angular separation between source and lens , implying that the dependence is on the angular diameter rather than on the solid angle , which would imply a quadratic dependence on the impact parameter . in contrast , the optical depth is a measure of the probability that a given star is microlensed at a given instant . this is proportional to the area of the einstein ring . to compute the rate of caustic crossing events , we compute the `` angular diameter '' of the caustic structure in direct analogy to the single lens case . we will call it the `` angular width '' of the caustic . we will appropriately average this `` angular width '' over the distribution of binary systems . the optical depth to caustic crossing would similarly depend on the solid angle enclosed by the caustic , but the rate of caustic crossing events depends on the angular width , as we now show . in a caustic crossing event , the source crosses the caustic an even number of times , half entering and half exiting the interior of the caustic . consider the rate @xmath57 at which the sources enter the region limited by the caustic . the number of caustic crossings per unit time , counting each caustic peak in the lightcurve , is then @xmath58 . if the orientation of the caustic is random , as is the case for binary systems whose separation has no preferred direction in the sky , we find @xmath59 where @xmath60 is the angle - averaged number flux of the sources , and @xmath61 is the `` width of the caustic , '' defined in terms of the caustic length @xmath62 by @xmath63 we take @xmath61 and @xmath62 in units of @xmath20 , and @xmath60 in the corresponding units . ( [ eq : gamma ] ) and ( [ eq : w ] ) are proven as follows . the flux of sources whose proper motion is in a direction forming an angle @xmath64 with the binary separation is @xmath65 . consider an infinitesimal piece of the caustic @xmath66 , whose interior normal makes an angle @xmath67 with the fixed axis . the number of sources entering the caustic through @xmath66 per unit time is @xmath68 the limits of integration restrict the flow of sources to those coming from one side of the caustic only . integration of eq . ( [ eq : dgamma ] ) along the caustic gives eqs . ( [ eq : gamma ] ) and ( [ eq : w ] ) . eqs . ( [ eq : gamma ] ) and ( [ eq : w ] ) apply to any closed curve in the plane . for example , consider the rate of single lens events . the critical curve of a single lens is a circle , namely the einstein ring . its circumference is its length @xmath62 , and eq . ( [ eq : w ] ) implies that @xmath61 is the diameter of the circle . then eq . ( [ eq : gamma ] ) states the obvious fact that the number of sources entering the circle equals the flux times the circle diameter . this method of computing the cross section of a closed curve counts all entries to the curve . a curve that is not convex , such as a caustic curve , can have multiple entries . the rate @xmath57 counts all of these entries . for example , a binary lens event with 4 caustic crossings is counted twice in @xmath57 . in other words , the rate @xmath57 is half the number of caustic peaks per unit time . we now discuss the observed physical parameters of binary systems relevant to our calculation . it is well known that a large fraction of stars have companions , perhaps the majority . a well known rule of thumb is that the distribution of periods @xmath69 for binary systems is constant in @xmath70 , with roughly 10@xmath71 of binaries in each decade from one third day to ten million years . less well known is the correlation between the masses of the bound stars . at high masses , a correlation is seen , but at low masses , the data are inconclusive . we proceed to express the separation of the binary pair in terms of its period and total mass . with semimajor axis @xmath72 , kepler s law reads @xmath73 the einstein radius is @xmath74 with @xmath75 , and @xmath76 is the distance modulus to the source . assuming that the projected separation is at maximum , the most probable situation , we find the relation @xmath77 in the range @xmath78 $ ] , the distribution in @xmath69 is simply @xmath79 , and is correctly normalized . fixing all other quantities , appropriate for the rate calculation we will do , we thus find the distribution of binaries in @xmath28 , @xmath80 . there is disagreement about the distribution in @xmath81 of binary systems , ranging from a linear rise @xmath82 ( mazeh et al . 1992 ) to a power law decline @xmath83 ( patience et al . 1998 ) , with other measurements falling between ( duquennoy & mayor 1991 ; trimble 1990 ) . in the next section we will find that the effects of the @xmath81 distribution of binaries are mild . we evaluate the mean width of the caustics @xmath84 for the binary systems described in the previous section . in the parametric representation of the caustic @xmath47 , its length in units of the einstein angle is @xmath85 we have computed the length of the caustic structure for various values of @xmath81 . the caustic length is not as sensitive to the @xmath81 distribution as it is to the lens separation @xmath28 , as we show in figure 1 . there we plot the width of the caustic as a function of @xmath28 , averaging over four distributions in @xmath81 : a linear rise @xmath82 , flat , flat in @xmath86 , and a power law @xmath87 ( in the last two distributions we take @xmath88 ) . the difference is only a few per cent . the peaks occur at the average values of the close intermediate and intermediate wide binary separations @xmath29 and @xmath30 . for small and large separations @xmath28 , we have found the leading behavior of the mean caustic width . in both cases the important caustics are approximately square . for @xmath89 , the linear dimension of the two additional triangular caustics is suppressed relative to the single , central square caustic by a factor of @xmath28 . the mean width is @xmath90 in the case @xmath51 , and it is @xmath91 when @xmath54 . thus , we see that the caustic structure is largest when @xmath28 is of order unity . averaging over the four @xmath81 distributions mentioned above ( namely @xmath92 , flat , @xmath93 , and @xmath87 , in this order ) we find @xmath94 , @xmath95 , @xmath96 , and @xmath97 in the case @xmath51 , and @xmath98 , @xmath99 , @xmath100 , and @xmath101 in the case @xmath102 . now that we have computed the cross section for caustic crossing events , we can compare with that of single lens events on resolved stars , whose cross section in einstein units is 2 . the cross section averaged over angles is @xmath103 , which needs to be averaged over @xmath28 and @xmath81 . we find the ratio of caustic crossing to single lens events to be approximately 6.3% , multiplied by the fraction of star systems that are binaries . this is in quite good agreement with the results of mao and paczyski ( 1991 ) . in addition to the mean width of the caustic structure , important for determining the cross section for caustic crossing events , we would like to know the mean length of chords through the caustic structure . this quantity is directly related to the mean time between caustic crossings . we proceed as follows . let @xmath104 be the projected width of the caustic structure at a fixed orientation @xmath67 , and let @xmath105 be the chord length at orientation @xmath67 and impact parameter @xmath46 . the area of the caustic structure is clearly @xmath106 where now @xmath107 is the mean chord length at orientation @xmath67 . the area is of course independent of @xmath67 . we want the mean chord length , weighted by the width of the caustic . the mean chord length over all orientations is now just @xmath108 the area divided by the mean cross section . note that on the occasions where there are multiple pairs of caustic crossings , the chord is counted as multiple separate pieces . we will need the area and projected diameter of the caustic structure . in vector notation , we have @xmath109 , which is easily transformed to the complex parameterization , @xmath110 @xmath111 the projected diameter along the real axis , at orientation @xmath67 , is simply @xmath112 @xmath113 @xmath114 note that if this expression is averaged over orientations @xmath67 , we recover the result for the mean width of the caustic structure being proportional to the length of the caustic . we have now assembled all of the necessary pieces for calculating the mean length of chords through the caustic curves . as for the mean diameter , we compute this quantity as a function of the lens spacing @xmath28 , averaging over a distribution in @xmath81 , which we take to be constant in @xmath81 . the mean chord length is plotted in figure 1 . we are furthermore interested in the full distribution of chord lengths for a given lens configuration . as is done by han , park & lee ( 2000 ) , we have performed a monte carlo simulation to find this distribution . the results are shown in figure 2 , with the exact mean shown as a short dashed line and the median shown as a long dashed line . from this simulation , it seems that using the mean chord length may not be an adequate prescription for designing a microlensing survey , since the distribution of chord lengths is quite complicated . however , we will find that the mean chord length prescription is usually quite good . we illustrate two lightcurves for caustic crossing events in figure 3 . other examples can be found in the literature cited above . in the upper panels we show the caustic structure together with the source trajectory in the plane @xmath115 . we place the lenses along the @xmath116-axis at @xmath117 . in the lower panels we plot the lightcurves as functions of @xmath118 , which for a uniformly moving source is related linearly to time . in the interior neighborhood of a caustic , the magnification diverges as @xmath119 , where @xmath120 is the angular distance perpendicular to the caustic , in units of the einstein angle ( chang & refsdal 1979 , kayser & witt 1989 ) . the flux factor @xmath121 depends on the lens map through the length of the tangent vector to the caustic , @xmath122 , as @xmath123 . in the complex parameterization , the tangent vectors to the critical curve and caustic are ( witt 1990 ) @xmath124 defining the einstein time as usual @xmath125 , and taking the angle to the normal as @xmath126 , the lightcurve near the caustic crossing at @xmath127 is , in the interior side of the caustic , @xmath128 with the @xmath129 sign ( @xmath130 sign ) when entering ( exiting ) the caustic . in pixel lensing @xmath131 is unknown ( but see the next paragraph for a possibility to determine it ) . however , there are two caustic crossings , and if the timescales of the two are compared , the ratio of the geometric factors @xmath132 at each of the two caustic crossings can be measured . this may provide some insight into the exact parameters of individual events . we may gain useful information by measuring the duration of the caustic crossing . this is the time over which the disk of the source star intersects the caustic , and is given simply by @xmath133 for star star lensing in a distant galaxy , @xmath134 , @xmath135 is known , and @xmath136 and @xmath137 are taken from known distributions . this can give a handle on the angular size of the source star . furthermore , if the caustic crossing is seen in multiple wavebands , the color(s ) of the source star can be determined . this is possible because the magnification is the same in all wavebands , thus comparing the flux _ increase _ in two bands is the same as comparing the actual ( unmeasured ) source flux in those bands . with the color and the angular size , a rough estimate of the absolute flux @xmath131 can be derived . it may also be possible to image the surface of these extragalactic stars as the caustic passes over them , as has been done towards the smc ( afonso et al . 1998 ; albrow et al . however , this would require intense monitoring to cover the actual caustic crossing . it is not the purpose of this paper to discuss this possibility further . the magnification during a caustic crossing is critically important to a calculation of the event rate . while the magnification of a point - like source is infinite , that of a real source is finite . since only magnifications higher than a threshold magnification can be observed , it is crucial to determine the fraction @xmath138 of caustic crossing events that exceed a fixed magnification @xmath139 . the rate of caustic crossings entering the caustic with magnification @xmath140 is then @xmath141 we now determine @xmath138 . the maximum magnification of a finite source of angular radius @xmath142 is given by ( schneider & wei 1987 ) @xmath143 where @xmath144 is a form factor depending on the luminosity profile of the source . for a uniform disk , @xmath145 , while a limb darkened profile @xmath146 gives @xmath147 ( kayser & witt 1989 ) , thus we see that the profile is relatively unimportant for our purposes . @xmath148 in eq . ( [ eq : amax ] ) is the highest magnification when the source crosses the caustic . it follows that the fraction of events that cross a given caustic with magnification @xmath140 is equal to the fraction of length of the caustic in which @xmath149 , @xmath150 taking a distribution of mass ratios constant in @xmath81 , we plot @xmath151 as a function of @xmath28 for several values of @xmath152 in figure 4 . we now compute the rate distribution of caustic crossing events . we let the velocity distribution of the source and lens population be maxwellian with characteristic velocity @xmath153 . we denote the transverse velocity of the lens @xmath154 , and the transverse velocity of the line - of - sight @xmath155 . we denote the total mass of the lens @xmath156 , and the apparent magnitude of the source @xmath157 . we define the mass function @xmath158 normalized to unity , and the luminosity function @xmath159 normalized to the surface brightness . we denote the distribution in @xmath28 as @xmath160 . the mean width of the caustic structure in einstein units is @xmath84 , and we choose a distribution constant in @xmath81 . we will compute the rate of caustic crossing events with respect to the time between caustic crossings @xmath161 in two ways . first we will use the mean length of chords to determine this timescale , which is not correct event by event , but may be acceptable for a rate distribution . if we define the effective time between caustic crossings @xmath162 where @xmath163 is the length of the caustic chord averaged over the @xmath81 distribution , we can write the rate distribution as @xmath164 @xmath165 with @xmath166 given below , @xmath167 and @xmath168 . also , @xmath169 , where @xmath170 is a modified bessel function of the second kind . we note that @xmath171 and @xmath172 . secondly , we will use the actual caustic crossing timescale , at the cost of doing an exhaustive monte carlo of chord lengths . here we define @xmath161 as in eq . ( [ eq : tc ] ) , except that now @xmath173 is the actual chord length . the rate distribution is then @xmath174 @xmath175 with @xmath176 , @xmath177 being the effective width as a function of chord length , @xmath178 @xmath61 is the projected diameter given in eq . ( [ eq : diameter ] ) , and @xmath179 is the probability distribution of chord lengths at a given @xmath180 , computed by monte carlo and normalized to unity at @xmath181 . we illustrate the function @xmath177 for several values of @xmath152 in figure 5 . to fix the minimum flux factor @xmath152 , we adopt the convention that the signal to noise totaled throughout the approximate lightcurve eq . ( [ eq : lightcurve ] ) , be greater than @xmath182 . we assume that samples are taken daily , ( except for the reference case of the milky way bulge , monitored continuously ) . the lightcurve is truncated first by assuming the first sample is taken at one half day away from the caustic crossing , and then by requiring that the stellar diameter is less than the length of the chord . we derive the signal to noise as follows . assume that the telescope collects photons from a source of magnitude @xmath157 at a rate @xmath183 . an integration of @xmath184 per observation is taken . the background galaxy light falling on the pixel has a magnitude @xmath45 . this gives a signal to noise per sample of @xmath185 , where we have only considered photon noise . ground based microlensing surveys have achieved noise levels within a factor of two of the photon noise , and we expect space based surveys to approach the photon noise limit . thus there is a simple relation between the minimum allowed @xmath186 and the minimum flux factor @xmath152 . the minimum @xmath28 allowed is determined by fixing the period at its minimum , @xmath187 yr , giving @xmath188 . in practice , fixing @xmath189 is fine , since @xmath28 lower than this gives a very small cross section , and for the mass we are concerned with , it gives a minimum lens source distance of 10 pc , which is negligible . in principle a maximum @xmath28 can be determined by the binding of the binary , but it is never important observationally . the density @xmath190 is the effective density of source lens systems at separation @xmath191 , and can be determined by integrating the source and lens positions along the line of sight through the target system . assuming a radial density profile of stars @xmath192 truncated at a radius @xmath193 , we calculate the column density along a line of sight with impact parameter @xmath46 , @xmath194 the density along the line of sight is used as a probability distribution for sources . using @xmath195 , and assuming @xmath196 ( which furthermore implies that @xmath197 only depends on @xmath120 and @xmath135 and not on @xmath198 ) , we find @xmath199 @xmath200 finally , we comment on the rate of binary events with resolved sources . in this case there is no maximum @xmath191 , as any caustic crossing event can be detected . thus the integral over the luminosity function @xmath159 is trivial . the remaining triple integral is simply multiplied by the number of monitored stars . in any microlensing survey , the issue of backgrounds is crucial . in considering binary microlensing events , the lightcurves are basically characterized by a double spike . such events in isolation should be exceedingly rare . the most pernicious of backgrounds for single lens events is the mira variables ( crotts et al . 2000 ) , which have a very long period , but as they are characterized by a single spike , we are unconcerned with them . however , if one of the two caustic crossings is missed , the characterization of the event as microlensing would be very difficult , and only possible with a very high signal to noise and/or long follow - up period . in uncovering variability , we must be concerned with the amount of blending of the lensed star s light . we will now make a simple estimate of this quantity . we want to know first the typical magnitude of the lensed star . studying a typical @xmath201-band luminosity function , such as that of terndrup , frogel & whitford ( 1990 ) , we see that the slope at the highest fluxes is shallower than 0.4 , meaning that the most light comes from the brightest stars . this means that the majority of detectable microlensing events will be microlensing of very bright stars . in fact , typically the rate of such events is expected to peak somewhere between the surface brightness fluctuation magnitude ( which for the terndrup et al . ( 1990 ) luminosity function is roughly @xmath202 ) and the tip of the red giant branch ( at roughly@xmath203 ) . as a rough estimate for m87 , if we assume the pixel scale of the advanced camera for surveys ( acs ) on the hst is 0.05 , and a surface brightness of roughly 21 mag arcsec@xmath204 for most of the field of view , the pixel brightness is 27.5 mag pixel@xmath205 , while the surface brightness fluctuation magnitude is roughly 29.5 for m87 , meaning that each pixel has the light of roughly 6 sbf magnitude stars . moreover , we assume a psf where roughly 25% of the light of a star falls on the central pixel . this means that if we desire the pixel flux to double , the typical required magnification ( for @xmath206 ) is a factor of 12 . in an 1800 second integration with the acs at this surface brightness , roughly 350 photons are collected , so a @xmath207 excess requires only a 50% magnification of the pixel flux , or a microlensing magnification that is typically of order six . in figure 6 , we illustrate four representative lightcurves of a binary event in m87 observed by the hst acs , sampling daily . in all cases the geometry is identical to the left panel of figure 3 , but with four different timescales between caustic crossings of 3 , 5 , 10 , and 20 days . with a timescale of three days , though the event is detected at a significant signal to noise , it is not convincingly a binary . with a timescale of five days or more , it seems more likely that binary events can be unambiguous . finite source size effects are not severe for the example lightcurves in figure 6 . the star of magnitude @xmath206 has a radius of roughly 65 @xmath208 ( gould 1995 ) , and for a solar mass lens system with @xmath209 kpc , the star subtends roughly one seventh of the angular distance between caustics along its path . the dotted curves are the theoretical lightcurves , including finite source effects for a uniform surface brightness star , computed according to gould & gaucherel ( 1997 ) . according to eq . ( [ eq : amax ] ) , the limiting counts should be roughly 475 on the left and 600 on the right . as shown , the full calculation is not quite that severe . that the lightcurves should not be seriously affected is also clear from the fact that the angular size of the star is small relative to the angular distance between the caustics . for the two longer events , it becomes possible to see the actual caustic crossing happening , though with one image a day the accuracy is poor . in principle , one could envision an alert system so that more frequent images are recorded during a suspected caustic crossing . we consider the case of star star lensing in m87 , a giant elliptical in the virgo cluster at a distance of 15.8 mpc . we take the radial profile of stars to be @xmath210^{-\alpha},\ ] ] with @xmath211,\ ] ] and @xmath212 @xmath213 pc@xmath214 , following tsai ( 1993 ) . we adopt the @xmath201-band luminosity function of terndrup et al . ( 1990 ) adjusted to yield a surface brightness fluctuation magnitude of @xmath202 , and a mass function @xmath215 with @xmath156 in solar masses ( miller & scalo 1979 ) . the velocity dispersion is taken to be @xmath216 km s@xmath205 . typically , stellar radii are @xmath217 , while the typical einstein radii are of order @xmath218 for solar mass lenses , a factor of twenty larger . we compute the rate of caustic crossing events observed in surveys using the acs on the hst , and on a model ngst . we compare with the rate of single lens events ( baltz & silk 2000 ) , requiring seven samples detected at @xmath219 , as criteria a of alcock et al . ( 2000b ) . we assume that the sensitivity of the acs is 4.5 times that of the wfpc2 in the i band , meaning that one photon per second is collected from a star of magnitude @xmath220 , integrated over the entire psf . we model the ngst as having seven times the efficiency of wfpc2 , but with nine times as much collection area . these rates are plotted in the upper panels of figure 7 . in a monitoring program of daily observations of m87 for a period of a month , the advanced camera for surveys on hst would detect of order one caustic crossing event , and the ngst could detect of order one dozen caustic crossing events . microlensing in m31 , the andromeda galaxy , has been searched for extensively using the pixel technique ( ansari et al . 1997 , 1999 ; melchior et al . 1998 , 1999 ; crotts and tomaney 1996 ; tomaney and crotts 1996 ) . we have computed the rate of caustic crossing events observed in the bulge of m31 , assuming a cfht class telescope and using the bulge model of kent ( 1989 ) , with a velocity dispersion of @xmath221 km s@xmath205 . the typical star has a radius @xmath222 , and the typical einstein radius is @xmath223 , again a factor of 20 larger . the fiducial star has magnitude @xmath224 , giving one photon per second integrated over the psf . furthermore , we assume that the errors are twice the photon counting noise , as found by the agape collaboration ( ansari et al . we again compare with the rate of single lens events , and plot the results in figure 7 . a cfht survey of the bulge of m31 would detect on the order of one binary event per month of observations . finally we compare the rates of single lens and caustic crossing events towards the bulge of the milky way . we take the simple bulge and disk model of evans ( 1994 ) , with no bar . these rates are plotted in the upper panel of figure 7 . note that typically @xmath225 stars are monitored in the galactic bulge . we notice that our rates are comparable to the the results of the macho project monitoring of the galactic bulge ( alcock et al . 1997 ; 2000a ) . while we find that roughly @xmath226 of events should exhibit caustic crossings in this case , the finite stellar radii remove a significant fraction of these events as the stellar radius becomes comparable to the distance between the caustics , and we find that roughly 2 - 3@xmath71 of events will exhibit detectable caustic crossings . we find that caustic crossing events typically make up @xmath227 , and as much as @xmath228 , of the total rate of stellar events in pixel microlensing surveys , assuming that all stars are binaries . this is significantly higher than the relative rate in microlensing with resolved sources , typically 2 - 3@xmath71 . thus , the larger effective cross section is more important than the more severe finite source effects . these estimates all assume only self lensing in the target galaxy . of course there is the possibility that lenses in the halo of the milky way could contribute . unless there is a significant population of halo lenses , the rate for stellar lenses in the milky way would be substantially smaller than the self lensing rate in the target galaxy , since the scale height of the disk is much smaller than the size of the spheroids of the target galaxies . from figure 7 we see that the time between caustic crossings is typically of order several days . sampling daily would allow a sizeable fraction of the lightcurves to be well measured . we also see that for long timescale events , the timescale @xmath229 gives a very good approximation to the true rate ( using @xmath161 ) , but for the shorter events , it is less accurate . we notice that the agreement between the two is excellent for the ngst and for the m31 bulge . in these cases , the required magnification is quite modest , of order a factor of two for a star with the surface brightness fluctuation magnitude . the acs requires significantly higher magnifications , spoiling the excellent agreement between the rate with respect to the two timescales @xmath161 and @xmath229 . for the milky way bulge , the agreement between the two timescales is quite good , though the ( quite broad ) peaks disagree by a factor of about two in timescale . we thus find that the timescale @xmath229 provides a surprisingly good estimate of the rate of events with respect to the time between caustic crossings . finally , we comment that with daily sampling , the fine details of the caustic crossing will not be well covered . if we choose a finer sampling scheme , with the same telescope resources , the expected number of events will be much smaller . we feel that it is important first to identify binary lensing events . if this is shown to be possible , then more sophisticated sampling strategies should of course be considered . as an aside , we apply the previous analysis to a pixel microlensing search of extragalactic planets . discovering planets outside our own solar system has been an important scientific goal for many years . the planets that have been discovered so far were all found by detecting the wobble of the star that they orbit . the first detection was of a planet approximately the mass of earth , in orbit around a pulsar ( wolszczan & frail 1992 ; wolszczan 1994 ) . more recently , planets have been detected around main sequence stars ( see marcy , cochran & mayor 2000 for a review ) . we would like to extend the range of planetary searches to distant galaxies . gott ( 1981 ) was the first to point out that gravitational lensing might be used to detect low mass objects in distant galaxies . here we make another suggestion : pixel microlensing surveys may be able to detect jupiter mass planets as far away as the virgo cluster , by capitalizing on the unique nature of binary microlensing events . previous work has shown that planets might be detected in microlensing events in the bulge of the milky way galaxy , or in the small and large magellanic clouds , which are small galaxies in orbit around the milky way ( mao & paczyski 1991 , gould a. & loeb , a. 1992 , griest , k. & safizadeh , n. 1998 ) . stars in the bulge and in the magellanic clouds can be resolved easily , and surveys routinely monitor of order ten million stars for microlensing events . evidence for a planet orbiting a binary star system in the milky way bulge has recently been claimed in a joint publication of the mps and gman collaborations ( bennett et al . 1999 ) , although the data can also be interpreted as a rotating binary system without a planet ( albrow et al . 2000 ) . a solar - type planetary system can be described to first approximation as a binary object , consisting primarily ( in mass ) of the sun and the planet jupiter . in such systems @xmath81 is very small , for example @xmath230 for the sun and jupiter . in figure 8 , we show two example lightcurves of microlensing events , together with the trajectories of the source stars relative to the caustic curves . in both cases the star has a companion one one - thousandth as massive , like the sun jupiter system . we have calculated the rate of planetary events observing m31 with a telescope like the canada - france - hawaii telescope ( cfht ) on mauna kea . we assume that every star has a companion that is one one - thousandth as massive , just like the sun and jupiter . furthermore , we assume , as is true for known binary stars , that the distribution of orbital periods is such that ten percent of such systems lie in each decade of period , from a third of a day to ten million years . this gives a 10% probability that a star has a companion between one and five au , in rough agreement with observational findings ( marcy , cochran & mayor 2000 ) . with a long term monitoring program observing four times daily ( using the necessary global network of telescopes ) , and relaxing the threshold to be three sigma in the two images appearing inside the caustic , we expect about one caustic crossing planetary event every two years this neglects other types of microlensing anomalies , such as those discussed by covone , de ritis & marino ( 1999 ) . to increase the chances to detect planetary systems in distant galaxies , we require a space telescope such as the proposed next generation space telescope , to be launched around 2009 . this will be a large ( about 7 meters in diameter ) infrared telescope at a lagrange point of the earth sun system , and it will be more than ten times as sensitive as the hubble space telescope . we have calculated the rate of events we might detect with the ngst observing the giant elliptical galaxy m87 in the virgo cluster , at a distance of 15.8 mpc . with the same assumptions as the m31 calculation , we find that an ngst survey of three month s duration , taking four images each day , should be able to detect of order four planetary systems . we find that such a survey is most sensitive to events where the separation between caustic crossings is about five days . an alert system for microlensing events would allow more frequent measurements of the light curve during the caustic crossings , with the possibility of determining the orbital parameters of the planetary system . our results for the rate of caustic crossing planetary events indicate that the rates are probably too small for this technique to be a feasible method of detecting extragalactic planets . we have made overly generous assumptions , and still the detection probabilities are marginal . we have found that caustic crossing binary events should typically make up @xmath227 , and as much as @xmath228 , of the total rate of events in pixel microlensing surveys . this is significantly higher than the relative rate in microlensing with resolved stars , which is typically 2 - 3@xmath71 . the enhancement is due to the larger magnification of caustic crossing events with respect to single - lens events . the suppression in the count rate expected from finite - source effects is not large enough to win over the enhancement from larger regions of high magnification . in examples of pixel lensing surveys , we find that on the order of one binary event per month of observation should be observable in a cfht survey of the bulge of m31 . also of the order of one caustic crossing event per month of daily observations of m87 should be detectable with the advanced camera for surveys on hst . the latter rate would increase to of order one dozen caustic crossing events per month with the ngst . searches for extragalactic planets with the microlensing technique discussed in this paper do not seem particularly promising . we would like to thank the referee for constructive comments that have added much to the original manuscript , and for making us aware of a simplification of eq . ( [ eq : dciiw ] ) . e. b. thanks p. podsiadlowski for productive conversations concerning binary star systems . we thank j. silk for useful comments and hospitality at oxford university , where this research was begun . e. b. was supported in part by grants from nasa and doe .	pixel microlensing , i.e. gravitational microlensing of unresolved stars , can be used to explore distant stellar systems , and as a bonus may be able to detect extragalactic planets . in these studies , binary - lens events with multiple high - magnification peaks are crucial . considering only those events which exhibit caustic crossings , we estimate the fraction of binary events in several example pixel microlensing surveys and compare them to the fraction of binary events in a classical survey with resolved stars . we find a considerable enhancement of the relative rate of binary events in pixel microlensing surveys , relative to surveys with resolved sources . we calculate the rate distribution of binary events with respect to the time between caustic crossings . we consider possible surveys of m31 with ground - based telescopes and of m87 with hst and ngst . for the latter , a pixel microlensing survey taking one image a day may observe of order one dozen binary events per month . to 0ptcwru - p11 - 00
future fifth generation ( 5 g ) mobile networks are expected to provide an unprecedented capacity in supporting the rapid growth of mobile data traffic with very limited spectrum resources . new multiple access technique , i.e. , non - orthogonal multiple access ( noma ) , which allow multiple concurrent communications , has been recognized as one of most efficient solutions to fulfill these requirements @xcite . recently , several non - orthogonal code division multiple access ( cdma ) schemes , named sparsely spread code division multiple access ( scdma ) @xcite@xcite , low - density spreading @xcite@xcite , and sparse code multiple access @xcite , have been developed for multiple access channels . all of these techniques rely on sparse signature sequences and near - optimal joint multi - user belief prorogation ( bp ) detections on sparse graphs . we collectively call these techniques scdma . it has demonstrated many advantages with respect to the capacity load and detection complexity over the conventional dense cdma and orthogonal multiple access schemes . in the downlink of a general scdma system , a base station simultaneously communicates with multiple users . data streams for the multiple users are first spread ( encoded ) into vectors by multiplying their signature sequences , which are sparse and the elements are usually selected from a given alphabet set . multiple data streams after spreading are superimposed at the base station and broadcasted to the users over a common channel , i.e. , using the common resources such as time and frequency . a multi - user bp detection is performed at each user to recover the data streams . the performance of scdma detection is mainly determined by a signature matrix that consists of all the users signature sequences as its row vectors . generally , the signature matrix should have a good sparsity , i.e. , without short cycles in the formed factor graph , to achieve a good bp detection performance . theoretically , if its factor graph has no cycles , the bp detection converges to the maximum likelihood ( ml ) detection performance @xcite . moreover , the equivalent signal constellation after spreading and superposition should have large euclidean distance which ultimately determines the performance bound of ml detection . this motivates us to design the elements in the signature matrix in scdma . signature design has been investigated for dense spreading in conventional cdma @xcite , where an orthogonal or low - correlated sequence set are constructed to maximize an equivalent cdma channel capacity . the problem becomes more complex for sparse spreading in scdma since the design should be implemented under the sparsity constraint of the signature matrix . the problem becomes even more difficult when a two - dimensional modulation scheme is employed as in the scenarios of @xcite and @xcite . works @xcite and @xcite show that a user constellation rotation significantly affects detection performance of multi - user superposition codes . convolutional code is employed for each user in @xcite @xcite and the multiple access scheme is referred to as trellis code multiple access ( tcma ) , which can be regarded as a spatial case of the scenarios in @xcite and @xcite with unitary spreading length . work @xcite considers two - user tcma and designs the user constellation rotation by maximizing an equivalent channel capacity . work @xcite considers a general multi - user scdma with a non - trivial spreading length . for a given regular factor graph structure , @xcite shows that a latin - rectangular signature matrix significantly outperforms a randomly generated signature matrix due to a large minimum code distance property . however , many open research problems , including how to efficiently find an optimal signature matrix with the maximum minimum code distance for an scdma system , how to efficiently estimate the ml detection performance , and how to design signature matrix that works well under both ml and bp detections , are still yet to be resolved . in this paper , we consider a general scdma system with a two - dimensional quadrature amplitude modulation ( qam ) and give a theoretical framework for signature design . we give a formal definition of scdma code distance and a distance enumerator analysis to estimate the ml detection performance . for a given factor graph structure of an scdma code , we design the optimal signature matrix with the maximum minimum code distance . we construct two scdma code families whose factor graphs have very few short cycles . the constructed scdma codes outperform the existing codes in terms of both word error rate ( wer ) performance and detection complexity . our numerical results show that their bp detections exactly converge to their ml detection performances with few iterations . simulations for turbo - coded scdma systems with variety communication rates are given to verify the validity of our design in more practical applications . the remainder of the paper is organized as follows . section [ sec : model ] describes the scdma system model and introduces three detection algorithms . section [ sec : distance ] defines the scdma code distance and some properties on code distance are shown . section [ sec : design ] gives the optimal signature matrix design for scdma codes . section [ sec : const ] gives two constructions of code families with few short cycles in their factor graph and large minimum code distance . section [ sec : simulation ] gives simulations for our design in both uncoded and turbo - coded scdma systems . section [ sec : conclude ] concludes this paper . figure [ fig : scma ] shows a @xmath0-user downlink scdma transmitter model at the base station . there are @xmath0 data streams to be transmitted to @xmath0 mobile users . after a forward error correction ( fec ) encoding , each user s data stream is modulated and spread by multiplying its signature sequence . figure [ fig : scma ] illustrates the spread processing for an individual symbol of each user s data stream . here we consider qam with @xmath1 , where @xmath2 is the imaginary unit . the output after spreading is @xmath3 for @xmath4 , where @xmath5 with @xmath6 or @xmath7 , is called a signature sequence of user @xmath8 . here we considered unitary energy for each nonzero element of the signature sequence . it should be emphasized that the spreading vector is sparse , i.e. , the majority of elements might be 0 . number of nonzero elements in a spreading vector is called an effective spreading length . the @xmath0 users data streams after spreading are superimposed and transmitted over @xmath9 orthogonal channel resources , e.g. , ofdma tones or mimo spatial layers . the transmitted vector is represented as @xmath10 which is referred to as an scdma codeword . note that there is a total of @xmath11 number of scdma codewords corresponding to the @xmath11 different variations of @xmath12 , where @xmath13 is the transpose of a matrix . matrix @xmath14 $ ] is sparse and is referred to as a signature matrix . by multiplexing @xmath0 users over @xmath9 channel resources , the load of the scdma code is @xmath15 . since for @xmath16 , we can use orthogonal spreading sequences to achieve near single - user performance , in this paper , we mainly consider overloaded scdma with @xmath17 . each user receives a noise - corrupted codeword @xmath18 , where @xmath19 , @xmath20 is a channel gain , and @xmath21 is a complex gaussian noise vector with each element an i.i.d . mean-0 variance-@xmath22 complex gaussian variable , i.e. , @xmath23 . a joint @xmath0-user scdma detection is performed to recover the data streams . -user scdma transmitter model.,width=297 ] each scdma code can be represented by a sparse factor graph . figure [ fig : factor ] gives an example of factor graph representation for a @xmath24-user @xmath25-resource scdma code proposed in @xcite@xcite , where data nodes @xmath26 denote data symbols of @xmath0 users , and code nodes @xmath27 denote @xmath9 scdma coded symbols . there is an edge between @xmath28 and @xmath4 , denoted as @xmath29 , if @xmath30 . let @xmath31 be the set that includes all the edges in the factor graph . edge @xmath32 is labeled by signature element @xmath33 . each code node is a superposition of its neighboring data nodes , i.e. , @xmath34 . the degree of a node is the number of edges incident with the node . the graph is called code - node regular of degree @xmath35 if all the the code nodes have degree @xmath35 . figure [ fig : factor ] is code - node regular of degree @xmath36 . if a factor graph is cycle - free , we call it a tree graph , and we call the corresponding code a tree scdma code . -user @xmath37-resource scdma code.,width=1 ] in this section , after we briefly reviewing two detection algorithms of scdma , the ml and bp detections , we give an approximate bp detection whose detection complexity linearly increases with the user number . for all these three detections , we assume the receiver knows channel gain @xmath20 perfectly . based on @xmath38 , the ml detection is @xmath39 where the ml detection is reduced to the minimum distance detection due to the memoryless gaussian channel . if more than one codeword satisfies ( [ eq : ml ] ) , we randomly select one of them as our decision with equal probability . here we assume that all scdma codewords are transmitted with equal probability , and thus , ( [ eq : ml ] ) is equivalent to the maximum a posteriori probability detection . based on the estimated scdma codeword @xmath40 , we can uniquely determine the transmitted data stream . the complexity of ml detection is @xmath41 for scdma with qam . a suboptimal scheme with lower complexity is bp detection . this detection is performed on the factor graph . the whole detection is performed iteratively in a belief - propagation manner . in each iteration , each node of the factor graph performs a local processing and exchanges message with its neighboring nodes . * code node processing : * consider the processing at code node @xmath28 . let @xmath42 , a priori probability of @xmath43 , be the message output from data nodes @xmath44 to code node @xmath28 at the @xmath45-th iteration . let @xmath46 be the set obtained by delating @xmath8 from @xmath47 . based on this priori probability , and the channel observation @xmath48 , code node @xmath28 outputs a probability message of @xmath49 to data node @xmath50 , which is a probability of @xmath43 . * data node processing : * data node @xmath4 combines the message obtained from the code nodes in its neighborhood and outputs a probability message of @xmath51 to its neighboring code node @xmath52 . * hard decision : * after a fixed number @xmath53 of iterations , hard decision is made for @xmath4 as @xmath54 the complexity of bp detection is dominated by the complexity of code node processing , whose complexity is @xmath55 , where @xmath35 is the maximum code node degree . the specific detection complexity is determined by the code node degree profile and the iteration number that is required for the detection . note that if the factor graph of an scdma code is a tree graph , its bp detection will converge to its ml detection with finite number of iterations . many works show that short cycles , such as length-4 cycles , significantly degrade the performance of bp detection @xcite . therefore , in this paper we only consider factor graphs without length-4 cycles . in the bp detection , the processing at the code node is a map processing , where the accurate probability about the estimated data is calculated . in this section , we give an approximate bp detection with a simplified code node processing . we regard the summation of interferences for each data as complex gaussian , so we only need to track a mean and variance message . take the processing at code node @xmath28 as an example . the associated receive at this node is @xmath56 where @xmath57 is the equivalent noise for @xmath4 . we approximately regards @xmath58 as complex gaussian , i.e. , @xmath59 with @xmath60=\sum_{j\in\kappa_{\backslash k}}s_{n , j}\textrm{e}[x_{j}]=\sum_{j\in\kappa_{\backslash k}}s_{n , j}\sum _ { \alpha\in\mathcal{x}}\textrm{p}_{x_{j},c_n}^{\ell-1}(x_{j}=\alpha)\alpha\nonumber\\ n_k\!\!\!\!\!\!\!\!\!\!\!&&\!=\!\textrm{e}[\|\xi_k\!-\!\mu_k\|^2]\!=\!\!\sum_{j\in\kappa_{\backslash k}}\!\textrm{e}[\|x_{j}\!-\!\textrm{e}[x_j]\|^2]\!+\!n_0\!=\!\sum_{j\in\kappa_{\backslash k}}\left(1\!-\!\left\|\textrm{e}[x_{j}]\right\|^2\right)\!+\!n_0\nonumber\\ \!\!\!\!\!\!\!\!\!\!\!&&=\sum_{j\in\kappa_{\backslash k}}\left(1-\left\|\sum _ { \alpha\in\mathcal{x}}\textrm{p}_{x_{j},c_n}^{\ell-1}(x_{j}=\alpha)\alpha\right\|^2\right)+n_0\nonumber\end{aligned}\ ] ] where @xmath61 $ ] takes the expectation of a random variable . therefore , code node @xmath28 outputs a probability message of @xmath62 to data node @xmath50 . the approximate bp detection may work well when the code node degree is large or the noise level is high since at these two cases , interference term @xmath58 is more like gaussian . the processing complexity of the code node reduces to @xmath63 , where @xmath35 is the maximum code node degree . in this section , we first define an scdma code distance and distance enumerator function , which is used to formulate a union bound for ml detection . some properties about scdma code distance enumerator function and the minimum code distance are derived . [ def : d ] distance between two scdma codewords @xmath64 is @xmath65 where @xmath66 is the scdma code set . @xmath67 is called the minimum distance of scdma code @xmath66 . applying ( [ eq : codeword ] ) to ( [ eq : distance ] ) , we obtain the following lemma immediately . [ lem : distance ] let @xmath68 and @xmath69 be the universal set of length @xmath0 vectors over @xmath70 . the minimum distance of the scdma code with spreading signature matrix @xmath71 is @xmath72 where @xmath73 and @xmath74 . to give a global description of the code distance spectrum of an scdma code , we have the following definition . [ eq : ad ] distance enumerator function for an scdma code with signature matrix @xmath71 is @xmath75 where @xmath76 is a dummy variable , and @xmath77 for qam is the cardinality of the code set . equation ( [ eq : enumerator ] ) in fact gives an average distance spectrum for all the codewords in the code set . the distance enumerator function of scdma code can be used to calculate a multi - user union bound developed in @xcite . it is an wer upper bound for ml detection . let @xmath78 be the distance enumerator function of an scdma code with signature matrix @xmath71 , where @xmath79 can be regarded as the average number of codeword pairs with distance @xmath35 . the wer @xmath80 under ml detection is upper bounded by + _ union bound @xcite : _ @xmath81 note that ( [ eq : union ] ) has different form as that in @xcite since the definition of code distance in this work has different form from that in @xcite . we give the following properties for the scdma code distance enumerator function and minimum code distance . [ lem : rot1 ] @xmath82 holds for @xmath83 . _ proof : _ equation ( [ eq : rot1 ] ) holds because @xmath84 holds for any @xmath85 . [ lem : rot2 ] @xmath86 holds for any @xmath87 , where @xmath76 is the set of integer numbers . _ proof : _ equation ( [ eq : rot2 ] ) holds because @xmath88 where @xmath89 , holds for any @xmath85 . [ lem : add ] for a give signature matrix @xmath71 , it holds that @xmath90 where @xmath91 and @xmath92 are signature matrices obtained by adding a row ( resource ) and column ( user ) to @xmath71 , respectively . similarly , we can obtain an opposite proposition of lemma [ lem : add ] by deleting a user or resource . [ col : bound ] for a give signature matrix @xmath71 , @xmath93 where @xmath94 is the minimum effective spreading length . _ proof : _ equation ( [ eq : bound ] ) is from the fact that @xmath95 is the minimum distance that is achieved by the matrix obtained by delating all the columns of @xmath71 except the one with the minimum effective spreading length . @xmath96 where @xmath97^t$ ] is a concatenation of @xmath98 . for a given factor graph structure , we design the optimal signature matrix with the maximum minimum scdma code distance . for factor graph @xmath99 , we have infinite number of scdma codes by varying its edge labels , i.e. , the phases of nonzero elements of the signature matrix . let @xmath100 be the universal set that includes all the possible signature matrices , edge labels , associated with @xmath99 . we aim to find the scdma code with the maximum minimum code distance . since a disconnected factor graph can be considered as multiple independent scdma codes , in the following we only consider connected factor graphs . @xmath101 is called an optimal signature matrix of @xmath99 . for a factor graph , there are infinite number of optimal signature matrices due to lemmas [ lem : rot1 ] and [ lem : rot2 ] . for a given factor graph @xmath99 , finding @xmath102 in @xmath100 is a non - convex problem with high complexity . however , if @xmath99 is cycle - free , the problem can be simplified . we first give the following theorem . [ thm : tree ] if @xmath99 is a cycle - free factor graph , for each signature matrix @xmath103 , there exists a matrix @xmath104 with @xmath105 . _ proof : _ we first prove that for each @xmath106 there exists @xmath107 with @xmath108 . using the row rotation invariance property of lemma [ lem : rot1 ] , we just need to show that for a given @xmath106 , there exists an @xmath109 which is a row rotation of @xmath71 . assume that @xmath71 is given . we determine @xmath110 as follows . since zero elements in @xmath110 are predetermined by the factor graph @xmath99 , we only determine the nonzero elements in @xmath110 in the following steps : \i . for each @xmath111 , the @xmath112-th row of @xmath110 is a rotation of the @xmath112-th row of @xmath71 , i.e. , @xmath113 for @xmath32 . find a column of @xmath110 that has only one determined nonzero element and at least one undetermined element . assume that the @xmath114-th column is found and the only determined nonzero element is @xmath115 . for each @xmath116 , the @xmath112-th row of @xmath110 is a rotation of the @xmath112-th row of @xmath71 , i.e. , @xmath117 for @xmath32 . \iii . if all the elements of @xmath110 are determined , terminate the procedure , otherwise , repeat step ii . + we show that step ii can always be successful if there exists undetermined elements in @xmath110 . first , if there exists undetermined element in @xmath110 , we can always find a column with both determined and undetermined nonzero elements since @xmath99 is connected and the nonzero elements are determined in a row - by - row manner according to the procedure . moreover , if a column has both determined and undetermined nonzero elements , the number of determined nonzero element must be one . suppose that there are more than one determined nonzero elements in column @xmath114 in step ii , i.e. , elements @xmath118 and @xmath119 , are determined . since the labeling begins from the edges incident with data node @xmath120 ( in step i ) , there should exist two paths from data node @xmath120 to both code nodes @xmath121 and @xmath122 , i.e. , there exists a path between @xmath121 and @xmath122 that goes through @xmath120 . since there exists another path of @xmath123 between @xmath121 and @xmath122 , which results a cycle in @xmath99 . this conflicts with the fact that @xmath99 is cycle - free . therefore , step ii can always be successful . it also holds that @xmath109 , since step i guarantees that the nonzero elements in the first column are 1 , and step ii guarantees that nonzero elements in each column are the same . note that each column except the first with more than one nonzero elements will be found in step ii , otherwise , we can show that a cycle exists in the graph similarly . since @xmath110 is a row rotation of @xmath71 , @xmath108 . direct applying the column rotation invariance property of lemma [ lem : rot2 ] , we can get a matrix @xmath124 through column rotations from @xmath110 with @xmath125 . thus , the theorem is proved . [ cor : tree ] if @xmath99 is a cycle - free factor graph , there exists an optimal signature matrix with @xmath126 . we consider a special tree factor graph with @xmath0 data nodes and one code node with load @xmath0 in which case the signature matrix become a vector . this is the scenario considered in tcma @xcite@xcite . we first consider the simplest case of @xmath127 . . one is for the optimal signature matrix @xmath128 $ ] and the other is for signature matrix @xmath129 $ ] used in @xcite@xcite.,width=3 ] .optimal signature matrix and minimum distance for @xmath130 . [ cols="^,^,^",options="header " , ] [ thm : two - user ] for the factor graph of an scdma system with @xmath127 users sharing @xmath131 resource , @xmath132 $ ] is an optimal signature matrix with @xmath133 . _ proof : _ see appendix [ app : two - user ] . using ( [ eq : enumerator ] ) , the distance enumerator function of the two - user scdma code with the optimal signature matrix is calculated as : @xmath134 , which could be used to estimate the wer performance based on the union bound ( [ eq : union ] ) . signature matrix @xmath129 $ ] is used in @xcite@xcite for two - user tcma system . it is suboptimal since it gives a smaller minimum distance of @xmath135 . we illustrated the superimposed constellation for both cases in fig . [ fig : constellation ] . it is difficult to formulate the optimal signature matrix for @xmath136 . we obtain the optimal signature matrix and minimum distance , denoted as @xmath137 , for @xmath138 in table [ tab : vector ] by numerical search . the minimum distance decreases as the number of users @xmath0 increases . in general , a signature matrix of a multi - resource scdma system with each row an optimal signature vector for a single - resource scdma system may not be optimal . we only have the following lower bound for the minimum code distance of such signature matrix . let @xmath139 be an index subset . subgraph @xmath140 is obtained by delating all the data nodes that are adjacent to the code nodes with index in @xmath141 and edges induced from these data nodes . [ lem : bound ] let factor graph @xmath99 be code - node regular degree of @xmath142 . let @xmath103 be a signature matrix with each row an optimal signature vector of a single - resource scdma system . it holds that @xmath143 where @xmath144 and @xmath145 are the numbers of code nodes in @xmath140 with degree one and larger than one , respectively . _ proof : _ see appendix [ app : bound ] . note that lemma [ lem : bound ] can be extended to the case of irregular code - node degree factor graph with a more complex expression . [ eg : tree ] we have the following @xmath0-user @xmath146-resource tree scdma code with each row an optimal two - user single - resource signature vector , @xmath147 this scdma code has a load of @xmath148 . using lemma [ lem : bound ] we have @xmath149 . also based on the proof of theorem [ thm : two - user ] in appendix [ app : two - user ] , we know that @xmath150 exactly holds . for @xmath151 , the minimum distance is @xmath152 which achieves the upper bound of corollary [ col : bound ] . in fact , we can further show that @xmath153 is also an optimal signature labeling for @xmath154 . for a factor graph with cycles , we can find an edge set so that after delating the edges in the set , the remaining graph is a tree . we simplify its labeling based on the following theorem . [ thm : nontree ] let @xmath99 be a factor graph with cycles and @xmath155 be an edge subset that after delating the edges @xmath156 , the remaining graph is a tree . for each signature matrix @xmath103 , there exists a matrix @xmath157 with @xmath105 . _ proof : _ see appendix [ app : nontree ] . [ cor : nontree ] for factor graph @xmath99 with cycles and @xmath156 defined in theorem [ thm : nontree ] , there exists an optimal signature matrix with @xmath158 . [ eg : optmat ] based on corollary [ cor : nontree ] , we can simplify the labeling for the @xmath159-user and @xmath37-resource scdma factor graph in fig . [ fig : factor ] proposed in @xcite@xcite as @xmath160 where @xmath161 and @xmath162 since after delating the edges in set @xmath163 , the factor graph will be a tree . its load is @xmath164 . we can find 4 length-6 cycles in its factor graph . through a full search based on ( [ eq : optmat ] ) , we obtain the following optimal signature matrix @xmath165 which has the minimum distance @xmath166 . note that the latin - rectangular labeling proposed in @xcite@xcite , which uses the elements @xmath167 with permutation for each row , only gives the minimum distance of @xmath168 . in this section , we give two constructions of code families whose factor graph has very few short cycles . we give examples to obtain the optimal signature labeling for these two constructions . [ const : improve ] @xmath169 where @xmath170 , \theta^1_j,\cdots , \theta^{k-1}_j\in [ 0 , \pi/2)$ ] for @xmath171 , and @xmath172 , for @xmath173 , @xmath174 , @xmath175 is a @xmath176 identity matrix , and @xmath177 is the following @xmath176 permutation matrix @xmath178 this scdma code family has a load of @xmath148 . vectors @xmath179 , should be optimized to achieve the maximum minimum distance . it is easy to see that there is only one length-@xmath180 cycle in its corresponding factor graph . since if we delete one edge in @xmath181 the graph will be cycle - free , we have used the simplified labeling for the remaining tree graph based on corollary [ cor : nontree ] . [ eg : plus46 ] consider @xmath182 in construction [ const : improve ] . the graph will be cycle - free by deleting @xmath183 . through a full search , we obtain the following optimal @xmath159-user , @xmath37-resource scdma code : @xmath184 with the minimum distance @xmath185 . if we delete the edges corresponding to @xmath186 in construction [ const : improve ] , the graph will become a tree , and the maximum minimum code distance will reduce to @xmath187 , which is achieved by allocating single - resource optimal signature vector to each row as in example [ eg : tree ] . introducing the part of @xmath186 in construction [ const : improve ] increases the minimum code distance for @xmath154 , and thus , improves the performance of ml detection . [ eg : plus68 ] similarly , by considering @xmath188 in construction [ const : improve ] , we obtain the following optimal @xmath189-user , @xmath159-resource scdma code : @xmath190 its load is @xmath191 and the minimum distance is @xmath192 , which achieves the upper bound of corollary [ col : bound ] . the following construction gives a higher load scdma code family . [ const : nontree ] @xmath193 where @xmath170 , \theta^k_j\in [ 0 , \pi/2 ) , k=1, ... ,k-1 $ ] , @xmath194 , \tau^k_j\in [ 0 , 2\pi ) , k=1,2 ... ,k-3 , j=1, ... ,q$ ] , and @xmath195 , are @xmath176 permutation matrices . this scdma code family has a load of @xmath196 . permutation matrices @xmath197 should be carefully selected to avoid short cycles , and vectors @xmath198 should be optimized to achieve the maximum minimum distance . since if we delete edges corresponding to @xmath199 , in construction [ const : nontree ] the graph will be cycle - free , we have used the simplified labeling for the remaining tree graph . [ eg:84 ] consider @xmath188 in construction [ const : nontree ] . by selecting @xmath200 and @xmath201 defined in ( [ eq : p ] ) , the generated factor graph has only one length-@xmath189 cycle . since the graph will be cycle - free by deleting the edge in @xmath202 , using theorem [ thm : nontree ] , we obtain the following optimal signature matrix @xmath203 where @xmath204 @xmath205 and @xmath206 . its load is @xmath207 and the minimum code distance is @xmath208 . if we allocate the single - resource optimal signature vector for each row of @xmath209 in construction [ const : nontree ] , using lemma [ lem : bound ] , we can show that the minimum distance @xmath210 . for @xmath211 , the minimum distance is @xmath152 , which achieves the upper bound of corollary [ col : bound ] . and optimal codes constructed in example [ eg : tree ] with @xmath212 , under ml detection and their union bounds . the wer of two - user suboptimal signature @xmath213 $ ] used in @xcite@xcite and its union bound are also illustrated.,width=3 ] -user @xmath37-resource scdma codes ( the optimal codes obtained in examples [ eg : optmat ] , [ eg : plus46 ] , and the code with latin - rectangular labeling proposed in @xcite@xcite ) under bp detections with @xmath214 iterations and their union bounds.,width=3 ] in this section we give error performance simulations for the ( uncoded ) scdma codes designed in sections [ sec : design ] and [ sec : const ] and turbo - coded scdma over awgn channel . figure [ fig : singlesource ] illustrates the wer curves ( solid lines ) and the union bounds ( dashed lines ) for optimal tree scdma codes , i.e. , two and three - user single - resource ( @xmath131 ) optimal codes obtained in table [ tab : vector ] and optimal codes constructed in example [ eg : tree ] with @xmath212 , under ml detections . for these codes , the ml and bp detections have exactly the same performance . it shows that the code with higher load has higher wer because higher load results in smaller minimum code distances and more error events ( larger distance enumerator coefficients ) . the two - user optimal signature @xmath215 $ ] has an asymptotic performance gain of near 2 db over the suboptimal signature of @xmath216 $ ] used in @xcite@xcite . all simulations coincide well with their union bound ( [ eq : union ] ) in most area except a little mismatch at the low wer regime . therefore , the union bound gives a good estimation for the wer of ml detection . figure [ fig : foursix ] illustrates the wer of @xmath159-user @xmath37-resource scdma codes : the optimal codes obtained in examples [ eg : optmat ] , [ eg : plus46 ] , and the code with latin - rectangular labeling proposed in @xcite@xcite under bp detections and their union bounds . the code obtained in example [ eg : optmat ] with the optimal signature has the best union bound since it has the maximum minimum code distance . the union bound is near 1 db better than that of the code with latin - rectangular labeling proposed in @xcite@xcite . for the wer simulation under bp detection with @xmath217 and @xmath189 iterations , the optimal signature has asymptotic performance gains of about @xmath218 db and @xmath219 db over the latin - rectangular labeling . however , both of their wer under bp detection with @xmath189 iterations is more than 1 db worse than their union bounds due to too many short cycles in their factor graph . although the code designed in example [ eg : plus46 ] of construction [ const : improve ] has a slightly worse union bound than that of the code obtained in example [ eg : optmat ] , its bp detection with 6 iterations converges to its union bound , which means that its bp detection may converge to its ml detection . it has a performance gain of more than 1 db under bp detection over the optimal code obtained in example [ eg : optmat ] with even fewer iterations . figure [ fig : eightuser ] illustrates the wer of the optimal 8-user 6-resource and 8-user 4-resource scdma codes designed in examples [ eg : plus68 ] and [ eg:84 ] under bp detections with @xmath220 iterations and their union bounds . with 6 iterations , their bp detections converge to or even exceed their union bounds which means that their bp detections may converge to their ml detections . let s consider their bp detection complexities . as mentioned in section [ sec : bp ] , the complexity is manly determined by the code node degrees and the iteration numbers . the factor graph of the code in example [ eg : optmat ] has 4 degree-3 code nodes , and its bp detection requires 8 iterations to converge ( according to our simulation observations ) . the factor graphs of the codes designed in examples [ eg : plus46][eg:84 ] have 2 , 2 , and 4 degree-3 code nodes , respectively , and their bp detections only require 6 iterations to converge . let @xmath221 denote the detection complexity of the code designed in example @xmath222 . based on a full consideration of their code node degree profile and iteration number , we can rank their complexities as @xmath223 . -user @xmath159-resource and @xmath189-user @xmath37-resource scdma codes obtained in examples [ eg : plus68 ] , [ eg:84 ] under bp detections with @xmath220 iterations and their union bounds.,width=3 ] and @xmath224 ) @xmath159-user @xmath37-resource scdma codes ( the optimal code obtained in example [ eg : optmat ] and the code with latin - rectangular labeling proposed in @xcite@xcite ) under bp and approximate bp ( abp ) detections with @xmath225 iterations.,width=3 ] in this section , we give bit - error - rate ( ber ) simulations of turbo - coded scdma systems with qam modulation , where the fec code in fig . [ fig : scma ] is realized by a turbo code . here the turbo code we considered consists two 8-state parallel concatenated convolutional codes with generator matrix @xmath226 $ ] , which is used in 3gpp let networks . by puncturing its parity bits , we can obtain different turbo encoding rates : @xmath227 . for all the simulations , the data stream length for turbo encoding of each user is @xmath228 . for both bp and approximate bp detections , the global decoding iteration ( each global iteration includes a turbo decoding iteration and an scdma iteration ) number is 30 , which is enough for all the considered decodings converge to their best performances . moreover , codes given by example [ eg : tree ] and constructions [ const : improve ] , [ const : nontree ] have irregular effective spreading profile , i.e. , effective spreading lengths for symbols of different users may be different . to realize user fairness , we alternately use column permutations of a signature matrix so that each user s symbol is spread with equal effective spreading length in average sense . for example , the signature matrix given in example [ eg : plus46 ] has effective spreading length profile @xmath229 for the six users . in our simulations , we divide the modulated symbol stream within a turbo codeword of each user into three sub - streams with equal length . the first sub - streams of the six users are spread based on signature matrix @xmath230 in example [ eg : plus46 ] . for the second and third sub - streams we use permuted matrices @xmath231 and @xmath232 , respectively , where @xmath233 is a @xmath234 column permutation matrix that swaps columns @xmath2 and @xmath222 . the permuted signature matrices have the same distance property with the original matrix but have the effective spreading length profiles @xmath235 and @xmath236 , respectively . by doing this , the average effective spreading length for each symbol , which is the same for each user , becomes @xmath237 . therefore , the detection error rate of each user will also be the same . figure [ fig : bercoded1 ] illustrates ber of rate-@xmath238 and @xmath224 turbo - coded @xmath159-user @xmath37-resource scdma systems under bp and approximate bp ( abp ) detections , where the optimal scdma code obtained in example [ eg : optmat ] and the code with latin - rectangular labeling proposed in @xcite@xcite are considered . the sum communication rates of these two turbo - coded scdma systems are @xmath239 bit / resource and @xmath240 bit / resource . the rate-@xmath238 turbo - coded scdma system with the optimal scdma code designed in example [ eg : optmat ] has a performance gain of about @xmath241 db over the same rate scdma system with latin - rectangular labeling under both bp and abp decodings . this gain increases if we considered higher rate turbo code , which works at higher @xmath242 regime , i.e. , the gain increases to @xmath243 db for the rate-@xmath224 turbo - coded scdma system . comparing with bp decoding , the performance loss of the abp is about @xmath241 db for rate-@xmath238 turbo - coded scdma system at low @xmath242 regime since the interference term in ( [ eq : interfer ] ) is very similar to gaussian . this performance loss increases to @xmath244 db at the high @xmath242 regime for the rate-@xmath224 turbo coded scdma system . figure [ fig : bercoded2 ] illustrates ber of rate-@xmath245 and @xmath246 turbo - coded scdma systems under bp detection , where the optimal codes obtained in examples [ eg : optmat ] , [ eg : plus46 ] and the code with latin - rectangular labeling are considered . the rate-@xmath245 turbo - coded scdma system with the optimal scdma codes designed in examples [ eg : plus46 ] has slightly better ber than the code designed in examples [ eg : optmat ] and has a performance gain of about @xmath219 db over the code with the latin - rectangular labeling . they have the sum communication rate of @xmath207 bit / resource . for the even higher encoding rate , i.e. , a rate-@xmath246 turbo - coded scdma system that works at higher @xmath242 regime , this gain increases and the code in example [ eg : plus46 ] has larger performance gains of about @xmath247 db and @xmath218 db over the code in example [ eg : optmat ] and the code with the latin - rectangular labeling . in this case , the sum communication rate reaches @xmath248 bit / resource . figure [ fig : bercoded3 ] compares four pairs of turbo - coded scdma systems : \a ) rate-@xmath245 turbo - coded scdma systems with @xmath249-user 2-resource optimal tree scdma code constructed in example [ eg : tree ] and @xmath24-user 4-resource scdma code designed in example [ eg : plus46 ] . their communication rate is @xmath207 bit / resource . \b ) rate-@xmath246 turbo - coded scdma systems with @xmath250-user 3-resource optimal tree scdma code constructed in example [ eg : tree ] and @xmath251-user 6-resource scdma code with the optimal scdma code designed in example [ eg : plus68 ] . their communication rate is @xmath252 bit / resource . \c ) rate-@xmath246 turbo - coded scdma systems with @xmath249-user 2-resource optimal tree scdma code constructed in example [ eg : tree ] and @xmath24-user 4-resource scdma code designed in example [ eg : plus46 ] . their communication rate is @xmath248 bit / resource . \d ) rate-@xmath246 turbo - coded scdma systems with two - user single - resource optimal scdma code obtained in theorem [ thm : two - user ] and @xmath251-user 4-resource scdma code designed in example [ eg:84 ] . their communication rate is @xmath253 bit / resource . each pair has the same communication rate but the code with more users has steeper ber cure , better asymptotic ber performance , due to the joint multi - user processing gain . the rate-@xmath246 turbo - coded two - user single - resource optimal scdma code still has a 1 db performance gain over the same rate turbo - coded suboptimal scdma code used in @xcite@xcite . and @xmath246 ) scdma codes ( the optimal codes obtained in examples [ eg : optmat ] , [ eg : plus46 ] and the code with latin - rectangular labeling ) under bp detection with @xmath225 iterations.,width=3 ] and @xmath246 ) scdma codes ( the optimal codes obtained in examples [ eg : tree ] , [ eg : plus46]-[eg:84 ] , and two - user single - resource scdma with optimal [ theorem [ thm : two - user ] ] and suboptimal @xcite@xcite labeling ) under bp detection with @xmath225 iterations.,width=3 ] we gave a code distance analysis and signature optimization for overloaded scdma systems . good scdma codes that work well under both bp and ml detections with low detection complexities are constructed . the constructed codes can support very diverse high - rate services . as an initial work , we only analyzed the code distance of uncoded scdma systems , i.e. , without fec code , and scdma with qam and equal power for each user . one possible extension is to do distance analysis for coded scdma systems , which leads to a joint fec and scdma code design . the new system can be treated as a concatenated code . some works related to concatenated code are given in @xcite . another possible extension is to consider a more general modulation and unequal - power user transmissions . although we focused on scdma systems , our design also apply to several similar well - documented system proposals , such as tcma @xcite and superposition modulation @xcite . _ proof : _ we first prove that there exists an optimal signature matrix in @xmath254 , \theta\in[0 , \pi/4)$ ] . according to corollary [ cor : tree ] , optimal signature matrix exists in @xmath255 , \theta\in[0 , \pi/2)$ ] . moreover , signature matrices @xmath256 , \theta\in[0,\pi/4)$ ] and @xmath257 $ ] give the same distance enumerator function since for any @xmath258\in\triangle\mathcal{x}^2 $ ] , @xmath259 holds with @xmath260\in\triangle\mathcal{x}^2 $ ] , where @xmath261 is the complex conjugate of @xmath262 . to continue prove theorem [ thm : two - user ] , we simplify the expression of minimum distance as @xmath263 due to the following facts : @xmath264 since for any @xmath265 , the following holds @xmath266 we obtain the final expression of minimum distance as @xmath267 since for @xmath265 , @xmath268 increases as @xmath269 increases , and @xmath270 decreases as @xmath269 increases , the optimal @xmath271 should satisfy @xmath272 , which leads to @xmath273 , i.e. , @xmath274 $ ] with @xmath133 . the theorem is proved . _ proof : _ let @xmath275 be an index subset . let @xmath276 , where @xmath277 is the complementary set of @xmath278 . from lemma [ lem : distance ] , @xmath279 where @xmath280 . since each row of @xmath71 is a length-@xmath281 optimal signature vector for a single - resource scdma system , for a given @xmath282 , @xmath283 hold for any @xmath284 , where @xmath285 if @xmath31 holds , otherwise , @xmath286 . since @xmath287 , @xmath288 decreases as @xmath289 increases . moreover , since @xmath290 , where @xmath291 is the degree of the @xmath112-th code node in @xmath292 , we have @xmath293 for any @xmath284 . since by varying @xmath278 , @xmath294 can be any proper subset of @xmath295 , by denoting @xmath296 we obtain @xmath297 the lemma is proved . _ proof : _ we first prove that for each @xmath106 there exists @xmath298 with @xmath108 . we just need to show that for a given @xmath106 , there exists an @xmath109 which is a row rotation of @xmath71 . assume that @xmath71 is given . we determine @xmath110 as follows . since the zero elements in @xmath110 are predetermined by the factor graph @xmath99 , we only determine the nonzero elements in @xmath110 . the procedure is similar as that in the proof of theorem [ thm : tree ] except some modifications . terminate the procedure , otherwise , repeat step ii . + note that if there are still undetermined labelings for edges in @xmath156 at the end of the procedure , we simply use the same labeling as in @xmath71 . moreover , the above procedure only applies to the case that the remaining graph after deleting edges in @xmath156 is connected . if it is not connected , i.e. , it contains multiple trees , we can label each of them independently in a similar way . k. alishahi , s. dashmiz , p. pad , and f. marvasti , design of signature sequences for overloaded cdma and bounds on the sum capacity with arbitrary symbol alphabets , " _ ieee trans . inf . theory 14411469 , mar .	sparsely spread code division multiple access ( scdma ) is a non - orthogonal superposition coding scheme that permits a base station simultaneously communicates with multiple users over a common channel . the detection performance of an scdma system is mainly determined by its signature matrix , which should be sparse to guarantee large euclidean distance for the equivalent signal constellation after spreading and superposition . good signature matrices that perform well under both belief prorogation and the maximum likelihood detections are designed . the proposed design applies to several similar well - documented schemes , including trellis code multiple access ( tcma ) , low density spreading , and superposition modulation systems . sparsely spread cdma , non - orthogonal multiple access , signature design , code distance .
the potential of graphene @xcite a two - dimensional ( 2d ) crystal of carbon atoms tightly packed in a honeycomb lattice in optoelectronics , photonics , and plasmonics is attracting a truly considerable attention @xcite . the particle - hole - symmetric spectrum of massless dirac fermions @xcite in graphene and its structureless optical conductivity @xcite are particularly suitable @xcite to realize detectors of radiation in a wide range of photon energies from visible to terahertz ( thz ) frequencies . as a consequence , the photoresponse of a graphene sheet has been the subject of truly intense investigations @xcite . three main photoresponse mechanisms have been identified in the literature to date . ( i ) a finite dc response to an oscillating radiation field can simply stem from the ordinary _ photovoltaic _ effect . photons impinging on the sample excite electron - hole pairs and their basic constituents ( the electron and the hole ) are separated by the electric field across a p - n junction . ( ii ) it has been recognized that hot - carrier - assisted transport plays an important role in graphene @xcite . due to the large optical phonon energy scale in this material , hot carriers created by the radiation field can remain at a temperature higher than that of the lattice for several tens of @xmath0 . equilibration with the lattice occurs indeed mainly because of scattering between carriers and acoustic phonons @xcite . these processes take place on a @xmath1 timescale , although they can experience a speed - up in the case of disorder - assisted collisions @xcite . hot carriers can therefore significantly alter the photoresponse of a graphene sheet @xcite by virtue of the _ photo - thermoelectric _ effect . carrier multiplication @xcite can greatly enhance the performance of graphene photodetectors operating on the basis of photovoltaic and photo - thermoelectric effects . ( iii ) finally , _ bolometric _ effects can play an important role in the photoresponse of a graphene sheet . in this case , the radiation impinging on the sample produces heating , which , in turn , affects the magnitude of the resistance . schematics of a dyakonov - shur single - layer - graphene field - effect transistor for photodetection . an oscillating radiation field coupled to a graphene field - effect transistor creates an oscillating potential difference @xmath2 between the back gate and source . the graphene flake is deposited on a substrate with dielectric constant @xmath3 and is separated by a distance @xmath4 from the back gate . the dielectric constant of the material above the graphene sheet is @xmath5 . the separation between source and drain is @xmath6 . because of the intrinsic nonlinear response of the two - dimensional electron fluid in the field - effect transistor channel , a finite dc potential difference @xmath7 , which is proportional to the power of the incident radiation ( i.e. @xmath8 ) , is measured between the source and drain at _ zero _ source - drain bias . ] in a series of pioneering papers @xcite , which appeared in the mid nineties , dyakonov and shur ( ds ) proposed a very elegant mechanism that yields a finite dc response to an oscillating radiation field . the ds photodetection mechanism is based on the fact that a field - effect transistor ( fet ) hosting a 2d electron gas ( 2deg ) acts as a cavity for plasma waves . when these are weakly damped , i.e. when a plasma wave launched at the source can reach the drain in a time shorter than the momentum relaxation time @xmath9 , the detection of radiation exploits constructive interference of the plasma waves in the cavity , which results in a resonantly enhanced response . this is the so - called _ resonant regime _ of plasma - wave photodetection . plasma waves propagating in the fet channel can not be simply identified with the well - known `` plasmons '' of a 2deg @xcite . indeed , plasma waves are collective oscillations that occur in a _ gated _ 2deg , whereby the long - range tail of the coulomb interaction among electrons is screened by the presence of a metal gate . plasma waves with wave vector @xmath10 ( where @xmath4 is the distance between the 2deg and the gate ) have a gapless dispersion relation @xmath11 ( at zero temperature and neglecting friction and viscosity @xcite ) which resembles that of sound waves in ordinary gases and liquids . on the contrary , plasmons in a 2deg with long - range coulomb interactions have a dispersion relation which is proportional to @xmath12 in the long - wavelength limit @xcite . ds showed @xcite that the photovoltage response of the 2deg in a fet , i.e. the electric potential difference between drain and source , contains a dc component even if the incoming field is ac , and thus provides rectification of the signal . in the resonant regime , the dc photoresponse is characterized by peaks at the odd multiples of the lowest plasma - wave frequency . this rectification mechanism is of purely dynamical origin and is not related to other rectification mechanisms ( occurring , for the example , at the contacts ) which could also be present in a real device . note that rectification of the signal is necessary to detect incoming radiation that exceeds the typical cutoff frequencies of circuit elements . the ds mechanism is therefore particularly useful to detect thz radiation . the ds mechanism relies on two facts . ( i ) the reflection symmetry corresponding to the exchange of source with drain in the fet channel ( see fig . [ fig : one ] ) is broken by the ds boundary conditions . these boundary conditions are unusual because ds fixed the value of the current at the drain and the value of the potential at the source ( instead of operating the device by fixing the current or the potential both at the source and at the drain , as is more customary ) . ( ii ) the fact that the photovoltage averaged over a cycle of the oscillating radiation field is finite ultimately stems from the nonlinearity of the continuity equation and can be viewed as a result of the gate modulating _ both _ the electron density and the drift velocity in the channel . a substantial amount of experimental work has been carried out on ds photodetection in 2degs in ordinary semiconductor heterojunctions : for recent reviews see e.g. refs . . recent experimental work @xcite has demonstrated that the ds photodetection mechanism is active also in the case of the 2d massless dirac fermion ( mdf ) fluid in a graphene fet . et al . _ @xcite have demonstrated room - temperature thz detectors based on antenna - coupled graphene fets , which exploit the ds mechanism but display also contributions of photovoltaic and photo - thermoelectric origin . the plasma waves excited by thz radiation in ref . are overdamped and the fabricated thz detectors , although enabling large area , fast imaging of macroscopic samples , do not _ yet _ operate in the aforementioned resonant regime . in this article we present a theory of ds plasma - wave photodetection in a graphene fet in the resonant regime . we take into account fundamental differences between ordinary parabolic - band 2degs @xcite and 2d mdf fluids . by employing the theory of relativistic hydrodynamics , we demonstrate in a fully analytical fashion that nonlinearities of purely relativistic origin substantially influence the response of graphene - based plasma - wave photodetectors . this article is organized as follows . in sec . [ sect : model ] we present a hydrodynamic theory of transport for mdfs in a gated graphene sheet . in sec . [ sect : instability ] we discuss the plasma - wave instability which arises when a graphene fet is subject to a small dc current bias . in sec . [ sect : photoresponse ] we use the hydrodynamic theory outlined in sec . [ sect : model ] to calculate the photoresponse of a graphene fet to an oscillating electromagnetic field when ds boundary conditions are applied . finally , in sec . [ sect : conclusions ] we summarize our main findings and draw our main conclusions . the theory of hydrodynamics @xcite can be applied to describe transport in a 2d electron fluid when electron - electron ( e - e ) collisions take place on a time scale @xmath13 which is much shorter than the typical time scale of the evolution of macroscopic variables ( density , current , and energy ) . at the same time , collisions with impurities and phonons , which spoil the conservation of momentum and energy , must be assumed to occur much less frequently . in the case of present interest , we require @xmath14 , where @xmath15 is the lowest frequency of a plasma wave in the fet . this requirement implies that plasma waves are collective modes which can be excited in the `` collisional '' regime , when e - e interactions dominate the dynamics . on the contrary , plasmons are collective modes which can be excited in the `` collisionless '' regime , because they can be understood , in the first instance and at long wavelengths , as the sloshing mode of the center of mass of the electron liquid . to estimate the order of magnitude of @xmath16 for plasma waves , we use the following result @xcite for the plasma - wave group velocity @xmath17 which will be re - derived below see eq . ( [ eq : plasmaspeed ] ) . here @xmath18 is the fermi velocity of mdfs in graphene , @xmath4 is the distance between graphene and the gate ( see fig . [ fig : one ] ) , @xmath19 is the fermi momentum corresponding to an average carrier ( electron or hole ) concentration @xmath20 , and @xmath21 is the number of fermion flavors in graphene ( due to spin and valley degrees of freedom ) . finally , @xmath22 is a dimensionless parameter that controls the strength of e - e interactions @xcite . a microscopic definition of @xmath22 will be given below after eq . ( [ eq : plasmaspeed ] ) . we take @xmath23 , @xmath24 , @xmath25 ( corresponding to a typical doping @xmath26 ) , and we find @xmath27 . we notice that @xmath28 with these parameters . the largest wave vector @xmath29 of a plasma - wave resonance supported by the electron liquid in the fet channel is @xmath30 , which corresponds to a standing wave with a node at the source and an antinode at the drain . taking @xmath31 for a typical device , we have @xmath32 ( @xmath33 ) , corresponding to @xmath34 . the transit time of a plasma wave in the device is @xmath35 . a rough estimate of @xmath13 can be obtained by dividing the typical distance between two electrons ( @xmath36 ) by the fermi velocity , i.e. @xmath37 . this is in agreement with recent ultrafast two - color pump - probe measurements that have reported a time scale of the order of tens of femtoseconds to establish thermal equilibrium in single - layer graphene @xcite . for plasma waves in a graphene fet channel we therefore obtain @xmath38 , which shows that the electron system in a typical graphene fet is in the collisional regime . on the contrary , to verify that , in the absence of a gate , plasmon excitations can not be described within a collisional hydrodynamic theory , we consider the dispersion @xcite @xmath39 , evaluated at @xmath40 . using the same parameters as above , we find @xmath41 ( @xmath42 ) and then @xmath43 . when the hydrodynamic assumption holds true , the system is in local equilibrium , i.e. the electron distribution is a fermi - dirac distribution and does not evolve in time due to e - e collisions . however , the parameters of the electron distribution change slowly in time and space and determine thermodynamic and hydrodynamic observables , which are not necessarily steady and evolve in time according to macroscopic conservation laws in the presence of suitable boundary conditions and slowly - varying external potentials . before concluding , we emphasize that for typical device lengths , the plasma - wave frequency @xmath44 is in the thz regime . for example , changing @xmath6 from @xmath45 to @xmath46 , the plasma - wave frequency in a graphene fet changes from @xmath47 to @xmath48 . for this reason , photodetectors based on the ds mechanism are naturally useful in the context of thz light detection . hydrodynamic equations @xcite can be derived from the boltzmann semiclassical equation following a standard procedure @xcite , which relies on the conservation of the particle number , total momentum , and total energy in e - e scattering processes . an outline of the derivation of hydrodynamic equations for 2d mdfs in graphene is reported in the appendix . the continuity equation takes the familiar form @xmath49 where @xmath50 is the total carrier density and @xmath51 is the electron ( hole ) density . carriers drift with the local average velocity @xmath52 . the continuity equation ( [ eq : continuity ] ) is identical in form to the corresponding equation for an ordinary parabolic - band 2deg . the euler and energy equations , on the contrary , are dramatically different from the 2deg case and are reported in the appendix . with the aim of investigating the dynamics of plasma waves in gated graphene flakes , we restrict here to the limit in which the drift velocity @xmath52 is much smaller than the fermi velocity @xmath53 . in this case the euler equation reads @xmath54 here , @xmath55 is the pressure and @xmath56 is the effective potential which acts on electrons , including the electrostatic potential generated by the plasma wave and the fields of nearby conductors . note that the `` free '' parts of the euler and energy equations [ eqs . ( [ eq : eulerfullrel ] ) and ( [ eq : energyfullrel ] ) in the appendix , respectively ] can be written in a very compact form by making use of a covariant notation , commonly employed in relativistic problems @xcite , where the fermi velocity @xmath53 plays the role of the speed of light @xmath57 . in the covariant notation , the euler and energy equations correspond to the space- and time - like component of the divergence of the relativistic energy - momentum tensor @xcite . this formal analogy stems from the highly - nonlinear relation [ see eq . ( [ eq : momentumrudin ] ) in the appendix ] between the average drift momentum @xmath58 and the drift velocity @xmath59 , which , in turn , stems from the linear mdf dispersion of carriers in graphene @xcite . the average drift velocity @xmath52 is smaller than the fermi velocity due to the large number of e - e scattering events that take place in the hydrodynamic limit in a period of a plasma wave . of course , the formal analogy with relativistic fluid dynamics does not mean that the fermi velocity is the largest velocity at which signals can causally propagate in a 2d mdf fluid . indeed , as shown above , the plasma - wave group velocity @xmath60 is typically larger than @xmath53 . furthermore , the coulomb interaction between electrons and all the external potentials propagate instantaneously and do not obey relativistic causality . the hydrodynamic theory of mdfs in graphene is therefore fundamentally different from the hydrodynamic theory of truly relativistic fluids , which is routinely employed to describe e.g. plasmas of astrophysical interest . we now focus our analysis on the limit @xmath61 ( limit of vanishingly small hole concentration ) and approximate @xmath62 . the local density determines the local chemical potential @xmath63 and the local fermi energy @xmath64 , @xmath65 we assume that the temperature @xmath66 is uniform and constant in the sample and that @xmath67 . the pressure [ see eq . ( [ eq : pressurefull ] ) in the appendix ] can then be approximated as @xmath68 the euler equation reduces to @xmath69 let us now consider a graphene flake , which lies in the plane @xmath70 and is subject to the presence of a gate located at @xmath71 . the gate will be modeled as a perfect conductor . according to fig . [ fig : one ] , the dielectric constant @xmath72 is equal to @xmath73 ( @xmath74 ) for @xmath75 ( @xmath76 ) . the effective potential is given by @xmath77 where @xmath78 is the three - dimensional ( 3d ) electric potential , @xmath79 being a coordinate in 3d space . at this stage it is convenient to introduce the so - called `` gate - to - channel swing '' @xcite , @xmath80 where @xmath81 is the uniform electric potential at the gate . from this definition , @xmath82 and @xmath83 is the electric field in the graphene sheet . the relation between the electric potential and the electron density is found by solving the poisson equation @xmath84 . the average density is @xmath85 , where @xmath86 is the geometrical capacitance per unit area . electric potential and density fluctuations are related by @xmath87 where @xmath88 is a 2d wave vector in the graphene sheet and @xmath89 and @xmath90 are the 2d fourier transforms of the electric potential and of the electron density , respectively . in the limit @xmath91 , in which the distance between the graphene sheet and gate is much smaller than the typical wavelength of density oscillations , the relation ( [ eq : channelpotential ] ) between the density and gate - to - channel swing assumes the local form @xmath92 the latter expression is known as the `` gradual channel '' approximation @xcite . in this regime , the long - range tail of the coulomb interaction is completely screened by the charges on the gate . using the gradual channel approximation , the euler and continuity equations read @xmath93 and @xmath94 in this section we make a crude approximation on the euler equation ( [ eq : eulergca ] ) for a 2d mdf fluid to recover earlier results pertaining to the 2deg literature . when the two terms on the second line of eq . ( [ eq : eulergca ] ) are neglected and the fermi energy in the denominator of the first term on the right - hand side of eq . ( [ eq : eulergca ] ) is assumed to be constant in space , @xmath95 , where @xmath96 is the fermi wave number corresponding to the average electron density @xmath97 , eq . ( [ eq : euler_complete ] ) reduces to @xmath98 which is identical to the euler equation used by ds @xcite with the bare electron mass @xmath99 replaced by the density - dependent cyclotron mass @xcite @xmath100 . the validity of this 2deg - like approximation to describe the plasma - wave photoresponse of a graphene fet will be critically examined below . in the following , we assume that our system is translationally invariant in one direction ( the one perpendicular to the source - drain direction ) and therefore specify eqs . ( [ eq : eulergca])-([eq : continuitygca ] ) to a one - dimensional ( 1d ) geometry in the @xmath101 direction , with @xmath102 . linearizing eqs . ( [ eq : continuitygca])-([eq : euler2deg ] ) we obtain @xmath103 and @xmath104 it is well known that these two coupled equations admit solutions in the form of traveling waves @xmath105 with @xmath11 . the plasma - wave velocity is given by @xcite @xmath106 where @xmath107 is a dimensionless coupling constant that controls the strength of e - e interactions @xcite . as shown above , for typical carrier densities and device sizes , @xmath28 . our derivation , however , does not formally rely on this inequality and applies also to the regime in which the fermi velocity and the plasma wave velocity are comparable , a situation which is achieved when @xmath108 . the euler equation ( [ eq : eulergca ] ) is not amenable to analytic treatment . we therefore introduce an expansion of this equation in powers of the deviation @xmath109 of the local density @xmath110 from its average value @xmath97 . because of the gradual channel approximation ( [ eq : gca ] ) , a similar expansion can be carried out in powers of @xmath111 . at this stage , it is also convenient to introduce _ dimensionless _ quantities by scaling lengths , electrical potentials , velocities , and time with @xmath6 , @xmath112 , @xmath60 , and @xmath113 , respectively . for the sake of notational simplicity , dimensionless variables will be denoted with the same symbols used for dimensionful variables . expanding the euler equation ( [ eq : eulergca ] ) up to _ second order _ in @xmath114 we obtain @xmath115 once again , all the quantities appearing in the previous equation ( @xmath101 , @xmath116 , @xmath117 , etc . ) are dimensionless . the first two terms in the right - hand side of eq . ( [ eq : eulermdf ] ) stem from the expansion in powers of @xmath114 of the term in eq . ( [ eq : eulergca ] ) which is proportional to @xmath118 . the third and the fourth term originate from @xmath119 and @xmath120 . the first line in eq . ( [ eq : eulermdf ] ) reproduces the 1d version of eq . ( [ eq : euler2deg ] ) . the terms contained in the last three lines of eq . ( [ eq : eulermdf ] ) are qualitatively new terms pertaining to the hydrodynamic theory of the 2d mdf fluid . they do not have an equivalent in a 2deg . the implications of these nonlinear terms on the plasma - wave photoresponse of a gated graphene sheet will be the subject of the reminder of this article . we now proceed to estimate the relative magnitude of each term in the right - hand side of eq . ( [ eq : eulermdf ] ) . we assume a profile @xmath121 with @xmath122 . in this case , the differential operators simplify to @xmath123 and @xmath124 , and @xmath29 can be collected as a common prefactor . we remind the reader that @xmath125 . as discussed above , typically @xmath126 ( the fermi velocity is much smaller than the plasma - wave group velocity when @xmath127 ) . the first nonlinear term on the right - hand side of eq . ( [ eq : eulermdf ] ) can be estimated as @xmath128 ^ 2~.\ ] ] similarly , @xmath129u(k,\omega)~. \end{split}\ ] ] we therefore see that the term ( [ eq : fluctuationsmass ] ) does not contain velocity factors . keeping only this term in eq . ( [ eq : eulermdf ] ) can be seen as a `` zeroth - order expansion '' in powers of both @xmath130 and @xmath126 . in this section we study the dispersion of plasma waves in a 2d mdf fluid that moves with an average constant speed @xmath131 , measured in units of the plasma - wave speed @xmath60 defined in eq . ( [ eq : plasmaspeed ] ) . we repeat the analysis of sect . [ ssec:2deglike ] , allowing , however , for @xmath132 . to this aim , we discard all the terms in eq . ( [ eq : eulermdf ] ) which are @xmath133 obtaining @xmath134 as usual , the previous equation must be supplemented by the continuity equation . we now seek solutions of the form @xmath135 and @xmath136 , with @xmath137 we find that for each value of @xmath138 , two values of @xmath29 are possible , corresponding to waves traveling in opposite directions . in the limit @xmath139 , the dispersion reads @xmath140 where @xmath141 the previous result differs from eq . ( [ eq : plasmaspeed ] ) because we have taken into account the contribution of the pressure to the restoring force , i.e. the term @xmath142 in eq . ( [ eq : eulerlin ] ) . the correction @xmath143 in eq . ( [ eq : relsoundspeed ] ) can therefore be understood from the following general argument . from eqs . ( [ eq : fermienergy ] ) and ( [ eq : pressure ] ) we see that the 2d mdf fluid fulfills a _ polytropic _ equation of state , @xmath144 , @xmath145 being the mass density and @xmath146 . then , the usual thermodynamic expression for the sound speed gives @xmath147 . before concluding , we would like to comment on the factor @xmath148 which appears in front of @xmath131 in eq . ( [ eq : relvelsum ] ) . from eq . ( [ eq : relvelsum ] ) with @xmath149 , we see that @xmath150 is the plasma - wave speed in a reference system which moves with velocity @xmath151 with respect to the laboratory rest frame . contrary to the 2deg case , however , the plasma - wave speed in the laboratory rest frame can not simply be obtained by performing a galilean transformation @xmath152 . in the case of present interest , we find that there is a correction of magnitude @xmath153 which reduces ( increases ) the plasma - wave speed , with respect to the result of the galilean transformation @xmath154 . this fact is well - known in the context of relativistic hydrodynamics @xcite . for future purposes , we introduce the parameter @xmath155 the dispersion relation ( [ eq : relvelsum ] ) characterizes the collective modes of a 2d mdf fluid in the presence of a metal gate and in the thermodynamic limit . ds showed @xcite that appropriate boundary conditions in a device of length @xmath6 define a resonator which supports discrete modes with frequencies @xmath156 . moreover , when the average drift velocity @xmath151 varies in a certain interval , the resonating modes become unstable and their amplitude increases exponentially with a rate @xmath157 , which is independent of @xmath158 . the ds boundary conditions for the instability are @xmath159 where @xmath160 is the current density . the boundary conditions correspond to a constant density at the source ( @xmath161 ) and a constant current at the drain ( @xmath162 ) . we study the ds instability with both space and time derivatives of the pressure taken into account . however , we neglect momentum relaxation , which takes place on a time scale @xmath9 . this approximation is consistent if we assume that the exponential growth of the amplitude of the modes is quenched by nonlinear terms in the equations of motion on a time scale shorter than @xmath9 . the boundary conditions for the fluctuations , which follow from eq . ( [ eq : bcinst ] ) , read @xmath163 to find the allowed values of @xmath138 , we insert the ansatz ( [ eq : travelingwaves ] ) into eq . ( [ eq : bclininst ] ) and solve for @xmath138 to linear order in @xmath151 . we find that the real part of the frequency is of the form @xmath164 where @xmath158 is a positive integer , while the imaginary part is @xmath165 . the fundamental @xmath166 plasma resonance defines the fermi - velocity - dependent plasma frequency @xmath167 with respect to the ordinary 2deg @xcite , the instability rate @xmath168 is reduced by a factor @xmath148 . as noted above , this effect is of purely relativistic origin and stems from the non - galilean nature of electron dynamics in graphene . fig02(2,60 ) we now consider the response of the system to a periodic modulation of the gate - to - channel swing at the source , with vanishing current at the drain . the periodic modulation at the source can be induced by an impinging radiation ( see fig . [ fig : one ] ) , collected , for instance , by means of an appropriate antenna @xcite . we want to solve eq . ( [ eq : eulermdf ] ) together with the continuity equation @xmath169 with the ds boundary conditions for detection : @xmath170 here @xmath171 is the frequency of the incoming radiation . our aim is to demonstrate that a dc potential difference @xmath7 is generated between the drain and source in response to the oscillating radiation and _ in the absence of a source - drain bias_. most importantly , we will elucidate the contributions to @xmath7 arising from the peculiar relativistic corrections to the euler equation of motion pertaining to graphene . following ds @xcite , we seek solutions of the euler ( [ eq : eulermdf ] ) and continuity ( [ eq : continuity1d ] ) equations in the form of power expansions : @xmath172 and @xmath173 here , @xmath174 is a dimensionless parameter that helps with the bookkeeping while solving eqs . ( [ eq : eulermdf ] ) and ( [ eq : continuity1d ] ) perturbatively . the functions @xmath175 , @xmath176 and @xmath177 , @xmath178 are assumed to oscillate in time with frequency @xmath171 and @xmath179 , respectively , while @xmath180 and @xmath181 are constant in time . with the ansatz ( [ eq : expansionu ] ) the dc photovoltage is @xmath182 . this perturbative approach is fully justified when the external perturbation is sufficiently weak , i.e. when @xmath183 . we find that @xmath184 , i.e. the dc photovoltage is proportional to the power of the incoming radiation . to linear order , the equations of motion read : @xmath185 and @xmath186 with boundary conditions @xmath187 note that in eq . ( [ eq : eulerlin ] ) we have included a linear phenomenological friction term , proportional to the inverse momentum relaxation time @xmath188 . the solution of the linear system of differential equations ( [ eq : eulerlin])-([eq : contlin ] ) is given by @xmath189 and @xmath190 with @xmath191 note that in the limit @xmath192 eq . ( [ eq : dispersionk ] ) coincides with eq . ( [ eq : relvelsum ] ) with @xmath149 . fig03a(2,60)(a ) fig03b(2,60)(b ) we then write the euler equation to second order in @xmath174 and average it over one period of the incoming radiation . we find @xmath193 where @xmath194 denotes the time average over one period @xmath195 of the external radiation . ( [ eq : euleraverage ] ) depends on @xmath180see eq . ( [ eq : expansionv ] ) . this quantity can be easily obtained by averaging over time the continuity equation , written up to second order in @xmath174 . we find @xmath196 . the final expression for the photovoltage is @xmath197 remarkably , we have found an analytical expression for @xmath7 which can be given in the following rather compact form : @xmath198 where @xmath199 , @xmath200 , @xmath201 , and @xmath202 ( @xmath203 ) is the real ( imaginary ) part of @xmath204 , which reads @xmath205 eq . ( [ eq : photovoltage ] ) reduces to the ds result for an ordinary 2deg @xcite ( with the cyclotron mass @xmath206 playing the role of the electron mass @xmath99 ) in the limit @xmath207 . the terms controlled by @xmath208 stem from the local fluctuations of the cyclotron mass . this coefficient tends to unity in the limit @xmath126 . the terms controlled by @xmath209 originate from the derivatives of the pressure and suppress the photoresponse with respect to the case @xmath210 . illustrative plots of @xmath7 as a function of the incoming radiation frequency @xmath171 are presented in fig . [ fig : photovoltage ] for a given value of the momentum relaxation time @xmath9 . the photoresponse features maxima at odd multiples of the plasma frequency ( [ eq : omegapvf ] ) . nervous readers can easily convert dimensionless values of the momentum relaxation time @xmath9 into values of the mobility @xmath211 . the mobility @xmath211 is obtained from eq . ( [ eq : eulerlin ] ) in the steady - state regime ( i.e. when the time derivative on the left - hand side is neglected ) and reads @xmath212 . restoring , for a moment , physical units , the previous relation reads : @xmath213 more conveniently , @xmath214 where @xmath9 , @xmath158 , and @xmath4 must be expressed in @xmath0 , @xmath215 , and @xmath216 , respectively . the photovoltage ( [ eq : photovoltage ] ) depends sensibly on the ratio @xmath217 between the fermi velocity @xmath53 and the effective plasma - wave speed @xmath150 . from eq . ( [ eq : xi ] ) we see that this ratio can be tuned in the range @xmath218 $ ] by changing the distance @xmath4 between graphene and the gate , the dielectric constant @xmath3 , and the average carrier density @xmath20 . [ fig : photovoltage ] shows that the photoresponse of a system that is described by relativistic hydrodynamics can be tuned to regimes where its resonant maxima have larger amplitude and its minima are shallower . we are therefore naturally led to introduce the visibility @xmath219 of the resonant photovoltage profile according to @xmath220 where @xmath221 ( @xmath222 ) is the maximum ( minimum ) of the photoresponse evaluated at @xmath223 $ ] ( @xmath224$])see eq . ( [ spagnasuca ] ) . illustrative plots of @xmath219 are shown in fig . [ fig : visibility ] . we see that , for all values of the momentum relaxation time @xmath9 , the visibility drastically increases as @xmath217 increases . in particular , in the limit @xmath225 the visibility reaches its maximum value . reaching values of @xmath217 of the order of @xmath226 requires to minimize the product @xmath227 . values of this product as small as @xmath228 can be reached by placing graphene on high - quality substrates like h - bn , where carrier densities as low as @xmath229 can be reached without entering the regime where disorder ( electron - hole puddles ) dominates . in these samples , gold metal gates can be placed as close as @xmath230 to the graphene sheet , as recently shown in a quantum capacitance measurement carried out in single - layer graphene @xcite . in this article we have presented a theory of a photoresponse mechanism which is based on the excitation of plasma waves in the channel of a graphene field - effect transistor ( fet ) . the fet is coupled to a source of radiation which periodically modulates the voltage difference between the gate and source . our main result is a fully analytical expression eq . ( [ eq : photovoltage])for the dc voltage difference @xmath7 between the drain and source , in response to the oscillating radiation field . plasma waves in the channel are described within a hydrodynamic model which takes into account the linear dispersion of massless dirac fermions in graphene . formally , the hydrodynamic theory of two - dimensional massless dirac fermions in graphene presents strong analogies with the hydrodynamic theory of a relativistic fluid @xcite , with the fermi velocity @xmath53 playing the role of the speed of light @xmath57 . in particular , we have shown that this yields a breakdown of galilean invariance whereby the speed of plasma waves in the frame of the moving fluid is not connected by a galilean transformation to that in the laboratory frame . moreover , the photoresponse substantially depends on the peculiar nonlinearities of the equations of motion which arise due to the formal relativistic nature of the energy - momentum dispersion of carriers in graphene . we have shown that it is possible to leverage the effect of these nonlinearities to increase the photoresponse of a gated graphene sheet with respect to that occurring in an ordinary two - dimensional electron gas . measuring the photoresponse of a graphene sheet in a fet geometry offers the opportunity to demonstrate the importance of relativistic corrections to the ordinary hydrodynamic theory . finally , we emphasize that the present theory assumes that the gate of the fet is as long as the channel . in certain experimental situations @xcite it may be more convenient to exploit gates which are much shorter than the fet channel . in this case the theory of this article needs to be changed . the dependence of the ds - like dc photovoltage on the position of the gate with respect to the source , say , may present non - trivial features . in the case of a short gate , however , the dc photovoltage will suffer @xcite from contributions of thermoelectric and photoconductive origin . last but not least , we would like to emphasize that vicarelli _ et al . _ @xcite have also shown that _ bilayer graphene _ fets can be used to fabricate plasma - wave thz photodetectors with low noise - equivalent power . in the case of bilayer graphene , we note that modulating the gate - to - source voltage will not only modulate the density in the fet channel but also the band gap @xcite . these issues are well beyond the scope of the present article and will be investigated in forthcoming publications . it is a pleasure to thank vittorio pellegrini and alessandro tredicucci for useful discussions . this work was supported by the italian ministry of education , university , and research ( miur ) through the program `` firb - futuro in ricerca 2010 '' grant no . rbfr10m5bt ( `` plasmograph : plasmons and terahertz devices in graphene '' ) . we have made use of free software ( www.gnu.org , www.python.org ) . for the sake of completeness , in this appendix we summarize the derivation of the hydrodynamic equations for graphene , starting from the boltzmann semiclassical equation @xcite for 2d mdfs @xcite : @xmath231~.\end{aligned}\ ] ] here , @xmath232 represents the probability to find an electron ( hole ) with momentum @xmath233 , at position @xmath234 and time @xmath116 . the relation between velocity @xmath235 and wave vector @xmath88 is @xmath236 it is important to note that this relation , peculiar to mdfs , is strongly nonlinear @xcite . in the ordinary schrdinger case @xmath237 . in eq . ( [ eq : kineticequation ] ) @xmath56 is the effective potential energy felt by an electron at position @xmath234 and time @xmath116 , which includes the effect of external potentials . the self - consistent electrostatic potential generated by the electron distribution ( i.e. the rpa contribution to the electron self - energy @xcite ) should be included in @xmath56 @xcite . the term @xmath238 $ ] on the right - hand side of eq . ( [ eq : kineticequation ] ) represents the boltzmann collision integral , which takes into account the effect of e - e scattering events on the time - evolution of the distribution function . in graphene , particular care has to be taken in the calculation of this integral @xcite due to the existence of peculiar scattering events ( auger processes ) which have a vanishing phase space . as discussed in sec . [ sect : model ] , the hydrodynamic theory relies on the assumption that the system is in local equilibrium with respect to e - e collisions . formally , this means that the distribution functions @xmath239 fulfill @xmath240=0 $ ] at each point in space and time . quite generally , since each e - e scattering event conserves the number of particles , total momentum , and total energy , the form of the distribution function which nullifies the collision integral is @xcite @xmath241 local macroscopic averages @xmath242 of wave - vector dependent quantities @xmath243 are defined by @xmath244 where @xmath245 indicates integration over momentum space and summation over spin and valley degrees of freedom . it is easy to see that the parameters @xmath52 , @xmath246 , and @xmath63 introduced in eq . ( [ eq : quasieqansatz ] ) have the physical meaning of local average velocity , temperature , and chemical potential , respectively . in the case of ordinary schrdinger fermions , the relation between average velocity and wave vector follows by linearity from @xmath247 and reads @xmath248 . in the 2d mdf case , on the contrary , one finds @xcite @xmath249 where @xmath250 and @xmath251 are the density and pressure , respectively . the pressure is given by @xcite @xmath252 where @xmath253 is the electron ( hole ) density , @xmath254 , @xmath255 , and @xmath256 is the polylogarithmic function . ( [ eq : pressurefull ] ) is the equation of state of 2d mdfs . the hydrodynamic equations for density ( i.e. the continuity equation ) , momentum ( i.e. the euler equation ) , and energy can be obtained following the standard procedure @xcite of multiplying the boltzmann equation by @xmath257 , @xmath258 , and @xmath259 , respectively , and averaging the resulting equations as from eq . ( [ eq : averaging ] ) . the euler equation reads @xmath260 we remind the reader that the euler equation is usually @xcite written in terms of the average velocity @xmath59 only . in the present case of a 2d mdf fluid , however , rewriting eq . ( [ eq : eulerfullrel ] ) in terms of @xmath52 generates a rather complicated expression , due to the nonlinear relation ( [ eq : momentumrudin ] ) . ( [ eq : euler ] ) represents the final result in the `` non - relativistic '' limit @xmath261 . finally , the energy equation reads @xmath262 with @xmath263 where @xmath264 ( @xmath265 ) is the @xmath266th component of @xmath234 ( @xmath267 ) . in this article we have not made use of the energy equation . we emphasize that a proper description of thermal transport and thermoelectricity requires to take into consideration energy dissipation ( e.g. to lattice phonons ) and heat sources . kuzmenko , e. van heumen , f. carbone , and d. van der marel , http://dx.doi.org/10.1103/physrevlett.100.117401 [ , 117401 ( 2008 ) ] ; r.r . nair , p. blake , a.n . grigorenko , k.s . novoselov , t.j . booth , t. stauber , n.m.r . peres , and a.k . geim , http://dx.doi.org/10.1126/science.1156965[science * 320 * , 1308 ( 2008 ) ] ; f. wang , y. zhang , c. tian , c. girit , a. zettl , m. crommie , and y.r . shen , http://dx.doi.org/10.1126/science.1152793[_ibid._ * 320 * , 206 ( 2008 ) ] . echtermeyer , l. britnell , p.k . jasnos , a. lombardo , r.v . gorbachev , a.n . grigorenko , a.k . geim , a.c . ferrari , and k.s . novoselov , http://dx.doi.org/10.1038/ncomms1464[nature commun . * 2 * , 458 ( 2011 ) ] . d. brida , a. tomadin , c. manzoni , y.j . kim , a. lombardo , s. milana , r.r . nair , k.s . novoselov , a.c . ferrari , g. cerullo , and m. polini , http://dx.doi.org/10.1038/ncomms2987[nature commun . * 4 * , 1987 ( 2013 ) ] . w. knap , m. dyakonov , d. coquillat , f. teppe , n. dyakonova , j. lusakowski , k. karpierz , g. valusis , d. seliuta , i. kasalynas , a. el fatimy , y.m . meziani , and t. otsuji , http://dx.doi.org/10.1007/s10762-009-9564-9[j . infrared millim . terahz waves * 30 * , 1319 ( 2009 ) ] . w. knap , s. rumyantsev , m.s . vitiello , d. coquillat , s. blin , n. dyakonova , m. shur , f. teppe , a. tredicucci , and t. nagatsuma , http://dx.doi.org/10.1088/0957-4484/24/21/214002[nanotech . * 24 * , 214002 ( 2013 ) ] . note that the coupling constant @xmath22 defined in eq . ( [ eq : alphaeegate ] ) differs from the usual one that appears in the theory of e - e interactions in single - layer graphene in the absence of the gate , which is given by @xcite @xmath268 , with @xmath269 . yu , r. jalil , b. belle , a.s . mayorov , p. blake , f. schedin , s.v . morozov , l.a . ponomarenko , f. chiappini , s. wiedmann , u. zeitler , m.i . katsnelson , a.k . geim , k.s . novoselov , and d.c . elias , http://dx.doi.org/10.1073/pnas.1300599110[proc . ( usa ) * 110 * , 3282 ( 2013 ) ] .	the photoresponse of graphene has recently received considerable attention . the main mechanisms yielding a finite dc response to an oscillating radiation field which have been investigated include responses of photovoltaic , photo - thermoelectric , and bolometric origin . in this article we present a fully analytical theory of a photoresponse mechanism which is based on the excitation of plasma waves in a gated graphene sheet . by employing the theory of relativistic hydrodynamics , we demonstrate that plasma - wave photodetection is substantially influenced by the massless dirac fermion character of carriers in graphene and that the efficiency of photodetection can be improved with respect to that of ordinary parabolic - band electron fluids in semiconductor heterostructures .
in recent years , networks have gained popularity as a tool to represent , organize , and interpret phenomena arising in many fields of science , including physics , biology , social sciences , etc . questions as diverse as the structure of the world wide web , the robustness of a nation s banking system or its power grid , or the mechanism of functions inside a cell can be expressed in terms of networks . these applications have led to networks whose structure and complexity have gone far beyond the examples studied before in the classical computer science literature . driven partly by the emergence of these new applications , research in network science has also undergone a revolutionary change in recent years . while traditional network science was basically a subject of graph theory and focused on networks with rather simple structure , recent studies often took the viewpoint of treating networks as complex systems , and used tools and concepts from statistical mechanics . while the structure and topology of networks has been under much investigation , the dynamics on the network is less well understood despite the fact that it leads to important and nontrivial questions . for example , any network with positive - weighted edges defines a markov jump process ( mjp ) ( and vice versa ) and in many applications , it is of interest to understand the interplay between the network structure and the dynamics of this mjp . our aim here is to address such questions within the framework of transition path theory ( tpt ) , originally introduced in @xcite ( see also @xcite for reviews ) and already used in @xcite in the context of networks and mjps the present work can be viewed as a continuation of this last paper . in a nutshell , the basic idea in tpt is to single out two specific sets of nodes and analyze the statistical properties of the reactive trajectories by which transitions between these sets occur if the sets are chosen appropriately , this permits to extract the most salient features of the dynamics on the network and relate them to its topology . this is like probing an electrical network by wiring it at different locations and analyzing how the current flow from the nodes wired positively to those wired negatively @xcite . tpt is also related to the potential - theoretic approach to metastability championed by bovier and collaborators @xcite , albeit the emphases of both approaches are different . the potential - theoretic approach has been introduced as a theoretical tool to obtain rigorous bounds on the low - lying eigenvalues that characterize the slowest relaxation phenomena in mjps displaying metastability @xcite . tpt on the other hand permits to characterize exactly the statistical properties of the transition pathways on complex networks that are not necessarily metastable , or such that the low - lying part of their spectrum is too complicated to be estimated analytically . importantly tpt can also be used as a computational tool in such situations . by being able to analyze the flow of transitions between specific parts of the network , for example by generating numerically reactive trajectories by which these transitions occur , or even no - detour transitions paths , and analyzing their statistical properties , tpt can provide invaluable information about the network and the dynamics it supports . ( a ) the two lowest minima of the potential energy of the lj@xmath1 . ( a ) : the face - centered cubic truncated octahedron with the point group @xmath2 is the lowest minimum . ( b ) : the icosahedral structure with the @xmath3 point group is the second lowest minimum . throughout this paper we refer to them as fcc and ico , respectively.,title="fig:",height=188 ] ( b ) the two lowest minima of the potential energy of the lj@xmath1 . ( a ) : the face - centered cubic truncated octahedron with the point group @xmath2 is the lowest minimum . ( b ) : the icosahedral structure with the @xmath3 point group is the second lowest minimum . throughout this paper we refer to them as fcc and ico , respectively.,title="fig:",height=188 ] to make this last point and illustrate the usefulness of the tools developed in this paper , we will apply them to analyze the network developed by david wales and collaborators to model the dynamics of lennard - jones clusters with 38 atoms ( @xmath0 ) @xcite . @xmath0 is a prototypical example illustrating how the complexity of a system s energy landscape ( and its associated network ) affects its dynamical properties , a feature that is also observed in other complex phenomena such as protein folding or glassy dynamics . @xmath0 has a double - funnel landscape : its global minimum , a face - centered - cubic truncated octahedron , lies at the bottom of one funnel , whereas its second lowest minimum , an incomplete mackay icosahedron , lies at the bottom of the other ( see fig . [ fcc_ico ] ) . the deeper octahedral funnel is also narrower , and believed to be mostly inaccessible from the liquid state . thus , when @xmath0 self - assembles by crystallization , it does so by reaching the bottom of the shallow but broader isocahedral funnel , and an interesting question is how does @xmath0 manage to subsequently find its ground state structure by travelling from the shallow funnel to the deep one ? this question of rearrangement is the one that we will address below . it is made complicated by the ruggedness of the energy landscape of @xmath0 , which has an enormous number of local minima separated by a hierarchy of barriers of different heights . the remainder of this paper is organized as follows . in sec . [ sec : tpt ] we summarize the main outputs of tpt . in sec . [ sec : current ] we introduce sampling tools based on the theory . in sec . [ sec : metastable ] we discuss the case of metastable networks , and establish connections between tpt and the potential theoretic approach to metastability as well as large deviation theory that arise in these situations . in sec . [ sec : lj38 ] we apply the tools introduced earlier to analyze the rearrangement of the @xmath0 network . finally , some concluding remarks are given in sec . [ sec : lj38 ] . tpt for networks and markov jump processes ( mjps ) is discussed in detail in @xcite ( see also @xcite ) . here we give a brief summary of the theory , then discuss algorithms based on it that can be used to characterize the flows on the network . we also comment on the connections between tpt and spectral approaches to network analysis , bovier s potential theoretic approach to metastability in mjps , and large deviation theory . we will consider mjps on a countable state - space @xmath4 with infinitesimal generator @xmath5 : @xmath6 where @xmath7 for @xmath8 denotes the probability that the process jumps from state @xmath9 to state @xmath10 in the infinitesimal time interval @xmath11 $ ] . any such mjp is equivalent to a network which we denote by @xmath12 : the set @xmath4 of states of the mjp is the set of nodes in the network , and @xmath13 is the set of edges , i.e. the set of ordered pairs @xmath14 with @xmath15 such that @xmath16 . conversely , any network with positive weighted edges is equivalent to an mjp by interpreting the weights on these edges as off - diagonal entries of the mjp generator . we assume that the generator is irreducible and that the mjp is ergodic with respect to the equilibrium probability distribution @xmath17 satisfying @xmath18 for simplicity , we also assume that the mjp is time - reversible , i.e. that the detailed balance property holds @xmath19 we denote by @xmath20 the instantaneous position of the mjp and following standard conventions we assume that the function @xmath21 is right - continuous with left limits ( _ cdlg _ ) . tpt is a framework to understand the mechanism by which transitions from any subset @xmath22 to any disjoint subset @xmath23 occur in the mjp . specifically , tpt analyzes the statistical properties of the _ reactive trajectories _ by which these transitions occur : if @xmath24 denotes an infinitely long equilibrium trajectory of the mjp , the reactive trajectories associated with it are the successive pieces of @xmath24 during which it has last left @xmath25 and is on its way to @xmath26 next . tpt gives explicit expressions for the probability distribution of the reactive trajectories , their probability current , their rate of occurrence , etc . besides the equilibrium probability distribution @xmath27 and the generator @xmath28 , the expressions for these quantities involve the committor @xmath29 , defined as the probability that the process starting at a state @xmath30 will first reach @xmath26 rather than @xmath25 : @xmath31 where @xmath32 denotes the first hitting time of set @xmath33 starting from @xmath9 : @xmath34 the committor is also known as equilibrium potential of the capacitor @xmath35 , and is denoted by @xmath36 in the collection of works of bovier _ et al . _ ( see e.g. @xcite ) . it satisfies @xmath37 and it can be used to estimate various statistical descriptors of the reactive trajectories . for example , the equilibrium probability to find the process in state @xmath9 and that it be reactive which is called the _ probability distribution of reactive trajectories _ is given by @xmath38 indeed , the equilibrium probability to find the trajectory in @xmath9 is @xmath39 , and the probability that it is reactive , is the product between @xmath40 , which gives the probability that it will reach @xmath26 rather than @xmath25 next , and @xmath41 , which by time - reversibility gives the probability that it came from @xmath25 rather than @xmath26 last . note that @xmath42 is only non - zero if @xmath43 . note also that this distribution is not normalized to one : the quantity @xmath44 gives the probability that the trajectory be reactive ( i.e. the proportion of time it spends traveling from @xmath25 to @xmath26 at equilibrium ) , and the probability to find the trajectory at state @xmath9 at equilibrium conditional on it being reactive is @xmath45 . similarly , we can calculate the average number of transitions per unit time that the reactive trajectories make from state @xmath9 to state @xmath46 : @xmath47 the additional factor @xmath48 beside the usual @xmath49 accounts for the requirement that , in order to be reactive , the trajectory must have reached @xmath9 coming from @xmath25 last and it must reach @xmath26 next after leaving @xmath10 . by antisymmetrizing @xmath50 we obtain the _ _ probability current of reactive trajectories _ _ : @xmath51 this current is key to understand the mechanism of the reaction as it permits to locate the productive channels by which this reaction occurs in contrast , both and indicate where the reactive trajectories go , but these locations may include many dynamical traps and/or deadends that these trajectories visit but do not contribute to their current towards @xmath26 . we will elaborate on these points in sec . [ sec : current ] . the current also permits to calculate the average number of transitions per unit time as the total current out of @xmath25 or into @xmath26 : @xmath52 this quantity is referred to as the _ reaction rate _ and it can also be expressed as @xmath53 follows from the detailed balance condition and the conservation of the current ( theorem 2.13 in @xcite ) : @xmath54 for all @xmath55 . the reaction rate @xmath56 should not be confused with the rates @xmath57 and @xmath58 defined respectively as the inverse of the average time it takes the trajectory to go back to @xmath26 after hitting @xmath25 or back to @xmath25 after hitting @xmath26 . these rates are given by @xmath59 where @xmath60 are the proportions of time such that the trajectory last hit @xmath25 or @xmath26 , respectively . in this section we show how the outputs of tpt can be used to understand the mechanism of the transitions from @xmath25 to @xmath26 . if we want to know where these trajectories go , this can be done by analyzing and . some of the locations visited by the reactive trajectories may be deadends , however , in the sense that not much current goes through them . in order to determine the productive paths ( in term of probability current ) taken by the reactive trajectories , we need to analyze the current . some tools to perform this analysis were already introduced in @xcite . for example , it was shown how to identify a dominant representative path , in the sense that this path maximizes the current it carries . while such a path can be informative about the mechanism of the reaction , it can also be misleading in situations where the probability current of reactive trajectories is supported on many paths which carry little current individually in other words , in situations where the reaction channel is spread out . here we introduce tools that are appropriate in these situations as well , since we expect them to be quite generic in complex networks . specifically , we provide ways to generate directly reactive trajectories that flow from @xmath25 to @xmath26 without even returning to @xmath25 , or even trajectories that only take productive steps towards @xmath26 . the statistical analysis of these trajectories then provides ways to analyze the flows in the network , which we also discuss . the following technical assumptions will be used below to simplify the discussion : * _ @xmath61 if @xmath62 and @xmath63 , i.e. the mjp can not jump directly from @xmath25 to @xmath26 with this condition , every reactive trajectory visits at least one state outside of @xmath64 . _ * _ @xmath65 and @xmath66 if @xmath67 . _ * _ @xmath68 if @xmath15 and @xmath69 . _ it is straightforward to generalize the statements in propositions [ th : tpp ] and [ th : lftpp ] below to situations where these assumptions do not hold , as indicated in the proofs , but it makes them slightly more involved . our first result is a proposition that indicates how to generate reactive trajectories directly . the main idea is to lump onto an artificial state @xmath70 all the pieces of the trajectory in the original mjp during which it is not reactive . we call the process obtained this way the transition path process , following the terminology introduced in @xcite , where a similar construction was made in the context of diffusions : [ th : tpp ] suppose that assumptions ( a ) and ( b ) hold , let @xmath71 , and consider the process on the state - space @xmath72 defined by the generator with off - diagonal entries given by @xmath73 where @xmath74 is the probability that the trajectory is reactive ( see eq . ) . then this process has the same law as the one obtained from the original mjp by mapping every non - reactive piece of its trajectory onto state @xmath70 . in particular , on @xmath75 the invariant probability distribution of the transition path process coincides with the probability distribution of the reactive trajectories given in , and the average number of transition per unit time that the transition path process makes between states in @xmath75 is given by and the associated current by . the proof of this proposition is given at the end of this section . note that we can supplement the transition path process with the information that when it jumps to @xmath76 from @xmath70 , it comes from state @xmath62 with probability @xmath77 and when it jumps to @xmath70 from @xmath78 , it reaches state @xmath63 with probability @xmath79 with this information added , the invariant probability current of the transition - path process is the same as the one in of the reactive trajectories even if we include edges that come out of @xmath25 or into @xmath26 . by construction , in the transition path process ( like in the reactive trajectories it represents ) , the trajectories go from @xmath25 to @xmath26 directly , without ever returning to @xmath25 in between in the transition path process , these returns arise through visits to state @xmath70 . in contrast , if we were to simply turn @xmath25 into a source and @xmath26 into a sink , the process one would obtain could take many steps to travel from @xmath25 to @xmath26 because it could revisit @xmath25 often before making an actual transition this problem is especially acute if @xmath25 and @xmath26 are metastable states since , by definition , @xmath25 is then revisited often before a transition to @xmath26 occurs ( more on metastability in sec . [ sec : metastable ] ) . in such situations , the reactive trajectories are much shorter since by construction they only contain this last transitioning piece . it should be stressed , however , that the reactive trajectories could still take many steps to travel from @xmath25 to @xmath26 and be complicated themselves . for example if the transition mechanism involves dynamical traps or deadends along the way , the reactive trajectories will wander a long time in the region between @xmath25 and @xmath26 before finally making their way to @xmath26 . in such situations , it is convenient to construct a process that carries the same probability current as the reactive trajectories , but makes no detour to go from @xmath25 to @xmath26 . by this we mean the following : if we look at the way the committor function varies along a reactive trajectory , it will start at 0 in @xmath25 and go to 1 in @xmath26 , but it will not necessarily increase monotonically between these values along the way . let us call the pieces of the reactive trajectories along which the committor increases the productive pieces , in the sense that they are the ones that bring these trajectories closer to the product @xmath26 , whereas they make a detour along any other piece . imagine patching together these productive pieces in such a way that the resulting process is markov and carries the same probability current as the reactive trajectories . it turns out that there is a precise way to do so , and this defines what we call the no - detour transition path process : [ th : lftpp ] suppose that assumptions ( a ) , ( b ) , and ( c ) hold , let @xmath80 and consider the process on the state - space @xmath81 defined by the generator with off - diagonal entries @xmath82 where @xmath83 and @xmath84 . then this process has the same stationary current as the transition path process , but the committor function increases monotonically along each of its paths on @xmath75 . in particular , these paths have no loops . the proof of this proposition is given at the end of this section . processes similar to the one in this proposition were introduced in @xcite . note that the equivalent of the no - detour transition path process for diffusions is somewhat trivial since the ` no - detour ' trajectories in this context are simply the flowlines of the probability current of reactive trajectories , which are deterministic . note also that we can again supplement this process with the information that when it jumps to @xmath76 from @xmath70 , it comes from state @xmath62 with probability , and when it jumps to @xmath70 from @xmath85 , it reaches state @xmath63 with probability . propositions [ th : tpp ] and [ th : lftpp ] can be used to generate reactive trajectories and no - detour reactive trajectories , which can then be analyzed using a variety of statistical tools to characterize the mechanism of the reaction . how to do so in practice will be illustrated on the example of @xmath0 in sec . [ sec : lj38 ] . particularly useful is to quantify how these trajectories go through specific cuts in the network , as we explain in sec . [ sec : cutstubes ] . under assumption ( b ) , the generator @xmath86 is irreducible because @xmath28 is . to prove the assertions of the proposition , we will verify that the invariant distribution of the transition path process is given by @xmath87 so that the average number of transitions per unit time it makes between any pair of states , that is , @xmath88 , is @xmath89 to show that is the invariant distribution of the transition path process , we consider two cases : @xmath76 and @xmath90 . for @xmath76 we have @xmath91 using the detailed balance condition , @xmath92 , a few terms cancel out and we are left with @xmath93 where we used @xmath94 if @xmath62 and @xmath95 if @xmath96 , and the last equality follows from the definition of the committor . for @xmath90 we have @xmath97 which terminates the proof . note that if assumption ( a ) does not hold , then we also need to account for the direct jumps from @xmath25 to @xmath26 in the original mjp as additional visits into state @xmath70 . if assumption ( b ) does not hold , we can fatten the states @xmath25 and @xmath26 to include all the nodes @xmath9 such that @xmath94 and @xmath98 , respectively . with this modification , the proposition is valid . the fact that the process has no loops follows directly from the form of its generator in particular the network defined by @xmath99 , @xmath100 , has no loops except for the ones through @xmath70 . the proof of the rest of the statement is similar to that of proposition [ th : tpp ] : under assumptions ( b ) and ( c ) , the generator @xmath99 is irreducible because @xmath28 is and we will show that the invariant distribution in the network with the generator @xmath101 in is equal to @xmath102 so that the average number of transitions per unit time it makes between any pair of states , that is , @xmath103 , is @xmath104 this will imply that the transition path process and the no - detour transition path process have the same stationary current , as claimed in the proposition . to show that is the invariant distribution , we consider again two cases : @xmath76 and @xmath90 . if @xmath76 we have @xmath105 using the detailed balance condition , @xmath92 , and the fact that @xmath106 we obtain @xmath107 where we used @xmath94 if @xmath62 and @xmath95 if @xmath96 , and the last equality follows from the definition of the committor . for @xmath90 we have @xmath108 which ends the proof . to remove assumption ( a ) , we need to account for the direct jumps from @xmath25 to @xmath26 in the original mjp as additional visits into state @xmath70 . to remove assumption ( b ) , we can fatten the states @xmath25 and @xmath26 to include into them all the nodes @xmath9 such that @xmath94 and @xmath98 , respectively . and to remove assumption ( c ) , we can restrict the statement of the proposition to the unique ergodic component of the chain with generator composed of all the states in @xmath75 that can be reached starting from @xmath25 . recall that a cut in a network @xmath109 is a partition of the nodes in @xmath4 into two disjoint subsets that are joint by at least one edge in @xmath13 . the set of edges whose endpoints are in different subsets of the partition is referred to as the cut - set . here we will focus on @xmath25-@xmath26-cuts that are such that @xmath25 and @xmath26 are on different sides of the cut - set . any @xmath25-@xmath26-cut leads to the decomposition @xmath110 such that @xmath111 and @xmath112 ( see fig . [ fig : cut ] ) . -@xmath26-cut between the sets @xmath25 and @xmath26 whose nodes are shown in blue and green respectively . the edges of the cut - set are shown with dashed lines.,scaledwidth=50.0% ] we can use cuts to characterize the width of the transition tube carrying the current of reactive trajectories . a specific set of cuts is convenient for this purpose , namely the family of isocommittor cuts which are such that their cut - set @xmath33 is given by @xmath113 the isocommittor cuts are the counterparts of the isocommittor surfaces in the continuous case . these cuts are special because if @xmath114 and @xmath115 , the reactive current between these nodes is nonnegative , @xmath116 , which also means that every no - detour transition path contains exactly one edge belonging to an isocommittor cut since the committor increases monotonically along these transition paths . therefore , we can sort the edges in the isocommittor cut @xmath117 according to the reactive current they carry , in descending order , and find the minimal number of edges @xmath118 carrying at least @xmath119% of this current . by doing so for each value of the committor @xmath120 and for different values of the percentage @xmath121 , one can then analyze the geometry of the transition channel - how broad is it , how many sub - channels there are , etc . the result of this procedure will also be illustrated on the example of @xmath0 in sec . [ sec : lj38 ] . finally note that the reaction rate can be expressed as the total current through any cut ( not necessarily an isocommittor cut ) as ( compare ) @xmath122 the proof of this statement is elementary and will be omitted . in this section , we briefly discuss the case of metastable networks . we start by giving a spectral definition of metastability , then discuss the connections of our results to the potential theoretic approach to metastability and to large deviation theory . metastable networks and mjps have been the subject of many studies ( e.g. @xcite ) . by definition , they are such that the spectrum of their generator contains one or more groups of low - lying eigenvalues . let us assume without loss of generality that @xmath123 or @xmath124 and denote by @xmath125 the solutions of the eigenvalue equation @xmath126 then the detailed balance condition implies that the eigenvalues are real , non - negative , and can be ordered as @xmath127 there is a low - lying group of eigenvalues if there exists a @xmath128 and an @xmath129 such that @xmath130 to see that this condition implies metastability , notice that the spectral decomposition of the generator , @xmath131 leads to the following expression for the transition probability distribution @xmath132 to find the walker at state @xmath10 at time @xmath133 if it was at @xmath9 initially : @xmath134 if holds , it means that on time - scales such that @xmath135 in @xmath136 , we have @xmath137 , and up to errors that are exponentially small in @xmath138 , the sum in can effectively be truncated at @xmath139 : @xmath140 in other words , on these time scales the fast processes described by the eigenvalues of index @xmath141 and above have already relaxed to equilibrium and what remains are the slow processes associated with the eigenvalues of index @xmath139 and below . this also means that the dynamics on these time scales can effectively be reduced to a markov jump processes on a state space with @xmath141 states . note that the spectral decomposition in also leads to a spectral decomposition for the current : @xmath142 where the eigencurrent associated with the pair @xmath143 is @xmath144 the eigencurrent should be compared to : as can be seen , can be obtained from by substituting the eigenvector @xmath145 for the committor @xmath40 . this suggests that if @xmath146 and corresponds to an eigencurrent associated with a slow process in the low - lying group , then it should be possible to find sets @xmath25 and @xmath26 such that the current of reactive trajectories @xmath147 between these two sets approximates @xmath148 . this is indeed the case , and this observation is at the heart of the potential theoretic approach to metastability developed by bovier and collaborators @xcite . in a nutshell , this approach says that , up to shifting and scaling , any low lying eigenvector @xmath149 can be approximated by the committor function for the reaction between two suitably chosen sets @xmath25 and @xmath26 . this observation is useful for analysis because it permits to focus on a specific eigenfunction / eigenvalue pair by studying the variational problem that the committor satisfies , that is , by minimizing the dirichlet form associated with the generator @xmath28 : @xmath150 over all @xmath151 subject to the boundary conditions that @xmath152 if @xmath62 and @xmath153 if @xmath96 . the minimizer of is the committor function and , by , its minimum is also the reaction rate @xmath56 . the discussion above makes a ( brief ) connection between the potential theoretic approach to metastability and tpt . in fact , tpt gives a way to reinterpret the various objects used in the potential theoretic approach in terms of exact statistical descriptors of the reactive trajectories . this reinterpretation is interesting because tpt applies regardless on whether the system is metastable or not . in other words , all of the formulas given in secs . [ sec : react ] and [ sec : current ] are exact no matter what the sets @xmath25 and @xmath26 are . this has the advantage that we can use the tools of tpt to analyze reactions even in situations where does not necessarily hold . more generally , our emphasis is different : we are mainly interested in using tpt to compute numerically the pathways for a reaction of interest between sets that are known before hand , rather than estimating analytically the low lying part of the spectrum . indeed , while this second goal rapidly becomes out of reach in practice for complex systems ( and typically require to make specific assumptions about the network like e.g. the ones discussed in sec . [ sec : ldt ] below ) , the first one remains achievable in a much broader class of situations , as will be illustrated in sec . [ sec : lj38 ] on the specific example of @xmath0 . another question of interest is when does condition applies ? one such situation occurs when the state - space is finite , @xmath154 , and the pairwise rates @xmath155 are logarithmically equivalent to @xmath156 in the limit as @xmath157 . the asymptotic properties of the eigenvalues in such systems , not necessarily with detailed - balance , was first established by a. wentzell @xcite using the tools from large deviation theory ( ldt ) developed in @xcite and summarized in @xcite ( see also @xcite ) . here we will focus on a sub - case of the one investigated by wentzell which is relevant in the context of @xmath0 , namely , when the generator of the mjp is of the form @xmath158 where @xmath159 , @xmath160 , @xmath161 and @xmath162 are parameters . the generator corresponds to a dynamics on the network where every node @xmath30 has an energy @xmath162 associated with it , and jumps between adjacent nodes on the network follow arrhenius law , with a rate depending exponentially on the energy barrier @xmath163 to hop from @xmath9 to @xmath10 : the information about the network topology is embedded in the energies by setting @xmath164 if @xmath9 and @xmath10 are not adjacent on the network , i.e if @xmath165 . the parameter @xmath166 plays the role of the temperature , and @xmath167 is a prefactor which we will assume temperature - independent . the generator satisfies the detailed balance condition with respect to be boltzmann - gibbs equilibrium probability distribution @xmath168 , @xmath10 , ... are the indices of these minima , there is an edge between any pair @xmath14 if there is a mep with a single saddle point along it connecting @xmath9 and @xmath10 . by using the energy of the saddle point @xmath169 as cost for the edge @xmath14 , one can find the minimal spanning tree of the network ( solid edges in the top right panel ) using e.g. kruskal algorithm , and thereby obtain its disconnectivity graph ( bottom right ) . on this disconnectivity graph , the pairs of numbers at the branching points indicate which of the nodes in the corrsponding bottom parts of the tree connect at that level of energy . using the dijkstra - based algorithm proposed in @xcite we can also calculated the minmax path connecting two states , for example between states 1 and 7 ( solid red path in the bottom left panel ) . this minmax path is relevant in regimes where ldt applies . , scaledwidth=100.0% ] in the set - up above , we can use the temperature @xmath166 as control parameter , in such a way that holds when @xmath157 . in that limit , for reasons that will become clear below , in general there are as many low - lying groups of eigenvalues as there are states ( i.e. @xmath170 as @xmath157 for all @xmath171 ) , and wentzell s approach provides a way to estimate each of these eigenvalues . to see how , it is convenient to organize the states of the chain on a disconnectivity graph , that is , a downward facing tree in which each node @xmath30 lies at the end of a branch at a depth equal to its energy @xmath162 , and branches in the tree are connected at the lowest energy barrier @xmath172 that connect all the nodes on one side of the tree to those on the other side a cartoon example is shown in fig . [ fig:7well ] : because this will be relevant in our analysis of @xmath0 , in this example we start from a continuous energy landscape that we convert into a network whose disconnectivity graph is then obtained from its minimal spanning tree ( bottom right , all solid edges ) calculated using e.g. kruskal s algorithm ( see e.g. @xcite ) . the eigenvalues can then be estimated recursively from the disconnectivity graph as follows : start by identifying the lowest barrier in the tree , i.e. the adjacent pair @xmath173 on the tree such that @xmath174 is minimum over all @xmath175 . the node @xmath176 identifies the well that the system can escape by crossing the barrier of minimum height , and the largest eigenvalue in the system corresponds to the inverse of the time scale of this escape , i.e. it can then be estimated as @xmath177 where the symbol @xmath178 means that the ratio of the logarithms of both sides of this equality tends to 1 as @xmath157 . now remove the node @xmath176 and its branch from the tree , and repeat the construction : that is , in the new tree find the pair @xmath179 such that @xmath180 is minimum over all @xmath181 , to obtain an estimate for the next largest eigenvalue , @xmath182 . by iterating upon this procedure , in @xmath183 steps we can then estimate @xmath184 for @xmath185 as @xmath186 intuitively , this procedure corresponds to lumping together the states that can be be reached on timescales of order @xmath187 or below , and analyzing what happens on the next timescale to get @xmath188 . after @xmath183 steps in the procedure we end up with a degenerate tree made of a single node lying at the very bottom of the original tree ( and of course we already know that @xmath189 ) . note that in the discussion above , we assumed that the barriers @xmath190 identified along the way are all different ( that is , strictly increasing with @xmath191 ) , which is the generic case and leads to eigenvalues that are all well - separated : if some of these barriers are equal , it means that some of the eigenvalues are asymptotically equivalent , and this case can be treated as well by generalizing the construction above . note also that estimates more precise than and can be obtained using the potential theoretic / tpt approach : in the present situation , at any stage in the iteration procedure , the states @xmath192 and @xmath193 are those that should be set as @xmath25 and @xmath26 , respectively . another interesting construction provided by ldt is the decomposition of the stochastic network into freidlin s cycles @xcite . for systems satisfying the detailed balance condition and with a rate matrix as in , the decomposition into cycles simplifies , as was recently discussed in @xcite . here we summarize this discussion and refer the interested reader to the original paper for details . in a nutshell , the decomposition into cycles focuses on which states are most likely to be reached from a given state : in the zero temperature limit , if the system is in state @xmath9 , with probability one it will reach next the state connected to @xmath9 by the smallest barrier , i.e. @xmath194 searching consecutively for the next most likely state defines a dynamics on the network that generically ends with cycles made of two states : each of these cycles contain a local minimum of energy on the disconnectivity graph ( that is , a state at the bottom of a group of branches on the tree ) , and the state connected to this minimum by the lowest barrier . these cycles are called 1-cycles by freidlin . once we have identified them , we can remove from the tree the state with highest energy in each of these 1-cycles , and repeat the construction iteratively . these gives 2-cycles , 3-cycles , etc . until we again end up with a tree with only 2 nodes on it . in this construction , we can also keep the information about the state in the original network by which any @xmath195-cycle is exited : with probability 1 as @xmath157 , this is the state whose barrier is the lowest to escape all the states contained in this @xmath195-cycle . a corollary of the fact that cycles are exited in a predictable way is that , between any two nodes in the network taken as sets @xmath25 and @xmath26 , there exists a single path on the network that concentrates all the current of the reactive trajectories as @xmath157 . this path has a minmax property : the maximal barrier separating every pair of states @xmath9 and @xmath10 on the path is minimal among the maximal barriers along all paths in the network connecting @xmath9 and @xmath10 ( see fig . [ fig:7well ] for an illustration ) . in @xcite , the construction of the hierarchy of freidlin s cycles was performed via a sequence of conversions of rate matrices into jump matrices followed by taking limits @xmath196 . relying on the properties of the hierarchy of cycles specific for the systems with detailed balance , an efficient dijkstra - based algorithm was also proposed for computing the minmax path . importantly , this algorithm did not built the whole hierarchy of cycles , but only computed the sub - hierarchy relevant to the transition process of interest , and did not require any pre - processing of the stochastic network . we conclude this section on ldt with a remark . as explained above , the ldt picture applies in the limit when @xmath157 , in which case the hierarchy of different barriers in the disconnectivity graph corresponds to timescales that become infinitely far apart as @xmath157 . while this picture is indeed correct at extremely low temperature , we do not expect it to remain valid as the temperature is increased , even if the system does remain strongly metastable ( i.e. such that some low - lying groups of eigenvalue do persist ) . rather , we expect that the transition channel will rapidly broaden if the networks is large , and that the mechanism of the reaction wil depart from that predicted by ldt . our analysis of @xmath0 by tpt will indeed confirm this picture . a lennard - jones cluster is made of particles ( or atoms ) interacting via the lennard - jones pairwise potential given by @xmath197.\ ] ] here @xmath198 denotes the positions of the @xmath199 particles in the cluster , @xmath200 is the distance between particles @xmath9 and @xmath10 , and @xmath201 and @xmath202 are parameters measuring respectively the strength and range of the interactions . at the most fundamental level , the finite - temperature dynamics of the cluster can be modeled as a continuous diffusion over the potential . this dynamics is extremely complicated owing to the multiscale nature of this potential which , when @xmath199 is large ( e.g. @xmath203 as we will consider below ) possesses an enormous number of local minima separated by a hierarchy of barriers of various height . a few thermodynamic properties of these clusters are known , however . first , it is known that the majority of global potential energy minima for lennard - jones clusters of various sizes involve an icosahedral packing @xcite . however , lennard - jones clusters with special numbers of atoms admit a high symmetry configuration based on a face - centered cubic packing , with a lower energy . the smallest cluster with this property contains @xmath203 atoms@xcite . the global potential energy minimum of the @xmath0 cluster is achieved by a truncated octahedron with the point group @xmath2 ( fig . [ fcc_ico ] ( a ) ) , which from now on we will simply refer to as fcc . the second lowest minimum is the icosahedral structure with the @xmath3 point group ( fig . [ fcc_ico ] ( b ) ) , which we will refer to as ico . it is also known that the basin around ico is much wider than that around fcc these two basins are usually referred to as funnels in the literature . this has thermodynamic consequences when the temperature of the system is non - zero . indeed , the fcc basin only remains the preferred basin for @xmath204 with @xmath205 ( here @xmath206 denotes boltzmann constant ) . at @xmath207 , the system undergoes a solid - solid phase transition where the ico basin becomes more likely due to its with greater configurational entropy ( see e.g. fig . 4 in @xcite ) . next , at @xmath208 , the outer layer of the cluster melts , while the core remains solid . then the cluster completely melts at @xmath209 @xcite . the difference of widths of the two basins also has dynamical consequences . indeed , due to its larger width , the ico basin is the one that is most likely to be reached by the system after crystallization even if @xmath210 . the question then becomes how does the system reorganize itself to get out of the dynamical trap around ico and in its preferred state around fcc ? it is also of interest to understand how this process is influenced by the temperature , since the rearrangement pathway is likely to be influenced by it . these are the type of questions that we will address in this section , as an illustration of the tpt - based network analysis tools presented earlier . this study is complementary to those conducted by wales and collaborators in the same context @xcite using different tools @xcite . the problem of rearrangement of @xmath0 has been the object of much studies in the past 15 years ( see e.g. @xcite ) . an interesting approach to the problem has been proposed by david wales and collaborators , who undertook an ambitious program aiming at mapping the evolution of @xmath0 onto a network / mjp and reducing the analysis of the dynamics of @xmath0 to the study of this network . while this mapping is technically hard to perform in practice and required a lot of inventiveness , it is conceptually quite simple to understand . if the temperature of the system is small enough , it will spend a long time near the bottom of the energy well around the local minima it is currently in before a thermal fluctuation large enough will manage to push it above an energy barrier separating it from an adjacent well . the system will then fall near the bottom of this adjacent well and the process will repeat . in this regime , the dynamics can be reduced to a basin hoping : the local minima of the energy become the nodes on the network , two such nodes are connected by an edge if the system can transit from one minimum to the another by crossing a single barrier , and the rate / weight of the directed edge from one node to another involves ( via arhennius formula ) the height of the energy barrier(s ) that must be crossed to perform this transition this construction was illustrated on a toy example in fig . [ fig:7well ] . an additional simplification made in the case of @xmath0 is to lump together all the minima and saddle point that are equivalent by symmetry ( point group , permutation , etc . ) . all together this construction led to a network for @xmath0 that contains a single connected component with 71887 nodes associated with the lowest local minima on the landscape ( which include fcc and ico ) , and 119853 edges this information is publicly available from the @xmath0 database wales s website @xcite . the database also contains the information about the generator , whose off - diagonal entries are in a form consistent with @xcite @xmath211 here @xmath212 is proportional to the inverse of the system s temperature @xmath213 , @xmath214 , @xmath162 , and @xmath215 are , respectively , the point group order , the value of the potential energy , and the geometric mean vibrational frequency for the local minimum associated with node @xmath9 , @xmath216 , @xmath217 and @xmath218 are the same numbers for the transition state @xmath191 connecting the local minima @xmath9 and @xmath10 ( there may be more than one of them for every pair @xmath14 adjacent on the network ) , and @xmath219 is the number of vibrational degrees of freedom . as in , if there is no minimum energy path connecting the minima with index @xmath9 and @xmath10 via a single saddle point , we set @xmath220 and @xmath221 . note that , by construction , the generator defined by satisfies detailed - balance with respect to the following boltzmann - gibbs equilibrium distribution : @xmath222 the network representation of @xmath0 via will be our starting point here . the majority of local minima / nodes listed in wales database do not have special names for example , fcc and ico are simply listed 1st and 7th , respectively . except for these two , in the sequel we will simply refer to the other minima by their indices in the database . we also work in reduced units in which the temperature is measured in units of @xmath223 . since we are interested in the mechanism of rearrangement between ico and fcc , we take the nodes of these two states as sets @xmath25 and @xmath26 , respectively . we also checked that our results do not change significantly if we fatten these states by including in them the nodes that are in the connected component around them where all the nodes have energy within @xmath224 of that of fcc and ico , respectively . a key preliminary step in the application of tpt to @xmath0 is the calculation of the committor function . this calculation requires solving which , in the present case , is a system of @xmath225 linear equations with the same number of unknowns . the detailed balance property allows us to make the matrix in symmetric by multiplying each row by @xmath226 . the resulting system can then be solved using the conjugate gradient method with the incomplete cholesky preconditioning ( see e.g. @xcite ) . this works for @xmath227 . for lower values of the temperature , the scale separation between the possible values of @xmath228 for different @xmath9 becomes too large for the computer arithmetics . in order to overcome this difficulty we truncate the @xmath0 network by keeping only the nodes whose energy is below a given cap this is legitimate because , the lower the temperature , the least likely it is to observe a reactive trajectory venturing at energies much higher than @xmath224 above that of the overall barrier between ico and fcc . for each value of temperature we set this cap as high as possible while keeping the system nonsingular in the computer arithmetics . the energy caps and resulting network sizes for the different values of temperature that we considered are listed in table [ table1 ] . the values in parentheses are the difference between the caping energy and that of fcc , @xmath229 @xcite . all in all , we computed the committor for temperatures ranging from @xmath230 to @xmath231 using steps of @xmath232 . the disconnectivity graphs of the network we used at three different temperatures , @xmath233 , @xmath234 , and @xmath235 are shown in fig . [ fig : dgraph ] . on these figures , we only included the nodes through which at @xmath236 of the total current of reactive trajectories goes and we colored the branches of the graph according to the value of the committor of the nodes at the end of these branches . as can be seen , as the temperature increases , the committor function becomes less step - like , and a higher number of nodes gets values than are in between the extreme 0 and 1 . .energy caps and network sizes used for different value of the temperature . [ cols="^,^,^",options="header " , ] [ table1 ] ( left ) , @xmath234 ( center ) , @xmath235 ( right ) . each disconnectivity graph includes only those local minima through which at least 1% of the reaction pathways from ico to fcc pass.,scaledwidth=100.0% ] ico and @xmath237 fcc at different temperatures . these rates display an almost perfect arrhenius - like behavior in this temperature range , even though the mechanism of the rearrangement becomes increasingly complex as the temperature increases . the zoom shown in the inset shows that a cross - over between @xmath238 and @xmath239 occurs ar @xmath240 ( i.e. @xmath241 ) : this is the temperature above which ico becomes more favorable than fcc due to entropic effects related to the relative widths of the funnels around these two structures.,scaledwidth=100.0% ] . the edges shown carry at least 10% of the total reactive flux from ico to fcc the thickness of the arrow is proportional to the precentage of current the edge carries , and the actual percentage is also displayed next to it . the values of the committor at the nodes are show in greyscale , with the explicit values of @xmath242 given for some of them . the blue arrows show the minmax path from ldt : at this low temperature , most of the current goes along this path . the highest barrier crossed along the minmax path is between nodes @xmath243 and @xmath244 ( @xmath245 ) . also show in inset is the energy profile along the minmax path.,title="fig:",scaledwidth=100.0% ] at @xmath234 . at this higher temperature , most of the edges carry less than @xmath246 of the total current : in particular , we can no longer go from ico to fcc by following edges that carry at least @xmath246 of the current . this also implies that the minmax path from ldt is no longer relevant to explain the mechanism of the rearrangement at this temperature the edges along this path that carry more than @xmath246 of the current are still shown in blue . the edges between nodes 8 and 3223 and nodes 3223 and 354 carry less than @xmath247 of the current : we show them because these edges belong to the dominant representative path introduced , i.e. the path maximizes the current it carries . this path is different from the minmax path but , as can be seen in this example , it is not relevant either in situations where the transition channel becomes spread out . , title="fig : " ] . as the temperature increases , the no - detour paths tend to cross higher barriers . , scaledwidth=100.0% ] is plotted against the index of these edges ordered by this magnitude . the inset shows the empirical cumulative distribution function of the current through the edges in the cut . the number of edges that must be included to account for a given percentage of the total current increases rapidly with temperature , indicative of the broadening of the reaction channel for the rearrangement.,scaledwidth=100.0% ] at @xmath248 . the way this representation was constructed is explained in text.,scaledwidth=100.0% ] at @xmath249 . , scaledwidth=100.0% ] at @xmath234.,scaledwidth=100.0% ] at @xmath235.,scaledwidth=100.0% ] once the committor function has been calculated , we can use tpt to calculate the rate of rearrangement of @xmath0 and characterize its mechanism . using formulae with @xmath250 ico and @xmath251 fcc , we obtain the rates at which the system rearranges itself between these two states . these rates are shown in fig . [ fig : rate ] as a function of the inverse temperature @xmath252 . as can be seen , both rates are almost perfectly straight on a log - linear scale , and can be fitted by @xmath253 since the energy barriers between fcc and ico and ico and fcc are 4.219 and 3.543 , respectively , @xcite , these fits are consistent with arrhenius law . the fits in also compare well with the ones calculated in @xcite for the temperature range @xmath254 : @xmath255 and @xmath256 . note also that the rates cross at the value @xmath257 ( i.e. @xmath258 ) : this temperature is the one above which tpt predicts that ico becomes preferred over fcc , which is slightly higher than the value @xmath259 listed in sec . [ sec : thermo ] . this crossover is due to entropic effects related to the relative widths of the funnels around ico and fcc . the arrhenius - like nature of the rates may suggest that the mechanism of rearrangement of the @xmath0 cluster is quite simple , and dominated at all the temperatures that we considered by the hoping over the lowest saddle point separating ico and fcc . this impression , however , is deceptive . to see why , in figs . [ fig : cartoon1 ] and [ fig : cartoon2 ] let us compare cartoon representations of the current of reactive trajectories given in at two different temperatures , @xmath260 and @xmath234 . the way these representations were constructed is by plotting all the nodes in the network such that the current of reactive trajectories along the edges between them carry at least 10@xmath261 of the total current , and connecting these nodes by an arrow whose thickness is proportional to the magnitude of the current . as can be seen in fig . [ fig : cartoon1 ] , at @xmath260 , most of the current concentrate on a single path : this path coincides with the minmax path between ico and fcc predicted by ldt @xcite . at the higher temperature of @xmath234 , however , we see that this minmax path becomes mostly irrelevant , and in fact we can no longer go from ico to fcc following edges that carry at least 10@xmath261 of the current . the reason is that the current becomes very spread out among the edges of the network , indicative that the tube carrying most of the current of reactive trajectories also becomes quite wide . to quantify further this observation , we used proposition [ th : lftpp ] to generate @xmath262 samples of the no - detour transition path process at every temperature . ( in the present example , it turns out that the network is so complex that the reactive trajectories themselves , which we can in principle generate via proposition [ th : tpp ] , are too long to be sampled efficiently . this arises because these trajectories wander too often into quasi - deadends or in between intermediate structures , and this is why we focused on no - detour transition paths which are much shorter and can be generated in great number . ) we used this sample of no - detour transition paths to first analyze the height of the highest energy barrier along these paths measured with respect to @xmath263 . the empirical cumulative distribution functions of these barrier heights are shown in fig . [ fig : barriers ] . as can be seen , at the low temperature of @xmath260 , this distribution is very peaked around the value @xmath264 , which is the height of the lowest saddle point separating ico and fcc . at higher temperatures , however , this distribution broadens significantly , indicative that higher barriers become frequently crossed by the no - detour transition paths . this is an entropic effect : in essence , we can think of the height of the barrier in terms of ` bonds ' between the lennard - jones particles that need to be broken for the rearrangement to proceed . what our results show is that the number of no - detour paths increases very rapidly with the maximal number of bonds that are ever broken along them . at low temperature , the rearrangement proceed mostly by no - detour paths along which no more than about 4 bounds are broken , because these paths are energetically favorable . at higher temperature , however , no - detour paths along which 5 , 6 or even 7 bonds break start to matter : even though they are less favorable energetically , their sheer number means that they eventually carry more current globally . a consequence of this effect is that the width of the reaction channel also broadens significantly with temperature . this is quantified in fig . [ fig : fluxincut ] , where we analyze the current along the edges in the isocommittor cut @xmath265 . by ordering these edges by the magnitude of the current they carry , and plotting this current magnitude as a function of the edge index , we arrive at the plots on the main panel of fig . [ fig : fluxincut ] . as can be seen , as the temperature increases , these plots widen with temperature , and display a power law behavior for a range of edge indices . the inset of fig . [ fig : fluxincut ] shows the cumulative distribution of the current through the edges in the isocommittor cut @xmath265 , and show that the higher the temperature , the more edges need to be included to get a significant percentage of the total current : for example , at @xmath231 , thousands of edges in the cut ( that is , most of them ) need to be included in order to account for @xmath266 of the current . the mechanism of rearrangement thus departs significantly from the one predicted by ldt , even though the rates remain arrhenius - like even at this high temperature . we tried to capture visually the complexity of the mechanism of rearrangement using the representation of the network of current of reactive trajectories shown in figs . [ fig : net06][fig : net15 ] . these figures were constructed as follows . we plotted every node of the network through which at least @xmath267 of the total current went . we ordered these nodes along the @xmath268-axis according to the cumulative distribution function of their committor , using a coloring from blue to green to indicate their actual committor value . along the @xmath269-axis , we ordered the nodes according to the inverse of the magnitude of current of reactive trajectory , , they carry ( the higher the node , the least current it carries ) and we connected the nodes by lines whose darkness is proportional to the magnitude of the current between them . we also faded the color as this magnitude decreased . finally , we used dots of different sizes to represent the nodes : the bigger the node , the larger is the magnitude of the average number of transitions per unit time that the reactive trajectories make through this node , see . this is a way to try to capture deadends and dynamical traps on the network , i.e. node that the reactive trajectories visit often but through which little current of these reactive trajectories go . in the figures these deadends are nodes that are high and big . overall , what these figures confirm is that , as the temperature increases , the curent of reactive trajectories spreads more and more on the network , and the reaction channel broadens . it also confirms that there exists many deadends and dynamical traps on the network . this last aspect makes tpt particularly suitable to analyze the mechanism of rearrangement : indeed , a spectral analysis of the network along the lines discussed in sec . [ sec : spectral ] is both hard to perform in the present situation and uninformative because it is too global . we have presented a set of analytical and computational tools based on tpt to analyze flows on complex networks / mjps . we expect these tools to be useful in a wide variety of contexts . the network representation of @xmath0 that we used here as illustration is just a specific example of markov state model ( msm ) used to map a complex dynamical system onto a mjp ( see e.g. @xcite ) . during the last decade , such msms have emerged as a way to analyze timeseries data generated e.g. by molecular dynamics simulations of macromolecules , general circulation models of the atmosphere / ocean system , etc . in these contexts , massively parallel simulations , special - purpose supercomputers , and high - performance graphic processing units ( gpus ) permit to generate time series data in amounts too large to be grasped by traditional `` look and see '' techniques . msms provide a way to analyze these data by partitioning the conformation space of the molecular system into discrete substates , and reducing the original kinetics of the system to markov jumps between these states in other words , by interpreting the timeseries as some dynamics on a network , with the states in the msms playing the role of the nodes on the network , and the transition rates between these states being the weights of the directed edges between these nodes . while msms typically provide an enormous simplification of the original timeseries data , the associated networks are typically quite complex themselves , with many nodes , a nontrivial topology of edges between them , and rates / weights on these edges that can span a wide range of scales . the tools that we derived from tpt can be used for the nontrivial task of analyzing these networks / msms . more generally , we expect the tools developed in this paper to be useful to analyze and interpret other networks that have emerged in many areas as a way to represent complex data sets . we thank prof . david wales for providing us with the data of the @xmath0 network and miranda holmes for interesting discussions . m. c. held an sloan research fellowship and was supported in part by darpa yfa grant n66001 - 12 - 1 - 4220 , and nsf grant 1217118 . e. v .- e . was supported in part by nsf grant dms07 - 08140 and onr grant n00014 - 11 - 1 - 0345 . bowman , g. r. , pande , v. s. , and no , f. eds . : an introduction to markov state models and their application to long timescale molecular simulation . advances in experimental medicine and biology , 797 . springer ( 2014 ) . den hollander , f. : three lectures on metastability under stochastic dynamics . in methods of contemporary mathematical statistical physics ( r. kotecky , ed . ) . lecture notes in math . springer , berlin . ( 2009 ) . vanden - eijnden e.:transition path theory . in _ computer simulations in condensed matter : from materials to chemical biology . m ferrario , g ciccotti , k binder , pages 439-78 . springer , berlin ( 2006 ) .	a set of analytical and computational tools based on transition path theory ( tpt ) is proposed to analyze flows in complex networks . specifically , tpt is used to study the statistical properties of the reactive trajectories by which transitions occur between specific groups of nodes on the network . sampling tools are built upon the outputs of tpt that allow to generate these reactive trajectories directly , or even transition paths that travel from one group of nodes to the other without making any detour and carry the same probability current as the reactive trajectories . these objects permit to characterize the mechanism of the transitions , for example by quantifying the width of the tubes by which these transitions occur , the location and distribution of their dynamical bottlenecks , etc . these tools are applied to a network modeling the dynamics of the lennard - jones cluster with 38 atoms ( @xmath0 ) and used to understand the mechanism by which this cluster rearranges itself between its two most likely states at various temperatures .
since purcell predicted spontaneous emission rate could be changed by electromagnetic environment in 1946 @xcite , the effect of electromagnetic field on radiation properties has been thoroughly investigated barton , agarwala121475,kleppner , john , yokoyama , knoll , wang , yoshie , thon , and classified in the category of cavity quantum electrodynamics ( qed ) . characteristics of qed have been greatly investigated both theoretically and experimentally @xcite , and many kinds of devices @xcite based on this theory have been developed . concepts such as enhanced and inhibited spontaneous emission kleppner , john , reversible spontaneous emission @xcite , photon blockade @xcite , micromasers @xcite , low threshold lasers @xcite , etc . , have become very familiar . dipole - dipole interaction could also be greatly modulated by the electromagnetic environment . photon emitted by one dipole could be absorbed by the other or vice versa . the strength of interaction is decided by photon emission , transmission and absorption . many different kinds of electromagnetic environment can be used to control or change these characteristics , such as vacuum agarwala121475,lehmberga2883,agarwal , hettich , puhan , optical cavities goldstein , kobayashi , lehmberga2889,gsagarwal , optical lens @xcite , dielectric droplet @xcite , photonic materials bay , shughesdd , gallardo , laucht , reitzenstein , weiler , kim , metal surface metasurface1,metasurface2,metasurface3,metasurface4,metasurface5 , metamaterial @xcite and so on . for example , optical lens and waveguide have been designed to collect the emission photon and transfer it to the other dipole . optical cavity or metal surface can enhance the emission or absorption rate roughly by the ratio of quality factor @xmath0 and the mode voluum @xmath1 . photonic crystal nanocavity is one of the promising platform to investigate dipole - dipole interaction , because local coupling strength of dipole and photon can be tailored and integrated to photonic crystal waveguide is extremely convenient . high quality factor @xmath2 and small mode volume @xmath3 have been realized for photonic crystal cavity @xcite . numerical investigations show that ultra - high quality factor with @xmath4 can be designed through finely tuning the scatters around the cavity with little change of the mode volume snodaqt . furthermore , static and ultra fast dynamic control of the cavity frequency and the quality factor have been achieved vignolini , mosor , snodasq , snodadq , tanabedf . on the other hand , temperature @xcite , strainin @xcite , electric field @xcite , magnetic field @xcite are much suitable for fine tuning the energe levels of dipoles located at certain position in solid system . recent studies show that this kind of interaction could be used to implement quantum entanglement preparation and quantum information processing lukin0,lukin1,lukin2,isenhower , zhengsb , shughesl05,shughesoe , gonzalez - tudela , cooperative radiation @xcite , frster energe transfer @xcite , dipole nanolaser @xcite , and so on . furthermore , some novel quantum phenomenareinhard , johnson , saquet , harlander , miranda has been found . all of these applications and phenomena are related to the cooperative decay parameters or dipole - dipole interaction potential . according to eq . ( [ 6c ] ) , dipole - dipole interaction potential can be got through the cooperative decay parameters . in the previous theoretical studies , mode - expansion method goldstein , kobayashi , agarwal , bay , rist , arnold or green function method schmid , gsagarwal , shughesl05,shughesoe , gonzalez - tudela is often adopted to investigate this cooperative decay parameters . these two methods work well for electromagnetic environment with perfect boundary condition . because of the extremely complexity of finding eigen - mode , mode - expansion method can be used only for simple case such as vacuum or planar cavity . besides , the exact analytic green function is also hard to be obtained for complex electromagnetic environments . numerical method is necessary for studying this kind of interaction in photonic crystal nanostructure . in this paper , we put forward a simple numerical method to investigate dipole - dipole interaction in photonic crystal nanocavity through finite - difference time domain algorithm ( fdtd ) . we calculate the collective and individual radiation rates of classical dipoles by directly solving maxwell s equations in real space with a free - space boundary condition . by using the result of two dipoles radiation rate minus the sum of the two individual radiation rates , we get the cooperative decay parameters and dipole - dipole interaction potential . a similar method has been used for local density of photonic states calculation @xcite . it is direct for structures with arbitrary shape and physical quantities such as radiation power at different frequencies can be got in a single simulation run . numerical results in vacuum and planar conductor cavity show that our method works well for this kind of investigation . for dipoles located in photonic crystal cavity , the strength of dipole - dipole interaction depends heavily on the following four factors : the atomic position , the transition frequency , quality factor , mode volume and the cavity frequency . effect of these factors on the interaction strength are also shown . this paper is organized as follows : in sec . ii , we give the model of dipole - dipole interaction with some illuminations . numerical method is presented in sec . validation of this method is presented in sec . iva for dipoles in vacuum and in planar microcavity . in sec . ivb , applying our method to photonic crystal nanocavity , we investigate the effect of the atomic position , the transition frequency , quality factor and the cavity frequency on the dipole - dipole interaction strength . in sec . v , we give a brief conclusion . ( @xmath5 ) and the excited state @xmath6 ( @xmath7 ) . atom a is in the excited state latexmath:[$% @xmath5 . they both interact with electromagnetic field with eigen frequency @xmath9 . analogous to the process of lamb shift where both virtral and real photon exchange between the atom and the quantum electromagnetic field take effect , photom exchange between the two atoms contribute to dipole - dipole interaction . ] schematic diagram of dipole - dipole interaction is illustrated in fig . 1 . there are two two - level ` atoms ' a and b located at @xmath10 and @xmath11 respectively and they both interact with the electromagnetic field with eigen frequency @xmath12 . the atom a ( b ) has two states , the ground state @xmath13 ( @xmath5 ) and the excited state @xmath14 ( @xmath15 ) . the hamiltonian of the system in the rotation wave approach reads @xcite : @xmath16 . \label{1}\end{aligned}\ ] ] where @xmath17 and @xmath18 are , the photonic creation and annihilation operators , @xmath12 is the frequency of the eigen mode of the electromagnetic field , @xmath19 ( @xmath20 ) , is the the coupling coefficient , @xmath21 is the eigen mode , @xmath22 is the transition dipole moments of atoms @xmath23 , and @xmath24 is the vacuum permittivity . in eq . ( [ 1 ] ) , @xmath25 is the noninteracting hamiltonian , and @xmath1 represents the interaction between dipole and electromagnetic field . there are three states for the system considered : ( 1 ) atom a is in the excited state , and atom b is in the ground state , without any photon , i.e. , @xmath26 , ( 3 ) both atom a and atom b are in the ground state , with a photon of frequency @xmath12 , i.e. , @xmath27 . the initial state is prepared in @xmath28 . then , the state of the system evolves as @xmath29 where @xmath30 is the evolution operator and @xmath31 , @xmath32 , @xmath33 . to derive the equation for the atom - field dynamics nonperturbatively , the resolvent operator @xmath34 is adopted cohen - tannoudji . the corresponding advanced and retarded propagators are @xmath35 . then the evolution operator can be expressed by @xmath36\exp ( -i\omega t)/2\pi i. \label{02}\ ] ] the dynamic properties of this system are governed by @xmath37 and @xmath38 . the matrix elements of the resolvent reads @xmath39 where @xmath40 , @xmath41 , @xmath42 and @xmath43 , @xmath44 eliminating @xmath45 , we have @xmath46 @xmath47 is given by @xmath48[z-\hbar \omega _ { b}-w_{bb}]-w_{ab}w_{ba } \label{4}\ ] ] the local coupling between atom and the field ( @xmath49 ) or the dipole - dipole coupling between atom a and atom b ( @xmath50 ) are denoted by @xmath51 clearly , these terms can be written as @xmath52,\text { } \\ \gamma _ { ij}(\omega , \mathbf{r}_{i},\mathbf{r}_{j } ) & = 2\pi \sum_{n}g_{n}^{\ast } ( \mathbf{r}_{i})g_{n}(\mathbf{r}_{j})\delta ( \omega -\omega _ { n } ) , \label{6 } \\ \delta _ { ij}(\omega , \mathbf{r}_{i},\mathbf{r}_{j } ) & = \frac{1}{2\pi } \wp \int_{0}^{+\infty } dz\frac{\gamma _ { ij}(z,\mathbf{r}_{i},\mathbf{r}_{j})}{% \omega -z}. \label{6c}\end{aligned}\ ] ] to better understand the physics underlying of these equations , we make some illumination . we are able to self - consistently determine the roots of @xmath53 ( eq . ( [ 4 ] ) ) . the real parts are corresponding to the energy levels of these two atoms and the imaginary parts are the lifetime . different from the usual markov approximation method where @xmath54 @xmath55 are independent of @xmath56 and have been replaced by their approximate value @xmath57 and @xmath58 , this method is non - markov and this self - consistent process can give the rigorous dipole - dipole coupling . we can also make use of the markov approximationvalues @xmath54 @xmath55 in eq . ( [ 4 ] ) and get the same results as usual . however , for complex electromagnetic environment such as photonic crystal or photonic crystal nanocavity , @xmath54 may vary sharply around some frequency and non - markov is necessary . owing to @xmath54 , equation . ( [ 4 ] ) also shows that both the energy lever and the lifetime of the states are spliting . different coupling @xmath54 makes different energy lever spliting and different lifetime spliting . dipole blockade , which have been widely investigated in quantum information processing needs large energy gap . efficient superradiant emission , steady state entanglement preparation and fast frster energy transfer needs large lifetime spliting . inspired by these novel application , accurate values for @xmath54 are expected . furthermore , equation ( [ 5 ] ) shows that all of the photonic eigen modes contribute to the local coupling and dipole - dipole interaction and both real and virtual photon effects have been taken into account . insert @xmath59 @xmath60 into eq . ( [ 6 ] ) for @xmath61 , and define @xmath62 @xmath60 , where @xmath63 and @xmath64 are the size and the unit vector of transition dipole @xmath22 . then , the cooperative decay parameters reads : @xmath65 for simplicty , we denote @xmath66 and @xmath67 for @xmath68 and @xmath69 respectively . from equations 2.12a to 2.14a of reference @xcite , it is easy to see that if @xmath70 compose a complete set of eigen mode , @xmath71 is also a complete set of eigen mode . then @xmath72could also be written as : @xmath73so we have : @xmath74\delta ( \omega -\omega _ { n } ) . \label{8}\ ] ] once we get the cooperative decay parameters @xmath75 , the dipole - dipole interaction potential @xmath76 can be achieved through eq . ( [ 6c ] ) . here we propose a new method to rigorously get the cooperative decay parameters based on finite - difference time domain algorithm . we show that dipole - dipole interaction can be got through calculating the collective and individual radiation rates of classical dipoles . we begin with maxwell equations : @xmath77expand @xmath78 where @xmath79 is the same as the eigen modes at the quantum analysis section . if we let @xmath80 , then @xmath81 satisfies the following equation : @xmath82through eq . ( [ 12 ] ) , we get : @xmath83 if @xmath84 , the emission power is given by : @xmath85 in the two dipoles instance , i.e. , @xmath86 , @xmath87 , the total power is@xmath88 if there is only one dipole , i.e. , @xmath86 or @xmath89 , then @xmath90 combining eq . ( [ 17 ] ) , eq . ( [ 18 ] ) and eq . ( [ 19 ] ) , we difine the cooperative emission power as : @xmath91\delta ( \omega _ { 0}-\omega _ { n } ) . \label{20}\end{aligned}\ ] ] through eq . ( [ 20 ] ) and eq . ( [ 8 ] ) , we found that:@xmath92where @xmath93 and @xmath94 correspond to cooperative decay parameters and cooperative emission power of two unit classical dipoles in complex environment . @xmath95 correspond to the emission power of a unit classical dipole in vacuum . @xmath96 is the local coupling strength for an two level atom with transition dipole moment @xmath22 and transition frequency of bare atom @xmath97 in vacuum . using the result of @xcite , we get @xmath98 . in order to make the cooperative decay parameters more clearly , we define @xmath99 where @xmath100 is totally decided by the two level atom i. then we get @xmath101 equation . ( [ 22 ] ) is the main content of our method . @xmath102 can be obtained through calculating @xmath94 and @xmath103 by fdtd method . utilizing eq . ( [ 6c ] ) , we can get the dipole - dipole interaction potential @xmath67 . furthermore , we see that our method may be generalized to study many dipole interactions through this similar procedure . before applying our method to investigate dipole - dipole interaction in photonic crystal nanocavity , we first numerically validate our approach in the case for dipoles in vacuum or planar microcavity where analytic expression for the cooperative decay parameters @xmath93 can be easily got through mode - expansion method . in this section , for simplicity , we set @xmath104 and @xmath105 . in vacuum , the two transition dipole moments are parallel to the x axis and the interatomic separation axis is along the z axis . the eigen - function of vacuum reads @xmath106 . inserting this into eq . ( [ 8 ] ) and making use of @xmath107 , we get the analytic cooperative decay parameters @xmath66 : @xmath108 . \label{23}\]]where @xmath109 is the distance between the two dipoles . if @xmath110 , for the two dipole located at the same place , @xmath111 which is the radiation rate of the dipole in vacuum . we set @xmath112 and increase the interatomic separation @xmath113 to simplify our discussion . in fig . [ ssp2 ] , black line is for the analytic results ( eq . ( [ 23 ] ) ) and red boxed dots represent the calculated results ( eq . ( [ 22 ] ) ) . it can be clearly seen that @xmath66 got through our method agrees very well with the analytic results . as a function of the interatomic separation @xmath114 when @xmath115 ; black line is for the analytic results ( eq . ( [ 23 ] ) ) and red boxed dots represent the calculated results ( eq . ( [ 22 ] ) ) . the orientation of the two transition dipole moments are the same and perpendicular to their separation . ] . ( b ) cooperative decay parameters @xmath116 as a function of the interatomic separation @xmath113 when @xmath117 , @xmath118 ; black line is for the analytic results from eq . ( [ 24 ] ) and red boxed dots represent the calculated from eq . ( [ 22 ] ) . ] similar result is also got when the two point dipoles are located in a planar nanocavity . here , the cooperative decay parameters @xmath119 could also be got analytically . the structure is shown in fig . [ ssp3](a ) . the origin is at the center for x , y axis and at the bottom conductor for z axis . the cpml encloses the domain , as can be seen from the top view at the right - hand side . the orientation of the two transition dipole moments are the same and along z ( red arrowhead ) . they are arranged along x axis and their separation is @xmath113 . the analytic cooperative decay parameter @xmath66 for these two dipoles is:@xmath120}2[1-(\frac{n\lambda } { 2l})^{2}]\cos ( % \frac{n\pi z_{1}}{l})\cos ( \frac{n\pi z_{2}}{l})\}\times \notag \\ & \int_{0}^{2\pi } d\theta \cos ( 2\pi r\sqrt{(\frac{1}{\lambda } ) ^{2}-(\frac{n% } { 2l})^{2}}\cos \theta ) . \label{24}\end{aligned}\ ] ] where @xmath121 , @xmath122 $ ] is the largest integer less than @xmath123 , @xmath124 is spontaneous emission rate in vacuum for any one of the two same dipoles , and @xmath125 is the coordinate of z for dipole i , here , we set @xmath126 , @xmath113 is the distance between the two dipoles in xy plane . for @xmath110 , this leads to the well - known results @xmath127}2[1-(\frac{n\lambda_{0}}{2l}% ) ^{2}]cos^{2}(\frac{n\pi}{2})\}.\ ] ] in the calculation , we set @xmath128 , and @xmath129 is gradually increased from 0 to 49/41 for simplicity . cooperative decay parameters @xmath130 as a function of the interatomic separation @xmath113 is displayed in fig . [ ssp3](b ) . black line is for the analytic results from eq . ( [ 24 ] ) and red boxed dots represent the calculated from eq . ( [ 22 ] ) . we find that the numerical result from our method also agrees very well with the analytic solution . , the radii of the air hole is @xmath131 and the slab height is @xmath132 . the refractive index of the slab is 3.4 . there is a defect hole in the center with refractive index @xmath133 . its radii is the same as the air hole . the origin of the axes is set at the center of the photonic crystal nanocavity . the two dipoles are located at point a and b on the center plane of the slab and the two dipole moments are parallel to the x axis . ] through the above two numerical tests , we clearly shown that our method could be able to efficiently and exactly deal with the dipole - dipole interaction . in the following of this section , dipole - dipole interaction in photonic crystal cavity ( pcs ) has been numerically studied . the sketch of the photonic crystal cavity is displayed in fig . it consists of a thin dielectric slab with air holes arranged as triangular lattice . the lattice constant is @xmath134 , the radii of the air hole is @xmath131 and the slab height is @xmath132 . the refractive index of the slab is 3.4 . there is a defect hole in the center with refractive index @xmath133 . its radii is the same as the air hole . the two dipoles are located at point a and b on the center plane of the slab and the two dipole moments are parallel to the x axis . the origin of the axes is set at the center of the pcs cavity . for convenience , the special points are also drawn in fig . [ ssp4 ] . ` atom ' a located at @xmath135 , ` atom ' b located @xmath136 , where r represents the separation and varies from @xmath137 to @xmath138 . the two transition dipole moments are parallel to the x axis . thanks to the scaling law , the cooperative decay parameters @xmath66 , dipole - dipole interaction potential @xmath67 and frequency @xmath56 in unit of @xmath139 are dimensionless . and ( b ) the dipole - dipole interaction potential @xmath140 versus @xmath141 for atom a located at @xmath135 , atom b located at @xmath142 in the photonic crystal nanocavity . @xmath143 and @xmath140 are in unit of @xmath144 ( @xmath145 ) . the @xmath146 is in unit of @xmath147 . the two dipole moments are parallel and along the x axis . the insets show the behavior around the defect frequency ( @xmath148 ) of the cavity . the two transition dipoles are parallel and along the x axis . ] the frequency dependent characters for cooperative decay parameters @xmath119 and dipole - dipole interaction potential @xmath149 are presented in fig . [ ssp5](a ) and fig . [ ssp5](b ) for atom a located at @xmath135 and atom b located at @xmath150 . the insets show the behaviors for frequency around the resonance frequency ( @xmath151 ) . from fig . [ ssp5](a ) , we clearly see that @xmath152 oscillates remarkably when @xmath56 is far away from @xmath153 , which shows the powerful modulation ability of photonic crystal for the electromagnetic eigen mode and the density of photonic states . for @xmath154 ( the inset ) , @xmath155 for relatively large atomic separation ( @xmath156 ) , where @xmath157 is the largest @xmath158 for two dipoles located in vacuum for the atomic separation @xmath110 . this large @xmath159 is attributed to the enhancement of photon emission and reabsorbing rates , which can be roughly characterized by the ratio of q / v . aimed for quantum computation and quantum information processing @xcite where large @xmath159 or large @xmath160 is needed , we can increase the quality factor q through improving the nanocavity design . for @xmath161 ( the inset ) , @xmath152 varies sharply and is sensitive to the frequency . this implies that tuning @xmath153 of the cavity or @xmath97 of the dipole can both help to control @xmath159 . figure [ ssp5](b ) shows similar properties for the dipole - dipole interaction potential . the inset shows that repulsive or attractive potential can be got around the cavity frequency . and ( b ) the absolute maximum dipole - dipole interaction potential @xmath162 versus the interatomic separation @xmath113 for atom a located at @xmath135 and atom b located at @xmath163 in the pcs cavity . the maximum are calculated for @xmath164 around @xmath165 . @xmath166 and @xmath167 are in unit of @xmath168 ( @xmath169 ) and @xmath146 is in unit of @xmath170 . the two transition dipoles are parallel and along the x axis . ] the position dependent characters have been investigated . figure [ ssp6](a ) and [ ssp6](b ) show the absolute maximum value for @xmath56 around @xmath171 of cooperative decay parameters @xmath172 and dipole - dipole interaction potential @xmath173 @xmath174 as a function of their separation r ( atom a located at @xmath175 and atom b located at @xmath136 ) are show where the maximum are calculated for @xmath56 around @xmath153 . @xmath176 and @xmath177 are in unit of @xmath178 ( @xmath179 ) . the @xmath56 is in unit of @xmath139 . the maximum value @xmath176 and @xmath177 varies in a similar way except for the amplitude . the above phenomena can be understood as follows : @xmath180 depends much on the electric field of the defect mode . equation ( [ 8 ] ) shows that @xmath166 changes in the same way as the electric field of the cavity mode at the location of atom b. the cavity mode along the y axis looks much the same as fig . [ ssp6](a ) and we did not show it here . further , we clearly see that @xmath181 and @xmath182 have been realized for atom a and atom b located at positions with same electric field strength along the x axis . in summary , we have proposed a new efficient and rigorous numerical method to investigate dipole - dipole interaction in photonic crystal nanocavity . we calculate the collective and individual radiation rates of classical dipoles by directly solving maxwell s equations in real space with a free - space boundary condition . by using the result of two dipoles radiation rate minus the sum of the two individual radiation rates , we get the cooperative decay parameters and dipole - dipole interaction potential . through the self - consistent procedure ( eq . ( [ 4 ] ) ) , exact non - markov results can be got and both real and virtual photon effects have been taken into account . further , it can be applied for dipoles with different transition frequencies in both weak and strong coupling regimes . our investigation suggests that this method works well for dipoles located in media with arbitrary shape and may be generalized to calculate many dipoles interaction . numerical validation has been made in vacuum and planar nanocavity . both results agree very well with the analytic . applying this method to a simple photonic crystal nanocavity , it is found that the cooperative decay parameters and the dipole - dipole interaction potential strongly depend on the atomic position , the transition frequency , quality factor and the cavity frequency . for two dipoles arranged at positions with the same local coupling strength , the cooperative decay parameters is equal to the local coupling strength . large cooperative decay parameters is achieved at the resonance frequency . dipole - dipole interaction potential changes continuously from attractive to repulsive case for transition frequency varying in a domain of the cavity linewidth around the resonance frequency . larger value and sharper change of cooperative parameters and dipole - dipole interaction can be obtained for higher quality factor . owing to the photonic crystal nanostructure is one of the most promising platform . it offers many advantages , including high quality factor , small mode volume , integrability with waveguide and fixed position of dipoles . in addition , many ingenious schemes for static and ultra fast dynamic control of the quality factor , resonance frequency have been proposed . based on these techniques for photonic crystal nanocavity and methods for tuning the transition frequency of quantum dots , our results provide some manipulative approaches for dipole - dipole interaction with potential application in various fields such as quantum computation and quantum information processing based on solid state nanocavity and quantum dot system . this work was supported by nsfc under grants nos . 10934010 , 60978019 , the nkbrsfc under grants nos . 2009cb930701 , 2010cb922904 , 2011cb921502 , 2012cb821300 , and nsfc - rgc under grants nos . 11061160490 and 1386-n - hku748/10 . e. gallardo , l. j. martez , a. k. nowak , d. sarkar , h. p. van der meulen , j. m. calleja , c. tejedor , i. prieto , d. granados , a. g. taboada , j. m. garca , and p. a. postigo , phys . rev . b * 81 * , 193301 ( 2010 ) .	dipole - dipole interaction between two two - level ` atoms ' in photonic crystal nanocavity is investigated based on finite - difference time domain algorithm . this method includes both real and virtual photon effects and can be applied for dipoles with different transition frequencies in both weak and strong coupling regimes . numerical validations have been made for dipoles in vacuum and in an ideal planar microcavity . for dipoles located in photonic crystal nanocavity , it is found that the cooperative decay parameters and the dipole - dipole interaction potential strongly depend on the following four factors : the atomic position , the atomic transition frequency , the resonance frequency , and the cavity quality factor . properly arranging the positions of the two atoms , we can acquire equal value of the cooperative decay parameters and the local coupling strength . large cooperative decay parameters can be achieved when transition frequency is equal to the resonance frequency . for transition frequency varying in a domain of the cavity linewidth around the resonance frequency , dipole - dipole interaction potential changes continuously from attractive to repulsive case . larger value and sharper change of cooperative parameters and dipole - dipole interaction can be obtained for higher quality factor . our results provide some manipulative approaches for dipole - dipole interaction with potential application in various fields such as quantum computation and quantum information processing based on solid state nanocavity and quantum dot system .
we present a four mode example of simulated data @xmath101 for a system comprised of a device described by a haar random unitary @xmath102 with losses on inputs @xmath103 and outputs @xmath104 @xmath105 & = \begin{pmatrix } 0.46 & 0 . & 0 . & 0 . \\ 0 . & 0.65 & 0 . & 0 . \\ 0 . & 0 . & 0.41 & 0 . \\ 0 . & 0 . & 0 . & 0.37 \end{pmatrix } \begin{pmatrix } 0.245 & 0.54 & 0.537 & 0.601 \\ 0.492 & 0.377 + 0.192 i & -0.634 + 0.213 i & 0.027 -0.362 i \\ 0.662 & 0.007 + 0.119 i & 0.305 -0.339 i & -0.549 + 0.196 i \\ 0.509 & -0.633 - 0.34 i & -0.042 + 0.235 i & 0.398 + 0.095 i \end{pmatrix } \begin{pmatrix } 0.08 & 0 . & 0 . & 0 . \\ 0 . & 0.82 & 0 . & 0 . \\ 0 . & 0 . & 0.55 & 0 . \\ 0 . & 0 . & 0 . & 0.24 \end{pmatrix}\\[8pt ] & = \begin{pmatrix } 0.009 & 0.204 & 0.136 & 0.066 \\ 0.026 & 0.201 + 0.102 i & -0.227 + 0.076 i & 0.004 -0.057 i \\ 0.022 & 0.002 + 0.04 i & 0.069 -0.076 i & -0.054 + 0.019 i \\ 0.015 & -0.192 - 0.103 i & -0.009 + 0.048 i & 0.035 + 0.008 i \end{pmatrix}\end{aligned}\ ] ] @xmath107 \begin{pmatrix } 1 & 1 & 1 & 1 \\ 1 & 0.348 -0.177 i & -0.588 - 0.197 i & 0.022 + 0.300 i \\ 1 & 0.005 -0.082 i & 0.211 + 0.234 i & -0.338 - 0.121 i \\ 1 & -0.565 + 0.304 i & -0.038 - 0.211 i & 0.319 -0.076 i \end{pmatrix } \begin{pmatrix } 0.245^{2 } \\ 0.540^{2 } \\ 0.537^{2 } \\ 0.601^{2 } \end{pmatrix } = \begin{pmatrix } 1 \\ 0 \\ 0 \\ 0 \end{pmatrix}\end{aligned}\ ] ]	linear optical circuits of growing complexity are playing an increasing role in emerging photonic quantum technologies . individual photonic devices are typically described by a unitary matrix containing amplitude and phase information , the characterisation of which is a key task . we present a constructive scheme to retrieve the unitary matrix describing an arbitrary linear optical device using data obtained from one - photon and two - photon ensembles . the scheme is stable on the arbitrarily increasable length scale of the photon packet and independent of photon loss at input and output ports of the device . we find a one - to - one correspondence between ideal data and unitary matrix , and identify the class of non - unitary matrices capable of reproducing the data . the method is extended for coherent state probes , which can simulate two - photon statistics with a reduced visibility . we analyse the performance of reconstruction to simulated noise . photonic circuitry is set to play an important role in emerging quantum technologies @xcite , connecting or implementing quantum logic gates @xcite , enabling photonic simulators @xcite , and coding states for quantum communication @xcite , with integrated optics @xcite enabling large reconfigurable circuits @xcite . in these contexts , each lossless linear optical network is ideally described by a unitary matrix @xcite , the determination of which is a key task . while a general quantum optical process may be fully characterised with coherent light and homodyne detection @xcite , the device to be determined and the probing set - up must be stable with respect to one another on a sub - wavelength scale . since the phases imparted to each photon of a two - photon state combine to a single global phase , circuit probing via two - photon interferometry @xcite is stable on the scale of the photon packet , which is typically orders of magnitude greater than wavelength . photons generated from spontaneous parametric down conversion ( spdc ) can be lengthened through spectral filtering or with a cavity in narrowband spdc @xcite , while photons generated in an atom - cavity system can be many meters long @xcite . two - photon sources are commonplace in quantum optics labs , with quantum interference visibilities approaching unity @xcite , and future many - photon experiments are likely to be easily configured to produce subsets of data from one - photon and two - photon ensembles for characterisation purposes . additionally , two - photon quantum statistics can be classically simulated by two coherent beams with a mutually randomised phase @xcite so that any circuit information retrievable with a two - photon state is also available to classical states . in ref . @xcite the transfer matrix of a four - mode linear optical device was numerically approximated as the matrix that minimised the distance between a self - simulated data - set of all possible two - photon statistics and experimental data , with data from one - photon ensembles used as starting parameters . however , a numerical search among the space of @xmath0 dimensional unitary matrices based on the @xmath1 data elements associated with all possible two - photon statistics becomes challenging for large systems . more fundamentally , it has not been clear up to this point if data from one - photon ensembles ( one - photon data ) and data two - photon ensembles ( two - photon data ) are sufficient to uniquely determine the matrix description of a linear optical circuit . here , we give a constructive method to efficiently determine the set of matrices that produce a given set of ideal one - photon and two - photon data , for an arbitrary linear optical network . reconstruction is insensitive to fluctuating phases between input ports and between output ports , as shown in fig [ fgschematic ] , and therefore does not require active stabilisation or other techniques for stable classical interferometry . in addition , reconstruction is independent of the typically unknown and tomographically troublesome photon losses at device input and output ports . the scheme involves injecting one - photon and two - photon states into the linear optical network with correlated photon detection , as shown in fig [ fgschematic ] . matrix amplitude information is probed with one - photon ensembles , while specific phase parameters are targeted by specific two - photon ensembles , so that the required number of measurements scales quadratically with the number of modes , and linearly with the number of unknown parameters . assuming the device can be described by a unitary matrix , applying @xmath2 of the @xmath3 orthonormal constraints leads to a one - to - one mapping between idealised experimental data and unitary matrix , up to the fluctuating phases . if the port - losses are known to be zero , then a matrix can be recovered with no orthonormal constraints invoked or required . we consider non - ideal situations and simulate noisy data from circuits of up to @xmath4 optical modes to quantify the fidelity of reconstruction versus noise . an insightful example of the tools used in our method is the case of a two - mode device , or subsection of a larger device , as shown in fig [ fg2x2](a ) . @xmath5 photons are individually injected into input port @xmath6 , coupled with efficiency @xmath7 , transmitted to output port @xmath8 with probability @xmath9 , and detected with efficiency @xmath10 , giving a count rate of @xmath11 . with @xmath12 , four such measurements combine to constrain four transition probabilities @xmath13 independently of coupling efficiencies : @xmath14 \implies \frac{t_{1,1 } \ , t_{2,2}}{t_{1,2 } \ , t_{2,1 } } & = \frac{r_{1,1 } \ , r_{2,2}}{r_{1,2 } \ , r_{2,1}}. \label{eq2x2}\end{aligned}\ ] ] in the case that the @xmath15 device behaves unitarily , probability constraints ( e.g. @xmath16 ) allow description in terms of one parameter , @xmath17 to derive the experimentalist s loss - insensitive equation for the reflectivity of a beamsplitter , @xmath18 , @xmath19 where @xmath20 replaces the rhs of . an alternative way to find the reflectivity of a beamsplitter uses the visibility of quantum versus classical correlations in a hong ou mandel ( hom ) dip @xcite . more generally , the visibility of a hom dip can be used to apply loss - insensitive constraints to the @xmath15 submatrix @xmath21 that describes a two mode system , which may be part of a larger device . @xmath21 consists of real positive probability amplitudes @xmath22 and unit complex phase factors @xmath23 ; pre and post multiplication by diagonal matrices @xmath24 and @xmath25 includes coupling and detection efficiencies , to describe of the overall subprocess @xmath26 , @xmath27 as shown in fig [ fg2x2](b ) . indicating correlated detection . ] simultaneously injecting two photons into the @xmath28 device and counting subsequent coincidental detector clicks as a proportion , gives a quantum signal equal to the square of the absolute value of the permanent of @xmath28 : @xmath29 introducing a delay between the photons so that no quantum interference occurs gives the classical signal equal to the square of the permanent of the matrix containing absolute values of the entries in @xmath28 : @xmath30 & = & s_{h } \ , s_{k } \ , r_{g } \ , r_{j } \left(\tau^{2}_{j , k } \ ; \tau^{2}_{g , h } \ ; + \ ; \tau^{2}_{g , k } \ ; \tau^{2}_{j , h } \right ) \nonumber \\[6pt ] & = & r_{j , k } \ ; r_{g , h } \ ; + \ ; r_{g , k } \ ; r_{j , h } \label{eqc}\end{aligned}\ ] ] and can be calculated directly from one - photon data . the loss insensitive visibility is calculated as @xmath31 \label{eqvdef } & = & \ ; -2 \cos{(\alpha_{j , k } - \alpha_{j , h } - \alpha_{g , k } + \alpha_{g , h})}\nonumber\\[6pt ] & \times & \ ; \frac{\ ; \tau_{j , k } \ ; \tau_{j , h } \ ; \tau_{g , k } \ ; \tau_{g , h}}{\tau^{2}_{j , k } \ ; \tau^{2}_{g , h } \ ; + \ ; \tau^{2}_{g , k } \ ; \tau^{2}_{j , h}}\nonumber . \label{eqv}\end{aligned}\ ] ] probability amplitudes @xmath22 are the square root of the transition probabilities @xmath32 described in so can be constrained from @xmath33 \implies x_{g , h , j , k } & = \frac{\tau_{j , k } \ , \tau_{g , h}}{\tau_{j , h } \ , \tau_{g , k}}. \label{eq2x2g}\end{aligned}\ ] ] a loss - independent measurable constraint on @xmath21 phase angles then follows from @xmath34 \cos{(\alpha_{j , k } - \alpha_{j , h } - \alpha_{g , k } + \alpha_{g , h } ) } & = -\frac{v_{g , h , j , k } \ ; y_{g , h , j , k}}{2}. \label{eqang}\end{aligned}\ ] ] since cosine is a symmetric function , the sign of the argument is unknown . however , the sign for any one of the four phase angles , say @xmath35 , can be ascertained from if its absolute value @xmath36 is known , and the absolute values _ and _ signs of the other three ( reference ) angles are known , which will be the case , as we will see . the sign on @xmath35 can be found by labelling the argument from as @xmath37 then computing @xmath38 = \operatorname{sgn } [ \nonumber \\ & \|\beta_{g , h , j , k } - \|\alpha_{j , k } - \alpha_{j , h } - \alpha_{g , k } - \|\alpha_{g , h}\|\|\| \nonumber \\ - & \|\beta_{g , h , j , k } - \|\alpha_{j , k } - \alpha_{j , h } - \alpha_{g , k } + \|\alpha_{g , h}\|\|\| \;\ ; ] \label{eqsign}\end{aligned}\ ] ] assuming the 3 reference angles do not sum to zero . ) choices for a @xmath15 submatrix that contains @xmath35 . ] using the amplitude and phase constraints of and with the sign disambiguation of , we can reconstruct @xmath39 , the @xmath40 dimensional matrix description of an arbitrary @xmath40 mode optical network , up to @xmath41 of the @xmath42 parameters , independently of any unknown losses on input and output ports . when @xmath39 can be assumed to behave as a unitary transformation , constraints give a tractable solution system for the @xmath41 unknown parameters . we can specify some equivalence among matrices where corresponding linear optical networks within the equivalence class produce the same data . firstly , two matrices @xmath43 and @xmath44 are equivalent if there exist two diagonal unitary matrices @xmath45 and @xmath46 such that @xmath47 . in a linear optical network @xmath39 , these diagonal matrices can be regarded as the typically unknown and somewhat trivial phases on the input and output ports of the network , to which our data is insensitive , as long as photons are input in a fock state ( not superposed across modes ) and there is no post-@xmath39 interferometry before detection . these trivial phases may equally be regarded as a particular instance of the fluctuating phases from fig [ fgschematic ] . we encapsulate this idea in the standard way by insisting that our @xmath39 is _ real bordered _ , that is , the entries on the left - most column and uppermost row should be real and positive : @xmath48 . secondly , noting the equality of photon statistics under the complex conjugate transformation @xmath49 , we insist that the imaginary part of the element @xmath50 be non negative : @xmath51 . additionally , we make the following assumptions : for a straightforward description , we firstly assume that the two leftmost columns and two uppermost rows contain no vanishingly small elements if this were not the case , one - photon data would identify zero matrix elements and the reconstruction method can be straightforwardly adapted ; secondly , while our method is insensitive to loss on the input and output ports , we assume any losses are not near - total . with these equivalencies and assumptions , the matrix to be recovered can be parameterised as @xmath52 \tau_{2,1 } & \ , \tau_{2,2 } \ , e^ { i \alpha_{2,2 } } & \ , \cdots & \ , \tau_{2,m } e^ { i \alpha_{2,m } } \\[8pt ] \vdots & \ , \vdots & \ , \ddots & \ , \vdots \\[8pt ] \tau_{m,1 } & \ , \tau_{m,2 } e^ { i \alpha_{m,2 } } & \ , \cdots & \ , \tau_{m , m } e^ { i \alpha_{m , m } } \end{pmatrix } \label{eqm1}\ ] ] where , as before , the @xmath22 correspond to transition amplitudes and the @xmath53 correspond to phase angles . replacements are made on all elements of @xmath54 _ apart from _ the first column and first row . the @xmath22 amplitudes are replaced following fixing @xmath55 while the indices of @xmath56 are taken from the @xmath22 to be replaced . all @xmath22 not in the first column or row , become functions of the @xmath22 in the first row and column . phase angles @xmath57 are found ( up to a sign ) following with the same index pattern used for the @xmath22 , so that only one angle appears in the cosine . apart from the sign on @xmath58 , which is defined to be positive , all signs on the @xmath53 in @xmath54 are determined using . signs on the @xmath53 in the 2nd column are found by fixing @xmath59 while the indices of @xmath60 , moving from third row to row @xmath40 . similarly , signs on the 2nd row are found by fixing @xmath61 and @xmath62 , moving from third column to column @xmath40 . signs on the remaining @xmath53 are found by fixing @xmath63 and the indices of @xmath35 are taken from the @xmath53 to be replaced . this pattern ensures that signs on all references phases in are known , as is required . with these replacements , the transfer matrix can be parameterised by @xmath41 unknown border amplitudes @xmath64 \tau_{2,1 } & \ , \tilde{x}_{2,2,1,1 } e^ { i \tilde{\alpha}_{2,2 } } & \ , \cdots & \ , \tilde{x}_{2,m,1,1 } e^ { i \tilde{\alpha}_{2,m } } \\[8pt ] \vdots & \ , \vdots & \ , \ddots & \ , \vdots \\[8pt ] \tau_{m,1 } & \ , \tilde{x}_{m,2,1,1 } e^ { i \tilde{\alpha}_{m,2 } } & \ , \cdots & \ , \tilde{x}_{m , m,1,1 } e^ { i \tilde{\alpha}_{m , m } } \end{pmatrix}. \label{eqm2}\ ] ] where , for ease of notation , we have introduced @xmath65 \tilde{\alpha}_{g , h } & = \operatorname{sgn}[\alpha_{g , h } ] \arccos ( \frac{1}{2 } \;v_{g , h , j , k } \ ; y_{g , h , j , k})\nonumber , \label{eqanglabel}\end{aligned}\ ] ] highlighting that the @xmath53 are fully determined from experimental measurements and not functions of the @xmath22 . the matrix @xmath66 , describing the complete experimental set up comprising the device @xmath39 and losses on input and output ports , has transition probabilities corresponding to the set of one photon count rates @xmath67 , while phase angles are the calculated @xmath53 : @xmath66 is therefore fully characterised so that the output state can be calculated for any arbitrary photon number fock state input . this implies that probing the system with more than two photons , or using number resolving detectors , provides no extra information to solve the remaining @xmath41 unknown parameters in @xmath68 . in the case of no losses @xmath69 and full reconstruction has been achieved . however , zero losses on input and output ports is atypical , particularly when one considers collection efficiencies from a photon source and detection efficiencies . when @xmath39 can be assumed to behave unitarily , normalisation can be applied to the first column of @xmath68 and orthogonality constraints applied between the first column and the remaining @xmath70 columns , with the same applying for rows . the solution system is the matrix of coefficients in @xmath68 , @xmath71 1 & \ , x_{2,2,1,1 } \ , e^ { i \tilde{\alpha}_{2,2 } } & \ , \cdots & \ , x_{2,m,1,1 } e^ { i \tilde{\alpha}_{2,m } } \\[8pt ] \vdots & \ , \vdots & \ , \ddots & \ , \vdots \\[8pt ] 1 & \ , x_{m,2,1,1 } e^ { i \tilde{\alpha}_{m,2 } } & \ , \cdots & \ , x_{m , m,1,1 } e^ { i \tilde{\alpha}_{m , m } } \end{pmatrix},\nonumber \label{eqsolve1}\ ] ] applied to the transition probabilities , to give the orthonormal constraints , @xmath72 \tau^{2}_{2,1}\\[8pt ] \vdots \\[8pt ] \tau^{2}_{m,1 } \end{pmatrix } = \begin{pmatrix } 1\\[8pt ] 0\\[8pt ] \vdots \\[8pt ] 0 \end{pmatrix } , \;\;\;\;\;\ ; m_{\mu } \begin{pmatrix } \tau^{2}_{1,1}\\[8pt ] \tau^{2}_{1,2}\\[8pt ] \vdots \\[8pt ] \tau^{2}_{1,m } \end{pmatrix } = \begin{pmatrix } 1\\[8pt ] 0\\[8pt ] \vdots \\[8pt ] 0 \end{pmatrix}. \label{eqsolve1}\ ] ] a standard method for solving systems of linear equations may then be used . one - photon and two - photon statistics can be estimated using classical coherent states . the classical analogue of one - photon counts @xmath11 are intensities @xmath73 , measured at port @xmath8 from injecting a coherent beam of intensity @xmath74 into port @xmath6 : the one - photon counts in can be replaced by a ratio of classical intensities . the classical counterpart of hong ou mandel correlations are those of hanbury - brown and twiss @xcite arising from interference between two independent fields @xcite , which have already been exploited in linear optical devices to simulate two - photon statistics @xcite . the quantum coincidence rate @xmath75 can be accurately estimated by injecting two coherent states with the same mean photon number and a randomised relative phase into respective input ports @xmath76 and @xmath6 . the intensity correlation function , @xmath77 , measured at output ports @xmath78 and @xmath8 is @xmath79 so that the quantum correlation can be estimated from @xmath77 minus the product of individual intensities @xmath80 experimental noise and device imperfections will drive the reconstructed matrix away from that which best describes the optical device , and away from unitarity since we use only some of those constraints when solving for border amplitudes . the unitary matrix @xmath81 closest to the reconstructed matrix may be found from the polar decomposition @xcite , but in the presence of errors @xmath81 will not , in general , perfectly match any true underlying unitary matrix @xmath82 . we analysed the effect of noisy data on our reconstruction procedure , for devices from dimension @xmath83 to @xmath84 , over a range of noise level @xmath85 . noise was added to the simulated experimental values for one - photon count rates @xmath67 and quantum interference visibilities @xmath86 : the ideal values were multiplied by @xmath87 , where @xmath88 was chosen from a normal distribution centred at @xmath89 . the amount of noise @xmath85 was controlled by varying the width of the distribution with @xmath90 , the third standard deviation , making it very unlikely that experimental values are found to be much further than @xmath85 away from their ideal values . for a particular value of @xmath40 and @xmath85 , we generated @xmath91 haar random unitary matrices , each of which was pre and post multiplied by diagonal matrices with real random values to simulate loss . the algorithm was ran for each instance , with a unitary matrix extracted via the polar decomposition . the average trace distance was computed over all @xmath91 examples to give an average fidelity of reconstruction ; the variance of each @xmath91 example set was also computed . over the tested range of matrix size @xmath40 , we found the empirical relation between average fidelity of reconstruction @xmath92 and error level @xmath85 @xmath93 with @xmath94 giving good predictions for fidelities down to at least @xmath95 . figure [ fgfidnoise ] shows @xmath96 simulated data points between @xmath97 and @xmath98 for @xmath83 , and between @xmath97 and @xmath99 for @xmath84 , with error bars on each point corresponding to the variance of its @xmath91 example data set . the empirical curve @xmath92 is shown as the solid line . the effect of errors analysed here involved no attempt at error correction . a useful line of research would be to investigate how the impact of errors may be minimised by advantageously permuting the matrix prior to reconstruction . additionally , having used only @xmath100 two - photon measurements from the possible @xmath1 , it would be interesting to understand how extra subsets of two - photon data may increase robustness to noise , or whether resources should be directed toward reducing errors on the minimal set . we have presented a scheme in which the unitary matrix of a linear optical device may be constructively retrieved without the requirement for stable classical interferometry . the merits of probing with coherent states versus single photon states is interesting . since the coincidental signal falls off quadratically ( due to the number of ways of choosing 2 from @xmath40 modes ) an arbitrarily bright coherent state may seem preferable . however , two - photon states readily available within a larger multi - photon experiment may exhibit the same characteristics as those producing the larger data set , with statistics more representative of a device that is somewhat defined with respect to its photons ; for example , the reflectivity of a waveguide directional coupler is a function of wavelength . any advantage from an increase in signal and greater measurement precision with classical states should be weighed against potential inaccuracies if the crucial properties of those states do not match the properties of the photons , due to some experimental uncertainty . furthermore , the capability to re - take tomographic data as a subset of principal experimental data , for a freshly reconfigured system , during the course of an uninterrupted experiment , is appealing . we thank i. barr , j. carolan , m. lobino , e. martn lpez , j. matthews , n. russell , and t. rudolph for helpful discussions . this work was supported by epsrc , erc , and nsqi . j.l.ob . acknowledges a royal society wolfson merit award . 26 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , _ _ ( , , ) . , , , * * , ( ) . , , , , , * * , ( ) , issn . , , , , , , , , , , , * * , ( ) . , in _ _ ( , ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , * * , ( ) . , , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
due to their remarkable properties @xcite , rydberg atoms emerge as a versatile tool in ultra - cold atomic physics . they can be used for diverse topics , such as quantum information , atomic aggregates @xcite , the study of conical intersections @xcite , digital quantum simulations @xcite , electro - magnetically induced transparency @xcite , coherent population trapping @xcite , giant kerr non - linearities @xcite and ultra - long range molecules @xcite , since their long range interactions can be tuned over many orders of magnitude . however excited atoms in rydberg orbitals are more vulnerable to ionisation @xcite or spontaneous decay and more difficult to trap @xcite . recently , several groups proposed to combine the advantages of both , ground - state and rydberg atoms , through `` rydberg - dressing '' schemes @xcite . the core idea is to use off - resonant laser coupling between ground- and rydberg states to create eigenstates of the laser - coupled system in which a small rydberg component is admixed to the ground - state . these resulting _ dressed _ states inherit some of the extreme properties of the rydberg states , while preserving favourable properties of the ground state . moreover , the additional degrees of freedom provided by laser rabi - frequency and detuning increase flexibility and promise potential shaping as well as dynamical control . dressing has been extensively studied for van der waals ( vdw ) interactions , generated when virtual transitions between levels of a two - atom system yield a distance dependent energy shift . several articles have proposed rydberg dressing techniques to induce this kind of interactions for ground - state atoms @xcite . in this context it was shown that dressing ground - state atoms with rydberg levels allows the large rydberg vdw interaction energies to be reduced until comparable to typical energy scales of cold - atom traps , while simultaneously increasing ensemble life - times to those required for thermal equilibration effects @xcite . these are helpful features e.g. , for realising a supersolid phase in rydberg - dressed bose - einstein condensates @xcite . here , we demonstrate that the ideas behind dressing rydberg vdw interactions can also be applied to resonant dipole - dipole interactions of neutral atoms . they occur when transitions between different states of the two - atom system are energetically resonant . in this case , the existence of an interaction potential is linked to electronic state transfer between atoms @xcite . dressing dipole - dipole interactions requires the use of two laser couplings and four electronic states ( two ground- and two rydberg - states ) , in contrast to one coupling and two states that suffice for vdw dressing . hence , the scheme is more involved but also more flexible , offering a larger number of dressed states . it ultimately allows coherent quantum state transport between atoms in different long - lived ground - states over distances of @xmath0-@xmath1 @xmath2 m and for durations of many milli - seconds . similar ideas have previously been applied to trapped rydberg _ ions _ @xcite . however , we are motivated by the possibility to generate potentials in an atomic many - body system that induce different kinds of ( coupled ) electronic and atomic _ motion _ dependent on the overall quantum state of the system . useful applications of such potentials can be found in @xcite . in many systems this goal is easier to achieve with neutral atoms than with ions , since the motion of the latter is typically dominated by their coulomb repulsion or the indispensable trap . furthermore the mapping of state - transfer interactions to normally non - interacting ground states by the laser dressing can be presented more clearly in the atomic case , where complications by trapping , mixing of internal and external dynamics and additional laser induced interactions can be ignored at least for free atoms . finally , see ref . @xcite for dressing techniques targeting the control of atomic motion without linked excitation transport . our results provide an additional handle on the time - scale of excitation transport and life - times , enable time - dependent control over hopping - strengths and can be used to vary the order of magnitude of dipole - dipole forces . within some constraints @xcite , the dressing also enables the use of well established trapping methods . as recently shown , dressed dipole - dipole interactions are an important tool for the realisation of atomic ring trimers @xcite . in particular , the confinement of long - range interacting atoms on a ring , as required in ref . @xcite , is greatly facilitated by the techniques presented here . atomic ring trimers allow the detailed study of wave - function dynamics near and across conical - intersections . dressed dipole - dipole interactions may also prove useful for the study of exciton dynamics in rydberg chains @xcite . to demonstrate dressed dipole - dipole interactions , we employ numerical simulations of a model with four essential states per atom . in this case there are several dressed state manifolds , distinguished by the number of excited rydberg atoms which they contain in the limit of vanishing dressing . we apply van - vleck perturbation theory to obtain analytical expressions for the induced effective interaction and to determine the parameter range within which the scheme can function . the paper is organised as follows : in we present the atomic states considered , describe the resulting hamiltonian and outline the basic principles of the dressing scheme . in we consider an atomic dimer in detail , providing analytical expressions for the effective dressed interactions , diagonalising the full hamiltonian and discussing different available dressing manifolds . in we present two exemplary applications of our results , dressed ring trimers and exciton migration . we conclude in and discuss technical details of our calculations and the underlying atomic physics in the subsequent appendices . we consider a collection of @xmath3 alkali atoms , where the @xmath4th atom is located at the position @xmath5 . the distances between the atoms , @xmath6 , shall be so large that interaction between two atoms can be neglected unless _ both _ of them are in a rydberg state . in this case they experience strong dipole - dipole interactions . as discussed in @xcite the dipole - dipole interaction leads to exciton dynamics similar to that of molecular dye aggregates , molecular crystals or light harvesting systems @xcite . for practical reasons we restrict ourselves to situations where the dynamics can be adequately described using only two `` ground '' states and two rydberg states per atom . with the term `` ground - states '' we refer e.g. , to two different hyperfine levels of the actual alkali - atom ground state . we shall denote the ground states by @xmath7 and @xmath8 and the rydberg states by @xmath9 and @xmath10 . the latter correspond to the @xmath11 and @xmath12 angular momentum components of a rydberg level with large principal quantum number @xmath13 . schematically , the resonant dipole - dipole interaction @xmath14 leads to a coupling between states @xmath15 and @xmath16 , whose matrix elements in the hamiltonian take the form @xmath17 . the transition strength is @xmath18 where @xmath19 is the inter - atomic distance and @xmath2 quantifies the transition dipole moment between the states @xmath9 and @xmath10 . we assume that all atoms are prepared in the @xmath20 azimuthal quantum states and never acquire @xmath21 . with atoms constrained in a two - dimensional ( 2d ) plane , orthogonal to the quantisation axis , this ensures that there is no angular dependence of @xmath14 @xcite . such a geometry covers both scenarios described later in this article . the interaction between atoms @xmath22 and @xmath23 reads @xmath24 to map the strong interactions @xmath25 to the ground states , the atoms are irradiated with far - detuned dressing laser - fields that couple the ground - states and our two selected rydberg states coherently . the relevant level diagram is sketched in fig . [ fig : sketch ] ( a ) . in practice such a coupling is commonly achieved by two ( or multi)-photon transitions @xcite , so that the rabi - frequencies @xmath26 and detunings @xmath27 in the diagram have to be regarded as effective quantities . from the laser transition parameters , we assemble the effective dimensionless dressing parameters @xmath28 they are a measure of how `` far - detuned '' the laser coupling is , and will emerge as crucial quantities controlling the dressing , as it is the case for dressed vdw interactions @xcite . it is important that the state @xmath7 is directly coupled only to @xmath9 and @xmath8 only to @xmath10 . in order for this simple picture with four relevant states per atom to be valid , the detunings of both laser couplings have to be chosen such that all other transitions , coupling to further rydberg states or ground state levels , are so far detuned that they can be safely neglected and @xmath29 with uniquely assigned states @xmath30 . see [ full : levels ] for more details . ] . we show some realistic level diagrams for @xmath31li in fig . [ fig : sketch ] ( b ) to demonstrate how this constraint can be met in practice . throughout the article , we will refer to the states @xmath7 , @xmath9 as the `` s - pair '' and the states @xmath8 , @xmath10 as the `` p - pair '' . using the four states introduced above as a basis for the single atom , an @xmath3-body basis state @xmath32 is written as @xmath33 where @xmath34 describes the electronic state of the atom @xmath35 . for example we write @xmath36 when the first atom is in state @xmath7 , the second in @xmath8 and the third in @xmath9 . after defining operators @xmath37 with @xmath22 , @xmath38 @xmath39 where @xmath4 is the atom - index , the many - body hamiltonian can be written as @xmath40 with @xmath41 and @xmath42 we split @xmath43 into two parts , @xmath44 and @xmath45 , to facilitate our later application of perturbation theory . the atom - light coupling in has been treated in the dipole- and rotating wave approximations , as explained in [ rotwave ] . the _ exciton number _ operator @xmath46 measures the number of atoms in the p - pair of states . it is easy to see that @xmath47 commutes with @xmath43 and hence the exciton number is conserved , see also [ multi_exc ] . of course equation is a simplified description of a complicated multi - state system . for a more complete picture , one could include additional rydberg levels adjacent to @xmath9 , @xmath10 , as shown in fig . [ fig : sketch ] ( b ) , van der waals interactions of all four states , spontaneous decay and loss in the excited states . the latter arises as atoms in rydberg states are lost due to auto - ionization or blackbody radiation @xcite . importantly , parameter regimes can be found where these corrections are minor and suffices , for example for the states and parameters shown in fig . [ fig : sketch ] ( b ) and described in . consider an atomic sample such as a chain , prepared in states @xmath48 , in which the nth atom is in the upper ground state @xmath8 and all others in the absolute ground - state @xmath7 @xmath49 without dressing ( @xmath50 ) this is an eigenstate of the hamiltonian ( [ hamil_dressed ] ) . if we ignore dipole - dipole interactions for the moment , an adiabatic ramp - up of the dressing strengths @xmath26 will change this eigenstate into one with some admixture of rydberg population . allowing then for dipole - dipole interactions finally causes excitation transport via processes of the schematic form : @xmath51 . here @xmath52 symbolizes a transition of both atoms between the ground state manifold and the rydberg state manifold due to the dressing lasers . similarly @xmath53 stands for a transition occurring due to resonant dipole - dipole interactions between rydberg states . in we will confirm this expectation with numerical simulations and later justify and quantify it using perturbation theory . if the lasers are sufficiently off - resonant , the admixed fraction of excited state should remain small throughout . for this reason we can describe the process that we have just sketched as an effective interaction in the space spanned by the @xmath48 . we will refer to it as the _ ground - state manifold_. for later reference we also define spaces spanned by @xmath54 which we call _ gp ( hs ) single excitation manifold_. throughout this paper we focus on a specific realisation of our scheme , sketched in fig . [ fig : sketch ] ( b ) . we consider @xmath31li atoms , whose @xmath55 ground - state has a hyperfine - splitting of @xmath56 mhz @xcite . for the rydberg state @xmath9 we pick a principal quantum number @xmath57 and angular momentum @xmath11 state . for @xmath10 we choose @xmath57 and @xmath12 . the energy difference between the two rydberg states is @xmath58 ghz . we will assume detunings @xmath59 mhz and dressing parameters @xmath60 unless indicated otherwise . we will occasionally refer to @xmath61 ( without index ) , which implies @xmath62 . for the above values of @xmath61 and @xmath63 , our rabi - frequencies are @xmath64 mhz . for the most part , we illustrate the dressing scheme through an atomic dimer with @xmath65 @xmath2 m inter - particle separation . for this distance , the strength of the bare dipole - dipole interaction is @xmath66 mhz , just less than the doubly excited state detuning @xmath67 mhz . the reason for this choice will become clear later . we stress that we have chosen these parameters to be specific , while our findings are more general . the constraints under which the parameters can be varied without invalidating the dressing scheme have been discussed in . all following results are expected to hold for any alkali species if these constraints are met . for the principal demonstration of excitation transport through rydberg dressing , we first treat the simplest possible case , the atomic dimer . in this section , we show dynamics from the initial state @xmath68 . we consider two different ramps of the dressing couplings @xmath26 in the hamiltonian : ( i ) a sudden jump from zero to their final value and , ( ii ) an adiabatic ramp over a small finite time @xmath69 ms , effectively changing the initial state to @xmath70)/{\cal n}$ ] , where @xmath71 is a normalisation factor ( see also ) .. , the ramp ought to be slow on the time - scale of the laser - coupling but fast on that of effective dipole - dipole transport . thus @xmath72 with @xmath73 @xmath2s and @xmath74 ms ( the value is read off and defined as the time it takes for the excitation to oscillate from the first atom to the second and back . ) . ] we expand the time dependent quantum state as @xmath75 , where @xmath32 are the @xmath3-body basis vectors defined in . in this basis , we express the hamiltonian in matrix form and solve the schrdinger equation ( se ) @xmath76 by matrix exponentiation . using the four - state model introduced in we employ the full hamiltonian without further approximations . for two atoms placed a distance @xmath65 @xmath2 m apart , the resulting population - oscillations between the states @xmath77 and @xmath78 can be clearly seen in . on the logarithmic scale of panel ( b ) , we see that also states containing one rydberg excitation or more are populated , but their occupation is roughly suppressed by @xmath79 , indicated by the black solid line . the suppression of excited state population by a factor @xmath80 is well known from vdw dressing techniques @xcite and a basic feature of off - resonant rabi - coupling . as can be seen in , there will be many ground state population oscillations up to the expected spontaneous life - time of the dressed @xmath81-atom state , which we estimate to about @xmath82s as described in . due to the dressing , the time - scale on which the interaction transports the excitation has been massively increased compared to the `` bare '' population oscillation period between rydberg states @xmath83 , which would be @xmath84ns . the comparison of panels ( b ) and ( c ) of shows that the use of the more refined , adiabatically created , initial state has led to a more regular evolution of the excited state populations , whose oscillations now precisely follow those of the two ground states . the population of states with a single rydberg excitation never exceeds @xmath80 . doubly excited states are further suppressed . however , this careful initial state creation is not actually required in order to observe dressed population oscillations , as we have just seen . the results of the preceding section already unambiguously demonstrate excitation transport through full numerical solutions of the problem . to further understand the population - oscillations more intuitively , we now consider the ground - state manifold spanned by @xmath85 , @xmath86 as the _ system _ of interest and its hilbert - space complement as _ environment_. the coupling between the two is treated as a perturbation . we use van vleck perturbation theory @xcite , as outlined by shavitt _ @xcite to derive an effective hamiltonian in terms of the `` system '' only . this scheme conveniently takes care of degeneracies and generates an effective hamiltonian well suited to describe excitation transport . a full analytical diagonalisation of is impractical even for @xmath87 . in contrast the perturbative results for @xmath87 are intuitive and generalise straightforwardly to cases with more atoms . they are also valid over large ranges of parameter space as we shall show below . the aim of van vleck perturbation theory is to find a basis that block - diagonalises the hamiltonian ( [ hamil_dressed ] ) to a given order in the perturbation @xmath45 . for @xmath88 this is achieved by the basis @xmath32 introduced in , since @xmath44 is already fully diagonal . the relative importance of @xmath45 and @xmath44 is governed by the dressing parameters @xmath89 . as the @xmath89 increase , the basis that block - diagonalises @xmath43 becomes more complicated . perturbation theory will thus be valid as long as the dressing parameters @xmath89 are small . we will use the notation @xmath90 with @xmath91 , i.e. @xmath92 is a state whose leading component is @xmath32 as long as @xmath61 remains small and @xmath93 is small ( i.e. r is large ) . in this section we shall only be interested in the block of the effective hamiltonian that governs the dynamics of the effective ground - state manifold @xmath94 and @xmath95 . the details of the calculation are given in [ vanvleck ] , here we merely present the results . in the basis @xmath96 , @xmath97 one obtains @xmath98 where @xmath99 is the @xmath100 identity matrix . we have energy shifts @xmath101 and an explicitly distance dependent part @xmath102 . the light shifts @xmath103 ( @xmath104 ) , corresponding to second ( fourth ) order in @xmath45 , are given by [ eshifts ] @xmath105 @xmath106 where we have used the dressing parameters defined in . these light shifts could allow optical trapping using the same fields that provide the dressing coupling . the position dependent part of the hamiltonian , @xmath102 , can be written as @xmath107 with [ potentials ] @xmath108 @xmath109 where we have defined @xmath110 . as expected the states @xmath111 , @xmath112 acquire small excited state populations of the form : @xmath113 independent of @xmath19 . here @xmath114 is a normalisation factor . let us briefly mention some special cases : for large inter - atomic separations , such that @xmath115 , we can expand the expressions ( [ potentials ] ) and obtain @xmath116 and @xmath117 . the shift @xmath118 , which is independent of @xmath19 , can then be merged with @xmath104 . for shorter distances , the factors @xmath119 become relevant and diverge at @xmath120 , or equivalently at @xmath121 . we will see in the next section that this divergence can be traced to avoided crossings between the perturbed energy eigenvalues . we show in that ( [ potentials ] ) fails only in a fairly narrow region around this avoided crossing . an interesting special case is @xmath122 . here , the equations predict that the dressing effect on the ground - state manifold vanishes , which is confirmed by full simulations . we will however show in that this case has special appeal for dressing within the single excited state manifold . we also derive the contribution to @xmath123 of @xmath124th order in @xmath45 in [ sixth_order ] . we will make use of it in , to illustrate the rate of improvement that can be achieved through higher - order terms of perturbation theory . to get an overview of the consequences of dressing radiation beyond the ground state manifold , we now consider the full energy spectrum of the dimer as a function of inter - atomic separation . as we outline in [ multi_exc ] , the dimer hamiltonian can be brought into block - diagonal form , with three blocks describing exciton numbers of @xmath125 . of these , only the single exciton block is non - trivial , hence we only show the eigenspectrum of its corresponding sub - matrix @xmath126 ( [ hamiltonian_mainblock ] ) here . explicitly we solve @xmath127 besides providing information on the possibilities of dressed excitation transport , the energy eigenvalues as a function of separation can also be viewed as adiabatic ( born - oppenheimer ) potential surfaces for motion of the atoms , as long as this motion is sufficiently slow @xcite . the @xmath19-dependent energy eigenvalues are shown in for @xmath128 mhz and @xmath129 mhz , with other parameters as described in . the slight offset between @xmath130 and @xmath131 makes it easier to grasp the structure of the resulting spectrum . at large distances @xmath19 and for moderate dressing strength @xmath61 , the eigenstates are essentially superpositions of two basis states , and have odd or even symmetry under exchange of atoms @xmath132 and @xmath81 . these superpositions are listed in the caption of fig . [ spectrum_deltaeq ] ( a ) , disregarding normalisation . their energies are then mostly determined by the total detuning of the two superimposed basis states . as the distance @xmath19 is reduced , the dipole - dipole interaction energy of the states consisting mainly of two rydberg atoms ( blue dotted- and red solid lines ) becomes more prominent and eventually leads to avoided crossings between these and the other states of the spectrum . on the large energy scale of fig . [ spectrum_deltaeq ] ( a ) , @xmath133 , the dipole - dipole shift of the @xmath134 states is most prominent . however also the other states with only a small doubly excited state component acquire a space dependent dipole - dipole potential , as can be seen in the close - ups , panels ( b)-(d ) . the energy splitting of the states within the ground - state manifold , panel ( b ) , is of order @xmath135 while that in the singly excited state manifolds , panels ( c - d ) , is of order @xmath136 . in panel ( b ) we additionally display the energy eigenvalues obtained from the effective hamiltonian ( [ h_eff ] ) . the shape of the potential is well reproduced . the dressed potentials shown in ( b)-(d ) have at least two interesting features : ( i ) their overall strength scales like @xmath137 or @xmath138 , it can thus easily be adjusted by choice of laser parameters and even manipulated time - dependently . ( ii ) the shape of the potentials can be modified beyond a simple attractive or repulsive form @xmath139 , due to the appearance of avoided crossings . for example the attractive branch of potentials in the ground - state manifold , shown in fig . [ spectrum_deltaeq ] ( b ) , approaches a finite value for small separations @xmath19 as can be seen from its continuation in fig . [ spectrum_deltaeq ] ( a ) . this could avoid acceleration to high velocities during atomic collisions on the attractive potential , maybe reducing the probability of collisional ionisation @xcite . light induced modifications of potentials that would be strongly attractive in the absence of dressing have already been discussed for the case with a single ground state in ref . @xcite , and proposed as collisional shield for cold polar molecules in ref . however , in our case , once the separation of the atoms is much smaller than the minimal value included in , which is @xmath140 , we have to keep in mind that the simplified four - state model breaks down : the dipole - dipole shift of the dressed states then may become so large that they couple strongly to two - atom states that are not included in our model . let us briefly discuss the case when the detunings have equal magnitudes . as @xmath130 approaches @xmath131 , the asymptotic @xmath141 energy curve ( violet line ) in fig . [ spectrum_deltaeq ] ( a ) is squeezed between the neighbouring ones until the states become degenerate . if we change the sign of one of the detunings and consider the special case @xmath142 , we obtain a particularly symmetric energy spectrum , shown in . as predicted by , the dressing - induced state transfer interaction vanishes between the states @xmath143 and @xmath144 . however symmetrical induced potentials can now be found in the singly - excited state manifold . the strength of these potentials scales as @xmath80 , different from those previously obtained in the ground - state manifold . we will discuss this further in . in this section we revisit dressed dimer dynamics first presented in , to apply the results of . figure [ comparison_pt_fullh ] shows a comparison of the evolution governed by the effective perturbative hamiltonian with that following from the full hamiltonian . already fourth order perturbation theory describes the effective induced hopping period fairly well , as can be seen in panel ( b ) . when we move to the sixth order expressions , which can be found in [ vanvleck ] , the agreement becomes even better : the deviation never exceeds @xmath145 in the time - interval shown . in other simulations we found that perturbation theory agrees better with full calculations if @xmath61 is small , as is expected . even for parameters with poorer agreement between full- and perturbative dynamics ( for example larger @xmath61 or closer to an avoided crossing ) , we typically still obtain the primary features which make these induced potentials interesting : ( i ) we have persistent dressed hopping , ( ii ) the excited state occupancy is suppressed . since we have included the dipole - dipole interaction in the _ unperturbed _ part of the hamiltonian ( [ hunpert ] ) , we do not a priori require it to be small compared to the detuning for the perturbation theory in to be valid . in practice we find that expression ( [ h_eff ] ) works well even for @xmath146 , as long as one stays clear of the avoided crossings at @xmath147 . the large splitting of the eigenstates in the ground state manifold in the vicinity of the avoided crossing , see fig . [ spectrum_deltaeq ] ( b ) , results in comparably fast population - oscillations . the regime where we can find dressed excitation transport on time scales shorter than the life - time of the multi atom state hence is typically entered when @xmath148 approaches @xmath149 , but has not quite reached it yet . the effective population - oscillation - period in is @xmath150ms . here @xmath151 , and the correction factor is @xmath152 due to the vicinity of an avoided crossing . let us reconsider the doubly excited state populations in . for atomic distances where the system is far from all avoided crossings shown in , the doubly excited states would be suppressed by @xmath153 . this gets modified near these avoided crossings . for the distance @xmath65 @xmath2 m as , the system is fairly close to an avoided crossing , thus the doubly excited states are more strongly populated . to conclude this section , we would like to remark on the position degree of freedom of the atoms carrying the interaction . in this article we treat it classically , assuming atoms with a precisely defined separation @xmath19 . in practice , each atom will have a position uncertainty , for example due to the zero - point motion in a harmonic well . the resulting distribution of interatomic distances @xmath19 can lead to a fast de - phasing of population oscillations such as shown in . for the example shown , if we would assume a separation uncertainty of @xmath154 the oscillations de - phase after @xmath155 cycles . we however do not study de - phasing here , since it is not specific to the _ dressed _ interactions . if we considered direct dipole - dipole excitation transport , between the states @xmath15 and @xmath16 only , it would also de - phase after @xmath155 cycles , which however would take place on a timescale of @xmath156 ns due to the by @xmath157 stronger interaction . thus , whether or not de - phasing due to disorder poses a problem depends on the intended application of excitation transport and not on whether or not the interactions arise due to dressing . in particular ref . @xcite discusses an exemplary application where it poses no problem , simulations presented in ref . @xcite _ fully include atomic positioning uncertainties_. also in references @xcite we study systems with interesting combined excitation transport and atomic motion , with interactions that could be interpreted as arising via the techniques considered here . again , de - phasing effects do not pose a problem . the use of chains of rydberg atoms for random walks and other aggregate physics @xcite is limited by spontaneous decay and other incoherent loss processes from the rydberg states . when using a dressing scheme , these decay processes are reduced by construction since the atomic population in rydberg states is kept small . throughout the paper , _ except _ , we thus do not explicitly include loss in the presented dynamical calculations , but instead estimate the life - time of dressed rydberg - aggregates based on their content of actual rydberg population . for example the state @xmath111 will be assigned a total effective decay - rate @xmath158 for two atoms . this scaling of @xmath159 with the dressing parameters follows the same pattern as in vdw dressing @xcite . most of our simulations treat excitation transport via @xmath57 rydberg states of litihum , which have life - times @xmath160 @xmath2s and @xmath161 @xmath2s @xcite . this yields @xmath162ms for @xmath163 and @xmath87 . the corresponding exponential population decay is schematically indicated in fig . [ comparison_pt_fullh ] ( a ) . note , that simple estimates of the life - time of a dressed system using the relations of this section and the results of require excited state populations to be accurately described by perturbation theory . we have seen in that this is the case , even fairly close to the avoided crossings in the spectrum . to ascertain that the simple estimates of loss effects discussed above are valid for our cases , we also calculated results such as shown in with the inclusion of loss and spontaneous decay . consider the density matrix @xmath164 with basis - states as defined in . the prime on the summation symbol indicates that only states within the single exciton manifold ( set @xmath165 in [ multi_exc ] ) are considered . let the time - evolution of @xmath166 be given by the master equation @xmath167 } + \sum_{ak}\bar{\gamma}_{k}{\cal d}[\hat{d}_{k}^{(a)},\hat{\rho } ] + \gamma { \cal l}[\hat{l},\hat{\rho}],\end{aligned}\ ] ] where @xmath168 is the lindblad superoperator @xmath169 = \hat{a}\hat{\rho}\hat{a}^{\dagger } - \hat{a}^{\dagger}\hat{a}\hat{\rho}/2 - \hat{\rho}\hat{a}^{\dagger}\hat{a}/2 $ ] . for @xmath170 define @xmath171 , then @xmath172 acts like @xmath173 in the electronic space of atom @xmath174 and as unity elsewhere . these operators describe spontaneous decay . the superoperator @xmath175 has the form @xmath176 = \hat{a } \hat{\rho}-\hat{\rho}\hat{a}^{\dagger}$ ] , and the operator @xmath177 . here @xmath178 counts the number of excited atoms in the state @xmath32 . the purpose of @xmath175 is to describe loss from the rydberg states to states external to our model @xcite , for example to further rydberg states . for simplicity we assumed an identical loss rate @xmath179 from @xmath9 and @xmath10 . we consider two cases : ( i ) all the population loss from the rydberg levels described by the effective decay rates @xmath180 from @xcite , including black - body effects , ends up in the respective ground - states . here we set @xmath181 and @xmath182 . this is a worst case scenario for our purposes , as all incoherent processes within our four state model de - cohere excitation transport . ( ii ) we split @xmath183 from @xcite into the zero temperature component @xmath184 and the black - body induced component @xmath185 via @xmath186 . this yields @xmath187 @xmath2s , @xmath188 @xmath2s , @xmath189 @xmath2s , @xmath190 @xmath2s . we then assume that population returns to their respective ground - states with a rate @xmath191 , while black - body redistributed population is lost , either due to ionisation or because it leaves the single exciton manifold . ] . we thus set @xmath192 . these calculations , shown in , verify that loss out of the system from the rydberg states @xmath9 and @xmath10 has indeed no other consequences than an overall population decay . if spontaneous decay in the channels @xmath193 and @xmath194 is included , population oscillations additionally show a de - phasing . neither effect is dramatic on the time - scales considered here . in we have pointed out that dressing induced dipole - dipole interactions can be created in the ground - state manifold ( between the states @xmath143 , @xmath144 ) or in two different excited state manifolds ( between the states @xmath195 , @xmath196 or @xmath197 , @xmath198 ) . the latter requires significantly different detunings @xmath130 and @xmath131 ( such like @xmath142 ) to energetically separate the two different singly - excited state manifolds . we have explicitly verified population - oscillations for example between the states @xmath199 and @xmath200 . to know whether the ground or singly - excited state manifold would be more useful for a specific application it is important to consider the life - time of the dressed multi - atom state : the smaller @xmath61 , the longer the life - time , but also the longer the effective hopping period @xmath201 required for a complete population - oscillation that transfers the excitation from one atom in the dimer to its counter - part and back . here , the interaction @xmath202 denotes the off - diagonal entry in the effective hamiltonian of the corresponding manifold , e.g. @xmath203 in for the ground - state manifold . here we estimate the life - time of dressed states more roughly than described in , assuming that ground - state atoms ( @xmath7 , @xmath8 ) do not decay and excited state atoms ( @xmath9 , @xmath10 ) decay with the same rate @xmath204 . for a system of @xmath3 atoms in various states we then again add the decay rates . @xmath202 & @xmath205 + . @xmath206 & @xmath207 & @xmath93 & @xmath208 + . @xmath209 & @xmath210 & @xmath211 & @xmath212 + . @xmath213 & @xmath214 & @xmath215 & @xmath216 + for three different @xmath3-atom states , the life - times determined in this way and the ratio of life - time and hopping period @xmath205 are listed in . as long as @xmath217 , the ratio @xmath218 for the singly - excited state manifold can be larger than for the ground - state manifold . whether dipole - dipole interactions induced through dressing are advantageous over direct dipole - dipole interactions among rydberg states , strongly depends on the problem at hand , as we discuss now for three examples . ( i ) exciton migration : _ for the migration of a single or multiple excitons on a rigid chain , as studied in @xcite , the ratio @xmath218 introduced above gives a direct measure of how many sites the excitation can traverse within the life - time of the whole chain . evidently this measure is never improved by the present dressing technique , whose advantage for this scenario thus only lies in reduced ionization probabilites and simpler atom trapping . in we show an exemplary case where the overall life time in the presence of dressing is long enough to exploit these features . _ ( ii ) adiabatic transport : _ there are dynamical scenarios whose time - scale of interest is not directly given by @xmath219 , such as the adiabatic entanglement and excitation transport that we reported in @xcite . it can be seen that the ratio of the time - scale of transport dynamics to spontaneous life - time of the chain can be improved by factors of about 2 by working in the dressed singly - excited - state manifold . _ ( iii ) ring aggregates : _ in we consider systems of three dipole - dipole interacting atoms trapped in effectively one - dimension on a ring . for such a construction dressing can be highly beneficial . after choosing the dimer as the simplest case to work out the basic details of dressed dipole - dipole interactions , we now briefly present two exemplary applications of our results to larger systems . first , in , we consider a dressed ring - trimer ( @xmath220 atoms confined on a ring ) , a system of interest due to conical intersections of the adiabatic energy surfaces @xcite . then , in , we examine a dressed exciton on an atomic chain with @xmath221 . in ref . @xcite we show that circular rydberg trimers , consisting of three rydberg atoms trapped in one dimension on a ring , exhibit interesting quantum dynamics near conical intersections ( cis ) . for practical realisations , however , dipole - dipole forces are typically _ too large _ for ring confinement . in this case the reduction of interaction strengths , which limits the parameter range where dressing is beneficial for exciton migration , turns into a benefit . as dressing also increases the system s life - time , it greatly facilitates the practical creation of dipole - dipole ring trimers . consider three atoms tightly confined in one dimension on a ring , as sketched in fig . [ trimer : plots ] ( a ) . achieving such confinement is the first point in this section that is simpler for dressed ground state atoms than bare rydberg atoms . on the ring , the bare interaction between atoms @xmath25 is fully determined by the two relative angles @xmath222 and @xmath223 shown in the sketch . adiabatic energies , defined by , then form 2d surfaces @xmath224 . in the absence of laser - couplings , the three surfaces spanned by @xmath225 , @xmath226 , @xmath227 exhibit a conical intersection @xcite , where two of them become exactly degenerate , at @xmath228 . in the presence of laser - couplings , this feature is imprinted also onto surfaces spanned predominantly by ground states @xmath229 , @xmath230 , @xmath231 . conical intersections are of great interest in particular in chemical physics , as they strongly affect the outcome of photo - chemical reactions @xcite . in cold atomic gases of rydberg dressed atoms , they result in interesting non - adiabatic and geometric phase effects @xcite . for a detailed description of this system and the concept of conical intersections we refer to ref . @xcite and references therein . for the purpose of the present article , it is more important to point out that the techniques worked out here are highly beneficial for the practical realisation of this kind of ultra cold conical intersections . this conclusion is reached through an extensive but technical survey of parameter space . we defer full details to a specialised publication , and present here only the core points : in order to realise the scenario sketched in fig . [ trimer : plots ] ( a ) , initially at least three conditions have to be met : + our interest is in dynamics near the conical intersection , where the atoms form an equilateral triangle and are separated by @xmath232 , where @xmath19 is the ring radius . comparing the ( possibly dressed ) interaction strength with the strength of the ring trap , @xmath233 , we have @xmath234 , thus preventing either too small ring radii @xmath19 or to large interaction strength @xmath2 and hence principal quantum numbers @xmath235 . + we can estimate the order of magnitude of the ci transit time @xmath236 from the time - scale set by the classical equations of motion @xmath237 , which holds for a dimer of angular separation @xmath238 . this time scale is @xmath239 . adiabatic dynamics must fulfill @xmath240 , with @xmath219 as in . + thus @xmath241 . + attempting to fulfill the inequalities ( a - c ) for various atomic species leads to two main conclusions : ( i ) lighter atoms are favourable . ( ii ) without dressing ( @xmath242 ) , principal quantum numbers @xmath235 would be too high to avoid strong effects of black - body radiation . if we consider a dressed system however , one more constraint comes into play : + this is to stay clear of avoided crossings such as seen in . a set of parameters where these conditions are met is : @xmath243 , @xmath244 @xmath2 m , @xmath245 mhz and @xmath246 . were given as angular frequency @xmath247 . ] now we consider the dynamical evolution of an eigenstate of , when the interaction parameters @xmath248 are varied in time in a manner representative for near ci dynamics . to this end we extracted the trajectory @xmath249 from full quantum mechanical solutions of the time - dependent schrdinger equation in ref . @xcite , shown in the inset of fig . [ trimer : plots ] ( a ) as dashed line . a second trajectory is taken slightly offset ( dotted line ) . we find that the quantum state dynamics in both cases follows @xmath123 ( see equation ) well , justifying the simplified model used in @xcite . since the energy - spacing @xmath250 ( see ) for @xmath243 is @xmath251 mhz , we could have chosen a much larger detuning , yielding even better agreement between the perturbative and exact solutions . as we have seen in this section , dressing is threefold beneficial for atom trapping in the construction of conical intersections in dressed rydberg ring trimers : ( i ) it greatly widens the available scope for quasi-1d trapping of the atoms , ( ii ) reduces the involved interaction strength such that trapping can be realistically considered in the first place and ( iii ) simultaneously extends the available life - times . we discuss results for a linear chain of @xmath221 equidistant atoms in the following , considering the same physical situation as in , hence the spacing of atoms in the chain is also @xmath252 m . we create the initial state adiabatically as in . for this section only , we will define our excitation to be the state @xmath7 , such that e.g. @xmath253 . the corresponding dressed states have a longer life - time for larger chains . the dressing - induced excitation transport and a comparison of exact and perturbative evolution is shown in . it can be seen that the excitation is transported over many sites of the chain on time - scales shorter than the expected spontaneous life - time of the dressed @xmath0-atom state ( about @xmath254^{-1}=0.14 $ ] s ) . due to the dressing , the time - scale on which the interaction transports the excitation has been increased by a factor of @xmath255 , compared to the direct use of excited rydberg states . in we argued that dressing typically does not favourably alter the possible number of excitation hops , given by the ratio @xmath205 of life - time and hopping period . nonetheless the results of this section show that dressed excitation migration within the available life - time is possible . despite the detrimental effect on @xmath218 , the dressing still enables time - dependent control of the transport , and , as in the previous section , simplifies trapping efforts . we have shown that the dressing of alkali atoms in the ground - state with rydberg - excitations can yield dipole - dipole interactions with excitation - transfer for the ground - state atoms . this generalises existing results for dressing with rydberg van der waals interactions , that do not entail excitation transport . the scheme proposed here is an adaptation of a similar method , introduced in the context of trapped rydberg ions @xcite , to neutral atomic systems . it makes use of two effective laser couplings to rydberg - states , instead of the one required for van der waals dressing . we find that the dressed dipole - dipole transition matrix element between ground state atoms scales like @xmath153 with the dressing parameter @xmath61 , while the life - time of the dressed state increases like @xmath256 . dressed vdw interactions show the same scaling behaviour . we demonstrate in ref . @xcite that this re - arrangement of life - times and interaction time - scales allows the study of conical intersections using rydberg - dressed atoms on a ring . beyond this , the dressing scheme will enable a larger variety of trapping techniqes for the atoms and reduce the likelihood of ionisation . as further example of the flexibility afforded by the dressing , we show that it is possible to induce effective transport of a single rydberg excitation within a chain of ground state atoms . the amplitude of this process scales like @xmath80 . whether the effective interactions discussed in this article are more useful than the direct use of dipole - dipole interactions depends on details of the situation one studies . @xcite we provide an exemplary scenario where they can be highly beneficial . in this section , we derive the rotating - wave hamiltonian from more fundamental expressions for atom - laser interactions , reviewing text - book material , see e.g. @xcite . the hamiltonian for an atom in the presence of an electric field is @xmath257 this expression is valid in the dipole approximation @xmath258 , when the wavelength of the incident light is much larger than typical atomic distances . the operator @xmath259 is the full atomic hamiltonian in the absence of any external field , @xmath260 is the electron charge , @xmath261 the position operator and @xmath262 the electric field at the atomic nucleus , which is treated classically . consider for the moment a single atom only , with the four relevant states @xmath7 , @xmath8 , @xmath9 , @xmath10 and a level scheme as shown in fig . [ fig : sketch ] ( a ) . these states are eigenvectors of the free atomic hamiltonian @xmath259 , thus @xmath263 with @xmath264 . we add two coupling fields @xmath265 , with @xmath266 . here @xmath267 is a unit vector in the @xmath268-direction . for any two states @xmath269 , @xmath270 we define matrix elements @xmath271 . a general quantum state in the chosen subspace is @xmath272 , with equation of motion @xmath273 & = \left [ \begin{array}{ccccc } e_{g } & -\mu_{gs } { \cal e}_{s}(t ) & 0 & 0 \\ -\mu_{gs}^ { * } { \cal e}_{s}(t ) & e_{s } & 0 & 0 \\ 0 & 0 & e_{h } & -\mu_{hp } { \cal e}_{p}(t)\\ 0 & 0 & -\mu_{hp}^ { * } { \cal e}_{p}(t ) & e_{p } \\ \end{array } \right ] \left [ \begin{array}{c } c_{g}\\ c_{s}\\ c_{h}\\ c_{p}\\ \end{array } \right].\end{aligned}\ ] ] here we have used atomic units . now we change the variables @xmath274 of the coefficient vector to @xmath275 , where @xmath276 are arbitrary constants . if the coefficient vector @xmath277 evolved according to @xmath278 for some matrix @xmath279 , @xmath280 evolves according to @xmath281 with @xmath282 , where @xmath283 and @xmath284 are diagonal matrices with elements @xmath285 and @xmath286 . using the specific transformation vector @xmath287 , we obtain @xmath288 & = \left [ \begin{array}{c } 0 \\ -\frac{1}{2}\mu_{gs}^ { * } { \cal e}_{s } f_{-}(\omega_{s},\nu_{s } ) \\ 0 \\ 0 \\ \end{array } \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \begin{array}{c } -\frac{1}{2}\mu_{gs } { \cal e}_{s } f_{+}(\omega_{s},\nu_{s } ) \\ 0 \\ 0 \\ 0 \\ \end{array } \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \begin{array}{c } 0 \\ 0 \\ 0 \\ -\frac{1}{2}\mu_{hp}^ { * } { \cal e}_{p } f_{-}(\omega_{p},\nu_{p } ) \\ \end{array } \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \begin{array}{c } 0 \\ 0 \\ -\frac{1}{2}\mu_{hp } { \cal e}_{p } f_{+}(\omega_{p},\nu_{p } ) \\ 0 \\ \end{array } \right ] \left [ \begin{array}{c } d_{g}\\ d_{s}\\ d_{h}\\ d_{p}\\ \end{array } \right],\end{aligned}\ ] ] where @xmath289 , @xmath290 and @xmath291 stands for @xmath292 . under the _ rotating - wave approximation _ all terms @xmath293 are dropped , as they are oscillating too rapidly to have any effect . we also introduce the detunings @xmath294 and rabi - frequencies @xmath295 and @xmath296 . finally we perform one further variable - change , using @xmath297 to reach @xmath298 & = \left [ \begin{array}{ccccc } 0 & \frac{\omega_{s}}{2 } & 0 & 0 \\ \frac{\omega_{s}^{}}{2 } & -\delta_{s } & 0 & 0 \\ 0 & 0 & 0 & \frac{\omega_{p}}{2 } \\ 0 & 0 & \frac{\omega_{p}^{}}{2 } & -\delta_{p } \\ \end{array } \right ] \left [ \begin{array}{c } d'_{g}\\ d'_{s}\\ d'_{h}\\ d'_{p}\\ \end{array } \right].\end{aligned}\ ] ] the hamiltonian in ( [ effective_equation3 ] ) is the basic building block of , as long as only a single atom is concerned . even for more than one atom and with the inclusion of dipole - dipole interactions , the above procedure can be followed . instead of the single atom energies in the first transformation vector @xmath299 , we would employ many - atom energies . since the dipole - dipole interaction only couples energetically degenerate states like @xmath15 to @xmath16 , all complex phase - factors can finally be eliminated to arrive at a many - body version of , which is . in practice the scheme displayed in is complicated by selection rules and transitions are constrained by the availability of laser sources . the coupling between the states @xmath300 can be realised with now broadly established two - photon excitation schemes . these transitions are typically near resonant with some auxiliary middle level , such as @xmath301 in , in order to enhance transition amplitudes . for the right choice of parameters , coherent coupling between @xmath302 and @xmath303 is achieved , see e.g. @xcite . for the coupling between the states @xmath304 two - photon transitions are forbidden due to the selection rule @xmath305 . this necessitates one- or three photon coupling schemes , examples for which are included in . let us briefly discuss both options : + _ ( i ) single photon uv - transition : _ direct transitions from ground- to rydberg levels @xcite suffer from extremely small dipole - transition matrix elements . these scale like @xmath306 with the principal quantum number @xcite . from the matrix element @xmath307 @xcite , from oscillator strengths @xmath308 and transition energies @xmath309 using @xmath310 , @xmath311 is the electron mass . ] we extrapolate @xmath312 , where @xmath313 is the bohr radius . assuming @xmath314 mw of laser - power , focussed to a waist @xmath315 @xmath2 m , one achieves a rabi - frequency @xmath316 mhz as employed in this article . the intensity at the focus is @xmath317 , resulting in a rabi - frequency @xmath318 , where @xmath319 is the electric field , and @xmath260 , @xmath320 as well as @xmath321 fundamental constants . laser light at the uv wavelength is typically created through higher harmonic generation , reducing flexibility and making it more challenging to achieve the required power . + _ ( ii ) three photon transition : _ one would probably employ three different transitions , as indicated in , exploiting near resonant auxiliary levels to enhance the coupling amplitude . in principle it would also be possible to utilise three photons stemming all from the same laser . for li such three - photon , single colour , schemes still can proceed near resonant with the intermediate @xmath322 state @xcite . consider for example the excitation chain sketched in yellow and dark blue in , and denote the rabi - frequencies and detunings of the three transition as @xmath323 , @xmath324 with @xmath325 from lower to higher energies . since both intermediate states decay on a timescale of @xmath156 ns , they have to be far detuned with @xmath326 and @xmath327 of the order of @xmath328 for @xmath329 . here this makes sure that the effective decay rate of the intermediate levels is at least not larger than that due to the rydberg state . if we now consider the effective three - photon rabi frequency : @xmath331 we would require @xmath332 ghz to reach @xmath333 mhz as employed here , rather hard to achieve on the weak rydberg transition . there may though still be applications where much smaller effective rabi - frequencies @xmath334 are sufficient and the three - photon scheme has advantages over the uv transition . the hamiltonian given in is the @xmath3-body generalisation of the results of [ rotwave ] with added dipole - dipole interactions . it conserves the exciton - number , represented by the operator ( [ nexcitons ] ) . consequently the hamiltonian has a block - diagonal structure , with each block describing a given number of excitons . for @xmath87 atoms we explicitly have @xmath335,\:\ : m_{0}= \left [ \begin{array}{cccc } 0 & \frac{\omega_s^}{2 } & \frac{\omega_s^}{2 } & 0 \\ \frac{\omega_s}{2 } & -\delta_{s } & 0 & \frac{\omega_{s}^}{2 } \\ \frac{\omega_s}{2 } & 0 & -\delta_{s } & \frac{\omega_{s}^}{2 } \\ 0 & \frac{\omega_s}{2 } & \frac{\omega_s}{2 } & -2 \delta_{s } \\ \end{array } \right ] , \:\ : m_{2}=m_{0}\big\|_{s\rightarrow p}\end{aligned}\ ] ] @xmath336 . \end{aligned}\ ] ] the bases with respect to which the three blocks are written are @xmath337 , @xmath338 , @xmath339 . in we study the nontrivial part , @xmath126 , of this hamiltonian in more detail . the basic goal of van - vleck perturbation theory was outlined in : to find a basis that block - diagonalises the hamiltonian ( [ hamil_dressed ] ) to a given order in the perturbation @xmath340 . this appendix supplies the details necessary to understand the origin of the results presented in and [ sixth_order ] . we first partition our unperturbed basis @xmath341 into two sets @xmath342 and @xmath343 , the first of which shall span the `` system '' space of interest and the second is its complement in the full hilbertspace ( the `` environment '' ) . the specific basis , @xmath341 , used for the definition of system and environment is given in . this basis is also an eigenbasis of @xmath344 . then we can construct projection operators on the system subspace @xmath345 and its complement @xmath346 . here we consider the example where @xmath347 , thus our system space is the ground - state manifold . we then have @xmath348 . with respect to these partitions the hamiltonian matrix , or similarly any other operator , can be divided into four blocks @xmath349 , @xmath350 , @xmath351 , @xmath352 . out of these , we assemble a block diagonal part @xmath353 and a block - off - diagonal part @xmath354 . since @xmath344 is diagonal in its eigenbasis , we have @xmath355 , @xmath356 . we now aim to find a unitary transformation @xmath357 that yields a block diagonal hamiltonian @xmath358 to a given order in @xmath340 , thus @xmath359 . the effective hamiltonian in the space of interest that we seek , is then given by the block @xmath360 . next , we express the unitary transformation operator @xmath357 as @xmath361 with @xmath362 . we then introduce the condition @xmath363 and @xmath364 for the block diagonal and off - diagonal parts of @xmath365 . other choices are possible , the present one distinguishes the van vleck procedure from other related schemes @xcite . for all operators in the problem , we write a series expansion in orders of the perturbation @xmath340 , most notably @xmath366 , @xmath367 , @xmath368 . as outlined in @xcite , it is then possible to obtain recursive relations for @xmath369}$ ] as series of increasingly deeper nested commutators using the baker - campbell - hausdorff formula . by segregating expansion orders in @xmath340 , the equation is then turned into a recursive set of equations for the @xmath369}$ ] . an elegant method to perform this cumbersome routine is outlined in @xcite , to which we refer for further details . ] . one extracts the full matrix @xmath370 from @xmath369}$ ] as follows . let us introduce the notation @xmath371 with latin indices to denote an eigenvector of @xmath344 within @xmath343 and @xmath372 with greek indices for one within @xmath342 . since we demand @xmath363 we only require matrix elements like @xmath373 or @xmath374 in order to know the whole matrix form of @xmath370 . now we define the resolvent operator . @xmath375 which fulfills : @xmath376 } { \|\,{\alpha}\,\rangle}. \label{resolvent_eqn}\end{aligned}\ ] ] for any operator @xmath377 . substituting @xmath378 in and multiplying from the left with @xmath379 we obtain @xmath373 and also @xmath380 . from knowledge of @xmath365 one can infer the expansion orders @xmath381 through commutation relations . a series expansion of @xmath361 to the required order finally yields the perturbed eigenstates via @xmath382 we refer to ref . @xcite for all further technical details . the effective hamiltonian of the dimer in the ground - state manifold spanned by @xmath383 is given in up to fourth order , with odd orders vanishing . here we display the corresponding expression of sixth order , @xmath384 , where we use the definition @xmath385 . we can write @xmath386 with @xmath387 \bigg\ } / ( { { \bar{u } } _ { \mbox{\scriptsize 12}}}^2 - 1)^2 , \label{wpotentials6 } \\ \tilde{u}_{12}^{(6)}(r ) & = \alpha_{s}^2 \alpha_{p}^2 { { u } _ { \mbox{\scriptsize 12}}}\bigg [ \frac { \alpha_{s}^2 + \alpha_{p}^2 } { { { \bar{u } } _ { \mbox{\scriptsize 12}}}^2 - 1 } \cr & + \frac{-2 ( \alpha_{s}^2 + \alpha_{p}^2 ) + 2 { { \bar{u } } _ { \mbox{\scriptsize 12}}}^2 \frac{(\delta_{s}^2 + \delta_{p}^2)}{(\delta_{s } + \delta_{p } ) } \left ( \frac{\alpha_{p}^2}{\delta_{s } } + \frac{\alpha_{s}^2}{\delta_{p } } \right ) } { ( { { \bar{u } } _ { \mbox{\scriptsize 12}}}^2 - 1)^2 } \bigg ] . \label{vpotentials6}\end{aligned}\ ] ] recall that@xmath388 and @xmath389 . finally , for both cases shown in , the effective hamiltonian of has to be adjusted for @xmath390 . in terms of the basis @xmath391 we obtain : @xmath392 , \\ \tilde{u}_{jj}(r ) & = \alpha_{s}^2 \alpha_{p}^2 \left ( \sum_{k\neq j } \frac{1}{1 - { { \bar{u } } _ { \mbox{\scriptsize kj}}}^2 } \right)(\delta_{s } + \delta_{p } ) \delta_{ij } \cr & + \alpha_{s}^2 \alpha_{p}^2 \frac { { { u } _ { \mbox{\scriptsize ij } } } } { 1 - { { \bar{u } } _ { \mbox{\scriptsize ij}}}^2 } ( 1-\delta_{ij}).\end{aligned}\ ] ] 10 [ 1]`#1 ` urlstyle [ 1]doi:#1 m. d. lukin , m. fleischhauer , r.ct , l. m. duan , d. jaksch , j. i. cirac and p. zoller ; _ dipole blockade and quantum information processing in mesoscopic atomic ensembles _ ; phys . lett . * 87 * 037901 ( 2001 ) . h. schempp , g. gnter , c. s. hofmann , c. giese , s. d. saliba , b. d. depaola , t. amthor , m. weidemller , s. sevinli and t. pohl ; _ coherent population trapping with controlled interparticle interactions _ ; phys . lett . * 104 * 173602 ( 2010 ) . v. bendkowsky , b. butscher , j. nipper , j. b. balewski , j. p. shaffer , r. lw , t. pfau , w. li , j. stanojevic , t. pohl and j. m. rost ; _ rydberg trimers and excited dimers bound by internal quantum reflection _ ; phys . * 105 * 163201 ( 2010 ) . i. i. beterov , i. i. ryabtsev , d. b. tretyakov and v. m. entin ; _ quasiclassical calculations of blackbody - radiation - induced depopulation rates and effective lifetimes of rydberg @xmath393 , @xmath394 , and @xmath395 alkali - metal atoms with @xmath396 _ ; phys . a * 79 * 052504 ( 2009 ) . a. v. gorshkov , p. rabl , g. pupillo , a. micheli , p. zoller , m. d. lukin and h. p. bchler ; _ suppression of inelastic collisions between polar molecules with a repulsive shield _ ; phys . * 101 * 073201 ( 2008 ) . m. reetz - lamour , t. amthor , j. deiglmayr and m. weidemller ; _ rabi oscillations and excitation trapping in the coherent excitation of a mesoscopic frozen rydberg gas _ ; phys . lett . * 100 * 253001 ( 2008 ) . l. j. radziemski , r. engleman , jr . and j. w. brault ; _ fourier - transform - spectroscopy measurements in the spectra of neutral lithium , @xmath402li i and @xmath403li i ( li i ) _ ; phys . rev . a * 52 * 4462 ( 1995 ) . g. d. stevens , c. iu , t. bergeman , h. j. metcalf , i. seipp , k. t. taylor and d. delande ; _ precision measurements on lithium atoms in an electric field compared with r - matrix and other stark theories _ ; phys . rev . a * 53 * 1349 ( 1996 ) . y. ralchenko , a. e. kramida , j. reader and nist asd team ( 2010 ) ; nist atomic spectra database ( ver . 4.0.1 ) , [ online ] . available : http://physics.nist.gov/asd [ 2011 , may 27 ] . national institute of standards and technology , gaithersburg , md .	we show how to create long range interactions between alkali - atoms in different hyper - fine ground states , with the goal of coherent quantum transport . the scheme uses off resonant dressing with atomic rydberg states . we demonstrate coherent migration of electronic excitation through dressed dipole - dipole interaction by full solutions of models with four essential states per atom and give the structure of the spectrum of dressed states for a dimer . in addition we present an effective ( perturbative ) hamiltonian for the ground - state manifold and show that it correctly describes the full multi - state dynamics . we discuss excitation transport in detail for a chain of five atoms . in the presented scheme , the actual population in the rydberg state is kept small . dressing offers large advantages over the direct use of rydberg levels : it reduces ionisation probabilities and provides an additional tuning parameter for life - times and interaction - strengths .
although laser cooling generally assumes the presence of an almost closed transition @xcite , it is now possible to laser cool in a magneto - optical trap ( mot ) atomic species whose internal level structure is more complex than that of alkali atoms . a more complex electronic structure may lead to qualitatively new physics . in the case of @xmath3cr , for example , for which bose - einstein condensation was reached recently in the group of t. pfau in stuttgart , germany @xcite , anisotropic , long - range dipole - dipole interactions play a visible role in the expansion of the bec after it is released from its trap @xcite - contrary to all previous experimental data on becs , for which all interaction effects could be modeled by short - range isotropic interactions . in addition , the optical manipulation of new atoms may find technical applications , which is the case for chromium ( cr ) @xcite , but also for example , for group iii atoms ( such as al , ga , in ) and for fe @xcite . the optical manipulation of species with complex electronic structure can nevertheless be a technical challenge when the cooling transition is not closed , since more and more repumping lasers are then needed . for example , laser cooling of molecules has never been investigated because of the many vibrational levels associated with a given electronic transition . here , we report on the magneto - optical trapping of fermionic @xmath0cr atoms . as with bosonic @xmath3cr atoms , the cooling transition is not perfectly closed and atoms are depumped towards metastable dark states , so that two repumpers are required to effectively close this transition . in addition , and in contrast to @xmath3cr which has no hyperfine structure , @xmath0cr nuclear magnetic moment is @xmath4 , and its electronic ground state has four hyperfine components . further repumpers are therefore needed , to prevent optical pumping towards dark hyperfine states . obtaining a mot for @xmath0cr atoms is appealing for several reasons . starting from a laser - cooled atomic vapor , it has been possible to reach quantum degeneracy for nine bosonic atomic species , which is one of the major achievements of atomic physics in the last ten years . however , up to now only two fermionic species , @xmath5k and @xmath6li , were cooled down to quantum degeneracy @xcite . for both species , it was possible to reach the strongly interacting regime by use of feshbach resonances , and to study the cross - over between the bcs and the bec regimes . it is therefore interesting to add a new species in the ultra - cold fermionic playground . the main specificity of chromium is its large magnetic moment of 6 @xmath7 : in polarized degenerate fermi seas , dipole - dipole interactions will be the leading interaction term , which is expected to produce new quantum phases @xcite . in addition , the atomic ground state hyperfine structure together with its strong magnetic dipole moment should lead to a rich feshbach resonances spectrum . finally we report in this paper that it is possible to simultaneously trap @xmath0cr and @xmath3cr atoms , and discuss the perspectives to produce chromium bose - fermi mixtures in the quantum regime . figure 1 shows a simplified energy - level diagram for @xmath3cr and @xmath0cr atoms . the @xmath8s@xmath9p@xmath10 transition ( @xmath11 = 425.51 nm , natural linewidth @xmath12 5.02 mhz , saturation intensity @xmath13 = 8.52 mw.@xmath14 ) is used both to decelerate the two isotopes in a zeeman slower , and to cool and trap them in a mot . since no suitable diode laser source is yet available at 425 nm , we generate 350 mw of blue laser light by frequency doubling high - power light at 851 nm using a non - linear 8 mm long lbo crystal in a resonant optical cavity . the light at 851 nm is produced with an argon - ion laser ( 10.5 w ) pumping a commercial ti : sapphire laser from tekhnoscan . the infrared power output is typically 1.3 w. the frequency of the ti : sa laser is stabilized by locking it to a fabry - perot ( fp ) reference cavity ( whose finesse is 100 ) . the fp spacer is made of invar steel . the cavity is housed in an evacuated chamber , which is isolated acoustically and thermally . the length of the fabry perot cavity is locked to the @xmath8s@xmath15p@xmath10 transition of @xmath3cr by use of a standard saturation spectroscopy setup on an hollow cathode lamp . the doubling cavity is kept in resonance with the infra - red light using a hnsch - couillaud locking technique @xcite . we estimate the blue laser jitter to be around one mhz . several acousto - optic modulators ( aoms ) are necessary to generate the laser frequencies needed for slowing and trapping both chromium isotopes ( see fig [ schemaniveaux ] ) . the @xmath8s@xmath9p@xmath10 transition is slightly leaky since excited atoms can decay from @xmath8p@xmath10 to dark @xmath16d states ( see fig [ schemaniveaux ] and @xcite ) . to compensate for that effect , repumping lasers can be used to pump atoms from the metastable d states back to the ground @xmath8s@xmath17 state via the @xmath8p@xmath17 state . their wavelength are in the range of @xmath18 nm and are thus referred as the red repumpers in the following . in our setup they are produced by external cavity laser diodes , each locked to the fabry - perot reference cavity using a pound - drever - hall scheme @xcite . our chromium oven is a commercial addon high temperature effusion cell . the realization of a robust chromium source is a technical challenge . indeed , chromium is a very refractory metal , with a melting point of 1857 celsius . furthermore , cr tends to react with most of the materials used to construct high temperature ovens . for example , direct contact with tantalum ( ta ) or tungsten ( w ) must be avoided because cr forms alloys with low temperature melting with them . in our setup , the cell is resistively heated with two self supporting w filaments . the main crucible is in ta . a chromium bar of 20 g ( 99.7 percent pure ) is enclosed in a second crucible made of ultra pure alumina , which is inserted inside the ta crucible . the temperature is measured with a w / re thermocouple . a water cooled thermal shield maintains the vacuum chamber at a moderate temperature of 40 celsius . the effusive source is terminated by a 2 mm diameter aperture . a 4 mm diameter aperture , set at a distance of 5 cm from the emission point , defines the direction of the atomic beam . the cell works in the horizontal plane and is connected to the chamber through a cf40 orientable flange . this provides the necessary fine tuning for the orientation of the atomic beam in order to align the thermal beam along the axis of the one meter long zeeman slower tube . the atomic beam can be switched off and on within 200 ms with a mechanical shutter . the oven chamber is pumped with a 250 l.s@xmath19 turbo - pump . we usually work with an oven at a temperature of 1500 celsius , and the pressure in the oven chamber is then equal to @xmath20 mbar . a zeeman slower ( zs ) connects the oven chamber to the mot chamber . it is made of three sets of copper coils wrapped around a one - meter - long steel tube , which has an internal diameter of @xmath21 cm , and is connected to the oven chamber through a @xmath22 cm - long flexible hose . at the entrance of the zs tube , a @xmath23 cm - long differential pumping tube having a @xmath24 cm internal diameter allows limiting pressures in the experimental chamber below @xmath25 mbar when the oven is on . the experimental magnetic field profile achieved is a usual positive- to- negative square root profile , allowing one to get a well - defined final velocity ( see metcalf for a discussion on the extraction at the end of a zs , and @xcite for a recent and detailed paper on a zs ) . it consists in three parts : the `` branching zone '' at the beginning of the tube where the magnetic field rises and quickly reaches its maximum intensity b@xmath26 ; the `` slowing zone '' with a smooth decrease towards b@xmath27 ; and a `` final decrease zone '' from b@xmath28 to zero ( these three zones are shown in the insert of fig [ planmanip ] ) . the small diameter of the final coil makes it possible to obtain a short `` final decrease zone '' . this allows us to produce the mots close to the exit of the tube ( @xmath22 cm ) , which reduces losses due to transverse expansion at the end of the slower . the experimental magnetic b field values are b@xmath29 g and b@xmath30 g , which , together with the detuning of the zs laser beam ( @xmath31 mhz ) , define a capture velocity @xmath32 of @xmath33 m.s@xmath19 , and a final velocity of @xmath34 m.s@xmath19 . at the exit of the oven , the populations of the fermionic isotope ( respectively bosonic ) are equally distributed into the @xmath35 ( @xmath36 ) zeeman sublevels of the four ( one ) hyperfine ground states ( ground state ) . only the atoms reaching the `` slowing zone '' in the @xmath37 ( @xmath38 ) zeeman sublevel can be slowed down in the zs . in order to get a theoretical estimate of the fluxes of atoms slowed in the zs , we carried out optical pumping calculations for both the bosonic and fermionic isotopes . while the bosonic case is easy to deal with , the fermionic one is much more complicated . first the number of levels ( @xmath35 in the @xmath39 and @xmath40 in the @xmath41 ) is large . in addition , the zeeman energy shifts become comparable with the hyperfine structure for low values of the b field ( a few gauss for the excited @xmath41 state ) , so that the calculations must involve the true eigenstates , which are neither the hyperfine @xmath42 states nor the zeeman @xmath43 states . we found that most of the @xmath44 atoms whose velocity is smaller than or equal to @xmath32 are optically pumped into the @xmath45 state by the zs slowing beam inside the `` branching zone '' . on the contrary , no substantial accumulation in the @xmath46 state is expected for the @xmath47 atoms as they travel through the `` branching zone '' ( @xmath48 instead of @xmath49 ) . taking into account both this different behavior with respect to optical pumping and the isotopic proportion in natural chromium , we expect the ratio of the flux of slowed @xmath47 atoms to the flux of slowed @xmath44 atoms to be equal to @xmath50 . the experimental ratio , deduced from the analysis of mot loading sequences , is larger , almost equal to @xmath51 at 1500 celsius . it should be noted though that less laser power is available for the boson manipulation than for the fermion in our experiment . in the zs , the @xmath47 atoms experience a bad crossing pattern : for b@xmath52 g , the two excited eigenstates adiabatically connected respectively to @xmath53 and to @xmath54 at b@xmath55 g are degenerate , so that any @xmath56 polarized zs laser depumps the atoms towards the @xmath57 ground state . in our experiment we clearly do not expect the zs laser beam to be perfectly @xmath58 polarized : we use an undervacuum metallic al mirror to reflect the zs beam and align it along the chromium beam ( see fig [ planmanip ] ) ; this mirror is coated by chromium as time goes on , and its reflectivity properties slowly change in time @xcite . optical pumping calculations show that about @xmath59 of the atoms can be lost if @xmath60 of the zs power is @xmath61 polarized . to repump @xmath0cr atoms in the zs , we use the @xmath3cr zs beam , which is ( at 25 g ) resonant with the transition between the states adiabatically connected to @xmath62 and @xmath63 ( at b=0 g ) . experimentally we got the optimal @xmath0cr mot loading rates with a power of @xmath64 mw in the @xmath3cr zs beam , and @xmath65 mw in the @xmath0cr zs beam . the @xmath66 radius of the @xmath3cr zs beam and of the @xmath0cr zs beam is respectively 3 mm and 3.6 mm at the mot position , and both beams are approximately focused at the oven position . the presence of the @xmath3cr zs beam as a repumper in the fermionic zeeman slower results in an increase of about @xmath67 in the final number of atoms in the @xmath0cr mot . we also performed a standard 1d transverse cooling in the oven chamber along the horizontal direction , using a laser having the same frequency as the mots lasers . in order to compensate for the red detuning of these lasers we applied a b field of a few gauss along the transverse cooling beam direction . in the optimal experimental configuration , the transverse cooling laser has a power of 10 mw , and induces a gain of @xmath68 to @xmath69 in the steady state mot populations . finally we obtained mot loading rates of the order of @xmath70 @xmath71 atoms.s@xmath19 for @xmath44 @xmath72 at @xmath73 @xmath74 . our experimental chamber is a compact octogonal chamber with eight cf40 viewports in the horizontal plane , and two cf100 viewports for the coupling of vertical laser beams ( see fig [ planmanip ] ) . a 150 l.s@xmath19 ion pump and a ti - sublimation pump maintain the pressure at about @xmath25 mbar . two horizontal external coils in anti - helmholtz configuration create a quadrupole field at the mot position . the mot laser beam setup uses two independent arms : a vertical beam is retroreflected , and a single horizontal retroreflected beam provides all four horizontal cooling beams ( as shown in fig [ planmanip ] ) . the atomic clouds are imaged both on a photomultiplicator ( pm ) , and a ccd camera with a @xmath75 cm focal length lens . as a first step , we report the achievement of a magneto - optical trap with bosonic @xmath3cr atoms . after transverse optical cooling , the atoms are slowed in the zs , and trapped in the mot . because @xmath3cr has no hyperfine structure , a single frequency is sufficient to slow the atoms in the zs , and another to cool and trap atoms in a mot . the vertical magnetic field gradient is 18 g.cm@xmath19 , the mot beams are typically detuned by a few linewidths below the atomic resonance , and their @xmath66 radius is 7 mm . a usual calibration ( see for example @xcite ) of the mot fluorescence measured on the pm , which takes into account the collection solid angle of the imaging lens and the 3/7 average squared clebsch - gordan coefficient of the trapping transition , allowed us to estimate the following numbers . our loading rate is up to @xmath70 atoms per second , at an oven temperature of 1500 celsius . we measured a temperature of @xmath77 @xmath78 from the free expansion of the atomic cloud released from the mot . we obtained a maximum atom number of @xmath79 @xcite , corresponding to a peak atom density of 8@xmath80 @xmath81 ( with a 30@xmath82 systematic uncertainty ) . these results mainly reproduce similar measurements made in the groups of j. mcclelland at nist in gaithersburg , ma , usa , and t. pfau ( @xcite,@xcite ) . first , we observed a strong one - body loss , due to the fact that the cooling transition is not perfectly closed . atoms in the @xmath83 state can undergo an intercombination transition , which leaves them in either the @xmath84 or the @xmath85 states . these metastable dark states are insensitive to the mot light , but can remain trapped in the mot b field gradient . we checked that atoms in these two states can be repumped ( via the @xmath8p@xmath17 state ) , using either one of two single mode extended cavity laser diodes running at 653.97 nm and 663.18 nm , which leads to a slight increase in the total number of atoms in the steady state mot . the second striking feature of @xmath3cr mots is the strong two - body inelastic loss rate , which was measured in @xcite . we reproduced the measurements reported in this reference , and found an inelastic loss rate of @xmath86 @xmath87.s@xmath19 , when the mot beams are detuned by 10 mhz from resonance , and have a total intensity of 116 mw.@xmath14 . we briefly describe the experimental procedure below . the fermionic @xmath0cr isotope has already been cooled by collisions with a cold buffer gas @xcite , but we are the first group to report on properties of the magneto - optical trap of this isotope , which is the first important new experimental result of this paper @xcite . the optical manipulation of @xmath0cr is not a straightforward extension of the work concerning @xmath3cr . the main difference between these isotopes is the complex hyperfine structure of @xmath0cr , whose electronic ground state has four different hyperfine components . although it may be expected that up to three repumpers should be needed to operate the mot , as well as the zs , the experiment shows that only two repumpers are needed in the mot ( repumper @xmath88 and @xmath89 in fig [ schemaniveaux ] ) , and only one in the zs . all these repumpers are derived from a single laser beam , using aoms . in addition to these blue repumpers , three red repumpers are required to prevent atoms from ending up in the metastable states @xmath90 or @xmath91 . due to laser availability when we performed the reported experiments , we were only able to repump atoms from the @xmath92 state . we first achieved a mot with @xmath0cr without any red repumper . atoms exiting the oven are transversally cooled by a near resonant @xmath93 to @xmath94 laser beam orthogonal to the atom beam , before being decelerated in the zs . when no transverse cooling is achieved before the atoms enter the zs , the mot atom number is reduced by typically 60 percent . as explained earlier , a blue repumping laser detuned by 300 mhz from the slowing transition is mixed with the zs laser beam to prevent losses at the bad crossing point of the zs . typically a few mw is enough for this repumper , and without it the number of atoms in the mot is only reduced by 30 percent . the mot blue beams consist in three different frequencies . the cooling beam is typically detuned by a few linewidths from the atomic transition . two blue repumpers @xmath88 and @xmath89 respectively correspond to the transition @xmath95 to @xmath96 and to the transition @xmath97 to @xmath98 ( see fig [ schemaniveaux ] ) . no mot could be obtained without @xmath88 , whereas if @xmath89 is removed , the number of atoms in the mot is only reduced by 30 percent . we inferred from this observation that repumping on the transition @xmath99 to @xmath100 would not significantly increase the number of atoms in the mot . the power in @xmath88 and @xmath89 is respectively 10 mw and 3 mw . the @xmath66 radius of the @xmath0cr mot cooling beams is 4.5 mm , and the @xmath66 radius of @xmath89 is 5.2 mm . as for @xmath3cr , the mot magnetic field gradient is 18 g.cm@xmath19 , and a fluorescence analysis ( with an average squared clebsch - gordan coefficient of 2/5 for the trapping transition ) was used to get quantitative data . > from the depletion rate of the mot , we estimated the transition probability from @xmath8p@xmath10 to the metastable states @xmath16d@xmath101 to be 280 s@xmath19 ( with an uncertainty of @xmath102 mainly of statistical origin ) which is significantly higher than the corresponding number for the boson ( 169 s@xmath19 , see fig [ schemaniveaux ] ) . we were able to partially cancel these losses with the available red repumper ( repumping from the @xmath16d@xmath10 state ) . the steady state atom number in our mot is multiplied by a factor of 2 in presence of this repumper , and the repumping efficiency saturates for only 2 mw of red light ( with a @xmath66 radius of 2.9 mm ) . no significant increase in the mot fluorescence was observed when an other red beam , detuned by the 225 mhz hyperfine gap of the @xmath16d@xmath10 state shown in fig [ schemaniveaux ] , was added . we performed systematic studies to measure the total number of atoms in the mot in presence of the red repumping laser , as a function of the cooling laser intensity and detuning . figure [ figdet ] shows that the mot atom number and peak density reach a maximum for different values of the mot laser detunings . we observed up to @xmath103 atoms in the @xmath0cr mot . this steady state number is significantly higher than what is expected from a direct comparison with the boson case , if only the relative loading rate ( about 2@xmath82 ) and loss rate of the two mots are taken into consideration , showing the dramatic effect of inelastic collisions for the denser bosonic mot . figure [ figint ] demonstrates a saturation of the mot atom number with the mot laser intensity , while the peak density does nt saturate for the laser power available . the highest peak atomic density obtained was up to 2.5@xmath10410@xmath105 @xmath81 ( with a systematic uncertainty of 30@xmath82 ) . even with one red repumper on , the maximum number of @xmath0cr atoms in the steady state mot is below 10@xmath6 , which is most likely not sufficient to reach quantum degeneracy via an evaporative cooling procedure . as described below , a very large inelastic rate coefficient in the @xmath0cr mot , similar to the one measured for @xmath3cr , limits the maximum number of atoms in the steady state mot . therefore , we studied , similarly to the work reported in @xcite for the bosonic isotope , how fermionic metastable atoms can be accumulated in the magnetic trap formed by the quadrupole field of the mot , before being repumped back into the ground state . a similar accumulation process has been reported for strontium atoms in @xcite , but requires a much higher value of the b field gradient . chromium offers a nice way to decouple cooling and trapping : as stated above , atoms are slowly optically pumped into metastable states , whose lifetimes are very large ( and unknown ) , and whose magnetic moments are large enough to be trapped by the mot gradient magnetic field , provided they happen to be produced in a low - field seeking state . in our experiment , we first reproduced the results of @xcite , and used the same procedure to accumulate metastable @xmath0cr atoms in the magnetic quadrupole field of the mot . in the case of @xmath3cr , the number of trapped metastable atoms saturates in a timescale which is fixed by inelastic collisions with atoms in the excited @xmath8p@xmath10 state @xcite . in the case of @xmath0cr mot , the steady state number of atoms in the @xmath8p@xmath10 state is about ten times smaller , and we found that the timescale for accumulating @xmath0cr metastable atoms in the magnetic trap is set by collisions with fast atoms coming from the oven . it is found to be on the order of 8 s when the oven is operated at 1500 celsius , which is equal to the measured lifetime of the metastable fermions in the magnetic trap when the atom shutter is not closed ( and smaller than the background collision lifetime when the shutter is closed ) . by accumulating atoms in the metastable state @xmath8d@xmath10 , and after repumping them using a 10 ms long red pulse , we measured up to @xmath106 @xmath0cr atoms in the magnetically trapped @xmath8s@xmath17 ground state . this number was obtained for a detuning of the mot beams equal to 12.5 mhz , and a total intensity in the mot beams equal to 200 mw.@xmath14 , which corresponds almost to a tenfold increase compared to the steady state mot atom number for these parameters . this measurement was achieved close to resonance , and should not correspond to the optimal @xmath0cr atom number accumulated in the metastable state ( see fig [ figdet ] ) . in addition , such a number was measured by using only one red repumper , and the use of an other red repumper for the @xmath16d@xmath17 state should give us a reasonable starting point for further cold collision experiments , feshbach resonance studies , and for evaporative cooling down to quantum degeneracy . this larger number of @xmath0cr atoms allowed us to carefully study density dependent losses for the @xmath0cr mot . the two - body loss rate parameter measured for @xmath0cr is very large , typically ranging from @xmath107 to @xmath108 @xmath87.s@xmath19 for detuning of the mot laser varying from -20 mhz to -3mhz , and a total mot laser intensity of 170 mw.@xmath14 . these values are similar to the inelastic loss rates obtained for @xmath3cr mots . to our knowledge , these very high inelastic loss rates in chromium mots are not understood up to now , and will be studied in a forthcoming publication . here , we note that they are only slightly smaller than the langevin rate ( see @xcite ) . in this paper , we report on the first simultaneous magneto - optical trapping of @xmath3cr and @xmath0cr atoms . to perform this experiment , we make use of a fortuitous quasi coincidence between two transitions : the cooling transition for @xmath3cr and the @xmath88 transition for @xmath0cr . as a consequence , the cooling beam for @xmath3cr is used in the @xmath0cr mot as @xmath88 . another consequence is that the zeeman repumper of @xmath0cr is also simultaneously used as the zs beam for @xmath3cr , so that we do not need a separate aom to perform the experiment . figure [ doublemot ] ( a ) shows that we can load a @xmath3cr mot , and after it has reached its steady state number , superimpose on it a @xmath0cr mot by simply switching on the aoms for the cooling transition of the fermion , and @xmath89 . both mots form approximately at the same location according to our ccd fluorescence images . when the @xmath0cr mot laser beams are removed , the number of @xmath3cr atoms immediately regains the same steady state value . this demonstrates that neither these lasers nor the presence ( in this regime ) of @xmath0cr cold atoms induce any substantial loss for @xmath3cr atoms . however , the data in fig [ doublemot ] ( a ) correspond to very few atoms ( about 3@xmath10410@xmath109 @xmath3cr atoms , and 4@xmath10410@xmath109 @xmath0cr atoms ) . in a separate set of experiments , we increased the number of fermionic atoms , by first accumulating them in the trapped metastable states during 10 s. we then loaded a @xmath3cr mot , repumped @xmath0cr atoms from the @xmath110 state , thus superimposing about 7@xmath10410@xmath16 fermionic atoms on the bosonic mot , with an estimated peak density of fermionic atoms of about 4@xmath10410@xmath105 @xmath81 . we then observed substantial loss of @xmath3cr atoms , as shown in fig [ doublemot ] ( b ) , which is a proof that fermionic atoms provide a new loss channel for bosonic atoms . the corresponding loss rate is equal to @xmath111 , where @xmath112 is the interisotope loss coefficient . if the number of fermionic atoms stays almost constant , this rate , from the bosonic atoms point of view , corresponds to a linear loss rate , being equal to @xmath113 , where @xmath114 is the number of bosonic atoms , and @xmath115 is an average fermionic density at the bosonic mot position . @xmath116 is the number of fermionic atoms and @xmath117 is a volume depending on the spatial distributions of the two isotopes . writing the atomic densities as @xmath118 gives @xmath117 in terms of integrals of the normalized density distributions @xmath119 ( @xmath120 , @xmath121 is the peak density ) : @xmath122 experimentally , the mot density distributions are gaussian shaped . with the ccd camera we measured the 1/@xmath123 horizontal and vertical radii of the two mots ( @xmath124 @xmath125 m , @xmath126 @xmath125 m , @xmath127 @xmath125 m and @xmath128 @xmath125 m ) , and the small center position separation in between them ( @xmath129 @xmath125 m , along the vertical axis ) , with a precision of about @xmath130 for these numbers . > from these values the numerical evaluation of the integrals in ( [ defvbar53 ] ) is straightforward . for the data in fig [ doublemot ] , the number of @xmath3cr atoms is small enough that we can neglect light assisted collisions among them , so that the evolution equation for the @xmath3cr mot population therefore reads : @xmath131 where @xmath132 is the loading rate of the @xmath3cr mot , @xmath133 is the depumping time towards metastable states , and the last term is only present when the fermionic mot is applied ( between @xmath134 and @xmath135 in fig [ doublemot ] ( b ) ) . we fitted the loading sequence of the @xmath3cr mot ( @xmath136 ) to deduce @xmath132 and @xmath133 . we then adjusted @xmath112 to reproduce the reduction of @xmath3cr atoms when the @xmath0cr mot is removed at @xmath137 . from this analysis ( similar to the one performed for a k - rb mot in @xcite ) , we estimated the light assisted inelastic loss coefficient between @xmath3cr and @xmath0cr to be @xmath138 @xmath87.s@xmath19 , for @xmath3cr ( @xmath0cr ) mot beams detuned by 10 ( 12.5 ) mhz , and having a total intensity of 70 ( 200 ) mw.@xmath14 . the solid line in fig [ doublemot ] ( b ) is the result of our numerical fit . @xmath112 is on the same order of magnitude than the light assisted loss coefficient that we measured both in a @xmath3cr mot and in a @xmath0cr mot . however , physics of light assisted collisions in the case of a mixture is different than for a single - species mot . for homonuclear molecules , the long range part of the excited molecular potential asymptotically related to the @xmath139 dissociation limit is dominated by resonant dipole - dipole interaction , scaling as @xmath140 . in contrast , for heteronuclear molecules , the excited molecular potential is dominated by a van - der - waals interaction scaling as @xmath141 ( @xcite ) . thus , in a mot composed of two different species , inter - species light - assisted collisions typically occur at shorter interatomic distances than intra - species light assisted collisions . in addition , the @xmath142 coefficient for the molecular potentials related to the ( @xmath3cr,@xmath143 , @xmath0cr,@xmath83 ) dissociation limit is positive . the associated molecular potentials are repulsive and should not lead to light assisted losses in a mot ( as discussed in@xcite ) . therefore , we interpret losses related to light assisted collisions in our double mot as the consequence of a process involving molecular potentials related to @xmath3cr,@xmath83 and @xmath0cr,@xmath143 . however , because the shift between the cooling transitions for @xmath3cr and for @xmath0cr is small , the situation may be more complicated , because the molecular excited potentials should be in @xmath144 at large distance , with a very large @xmath142 , and in @xmath145 at shorter distances . this situation is similar to dipole - dipole interactions in cold rydberg gases @xcite . a complication for continuously collecting large numbers of atoms in our mixed - species mot arises from the fact that the @xmath3cr mot is substantially altered by the presence of the @xmath0cr zs beam . we attribute this effect to the fact that slowed @xmath3cr atoms are pushed by the @xmath0cr zeeman slower beam as they approach the mot region ( as the zs magnetic field goes to zero , the bosonic atoms slowly travel through a region where they are resonant with the @xmath0cr zs laser frequency ) . our observations suggest that the best strategy to obtain large samples of cold @xmath0cr and @xmath3cr atoms is to sequentially accumulate them in the metastable states , before repumping both isotopes in their electronic ground state . we have reported the magneto - optical trapping of cold fermionic @xmath0cr atoms . magnetic trapping of metastable atoms in the quadrupole field of the mot was demonstrated , yielding an increase in the number of atoms by a factor of seven . these features are encouraging as a first step in producing a degenerate dipolar fermi gas . furthermore we demonstrated the achievement of a dual - isotope boson - fermion mot . this opens the way to two isotope collision studies , including the search for inter isotope feshbach resonances , and the obtention of sympathetic cooling . it also opens the way to the potential realization of a quantum degenerate boson - fermion mixture involving dipolar species . acknowledgements : lpl is unit mixte ( umr 7538 ) of cnrs and of universit paris nord . we acknowledge financial support from conseil rgional dile - de - france ( contrat sesame ) , ministre de leducation , de lenseignement suprieur et de la recherche , and european union ( feder - objectif 2 ) . we also thank c. chardonnet , v. lorent and h. perrin for their help in starting this project , as well as j. mcclelland and t. pfau for their many friendly advices . b. de marco and d. s. jin , science * 285 * , 1703 ( 1999 ) , f. schreck et al . lett . * 87 * , 080403 ( 2001 ) , a. g. truscott et al . , science * 291 * , 2570 ( 2001 ) , s. r. granade et al . lett , * 88 * , 120405 ( 2002 ) . cold @xmath146cr , @xmath3cr , and @xmath0cr atoms , at temperatures as low as 10 mk , were also obtained in the group of j. doyle : the atoms were magnetically trapped , after being sympathetically cooled by a cryogenic helium buffer gas . see j. d. weinstein et al . rev . a * 65 * , 021604(r ) ( 2002 ) .	we report on magneto - optical trapping of fermionic @xmath0cr atoms . a zeeman - slowed atomic beam provides loading rates up to 3@xmath1 . we present systematic characterization of the magneto - optical trap ( mot ) . we obtain up to 5@xmath2 atoms in the steady state mot . the atoms radiatively decay from the excited p state into metastable d states , and , due to the large dipolar magnetic moment of chromium atoms in these states , they can remain magnetically trapped in the quadrupole field gradient of the mot . we study the accumulation of metastable @xmath0cr atoms into this magnetic trap . we also report on the first simultaneous magneto - optical trapping of bosonic @xmath3cr and fermionic @xmath0cr atoms . finally , we characterize the light assisted collision losses in this bose - fermi cold mixture .
domain walls are localized structures typical of spatially extended systems with broken phase invariance , where two or more homogeneous states with different phases occupy different patches , the walls being defects connecting two of such states or domains . two different types of walls can exist in systems described by a complex order parameter , namely ising and bloch walls , which differ in the way the phase changes between the two domains as the wall is crossed : an ising wall is characterized by a discontinuous variation of the field phase across the wall , whereas in a bloch wall the phase angle rotates continuously accross the wall . as this rotation can occur in two different senses bloch walls are chiral . ( contrarily , the ising wall is not chiral . ) alternatively , the order parameter is null at the core of an ising wall , while in a bloch wall the order parameter does not vanish at any point . for this reason , ising and bloch walls are often referred to as dark and grey solitons , respectively . when the system dynamics does not derive from a potential other crucial difference between ising and bloch walls refers to their dynamics : ising walls are stationary ( static ) , while bloch walls move with a velocity related to their chirality . both types of domain walls may exist in different parameter regions , and in this case they bifurcate one into another via a nonequilibrium ising - bloch ( ib ) transition @xcite , which can be interpreted as a bifurcation of the wall chirality @xcite . the ib transition has been found in systems of very different nature , such as nematic liquid crystals @xcite or reaction diffusion systems @xcite . in the optical context , the phenomenon has been predicted to occur in type i @xcite and type ii @xcite optical parametric oscillators ; from the experimental point of view it is to be mentioned a related study @xcite where larionova _ et al _ have analysed the dynamics of two - dimensional phase domains in a photorefractive oscillator in a degenerate four wave mixing configuration : _ _ within the same ( closed ) wall they found both bloch type and ising type segments . they did not observe , however , the nonequilibrium ib transition , probably because of the 2d character of their system , which complicates the dynamics of domain walls with curvature effects . in fact the first experimental observation of the nonequilibrium ib transition in optics has been confirmed very recently in the ( transversely ) one dimensional version of the same device @xcite . the possibility of domain walls as stable states has a particular interest in nonlinear optical systems , given the potential use of localized structures in all optical signal processing . in nonlinear optical cavities , the structures develop in the transverse plane , perpendicular to the resonator axis , and can be controlled by external parameters @xcite . optical systems with a kerr nonlinearity have been shown to exhibit a rich spatio temporal dynamics , see @xcite and references therein for details . concerning the nonequilibrium ising bloch transition , the kerr cavity , consisting of an optical resonator filled with a kerr medium and driven by a external coherent field , is a good candidate for exhibiting it , as this system presents the basic requirement of broken phase invariance . the original model proposed by lugiato and lefever @xcite was later extended in @xcite to include the vector character of the light fields , what allows to describe instabilities of the light polarization state . this work was extended in @xcite where the case of elliptically polarized input was considered and in @xcite where two dimensional domain walls in the form of dark - ring solitons were studied . recently these works were generalized by considering the possibility of dichroism and/or birefringence in the optical cavity for the two linear polarization components @xcite . it was shown that large enough cavity anisotropy or birefringence substantially modifies the dynamics of the system . in particular , they allow for the polarization instability to rule pattern formation in self - focusing kerr cavities , something that does not occur in isotropic cavities ( without birefringence or dichroism ) @xcite . moreover , as the polarization instability can be subcritical for large enough dichroism or birefringence @xcite bright cavity solitons can exist , which in this case are polarization solitons @xcite . in @xcite it was also shown that in the limit of large anisotropy ( when the losses of the two polarization components are very different in magnitude ) , the dynamics of the system can be described by a single order parameter obeying a universal equation , namely the parametrically driven , damped nonlinear schrdinger equation ( pdnlse ) . for our purposes the recent prediction @xcite that the pdnlse contains an ib transition and that ising and bloch domain walls can connect either spatially uniform or patterned states is of special relevance . this equation has been derived in different contexts ( see @xcite ) . in particular , in the optical context , the pdnlse has been shown to describe ( apart from the anisotropic kerr resonator @xcite ) the degenerate optical parametric oscillator @xcite and optical fibre loops with parametric amplification @xcite . however , despite the generality of the results obtained in @xcite , their applicability to the kerr cavity is strictly valid only in the case of strong anisotropy , a very restrictive condition . in this paper we consider the more realistic case of a kerr cavity with moderate anisotropy as discussed in sec . [ model ] , where the symmetries of the model and the feasibility of domain walls are discussed . the pattern forming properties of this system are investigated in sec . [ instabilities ] . ising and bloch walls are then shown to be solutions of the system and the ib transition between them is characterized in sec . [ ib ] . for small detunings , the usual scenario of ib transition and bloch wall dynamics is found . for larger detunings , the dynamics of domain walls shows novel features typical of excitable systems , such as spiking and bursting during the domain wall evolution , as discussed in sec . [ spiking ] . the collisions between walls and the formation of bound states is preliminary considered in sec . [ bound ] . finally the main conclusions of the work are highlighted in sec . [ conclusions ] . the system considered in this paper is an optical resonator with plane mirrors filled with an isotropic @xmath0 medium of self - focusing type and driven by a spatially homogeneous linearly polarized coherent field of amplitude @xmath1 ( taken real without loss of generality ) that propagates along the resonator axis @xmath2 . the resonator is anisotropic ( dichroic ) , _ i.e. _ , the two intracavity field polarization components @xmath3 and @xmath4 ( parallel and orthogonal to the input field , respectively ) experience different linear losses ( with associated cavity linewidths @xmath5 and @xmath6 respectively ) . in order to avoid curvature effects that strongly influence domain wall dynamics , we assume a transversely 1d problem ( that can be experimentally achieved with a slab waveguide geometry that confines the fields along one transverse dimension ( say @xmath7 ) , or with rectangular slits placed in appropriate planes @xcite ) . the adimensional model equations for such a system in the mean field limit read @xcite:@xmath8 where @xmath9 and @xmath10 are normalized cavity detunings ( @xmath11 is the angular frequency of the input field and @xmath12 are the frequencies of the cavity longitudinal modes with polarization parallel and orthogonal to the input closest to @xmath11 ) , @xmath13 is the cavity anisotropy parameter ; @xmath14 accounts for diffraction ( the transverse spatial coordinate @xmath15 is normalized to the diffraction coefficient ) and @xmath16 is time normalized to @xmath17 . finally , @xmath18 and @xmath19 are the maker and terhune coefficients , which verify @xmath20 for isotropic media @xcite , which we consider . for details on the normalizations see @xcite . in refs . @xcite , the pattern formation properties of this model have been studied for @xmath21 and @xmath22 . we extended the model to the anisotropic cavity ( @xmath23 ) first considering the plane wave model ( no diffraction ) @xcite and then considering pattern formation @xcite , where we concentrated in the limit of large cavity anisotropy and demonstrated that the pdnlse describes the system in this limit . later , in @xcite we studied patern formation and localized structures due to birefringence ( @xmath24 ) . in the following , the case @xmath25 ( no birefringence ) is considered . also we restrict the study to the case when the nonlinear material is a liquid ( e.g. , @xmath26 ) , for which @xmath27 and @xmath28 @xcite . this makes that our results could apply , e.g. , to cells of liquid crystal in the isotropic phase ( nematic liquid crystals , which have a much larger nonlinearity , are not covered by our analysis as they are _ anisotropic _ nonlinear media ) . most relevant for our study are the symmetries supported by eqs . ( [ modelo1],[modelo2 ] ) . in particular the term @xmath29 in eq . ( [ modelo1 ] ) completely breaks the phase symmetry of field @xmath3 . nevertheless , the model still supports the discrete symmetry @xmath30 . ( note that the four - wave mixing term -the one multiplied by @xmath31 in eq . ( [ modelo2])- breaks the continuous phase symmetry . ) this symmetry means that whenever a state @xmath32 is a solution of the system , another solution @xmath33 exists as well that is dynamically equivalent ( has the same dynamical properties like stability , etc . ) to the former . as @xmath3 and @xmath4 correspond to the two orthogonal components of the light electric field vector , the above symmetry relates two equivalent solutions having opposite helicity , appart from a different polarization ellipse orientation . this symmetry thus opens the possibility of exciting domain walls that join asymptotically two of such symmetric states . on the other hand , the reflection @xmath34 and translation @xmath35 invariances of the problem imply that if a bloch wall of given chirality exists , another , equivalent one of opposite chirality also exists , and both move in opposite directions @xcite . the existence conditions and dynamic behaviour of domain walls , the main subject of this article , is strongly related with the stability properties of the homogeneous solutions connected by the walls . in the system described by eqs . ( [ modelo1 ] ) and ( [ modelo2 ] ) , these solutions have been analyzed in @xcite , and we review here the main results . according to the polarization state of the intracavity field two kinds of steady homogeneous solutions are possible : ( i ) the _ linearly polarized state _ , with intensities @xmath36 and @xmath37 given by the solutions of @xmath38 i_{0 } , \label{linhom}%\ ] ] and ( ii ) the _ elliptically polarized state _ , with intensities determined by @xmath39 ^{2},\label{pol.hom.1}\\ i_{1 } & = \delta-\mathcal{a}i_{0}\pm\sqrt{\left ( \frac{\mathcal{b}}{2}% i_{0}\right ) ^{2}-1}. \label{pol.hom.2}%\end{aligned}\ ] ] the linearly polarized solution eq . ( [ linhom ] ) shows a multivalued character when @xmath40 @xcite . the elliptically polarized solution eq . ( [ pol.hom.1 ] ) was analyzed in detail in @xcite and can be a multivalued function as well . the cavity anisotropy parameter @xmath41 plays an important role on the character of the polarization instability ( i.e. , the bifurcation affecting the linearly polarized solution that sets the onset of eliptically polarized emission ) , as shown in @xcite . in particular the polarization instability can become subcritical for @xmath42 ( the exact value at which this occurs depends on the detuning value ) whenever @xmath43 . in this case bright cavity solitons can be supported by the system at low pumpings @xcite . more important for the present study is the fact that for large @xmath41 the anisotropic kerr cavity model can be reduced to a pdnlse , which exhibits an ib transition @xcite . this implies that the anisotropic kerr cavity will exhibit the same phenomenon for large enough @xmath41 . in order to test the universality of this phenomenon we investigate here a cavity with moderate anisotropy and take @xmath44 for definiteness . we note that for this value of @xmath41 the model can not be rigorously reduced to a pdnlse and then some extra features can be expected . the stability of the linearly polarized solution ( [ linhom ] ) against space - dependent perturbations was analyzed in @xcite , where analytical expressions for the different boundaries and wavenumber of the emerging patterns were obtained . in fig . 1 the stability of the different solutions is shown on the plane @xmath45 , which are the only free parameters . in the figure , the linearly polarized solution ( [ linhom ] ) is stable below curve @xmath46 , which corresponds to the polarization instability . above this curve the linearly polarized state is no more a stable solution and gives rise to the elliptically polarized solution ( [ pol.hom.1],[pol.hom.2 ] ) . in its turn , this last solution exists above line @xmath47 , i.e. , there is a domain of coexistence between the linearly and elliptically polarized solutions , marked in the figure as bs , between lines @xmath46 and @xmath47 . we consider now the existence of pattern forming instabilities of the elliptically polarized solution ( [ pol.hom.1],[pol.hom.2 ] ) , which are of relevance for the analysis of the dynamics of domain walls performed below . following the usual procedure , we consider perturbations of the homogeneous solutions in the form @xmath48 , and re@xmath49 signals a bifurcation . unlike the case of the linearly polarized solution @xcite , the stability analysis is now quite involved and analytic expressions can not be obtained . instead , we perform a numerical analysis of the eigenvalues @xmath50 to determine the instability boundaries . the pattern formation instability boundary affecting the elliptically polarized solution ( [ pol.hom.1],[pol.hom.2 ] ) corresponds to the piecewise curve ( with dashed and dotted segments ) joining the points @xmath51 . beyond this line ( shadowed area ) , the elliptically polarized solution is modulationaly unstable . the complex form of this boundary follows from the dependence of the real part of the largest eigenvalue with the wavenumber of the perturbation . as shown in fig . 2 , the real part of the eigenvalue ( solid line ) evaluated for parameters corresponding to point ( b ) in fig . 1 shows two maxima at different wavenumbers . depending on the parameter setting the threshold is minimum for the smallest wavenumber ( dashed lines @xmath52 and @xmath53 ) or for the largest wavenumber ( dotted lines @xmath54 and @xmath55 ) . consequently , the points denoted by @xmath56 , @xmath57 and @xmath58 in the figure correspond to codimension-2 points , where the instability is reached at two different wavenumbers simultaneously . furthermore , the imaginary part of the eigenvalue is null in the first cases ( dashed lines @xmath52 and @xmath53 ) , thus corresponding to the emergence of stationary patterns , but non null in the second cases ( dotted lines @xmath54 and @xmath55 ) , see fig . 2 , thus corresponding to a hopf bifurcation that gives rise to the appearance of dynamic patterns whose amplitude oscillates in time . finally , as stated , the region marked with in fig . 1 denotes the domain of coexistence between linearly and elliptically polarized states . when this coexistence is bistable and one of the states ( the elliptically polarized state in this case ) is spatially modulated ( shadowed region ) , bright cavity solitons can be also excited @xcite . we performed the numerical integration of eqs . ( [ modelo1 ] ) and ( [ modelo2 ] ) with a split step algorithm , for @xmath27 , @xmath28 , @xmath25 , and @xmath44 . periodic boundary conditions were used that impose an even number of walls in the transverse domain . a spatial grid of 2048 points was used and the temporal step was lowered down to @xmath59 in order to otain @xmath60independent results . as an initial condition , two walls were placed symmetrically with respect to the center of the integration window . several integration region lengths @xmath61 were investigated . results shown here correspond to the choice @xmath62 , which ensured enough spatial separation between the two walls in order to avoid their mutual interaction during the initial stage of the evolution . for some parameter settings the walls reached a static configuration after a transient , having fixed their position across the transverse plane . in such cases it was assessed that walls were ising ones by verifying that the complex field @xmath4 was zero at the wall core . for other parameter values walls reached a moving configuration and were identified as bloch walls as there was no point in the transverse plane where the complex field @xmath4 was zero . note that the following analysis of the results refers to the cross polarized component @xmath4 that is the one in which domain walls can be clearly identified as dark solitons @xcite as walls join two domains where the values of @xmath4 have opposite signs ( phases ) as discussed . figure 1 summarizes our numerical findings . for low values of both pump @xmath1 and detuning @xmath63 ( inside the region labeled `` ising '' , delimited by the solid lines joining points @xmath64 , and line @xmath47 ) stable ising walls are found . these ising walls connect homogeneous states in most of this region , except in the thin white areas at the right of the dashed line @xmath53 and the dotted line @xmath55 , where they connect patterned states , in agreement with the linear stability analysis discussed above . by increasing the pump or the detuning from this region we observe ib transitions , marked with solid lines @xmath65 , where ising walls are replaced by bloch walls . beforme commenting the differences between the various ising bloch transition lines ( @xmath66 , @xmath67 and @xmath68 ) , let us comment about what happens in the small region above line @xmath54 , where the behaviour is somewhat anomalous . in the small domain between this line and the dark grey shadowed area marked as `` '' the walls are unstable , and the single pattern supported by the system are rolls . then , strictly speaking , the domain `` '' extends until the line @xmath54 . if we have let this domain without including it in the `` '' domain is just because in it a very long transient behaviour in the dynamics of the walls is observed until walls eventually dissapear and the system develops a spatially periodic state . this behaviour ( longest transient ) is in contrast with what happens in the `` '' domain , where domain walls dissapear sharply . at the right of the point @xmath69 , when increasing the pump , two different regimes can be clearly distinguished , depending on detuning . for small detunings the transition is denoted by @xmath66 in fig . 1 ( continuous line ) , and in this case the bloch walls ( that exist inside the `` triangle '' @xmath70 ) connect homogeneous states . for the chosen values of the parameters , this regime exists up to @xmath71 for larger detunings , the transition is mainly ruled by the pattern forming instability experienced by the domains joined by the walls ( @xmath67 in fig . 1 , represented by continuous curve ) . note that , for moderate detunings ( in the center of the plot ) , the @xmath72 and pattern forming boundaries are nearly coincident . in this case the bloch walls always connect patterned states . finally , for small pump but large detunings , another region @xmath68 of bloch walls is found ( right side of the curve @xmath73 ) . the bloch walls in this region move with a extremelly small , random velocity , similarly to what happens in the pdnlse without diffusion or saturation terms @xcite . an example of a bloch wall corresponding to the transition @xmath66 , obtained for @xmath74 and @xmath75 , is shown in fig . 3 . in fig . 3(a ) the intensity distribution in transverse space near the core of the wall is given , and in fig . 3(b ) we show the corresponding phase portrait ( i.e. , a plot of the real versus the imaginary part of the field ) . both pictures are alternative representations where the bloch character of the wall is evidenced : the intensity at the wall core is small , but non null , fig . 3(a ) , and a smooth variation of the phase between the domains separated by the wall is appreciated in the parametric plot in fig . for these parameters , the homogeneous solutions which constitute the domains connected by the wall are modulationally stable . as stated in the introduction , an essential feature of bloch walls in nonvariational systems is that , contrary to ising walls , they move in the transverse plane , with a velocity that depends on the parameters . for a fixed value of the detuning @xmath76 , the dependence of the wall velocity with the pump is shown in fig . 4 . for this particular value of the detuning , in the region where the background solutions are modulationally unstable , the walls no more exist as commented , since spatial modulations of the background grow and fill completely the transverse space leading eventually to a roll pattern . let us now describe the dynamic behaviour of bloch walls . as stated in the previous section the use of periodic boundary conditions forces that the number of domain walls that can exist within the integration window be an even number , two is the minimum . on the other hand bloch walls move with a velocity whose sign depends on their chirality . then two situations are possible , namely , that the two bloch walls have either the same or different chiralities ( notice that the _ sign _ of the chirality of the bloch walls is fixed by the initial conditions ) . when the chiralities are the same , the two bloch walls move along parallel paths and no interaction between them appears ( at least when their movement is regular : when it is highly irregular the chirality of the walls can change independently of each other @xcite ) . contrarily , when the chiralities are opposite the paths followed by the two bloch walls intersect and a collision occurs . as a result a new localized structure may appear and a bound state is formed . this bound state is a cavity soliton different from a wall . although a detailed study of these objects falls outide the scope of the present work , and without trying to be exhaustive in their characterization , we just note that two different behaviours of the new cavity soliton have been identified . for small detunings ( i.e. inside the triangle @xmath70 above the region @xmath66 in fig . 1 ) , after the collision the positions of the walls delimiting the cavity soliton perform small amplitude antiphase periodic oscillations , as shown in fig . 4 ; hence the cavity soliton performs a breathing dynamics , remaining constant its position in the transverse plane . for higher detunings ( above the region @xmath67 ) bound states are also formed as a result of the interaction , but their dynamics is different . in these cases , after the collision one wall is dragged by the other and the resulting bound state drifts with the velocity of one of the original bloch walls as shown in fig . 5 . finally , we note that near but below the boundary @xmath54 , the walls do not form bound states after the interaction , but instead bounce and exchange their chiralities . such behaviour is shown in fig . although it is difficult to determine the origin of these behaviours after a collision is produced , the reason for these different behaviours is very likely linked to the different dynamics of the patterned state that the walls connect , whose complicated spatio temporal dynamics depends on the parameter set . in the next section , where we concentrate in the dynamic behaviour of isolated bloch walls , we give some clues on how the pattern dynamics affects the wall behaviour . in the small detuning regime , bloch walls behave in a regular manner and the motion occurs at a constant velocity as in the case shown in fig . 7 . for higher detunings however , the wall dynamics shows features which are characteristic of excitable systems . the wall motion is in fact regular only close to the @xmath67 transition , which nearly coincides with the pattern formation boundary , see fig . 1 . for higher pump values , wall dynamics is characterized by an irregular behaviour of the wall position . we report next several numerical examples of such irregular motion , obtained for @xmath77 and different values of the detuning @xmath63 . in fig . 8 , a bursting phenomenon ( the appearance of almost periodic oscillations during time intervals of arbitrary duration ) , is observed both in the wall position , fig . 8(a ) , and chirality , fig . 8(b ) , for @xmath78 . ( we used the definition of chirality @xmath79 @xcite where @xmath80 denotes the point where the wall intensity @xmath81 is at its minimum ) . after experiencing several bursts , the wall velocity turns to be constant again , but in this case with a chirality of opposite sign with respect to the initial one . during the steady motion , the chirality takes a very small value , @xmath82 , and consequently the change in sign is not appreciated in fig . 8(a ) , owing to the scale imposed by the bursting events . however , this change is manifested in fig . 8(b ) as a change in the sign of the wall velocity . the number of bursts and final chirality depends in general on the parameter values and initial conditions , such as the amplitude and relative position of the walls . a slight increase in the detuning leads to a qualitatively different behaviour , namely a spiking in the wall position . an example is shown in fig . 9 , obtained for @xmath83 . in this case the bursts appear periodically in time , in the form of spikes . the intensity distribution of the bloch wall during the spiking regime is shown in fig . the wall connects two patterned states with different spatial distributions . the pattern on the right side of the wall in fig . 10 is spatially harmonic , contains a single spatial frequency , while the pattern of the left side of the wall is bi - periodic , and both a fundamental and a small amplitude second spatial harmonic are present . this particular structure of the domain walls has been observed in all the numerical simulations in the irregular regime , and seems to be at the root of the complex behaviour exhibited by the wall dynamics . these numerical results suggest that the complex dynamic behaviour of bloch walls is related with secondary instabilities of the extended roll patterns which form the domains at both sides of the wall . in particular , numerics show that bursting and spiking of the wall position always develop in coincidence with the appearance of a second spatial frequency in the intensity of the roll pattern forming one of the domains . to check this statement , eqs . ( [ modelo1 ] ) and ( [ modelo2 ] ) were integrated in the absence of walls , and the spatial distribution of roll patterns was studied . a summary of results is given in fig . 11 , corresponding to roll patterns obtained for a fixed value of the pump @xmath77 and for different detunings , in accordance with figs . the distributions shown in figs . 11(a)(c ) correspond , from top to bottom inside each figure , to the intensity , the real and the imaginary parts of the extended patterns . figure 11(a ) has been obtained for @xmath84 , and the resulting intensity pattern is a stationary roll , with a single spatial frequency . in fig . 11(b ) , where @xmath85 , the intensity pattern develops a second spatial frequency of small amplitude . this value of detuning nearly corresponds to the onset of bursting phenomena of the wall . finally , in fig . 11(c ) , for @xmath86 the resulting intensity pattern shows the coexistence , in different spatial domains , of rolls with different periodicity and equal phases ( compare with fig . 10 where a domain wall separates oppositely phased patterns ) . clearly , in this parameter region there exists bistability between different spatial structures . this bistability seems to be a requirement for the existence of irregular dynamics . it is interesting to interpret these results in terms of the real and imaginary parts of the patterns shown in the bottom parts of figs . 11(a)(c ) . the second spatial frequency of the intensity pattern , and consequently the irregular dynamical behaviour , appears when real and imaginary parts of the pattern distributions cross , fig . 11(b ) and left part of fig . we have studied the dynamics of domain walls in an anisotropic optical kerr cavity . both ising and bloch walls , and the transition between them , have been reported in the case of moderate cavity anisotropy . the stability of the homogeneous solutions against pattern forming instabilities has been also analyzed . these results show a complex scenario of spatio - temporal evolution of patterns in this system . domain wall dynamics is shown to be related with the stability of the domain ( background ) solutions . a numerical study shows the existence of different domains of behaviour , depending on parameters . besides the typical evolution of bloch walls , with a drift at nearly constant velocity , we have observed regimes in which the behaviour of the bloch wall parameters ( position , velocity , chirality and intensity at the core ) is irregular , and analogous to that found in the temporal dynamics of excitable systems . these regimes , namely bursting and spiking , are reported for the first time in the case of domain walls in optical cavities . finally new cavity solitons formed by the interaction of two domain walls have been identified and two dynamical regimes of wall collision have been envisaged . we gratefully acknowledge helpful comments by pier luigi ramazza . this work was financially supported by the spanish ministerio de ciencia y tecnologa and european union feder , under projects bfm2002 - 04369-c04 - 01 and bfm2002 - 04369-c04 - 04 . figure 2 . growth rate of the perturbations of the homogeneous solutions as a function of the spatial wavenumber , near a codimension - two point for @xmath88 and @xmath77 . the rest of parameters as in fig.1 . example of a pair of bloch walls numerically obtained for the parameters @xmath74 and @xmath75 ( the rest of parameters as in fig.1 ) intensity distribution ( a ) and parametric representation ( b ) . roll patterns obtained at different detunings , for @xmath77 ( the rest of parameters as in fig.1 ) : ( a ) stationary roll pattern with a single spatial frequency ( @xmath84 ) ; ( b ) a second spatial frequency appears ( @xmath85 ) ; ( c ) bistability of roll patterns with different structure ( @xmath78 ) . patterns in ( b ) and ( c ) are dynamic . interaction of two bloch walls with opposite chirality for @xmath76 and @xmath89 ( the rest of parameters as in fig.1 ) . intensity ( b ) and real part ( c ) distribution of the bound state . the inset shows the oscillatory evolution of the wall position after the interaction	an anisotropic ( dichroic ) optical cavity containing a self - focusing kerr medium is shown to display a bifurcation between static ising and moving bloch domain walls , the so - called nonequilibrium ising bloch transition ( nib ) . bloch walls can show regular or irregular temporal behaviour , in particular , bursting and spiking . these phenomena are interpreted in terms of the spatio - temporal dynamics of the extended patterns connected by the wall , which display complex dynamical behaviour as well . domain wall interaction , including the formation of bound states is also addressed .
modeling emission from an optically - thin collisionally - ionized hot plasma has been an on - going problem in astrophysics ( cox & tucker 1969 ; cox & daltabuit 1971 ; mewe 1972 ; landini & monsignori fossi 1972 ; raymond & smith 1977 ; brickhouse , raymond & smith 1995 ) . over time , the codes used in these papers have been updated or completely rewritten ; current versions are the spexcode ( kaastra , mewe & nieuwenhuijzen 1996 ) and the chianti code ( dere et al.1997 ; landi et al.1999 ) . the frequently used mekal ( mewe - kaastra - liedahl ) code ( kaastra 1992 , liedahl , osterheld , & goldstein 1995 ) embedded in xspec ( arnaud 1996 ) uses data that are now in the spexcode ; its results are similar to spex . we describe here a new plasma emission code apec(atomic plasma emission code ) along with the atomic plasma emission database ( aped ) . our primary goal is to create plasma emission models that can be used to analyze data from the high resolution x - ray spectrometers on the _ chandra_and _ xmm - newton_telescopes . due to limitations on computer speed , memory size , and the available atomic data , early plasma codes included only the strongest emission lines from each ion , and `` bundled '' nearby lines , reporting only their summed emission at a single wavelength . in addition , the data were stored in the code itself , and could not be easily separated and studied . in contrast , we have strictly separated the atomic data from the code , storing the atomic data in fits format files , which collectively form aped . we have endeavored to maintain the data in apedas close as possible to the original form , for simplicity and ease in error checking . in addition , apedcontains error estimates for many wavelengths ; errors on other values in the database are in progress . currently , apedcontains data for well over a million lines , although many of these are too weak to be observed individually . an apeccalculation done at low electron density ( 1 @xmath2 ) , with cosmic abundances ( anders & grevesse 1989 ) contains over 32,000 unique lines whose peak emissivities at temperatures between @xmath3k exceed @xmath4photons @xmath5s@xmath6 . although the literature of theoretical calculations for observable atomic transitions is far from complete , multiple calculations for some important rates do exist . apedcontains all the different datasets we have collected ( _ e.g. _ the entire chianti v2.0 database ( landi et al.1999 ) , as well as the data referred to in this paper ) to allow easy comparison of different rates , and of the effect of different assumptions on the emissivities of selected lines . in a subsequent paper , we plan to present a complete overview of the data in apedand the emissivity tables over the x - ray range . here we discuss what effect changing the underlying atomic data has on hydrogen - like ( h - like ) and helium - like ( he - like ) oxygen ions . apeccalculates both line and continuum emissivities for a hot , optically - thin plasma which is in collisional ionization equilibrium . although apeccan calculate the ionization balance directly ( and thus handle non - equilibrium conditions ) , apeddoes not yet contain all the necessary ionization / recombination rates , so we use tabulated values for the ionization balance in thermal collisional equilibrium . we primarily use the ionization balance calculated by mazzotta et al.(1998 ; hereafter mm98 ) because it is a recent compilation that self - consistently treats all the astrophysically relevant ions . apeditself contains recombination ( radiative and dielectronic ) rate coefficients for oxygen . we combined these with the ionization rates from mm98 ) to create our own self - consistent ionization balance for these ions and found some small differences ( @xmath7% ; see figure [ fig : lineratio](b ) ) . we do plan to include in apec / apeda self - consistent calculation of the ionization balance and level population in the future . since the term `` emissivity '' has a number of definitions , we give our definition explicitly . similar to the terminology of raymond & smith ( 1977 ) ( but in photon instead of energy units ) , the emissivity of a spectral line is the total number of radiative transitions per unit volume , divided by the product of the electron density @xmath8 and the hydrogen ( neutrals and protons ) density @xmath9 in the astrophysical plasma . the line emissivity therefore has units of photons @xmath5s@xmath6 . since the number of photons emitted is actually proportional to the density of the ions involved , this definition implicitly requires an elemental abundance and ionization balance for the relevant ion to be specified . for a given electron temperature @xmath10 and electron density @xmath8 , the level populations for each ion are calculated from the collisional ( de-)excitation rate coefficients , the radiative transition rates , and the radiative and dielectronic recombination rate coefficients . excitation - autoionization processes have not yet been included in aped , but these are not significant for the h- and he - like ions in the equilibrium plasmas discussed here . for the h- and he - like isosequences , we have data for all singly - excited levels up to principal quantum number @xmath11 with the exception of , where we have data up to @xmath12 . for the he - like ion , the collisional excitation rate coefficients are from sampson , goett & clark ( 1983 ) , kato & nakazaki ( 1989 ) , and zhang & sampson ( 1987 ) for the levels up to @xmath11 , and from hullac ( bar - shalom , klapisch , & oreg 1988 ; klapisch et al.1977 ) calculations for @xmath0 . the radiative transition rates for @xmath13 are from topbase ( fernley , taylor & seaton 1987 ) , derevianko & johnson ( 1997 ) , lin , johnson , & dalgarno ( 1977 ) , and from the nist database ; for @xmath14 , the rates are again from hullac . the wavelengths are taken from drake ( 1988 ) . for the h - like ion , we use the collisional rate coefficients from sampson , goett & clark ( 1983 ) . we have also compared with scaled values from kisielius , berrington & norrington ( 1996 ) , but find few significant differences in the low - temperature region where the data overlap . the radiative transition rates are again from topbase and shapiro & breit ( 1959 ) for the two - photon ( @xmath15s@xmath16s@xmath17 ) rate . the wavelengths are from ericsson ( 1977 ) . for both ions , we use the dielectronic recombination ( dr ) rate coefficients and satellite line wavelengths from vainstein & safronova ( 1978 , 1980 ) for @xmath18 . for @xmath19 we used data from safronova , vasilyev & smith ( 2001 ) . dielectronic recombination of into levels @xmath0of are from safronova ( 2001 , private communication ) . radiative recombination is calculated using the milne relation and the photoionization cross - sections of verner & yakovlev ( 1995 ) for the ground states and clark , cowan , & bobrowicz ( 1986 ) for the recombination to excited states . we choose these oxygen ions for our comparisons since they are relatively simple and have strong emission lines in many astrophysical plasmas . we will concentrate on two line ratios : for h - like ions , ly@xmath20(16.020)/ly@xmath21(18.987 ) , and for he - like ions , the frequently used @xmath22 ratio , where @xmath23 ( 22.098 ) is the forbidden transition @xmath24 , @xmath25 ( 21.804 , 21.801 ) is the sum of the two intercombination transitions @xmath26 , and @xmath27 ( 21.602 ) is the resonance transition @xmath28 . figure [ fig : lineratio](a ) shows the h - like ly@xmath20/ly@xmath21line ratio from apec(with the mm98 ionization balance ) , spex1.1 , and raymond - smith . the mekal results track those of spex1.1 , and so are not independently considered . we also did a `` pure - apec '' calculation where the ionization balance was found using recombination rate coefficients taken from aped . however , this did not affect the ly@xmath20/ly@xmath21 ratio and so it is not shown . figure [ fig : lineratio](b ) shows the he - like g ratio from the same calculations plus those of bautista & kallman ( 2000 ) and the pure - apeccalculation . the difference between the apec+mm98 and the pure - apeccalculation for , and the lack of a change in , can be understood in terms of the excitation mechanisms for the lines , as will be discussed below . figure [ fig : lineratio ] demonstrates two important points . first , near the peak emissivity of the emission lines , the code predictions agree to within 25% . second , substantial disagreements ( exceeding 50% ) exist even for these simple ions when they are not near peak emissivity . of course , comparing three theoretical models against each other does not provide an accurate estimate of the model error . for one limiting case , we can compare with analytic results . in figure [ fig : lineratio](a ) , the bethe limit of 0.14 [ photon units ] ( burgess & tully 1978 ) is also plotted ; this is the ratio of the electron collisional excitation rates in the high temperature limit , taking into account the branching ratio of ly@xmath20 from the @xmath29level . in the h - like case , this limit should be a good estimate of the ly@xmath20/ly@xmath21 ratio since collisional excitation is the dominant method of populating excited levels . we calculate that cascades from collisional excitation to higher levels as well as radiative recombination contribute only @xmath30% of the direct rate to each line , somewhat more for ly@xmath20 than ly@xmath21 . since dielectronic recombination can not occur onto bare ions , it is not an issue . the agreement between the apecresult and the bethe limit is thus reassuring ; as expected , apecis slightly larger due to cascades and radiative recombination . this also explains why changing the ionization balance ( the pure - apeccase in figure [ fig : lineratio](a ) ) does not affect the ly@xmath20/ly@xmath21 ratio . both lines are primarily populated by direct collisions of electrons with , so small changes in the abundance cancel . some _ chandra_and _ xmm - newton_measurements of the ly@xmath20/ly@xmath21 ratio from astrophysical plasmas already exist . the _ chandra_hetg observation of the rs cvn star ii peg obtained a value of @xmath31 for this ratio ( huenemoerder 2000 , private communication ) . in addition , both _ chandra_hetg and _ xmm - newton_rgs observations of the oxygen - rich snr e0102 have been made , and although the extended nature of the source makes the analysis more difficult , the ly@xmath20/ly@xmath21 ratio is measurable . chandra_hetg obtained @xmath32 ( davis et al.2001 ) , while the _ xmm - newton_rgs1 and rgs2 instruments obtained @xmath33 ( rasmussen et al.2001 ) . the discrepancy between these results is not yet understood . in addition , some laboratory measurements with the llnl ebit of oxygen ly@xmath20/ly@xmath21 have been made . they find a range of values from 0.16 to 0.24 for electron energies above 1 kev ( gendreau 2001 , private communication ) . although the bethe limit of 0.14 for the ly@xmath20/ly@xmath21 ratio is a good estimate if collisional excitation dominates the excitation , other effects could change the observed value . line blending , especially of and lines with o ly@xmath20 , can increase the apparent ly@xmath20 emission . pure radiative recombination also leads to a larger value of ly@xmath20/ly@xmath21 in h - like ions , similarly to the well - known effects on line ratios for he - like ions in photoionized plasmas ( porquet & dubau 2000 ; bautista & kallman 2000 ) . however , tests with apecshow that a plasma has to be nearly pure o@xmath34 for this to occur in a hot ( @xmath35k ) plasma . charge exchange into excited levels has been suggested by rasmussen et al.(2001 ) as a possible way of increasing the ratio in e0102 , but this requires mixing of neutral material directly into highly ionized regions ; furthermore , it is difficult to imagine how this could be occurring in ii peg . finally , resonance scattering could also cause a change in the observed ratio , with the size and direction of the effect dependent on the spatial configuration of the plasma ( e.g. wood & raymond 2000 ) . of course , the bethe limit is only applicable if the upper levels of both emission lines are populated by direct excitation from the ground state , and depopulated by radiative transitions . in the he - like system , other processes are more important than direct excitation at high temperatures for populating the @xmath36 level ( see figure [ fig : excrate](b ) ) . the differences in the g ratio among the codes at high temperature appear to be largely due to differences in the dielectronic recombination rates . figure [ fig : excrate ] shows the level population mechanisms for the @xmath37 level ( which forms the resonance line ) and the @xmath36 level ( which decays to the ground state as the forbidden line ) . the total excitation rates for these levels are equivalent to the line emissivities , since radiative decay to the ground state is the primary de - excitation process at low electron density for both levels . in figure [ fig : excrate](a ) , we see that direct excitation dominates all other mechanisms for populating the @xmath37 level . however , figure [ fig : excrate](b ) shows that for the @xmath38 level , direct excitation is important only at low temperatures . at the peak emissivity , the main excitation mechanism is cascades from higher levels . these higher levels are primarily populated by dielectronic recombination , which we have checked by comparing the computed rates for all the processes . as a result , the bethe limit can not be used as a check on the high - temperature limits of this he - like line ratio . this can also explain the difference between the apec+mm98 and pure - apeccalculations shown in figure [ fig : lineratio](b ) . as a result of the different recombination rates ( @xmath39% variation ) , at temperatures above @xmath40k , the pure - apeccode has a slightly larger population and slightly smaller population . thus the resonance line r emissivity ( which is proportional to the abundance ) is increased , while the f+i emission , due largely to dielectronic recombination from , is reduced . this leads overall to a small decrease in the g ratio . for the h- and he - like isosequences of ions with significant cosmic abundances ( c , n , o , ne , mg , al , si , s , ar , ca , fe , and ni ) apedincludes atomic calculations for principal quantum numbers @xmath41 . we have tested the adequacy of this model by running apecrepeatedly using only the @xmath42 , @xmath18 , and the @xmath43 levels of , and find that , above the temperature of peak emissivity , our results are not convergent . we therefore obtained data for @xmath44 for ; the collisional excitation rate coefficients and radiative rates are from hullac and the dr satellite line data are from safronova ( 2001 , private communication ) . the g ratio for each model , with @xmath45 and 10 is shown in figure [ fig : ratio ] . clearly , the convergence including all @xmath46 is inadequate . by including all levels up to @xmath12 we converge to within the 10%-20% accuracy of the underlying atomic data . in the case of the @xmath47 ratio , which is sensitive to the electron density for @xmath48@xmath2 , fewer levels are required to reach convergence at @xmath49k , as can be seen in figure [ fig : ratio](b ) . at higher temperatures , such as @xmath50k , the spread among models is slightly larger , but including only levels with @xmath41 is still adequate . this is the first in a sequence of papers on the new collection of atomic data apedand collisional plasma code apec . by separating the code and data we can more easily test the convergence of our models and compare atomic data from different sources . we have shown that , for , using newer atomic data leads to substantially different results in the high - temperature limit for the ly@xmath20/ly@xmath21 ratio , which agrees with some recent ebit measurements . in addition , for he - like ions , high-@xmath51 dielectronic recombination from the h - like ion can significantly affect the low-@xmath51 line ratio g. careful treatment of recombination to individual levels and cascades is necessary for other diagnostic line ratios ; _ e.g. _ neon - like ( liedahl 2000 ) . the methods described in this paper can be used to test the importance of such detailed treatment . furthermore , in cases where recombination and cascades contribute significantly to the level population , the accuracy of the ionization balance , not explicitly considered in this paper , will contribute to the total uncertainty in the model line ratio . we thank priyamvada desai , richard edgar , kate kirby , brendan mclaughlin , and ulyana safronova for their assistance over the course of creating apec / aped . we gratefully acknowlege support from nasa grant nag5 - 3559 and the chandra x - ray center . work at llnl was performed under the auspices of the u. s. department of energy by university of california lawrence livermore national laboratory under contract no . w-7405-eng-48 . davis , d. s. , flanagan , k. a. , houck , j. c. , canizares , c. r. , allen , g. e. , schulz , n. s. , dewey , d. & schattenburg , m. l. 2001 , `` young supernova remnants : eleventh astrophysics conf . , '' eds . s. s. holt and u. hwang ( new york : american institute of physics ) , p. 230	new x - ray observatories ( _ chandra_and _ xmm - newton _ ) are providing a wealth of high - resolution x - ray spectra in which hydrogen- and helium - like ions are usually strong features . we present results from a new collisional - radiative plasma code , the astrophysical plasma emission code ( apec ) , which uses atomic data in the companion astrophysical plasma emission database ( aped ) to calculate spectral models for hot plasmas . apedcontains the requisite atomic data such as collisional and radiative rates , recombination cross sections , dielectronic recombination rates , and satellite line wavelengths . we compare the apecresults to other plasma codes for hydrogen- and helium - like diagnostics , and test the sensitivity of our results to the number of levels included in the models . we find that dielectronic recombination with hydrogen - like ions into high ( @xmath0 ) principal quantum numbers affects some helium - like line ratios from low - lying ( @xmath1 ) transitions .
the discovery of extrasolar planets around solar - type stars has now become a fairly regular occurrence ( e.g. , see the extrasolar planets encyclopaedia maintained by j. schneider at www.obspm.fr/encycl/encycl.html ) . in addition to these , there have now been numerous detections of planetary companions to cool _ giant _ stars , ( early detections include frink et al . 2002 , setiawan et al . 2003 , and sato et al . 2003 ) . with the ansatz that the formation of planetary systems around solar type stars is a relatively common occurrence , plus the recognition that such stars will evolve to become giants that experience significant mass loss on the way to becoming white dwarf stars , it becomes interesting to consider how giant star winds might affect substellar companions . this paper is not the first to entertain questions about the eventual evolution of planetary systems during late stellar phases . several authors have considered the angular momentum transfer between the orbit of a planet or brown dwarf with a red giant and/or its wind ( livio & soker 1984 ; livio , soker , & harpaz 1984 ; livio & soker 2002 ) . such effects could lead to `` sculpting '' of planetary nebulae ( e.g. , soker 2001 ) , which could explain the non - spherical but axisymmetric shapes that are so commonly observed ( e.g. , balick 1987 ) . rasio et al . ( 1996 ) have discussed the tidal decay of planetary orbits during the red giant phase . indeed , even as far back as 1924 , jeans examined the evolution of binary orbits due to stellar mass - loss ( only in that paper , the mass loss being considered was in the form of the conversion of matter to energy via nuclear fusion ) . moreover , several authors have considered the possibility of detecting planetary companions to white dwarf stars , with the goal of constraining the evolution of planetary systems through empirical means ( zuckerman & becklin 1987 ; livio , pringle , & saffer 1992 ; li , ferrario , & wickramasinghe 1998 ; chu et al . 2001 ; ignace 2001 ; burleigh , clarke , & hodgkin 2002 ) . recently , a substellar companion to a subdwarf in a post - red giant phase of evolution was reported ( geier et al . this is particularly interesting as an example of a low mass companion that avoided being engulfed by the bloated red giant star to survive in a small period orbit around the remnant . such occurrences bolster the need for the consideration of detecting substellar companions to red giants to understand better the connections between stellar evolution and planetary evolution . in this contribution one single fact is emphasized : when stars like the sun evolve to become red giants , the mass loss increases substantially by several orders of magnitude which may have observational consequences for detecting substellar companions and detailing their physical and orbital properties . already the influence of a strong wind on a substellar companion has been considered by struck , cohanim , & willson ( 2003 ) , who discuss the possibility of wind accretion by brown dwarf companions during the asymptotic giant branch ( agb ) phase . in terms of detecting extrasolar planets , a number of researchers have explored radio emissions from substellar companions to stars by extrapolating the radio properties of solar system planets to other star systems ( e.g. , see zarka 2007 and references therein ) . the radio emission of the earth and other solar system planets with significant magnetic fields are dominated by the cyclotron maser instability ( cmi ) process in which the low frequency spectrum below about 100 mhz is dominated by coherent cyclotron emission from mildly relativistic electrons ( gurnett 1974 ) . a key point is that electrons are fed to the magnetic poles where the cmi is strong , and that the coherent cyclotron emission dominates the incoherent emission by several orders of magnitude . a greater understanding of the detailed energetics and emissive mechanism has come about fairly recently in relation to the terrestrial auroral kilometric radiation ( akr ) by mutel et al . ( 2007 ) and mutel , christopher , & pickett ( 2008 ) . in applications to extrasolar planets , jupiter has been prominent in these considerations , since it shows bursting behavior in the radio band that can be quite bright , partly in relation to interactions occuring between jupiter and io ( e.g. , zarka 1998 , 2004 ) . griessmeier , zarka , & spreeuw ( 2007 ) have summarized various mechanisms that may contribute to radio emissions in the case of exoplants around main sequence stars . for example , coronal mass ejections and in some cases magnetic field interactions between the star and a short - period planet can produce strong radio signals . our contribution is to extend the use of the radiometric `` bode s law '' for late type main sequence stellar winds to those of giant stars for which several new considerations must be taken into account . this law is an empirically determined , and theoretically motivated , power - law relation between planetary radio emissions and the solar wind kinetic energy flux . the radio emissions occur at low frequencies of about 0.340 mhz , of which the upper end is observable with lofar ( farrell et al . a lunar radio observatory such as the proposed dark ages lunar interferometer ( dali ) would be sensitive to almost that entire range ( see lazio et al . 2007 for a summary of the capabilities of lunar based radio observatories ) . giant star winds differ from those of low - mass main sequence stars ( i.e. , solar - type winds ) in important ways . giants that are warmer than early k tend to be x - ray emitters , whereas those that are cooler are x - ray faint . the distinction in the x - ray properties is known as the linsky - haisch `` dividing line '' ( linsky & haisch 1979 ) . although these x - ray faint giants possess chromospheres , their winds are largely neutral ( e.g. , reid & menten 1996 ) , whereas the earlier giants and the winds of low - mass main sequence stars are ionised plasma flows . consequently , in the case of the neutral giant star winds , the flow can _ penetrate _ the magnetosphere of a substellar companion , and the supersonic wind flow will set up a bow shock in the vicinity of that companion . cassinelli et al . ( 2008 ) have considered a closely related scenario . interested in x - ray emissions , those authors used hydrodynamic simulations to model the bow shock for an early - type stellar wind interacting with a dense spherical gaseous clump . with a wind speed of order @xmath0 km s@xmath1 , and assuming adiabatic cooling , they were able to derive a power - law dependence for the differential emission measure with temperature of the form @xmath2 , with peak hot gas temperatures of order @xmath3 k. that simulation remains relevant for the case of a substellar companion to a cool giant star . although cool star winds are far slower than hot star winds , with values of 15 km s@xmath1 for agb stars up to perhaps 100 km s@xmath1 for some red giants , the winds are themselves much cooler , with sound speeds of a few km s@xmath1 . as a result , the wind is still quite supersonic . however , this does not mean that bow shocks to substellar companions will produce x - ray emissions for the cooler giant star systems . these winds are indeed much slower than those of early - type stars , and so the peak post - shock temperature will be considerably smaller . for a strong shock , as considered by cassinelli et al . , the post - shock temperature at the bow shock head ( or `` apex '' ) @xmath4 follows the well - known rankine - hugoniot relation , with @xmath5 where @xmath6 is the mean molecular weight for the gas , @xmath7 is the mass of hydrogen , and @xmath8 is the relative speed between the wind and the companion . the orbital motion can be non - trivial compared to the wind , but considering only the wind speed for the moment , one expects apex temperatures on the order of @xmath9 k for speeds of 15100 km s@xmath1 . at the low end , orbital motion will likely dominate the incident shock speed in the rest frame of the companion ( e.g. , the earth orbits at about 30 km s@xmath1 ) . as a result , post - shock temperatures of @xmath10 k are expected , sufficient to ionise hydrogen , thus leading to a substantial reservoir of electrons that can interact with a magnetosphere to produce radio emissions . the scenario for the giant stars is thus significantly different than for main sequence stars . first , known low - mass companions to giants have no examples of extremely short period orbits ( a week or less ) as in the main sequence case . second , giants with the most massive winds will be neutral , and thus have no counterpart in the solar system or to current applications among late - type main sequence stars . third , for the x - ray emitters ( or `` coronal '' giants ) , there has been no evidence for coronal mass ejections as in the solar case , so that the application of the radiometric bode s law is likely the most important emissive mechanism in the long wavelength radio band . it is worth noting that griessmeier et al.(2007 ) have evaluated the expected radio emissions for all known exoplanets at that time . their list includes companions to giant stars that had been reported at that time , and these do not produce detectable radio emissions . however , those authors appear to have applied a main sequence solar wind model in relation mass - loss rate and wind speed which is inapplicable to giant star winds . here we use empirical relations that are more appropriate for evolved cool star winds . to explore the radio emissions from companions to giant stars , the next section summarizes the properties of the host stars and the orbital properties of their companions . in section 3 , we apply the radiometric bode s law to giant star systems , comparing emissions between stars on either side of the linsky - haisch dividing line . in section 4 , we consider the bow shock scenario for the case of neutral winds that penetrate deep into a magnetosphere . we use the results of cassinelli et al . ( 2008 ) to estimate the incoherent cyclotron emission . results of our study are summarized in section 5 , where we conclude that new simulations are required to consider properly the operation of the cmi in the bow shock scenario . an appendix details our derivation for the cyclotron emission from a bow shock . [ cols="<,^,^,^,^,^,^,^,^,^ , > " , ] @xmath11references : d09 = dollinger et al . ( 2009b ) ; d09a = dollinger et al . ( 2009a ) ; n09a = niedzielski et al . ( 2009a ) ; l08 = liu et al . ( 2008 ) ; s08a = sato et al . ( 2008a ) ; s08b = sato et al ( 2008b ) ; s07 = sato et al ( 2007 ) n09b = niedzielski et al . ( 2009b ) ; n07 = niedzielski et al . ( 2007 ) ; d07 = dollinger et al . ( 2007 ) ; j06 = johnson , et al . ( 2006 ) ; se05 = setiawan et al . ( 2005 ) ; h05 = hatzes et al . ( 2005 ) ; sa03 = sato et al . ( 2003 ) ; f02 = frink et al . ( 2002 ) ; se03 = setiawan et al . ( 2003 ) ; h06 = hatzes et al . ( 2006 ) ; r06 = reffert et al . ( 2006 ) ; h03 = hatzes et al . ( 2003 ) ; j07 = johnson et al . ( 2007 ) ; j08 = johnson et al . ( 2008 ) . + @xmath12 luminosity computed from values of @xmath13 and @xmath14 given by respective reference . + @xmath15 luminosity estimate based on spectral type . + @xmath16 eccentricity fixed to zero for orbital solution . a large number of substellar companions have been identified in radial velocity surveys of giant stars . from the literature we have gathered information about their physical properties which are displayed in table 1 . distances were taken from the simbad database that lists values from the hipparcos survey . references for the stellar and planetary properties appear in the last column . note that some of the values in the table are quite uncertain , and the reader is strongly urged to consult the references for further details of their evaluation . our goal here is to use these values as a guide for our application to the radio emissions . with these stellar parameters we use a standard form of reimer s law to estimate wind mass - loss rates @xmath17 ( reimers 1975 ; lamers & cassinelli 1999 ) : @xmath18 where @xmath19 is a scaling parameter between about 0.3 and 3 . here , we adopt @xmath20 since only mass - loss rate estimates will be needed . computed mass - loss rate values are given in table 1 in the seventh column , scaled to @xmath21 yr@xmath1 . note that no value is given for hd13189 since no radius or temperature value was quoted for this star . the mass loss is important since it sets the scale of the wind density at the companion where radio emission will be produced . table 1 shows that there is a significant spread in values fro the giants with planets detected so far , ranging from @xmath22 yr@xmath1 for hd47536 down to about @xmath23 yr@xmath1 for the subgiant hd185269 . however , there is controversary about reimer s law and its applicability . an understanding of it remains a topic of current research ( e.g. , schrder & cuntz 2005 ) . the law has application to more luminous cool stars , but its extension to their lower luminosity cousins is less clear . also ongoing are attempts to understand the linsky - haisch `` dividing line '' that separates giant stars earlier than about k0 that show x - ray emissions from the later types that are not coronal but which do have chromospheres ( suzuki 2007 ) . again , the main value of table 1 is the determination of representative stellar and planetary parameters and their spread for the sample as a whole . in relation to the population of host giant stars , the notable points are that the median distance is around 100 pc , about 10 times farther than main sequence host stars of substellar companions . also , the mass - loss rates are typically of order @xmath24 yr@xmath1 , about 4 orders of magnitude larger than the solar wind , but lower by a similar factor from the agb winds . finally , the distribution of orbital semi - major axes and eccentricities reflects the selection effects of the radial velocity study : values of @xmath25 around 1 au or less , and values of @xmath26 that can deviate significantly from circular orbits . radio studies of single red giant stars reveal them to be faint radio emitters ( spergel , giuliana , & knapp 1983 ; reid & menten 1996 ; gdel 2002 ) . for the brighter sources , observations at different radio bands indicate a flux density spectrum of the form @xmath27 , consistent with the rayleigh - jeans limit and suggestive of photospheric emission , although there are some exceptions indicating extended radio photospheres . the main conclusion is that the red giant star winds that show chromospheric signatures but not coronal x - rays are largely neutral , yet some metals are ionised , resulting in low level ionisations of the wind material at the level of 0.01 - 0.1% ( reid & menten 1996 ) . these ionisations are 3 to 4 orders of magnitude below that of the sun s wind . however , the mass - loss rate from giant stars is 4 or more orders of magnitude larger than the solar wind . interestingly , judge & stencel ( 1991 ) have reported on larger ionisations in some red giant stars . it is true that the earlier red giants , earlier than about k2 that show x - ray emissions , have fully ionised winds . unfortunately , their mass - loss rates are relatively uncertain . it is useful to consider how even the paltry ionised component of a giant star wind might interact with planetary magnetospheres so that at least lower limits to the radio emissions may be derived . the topic of radio emissions in the solar system and from extrasolar planets has been studied extensively . to estimate the flux density from the giant star wind impinging on a planetary magnetosphere , we use the radiometric `` bode s '' law of equation ( 3 ) from lazio et al . ( 2004 ) based on scaling relations from farrell , desch , & zarka ( 1999 ) and zarka et al . the median radio power from the stellar wind interaction with the planet is @xmath28 where @xmath29 is the planet s rotation rate , @xmath30 is its mass , @xmath31 is the orbital distance of the planet from its star , @xmath32 and @xmath33 are the density and speed of the stellar wind , @xmath34 and @xmath35 are the density and speed of the solar wind , and `` j '' subscripts indicate values for jupiter . the density @xmath32 must be corrected for the lower ionisation level of a late giant star wind as compared to a main sequence star . introducing @xmath19 as a ratio of stellar wind properties to that of the sun , and assuming a spherically symmetric wind with @xmath36 , we can derive that @xmath37 where @xmath38 is the ionisation correction factor , with a value of about @xmath39 to @xmath40 . with @xmath41 yr@xmath1 and @xmath42 km s@xmath1 for the sun , and @xmath43 yr@xmath1 and @xmath44 km s@xmath1 for a giant star wind with @xmath45 , we arrive at @xmath46 for a red giant with fairly strong mass loss . some red giants have lower @xmath17 values and faster speeds , in which case @xmath19 would remain comparable to the above value . agb stars on the other hand are slower by a factor of two but have higher mass loss by a factor of @xmath0 , and so @xmath19 can become quite large at around @xmath0 . assuming @xmath47 , and the other scaling ratios are unity in equation ( [ eq : lrad ] ) , then the total radio luminosity would be approximately @xmath48 erg s@xmath1 . using equation ( 4 ) of lazio et al . for a typical emission frequency of @xmath49 mhz , combined with their equation ( 5 ) for the flux density under the approximation that the radio luminosity is emitted isotropically in a bandwidth of @xmath50 , the expected radio brightness level will be @xmath51 where @xmath52 is the earth - star distance , and @xmath53 is the beaming of the radiation relative to isotropic . if all of the parenthetical factors were unity , then a jupiter - like companion to a red giant star at 100 pc would have a flux density of @xmath54jy . this is too faint for detection by current or near - future radio telescopes . however , it should be noted that we have assumed nearly neutral winds , referring to late red giants , those that lie on the cool side of the linsky - haisch dividing line . red giants of earlier spectral type that are x - ray emitters likely have winds similar to that of the sun , namely fully ionised plasmas . for such stars @xmath55 , with a gain factor in the radio emission of 3 orders of magnitude , bringing the flux density up to about 1 mjy , which is within the realm of detection by facilities such as lofar . unfortunately , most giant stars that are known host stars for substellar companions have much lower mass - loss rate values than assumed in equation ( [ eq : snu ] ) , closer to @xmath56 yr@xmath1 . for an ionised red giant wind with @xmath57 , this level of mass loss pushes the expected flux density down to about 10 @xmath6jy , again well below detection thresholds . however , some g giants can have @xmath17 values closer to @xmath58 yr@xmath1 ( e.g. , 11 com in tab . 1 ) . it is easy to imagine favorable cases with an early giant star wind with somewhat large mass loss and higher wind speed along with a companion in a sub - au orbit ( possibly eccentric see sect . 5 ) that could return the radio flux density closer to the mjy level . in addition , it is possible that in fact the majority of giant stars maintain planetary systems or brown dwarf companions . substantially higher radio flux densities would then be expected from giants with higher mass - loss rates , ones that have not been included in the radial velocity surveys . for example , agb stars with @xmath59 yr@xmath1 could represent a new stellar population for detecting substellar companions in later stages of stellar evolution than has so far been targeted . we now turn to a new consideration for generating radio emissions . since some giant star winds are neutral , they will penetrate companion magnetospheres and can be expected to set up bow shocks in close vicinity of the substellar object . cassinelli et al . ( 2008 ) describe a hydrodynamic simulation of a massive star wind impinging on a spherically symmetric `` hard '' clump . for these authors the focus was on explaining x - ray emissions from hot star winds . but the key results of the simulations pertain to emission measure and temperature distributions , and so the results have relevance for bow shocks formed from red giants that intercept substellar companions . to explore the observational consequences of this scenario , we adopt expressions for the emission measure ( @xmath60 ) distribution of post - shocked gas as a function of temperature ( @xmath61 ) from equations ( 23 ) and ( 24 ) of cassinelli et al.(2008 ) : @xmath62 where @xmath4 is given from equation ( [ eq : ta ] ) . the emission measure scaling factor @xmath63 is set by the square of the wind number density ( times four owing to a strong shock ) at the location of the companion multiplied by the volume of the companion . its value is given by @xmath64 with @xmath65 the planet radius , @xmath66 the wind mass - loss rate divided by @xmath67 yr@xmath1 , @xmath68 the radial wind speed divided by 30 km s@xmath1 , and @xmath69 the orbital distance of the planet in au . notice that the em is a strong function of the planet size ( as the cube ) and the orbital distance ( as the fourth power ) . the sizes of jovian planets and brown dwarfs vary slowly with radius , except for short period companions where x - ray and uv heating and tidal effects can enlarge the effective radius of the planet ( e.g. , guillot et al . 1996 ; lammer et al . 2003 ; burrows et al . we assume a nominal value of @xmath70 cm . the scenario that we envision is one where the largely neutral wind penetrates a planetary or brown dwarf magnetosphere . on the scale of the companion size , the wind flow is approximately plane parallel as in the cassinelli et al . ( 2008 ) simulation for a wind clump . the peak temperature achieved at the bow head will be of order @xmath71 k or more . the hydrogen gas can become ionised , and the bow shock has a decreasing temperature distribution along its length downstream . the shock becomes increasingly oblique until it drops down to around 3,000 k , where we assume that hydrogen is no longer ionised at the shock front . this reservoir of electrons finds itself embedded deep inside a magnetosphere that is sweeping past them . ignoring the details of the flow dynamics , we make a lower limit estimate for the radio emission as arising from a non - relativistic , incoherent cyclotron process . this emission will occur at a characteristic frequency of @xmath72 which is about 28 mhz for a field strength of 10 g. in fact , the emission occurs within the volume of the bow shock , predominantly in a higher density region that hugs the bow shock itself ( see cassinelli et al . a derivation for the flux density and spectral shape of the thermal , non - relativistic cyclotron emission is given in the appendix . the result is reproduced here ( see eq . [ [ eq : snubow ] ] ) , with the flux density @xmath73 where @xmath74 , and @xmath75 as described in equations ( [ eq : xi ] ) and ( [ eq : y ] ) of the appendix . the scale of this emission is sub - microjansky , well below the detectability of current instrumentation . however , there are several key points to note . first and foremost , the above calculation should be considered as a minimum flux density in the sense that it does not take account of cmi , or any bursting behavior such as is observed in jupiter , and such emissive processes will be stronger by orders of magnitude over the thermal cyclotron emission that we have considered . nor does it account for the possible influence of moons , each of which would have its own bow shock . ( of course , being a smaller target , the overall em from a moon s bow shock would be much smaller than for a jovian - like object , yet the shock could have hotter gas owing to its circumplanetary orbital motion . ) the ionisation is set only by the relative flow between the blunt object and the wind . although the em of the bow shock for a moon would be insignificant as compared to the companion bow shock , its primary influence may be in the form of providing an injection mechanism of electrons to the polar field regions of the companion where the cmi operates . it should be noted that much of what is being proposed for the bow shock is qualitative only . for a magnetosphere with a typical field strength of order 10 g in the vicinity of the bow shock , the magnetic energy density @xmath76 erg @xmath77 is vastly larger than the ram pressure of the wind flow @xmath78 erg @xmath77 . the implication is that a very strong field is rotating past the bow shock , that in essence acts as an ionisation front . the post - shocked plasma is essentially being `` slammed '' by the magnetosphere and rapidly accelerated . it is unclear what effect this will have on the bow shock structure itself . a fully consistent mhd simulation of this scenario needs to be carried out . it is a situation unlike anything occuring in the solar system where the solar wind is everywhere extremely fast and highly ionised in its interactions with solar system bodies . importantly , it is unclear in the neutral giant wind case how or if electrons can be accelerated to mildly relativistic values and also channeled to the polar regions of the field axis for the cmi to operate . however , it is encouraging that there is current interest in both theoretical and experimental work for understanding the cmi process . for example , mcconville et al . ( 2008 ) conducted a laboratory experiment for a scaled version of the terrestrial akr that yielded results in basic agreement with satellite measurements . their work revealed that approximately 1% of the electron kinetic energy pool was converted to radiation via the cmi . these results appear reproducible in 3d numerical simulations ( gillespie et al . such work supports the ansatz that a similar level of efficiency can be used in applications to extrasolar planets ( e.g. , jardine & cameron 2008 ) . to advance modeling of the radio emissions in the red giant wind case , mhd simulations will be necessary to assess how electrons can be fed to the magnetic poles of the companion , and then combine that information with the body of work just described for the akr mechanism . it seems likely that such an approach will be tractable . on the whole our analysis for the detection of radio emissions from the interaction between the wind of a red giant star and its substellar companion is largely negative . cyclotron emissions appear to be at the micro - jansky level , orders of magnitude below current or near - future detection thresholds . however , we have taken the most conservative and pessimistic approach in our estimates . we have considered red giant winds that are largely neutral ( i.e. , those on the cool side of the `` dividing line '' ) , yet the winds of earlier type red giants may be highly ionised with large gain factors in the resultant radio emissions . there have been radio - wavelength surveys ( typically at 2 and 6 cm ) of red giant stars including stars on the asymptotic giant branch of the h - r diagram ( see drake & linsky 1983 , 1986 ; drake , linsky , & elitzur 1987 ; drake et al . 1991 ; luttermoser & brown 1992 ) . typically evolved stars on the cool side of the `` dividing line '' are either not detected at these wavelengths or are very weak sources giant stars of m6 or warmer have been detected ( drake et al . the weak radio emission of these non - mira giants are thought to be due to partially ionised winds . there have been some detections of giants on the cool side of m6 iii . for example @xmath79 ceti ( mira ) appears variable in its radio emission . mira was detected at 6 cm by spergel , giuliana , & knapp ( 1983 ) as a @xmath80 event ; however , drake et al . ( 1987 ) failed to detect this pulsating long - period variable ( lpv ) at either 2 cm or 6 cm . since mira has a hot companion , it is not clear if the emission arose from the lpv or from the companion . in a survey of seven n - type carbon stars at 3.6 cm , luttermoser & brown ( 1992 ) only detected v hya , which is a peculiar carbon star that shows evidence of bipolar outflow ( tsuji et al . 1988 ) . in the case of a neutral wind , we have argued that a bow shock will result in the vicinity of the companion object . if the relative flow speed is enough , one can expect post - shock ionisation of the wind material . for incoherent cyclotron emission , the radio flux of the ionised component will be quite weak . however , it may be possible that some fraction of these newly created electrons are fed to the polar regions of the companion where the cmi mechanism can operate leading to much stronger radio emission . in addition , analogous to the jupiter - io interaction , moons of substellar companion may also produce bow shocks , and these might provide a source of electrons that could be accelerated in the magnetosphere and feed the cmi . our model for the bow shock is not adequate to address these possibilities . new self - consistent mhd calculations will be needed to model the interaction of the bow shock with the companion s magnetosphere and to assess the channeling of electrons for use with the cmi . of course , all of the effects described so far would be affected by orbital eccentricity of the companion . for orbits of a fixed semi - major axis , the radio flux density will be significantly higher near periapse as compared to apapse for increasingly eccentric orbits . but then of course , gains in signal strength near periapse will occur over a diminishing fraction of the orbital period as eccentricity increases . although these effects should be explored , they are of secondary importance to substantial task of modeling the flow dynamics more accurately . finally , it is possible that the agb stars might be targeted for long wavelength radio studies . the agb winds are certainly neutral so that the bow shock scenario would be of interest . the mass - loss rates can be extremely large at around @xmath81 yr@xmath1 . since the radio emission scales with @xmath82 , there would be a gain factor of 6 orders of magnitude above our estimates for the radio emission , and that is just for incoherent cyclotron emission . of course , agb stars are relatively rare and so tend to be more distant . plus the companion will probably need to be in an orbit that is greater than 1 au , otherwise it could be engulfed by the star . with a wind speed of just 15 km s@xmath1 , the astute reader may recognize that the apex temperature will hardly be large enough to ionise hydrogen . however , there are actually two effects to mitigate the low wind speed . first is that the orbital speed could be larger by perhaps a factor of 2 or more . second , the hydrodynamical simulations of cassinelli et al . ( 2008 ) are for clumps treated as blunt obstacles without self - gravity . jovian and brown dwarf mass objects have surface escape speeds of 30 km s@xmath1 and higher . given that the bow shock apex is at around 1.5 times the radius of the blunt object in the cassinelli et al . simulations , a slow agb wind can double and triple in speed as it falls into the gravitational potential of the substellar companion ( i.e. , similar to accretion described by bondi & hoyle 1944 ) . of course , inclusion of gravity changes the details of the bow shock , but the main point is that significant ionisation would still be viable . collecting the various factors , @xmath83 of around 0.1 mjy might be achievable . it is clear that long wavelength observations will provide new opportunities for detecting substellar companions and for studying their magnetospheres . companions to red giants should not be neglected in these efforts . an interesting result of our study has been the relatively novel consideration of neutral winds interactions with substellar companions , completely unlike the solar system case . the prospects for future observations are certainly exciting , but there is a clear need for more detailed modeling of mhd bow shocks in relation to the cmi to explore the viability of radio detection . the authors are gratefully indebted to bob mutel for discussions that clarified the akr process as well as an anonymous referee whose comments led to improvements of this paper . this research has made use of the simbad database , operated at cds , strasbourg , france . the steps for determining the radio flux density in the bow shock model is detailed . the differential luminosity @xmath84 of cyclotron emission from a small unit of volume @xmath85 is given by where @xmath89 is the cyclotron frequency . for the bow shock model of cassinelli et al . ( 2008 ) , the em @xmath60 is dominated in large part by the region along the bow shock . we can make the following approximation for the cyclotron emitting volume , this last expression takes advantage of the `` on the shock approximation '' ( otsh ) of cassinelli et al . ( 2008 ) , for which @xmath91 along the bow shock in the strong shock limit . this occurs under the assumption that the extent of the bow shock is relatively small compared to the orbital radius . where we have taken @xmath93 as a ratio of parametric form in the temperature @xmath61 . since @xmath61 is a smoothly decreasing function along the bow shock surface , it acts effectively as a mapping coordinate in our prescription for the thermal cyclotron emission . the differential em is simply a power law in temperature given by in order to determine @xmath95 , we require two pieces of information . the first is how the field varies in space around the companion object . we will assume that the field is a dipole . since we are seeking mainly an order of magnitude estimate of the emission level at long wavelengths in the radio , we will ignore latitudinal variations of the field and simply adopt @xmath96 , where @xmath65 is the companion radius . the second piece is the mapping of @xmath97 along the bow shock as it cuts through the dipole field . from cassinelli et al . ( 2008 ) , the bow shock shape is close to a parabola ( especially near the bow head , that is most relevant for our work ) , which is a convenient form to work with . again , with the goal of obtaining an order of magnitude estimate , we approximate the bow shock geometry using where @xmath99 and @xmath100 , where @xmath101 are cylindrical coordinates for the axisymmetric bow shock with @xmath102 the symmetry axis . the way to obtain @xmath103 is a two step process . the first step is to derive @xmath104 ; the second is to use @xmath105 from cassinelli et al .. combining the preceding relations , noting that the flux density is @xmath116 for distance @xmath52 , and introducing a maximum frequency of emission @xmath117 corresponding to the apex of the bow shock that temperature @xmath4 and minimum radius @xmath118 , the flux density becomes the total emission sub - microjansky and peaks close to the maximum frequency . an example spectrum is shown in figure [ fig2 ] . the flux density is normalized to 0.3 @xmath6jy with nominal values assumed for the physical parameters in equation ( [ eq : snubow ] ) . the lower temperature bound is taken as @xmath120 k. the solid curve is the for the frequency spectrum ; the dashed curve plots the emission against the temperature distribution with scale given at top . it is assumed that hydrogen is completely ionised over this temperature range . balick , b. 1987 , aj , 94 , 671 bondi , h. , & hoyle , f. 1944 , mnras , 104 , 273 burleigh , m. r. , clarke , f. j. , & hodgkin , s. t. 2002 , mnras , 331 , l41 burrows , a. , hubbard , w. b. , lunine , j. i. , & liebert , j. 2001 , rev . 73 , 719 burrows , a. , hubeny , i. , budaj , j. , hubbard , w. b. , 2007 , apj , 661 , 502 cassinelli , j. p. , ignace , r. , waldron , w. l. , cho , j. , murphy , n. a. , lazarian , a. , 2008 , apj , 683 , 1052 chu , y .- h . , dunne , b. c. , gruendl , r. a. , & brandner , w. 2001 , apj , 546 , l61 dollinger , m. p. , hatzes , a. p. , pasquini , l. , guenther , e. w. , hartmann , m. , girardi , l. , esposito , m. , 2007 , a&a , 472 , 649 dollinger , m. p. , hatzes , a. p. , pasquini , l. , guenther , e. w. , hartmann , m. , girardi , l. , 2009a , a&a , 499 , 935 dollinger , m. p. , hatzes , a. p. , pasquini , l. , guenther , e. w. , hartmann , m. , 2009b , astroph/0908.1753 drake , s. a. , & linsky , j. l. 1983 , apj , 274 , l77 drake , s. a. , & linsky , j. l. 1986 , aj , 91 , 602 drake , s. a. , linsky , j. l. , & elitzer , m. 1987 , aj , 94 , 1280 drake , s. a. , linsky , j. l. , judge , p. g. , & elitzer , m. 1991 , aj , 101 , 230 farrell , w. m. , desch , m. d. , zarka , p. , 1999 , jgr , 104 , 14025 farrell , w. m. , lazio , t. j. w. , zarka , p. , bastian , t. j. , desch , m. d. , ryabov , b. p. , 2004 , p&ss , 52 , 1469 frink , s. , mitchell , d. s. , quirrenbach , a. , fischer , d. a. , marcy , g. w. , butler , r. p. , 2002 , apj , 576 , 478 geier , s. , edelmann , h. , heber , u. , morales - rueda , l. , 2009 , apj , 702 , l96 gillespie , k. m. , mcconville , s. l. , speirs , d. c. , ronald , k. , mcconville , s. l. , phelps , a. d. r. , et al . , 2008 , plasma phys . control . fusion , 50 , 124038 griessmeier , j .- m . , zarka , p. , spreeuw , h. , 2007 , a&a , 475 , 359 gdel , m. , 2002 , araa , 40 , 217 guillot , t. , burrows , a. , hubbard , w. b. , lunine , j. i. , saumon , d. , 1996 , apj , 459 , l35 gurnett , d. a. , jgr , 79 , 4227 hatzes , a. p. , cochran , w. d. , endl , m. , et al . , 2003 , apj , 599 , 1383 hatzes , a. p. , guenther , e. w. , endl , m. , cochran , w. d. , dollinger , m. p. , bedalov , a. , 2005 , a&a , 437 , 743 hatzes , a. p. , cochran , w. d. , endl , m. , et al . , 2006 , a&a , 457 , 335 hubbard , w. b. 1974 , a. zh . , 51 , 1052 ignace , r. 2001 , pasp , 113 , 1227 jardine , m. , cameron , a. c. , 2008 , a&a , 490 , 843 jeans , j. h. 1924 , mnras , 85 , 2 johnson , j. a. , marcy , g. w. , fischer , d. a. , henry , g. w. , wright , j. t.,isaacson , h. , mccarthy , c. , 2006 , apj , 652 , 1724 johnson , j. a. , fischer , d. a. , marcy , g. w. , et al . , 2007 , apj , 665 , 785 johnson , j. a. , marcy , g. w. , fischer , d. a. , wright , j. t. , reffert , s. , kregenow , j. m. , williams , p. k. g. , peek , k. m. g. , 2008 , apj , 784 judge , p. g. , stencel , r. e. , 1991 , apj , 371 , 357 lamers h. j. g. l. m. , cassinelli j. p. 1999 , introduction to stellar winds , new york , cambridge university press lammer , h. , selsis , f. , ribas , i. , guinan , e. f. , bauer , s. j. , weiss , w. w. , 2003 , apj , 598 , l121 lazio , j. , macdowall , r. j. , burns , j. , demaio , l. , jones , d. l. , weiler , k. w. , 2007 , to appear in astrophysics enabled by the return to the moon , astroph/0701770 lazio , t. j. w. , farrell , w. m. , dietrick , j. , greenless , e. , hogan , e. , jones , c. , hennig , l. a. , 2004 , apj , 612 , 511 li , j. , ferrario , l. , & wickramasinghe , d. 1998 , apj , 403 , l151 linsky , j. l. , & haisch , b. m. 1979 , apj , 229 , l27 liu , y .- j . , sato , b. , zhao , g. , et al . , 2008 , apj , 672 , 553 livio , m. , & soker , n. 1984 , mnras , 208 , 763 livio , m. , pringle , j. e. , & saffer , r. a. 1992 , mnras , 257 , 15 livio , m. , & soker , n. 2002 , apj , 571 , l161 luttermoser , d.g . , & brown , a. 1992 , apj , 384 , 634 mcconville , s. l. , speirs , d. c. , ronald , k. , phelps , a. d. r. , cross , a. w. , et al . , 2008 , plasma phys . fusion , 50 , 074010 mutel , r. l. , peterson , w. m. , jaeger , t. r. , scudder , j. d. , 2007 , jgr , 112 , a07211 mutel , r. l. , christopher , i. w. , pickett , j. s. , 2008 , geophy . letters , 35 , l07104 niedzielski , a. , konacki , m. , wolszczan , a. , et al . , 2007 , apj , 669 , 1354 niedzielski , a. , nowak , g. , adamow , m. , wolszczan , a. , 2009a , astroph/0906.1804 niedzielski , a. , gozdziewski , k. , wolszczan , a. , konacki , m. , nowak , g. , zielinski , p. , 2009b , apj , 693 , 276 rasio , f. a. , tout , c. a. , lubow , s. h. , & livio , m. 1996 , apj , 470 , 1187 reffert , s. , quirrenbach , a. , mitchell , d. s. , albrecht , s. , hekker , s. , fischer , d. a. , marcy , g. w. , butler , r. p. , 2006 , apj , 652 , 661 reid , m. j. , menten , k. m. , 1996 , in radio emission from the stars and sun , ( eds ) a. r. taylor and j. m. pardes , asp conf . # 93 , 87 reimers , d. , 1975 , mem . liege , 63 ser . 8 , 369 sato , b. , ando , h. , kambe , e. , takeda , y. , izumiura , h. , masuda , s. , et al . , 2003 , apj , 597 , l157 sato , b. , izumiura , h. , toyota , e. , et al . , 2007 , apj , 661 , 527 sato , b. , izumiura , h. , toyota , e. , et al . , 2008a , pasj , 60 , 539 sato , b. , toyota , e. , omiya , m. , et al . , 2008b , pasj , 60 , 1317 schrder , k .- p . , cuntz , m. , 2005 , apj , 630 , l73 setiawan , j. , hatzes , a. p. , von der le , o. , pasquini , l. , naef , d. , da silva , l. , udry , s. , queloz , d. , et al . , 2003 , a&a , 398 , l19 setiawan , j. , rodmann , j. , da silva , l. , et al . , 2005 , a&a , 437 , l31 soker , n. , livio , m. , & harpaz , a. 1984 , mnras , 210 , 189 soker , n. 2001 , mnras , 324 , 699 spergel , d. n. , giuliani , jr , j. l. , knapp , g. r. , 1983 , apj , 275 , 330 struck , c. , cohanim , b. e. , & willson , l. a. 2004 , mnras , 347 , 173 suzuki , t. k. , 2007 , apj , 659 , 1592 tsuji , t. , unno , w. , kaifu , n. , izumiura , h. , ukita , n. , cho , s. , & koyama , k. 1988 , apj , 327 , l23 zarka , p. , 1998 , jgr , 103 , 20159 zarka , p. , treumann , r. a. , ryabov , b. p. , ryabov , v. b. , 2001 , ap&ss , 277 , 293 zarka , p. , 2004 , p&ss , 52 , 1455 zarka , p. , 2007 , p&ss , 55 , 598 zuckerman , b. , & becklin , e. e. 1987 , nature , 330 , 138	a number of substellar companions to evolved cool stars have now been reported . cool giants are distinct from their progenitor main sequence ( ms ) low - mass stars in a number of ways . first , the mass loss rates of cool giant stars are orders of magnitude greater than for the late - type ms stars . second , on the cool side of the linsky - haisch `` dividing line '' , k and m giant stars are not x - ray sources , although they do show evidence for chromospheres . as a result , cool star winds are largely neutral for those spectral types , suggesting that planetary or brown dwarf magnetospheres will not be effective in standing off the stellar wind . in this case one expects the formation of a bow shock morphology at the companion , deep inside its magnetosphere . we explore radio emissions from substellar companions to giant stars using ( a ) the radiometric bode s law and ( b ) a model for a bow shock morphology . stars that are x - ray emitters likely have fully ionised winds , and the radio emission can be at the milli - jansky level in favorable conditions . non - coronal giant stars produce only micro - jansky level emissions when adjusted for low - level ionisations . if the largely neutral flow penetrates the magnetosphere , a bow shock results that can be strong enough to ionise hydrogen . the incoherent cyclotron emission is sub - microjansky . however the long wavelength radio emission of solar system objects is dominated by the cyclotron maser instability ( cmi ) mechanism . our study leads to the following two observational prospects . first , for coronal giant stars that have ionised winds , application of the radiometic bode s law indicates that long wavelength emission from substellar companions to giant stars may be detectable or nearly detectable with existing facilities . second , for the non - coronal giant stars that have neutral winds , the resultant bow shock may act as a `` feeder '' of electrons that is well - embedded in the companion s magnetosphere . incoherent cyclotron emissions are far too faint to be detectable , even with next generation facilities ; however , much brighter flux densities may be achievable when cmi is considered . stars : late type stars : mass loss stars : planetary systems radio continuum : stars
o and b stars are not distributed at random in the sky , but instead are usually concentrated in loose co - moving groups that @xcite first referred to as ob associations . ob associations are generally distributed along spiral arms , and confined in the galactic plane . they have low stellar densities of order 0.1 m@xmath2 pc@xmath3 , and are easily disrupted by galactic tidal forces @xcite so must be relatively young with ages @xmath4 myr ( see , e.g. , reviews by * ? ? ? some of the best evidence that the formation of massive stars can induce the formation of other low- to high - mass stars have been found in ob associations . on scales up to a few tens of parsecs , massive stars can ionize the surfaces of small molecular clouds ( globules ) in their vicinity @xcite ; the pressure from such ionization fronts can then propagate through the globule to help overcome the magnetic , turbulent , and thermal pressure that supports it against collapse , thereby inducing star formation . this process of radiation - driven implosion has been invoked to trigger star formation in many globules within ob associations @xcite , not least by one of us in the orion region @xcite and lac ob1 @xcite . on the other hand , star - forming globules in ob associations may simply be dense cores that have spontaneously collapsed to form stars , and were then uncovered by advancing ionization fronts from nearby massive stars . in the examples described in @xcite and @xcite , classical t tauri stars ( cttss ) are found only between ob stars and bright - rimmed ( i.e. , ionized surface of ) molecular clouds , not within these clouds ; furthermore , those cttss closer to the bright - rimmed clouds are progressively younger , arguing for a genuine sequence in star formation . on larger scales of tens of parsecs or more , expanding superbubbles driven by ionizing radiation and stellar winds from massive stars , together perhaps with supernova explosions , have been invoked to both sweep up and compress gas , and subsequently induce star formation ( see , e.g. , review by * ? ? ? such large - scale triggerings of star formation may be required to explain co - moving ob subgroups with either identical ages or arranged in an apparent age sequence , with at least some subgroups having an age spread that is much smaller than their stellar crossing time . one of the best such examples is the scorpius - centaurus ( scocen ) ob association ( see , e.g. , review by * ? ? ? * ) , which is the closest ob association to the sun . it comprises the three co - moving subgroups upper scorpius ( usco ) with an age of @xmath55 myr , upper centaurus - lupus ( ucl ) with an age of @xmath517 myr , and lower centaurus - crux ( lcc ) with an age of @xmath516 myr distributed from approximately east to west in galactic longitude over a spatial extent of @xmath5130 pc @xcite . the age dispersion in the usco subgroup ( the youngest ob subgroup ) is no more than @xmath512 myr , compared with a stellar crossing time of @xmath520 myr @xcite . @xcite and @xcite proposed that subsequent supernova explosions in lcc induced star formation in the molecular cloud that formed usco . the abovementioned picture of triggered star formation is based on an updated variation of the collect - and - collapse model of @xcite , where a propagating disturbance first collects gas into dense clouds and subsequently induces star formation . here we present our study of the perseus ob1 ( per ob1 ) association , which as shall show has properties that can be most naturally explained by the cloud - shuffling model of triggered star formation as first proposed by @xcite . cloud shuffling takes place on the periphery of an ob association where a superbubble pushes and accelerates a giant molecular cloud outwards , triggering star formation at the dense cloud interface @xcite . until now , cepheus ob3 ( cep ob3 ) has been regarded as one of the best examples where star formation has been triggered by cloud shuffling @xcite . cep ob3 contains two ob subgroups , the younger ( @xmath54 myr ) of which is closer to and has a radial velocity resembling the cep b molecular cloud , and the older ( @xmath5 8 myr ) further away and has a radial velocity resembling the rest of the cloud . @xcite and @xcite proposed that ionizing radiation and stellar winds from the older ob subgroup pushed and accelerated the cep b cloud away , as well as induced star formation in this cloud to form the younger subgroup . our paper is organized as follows . in 2 we collect together previous studies that compile the proposed members of per ob1 , and describe our study of the proper motion of those members included in the _ hipparcos _ catalog ; we also describe our search for low to intermediate - mass pre - main - sequence stars towards per ob1 , and our identification of possible remnant molecular clouds in this association . in 3 we interpret and discuss our results in the context of current star formation models . in 4 we briefly summarize the most important results of our study and their chief implications . the per ob1 association , located in the region of the perseus spiral arm , is commonly ( but not without argument ) thought to be centered about the @xmath0 and @xmath1 per double cluster ( see discussion in * ? ? ? . widely referred to as the double cluster , @xmath0 and @xmath1 per are among the brightest , densest , and closest of open clusters containing moderately massive ( main - sequence b , along with supergiant ) stars . both clusters have been studied extensively over the last century ( see references in * ? ? ? the latest study by @xcite conclude that the two clusters have essentially identical ages of @xmath6 myr , distances of @xmath7 kpc , comparable initial masses ( integrating between 1120 m@xmath2 ) of 3700 m@xmath2 for @xmath0 per and 2800 m@xmath2 for @xmath1 per , and a salpeter imf . although containing more than ten o stars , per ob1 is not associated with any known regions . furthermore , unlike virtually all other ob associations studied in molecular gas , per ob1 is not associated with any known giant molecular cloud . @xcite quote a failed attempt to detect molecular gas in unpublished @xmath8co observations by l. a. hillenbrand & j. m. carpenter using the fcrao . in addition , they find that the reddening to the @xmath0 and @xmath1 per double cluster is indistinguishable and produced entirely by line - of - sight extinction to the perseus spiral arm , with the intracluster reddening essentially zero . @xcite collected from the literature all the then known early and late - type supergiant stars , o stars , and both main - sequence as well as giant b0b1 stars that were proposed to be members of this association . she recomputed their photometric distances using the then latest available luminosity calibration for o and b stars , as well as for late - type supergiants . a total of 105 members were identified , all of which also have comparable radial velocities . all the member stars lie within the range in galactic longitude of @xmath9 and galactic latitude of @xmath10 . the most recent compilation from the literature of all proposed early - type luminous members of per ob1 was made by @xcite . by studying where the early - b ( b0b2 ) main - sequence stars lie on the color - magnitude diagram ( i.e. , cluster fitting method ) , @xcite derived a distance modulus for related members of @xmath511.8 , which corresponds to a distance of @xmath52.3 kpc . next to trigonometric parallaxes , this is the most secure method for identifying cluster members . @xcite also identified possible members from among the early - type giant to supergiant stars based on the then best available color - magnitude relations . in this way , they identified a total of 149 members in per ob1 , including @xmath5@xmath11 of the stars listed in @xcite as members of this association . note that @xcite only considered early - type ( o a ) stars , whereas the cataloged members of @xcite also include late - type ( supergiant ) stars . the per ob1 members identified by @xcite span a significantly larger range in galactic longitude of @xmath12 as well as galactic latitude of @xmath13 . in figure 1 , we plot the proposed members of per ob1 as cataloged by @xcite and @xcite in galactic coordinates . [ we exclude hd 237019 , whose position has been incorrectly entered in table 2a of @xcite . this star is a member of cas ob6 @xcite . ] a clustering of b stars ( green circles ) can be seen towards two closely separated positions , corresponding to the @xmath0 and @xmath1 per double cluster ( indicated by large black circles ) . no clear clustering is seen for the o stars ( blue circles ) , nor the supergiant stars ( red circles ) . the relative deficiency in cluster members at low galactic latitudes ( @xmath14 ) appears to be real ; as can be seen in figure 1 of @xcite for per ob1 , many of the early - type stars at @xmath15 were found to be either field stars or members of other associations . on the other hand , it is not clear to us whether the identification of luminous early - type stars is as complete at low galactic heights ( given the larger extinction and greater confusion ) as at larger heights . from a search of the literature since the work by @xcite , we have found three more luminous stars toward per ob1 that have a comparable distance modulus as inferred by @xcite for this association . these are the wc6 star hd 17638 ( distance 1.91 kpc , * ? ? ? * ) , the o9.5iii(n ) star hd 15137 ( distance 2.65 kpc , * ? ? ? * ) , and the on8v star hd 14633 ( distance 2.15 kpc , * ? ? ? the locations of these three stars are indicated in figure 2 . the radial velocities of hd 15137 and hd 14633 @xcite are closely comparable with the range observed for the per ob1 association . these two stars , especially hd 14633 , lie at a significantly higher galactic latitude than those cataloged by either @xcite or @xcite . the hipparcos astrometry mission @xcite permits a study of the proper motion of stars in ( or towards ) per ob1 . note that the stars in this association are too far away for hipparcos to obtain a measurement of their trigonometric parallax , which is predicted to be @xmath50.4 mas compared with measurement uncertainties of @xmath51 mas @xcite . sixty - one out of a total of 180 stars cataloged by @xcite and @xcite as members of per ob1 , as well as the three new additions mentioned in @xmath16 , are contained in the hipparcos catalog . the positions of these stars , converted to galactic coordinates , are listed in table [ table : members ] together with their hipparcos catalog ( hip ) number , proper motion in galactic longitude and latitude along with their associated standard errors , star name in the henry - draper ( hd ) or bonner durchmusterung ( bd ) catalog , spectral type , and whether these stars are listed as members by @xcite or @xcite or both . most of these stars are either early- or late - type giants and supergiants . in figure 2 , we plot the proper motion of the stars listed in table [ table : members ] in galactic coordinates . the blue , green , and red arrows are o , b , and supergiant stars respectively . the arrows are plotted with lengths proportional to the magnitude of their proper motions : note , however , that the uncertainty in proper motion changes from star to star , and the uncertainty in galactic longitude can be quite different to that in galactic latitude . we also plot the proper motion of the @xmath0 and @xmath1 per clusters ( black arrows ) as determined by @xcite based on space mission and ground - based observations . h per has a proper motion in ( @xmath17 , @xmath18 ) of ( @xmath19 mas yr@xmath20 , @xmath21 mas yr@xmath20 ) , and @xmath1 per a proper motion of ( @xmath22 mas yr@xmath20 , @xmath23 mas yr@xmath20 ) . for comparison , we also plot the proper motions of all the other early - type ( b0b9 ) stars ( yellow arrows ) in the hipparcos catalog within the field encompassed by figure 2 . these stars are referred to hereafter as `` field '' stars , although some may be members of other clusters or associations , and yet others perhaps previously unidentified members of per ob1 as we explain below . the white arrow oriented at a position angle of @xmath24 ( measured anticlockwise from galactic north ) indicates the direction towards the solar apex ; i.e. , the direction ( as projected in the sky ) towards which the sun is moving relative to the local standard of rest . as can be seen in figure 2 , some of the field stars ( yellow arrows ) have very large proper motions that in many cases lie at position angles of roughly @xmath25 . these are relatively nearby stars ( as confirmed from an examination of their trigonometric parallaxes ) that , because of our motion relative to the local standard of rest , appear to be moving in a direction opposite to that of the sun ( white arrow in fig . 2 ) . the proposed members of the per ob1 association , being much more distant , do now show any such appreciable motion . to study the actual proper motion of the proposed per ob1 members relative to the galaxy at their ( presumed ) distance ( 2.3 kpc ) and direction , we subtract our motion relative to the local standard of rest and galactic rotation at a distance of 2.3 kpc in the direction of the individual members . the correction for @xmath0 per is 0.27 mas yr@xmath20 in @xmath17 and @xmath26 mas yr@xmath20 in @xmath18 , giving an actual proper motion of ( @xmath27 mas yr@xmath20 , @xmath28 mas yr@xmath20 ) ; for @xmath1 per , the correction is @xmath29 mas yr@xmath20 in @xmath17 and @xmath26 mas yr@xmath20 in @xmath18 , giving an actual proper motion of ( @xmath30 mas yr@xmath20 , @xmath31 mas yr@xmath20 ) . thus , the double cluster does not exhibit significant motion relative to the galaxy rotation and galactic plane at its location . we tabulate the average measured proper motion of the remaining luminous member stars in galactic latitude strips of width @xmath32 from latitudes of @xmath33 to @xmath34 in table [ table : pm ] . ( note that the measured proper motion of the individual stars is in general of low statistical significance ) . the average proper motion in longitude is not significantly ( i.e. , @xmath35 ) different from zero except for the two individual stars at the two highest latitudes ( hd 15137 at @xmath36 and hd 14633 at @xmath37 ) . the average proper motion in latitude is significantly ( at least 4@xmath38 ) different from zero at latitudes @xmath39 , including the two individual stars at the two highest latitudes . to better visualize the results , in figure 3 we plot the proper motion of the individual members in galactic latitude , as well as their proper motion averaged over the different latitude strips . apart from the abovmentioned motion away from the galactic plane at @xmath39 , there is also a clear trend towards larger motions away from the galactic plane at higher latitudes , with the star at the highest latitude ( hd 14633 ) exhibiting the largest proper motion away from the galactic plane . note that the members of per ob1 cataloged by @xcite have in most cases radial velocities in the range @xmath40 to @xmath41 , the same range in radial velocities as the perseus spiral arm . we therefore conclude that the bulk motion of the stars in per ob1 is predominantly away from the galactic plane , with velocities up to a few tens of km s@xmath20 for most of the stars with measured proper motions . a visual inspection of figure 2 reveals that many of the `` field '' stars at latitudes @xmath42 ( yellow arrows ) spanning approximately the same longitude range as the proposed association members also exhibit a preferential motion away from the galactic plane . the average proper motion of the stars in this latitude strip and in the longitude range @xmath43 is ( @xmath44 mas / yr , @xmath45 mas / yr ) . their average proper motion in galactic latitude is comparable with the values tabulated in table [ table : members ] for the proposed member stars in this latitude strip , and hence these stars are likely genuine members of per ob1 . most of these stars have spectral types in the range b5b9 , and would therefore not have been included in the catalogs of either @xcite or @xcite . to study whether there has been any recent ( within the last @xmath51 myr or so , compared with an age for the @xmath0 and @xmath1 per double cluster for @xmath5 13 myr ) formation of low- to intermediate - mass stars in per ob1 , we have searched the two micron all - sky survey ( 2mass ) point source catalog @xcite for pre - main - sequence ( pms ) star candidates . our selection of such candidate stars was based on the empirical criteria developed by @xcite for low - mass cttss and intermediate - mass herbig ae / be ( haebe ) stars , but with improved criteria here for better selecting the latter stars . with this method , we first select stars showing near - infrared ( near - ir ) excess thereby preferentially picking out pms stars . because haebe stars generally exhibit a stronger near - ir excess than ctts , we then separated these two stellar classes based on their 2mass colors as described below . our initial list comprised all stars in the 2mass catalog at @xmath46 and @xmath47 ( i.e. , spanning a much wider area in the sky than the members of per ob1 as identified by * ? ? ? * ) that exhibit a near - ir excess . in figure 4 , we plot the ( j - h ) versus ( h - k ) color diagram for selecting pms star candidates . ctts candidates are those defined to lie between the two parallel lines @xmath48 and @xmath49 , and above the dereddened ctts locus @xmath50 @xcite . haebe candidates are defined as those lying between the two parallel lines @xmath49 and @xmath51 , and above the line @xmath52 . the line @xmath52 helps discriminates classical be stars from haebe stars . a full explanation for the adopted criteria will be presented elsewhere @xcite . in this way , we found a total of 399 candidate ctts and 264 candidate haebe stars . in figure 5 , we plot the spatial distribution of the candidate pms stars ( red for candidate ctts , green for candidate haebe stars , and blue for early - type members of per ob1 ) . at low galactic latitudes of @xmath53 , apart from three clusters associated with the known hii regions w3 , w4 , and w5 of the cas ob6 association , the candidate pms candidates ( which are presumably confused with cool evolved stars ) are roughly equally distributed in galactic longitude . in figure 5 , we also plot ( in grayscale ) the @xmath54 100-@xmath55 map ( dominated by emission from interstellar dust ) of the per ob1 region . the map in the right - hand panel has been stretched to accentuate features at high galactic latitudes ( @xmath53 ) coincident with the candidate pms stars selected in the manner described above . as can be seen , around the region occupied by the per ob1 association , a number of the candidate pms stars at galactic latitudes as low as about @xmath56 appear to be close to or coincident with dust clouds . an examination of optical images from the 2nd digitized sky survey ( dss2 ) as shown in figure 6 reveals bright - rimmed clouds , comet - shaped clouds , and reflection clouds coincident with these dust clouds . the abovementioned clouds seen in the dss2 images resemble remnant molecular clouds ( rmcs ; ogura et al . rmcs are small molecular clouds that can be commonly found in the vicinity of ob associations . their outlines are shaped by ionization fronts from massive stars , and hence they often have cometary morphologies with heads pointing towards the ionizing source . there is at least one such previously known example in per ob1 , referred to as red nebulous object 6 ( rno6 ; * ? ? ? @xcite found two molecular clouds , each having a mass of @xmath5@xmath57 , around rno6 . they showed that these clouds have cometary morphologies with their heads pointing towards the @xmath0 and @xmath1 per clusters . they also confirmed that rno6 hosts an embedded ir cluster that includes a hbe star responsible for illuminating rno6 , and suggested that the recent star formation in this rmc was triggered by radiatively driven implosion . as can be seen in figure 6 , we find another candidate haebe apparently associated with rno6 . ( the known hbe star in rno6 does not exhibit a near - ir excess in 2mass . ) based on figure 5 , we have identified four other candidate rmcs towards per ob1 that appear to be closely associated with candidate pms stars . ( we emphasize that our selection method does not provide a complete census of rmcs in this region . ) they are henceforth referred to as rmc1rmc4 with locations as indicated in figure 5 and tabulated in table [ table : rmc ] . a search of the catalog of bright nebulae compiled by @xcite reveals that rmc1 , rmc2 , and rmc3 are associated with lbn 638 , lbn 640 , and lbn 666 respectively . all four candidate rmcs have cometary morphologies as shown in figure 5 and 6 with heads pointing towards the galactic plane , rather than to any particular o stars in per ob1 . to determine if there are other sites of recent star formation in per ob1 apart from rno6 , we performed spectroscopic observations of the candidate pms stars close to or coincident with the abovementioned candidate rmcs ( including rno6 ) . the location of these stars are plotted in figure 6 , and tabulated in table [ table : spectra ] . we also observed four stars apparently associated with reflection nebulae located in the vicinity of , and likely associated with , the rmcs . these stars do not exhibit detectable near - ir excess in 2mass , and hence are not included in our list of candidate pms stars . the locations of these stars also are shown in figure 7 , and tabulated in table [ table : spectra ] . these stars will prove useful in obtaining a distant estimate to the rmcs . we obtained spectra of both the candidate pms stars and stars associated with reflection nebulae at the beijing astronomical observatory ( bao ) in 2005 and 2006 . low - dispersion spectra with a dispersion of 200 mm@xmath20 or 4.8 pixel@xmath20 were taken with a 2.16-m optical telescope on 2005 december 68 . we used an omr ( optomechanics research , inc . ) spectrograph together with a tektronix 1024@xmath581024 ccd detector covering the wavelength range 4000 - 9000 . on 2006 november 1113 , we used the same spectrograph but now deployed with a new princeton instrument spec10 400@xmath581300b ccd detector camera covering the wavelength range 3500 - 8500 . we processed the spectroscopic data in the standard manner using the noao / iraf package . after correcting for bias and flat - fielding , we used the iraf program kpnoslit to extract and calibrate the wavelength and intensity of each spectrum . the results for the individual stars are presented in table [ table : spectra ] . ctts and haebe stars are identified by the presence of h@xmath59 emission , and separated by the presence of other balmer lines seen in absorption in haebe stars . we separate intermediate - mass stars into haebe , b , and a stars based on the spectral classification scheme of @xcite . we do not , however , attempt to determine the spectra types of ctts because their spectra exhibit heavy veiling . of the nine candidate pms stars selected by 2mass colors and observed , we confirmed that four are haebe stars and five are possibly t tauri stars . detection of the li line , impossible in our low - dispersion spectra , is required to confirm the nature of the ctts candidates . we find pms stars in the vicinity of or coincident with all four rmcs . as there are no known foreground ( or background ) star - forming regions in their directions , and given their large heights above the galactic plane ( see more detailed discussion in @xmath60 ) , their association with the rmcs is likely genuine . the rmcs are therefore likely to be regions of recent or ongoing star formation . of the four stars ( stars 5 , 6 , 8 , and 10 ) associated with reflection nebulae studied , three were found to be b main - sequence stars ( stars 5 , 6 , and 10 ) and one an a main - sequence star ( star 8) . we now use the b or a stars associated with reflection nebulae to determine a photometric distance to these stars and hence to two of the candidate rmcs . specifically , we will focus on 2mass j01494099@xmath615341341 ( b7v ; star 5 in table [ table : spectra ] ) , 2mass j01500456 + 5354007 ( b3v ; star 6 ) and 2mass j01501546@xmath615341350 ( a2v ; star 8) apparently associated with rmc2 , and 2mass j02220484@xmath615038130 ( b8v ; star 10 ) with rmc3 . to derive the absolute magnitudes in the j - band of stars with the abovementioned spectral classes , we selected eleven b3v , three b7v , forty - one b8v , and 128 a2v stars with good parallax ( @xmath62 ) measurements , [ relative error in parallax of @xmath63 @xcite , goodness - of - fit parameter @xmath64 3 , and percentage of rejected data @xmath65 @xcite ] , and good 2mass photometric qualities . by selecting only stars with small relative errors in parallax , we hope to minimize the bias in determining absolute magnitudes and hence distances using this method . as pointed out by @xcite , the inclusion of stars with large relative errors in parallax biases the final result to a lager parallax than the true parallax . based on measurements of their apparent magnitudes in the 2mass catalog , we then derived median absolute magnitudes of @xmath660.652 , 0.131 , 0.484 , and 1.176 for the b3v , b7v , b8v , and a2v stars respectively in the j - band . based on their apparent magnitudes in the j - band in the 2mass catalog , the corresponding distance modulus of 2mass j01500456 + 5354007 ( b3v ) is 11.324 , 2mass j01494099@xmath615341341 ( b7v ) is 11.831 , and 2mass j01501546@xmath615341350 ( a2v ) is 12.457 ; i.e. , a distance of 1840 pc , 2323 pc , and 3100 pc to rmc2 . similarly , the distance modulus of 2mass j02220484@xmath615038130 ( b8v ) is 12.327 ; i.e. , a distance of 2920 pc to rmc3 . all the estimated distance module agree with the range measured by @xcite to per ob1 of magnitude 10.83 12.97 . the inferred distances to rmc2 and rmc3 are therefore compatible with their location in per ob1 . as mentioned above , all four candidate rmcs appear to be associated with pms stars . if at the distance of the per ob1 association , then these star - forming regions are located at unprecedented heights above the galactic plane for known clusters embedded in molecular clouds . in figure 8 , we plot the height above the galactic plan of seventy - six such embedded clusters compiled by @xcite within @xmath5 2 kpc of the sun , as well as the four rmcs studied here ( assuming a distance to these rmcs of 2.3 kpc ) . as can be seen , @xmath5@xmath67 of these clusters are located within @xmath5160 pc of the galactic plane . only two of the embedded clusters cataloged by @xcite are located beyond a height of 200 pc , ngc 281w and ngc 281e , both of which extend no one higher than 240 pc . by comparison , all four rmcs studied here are situated more than 280 pc away from galactic plane , with rmc3 having the largest height of @xmath5390 pc . as mentioned in @xmath68 , per ob1 is unusual in that it is not associated with a giant molecular cloud . based on the results described in @xmath69 , we now find that per ob1 is unusual in at least two other respects . first , most of its members exhibit a bulk motion away from the galactic plane with those at larger heights moving faster ( on average ) away from the galactic plane . second , star formation has recently occurred ( and may still be occurring ) at unusually large heights of 280400 pc above the galactic plane . we now examine whether all of these features can be explained , and if so how , in the framework of current star formation models . star formation at the present epoch is thought to be relatively inefficient in converting molecular gas to stars , even more so in models that invoke spontaneous as opposed to triggered star formation . to explain the lack of a giant molecular cloud associated with per ob1 without invoking an extraordinarily effective dispersement or destruction mechanism , one would have to propose that the star formation in this association has been more efficient than normal in converting molecular gas to stars . this may , but is not necessarily required to , explain the unusually large stellar densities and masses of the @xmath0 and @xmath1 per double cluster . the high number density of stars and their resultant bipolar outflows , stellar winds , uv ionization , and/or supernova explosions would then be more effective also in sweeping away or destroying much of the remaining molecular gas . in a search for shells and supershells in the galaxy , @xcite identified a large shell of atomic hydrogen ( ) gas in the direction towards per ob1 . @xcite show that this shell occupies the same range in radial velocities as per ob1 , and forms a quasi - circular structure of size @xmath5@xmath70 pc centered at ( @xmath71)=([email protected] , @xmath73.5 ) . the center of this shell therefore lies quite close to the locations of the @xmath0 and @xmath1 per double cluster . @xcite estimate a total mass of atomic gas in the shell of order @xmath74 m@xmath2 . they infer an an upper limit of 10 km s@xmath20 for the expansion velocity of the shell , and a kinetic energy for the shell of @xmath75 ergs . they also compute a minimum mechanical energy to blow the shell of @xmath76 ergs , compared with an energy released through stellar winds by all the known member b stars of @xmath77 ergs . thus , stellar winds from b stars alone in per ob1 is sufficient to blow the observed shell , let alone an additional energy from sources such as supernova explosions . we note that the total kinetic energy of the member stars due to their motion away from the galactic plane ( @xmath78 ergs ) adds insignificantly to the overall energetics . @xcite show from the _ iras _ 100-@xmath79 m image that emission from dust traces the inner border of the shell in the portion at large heights above the galactic plane ( any dust associated with the shell close to the galactic plane is confused with galactic dust ) . this dust feature can be seen as a nearly semi - circular arc in figure 2 of @xcite , extending to a galactic latitude of about @xmath80 at its largest height above the galactic plane . we speculate that this dust shell may have formed from the swept - up giant molecular cloud . we have checked the literature to see if the presumed hot bubble of gas responsible for sweeping up the abovementioned hi and dust shell is visible in x - rays . such x - ray emission is detectable in the rosat all sky survey ( rass ) @xcite from the orion - eridanus superbubble @xcite , which is powered by ori ob1 ( which contains approximately the same number of o stars as per ob 1 ) . this superbubble has approximately the same physical size as that enclosed by the hi shell associated with per ob1 . the kinematic energy of the orion - eridanus superbubble , as derived from its associated expanding shell , is about @xmath81 ergs @xcite , which is between one and two orders of magnitude larger than that of the presumed per ob1 superbubble . if the orion - eridanus superbubble was located at the distance of per ob1 , the count rate range of 3/4 kev band would be about 515 counts s@xmath20 . by comparison , the background count rate towards the presumed per ob1 superbubble is about 110 counts s@xmath20 . perhaps not surprisingly then , we did not find any significant x - ray emission from rass in 1/4 , 3/4 , and 1.5 kev bands towards the presumed per ob1 superbubble . we have searched the literature to find maps in molecular gas towards per ob1 . @xcite has mapped in co(1 - 0 ) the sky in the direction towards per ob1 at an angular resolution of 85 . compact and isolated co emission is detectable towards rmc1 , rmc2 , and rmc3 , but not towards rmc4 . if these sources are genuinely associated with the abovementioned remnant molecular clouds , the corresponding molecular gas mass is in these clouds @xmath5 8000 m@xmath2 . no co emission was detected towards rno6 , which has a molecular gas mass of @xmath5 400 m@xmath2 , consistent with the upper limit in sensitivity of the @xcite map . no co emission was detected towards the dust/ shell either ; assuming a co linewidth of 10 km s@xmath20 , the corresponding 3@xmath38 upper limit in molecular gas along this shell is @xmath82 m@xmath2 . if this estimate is accurate , the bulk of the molecular gas in this shell may therefore have been dissociated into atomic hydrogen gas . much of the star formation in the per ob1 association appears to have occurred in or within the vicinity of the @xmath0 and @xmath1 per double cluster . if so , then the observed increase ( on average ) in velocity with height ( acceleration ) away from the galactic plane may be a natural consequence of the dissolution of these clusters @xcite . over time , stars moving faster away from the double cluster would travel furthest from their initial locations , thus producing the observed linear increase in velocity with height above the galactic plane . although this model could explain the velocity pattern of the b main - sequence and supergiant stars , which can have ages up to @xmath5 10 myr of more , it can not easily explain the same velocity pattern observed for the o stars . the latter stars have ages of only a few millions years , much younger than the b stars in the @xmath0 and @xmath1 per double cluster ( @xmath5 13 myr ) , and could not have reached their present locations if traveling at their observed velocities from these clusters . this model also predicts that the member stars should form an approximately spherical distribution moving away from the @xmath0 and @xmath1 per double cluster , which is not seen ( c.f . fig . 2 ) . given that the molecular gas from which the stars in per ob1 formed appears to have been effectively swept outwards by the luminous stars in this association ( @xmath83 ) , we now consider whether the observed velocity dependence with height could simply be a reflection of stars forming in this swept - up gas . the situation envisaged is akin to that proposed in models of triggered star formation , in particular the scenario of cloud shuffling . cloud shuffling takes place on the periphery of an ob association where an over - pressured superbubble of accumulated stellar winds , supernova explosions , and/or ionized gas pushes and accelerates a giant molecular cloud outwards @xcite . such cloud shuffling is proposed to induce star formation , and also extend the lifetimes of molecular clouds by pushing them away from massive stars and their destructive effects . in the present case , the dust shell at the inner boundary of the shell presumably lies just beyond the expansion front of the superbubble associated with per ob1 . assuming typical physical conditions in ob associations , @xcite suggests that cloud shuffling can accelerate molecular gas by about 1 km s@xmath20 myr@xmath20 ( or equivalently 1 pc myr@xmath84 ) . let us assume that the @xmath0 and @xmath1 per double cluster was the original source of the superbubble ( with luminous stars formed later from the swept - up molecular gas adding to the energy of this superbubble ) , consistent with its location close to the center of the shell . the age of this double cluster is estimated to be @xmath513@xmath851 myr @xcite , allowing sufficient time for the molecular gas at the edge of the superbubble to be accelerated by over @xmath86 . this is the correct order of magnitude to produce , given the uncertainties involved in estimating the magnitude of the acceleration and its possible variation with time , the observed velocity dependence with height of the stars in per ob1 . the proposed scenario provides a natural explanation for the relatively large height of hd 15137 above the galactic plane , and its motion . @xcite suggest that both the massive stars hd 15137 and hd 14633 are runaway stars ejected from the open cluster ngc 654 in the perseus spiral arm . this suggestion , however , has a serious timescale problem , with inferred flight times for both stars that are longer than their ages @xcite . if hd 15137 is a member of per ob1 as proposed in @xmath87 , then in our proposed picture it formed close to its present location and inherited its relatively large velocity away from the galactic plane from its swept - up parent molecular cloud . note that the o star lying at the next highest galactic latitude , hd 15137 ( location as indicated in fig . 2 ) , has a comparably large proper motion away from the galactic plane . this explanation , however , can not be simply applied to hd 14633 , which we also proposed in @xmath87 to be a possible member of per ob1 . this star is located at a galactic latitude of @xmath88 , which is far beyond the presumed outer edge of the superbubble at a galactic latitude of about @xmath80 . the triggered formation of stars in the cloud - shuffling model naturally predicts an age gradient from the center of the superbubble , such that stars at larger radii have younger ages . if this model is applicable , the supergiant stars in per ob1 should therefore be closer to the @xmath0 and @xmath1 per double cluster ( the presumed center of the superbubble ) than the o stars . in figure 9 we plot the spatial distributions of o stars ( dotted lines ) and supergiant stars ( dashed lines ) proposed to be members of per ob1 from @xcite . using the kolmogorov - smirnov test @xcite , we find a @xmath89 probability that these two populations have different distributions in galactic height , such that the younger o tend to be located higher above the galactic plane than the older supergiant stars ( the visual impression given in fig . thus , there does seem to be a tendency , albeit only of marginally significance , for younger stars to be located at larger heights above the galactic plane . in the cloud shuffling model as applied to per ob1 , the relatively recent ( within the last @xmath51 myr ) formation of low- to intermediate - mass stars in remnant molecular clouds at unusually large heights above the galactic plane simply reflects the transport of molecular gas to such heights by the expanding superbubble . the dust and shells described in @xmath83 extend to galactic latitudes of about @xmath80 ( height @xmath5 450 pc above the galactic plane ) , and so the remnant molecular clouds described in @xmath90 lies in projection inside the superbubble . these clouds may actually comprise swept - up molecular gas located at the periphery of the superbubble . alternatively , they may comprise the densest condensations in the giant molecular cloud that were not completely swept up by the expanding superbubble . as mentioned in @xmath91 , all the rmcs have head - tail morphologies with their heads pointing towards the galactic plane rather than in the direction of any particular o star . these morphologies of these rmcs may therefore have been sculpted by the expanding superbubble rather than ionizing radiation from o stars . the per ob1 association comprises the remarkable @xmath0 and @xmath1 per double cluster ( among the brightest and densest open clusters known ) , surrounded by luminous o / b and both early- and late - type supergiant stars . unlike virtually all other known ob associations in the galaxy , there is no known giant molecular cloud in the vicinity of per ob1 . we have gathered together the available information on the stellar population of per ob1 , primarily from the work by @xcite and @xcite . for those luminous early- and late - type stars included in the hipparcos astrometry mission , we have extracted measurements of their proper motions . we found that : we also searched for candidate intermediate - mass ( haebe ) and low - mass ( classical t tauri star ; ctts ) pre - main - sequence stars in per ob1 . we find a number of these candidate stars to be associated with four candidate remnant molecular clouds ( rmcs ) identified as dust clouds in _ iras _ 100-@xmath79 m images and optical nebulae in the 2nd digitized sky survey . all these rmcs are located high ( 280400 pc in projection ) above the galactic plane . from our spectroscopy of both candidate pre - main - sequence as well as main - sequence b a stars apparently associated with these rmcs , we : 1 . show that the distance moduli to the main - sequence b a stars are consistent with the rmcs being located in per ob1 2 . confirm that some of the candidate pre - main - sequence stars are indeed haebe stars and identify possible cttss , implying that there has been recent ( within the last 1 myr ) star formation at galactic heights far above the locations of known luminous members in per ob1 1 . point out that the large shell of atomic hydrogen ( ) gas ( mass @xmath5@xmath92 ) enclosing a slightly smaller shell of dust , studied by @xcite and which they propose to be created by a stellar - wind - blown superbubble from the @xmath0 and @xmath1 per double cluster and ob stars in per ob1 , likely comprises ( in part ) the swept - up remains of the giant molecular cloud from which per ob1 formed 2 . argue that if the observed motion of the luminous member stars is attributed to the dissolution of the @xmath0 and @xmath1 per double cluster ( at their outskirts ) , this can only explain the motion of the main - sequence b and late - type supergiant stars . the main - sequence o stars , which are much younger than the double cluster , would not have had sufficient time to reach their present locations high above the galactic plane if they originated from the close vicinity of the double cluster 3 . show that the observed motion of the luminous member stars is better explained in the cloud - shuffling model @xcite , in which the abovementioned superbubble pushes and accelerates the giant molecular cloud outwards and at the same time induces star formation at the dense cloud interface . this would naturally result in those stars forming locally in the compressed molecular gas to have increasingly large velocities away from galactic plane with height 4 . show that there is a weak trend for the younger o member stars to lie at higher galactic latitudes than the older late - type supergiant member stars , consistent with the simplest prediction of the cloud - shuffling but not cluster dissolution model 5 . suggest that the recent star formation in the remnant molecular clouds that we found at very large heights above the galactic plane , far above the locations of the luminous member stars , likely occurred in the swept - up molecular gas at the periphery of the superbubble , or in the densest condensations that were not entirely swept away by the expanding superbubble h .- lee thanks wen - ping chen for sharing observation time , wen - shan hsiao for helping spectral observations , rue - ron hsu and pin - gao gu for discussing , and the staff at the beijing astronomical observatory for their assistance during observing runs . we are grateful to the anonymous referee for constructive suggestions . this research makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation . this research made use of montage , funded by the national aeronautics and space administration s earth science technology office , computation technologies project , under cooperative agreement number ncc5 - 626 between nasa and the california institute of technology . montage is maintained by the nasa / ipac infrared science archive . this research made use of data products from the digitized sky surveys produced at the space telescope science institute . ambartsumian , v. a. 1947 , in stellar evolution and astrophysics , armenian acad . of sci . ( german transl . , 1951 , sowjetischen astron , 1 , 33 ) bachiller , r. , fuente , a. , & kumar , m. s. n. 2002 , , 381 , 168 bok , b. j. 1934 , harvard college observatory circular , 384 , 1 boyajian , t. s. , et al . 2005 , , 621 , 978 bragg , a. e. , & kenyon , s. j. 2005 , , 130 , 134 brown , a. g. a. , hartmann , d. , & burton , w. b. 1995 , , 300 , 903 brown , a. g. a. , arenou , f. , van leeuwen , f. , lindegren , l. , & luri , x. 1997 , hipparcos - venice 97 , 402 , 63 cappa , c. e. , & herbstmeier , u. 2000 , , 120 , 1963 cohen , m. 1980 , , 85 , 29 cutri , r. m. , et al . 2003 , 2mass all sky catalog of point sources ( pasadena : ipac ) dame , t. m. , hartmann , d. , & thaddeus , p. 2001 , , 547 , 792 de geus , e. j. , de zeeuw , p. t. , & lub , j. 1989 , , 216 , 44 de zeeuw , p. t. , hoogerwerf , r. , de bruijne , j. h. j. , brown , a. g. a. , & blaauw , a. 1999 , , 117 , 354 elmegreen , b. g. & lada , c. j. 1977 , , 214 , 725 elmegreen , b. g. 1979 , , 231 , 372 elmegreen , b. 1992 , in star formation in stellar systems , ed . g. tenorio - tagle , m. prieto , & f. sanchez ( cambridge : cambridge univ . press ) , 381 esa . 1997 , the hipparcos and tycho catalogs , 1997 , esa sp-1200 garmany , c. d. , & stencel , r. e. 1992 , , 94 , 211 geyer , m. p. , & burkert , a. 2001 , , 323 , 988 heiles , c. 1979 , , 229 , 533 humphreys , r. m. 1978 , , 38 , 309 jacoby , g. h. , hunter , d. a. , & christian , c. a. 1984 , , 56 , 257 keller , s. c. , grebel , e. k. , miller , g. j. , & yoss , k. m. 2001 , , 122 , 248 kharchenko , n. v. , piskunov , a. e. , rser , s. , schilbach , e. , & scholz , r .- d . 2005 , , 438 , 1163 kovalevsky , j. 1984 , space science reviews , 39 , 1 kroupa , p. , aarseth , s. , & hurley , j. 2001 , , 321 , 699 lada , c. j. , & lada , e. a. 2003 , , 41 , 57 lee , h .- t . , chen , w. p. , zhang , z. , & hu , j. 2005 , , 624 , 808 lee , h .- , & chen , w. p. 2007a , , 657 , 884 lee , h .- t . et al . 2007b , in preparation lutz , t. e. , & kelker , d. h. 1973 , , 85 , 573 lynds , b. t. 1965 , , 12 , 163 mamajek , e. e. , meyer , m. r. , & liebert , j. 2002 , , 124 , 1670 meyer , m. r. , calvet , n. , & hillenbrand , l. a. 1997 , , 114 , 288 ogura , k. , & sugitani , k. 1998 , publ . astron . australia , 15 , 91 oudmaijer , r. , groenewegen , m. a. t. , & schrijver , h. 1999 , , 341 , l55 perryman , m. a. c. , et al . 1997 , , 323 , l49 preibisch , t. , & zinnecker , h. 1999 , , 117 , 2381 preibisch , t. , brown , a. g. a. , bridges , t. , guenther , e. , & zinnecker , h. 2002 , , 124 , 404 preibisch , t. , & zinnecker , h. 2007 , iau symposium , 237 , 270 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. numerical recipes : the art of scientific computing ( 3rd ed . ; cambridge university ) reipurth , b. 1983 , , 117 , 183 sargent , a. i. 1979 , , 233 , 163 slesnick , c. l. , hillenbrand , l. a. , & massey , p. 2002 , , 576 , 880 snowden , s. l. , et al . 1995 , , 454 , 643 snowden , s. l. , burrows , d. n. , sanders , w. t. , aschenbach , b. , & pfeffermann , e. 1995 , , 439 , 399 sugitani , k. , fukui , y. , & ogura , k. 1991 , , 77 , 59 van der hucht , k. a. 2001 , new astronomy review , 45 , 135 van steenberg , m. e. , & shull , j. m. 1988 , , 67 , 225 per ( east ) and @xmath1 per ( west ) double cluster . the locations of remnant molecular clouds ( rmcs ) identified in this work , as well as the remnant molecular cloud rno6 are indicated . the dashed line indicates the approximate geometry of the shell found by @xcite and described in 3.1 . ] and @xmath1 per double cluster . yellow arrows are for other stars in the _ hipparcos _ catalog lying in the region shown . the length of each arrow is proportional to the magnitude of its proper motion . the grayscale is the @xmath54 100 @xmath55 image . ] and @xmath1 per double cluster ( triangles ) by @xcite . also plotted are the tangential velocities for all proposed members in the _ hipparcos _ averaged over latitude strips of @xmath93 ( squares ) . note that all of the proper motions have been corrected for the solar motion . ] 100 @xmath55 image using two different stretches . the left panel emphasizes regions with the strongest extinction , and the right panel regions with lower extinctions to more clearly show the rmcs identified in this work . note that only those candidate ctts and haebe stars lying close to or against the rmcs were selected for follow - up spectroscopy . ] 100 @xmath55 image . locations of the rmcs identified in this work , as well as rno6 , are indicated . also indicated are the locations of main - sequence b or a stars ( blue ) used to estimate distances to the rmcs . the stars are numbered according to their identification numbers in table 4 . ] 130.900 & -3.441 & 8725 & 1.05 & 2.49 & 1.38 & 1.13 & hd 236894 & o8v & a , b + 131.479 & -2.433 & 9211 & 0.65 & -1.88 & 1.03 & 0.87 & hd 236915 & m2iab & b + 131.867 & -5.138 & 9017 & -0.85 & -3.01 & 1.46 & 1.33 & bd+55 441 & b1vpe & a + 132.691 & -2.588 & 9886 & -0.12 & -1.71 & 1.11 & 1.05 & hd 236947 & m2iab - ib & b + 132.742 & -1.700 & 10078 & 1.40 & -0.03 & 1.49 & 1.10 & hd 13036 & b0.5iii & a , b + 132.975 & -5.802 & 9523 & 0.77 & -5.80 & 1.00 & 1.02 & hd 12323 & on9v & a , b + 132.977 & -2.504 & 10060 & -0.13 & -0.26 & 1.00 & 0.98 & hd 13022 & o9.5ia & a , b + 132.981 & -2.880 & 9990 & -0.30 & -1.80 & 0.58 & 0.61 & hd 12953 & a1ia & a , b + 133.105 & -4.282 & 9826 & 0.72 & -3.10 & 1.12 & 0.98 & hd 12727 & b2iii & a + 133.173 & -6.153 & 9582 & 1.25 & -3.21 & 0.71 & 0.78 & hd 12401 & m4ib & b + 133.168 & -1.650 & 10323 & -0.30 & -1.28 & 1.49 & 1.10 & hd 13402 & b0.5ib & a , b + 133.174 & -3.334 & 10024 & -0.87 & -1.39 & 1.16 & 0.99 & hd 12993 & o6.5v & a , b + 133.270 & -4.170 & 9935 & -0.51 & -2.15 & 0.94 & 0.87 & hd 12856 & b0iiipe & a , b + 133.505 & -4.217 & 10055 & 0.72 & -2.47 & 1.09 & 0.98 & hd 13051 & b1ive & a + 133.565 & -2.549 & 10379 & 0.09 & -0.54 & 0.65 & 0.65 & hd 13476 & a3iab & a , b + 133.571 & -3.509 & 10227 & -0.14 & -1.92 & 0.66 & 0.61 & hd 13267 & b5ia & a , b + 133.888 & -2.882 & 10489 & -3.66 & 0.02 & 1.16 & 1.11 & hd 13658 & m1iab & b + 134.013 & -1.219 & 10904 & -0.90 & 0.50 & 1.69 & 1.43 & hd 14242 & m2iab & b + 134.030 & -4.924 & 10228 & 1.90 & -2.83 & 0.92 & 0.74 & hd 13268 & o7 : & a , b + 134.064 & -3.223 & 10527 & 0.90 & -2.15 & 1.03 & 0.90 & hd 13716 & b0.5iii & a , b + 134.130 & -1.931 & 10829 & 1.78 & -2.03 & 1.36 & 1.08 & hd 14142 & m2iab & b + 134.445 & -3.868 & 10624 & 0.23 & -2.39 & 0.77 & 0.69 & hd 13841 & b2ib & a , b + 134.448 & -3.840 & 10633 & -0.60 & -1.02 & 0.54 & 0.52 & hd 13854 & b1iab & a , b + 134.525 & -4.149 & 10615 & -0.51 & -2.68 & 0.86 & 0.81 & hd 13831 & b0iiie & a , b + 134.571 & -4.154 & 10641 & -0.85 & -1.85 & 0.68 & 0.66 & hd 13866 & b2ib & a , b + 134.649 & -4.891 & 10541 & 1.16 & -4.52 & 0.83 & 0.76 & hd 13745 & o9.7ii((n ) ) & a , b + 134.685 & -2.129 & 11093 & 3.21 & -2.09 & 1.98 & 1.71 & hd 14528 & m4ie & b + 134.705 & -5.589 & 10443 & -0.90 & -2.69 & 0.99 & 0.92 & hd 13621 & b0.5iii - iv((n ) ) & a + 134.705 & -3.665 & 10805 & 0.24 & -2.03 & 0.68 & 0.59 & hd 14134 & b3ia & a , b + 134.714 & -3.626 & 10816 & -0.64 & -1.76 & 0.66 & 0.59 & hd 14143 & b2ia & a , b + 134.792 & -2.861 & 10995 & 0.56 & -3.45 & 0.84 & 0.77 & hd 14404 & m2iab & b + 135.033 & -3.433 & 11020 & -0.50 & -2.13 & 0.61 & 0.57 & hd 14433 & a1ia & a , b + 135.058 & -1.678 & 11394 & 0.21 & -0.20 & 1.33 & 0.97 & hd 14947 & o5if+ & a , b + 135.120 & -3.247 & 11115 & -0.84 & -1.52 & 0.72 & 0.63 & hd 14542 & b8ia & a , b + 135.143 & -3.752 & 11018 & 1.46 & -2.86 & 0.99 & 0.84 & hd 14434 & o6.5 : & a , b + 135.155 & -3.385 & 11098 & -1.34 & -1.31 & 0.75 & 0.66 & hd 14535 & a2iab & a , b + 135.334 & -4.743 & 10926 & 0.10 & -1.02 & 0.79 & 0.63 & hd 14322 & b8ib & a , b + 135.381 & -4.815 & 10935 & 0.90 & -1.00 & 1.05 & 0.83 & hd 14331 & b0iii & a , b + 135.400 & -3.088 & 11284 & 2.05 & -1.52 & 1.32 & 1.36 & hd 14826 & m2iab & b + 135.488 & -2.795 & 11391 & 0.94 & -1.21 & 0.62 & 0.59 & hd 14956 & b2ia & a , b + 135.573 & -4.724 & 11060 & 0.76 & -0.99 & 0.66 & 0.54 & hd 14489 & a2iab & a , b + 135.683 & -3.866 & 11279 & -0.13 & -0.75 & 0.51 & 0.50 & hd 14818 & b2ia & a , b + 135.835 & -2.506 & 11625 & 0.07 & 0.58 & 0.68 & 0.63 & hd 15316 & a2ia & a , b + 136.119 & -2.524 & 11769 & 0.55 & 0.37 & 0.73 & 0.67 & hd 15497 & b6ia & a , b + 136.128 & -1.754 & 11953 & 0.73 & -0.30 & 1.21 & 0.98 & hd 15752 & b0iiin & a , b + 136.152 & -2.260 & 11841 & -0.92 & 0.24 & 1.01 & 0.90 & hd 15620 & b8iab & a , b + 136.323 & -0.371 & 12416 & -2.23 & -1.21 & 1.23 & 1.13 & bd+58 501 & m2iab & b + 136.351 & -3.293 & 11722 & 1.10 & -1.38 & 1.22 & 1.14 & hd 15450 & b1iiie & a + 136.385 & -2.587 & 11898 & 0.33 & -3.51 & 1.88 & 1.75 & hd 15690 & b2iab & a , b + 136.535 & -3.468 & 11782 & 0.34 & -1.86 & 1.37 & 1.30 & hd 15548 & b1v & a + 136.695 & -0.046 & 12750 & -0.06 & -0.04 & 0.74 & 0.67 & hd 16778 & a2ia & b + 137.159 & -4.660 & 11850 & 0.05 & -3.11 & 1.11 & 1.05 & hd 15642 & o9.5iii : n & a , b + 137.185 & -2.779 & 12302 & 1.33 & -3.09 & 0.86 & 0.86 & hd 236979 & m1iab & b + 137.351 & -1.300 & 12769 & -0.84 & -1.73 & 1.50 & 1.20 & hd 16808 & b0.5ib & a , b + 137.367 & -1.044 & 12840 & -0.58 & -1.36 & 1.27 & 1.09 & hd 236995 & a0ia & a , b + 137.532 & -7.507 & 11473 & 2.53 & -4.48 & 0.67 & 0.72 & hd 15137 & o9.5iii(n ) & c + 137.796 & -2.660 & 12676 & -0.65 & -4.71 & 1.04 & 1.14 & hd 16691 & o5f & a + 137.932 & -1.681 & 12972 & 1.01 & -1.89 & 0.94 & 0.93 & hd 17088 & b9ia & a , b + 138.022 & -1.701 & 13022 & 0.69 & -1.81 & 1.10 & 1.13 & hd 17145 & b8ia & a , b + 138.383 & -1.326 & 13290 & -0.36 & -1.51 & 1.61 & 1.46 & hd 237010 & m2iab & b + 138.537 & -2.110 & 13178 & -0.03 & -1.72 & 0.63 & 0.69 & hd 17378 & a5ia & a , b + 138.718 & -2.133 & 13262 & 1.21 & -5.23 & 1.01 & 1.03 & hd 237008 & m3iab & b + 138.930 & -2.081 & 13380 & 1.49 & -3.08 & 1.44 & 1.54 & hd 17638 & wc6 & c + 140.863 & -18.126 & 11099 & 3.13 & -6.05 & 0.87 & 0.69 & hd 14633 & on8v & c + [ table : members ] references(a ) @xcite ; ( b ) @xcite ; ( c ) this paper rrrrrr 0 @xmath94 & @xmath95 & @xmath96 & 0.72 & 0.66 & 2 + @xmath94 @xmath97 & 0.44 & @xmath98 & 0.42 & 0.35 & 11 + @xmath97 @xmath99 & 0.46 & @xmath100 & 0.26 & 0.24 & 18 + @xmath99 @xmath101 & 0.41 & @xmath102 & 0.24 & 0.22 & 16 + @xmath101 @xmath103 & 0.67 & @xmath104 & 0.28 & 0.25 & 11 + @xmath103 @xmath105 & 0.06 & @xmath106 & 0.68 & 0.64 & 3 + @xmath105 @xmath107 & 1.64 & @xmath108 & 0.71 & 0.78 & 1 + @xmath107 @xmath109 & 2.58 & @xmath110 & 0.67 & 0.72 & 1 + ......... + @xmath111 @xmath112 & 2.93 & @xmath113 & 0.87 & 0.69 & 1 + [ table : pm ] 1 & j01380790 + 5453463 & h@xmath114 & haebe ( a0 ) & around rmc1 + 2 & j01424061 + 5449135 & h(@xmath6652.5 ) & ctts & in rmc1 + 3 & j01433441 + 5451323 & h(@xmath666.1 ) & ctts & in rmc1 + 4 & j01435138 + 5526098 & h@xmath114 & haebe ( b3 ) & in rmc1 + 5 & j01494099 + 5341341 & & b ( b7v ) & with a reflection nebula in rmc2 + 6 & j01500456 + 5354007 & & b ( b3v ) & with a reflection nebula in rmc2 + 7 & j01501404 + 5341107 & h(@xmath6689.1 ) & ctts & in rmc2 + 8 & j01501546 + 5341350 & & a ( a2v ) & with a reflection nebula in rmc2 + 9 & j02151126 + 5519500 & h@xmath114 & haebe ( b8 ) & around rno 6 + 10 & j02220484 + 5038130 & & b ( b8v ) & with a reflection nebula in rmc3 + 11 & j02220662 + 5040013 & h(@xmath662.2 ) & wtts & in rmc3 + 12 & j02261522 + 5243131 & h(@xmath6616.1 ) & ctts & with a reflection nebula in rmc4 + 13 & j02262049 + 5244288 & h@xmath114 & haebe ( b6v ) & with a reflection nebula in rmc4 + [ table : spectra ]	the per ob1 association , which contains the remarkable double cluster @xmath0 and @xmath1 per , is unusual in not having a giant molecular cloud in its vicinity . we show from _ hipparcos _ data that the luminous members of this association exhibits a bulk motion away from the galactic plane , such that their average velocity increases with height above the galactic plane . we find haebe and t tauri stars towards probable remnant molecular clouds associated with per ob1 . these star - forming regions lie well beyond the location of the luminous member stars at heights of 280400 pc above the galactic plane , far higher than that previously found for embedded clusters . we argue that the observed motion of the luminous member stars is most naturally explained if many formed from molecular gas pushed and accelerated outwards by an expanding superbubble driven presumably by stellar winds and perhaps also supernova explosions . a large shell of atomic hydrogen gas and dust that lies just beyond the remnant molecular clouds , believed to be driven by just such a superbubble , may comprise the swept - up remains of the parental giant molecular cloud from which this association formed . in support of this picture , we find a week trend for the younger o star members to lie at higher galactic latitudes than the older supergiant members . the star - forming regions located at even larger heights above the galactic plane presumably correspond to more recent episodes of star formation at or near the periphery of this superbubble .
percolation@xcite is a leading paradigm for disorder . it provides an intuitively appealing and transparent model of the irregular geometry which occurs in disordered systems . moreover , it is a prototype of a phase transition . though percolation represents the simplest model of a disordered system , it has many applications , e.g. , polymerization , porous and amorphous materials , thin films , spreading of epidemics etc . in particular the transport properties of percolation clusters have gained a vast amount of interest over the last decades . random resistor networks ( rrn ) are a prominent model for transport on percolation clusters . by means of rrn one can study the conductivity of disordered media which might be important for technical applications . nonlinear random resistor networks , for which the voltage drop over an individual resistor depends on some power of the current flowing through it , can be exploited to derive various fractal dimensions of percolation clusters . from the conceptual point of view , rrn have the advantage that one can formulate a field theoretic hamiltonian amenable to renormalization group analysis . via rrn one can learn about diffusion on disordered substrates , since the diffusion constant @xmath8 and the conductivity @xmath9 of the system are related by the einstein relation @xmath10 where @xmath11 and @xmath12 denote the charge and the density of the mobile particles . the connection of the two problems is particularly important , since up to date no direct approach to diffusion on percolation clusters by means of a dynamic field theory exists . in this paper we study the distribution of currents in rrn . the current distribution has many interesting features , one of which is multifractality@xcite . this means that the distribution is not controlled by one or two relevant length scales , but rather by an infinite hierarchy of such length scales . the concept of multifractality was introduced for turbulence@xcite . it has been applied successfully in diverse areas including diffusion near fractals@xcite , electrons in disordered media@xcite , polymers in disordered media@xcite , random ferromagnets@xcite , chaotic dissipative systems@xcite , and heartbeat@xcite . due to the multifractality infinitely many exponents are needed to characterize the current distribution . consider two connected terminals @xmath0 and @xmath1 of the network . suppose a current @xmath13 is inserted at @xmath0 and withdrawn at @xmath1 . the @xmath2th moment of the current distribution given ( apart from technical details , c.f . [ momentsofthecurrentdistribution ] ) by , @xmath14 where the sum runs over all current carrying bonds ( the backbone ) , @xmath15 stands for the average over all diluted configurations and @xmath16 is an abbreviation for @xmath17 , scales at criticality as @xcite @xmath18 the @xmath19 constitute an infinite set of exponents which are not related to each other in a linear fashion , i.e. , the multifractal moments do not show the usual gap scaling commonly encountered in critical phenomena . each of the @xmath20 is associated with a particular subset of backbone bonds having its distinct fractal dimension . let @xmath21 be the number of bonds carrying current @xmath22 . upon applying the saddle point method on finds that the main contribution to the @xmath2th moment is given by @xmath23 with the multifractal spectrum @xmath24 and the multifractal exponents @xmath19 being related to each other by a legendre transformation . @xmath24 can be interpreted as the fractal dimension of the subset of bonds dominating @xmath20 . an elegant approach for studying the multifractal moments is to consider rrn with microscopic noise , i.e. , random networks in which the conductances of the individual resistors fluctuate about some mean . these noisy rrn were originally introduced by rammal _ et al_.@xcite to study the effects of flicker ( @xmath25 ) noise . flicker noise refers to the low frequency spectrum of excess voltage fluctuations measured when a constant current is applied to a resistor . the @xmath2th noise cummulant @xmath26 of the total resistance between the terminals @xmath0 and @xmath1 is proportional to @xmath27 by virtue of cohn s theorem@xcite . historically , the existence of the set of multifractal exponents @xmath28 was proposed by rammal _ et al_@xcite . the authors determined several of their exponents for two dimension by numerical simulations . a set of exponents @xmath29 equivalent to @xmath30 , where @xmath4 is the correlation length exponent for percolation , was also proposed by arcangelis _ et al._@xcite . theses authors derived their exponents for several hierarchical structures analytically . the field theoretic description of multifractality in rrn was pioneered by park , harris and lubensky ( phl)@xcite . based on an approach by stephen@xcite they formulated a @xmath31-fold replicated hamiltonian for noisy rrn . the contributions to the hamiltonian leading to multifractal behavior contain powers of replica space gradients analog to powers of real space gradients , which were accounted for as an origin of multifractality by duplantier and ludwig@xcite . phl introduced a set of exponents @xmath32 identical to the set @xmath33 and calculated it to first order in @xmath34 , where @xmath35 denotes the spatial dimension . later on fourcade and tremblay@xcite gave a reinterpretation of the work by phl . _ computed several multifractal exponents for @xmath36 by numerically solving kirchhoff s equations . recently barthlmy _ et al . _ performed simulations indicating that in the thermodynamic limit the @xmath20 do not exist for @xmath37 . in this article we study the moments of the current distribution by renormalized field theory . we extend our real - world interpretation of feynman diagrams@xcite to rrn with noise . upon introducing multifractal moments for feynman diagrams we reformulate the field theory of phl in a way that to our opinion is less complex and more intuitive . by carefully analyzing the relevance of the field theoretic operators related to the noise cumulants , we show that the multifractality is associated with dangerously irrelevant master operators@xcite . we calculate the set @xmath5 for @xmath38 to second order in @xmath39 . finally , we compare our result to numerical simulations . this section provides background on noisy rrn . it is guided by the work of stephen@xcite and phl@xcite . consider a @xmath35-dimensional lattice , where bonds between nearest neighboring sites @xmath22 and @xmath40 are randomly occupied with probability @xmath41 or empty with probability @xmath42 . each occupied bond @xmath43 has a conductance @xmath44 . unoccupied bonds have conductance zero . the bonds obey ohm s law @xmath45 where @xmath46 is the current flowing through the bond from @xmath40 to @xmath22 and @xmath47 is the potential at site @xmath22 . suppose a current @xmath13 is injected into a cluster at site @xmath0 and withdrawn at site @xmath1 . the union of all sites belonging to all self avoiding paths between @xmath0 and @xmath1 is refered to as the backbone between @xmath0 and @xmath1 . the power dissipated on the backbone is by definition @xmath48 using ohm s law , it may be expressed entirely in terms of voltages as @xmath49 here @xmath50 is the total resistance of the backbone , the sum is taken over all nearest neighbor pairs on the cluster and @xmath51 denotes the corresponding set of voltages . as a consequence of the variation principle @xmath52 = 0 \ , \end{aligned}\ ] ] one obtains kirchhoff s law @xmath53 where @xmath54 and the summations extend over the nearest neighbors of @xmath22 . alternatively to eq . ( [ noisypowerintermsofv ] ) the power can by rewritten in terms of the currents as @xmath55 with @xmath56 denoting the set of currents flowing through the individual bonds , @xmath57 , and @xmath58 . obviously the cluster may contain closed loops as subnetworks . suppose there are currents @xmath59 circulating independently around a complete set of independent closed loops . then the power is not only a function of @xmath13 but also of the set of loop currents . the potential drop around closed loops is zero . this gives rise to the variation principle @xmath60 eq . ( [ noisyvariationprinciple2 ] ) may be used to eliminate the loop currents and thus provides us with a method to determine the total resistance of the backbone via eq . ( [ noisypowerintermsofi ] ) . since the resistance of the backbone depends on the configurations @xmath61 of the randomly occupied bonds , one introduces an average @xmath15 over these configurations . it is important to recognize that the resistance between disconnected sites is infinite . therefore one considers only those sites @xmath0 and @xmath1 known to be on the same cluster . practically this is done by introducing the indicator function @xmath62 which , for a given configuration @xmath61 , is unity if @xmath0 and @xmath1 are connected and zero otherwise . then the @xmath2th moment of the resistance @xmath63 with respect to the average @xmath15 subject to @xmath0 and @xmath1 being on the same cluster is given by @xmath64 in the following we consider rrn with noise in the sense that the conductances @xmath65 of occupied bonds fluctuate about some mean . to be specific , the @xmath65 are equally and independently distributed random variables with mean @xmath66 and higher cumulants @xmath67 . the distribution function @xmath68 might for example be gaussian . nevertheless , our considerations are not limited to this particular choice . in order to suppress unphysical negative conductances , the assumption @xmath69 is made . in general the backbone resistance will depend on the set of conductances of occupied bonds @xmath70 . its noise average will be denoted by @xmath71 and the corresponding cumulants by @xmath72 \right\}_f \right\|_{\lambda = 0 } \ .\end{aligned}\ ] ] both kinds of disorder , the random dilution of the lattice and the fluctuation of the bond conductances about their mean @xmath66 , influence the statistical properties of the backbone resistance . they are reflected by the moments @xmath73 and the cumulants @xmath74 the noise cumulants @xmath75 characterize the distribution of currents flowing through the network . this section provides a relation between the @xmath75 and the moments of the current distribution . equation ( [ defofnoiseaverage ] ) defines the noise average as an average with respect to the distribution of the bond conductances @xmath65 . equally well one might express the backbone resistance in terms of the bond resistances and average over the distribution of the @xmath76 . since the @xmath65 are independently and equally distributed , the @xmath76 are distributed by the same means . assume that the distribution function of the deviations @xmath77 of the resistance of each bond from its average @xmath78 has the form @xmath79 and that @xmath80 @xmath81 is a variable with units of resistance which sets the scale of the distribution . with this form of @xmath82 , the @xmath12th cumulant @xmath83 of @xmath84 tends to zero as @xmath85 . this follows from the generating function @xmath86 of the @xmath83 : @xmath87 = \left\ { \exp \left ( \lambda \delta \rho_b \right ) \right\}_f = \int dy \ h \left ( y \right ) \exp \left ( \lambda s y \right ) = \exp \left ( \sum_{n=1}^\infty \frac{\lambda^n}{n ! } v_n\right ) \ , \end{aligned}\ ] ] where @xmath88 with @xmath89 being constants . in general @xmath90 depends on the entire set of cumulants @xmath91 . however , in the limit @xmath92 the leading term is proportional to @xmath93 as we will see immediately . consider the generating function @xmath94 of the cumulants @xmath95 , @xmath96 = \int \prod_b d \delta \rho_b \ g_s \left ( \delta \rho_b \right ) \exp \left [ \lambda r ( x , x^\prime ) \right ] \ .\end{aligned}\ ] ] expansion of the backbone resistance in a power series in the @xmath84 leads to @xmath97 = \int \prod_b d y_b \ h \left ( y_b \right ) \exp \left [ \lambda r^0 ( x , x^\prime ) + \lambda \sum_{k=1}^\infty \sum_{b_1 , \cdots , b_k } \frac{s^k}{k ! } \left . \frac{\partial^k r ( x , x^\prime ) } { \partial \rho_{b_1 } \cdots \partial \rho_{b_k } } \right\|_{\overline{\rho } } y_{b_1 } \ldots y_{b_k } \right ] \ , \end{aligned}\ ] ] where @xmath98 is the resistance when @xmath99 for every bond @xmath100 . equation ( [ schubidu ] ) can be rearranged as @xmath96 & = & \exp \left [ \lambda r^0 ( x , x^\prime ) + \lambda \sum_{k=2}^\infty \sum_{b_1 , \cdots , b_k } \frac{s^k}{k ! } \left . ( x , x^\prime ) } { \partial \rho_{b_1 } \cdots \partial \rho_{b_k } } \right\|_{\overline{\rho } } \frac{\partial^k } { \partial z_{b_1 } \cdots \partial z_{b_k } } \right ] \nonumber \\ & \times & \prod_b \exp \left [ c \left ( z_b \right ) \right ] \bigg\|_{\lambda s \sum_b \left . \frac{\partial r ( x , x^\prime ) } { \partial \rho_b } \right\|_{\overline{\rho } } } \nonumber \\ & = & \exp \left [ \lambda r^0 ( x , x^\prime ) + \sum_{l=1}^\infty \left ( \lambda s \right)^l c_l \sum_b \left ( \left . \frac{\partial r ( x , x^\prime ) } { \partial \rho_b } \right\|_{\overline{\rho } } \right)^l + \sum_{i=2}^\infty f_i \left ( \lambda s^i \right ) \right ] \ , \end{aligned}\ ] ] where @xmath101 are functions of @xmath102 . hence for @xmath103 , @xmath104 in the limit @xmath92 the leading term is @xmath105 where we have used cohn s theorem eq . ( [ originalcohn ] ) . upon substitution of eq . ( [ gleichfa ] ) into eq . ( [ defcumulant ] ) one finds for the noise cumulants @xmath106 i.e. , the noise cumulant @xmath75 is proportional to the @xmath2th multifractal moment @xmath107 of the current distribution . our aim is to determine @xmath75 . hence the task is to solve the set of kirchhoff s equations ( [ noisycirquitequations ] ) and to perform the averages over the diluted lattice configurations and the noise . it can be achieved by employing the replica technique@xcite . in order to treat the averages @xmath108 and @xmath109 separately , phl introduced @xmath110-fold replicated voltages , @xmath111 note from the definitions eq . ( [ defcumulant ] ) and eq . ( [ defofnoisecumulant ] ) that one has to treat the two averages independently in the calculation of @xmath112 . in contrast , for calculating @xmath113 it is not necessary to distinguish between the two averages because one could also introduce a composite distribution function @xmath114 and a single , say @xmath8-fold , replication would be sufficient . to construct a generating function for the noise cumulants one introduces @xmath115 where @xmath116 and @xmath117 . the corresponding correlation functions @xmath118 are defined as @xmath119 \bigg\}_f \bigg\rangle_c \ .\end{aligned}\ ] ] here @xmath120 , @xmath121 with @xmath122 , and @xmath123 is the normalization @xmath124 \ .\end{aligned}\ ] ] note that we have introduced an additional power term @xmath125 . this is necessary to give the integrals in eqs . ( [ noisyerzeugendefunktion ] ) and ( [ noisynorm ] ) a well defined meaning . without this term the integrands depend only on voltage differences and the integrals are divergent . physically the new term corresponds to grounding each lattice site by a capacitor of unit capacity . the original situation may be restored by taking the limit of vanishing frequency , @xmath126 . the integrations in eq . ( [ noisyerzeugendefunktion ] ) can be carried out by employing the saddle point method . since the integrations are gaussian the saddle point method is exact in this case . the saddle point equation is identical to the variation principle stated in eq . ( [ noisyvariationprinciple1 ] ) . thus the maximum of the integrand is determined by the solution of kirchhoff s equations ( [ noisycirquitequations ] ) and @xmath127 \right\}_f \right\rangle_c \ .\end{aligned}\ ] ] the right hand side of eq . ( [ noisygenfkt ] ) may be expanded in term of the cumulants defined in eq . ( [ defofnoisecumulant ] ) . this gives @xmath128 \right\rangle_c \ , \end{aligned}\ ] ] where @xmath129 is defined by @xmath130^l \ .\end{aligned}\ ] ] we learn that the correlation function @xmath131 can be exploited as a generating function for the noise cumulants via @xmath132 note that @xmath133 . since infinite voltage drops between different clusters may occur , it is not guaranteed that @xmath123 stays finite , i.e. , the limit @xmath134 is not well defined . moreover , @xmath135 has to be excluded properly . both problems can be handled by resorting to a lattice regularization of the integrals in eqs . ( [ noisyerzeugendefunktion ] ) and ( [ noisynorm ] ) . one switches to voltage variables @xmath136 taking discrete values on a @xmath31-dimensional torus , i.e. , @xmath137 is chosen to be an @xmath31-dimensional integer with @xmath138 and @xmath139 . @xmath140 is the gap between successive voltages and @xmath141 is the voltage cutoff . the continuum may be restored by taking @xmath142 and @xmath143 . by setting @xmath144 , @xmath145 , and , respectively , @xmath146 , the two limits can be taken simultaneously via @xmath147 . since the voltage and current variables are conjugated @xmath148 is affected by the discretization as well : @xmath149 where @xmath150 is a @xmath151-dimensional integer taking the same values as @xmath152 . this choice guarantees that the completeness and orthogonality relations [ noisycomplete ] @xmath153 and @xmath154 do hold . equation ( [ noisycomplete ] ) provides us with a fourier transform in replica space . in this discrete picture there are @xmath155 independent state variables per lattice site . upon fourier transformation one introduces the potts spins@xcite @xmath156 subject to the condition @xmath157 . now we revisit eq . ( [ noisyerzeugendefunktion ] ) . carrying out the average over the diluted lattice configurations and the noise provides us with the weight @xmath158 of the average @xmath159 , @xmath160 \right\}_f \right\rangle_c \nonumber \\ & = & - \sum_{\langle i , j \rangle } \ln \left\langle \prod_{\beta = 1}^e \left\ { \exp \left [ - \frac{1}{2 } \sigma_{i , j}^{(\beta ) } \left ( \vec{\theta}_{i}^{(\beta ) } - \vec{\theta}_{j}^{(\beta ) } \right)^2 \right ] \right\}_f \right\rangle_c - \frac{i\omega}{2 } \sum_i \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta}_{i}^2 \ .\end{aligned}\ ] ] by dropping a constant term @xmath161 , with @xmath162 being the number of bonds in the undiluted lattice , one obtains @xmath163 where @xmath164 and @xmath165 \right\}_f \right\ } \nonumber \\ & = & \ln \left\ { 1 + \frac{p}{1-p } \exp \left [ \sum_{l=1}^\infty \frac{(-1/2)^l}{l ! } \delta^{(l ) } k_l \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta } \right ) \right ] \right\ } \ .\end{aligned}\ ] ] in the limit of perfect transport , @xmath166 , @xmath167 goes to its local limit @xmath168 , with @xmath169 being a positive constant . the interaction part of the hamiltonian reduces to @xmath170 this represents nothing more than the @xmath171 states potts model which is invariant against all @xmath172 permutations of the potts spins @xmath173 . in the case of imperfect transport this @xmath174 symmetry is lost . for finite @xmath66 and @xmath175 , @xmath167 is an exponentially decreasing function in replica space with a decay rate proportional to @xmath176 . then , for large @xmath66 , the hamiltonian @xmath177 describes a translationally and rotationally invariant short range interaction of potts spins in real and replica space with an external one site potential @xmath178 . admitting fluctuations of the resistances , @xmath179 , results in breaking the rotational @xmath180 replica space symmetry of the interaction part of the hamiltonian . the fourier transform of @xmath181 , @xmath182 is expediently evaluated by switching back to continuous voltages , @xmath183 \right\ } \ , \end{aligned}\ ] ] where we have dropped a factor @xmath184 . taylor expansion of the logarithm yields a series of terms of the form @xmath185 \ , \end{aligned}\ ] ] where @xmath186 the @xmath187 are constants of order @xmath188 . in addition to the expansion of the logarithm we expand in a power series in @xmath81 , @xmath189 \left\ { 1 + \sum_{l=2}^\infty \left ( \overline{\sigma } s \right)^l p_l \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta } \right ) \right\ } \ . \end{aligned}\ ] ] here the @xmath190 are homogeneous polynomials of order @xmath191 in @xmath192 which are a sums of terms proportinal to @xmath193 such that @xmath194 . completing squares in the exponential in eq . ( [ s2 ] ) gives @xmath195 \int_{-\infty}^\infty d \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta } \ \exp \left [ - a \overline{\sigma } \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta}^2 \right ] \left\ { 1 + \sum_{l=2}^\infty \left ( \overline{\sigma } s \right)^l p_l \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\theta } - i \frac{\stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda}}{2 a \overline{\sigma } } \right ) \right\ } \nonumber \\ & = & \exp \left [ - \frac{\stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda}^2}{4 a \overline{\sigma } } \right ] \left\ { 1 + \sum_{l=2}^\infty \left ( \overline{\sigma } s \right)^l \bigg [ p_l \left ( \frac{\stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda } } { \overline{\sigma } } \right ) + \cdots + \overline{\sigma}^{(-r ) } p_{l - r } \left ( \frac{\stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda } } { \overline{\sigma } } \right ) + \cdots \bigg ] \right\ } \ , \end{aligned}\ ] ] where we have omitted multiplicative factors decorating the @xmath190 . due to the homogenity of the @xmath190 , eq . ( [ s3 ] ) can be rearranged as @xmath196 \left\ { 1 + \sum_{l=2}^\infty s^l \bigg [ \overline{\sigma}^{-l } p_l \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda } \right ) + \cdots + \overline{\sigma}^{-(l - r)}p_{l - r } \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda } \right ) + \cdots \bigg ] \right\ } \nonumber \\ & = & \exp \left [ - \frac{\stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda}^2}{4 a \overline{\sigma } } \right ] \left\ { 1 + \sum_{l^\prime = 1}^\infty \left ( \frac{s}{\overline{\sigma } } \right)^{l^\prime } \bigg [ 1 + { \sl o } \left ( s \right ) \bigg ] p_{l^\prime } \left ( \stackrel{\mbox{\tiny $ \leftrightarrow$}}{\lambda } \right ) \right\ } \ , \end{aligned}\ ] ] up to multiplicative factors . by keeping only the leading contributions , one finds that @xmath197 can be expanded as @xmath198 with @xmath199 , @xmath200 , and @xmath201 being expansion coefficients . it is known that the terms @xmath202 are irrelevant in the renormalization group sense for @xmath203 ( see , e.g.@xcite ) . from sec . [ relevanceofthenoiseterms ] can be inferred that the @xmath204 are irrelevant as well . however , the terms proportional to @xmath205 are indispenseble in studying the noise cumulants ; they are dangerously irrelevant . therefore , we restrict the expansion of @xmath206 to @xmath207 with @xmath208 , and @xmath209 . nevertheless the neglected terms will regain some importance later on since they are required for the renormalization of the @xmath93 . the @xmath129 are homogeneous polynomials of order @xmath191 . for @xmath103 they are possessing a @xmath210 $ ] symmetry . thus , allowing for @xmath211 results in loosing the rotational @xmath212 in favor of the @xmath213 $ ] symmetry . it is worth pointing out that @xmath214 , i.e. , the condition @xmath166 translates into @xmath215 . consequently one has to take the limit @xmath216 before the limit @xmath217 in calculating the exponents associated with the @xmath93 . we proceed with the usual coarse graining step and replace the potts spins @xmath218 by order parameter fields @xmath219 which inherit the constraint @xmath220 . we model the corresponding field theoretic hamiltonian @xmath221 in the spirit of landau as a mesoscopic free energy from local monomials of the order parameter field and its gradients in real and replica space . the gradient expansion is justified since the interaction is short ranged in both spaces . purely local terms in replica space have to respect the full @xmath174 potts symmetry . after these remarks we write down the landau - ginzburg - wilson type hamiltonian @xmath222 where @xmath223^{l } \ .\end{aligned}\ ] ] in eq . ( [ noisyfinalhamil ] ) we have neglected terms of order @xmath224 or higher which are irrelevant in the renormalization group sense . @xmath199 , @xmath225 and @xmath93 are now coarse grained analogues of the original coefficients appearing in eq . ( [ noisytaylorexp ] ) . note again that @xmath221 reduces to the usual @xmath226 states potts model hamiltonian by setting @xmath227 and @xmath228 as one retrieves purely geometrical percolation in the limit of vanishing @xmath93 and @xmath225 . irrelevant variables that can not be taken to zero because the quantity one is looking at either vanishes or diverges in this limit have been given the name _ dangerously irrelevant variables _ by fisher@xcite . later on this notion was introduced into field theory by amit and peliti@xcite . a characteristic feature of dangerously irrelevant variables is that corrections due to them determine the asymptotic behavior of quantities with the above property , so that their effect is felt arbitrarily close to the transition@xcite . in this section we show that the @xmath93 are dangerously irrelevant . they are irrelevant on dimensional grounds , i.e. , they are associated with a negative naive dimension . however , we can not simply take the @xmath93 to zero by appealing to their irrelevance , because the amplitudes of the noise cumulants vanish in this limit . in the remainder of this article we focus on vanishing frequency , @xmath229 . let @xmath230 denote the set of parameters @xmath231 . we introduce a scaling factor @xmath100 for the voltage variable : @xmath232 . by substitution of @xmath233 the hamiltonian turns into @xmath234 = \int d^dx \sum_{\stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } } \left\ { \frac{1}{2 } \varphi^\prime \left ( { \rm{\bf x } } , b \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } \right ) k \left ( \delta , \nabla_{\stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } } \right ) \varphi^\prime \left ( { \rm{\bf x } } , b \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } \right ) + \frac{g}{6}\varphi^{\prime } \left ( { \rm{\bf x } } , b \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } \right)^3 \right\ } \ .\end{aligned}\ ] ] renaming the scaled voltage variables @xmath235 yields @xmath236 = \int d^dx \sum_{\stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta}^\prime } \left\ { \frac{1}{2 } \varphi^\prime \left ( { \rm{\bf x } } , \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta}^\prime \right ) k \left ( \delta , b \nabla_{\stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta}^\prime } \right ) \varphi^\prime \left ( { \rm{\bf x } } , \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta}^\prime \right ) + \frac{g}{6}\varphi^{\prime } \left ( { \rm{\bf x } } , \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta}^\prime \right)^3 \right\ } \ .\end{aligned}\ ] ] obviously the voltage cutoff is affected by the scaling as well : @xmath237 . however , if the limits are taken in the appropriate order , namely @xmath238 and then @xmath147 , the dependence of the theory on the cutoff drops out . thus , we can identify @xmath239 and @xmath240 and hence @xmath241 = { \mathcal{h } } \left [ \varphi \left ( { \rm{\bf x } } , \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\theta } \right ) , p^\prime \right ] \ , \end{aligned}\ ] ] where @xmath242 . now consider correlation functions @xmath243 \right ) \ , \end{aligned}\ ] ] where @xmath244 indicates an integration over the set of variables @xmath245 for all @xmath246 and @xmath240 . equation ( [ noisyrelforh ] ) implies @xmath247 the two - point correlation function @xmath248 is the fourier transform of @xmath249 . we deduce from eq . ( [ cumulantgenfkt ] ) that @xmath250 we are free to choose @xmath251 . this gives @xmath252 where @xmath253 is a scaling function . we learn from eq . ( [ cumulantscaling ] ) that the coupling constants @xmath254 appear only as @xmath255 . dimensional analysis of the hamiltonian shows that @xmath256 and @xmath257 , where @xmath258 is an inverse length scale , i.e. , @xmath259 and @xmath260 have a naive dimension 2 . thus @xmath261 and hence the @xmath255 have a negative naive dimension . this leads to the conclusion that the @xmath262 are irrelevant couplings . though irrelevant , one must not set @xmath263 in calculating the noise exponents . in order to see this we expand the scaling function @xmath253 in eq . ( [ cumulantscaling ] ) , @xmath264 with @xmath265 being expansion coefficients depending on @xmath266 , @xmath267 , and @xmath199 . it is important to recognize that @xmath268 because the corresponding terms are not generated in the perturbation calculation . equation ( [ expofcumulantscaling1 ] ) can be rewritten as @xmath269 where the first term on the right hand side gives the leading behavior . thus @xmath75 vanishes upon setting @xmath270 and we can not gain any further information about @xmath75 . in particular we can not determine the associated noise exponent . in other words , the @xmath271 are dangerously irrelevant in investigating the critical properties of the @xmath272 . the diagrammatic elements contributing to our renormalization group improved perturbation calculation are the three point vertex @xmath273 and the propagator @xmath274 note that we have switched to a @xmath275-representation by employing fourier transformation in real and replica space . the notation in eq . ( [ propagatordecomp ] ) is somewhat symbolic . to treat the irrelevant terms @xmath276 properly , we have to expand the propagator in a power series in the @xmath93 and discard all contributions of higher than linear order in the @xmath93 . in other words : the irrelevant terms have to be treated as insertions . ( [ propagatordecomp ] ) shows that the principal propagator decomposes into a propagator carrying @xmath148 s ( conducting ) and one not carrying @xmath277 s ( insulating ) . this allows for a schematic decomposition of principal diagrams into sums of conducting diagrams consisting of conducting and insulating propagators . to two - loop order , we obtain the conducting diagrams listed in fig . 1 . from the decomposition in sec . [ noisydiagrammaticexpansion ] a real - world interpretation of the conducting feynman diagrams emerges@xcite . they may be viewed as resistor networks themselves with conducting propagators corresponding to conductors and insulating propagators corresponding to open bonds . the parameters @xmath81 appearing in a schwinger parametrization of the conducting propagators , @xmath278 \ , \end{aligned}\ ] ] correspond to resistances and the replica variables @xmath279 to currents . the replica currents are conserved in each vertex and we may write for each edge @xmath22 of a diagram , @xmath280 , where @xmath148 is an external current and @xmath281 denotes a complete set of independent loop currents . the real - world interpretation suggests an effective way of computing the conducting diagrams . we learn from the discussion above , that the irrelevant terms have to be treated by means of insertions @xmath282 where @xmath283 denotes the fourier transform of @xmath284 . the resulting diagrams are of the type displayed on the left hand side of fig . 2 . we express the current dependend part of such a diagram in terms of its power @xmath230 , @xmath285 = - s_i v_l \sum_{\big\ { \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } \big\ } } k_l \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}_i \right ) \exp \left [ w p \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } , \left\ { \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } \right\ } \right ) \right ] \ .\end{aligned}\ ] ] the summation is carried out by completing the squares in the exponential . the corresponding shift in the loop currents is given by the minimum of the quadratic form @xmath230 which is determined by a variation principle completely analogous to the one stated in eq . ( [ noisyvariationprinciple2 ] ) . thus , completing of the squares is equivalent to solving kirchhoff s equations for the diagram . it leads to @xmath286 \ .\end{aligned}\ ] ] @xmath287 is the current induced by the external current into edge @xmath22 . @xmath288 and @xmath289 are homogeneous functions of the schwinger parameters of degree zero . @xmath290 and the total resistance of the diagram @xmath291 are homogeneous functions of the schwinger parameters of degree one . by a suitable choice of the @xmath292 the matrix constituted by the @xmath293 is rendered diagonal , i.e. , @xmath294 . at this stage it is convenient to switch to continuous currents and to replace the summation by an integration , @xmath295 where @xmath296 is an abbreviation for @xmath297 and @xmath298 stands for the number of independent conducting loops . this integration is gaussian and therefore straightforward . in the limit @xmath238 one obtains @xmath299 the terms neglected in eq . ( [ huhu ] ) are not required in calculating the @xmath300 . this issue is discussed in detail in sec . [ noisyrenormalization ] . diagrammatically , the calculation scheme can be condensed into fig . 2 . appendix [ app : calcschemeex ] illustrates the calculations in terms of an example . so far we have inserted @xmath301 only in one of the conducting propagators . however , each of them has to get an insertion . moreover , the integrations over loop momenta and schwinger parameters remain to be carried out . all in all , each diagram can be written as @xmath302 d \left ( { \rm{\bf p}}^2 , \left\ { s_i \right\ } \right ) \ .\end{aligned}\ ] ] here @xmath303 stands for the integrand one obtains upon schwinger parametrization of the corresponding diagram in the usual @xmath304 theory . @xmath305 is defined as @xmath306 where the sum runs over all conducting propagators of the diagram . notice the analogy of the @xmath305 to the generalized multifractal moments we introduce in app . [ app : generalization ] . thus , we refer to the @xmath307 as multifractal moments of conducting feynman diagrams . by employing dimensional regularization and minimal subtraction we proceed with standard techniques of renormalized field theory@xcite . the renormalization of the @xmath93 , however , involves some peculiarities that we will discuss in this section . an operator @xmath308 of a given naive dimension @xmath309 $ ] inserted one time in a vertex function generates in general new primitive divergencies corresponding to all operators of equal or lower naive dimension . thus , one needs these newly generated operators as counterterms in the hamiltonian . the operators of lower naive dimension can be isolated by additive renormalization , @xmath310 < \left [ { \mathcal{o}}_i \right ] } x_{i , j } { \mathcal{o}}_j \ .\end{aligned}\ ] ] dimensional regularization in conjunction with minimal subtraction leads to @xmath311 containing at least a factor @xmath199 . these @xmath311 vanish at the critical point . hence , the operators of lower naive dimension will not be considered in the following . as argued in sec . [ multifractalmomentsoffeynmandiagrams ] the term proportional to @xmath93 in eq . ( [ noisyexpansionofdiagrams ] ) is generated by inserting the operator @xmath301 . inserting @xmath301 into a diagram with @xmath12 external legs , see fig . 3 , generates primitive divergencies which must be cancelled by counter terms of the structure @xmath312 where @xmath313 with @xmath314 , is a homogeneous polynomial of degree @xmath315 . note , that the notation we use here and in the following is symbolic , since such a counter term has to depend on the entire set of external momenta and currents . @xmath316 is for example an abbreviation for @xmath317 . the leading contribution comes from operators having the same naive dimension as @xmath301 , i.e. , those satisfying @xmath318 here we expressed the naive dimension with help of eq . ( [ naivedimofinsertion ] ) . for @xmath319 one is led to @xmath320 , i.e. , the insertion of @xmath301 generates operators containing homogeneous polynomials in the replica currents of degree equal or lower @xmath191 . in particular @xmath301 generates an operator of type @xmath321 the important question now is , if the other operators generated by @xmath301 generate operators of this type , too . consider @xmath322 . with help of eq . ( [ blabla ] ) one obtains @xmath323 , where the second inequality is a consequence of the limit @xmath324 . bearing in mind that maximal homogeneous polynomials of degree @xmath325 in @xmath148 are generated , we reinsert these operators of the type in eq . ( [ generaloperator ] ) with @xmath322 into two - leg diagrams , see fig . the resulting terms are of the form @xmath326 with the leading contributions satisfying @xmath327 . thus , @xmath328 , i.e. , the homogeneous polynomials in @xmath329 may have a higher degree than @xmath191 . however , they are of the type @xmath330 with @xmath331 and @xmath332 . these polynomials have a higher symmetry than the original @xmath129 . we conclude that @xmath301 generates itself and an entire family of new operators but these in turn do not generate @xmath301 . in principle , the entire family of operators associated with @xmath301 has to be taken into account in the renormalization proceedure , leading to a renormalization in matrix form @xmath333 the vector @xmath334 contains the family associated with @xmath301 . for the remaining renormalizations , we employ the same scheme as in @xcite , [ renormalizationscheme ] @xmath335 in eq . ( [ renormalizationscheme ] ) @xmath39 stands for @xmath336 and the factor @xmath337 , with @xmath338 denoting the gamma function , is introduced for convenience . according to the arguments given above the renormalization matrix @xmath339 has a particularly simple structure , @xmath340 @xmath341 stands for the unit matrix and @xmath342 symbolizes elements that we do not evaluate . in this paper , we determine @xmath343 to the order of two loops . @xmath123 , @xmath344 and @xmath345 are the usual potts model @xmath123 factors . they have been computed to three - loop order by de alcantara bonfim _ et al_@xcite . @xmath346 is known to two - loop order@xcite . the unrenormalized theory has to be independent of the length scale @xmath347 introduced by renormalization . in particular , the connected @xmath348 point correlation functions with an insertion of @xmath349 must be independent of @xmath258 , i.e. , @xmath350 for all @xmath348 . ( [ noisyindependence ] ) translates via the wilson functions @xmath351 where the bare quantities are kept fix while taking the derivatives , into the gell - mann - low renormalization group equation @xmath352 \underline{\underline{1 } } + \underline{\underline{\gamma}}^{(l ) } \right\ } } \nonumber \\ & & \times \ , g_n \left ( \left\ { { \rm{\bf x } } , w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \right\ } ; \tau , u , \mu \right)_{\underline{\hat{\mathcal{o}}}^{(l ) } } = 0 \ .\end{aligned}\ ] ] the particular form of the wilson functions can be extracted from the renormalization scheme and the @xmath123 factors . at the infrared stable fixed point @xmath353 , determined by @xmath354 , the renormalization group equation reduces to @xmath355 \underline{\underline{1 } } + \underline{\underline{\gamma}}^{(l)\ast } \right\ } } \nonumber \\ & & \times \ , g_n \left ( \left\ { { \rm{\bf x } } , w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \right\ } ; \tau , u^\ast , \mu \right)_{\underline{\hat{\mathcal{o}}}^{(l ) } } = 0 \ , \end{aligned}\ ] ] where @xmath356 , @xmath357 , @xmath358 , and @xmath359 . the matrix @xmath360 inherits the simple structure of @xmath361 , @xmath362 owing to this structure , @xmath363 is a right eigenvector of @xmath364 with eigenvalue @xmath365 . the remaining right eigenvectors with eigenvalues @xmath366 , @xmath367 , we denote by @xmath368 . the left eigenvectors of @xmath364 are @xmath369 and @xmath370 . in terms of the eigenvectors , @xmath364 can be spectrally decomposed into @xmath371 now it is important to realize , that @xmath372 i.e. , only @xmath373 contains the operator @xmath301 we are interested in . we substitute eq . ( [ spectraldeco ] ) into the renormalization group equation ( [ fixnoisyrgg ] ) and act on the entire equation with @xmath374 . the result is @xmath375 g_n \left ( \left\ { { \rm{\bf x } } , w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \right\ } ; \tau , u^\ast , \mu \right)_{{\mathcal{a}}^{(l ) } } = 0 \ .\end{aligned}\ ] ] equation ( [ noisyrggsimpler ] ) is solved by the method of characteristics . the solution reads @xmath376 to derive a scaling relation for the correlation functions , a dimensional analysis remains to be performed . it yields @xmath377 from eqs . ( [ noisysolofrgg ] ) and ( [ noisydimana ] ) we deduce the scaling behavior @xmath378 @xmath379 and @xmath380 are the well known critical exponents for percolation . they are known to third order in @xmath39@xcite : @xmath381\epsilon^3 + { \sl o } \left ( \epsilon^4 \right ) \ , \end{aligned}\ ] ] and @xmath382\epsilon^3 + { \sl o } \left ( \epsilon^4 \right ) \ .\end{aligned}\ ] ] note that @xmath383 in eqs . ( [ exponenteta ] ) and ( [ exponentnu ] ) stands for the riemann zeta function and should not be confused with the wilson function defined above . @xmath384 is the resistance exponent known to second order in @xmath39@xcite , @xmath385 the noise exponents @xmath300 are defined by @xmath386 . the expansion of @xmath300 to second order in @xmath39 is given below . now we are in the position to derive the scaling behavior of @xmath75 . from eq . ( [ noisyscalingrel ] ) we find upon choosing @xmath387 and taylor expanding that the two point correlation function @xmath388 scales at criticality as @xmath389 where we have dropped several arguments for notational simplicity . with eq . ( [ exploitgenfkt ] ) the desired scaling behavior of @xmath75 is now readily obtained as @xmath390 at this point , we emphasize once more the outstanding role of the @xmath301 , which warrants calling them master operators@xcite . each multifractal moment @xmath391 has a master operator as field theoretic counterpart . the master operators are highly and dangerously irrelevant in the renormalization group sense . therefore , each master operator needs in general a myriad of other irrelevant operators for renormalization . however , the renormalization of these servant operators does not induce their master . it follows , that the servant operators can be neglected in determining the scaling index of their master operator , i.e. , one is spared the computation and diagonalization of giant renormalization matrices . our @xmath39-expansion result for the noise exponents reads @xmath392 \big\ } \bigg\ } \nonumber \\ & - & 672\left ( 1 + l \right)^2 \left ( 1 + 2l \right)^2 \psi(1 + 2l ) \bigg\ } + { \em o } \left ( \epsilon^3 \right)\end{aligned}\ ] ] in agreement to first order in @xmath39 with the one - loop calculation by phl . @xmath393 denotes euler s constant and @xmath394 stands for the digamma function@xcite . equation ( [ monsterexponent ] ) is valid not only for @xmath395 since it can be continued analytically down to @xmath396 . a plot of @xmath300 versus @xmath39 is given in fig . we point out that eq . ( [ monsterexponent ] ) evaluated at @xmath397 is in conformity with the result for @xmath398 stated in eq . ( [ exponentphi ] ) , i.e. , our result for @xmath300 satisfies an important consistency check steaming from @xmath399 . et al._@xcite proved that @xmath300 is a convex monotonically decreasing function of @xmath2 . note from fig . 5 that our result for @xmath300 captures this feature for reasonable values of @xmath39 . it reduces to unity in the limit @xmath400 as one expects from the relation of @xmath401 to the fractal dimension of the red bonds ( see app . [ app : relationtothebackbonedimension ] ) . moreover , analytic continuation of @xmath300 to @xmath396 shows that @xmath402 up to order @xmath403 as expected ( see app . [ app : relationtothebackbonedimension ] ) . in this section we compare our result for @xmath300 to numerical values . instead of working with @xmath300 directly we compare @xmath404 because data for exponents of this type is available in the literature . for the comparison it is not sufficient to simply evaluate eq . ( [ monsterexponent ] ) at @xmath405 or @xmath406 . the pure @xmath39-expansion gives for small spatial dimension poor quantitative predictions . however , one can improve the @xmath39-expansion by incorporating rigorously known features . we carry out a rational approximation which takes into account that @xmath407 in one dimension . practically this is done by adding an appropriate third order term to the @xmath39-expansion of @xmath404 . we refrain from stating the so obtained formula explictly because it is a little lengthy . instead we plot it for @xmath408 , i.e. @xmath36 , in fig . our analytic result shows remarkable agreement with the the available numerical data for @xmath36 . for @xmath396 our result lays slightly outside the error bars of the simulations . however , the deviation of the values is less than @xmath409 . for @xmath410 our result is within the error bars of the simulation . there are also numerical values available for @xmath411@xcite . here , however , the agreement is much less pronounced . for @xmath412 we find a deviation of the order of @xmath413 . it appears that the dependence of @xmath300 on dimensionality is too rich in structure to be approximated well at @xmath411 by a series of a few terms . we studied the multifractal moments of the current distribution in rrn by renormalized field theory . our approach thrived on two cornerstones . first , the feyman digrams for rrn can be interpreted as being resistor networks themselves . in this paper we extended our real - world interpretation by introducing multifractal moments for feynman diagrams . the real - world interpretation proves to be a powerful tool which allows a highly efficient calculation of the diagrams . the second cornerstone was our concept of master operators . whereas the field theoretic operator associated with the resistance exponent @xmath398 is relevant in the renormalization group sense , the operators associated with the @xmath414 are dangerously irrelevant master operators . due to their irrelevance the master operators generate a multitude of other irrelevant operators , the servants , which in principle must all be taken into account in the renormalization procedure . the servants however , do not influence the scaling index of their master . without this property , one would have to compute and diagonalize entire renormalization matrices for determining the @xmath415 . these renormaliztion matrizes are giants for large @xmath2 . without the master property it would be practially impossible to compute the @xmath415 for arbitrary @xmath2 . to our knowledge this is the first time that an entire family of multifractal exponents has been calculated to two - loop order , at least for percolation . our result is for dimensions near the upper critical dimension 6 the most accurate analytic estimate for the @xmath415 that we know of . it fulfils several consistency checks . moreover , it agrees remarkably well with numerical data for @xmath36 . as one expects , the agreement suffers by further decreasing the dimension . the dependence of the @xmath300 on dimensionality appears to be too complex for being approximated well at @xmath411 by a series of a few terms . we expect that our concept of master operators can be applied to other systems showing multifractality . it works for example to describe the moments of the current distribution in random resistor diode networks . a two - loop calculation of the corresponding family of mutifractal exponents will be reported in the near future@xcite . another example for the applicability of the concept of master operators is the problem of diffusion near polymers . for this problem von ferber and holovatch@xcite formulated a field theory which comprises dangerously irrelevant operators . due to the symmetry properties of their operators no other irrelevant operators are generated in the perturbation calculations of these authors . thus , the operators studied by von ferber and holovatch are particular simple instances of master operators . their scaling index is not influenced by any other operator simply because the number of their servants is zero . in this sense these master operators may be called poor . it might turn out that the field theoretic operators associated with multifractal quantities are in general master operators . in this case the concept of master operators would be a key in understanding the origin of multifractality , at least from a field theoretic point of view . we acknowledge support by the sonderforschungsbereich 237 `` unordnung und groe fluktuationen '' of the deutsche forschungsgemeinschaft . consider a generalized power of the form @xmath416 where @xmath417 is some function of bond currents @xmath418 . as argued in sec . [ noisymodel ] , @xmath418 is in general a function of the external current @xmath13 and a complete set of loop currents @xmath419 . we can exploit the variation principle ( [ noisyvariationprinciple2 ] ) to eliminate the loop currents . as a result we obtain the @xmath418 as a function of @xmath13 and @xmath420 only . the solutions we denote by @xmath421 . the derivative of the so obtained power @xmath422 with respect to a bond resistance @xmath76 reads @xmath423 the second term on the right hand side vanishes by virtue of the variation principle ( [ noisyvariationprinciple2 ] ) , @xmath424 renaming @xmath425 we finally obtain @xmath426 as a generalization of cohn s theorem . for @xmath427 one retrieves the original theorem @xmath428 having generalized the power it is natural to generalize the multifractal moments as well . consider the cumulants of the generalized power , @xmath429 in analogy to the resistance cumulants one finds for the leading behavior in the limit @xmath430 , @xmath431 for networks like those described in sec . [ noisymodel ] one has @xmath432 . however , in a more general situation the individual bonds may be composed of a series of elementary resistors . the elementary resistors are assumed to have independently and identically distributed resistances with mean @xmath78 and higher cumulants @xmath433 . then @xmath434 where @xmath435 denotes the number of elementary resistors constituting bond @xmath100 and @xmath436 is the average resistance of that bond . upon incorporating a factor @xmath437 into the constants @xmath83 we finally find @xmath438 with @xmath439 being the @xmath12th multifractal moment of @xmath440 . in this appendix we illustrate the calculation scheme sketched in sec . [ multifractalmomentsoffeynmandiagrams ] at the instance of diagram a. for the sake of simplicity , we focus on its contribution to the renormalization of @xmath441 . we neglect all other parts and obtain @xmath442 \nonumber \\ & \times & \left\ { s_1 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } \right ) + s_2 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } + \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) \right\ } \ , \end{aligned}\ ] ] where @xmath443 is , according to the real - world interpretation , the power of the diagram and @xmath444 stands for @xmath445 \ , \end{aligned}\ ] ] with @xmath446 being an abbreviation for @xmath447 . it is convenient to switch back to continuous currents and replace the summation over the loop current by an integration , @xmath448 \nonumber \\ & \times & \bigg\ { s_1 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } \right ) + s_2 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } + \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) \bigg\ } \ .\end{aligned}\ ] ] the integration over the loop current is simplified by completing the squares in the exponential . one looks for the minimum of the quadratic form @xmath449 . the minimum is determined by a variation principle completely analogous to the one stated in eq . ( [ noisyvariationprinciple2 ] ) . thus completing the squares is equivalent to solving kirchhoff s equations for the diagram . we obtain @xmath450 \int_{-\infty}^\infty d \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } \exp \left [ - \left ( s_1 + s_2 \right ) w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa}^2 \right ] \nonumber \\ & \times & \bigg\ { s_1 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } - \frac{s_2}{s_1 + s_2 } \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) + s_2 v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\kappa } + \frac{s_1}{s_1 + s_2 } \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) \bigg\ } \ , \end{aligned}\ ] ] where @xmath451 is the total resistance of the diagram . note that @xmath452 is , appart from a factor @xmath22 , the replica current induced by the external replica current @xmath329 into the propagator parametrized by @xmath453 . @xmath454 is the replica current induced into the propagator parametrized by @xmath455 . in the limit @xmath238 we find @xmath456 \nonumber \\ & \times & \bigg\ { \frac{s_1 s_2 ^ 4}{\left ( s_1 + s_2 \right)^4 } \ , v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) + 2 \frac{s_1 s_2 ^ 2}{\left ( s_1 + s_2 \right)^3 } \ , \frac{v_2}{w } \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \bigg\ } \ , \end{aligned}\ ] ] where we have carried out the momentum integration as well . expanding the exponential and keeping only the terms proportional to @xmath441 gives @xmath457 \nonumber \\ & \times & \bigg\ { \frac{s_1 s_2 ^ 4}{\left ( s_1 + s_2 \right)^4 } \ , v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) + 2 \frac{s_1 s_2 ^ 2}{\left ( s_1 + s_2 \right)^3 } \ , \frac{v_2}{w } \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \nonumber \\ & - & 2 \frac{s_1 ^ 2 s_2 ^ 3}{\left ( s_1 + s_2 \right)^4 } \ , \frac{v_2}{w } \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \left ( { \rm{\bf p}}^2 + w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \right ) - \frac{s_1 ^ 2 s_2 ^ 5}{\left ( s_1 + s_2 \right)^5 } \ , v_2 k_2 \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) \left ( { \rm{\bf p}}^2 + w \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda}^2 \right ) \bigg\ } \ .\end{aligned}\ ] ] the integral over the last term is convergent and therefore neglected . the remaining integrations are rendered straightforward by the change of variables @xmath458 and @xmath459 . upon expanding the result for small @xmath460 we obtain @xmath461 where we have introduced @xmath462 for convenience . we learn , that not only primitive divergencies proportional to @xmath463 , but also proportional to @xmath464 , @xmath465 and @xmath466 are generated . consider the insertion of a local operator @xmath467 by adding a term to the hamiltonian , @xmath468 where @xmath467 is a local monomial of degree @xmath12 in the fields @xmath469 with @xmath470 derivatives in real and @xmath471 derivatives in replica space . in a diagram composed of @xmath230 propagators , @xmath472 three - leg vertices and the insertion there are @xmath473 loops . the topological relation @xmath474 balances the number of legs . each propagator behaves for large momenta as @xmath475 and hence reduces the superficial degree of divergence of the diagram by 2 . the insertion increases it by @xmath476 . thus the superficial degree of divergence @xmath477 $ ] of the diagram with insertion is @xmath478 = dl + a + b - 2p \ .\end{aligned}\ ] ] with help of eqs . ( [ loopnumber ] ) and ( [ toporel ] ) one finds @xmath478 = \frac{d-6}{2 } v + \frac{d-2}{2 } n + \frac{2-d}{2 } e + a + b \ .\end{aligned}\ ] ] in contrast , the superficial degree of divergence @xmath479 of the diagram without insertion is @xmath480 the difference @xmath478 - \delta = \frac{d-2}{2 } n + a + b - d \end{aligned}\ ] ] is identical to the naive dimension @xmath481 $ ] of the insertion . for @xmath482 it reduces to @xmath483 = 2 n + a + b - d \ .\end{aligned}\ ] ] here we give details on the calculation of the conducting feynman diagrams in listed in fig . 1 . we focus on the contributions of the diagrams to the renormalization of the @xmath93 , i.e. , those terms appearing in eq . ( [ noisyexpansionofdiagrams ] ) proportional to @xmath93 . the other terms appearing in eq . ( [ noisyexpansionofdiagrams ] ) will be omitted throughout the entire appendix for the sake of notational simplicity . for details on the calculation of the contributions to the renormalization of @xmath225 we refer to@xcite . the @xmath148-independent parts of the conducting diagrams correspond to the usual diagrams found in the literature on the potts model@xcite and can be calculated by standard proceedures@xcite . we start with diagram a. the part of a required in the calculation of @xmath300 reads for vanishing external momentum @xmath484 \bigg\ { s_1 \left ( \frac{s_2}{s_1 + s_2 } \right)^{n } + s_2 \left ( \frac{s_1}{s_1 + s_2 } \right)^{n } \bigg\ } \ , \end{aligned}\ ] ] where @xmath485 . carrying out the momentum integration gives @xmath486 \frac{s_1 s_2^n}{\left ( s_1 + s_2 \right)^n } \ .\end{aligned}\ ] ] changing variables , @xmath487 and @xmath488 , leads to @xmath489 expansion for small @xmath490 yields @xmath491 where we have introduced @xmath492 for convenience . the calculation of b is particularly simple . thus we merely state the result @xmath493 now we turn to the two - loop diagrams . as an example , we consider the diagram c. as a first step , we determine the currents flowing through the conducting propagators . kirchhoff s law eq . ( [ noisycirquitequations ] ) applies to the 4 vertices of the diagram . this allows us to eliminate 3 of the 5 unknown currents ( one of the vertices is inactive with respect to this purpose since the external current @xmath329 must be conserved ) . the potential drop around closed loops is zero . hence we can eliminate the two remaining unknown currents and express all currents flowing through conducting propagators in terms of the schwinger parameters and @xmath148 . the momentum integrations are straightforward . they can be done by using the saddle point method which works exactly here since the momentum dependence is purely quadratic . after the momentum integration we have @xmath494^{d/2 } } \nonumber \\ & \times & \bigg\ { s_1 \left [ \frac{s_2 \left ( s_3 + s_4 + s_5 \right ) + s_4 s_5}{\left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 } \right]^n + s_2 \left [ \frac{s_1 \left ( s_3 + s_4 + s_5 \right ) + s_3 s_5}{\left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 } \right]^n \nonumber \\ & & + s_3 \left [ \frac{s_4 \left ( s_3 + s_4 + s_5 \right ) + s_2 s_5}{\left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 } \right]^n + s_4 \left [ \frac{s_3 \left ( s_3 + s_4 + s_5 \right ) + s_1 s_5}{\left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 } \right]^n \nonumber \\ & & + s_5 \left [ \frac{s_2 s_3 - s_1 s_4}{\left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 } \right]^n \bigg\ } \nonumber \\ & = & - \frac{g^4}{2 } v_l k_l \left ( \stackrel{\mbox{{\tiny $ \leftrightarrow$}}}{\lambda } \right ) \frac{1}{\left ( 4 \pi \right)^{d } } \int_0^\infty \prod_{i=1}^5 ds_i \ \frac{\exp \left ( - \tau \sum_{i=1}^5 s_i \right)}{\left [ \left ( s_1 + s_2 + s_5 \right ) \left ( s_3 + s_4 + s_5 \right ) - s_5 ^ 2 \right]^{d/2+n } } \nonumber \\ & \times & \bigg\ { 4 s_1 \left [ s_2 \left ( s_3 + s_4 + s_5 \right ) + s_4 s_5 \right]^n + s_5 \left [ s_2 s_3 - s_1 s_4 \right]^n \bigg\ } \ .\end{aligned}\ ] ] at this stage , the change of variables @xmath495 , @xmath496 , @xmath497 , @xmath498 , and @xmath499 turns out to be useful . it leads to @xmath500}{\left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{d/2+n } } \ t_1 t_2 \nonumber \\ & \times & \bigg\ { 4 t_1 \left ( 1-x \right ) \left [ x t_1 \left ( t_2 + t_3 \right ) + y t_2 t_3 \right]^n + t_3 \left ( t_1 t_2 \right)^n \left [ x - y \right]^n \bigg\ } \ .\end{aligned}\ ] ] the integrations over @xmath0 and @xmath501 are straightforward and can be basically looked up in a table@xcite . after some additional algebra we obtain @xmath502}{\left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{d/2 } } \nonumber \\ & \times & \bigg\ { \frac{2}{\left ( n+1 \right ) \left ( n+2 \right ) } \ \frac{t_3 \left ( t_1 t_2 \right)^{n+1}}{\left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{n } } \nonumber \\ & & + \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \ \frac{t_3 \left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^2}{t_1 \left ( t_1 + t_2\right ) } \nonumber \\ & & + \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \ \frac{t_1 \left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^2}{\left ( t_1 + t_2 \right)^2 } \nonumber \\ & & - \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \ \frac{t_3^{n+3 } \left ( t_1 + t_2 \right)^{n+1}}{t_1 \left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{n } } \nonumber \\ & & - \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \ \frac{t_1^{n+3 } t_2^{n+2}}{\left(t_1 + t_2 \right)^2 \left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{n } } \nonumber \\ & & - \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) } \ \frac{t_3 t_1^{n+2 } t_2^{n+1}}{\left(t_1 + t_2 \right ) \left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{n } } \bigg\ } \ , \end{aligned}\ ] ] we find it convenient to express the remaining integrals in terms of the parameter integrals given in app . [ app : parameterintegrals ] . for the sake of notational simplicity we introduce the notation @xmath503 where @xmath504 , and @xmath505 where @xmath506 . in terms of the parameter integrals we obtain @xmath507 \nonumber \\ & + & \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m + m_{2,0,0}^3 + m_{1,1,0}^3 \nonumber \\ & - & m_{1,0,0}^6 \left ( n \right ) - m_{0,1,0}^6 \left ( n \right ) - m_{3,0,0}^7 \left ( n \right ) - m_{2,1,0}^7 \left ( n \right ) \bigg ] \bigg\ } \ .\end{aligned}\ ] ] the final result reads @xmath508 the diagrams d to g can be evaluated in the same fashion . as another example we consider diagram h. determination of the noise cumulants of h leads to @xmath509^{d/2+n } } \nonumber \\ & \times & \bigg\ { 2 s_1 \left [ s_5 \left ( s_3 + s_4 \right ) \right]^n + 2 s_3 \left[s_4 s_5 \right]^n + s_5 \left [ \left ( s_1 + s_2 \right ) \left ( s_3 + s_4 \right ) + s_3 s_4 \right]^n \bigg\ } \ .\end{aligned}\ ] ] here , the change of variables @xmath510 , @xmath511 , @xmath512 , @xmath513 , and @xmath514 simplifies the integration . we obtain @xmath515}{\left [ t_1 t_2 + t_2 t_3 + t_1 t_3 \right]^{d/2+n } } \ t_1 ^ 2 \nonumber \\ & \times & \bigg\ { 2 y t_1 \left [ x t_1 \left ( t_2 + t_3 \right ) \right]^n + 2 t_2 \left [ x t_1 t_3 \right]^n + x t_1 \left [ \left ( 1-x \right ) t_1 \left ( t_2 + t_3 \right ) + t_2 t_3 \right]^n \bigg\ } \ .\end{aligned}\ ] ] the integrations over @xmath0 and @xmath501 are again straightforward . in terms of the parameter integrals we find @xmath516 \nonumber \\ & + & \frac{1}{\left ( n+1 \right ) \left ( n+2 \right ) } \bigg [ 2 m_{0,1,0}^8 \left ( n \right ) + m_{2,0,1}^7 \left ( n \right ) + m_{1,1,1}^7 \left ( n \right ) \bigg ] \nonumber \\ & + & \frac{1}{\left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m_{1,0,1}^3 + m_{0,1,1}^3 + \left ( n+4 \right ) m_{1,1,0}^3 \bigg ] \nonumber \\ & + & \frac{2}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m_{1,0,0}^6 \left ( n \right ) + m_{2,1,0}^7 \left ( n \right ) \bigg ] \bigg\ } \ .\end{aligned}\ ] ] finally we obtain @xmath517 the diagrams i to l can be treated in a similar manner . this appendix contains a list of the parameter integrals we use in the calculation of the noise exponents . the results stated are obtained by employing the dimensional regularization scheme . the parameter integral @xmath518 given below was introduced by breuer and janssen@xcite . the notation @xmath519 we reserved for a parameter integral we introduced in@xcite but which is not used here . for notational briefness we define @xmath520 @xmath521 is related to the digamma function @xmath394 via @xmath522}{(m-1)!\ , ( 1-m+n ) ! } \ .\end{aligned}\ ] ] the parameter integrals we use in calculating the @xmath300 are : @xmath523 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2 } } \nonumber \\ & = & \frac{g_\epsilon^2}{6\epsilon } \left\ { \left ( \frac{1}{\epsilon } + \frac{25}{12 } \right ) \left ( a^{3-\epsilon } + b^{3-\epsilon } + c^{3-\epsilon } \right ) \right . \nonumber \\ & - & \left . \left ( \frac{3}{\epsilon } + \frac{21}{4 } \right ) \left [ a^{2-\epsilon } \left ( b + c \right ) + b^{2-\epsilon } \left ( a + c \right ) + c^{2-\epsilon } \left ( a + b \right ) \right ] - 3abc \right\ } \ , \end{aligned}\ ] ] @xmath524 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2 - 2 } } \ \frac{1}{\left [ t_1 + t_2 \right]^3 } } \nonumber \\ & = & \frac{g_\epsilon^2}{2 \epsilon } \bigg\ { c^{2-\epsilon } \left ( \frac{1}{15\epsilon } + \frac{46}{450 } \right ) - c^{1-\epsilon } \left ( a+b \right ) \left ( \frac{1}{3\epsilon } + \frac{4}{9 } \right ) \nonumber \\ & + & c^{-\epsilon } \left ( a^2+b^2 \right ) \left ( \frac{2}{3\epsilon } + \frac{13}{18 } \right ) + c^{-\epsilon } ab \left ( \frac{2}{3\epsilon } + \frac{5}{9 } \right ) \bigg\ } \ , \end{aligned}\ ] ] @xmath525 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2 - 1 } } \ \frac{t_1t_2}{\left [ t_1 + t_2 \right]^3 } } \nonumber \\ & = & \frac{g_\epsilon^2}{2 \epsilon } \bigg\ { - c^{2-\epsilon } \left ( \frac{2}{15\epsilon } + \frac{107}{450 } \right ) - c^{1-\epsilon } \left ( a+b \right ) \left ( \frac{1}{3\epsilon } + \frac{4}{9 } \right ) \nonumber \\ & + & \frac{1}{3 } \left ( a^2+b^2 \right ) + \frac{1}{3 } ab \bigg\ } \ , \end{aligned}\ ] ] @xmath526 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2 } } \ \frac{t_1^n}{\left [ t_1 + t_2 \right]^n } } \nonumber \\ & = & - \frac{g_\epsilon^2}{6\epsilon } \bigg\ { a^{3-\epsilon } \left [ - \frac{n+1}{\epsilon } - \frac{25 \left ( n+1 \right)}{12 } + \frac{1}{2 } f_2 \left ( n+1 \right ) \right ] \nonumber \\ & + & c^{3-\epsilon } \frac{1}{\left ( n+2 \right ) \left ( n+3 \right ) } \left [ - \frac{6}{\epsilon } - \frac{15}{2 } + 3 \psi \left ( n+2 \right ) + 3 \psi \left ( 2 \right ) - 6 \psi \left ( n+4 \right ) \right ] \nonumber \\ & + & a^{2-\epsilon } b \left [ \frac{3}{\epsilon } + \frac{21}{4 } + \frac{3}{2 } f_1 \left ( n \right ) \right ] + a^{2-\epsilon } c \left [ \frac{3}{\epsilon } + \frac{27}{4 } + \frac{3}{2 } f_1 \left ( n+1 \right ) \right ] \nonumber \\ & + & c^{2-\epsilon } a \frac{1}{n+2 } \left [ \frac{6}{\epsilon } + \frac{9}{2 } - 3 \psi \left ( n+2 \right ) + 6 \psi \left ( n+3 \right ) + 3\gamma \right ] \nonumber \\ & + & c^{2-\epsilon } b \frac{1}{\left ( n+1 \right ) \left ( n+2 \right ) } \left [ \frac{6}{\epsilon } + \frac{9}{2 } - 3 \psi \left ( n+1 \right ) - 3 \psi \left ( 2 \right ) + 6 \psi \left ( n+3 \right ) \right ] \nonumber \\ & + & b^3 \frac{1}{2 \left ( n-1 \right ) n } + b^2 a \frac{3}{2n } + b^2 c \frac{3}{2n \left ( n+1 \right ) } - 3abc \frac{1}{n+1 } \bigg\ } \ , \end{aligned}\ ] ] @xmath527 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2+n } } \ \frac{t_3^n \left [ t_1 + t_2 \right]^n}{t_1 } } \nonumber \\ & = & - \frac{g_\epsilon^2}{\epsilon } \bigg\ { c^{1-\epsilon } \left [ \frac{\left ( n+2 \right ) \left ( n+3 \right ) } { 12\epsilon } + \frac{n+3}{6 \epsilon } + \frac{11 \left ( n+2 \right ) \left ( n+3 \right ) } { 72 } + \frac{2 \left ( n+3 \right ) } { 9 } - \frac{1}{24 } \right . \nonumber \\ & + & \left . \frac{1}{12 } \left ( f_3 \left ( n+3 \right ) - f_2 \left ( n+3 \right)\right ) \right ] \nonumber \\ & + & c^{-\epsilon } a \left [ - \frac{n+3}{3 \epsilon } - \frac{5 \left ( n+3 \right ) } { 18 } + \frac{1}{6 } + \frac{1}{6 } f_2 \left ( n+3 \right ) \right ] \nonumber \\ & + & c^{-\epsilon } b \left [ - \frac{2 \left ( n+3 \right)}{3 \epsilon } - \frac{13 \left ( n+3 \right ) } { 18 } + \frac{1}{6 } + \frac{1}{3 } f_2 \left ( n+3 \right ) \right ] \nonumber \\ & + & c^{-(1+\epsilon ) } b^2 \left ( \frac{2}{\epsilon } + \frac{3}{2 } \right ) - \frac{1}{2 } c^{-(1+\epsilon ) } a^2 - c^{-(1+\epsilon ) } ab \bigg\ } \ , \end{aligned}\ ] ] @xmath528 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2+n } } \ \frac{t_1^{n+1}t_2^{n+2}}{\left [ t_1+t_2 \right]^3 } } \nonumber \\ & = & - \frac{g_\epsilon^2}{6\epsilon } \bigg\ { b^{3-\epsilon } \left [ \frac{2}{\left ( n+2 \right ) \epsilon } + \frac{1}{\left ( n+2 \right)^2 } + \frac{11}{6 \left ( n+2 \right ) } \right]+ a^3 \frac{1}{3 \left ( n+2 \right ) } \nonumber \\ & + & c^3 \frac{1}{10 \left ( n-1 \right ) n\left ( n+1 \right ) \left ( n+2 \right ) } + a^2 b \frac{1}{2 \left ( n+2 \right ) } + a^2 c \frac{1}{4 \left ( n+1 \right ) \left ( n+2 \right ) } \nonumber \\ & + & b^2 a \frac{1}{n+2 } + b^2 c \frac{3}{4 \left ( n+1 \right ) \left ( n+2 \right ) } + c^2 a \frac{1}{5 n \left ( n+1 \right ) \left ( n+2 \right ) } \nonumber \\ & + & c^2 b \frac{3}{10 n \left ( n+1 \right ) \left ( n+2 \right ) } + abc \frac{1}{2 \left ( n+1 \right ) \left ( n+2 \right ) } \bigg\ } \ , \end{aligned}\ ] ] @xmath529 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2+n } } \ t_1^{n}t_2^{n+1}t_3 } \nonumber \\ & = & - \frac{g_\epsilon^2}{\epsilon } \bigg\ { b^{1-\epsilon } \left [ \frac{2}{\left ( n+1 \right ) \left ( n+2 \right ) \epsilon } + \frac{2n+3}{\left ( n+1 \right)^2 \left ( n+2 \right)^2 } + \frac{1}{\left ( n+1 \right ) \left ( n+2 \right ) } \right ] \nonumber \\ & + & a \frac{1}{\left ( n+1 \right ) \left ( n+2 \right ) } + c \frac{1 } { n \left ( n+1 \right ) \left ( n+2 \right ) } \bigg\ } \ .\end{aligned}\ ] ] in addition to the parameter integrals we use @xmath530 } { \left [ t_3 t_1 + t_3 t_2 + t_1 t_2 \right]^{d/2 - 2 } } \ \frac{t_3}{t_1 \left ( t_1 + t_2 \right ) } \nonumber \\ & = & \frac{g_\epsilon^2}{\epsilon } \tau^{-\epsilon } \left ( \frac{3}{\epsilon } - \frac{1}{2 } \right ) \ .\end{aligned}\ ] ] here we list our results for the conducting two - loop diagrams in terms of the parameter integrals given in app . [ app : parameterintegrals ] . for notational simplicity , we show only the parts of the diagrams proportional to @xmath276 : @xmath531 \nonumber \\ & + & \frac{4}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m + m_{2,0,0}^3 + m_{1,1,0}^3 \nonumber \\ & - & m_{1,0,0}^6 \left ( n \right ) - m_{0,1,0}^6 \left ( n \right ) - m_{3,0,0}^7 \left ( n \right ) - m_{2,1,0}^7 \left ( n \right ) \bigg ] \bigg\ } \ , \end{aligned}\ ] ] @xmath532 @xmath533 \bigg\ } \ , \end{aligned}\ ] ] @xmath534 @xmath535 @xmath536 \nonumber \\ & + & \frac{1}{\left ( n+1 \right ) \left ( n+2 \right ) } \bigg [ 2 m_{0,1,0}^8 \left ( n \right ) + m_{2,0,1}^7 \left ( n \right ) + m_{1,1,1}^7 \left ( n \right ) \bigg ] \nonumber \\ & + & \frac{1}{\left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m_{1,0,1}^3 + m_{0,1,1}^3 + \left ( n+4 \right ) m_{1,1,0}^3 \bigg ] \nonumber \\ & + & \frac{2}{\left ( n+1 \right ) \left ( n+2 \right ) \left ( n+3 \right ) } \bigg [ m_{1,0,0}^6 \left ( n \right ) + m_{2,1,0}^7 \left ( n \right ) \bigg ] \bigg\ } \ , \end{aligned}\ ] ] @xmath537 @xmath538 \bigg\ } \ , \end{aligned}\ ] ] @xmath539 @xmath540 from eq . ( [ finalcumulant ] ) it is evident that only those bonds with @xmath541 contribute to @xmath542 . consequently , @xmath401 is related to the fractal dimension @xmath543 of the singly connected ( red ) bonds via @xmath544 . coniglio @xcite proved that @xmath545 , which in turn leads to @xmath546 . as mentioned above , our result for @xmath300 matches this consistency requirement . another trivial consequence of eq . ( [ finalcumulant ] ) is , that @xmath547 is proportional to the average number of bonds ( the mass ) of the backbone . hence @xmath548 is related to the backbone dimension @xmath549 by @xmath550 this relation can also be obtained on the level of feynman diagrams . revisit the definition of the noise cumulants for feynman diagrams eq . ( [ diagramcumulant ] ) . in the limit @xmath551 the noise cumulant reduces to the sum of schwinger parameters of conducting propagators , @xmath552 now we take a short detour to our renormalized field theory of diluted networks in which the occupied bond obey a generalized ohm s law @xmath553@xcite . in these networks , the nonlinear resistance @xmath554 averaged subject to @xmath0 and @xmath1 being on the same cluster , @xmath555 obeys at criticality @xmath556 in analogy to @xmath557 we introduced the notion of the nonlinear resistance @xmath558 of feynman diagrams . in the limit @xmath559 we found @xmath560 hence we can identify @xmath561 and @xmath562 . this leads to the conclusion that @xmath563 . @xmath564 is related to the fractal dimension @xmath549 of the backbone via @xmath565 , and hence we obtain once more eq . ( [ psinulldb ] ) . equation ( [ psinulldb ] ) provides another consistency check for our result ( [ monsterexponent ] ) , which is indeed fulfilled . moreover , eq . ( [ psinulldb ] ) can be used to calculate @xmath548 to third order in @xmath39 from our three - loop result for @xmath549@xcite , @xmath566 we obtain @xmath567 for reviews on percolation see , e.g. , a. bunde and s. havlin , _ fractals and disordered systems _ ( springer , berlin , 1991 ) ; d. stauffer and a. aharony , _ introduction to percolation theory _ ( taylor&francis , london , 1992 ) ; b. d. hughes , _ random walks and random environments _ ( clarendon , oxford , 1995 ) . for general aspects of multifractals , see , c. j. c. evertsz and b. b. mandelbrot , in _ chaos and fractals : new frontiers of science _ , edited by h. o. peitgen , h. jurgens , d. saupe ( springer , new york , 1992 ) , p. 849 . for a review in the context of rrn , see , a. hansen , in _ statistical models for the fracture of disordered media _ , edited by h. j. herrmann and s. roux ( elsevier , amsterdam , 1990 ) , p. 115 b. b. mandelbrot , j. fluid mech . * 62 * , 331 ( 1974 ) ; u. frisch and g. parisi , in _ turbulence and predictability in geophysical fluid dynamics and climate dynamics _ , proceedings of the international school of physics enrico fermi , course lxxxviii , varenna , italy , 1983 , edited by m. ghil ( north - holland , new york , 1985 ) , p. 84 ; r. benzi , g. paladin , g. parisi , and a. vulpiani , j. phys . a * 18 * , 3521 ( 1984 ) . m. e. cates and t. a. witten , phys . a * 35 * , 1809 ( 1987 ) . f. wegner , nucl . * b280 * , 193 ( 1987 ) ; c. castellani and l. peliti , j. phys . a * 19 * , l429 ( 1986 ) . d. gersappe , j. m. deutsch , and m. olvera de la cruz , phys . lett . * 66 * , 731 ( 1991 ) . a. w. w. ludwig , nucl . * b285 * , 97 ( 1987 ) ; * b330 * , 639 ( 1990 ) . t. c. halsey , m. h. jensen , l. p. kadanoff , i. procaccia , and b. shraiman , phys . a * 33 * , 1141 ( 1986 ) . et al_. , nature ( london ) * 399 * , 461 ( 1999 ) . l. de arcangelis , s. redner , and a. coniglio , phys . b * 34 * , 4656 ( 1986 ) . r. rammal , c. tannous , and a .- s. tremblay , phys . a * 31 * , 2662 ( 1985 ) ; r. rammal , c. tannous , p. brenton , and a .- m . s. tremblay , phys . * 54 * , 1718 ( 1985 ) . r. m. cohn , am . math . * 1 * , 316 ( 1950 ) . l. de arcangelis , s. redner , and a. coniglio , phys . b * 31 * , 4725 ( 1985 ) . y. park , a. b. harris and t. c. lubensky , phys . b * 35 * , 5048 ( 1987 ) . m. j. stephen , phys . b * 17 * , 4444 ( 1978 ) . b. duplantier and w. w. ludwig , phys . * 66 * , 247 ( 1991 ) . b. fourcade and a .- m . s. tremblay , phys . e * 51 * , 4095 ( 1995 ) . g. g. batrouni , a. hansen , and b. larson , phys . e * 53 * , 2292 ( 1996 ) . m. barthlmy , s. v. buldyrev , s. havlin , and h. e. stanley , phys . e * 61 * , r3283 ( 2000 ) . o. stenull , h. k. janssen , and k. oerding , phys . e * 59 * , 4919 ( 1999 ) . h. k. janssen , o. stenull , and k. oerding , phys . e * 59 * , r6239 ( 1999 ) . h. k. janssen and o. stenull , phys . e * 61 * , 4821 ( 2000 ) . o. stenull and h. k. janssen , europhys . * 51 * , 539 ( 2000 ) . see , e.g. , r. k. p. zia and d. j. wallace , j. phys . a * 8 * , 1495 ( 1975 ) . m. e. fisher , in _ renormalization group in critical phenomena and quantum field theory - proceedings of a conference _ , edited by j. d. gunton and m. s. green ( temple university press , philadelphia , 1974 ) . d. j. amit and l. peliti , ann . phys . ( usa ) * 140 * , 207 ( 1982 ) . h. w. diehl , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic press , london , 1986 ) , vol . 10 , p. 171 . see , e.g. , d. j. amit , _ field theory , the renormalization group , and critical phenomena _ ( world scientific , singapore , 1984 ) ; j. zinn - justin , _ quantum field theory and critical phenomena _ ( clarendon , oxford , 1989 ) . o. f. de alcantara bonfim , j. e. kirkham and a. j. mckane , j. phys . a * 13 * , l247 ( 1980 ) ; * 14 * , 2391 ( 1981 ) . t. c. lubensky and j. wang , phys . rev . b * 33 * , 4998 ( 1985 ) . see , e.g. , m. abramowitz and i. a. stegun , _ handbook of mathematical functions _ ( dover publications , new york , 1965 ) . r. blumenfeld , y. meir , a. aharony , and a. b. harris , phys . b * 35 * , 3524 ( 1987 ) . c. moukarzel , int . j. mod . c * 8 * , 887 ( 1998 ) . p. grassberger , physica a * 262 * , 251 ( 1999 ) . o. stenull and h. k. janssen , forthcoming paper . see , e.g. , i. s. gradshteyn , i. m. ryzhik , _ table of integrals , series , and products _ ( academic press , new york , 4 . edition , 1980 ) . n. breuer and h. k. janssen , z. phys . b * 41 * , 55 ( 1981 ) . a. coniglio , phys . lett . * 46 * , 250 ( 1981 ) . a. coniglio , j. phys . a * 15 * , 3829 ( 1982 ) . c. von ferber and y. holovatch , europhys . 39 * , 31 ( 1997 ) ; in _ renormalization group 97 _ , dubna conf . ( world scientific , singapure , 1997 ) ; phys . e * 59 * , 6914 ( 1999 ) .	we study the multifractal moments of the current distribution in randomly diluted resistor networks near the percolation treshold . when an external current is applied between to terminals @xmath0 and @xmath1 of the network , the @xmath2th multifractal moment scales as @xmath3 , where @xmath4 is the correlation length exponent of the isotropic percolation universality class . by applying our concept of master operators [ europhys . lett . * 51 * , 539 ( 2000 ) ] we calculate the family of multifractal exponents @xmath5 for @xmath6 to two - loop order . we find that our result is in good agreement with numerical data for three dimensions . # 1@xmath7#1
the karmen collaboration , which studies the interactions of neutrinos from the stopped @xmath6 decay chain at rutherford appelton laboratory , has reported an anomaly @xcite in the time distribution of neutrino events in the karmen calorimeter . there is a bump of events in the distribution which was expected to be well described by a single exponential with a time constant equal to the muon decay lifetime . it was suggested that a new weakly - interacting half spin particle with a mass 33.9 mev / c@xmath7 from initial @xmath8 decay , which travels with velocity @xmath9 and decays in the karmen calorimeter after a mean flight path of 17.5 metres , can be a possible candidate for an explanation of this anomaly . since the visible energy in the calorimeter was in the range from 11 to 35 mev it was also assumed that these new particles deposited only a fraction of their full energy in the calorimeter . the hypothetical @xmath8 decay has been searched for at psi by studying muons from pion decay in flight @xcite . the result gives a branching ratio of less than @xmath10 at 95@xmath11 confidence level ( c.l . ) for this decay mode leaving only a narrow allowed region in the branching ratio if combined with the results of ref . @xcite . another possible interpretation of the anomaly proposed by the karmen collaboration is related to the direct production of a new particle in the proton collision with the neutrino target . however , the collaboration has not yet stated whether it has devised one . concerning the possible explanation of the anomaly by exotic muon decays it was concluded that `` ... the narrow clustering of the anomalous events rules out contributions from @xmath0 decay , which spreads over a much longer time interval ( compared to the @xmath6 - life time of @xmath12 ) up to 10 @xmath1 after beam - on target . '' @xcite . this conclusion is the motivation for present letter , where we show that the karmen anomaly could be explained by contributions from the exotic @xmath0 decays . the paper is organised as follows . in section 2 we analyse the time distribution obtained by the karmen collaboration and give a possible explanation of the anomaly . as an example we consider the @xmath3 decay into a positron and a new boson with the mass of 103.9 mev/[email protected] sections 3,4 the possible x - boson phenomenologies and experiments on searching for the rare decay @xmath3 are discussed , respectively . section 5 contains concluding remarks . figure 1 shows the time distribution found by karmen with respect to beam - on target events . analysis of this time distribution was performed in ref.@xcite after subtraction of the cosmic ray induced background , which was found to be constant and equal to 60 events per 0.5 @xmath1 . this spectrum was fitted in ref . @xcite by a superposition of gaussian and exponential signals from muon decay neutrinos having a 2.2 @xmath1 decay constant . the fit procedure performed with the minimization routines of minuit @xcite results in @xmath13 of 9.7 for 14 degrees of freedom ( df ) corresponding to 79@xmath11 c.l . for this hypothesis.gaussian , which is centred at @xmath14 with a full width at half maximum(fwhm ) of @xmath15 consisted of 83@xmath1628 events associated with the signals from decay in flight of the proposed particle in the fiducial volume of the karmen calorimeter . we tried to reproduce these results by using the same minuit minimizing routines . our results and results of the karmen fit were found to be in a good agreement : our fit procedure yielded @xmath13 of 9.8 for 14 degrees of freedom ( df ) corresponding to 77@xmath11 c.l . for this hypothesis.gaussian was found to be centred at @xmath17 with a fwhm of @xmath15 and with 85@xmath1632 events . the result of our fit procedure is shown in figure 1 . having this good agreement in mind , we then performed an exponential fit of the karmen time distribution assuming that this spectrum is described by a sum of two exponentials : the first exponential describes the time distribution in the interval 1 - 3.3 @xmath1 , and the second one describes the time distribution in the interval 3.3 - 10 @xmath1 . figure 2 shows the result of the fit ( solid lines).the minuit minimizing routines yield a @xmath18 of 9.7 for 15 df corresponding to 84@xmath11 c.l . , and time constants of @xmath19 and @xmath20 for the first and the second exponentials respectively . note that there is no significant difference in confidence levels of our and karmen s hypotheses . the time width of the transition region between two exponentials is rather small and corresponds to @xmath21 . we try to fit this region more precisely using a gaussian shape of the transition curve instead of a step - function . however , no significant difference between two results has been found . a similar description of the time distribution was obtained by superimposing an exponential describing the distribution of events in the time interval 3.3 - 10 @xmath1 upon an exponential signal having a 2.2 @xmath1 decay constant for the full time interval . figure 3 shows the corresponding time distribution of events after subtraction of the exponential part with 2.2 @xmath1 . a direct exponential fit of the time distribution of events in this region gives a time constant of 2.26 @xmath1 close to the value of muon decay constant ; however the error is rather large because of poor statistics . since the time constants were found to be equal within the error limits to the muon decay constant we may assume that all events in the karmen time distribution originate from the muon decays . thus , our interpretation of the karmen anomaly is the following . positive muons from @xmath6 decays stop in the target ( muon range in the target is @xmath22 1 cm ) and decay giving a flux of neutrinos which decreases in time exponentially with the muon decay constant . the first exponential shown in the time spectrum in figure 2 is related to the interactions of this flux of neutrinos in the karmen calorimeter.an excess of events after @xmath23 3.6 @xmath1 shown in figure 3 is related to the decays of slow moving particles produced in muon decays.therefore , the flux of these particles and fraction of their decays in the fiducial volume of the karmen calorimeter also decrease in time exponentially with the muon decay constant , resulting in the second time distribution but shifted by the time of flight with respect to the time distribution of neutrino events . since the time width of the transition region between two exponentials is rather narrow ( see figure 2 ) , we suggest that these new particles originate from the two - body decays of positive muons , namely @xmath24 . we note that the rms of fluctuations of time needed for @xmath4 bosons to pass from the neutrino target to the calorimeter and decay inside its volume was estimated to be of the order of 0.2 @xmath1 . the effect of the convolution of these fluctuations and the exponential from muon decay on the exponential shape of this curve and its decay constant was found to be small . the mass of the @xmath4 boson calculated for @xmath25 is found to be ( [email protected] ) mev / c@xmath7 , where the error corresponds to the spread of @xmath4 s velocity estimated from the time width of the transition region between two exponentials . the excess of events after subtraction of the contribution from the neutrino was found to be @xmath26 events . the limit on branching ratio @xmath27 as a function of the @xmath4-boson lifetime can be recalculated from the karmen result for @xmath28 decay mode taking into account the difference in the number of excess events found after the fit procedure . the result is shown in figure 4 . suppose that x - boson is a neutral scalar particle . the lagrangian describing the @xmath29 decay has the form @xmath30\ ] ] the @xmath29 decay width is given by the formula @xmath31^{\frac{1}{2}}\ ] ] for a non - relativistic x - boson with a speed @xmath32 we find that its mass is @xmath33 mev and the @xmath29 branching ratio is @xmath34\ ] ] for the interaction ( 1 ) the x - boson decays via the virtual @xmath35 meson @xmath36 and its decay width is @xmath37 as it follows from the observed number of the anomalous events due to the x - boson decay in the karmen experiment , the ratio @xmath38 is @xmath39 from the formulae ( 2 -6 ) we find that @xmath40 and @xmath41 , that is probably at the level of the contradiction with the experimental bound @xcite @xmath42 which is valid however for @xmath43 mev ( see section 4 ) . it is possible to invent a model where the @xmath44 is less than @xmath45 . let us introduce the hypothetical heavy charged lepton l and consider the decay of the x - boson through the virtual charged l - lepton @xmath46 for the l - lepton mass @xmath47 gev , the coupling constant of l - lepton with electron and x - boson @xmath48 , the l - lepton mixing with electron @xmath49 , we find that @xmath50 . so , if our interpretation of the karmen anomaly is correct we expect a rather large branching ratio @xmath51 . therefore , it would be very interesting to perform an experiment on the search for @xmath29 decay with improved sensitivity . these scenarios predict the x - boson lifetime @xmath52 s that probably leads to problems with astrophysical and cosmological constraints . moreover , these scenarios predict an anomalously high x - boson decay energy @xmath53 mev compared to the observed visible energy @xmath54 mev . to avoid these problems , let us consider the model with two additional neutral scalar bosons @xmath55 and @xmath56 and suppose that the @xmath55-boson is lighter than the x - boson and the @xmath56-boson is heavier than the x - boson . the lagrangian describing the x - boson decay has the form @xmath57 the x - boson decays via the reaction @xmath58 through the virtual @xmath56-boson and the neutral @xmath55-boson decays into neutrino antineutrino pair . for such a scenario it is possible to have @xmath59 mev and @xmath60 s. for instance , for @xmath33 mev , @xmath61 mev , @xmath62 gev , @xmath63 , @xmath64 and @xmath65 we find that the average visible energy in the x - decay is @xmath66 mev and @xmath67 s , @xmath68 s , @xmath69 , @xmath70 s. so in this rather artificial model all particles are not long - lived and it is possible to have @xmath71 in agreement with karmen data . note also that it is possible to have a model where the @xmath56-boson is the higgs boson . for instance , for the higgs boson mass @xmath72 gev and @xmath73 we find that @xmath74 s and @xmath75 . in this case to have @xmath60 s we have to postulate an additional interaction of the @xmath4-boson with neutrinos . since @xmath4 s are weakly - interacting and relatively long lived particles , their flux would penetrate any type of calorimeter without significant attenuation and can not be observed effectively in the detector via its decays . a more effective experimental signature for @xmath76 decays would be a peak from mono - energetic positrons in the low energy part of the positron spectrum corresponding to the principal mode of @xmath77 decays . direct experimental searches for a peak from @xmath78 decays were performed in a few experiments @xcite . for the lowest positron energy region from 1.6 to 6.8 mev , corresponding to the mass region of 103.5 - 98.3 mev / c@xmath7 , the branching ratio limit of @xmath79 was obtained in ref . @xcite from the analysis of published data on muon decays in the hydrogen bubble chamber obtained by derenzo @xcite . the result is limited by a poor ( @xmath22 30 - 40@xmath11 ) momentum resolution in this region . it would be interesting to search for a peak in the low energy part of the positron spectrum by using , for example , a high resolution ge - detector ( nai detector ) with a typical energy resolution of the order of kev ( @xmath22 100 kev ) . the low energy muons could be stopped in the active volume of the detector . in this case the low energy spectrum of the positrons could be measured without significant distortions . the estimate of the rate from the principal muon decays shows that , using ge - detector , one could expect a sensitivity to the @xmath76 decay branching ratio of better than @xmath80 . here it is assumed that the intensity of stops is of the order of @xmath81/s and that the exposure to the muon beam is @xmath22 1 month . the anomalous time distribution of the events from the karmen detector can be acceptably described by a superposition of the time distribution of neutrino signals following the expected 2.2 @xmath1 muon decay constant and a weaker exponential signal from delayed decays of the new massive neutral particles also originated from the muon decay at rest and shifted in time due to their time of flight . it is proposed that a new boson @xmath4 with the mass of 103.9 mev / c@xmath7 which might be produced in the decays of positive muons @xmath3 could be a possible candidate for an explanation of the anomaly . the visible energy from the possible decay modes of @xmath4 s can be within the allowed region from 11 to 35 mev / c@xmath7 . this hypothesis can be experimentally tested in the near future at a level of sensitivity of better than 10@xmath82 . it should be noted that the original interpretation of the karmen anomaly as a new neutral fermion @xmath83 produced in @xmath84 decay with a mass @xmath85 mev has been discussed in ref . it has been shown that a mainly sterile neutrino scenario is compatible with all laboratory constraints , although there are problems with astrophysical and cosmological constraints . our interpretation with the scalar @xmath4-boson produced in the @xmath86 decay is in the spirit of the familon scenario @xcite . we thank prof . v.matveev for useful discussions and remarks , prof . f.wilczek for useful comment and a.toropin for help in calculations . one of the authors ( s.n.g . ) would like to thank his colleagues from the nomad collaboration at cern for fruitful discussions . k. eitel , talk given at cern on february 17 , 1998 , where reality of the anomaly was further confirmed by still more such events from the karmen detector . r. bilger at el . , phys.lett.b363(1995)41 . v. barger , r.j.n . phillips and s. sarkar , phys.lett.b352(1995)365 . f. james and m. roos , cernlib write up no . d506 , 1977 . particle data group , phys . v. d54(1996 ) part 1 . bryman and e.t.h . clifford , phys . derenzo , phys . rev . 181 * ( 1969 ) 1854 . f. wilczek , phys.rev.lett.49(1982)1549 .	an anomaly in time distribution of neutrinos from the isis pulsed beam stop source observed by the karmen collaboration is discussed . we show that the anomaly can be interpreted as a superposition of two exponentials , both having time constants consistent with the @xmath0 lifetime of 2.2 @xmath1.it is assumed that they both originate from muon decays at rest . one of them describes the time distribution of the prompt neutrino events , while the other describes the time distribution of events from delayed decays of slowly moving ( @xmath2 ) particles in the karmen calorimeter . we propose here that these particles are produced in exotic decays of positive muons @xmath3 , resulting in the second exponential time distribution shifted by the time of flight with respect to the time distribution of neutrino events . this model gives an aceptable fit to the karmen data if @xmath4 has a mass of 103.9 mev . the possible decay modes of this new massive neutral particle are discussed . this hypothesis can be experimentally tested in the near future by studying the low energy part of the @xmath5 spectrum in the @xmath0 decays . -0.5 in 8.5 in 5.7 in * exotic muon decays and the karmen anomaly * s.n . gninenko and n.v . krasnikov + _ institute for nuclear research of the russian academy of sciences , moscow 117312 _
let @xmath18 be a finite set of points in an axis - aligned bounding rectangle @xmath19 . an _ anchored rectangle packing _ for @xmath20 is a set of axis - aligned empty rectangles @xmath21 that lie in @xmath19 , are interior - disjoint , and @xmath3 is one of the four corners of @xmath4 for @xmath22 ; rectangle @xmath4 is said to be _ anchored _ at @xmath3 . for a given point set @xmath23 , we wish to find the maximum total area @xmath24 of an anchored rectangle packing of @xmath20 . since the ratio between areas is an affine invariant , we may assume that @xmath25 ^ 2 $ ] . however , if we are interested in the maximum area of an _ anchored square packing _ , we _ must _ assume that @xmath25 ^ 2 $ ] ( or that the aspect ratio of @xmath19 is bounded from below by a constant ; otherwise , with an _ arbitrary _ rectangle @xmath19 , the guaranteed area is zero ) . finding the maximum area of an anchored rectangle packing of @xmath6 given points is suspected but not known to be np - hard . balas and tth @xcite observed that the number of distinct rectangle packings that attain the maximum area , @xmath24 , can be exponential in @xmath6 . from the opposite direction , the same authors @xcite proved an exponential upper bound on the number of maximum area configurations , namely @xmath26 , where @xmath27 is the @xmath6th catalan number . note that a greedy strategy may fail to find @xmath24 ; see fig . [ fig : greedyfail ] . , with @xmath28 and @xmath29 , a greedy algorithm selects rectangles of area @xmath30 ( left ) , which is less than the area @xmath31 of the packing on the right . ] [ [ variants - and - generalizations . ] ] variants and generalizations . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider three additional variants of the problem . an _ anchored square packing _ is an anchored rectangle packing in which all rectangles are squares ; a _ lower - left anchored rectangle packing _ is a rectangle packing where each point @xmath32 is the lower - left corner of @xmath4 ; and a _ lower - left anchored square packing _ has both properties . we suspect that all variants , with rectangles or with squares , are np - hard . here , we put forward several approximation algorithms , while it is understood that the news regarding np - hardness can occur at any time or perhaps take some time to establish . the problem can be generalized to other geometric shapes with distinct representatives . let @xmath33 be a finite set of points in a compact domain @xmath34 , and let @xmath35 be @xmath6 families of measurable sets ( , rectangles , squares , or disks ) such that for all @xmath36 , we have @xmath37 and @xmath38 is the measure of @xmath39 . an _ anchored packing for @xmath40 _ is a set of pairwise interior - disjoint representatives @xmath21 with @xmath41 for @xmath5 . we wish to find an anchored packing for @xmath40 of maximum measure @xmath42 . while some variants are trivial ( , when @xmath25 ^ 2 $ ] and @xmath43 consists of all rectangles containing @xmath3 ) , there are many interesting and challenging variants ( , when @xmath43 consists of disks containing @xmath3 ; or when @xmath19 is nonconvex ) . in this paper we assume that the domain @xmath19 and the families @xmath44 are axis - aligned rectangles or axis - aligned squares in the plane . + ' '' '' guaranteed max . area & @xmath45 & @xmath46 & @xmath47 & @xmath47 + greedy approx . ratio & @xmath48 & @xmath14 & @xmath49 @xcite & @xmath15 + approximation scheme & qptas & ptas & qptas & ptas + [ [ contributions . ] ] contributions . + + + + + + + + + + + + + + our results are summarized in table [ table:1 ] . \(i ) we first deduce upper and lower bounds on the maximum area of an anchored rectangle packing of @xmath6 points in @xmath1 ^ 2 $ ] . for @xmath50 , let @xmath51 . we prove that @xmath52 for all @xmath8 ( sections [ sec : upper ] and [ sec : lower ] ) . \(ii ) let @xmath53 be the maximum area of an anchored square packing for a point set @xmath20 , and @xmath54 . we prove that @xmath46 for all @xmath6 ( sections [ sec : upper ] and [ sec : lower : sq ] ) . \(iii ) the above constructive lower bounds immediately yield constant - factor approximations for computing anchored packings of maximum area ( @xmath12 for rectangles and @xmath10 for squares ) in @xmath13 time ( sections [ sec : lower ] and [ sec : lower : sq ] ) . in section [ sec : approx ] we show that a ( natural ) greedy algorithm for anchored square packings achieves a better approximation ratio , namely @xmath55 , in @xmath56 time . by refining some of the tools developed for this bound , in section [ sec : approx : ll ] we prove a tight bound of @xmath15 for the approximation ratio of a greedy algorithm for lower - left anchored square packings . \(iv ) we obtain a polynomial - time approximation scheme ( ptas ) for the maximum area anchored square packing problem , and a quasi - polynomial - time approximation scheme ( qptas ) for the maximum area anchored rectangle packing problem , via a reduction to the maximum weight independent set ( mwis ) problem for axis - aligned squares @xcite and rectangles @xcite , respectively . given @xmath6 points , an @xmath57-approximation for the anchored square packing of maximum area can be computed in time @xmath16 ; and for the anchored rectangle packing of maximum area , in time @xmath17 . both results extend to the lower - left anchored variants ( section [ sec : schemes ] ) . [ [ motivation - and - related - work . ] ] motivation and related work . + + + + + + + + + + + + + + + + + + + + + + + + + + + + packing axis - aligned rectangles in a rectangular container , albeit without anchors , is the unifying theme of several classic optimization problems . the 2d knapsack problem , strip packing , and 2d bin packing involve arranging a set of given rectangles in the most economic fashion @xcite . the maximum area independent set ( mais ) problem for rectangles ( squares , or disks , etc . ) , is that of selecting a maximum area packing from a given set @xcite ; see classic papers such as @xcite and also more recent ones @xcite for quantitative bounds and constant approximations . these optimization problems are np - hard , and there is a rich literature on approximation algorithms . given an axis - parallel rectangle @xmath19 in the plane containing @xmath6 points , the problem of computing a maximum - area empty axis - parallel sub - rectangle contained in @xmath19 is one of the oldest problems studied in computational geometry @xcite ; the higher dimensional variant has been also studied @xcite . in contrast , our problem setup is fundamentally different : the rectangles ( one for each anchor ) have variable sizes , but their location is constrained by the anchors . map labeling problems in geographic information systems ( gis ) @xcite call for choosing interior - disjoint rectangles that are incident to a given set of points in the plane . gis applications often impose constraints on the label boxes , such as aspect ratio , minimum and maximum size , or priority weights . most optimization problems of such variants are known to be np - hard @xcite . in this paper , we focus on maximizing the total area of an anchored rectangle packing . in a restricted setting where each point @xmath3 is the lower - left corner of the rectangle @xmath4 and @xmath58 , allen freedman @xcite conjectured almost 50 years ago that there is a lower - left anchored rectangle packing of area at least @xmath59 . the current best lower bound on the area under these conditions is ( about ) @xmath49 , as established in @xcite . the analogous problem of estimating the total area for lower - left anchored square packings is much easier . if @xmath20 consists of the @xmath6 points @xmath60 , @xmath61 , then the total area of the @xmath6 anchored squares is at most @xmath62 , and so it tends to zero as @xmath6 tends to infinity . a looser anchor restriction , often appearing in map labeling problems with square labels , requires the anchors to be contained in the boundaries of the squares , however the squares need to be congruent ; see , @xcite . in the context of _ covering _ ( as opposed to _ packing _ ) , the problem of covering a given polygon by disks of given centers and varying radii such that the sum of areas of the disks is minimized has been considered in @xcite . in particular , covering @xmath1 ^ 2 $ ] with @xmath63-disks of given centers and minimal area as in @xcite is dual to the anchored square packings studied in this paper . [ [ notation . ] ] notation . + + + + + + + + + given an @xmath6-element point set @xmath20 contained in @xmath25 ^ 2 $ ] , denote by @xmath64 a packing ( of rectangles or squares , as the case may be ) of maximum total area . an algorithm for a packing problem has approximation ratio @xmath65 if the packing it computes , @xmath66 , satisfies @xmath67 , for some @xmath68 . a set of points is _ in general position _ if no two points have the same @xmath69- or @xmath70-coordinate . the boundary of a planar body @xmath71 is denoted by @xmath72 , and its interior by @xmath73 . [ prop:2/3 ] for every @xmath8 , there exists a point set @xmath74 such that every anchored rectangle packing for @xmath74 has area at most @xmath75 . consequently , @xmath76 . consider the point set @xmath77 , where @xmath78 , for @xmath79 ; see fig . [ f3](left ) . let @xmath80 be an anchored rectangle packing for @xmath20 . since @xmath81 , any rectangle anchored at @xmath82 has height at most @xmath83 , width at most @xmath83 , and hence area at most @xmath84 . for @xmath85 , the @xmath69-coordinate of @xmath3 , @xmath86 , is halfway between 0 and @xmath87 , and @xmath88 is halfway between 0 and @xmath89 . consequently , if @xmath3 is the lower - right , lower - left or upper - left corner of @xmath4 , then @xmath90 . if , @xmath3 is the upper - right corner of @xmath4 , then @xmath91 . therefore , in all cases , we have @xmath92 . the total area of an anchored rectangle packing is bounded from above as follows : @xmath93 = 3.3 in [ prop:7/27 ] for every @xmath8 , there exists a point set @xmath74 such that every anchored square packing for @xmath74 has area at most @xmath94 . consequently , @xmath95 . consider the point set @xmath77 , where @xmath96 , for @xmath97 ; see fig . [ f3](right ) . let @xmath98 be an anchored square packing for @xmath20 . since @xmath99 and @xmath100 , any square anchored at @xmath82 or at @xmath101 has side - length at most @xmath102 , hence area at most @xmath103 . for @xmath104 , the @xmath69-coordinate of @xmath3 , @xmath86 , is halfway between 0 and @xmath87 , and @xmath88 is halfway between 0 and @xmath89 . hence any square anchored at @xmath3 has side - length at most @xmath105 , hence area at most @xmath106 . the total area of an anchored square packing is bounded from above as follows : @xmath107 [ [ remark . ] ] remark . + + + + + + + stronger upper bounds hold for small @xmath6 , , @xmath108 . specifically , @xmath109 attained for the center @xmath110 ^ 2 $ ] , and @xmath111 and @xmath112 attained for @xmath113 . in this section , we prove that for every set @xmath20 of @xmath6 points in @xmath1 ^ 2 $ ] , we have @xmath114 . our proof is constructive ; we give a divide & conquer algorithm that partitions @xmath19 into horizontal strips and finds @xmath6 anchored rectangles of total area bounded from below as required . we start with a weaker lower bound , of about @xmath59 , and then sharpen the argument to establish the main result of this section , a lower bound of about @xmath115 . [ pn/2(n+1 ) ] for every set of @xmath6 points in the unit square @xmath1 ^ 2 $ ] , an anchored rectangle packing of area at least @xmath116 can be computed in @xmath13 time . let @xmath77 be a set of points in the unit square @xmath1 ^ 2 $ ] sorted by their @xmath70-coordinates . draw a horizontal line through each point in @xmath20 ; see fig . [ fig : horizontalstrips](left ) . these lines divide @xmath1 ^ 2 $ ] into @xmath117 horizontal strips . a strip can have zero width if two points have the same @xmath69-coordinate . we leave a narrowest strip empty and assign the remaining strips to the @xmath6 points such that each rectangle above ( resp . , below ) the chosen narrowest strip is assigned to a point of @xmath20 on its bottom ( resp . , top ) edge . for each point divide the corresponding strip into two rectangles with a vertical line through the point . assign the larger of the two rectangles to the point . ( here @xmath118 ) . right : an example of the partition for even @xmath6 ( here @xmath119 ) . the strip that is discarded is shaded in the drawing . ] the area of narrowest strip is at most @xmath120 . the rectangle in each of the remaining @xmath6 strips covers at least @xmath83 of the strip . this yields a total area of at least @xmath116 . for a stronger lower bound , our key observation is that for _ two _ points in a horizontal strip , one can always pack two anchored rectangles in the strip that cover strictly more than half the area of the strip . specifically , we prove the following easy - looking statement with @xmath121 points in a rectangle ( however , we did not find an easy proof ! ) . the proof of lemma [ lem : rectangle ] is deferred to appendix [ sec : rectangle ] . [ lem : rectangle ] let @xmath122 be two points in an axis - parallel rectangle @xmath123 such that @xmath82 lies on the bottom side of @xmath123 . then there exist two empty rectangles in @xmath123 anchored at the two points of total area at least @xmath124 , and this bound is the best possible . in order to partition the unit square into strips that contain two points , one on the boundary , we need to use parity arguments . let @xmath20 be a set of @xmath6 points in @xmath1 ^ 2 $ ] with @xmath70-coordinates @xmath125 . set @xmath126 and @xmath127 . for @xmath128 , put @xmath129 , namely @xmath130 is the @xmath131th vertical gap . obviously , we have @xmath132 parity considerations are handled by the following lemma . [ lem : partition ] ( i ) if @xmath6 is odd , then at least one of the following @xmath133 inequalities is satisfied : @xmath134 \(ii ) if @xmath6 is even , then at least one of the following @xmath135 inequalities is satisfied : @xmath136 assume first that @xmath6 is odd . put @xmath137 and assume that none of the inequalities in is satisfied . summation would yield @xmath138 which is an obvious contradiction . assume now that @xmath6 is even . put @xmath139 and assume that none of the inequalities in is satisfied . summation would yield @xmath140 an obvious contradiction . we can now prove the main result of this section . [ thm:7/12 ] for every set of @xmath6 points in the unit square @xmath1 ^ 2 $ ] , an anchored rectangle packing of area at least @xmath141 when @xmath6 is odd and @xmath142 when @xmath6 is even can be computed in @xmath13 time . let @xmath77 be a set of points in the unit square @xmath1 ^ 2 $ ] sorted by their @xmath70-coordinates with the notation introduced above . by lemma [ lem : partition ] , we find a horizontal strip corresponding to one of the inequalities that is satisfied . assume first that @xmath6 is odd . draw a horizontal line through each point in @xmath143 , for @xmath144 even , as shown in fig . [ fig : horizontalstrips ] . these lines divide @xmath1 ^ 2 $ ] into @xmath145 rectangles ( horizontal strips ) . suppose now that the satisfied inequality is @xmath146 for some odd @xmath131 . then we leave a rectangle between @xmath147 and @xmath148 empty , , @xmath4 is a rectangle of area @xmath47 . for the remaining rectangles , we assign two consecutive points of @xmath20 such that each strip above @xmath148 ( resp . , below @xmath147 ) is assigned a point of @xmath20 on its bottom ( resp . , top ) edge . within each rectangle @xmath123 , we can choose two anchored rectangles of total area at least @xmath124 by lemma [ lem : rectangle ] . by lemma [ lem : partition](i ) , the area of the narrowest strip is at most @xmath149 . consequently , the area of the anchored rectangles is at least @xmath150 . assume now that @xmath6 is even . if the selected horizontal strip corresponds to the inequality @xmath151 , then divide the unit square along the lines @xmath152 , where @xmath131 is odd . we leave the strip of height @xmath153 empty , and assign pairs of points to all remaining strips such that one of the two points lies on the top edge of the strip . we proceed analogously if the inequality @xmath154 is satisfied . suppose now that the satisfied inequality is @xmath155 . if @xmath131 is odd , we leave the strip of height @xmath156 ( between @xmath147 and @xmath152 ) empty ; if @xmath131 is even , we leave the strip of height @xmath157 ( between @xmath152 and @xmath158 ) empty . above and below the empty strip , we can form a total of @xmath159 strips , each containing two points of @xmath20 , with one of the two points lying on the bottom or the top edge of the strip . by lemma [ lem : partition](i ) , the area of the empty strip is at most @xmath160 . consequently , the area of the anchored rectangles is at least @xmath161 , as claimed . given a set @xmath162 ^ 2 $ ] of @xmath6 points , we show there is an anchored square packing of large total area . the proof we present is constructive ; we give a recursive partitioning algorithm ( as an inductive argument ) based on a quadtree subdivision of @xmath19 that finds @xmath6 anchored squares of total area at least @xmath10 . we need the following easy fact : [ obs : adjacent1 ] let @xmath163 be two congruent axis - aligned interior - disjoint squares sharing a common edge such that @xmath164 and @xmath165 . then @xmath166 contains an anchored empty square whose area is at least @xmath167 . let @xmath168 denote the side - length of @xmath169 ( or @xmath170 ) . assume that @xmath170 lies right of @xmath169 . let @xmath171 be the rightmost point in @xmath169 . if @xmath172 lies in the lower half - rectangle of @xmath169 then the square of side - length @xmath173 whose lower - left anchor is @xmath172 is empty and has area @xmath174 . similarly , if @xmath172 lies in the higher half - rectangle of @xmath169 then the square of side - length @xmath173 whose upper - left anchor is @xmath172 is empty and has area @xmath174 . [ thm:5/32 ] for every set of @xmath6 points in @xmath25 ^ 2 $ ] , where @xmath175 , an anchored square packing of total area at least @xmath10 can be computed in @xmath176 time . we first derive a lower bound of @xmath177 and then sharpen it to @xmath10 . we proceed by induction on the number of points @xmath6 contained in @xmath19 and assigned to @xmath19 ; during the subdivision process , the rle of @xmath19 is taken by any subdivision square . if all points in @xmath20 lie on @xmath19 s boundary , @xmath178 , pick one arbitrarily , say , @xmath179 ; assume @xmath180 . ( all assumptions in the proof are made without loss of generality . ) then the square @xmath181 \times [ 0,1/2]$ ] is empty and its area is @xmath182 , as required . otherwise , discard the points in @xmath183 and continue on the remaining points . if @xmath184 , we can assume that @xmath185 . then the square of side - length @xmath59 whose lower - left anchor is @xmath172 is empty and contained in @xmath19 , as desired ; hence @xmath186 . if @xmath187 let @xmath188 be the widths of the @xmath189 vertical strips determined by the two points , where @xmath190 . we can assume that @xmath191 ; then there are two anchored empty squares with total area at least @xmath192 , as required . assume now that @xmath193 . subdivide @xmath19 into four congruent squares , @xmath194 , labeled counterclockwise around the center of @xmath19 according to the quadrant containing the square . partition @xmath20 into four subsets @xmath195 such that @xmath196 for @xmath197 , with ties broken arbitrarily . we next derive the lower bound @xmath198 . we distinguish @xmath199 cases , depending on the number of empty sets @xmath200 . * case 1 : precisely one of @xmath195 is empty . * we can assume that @xmath201 . by observation [ obs : adjacent1 ] , @xmath202 contains an empty square anchored at a point in @xmath203 of area at least @xmath204 . by induction , @xmath205 and @xmath206 each contain an anchored square packing of area at least @xmath207 . overall , we have @xmath208 , which holds for @xmath209 . * case 2 : precisely two of @xmath195 are empty . * we can assume that the pairs @xmath210 and @xmath211 each consist of one empty and one nonempty set . by observation [ obs : adjacent1 ] @xmath202 and @xmath212 , respectively , contain a square anchored at a point in @xmath203 and @xmath213 of area at least @xmath214 . hence @xmath215 . * case 3 : precisely three of @xmath195 are empty . * we can assume that @xmath216 . let @xmath217 be a maximal point in the product order ( e.g. , the sum of coordinates is maximum ) . then @xmath218\times [ b , b+\frac12]$ ] is a square anchored at @xmath219 , @xmath220 ^ 2 $ ] since @xmath221 , and @xmath222 . hence @xmath223 . * case 4 : @xmath224 for every @xmath197 . * note that @xmath225 , where the squares anchored at @xmath200 are restricted to @xmath226 . induction completes the proof in this case . in all four cases , we have verified that @xmath227 , as claimed . the inductive proof can be turned into a recursive algorithm based on a quadtree subdivision of the @xmath6 points , which can be computed in @xmath13 time @xcite . in addition , computing an extreme point ( with regard to a specified axis - direction ) in any subsquare over all needed such calls can be executed within the same time bound . note that the bound in case 3 is at least @xmath10 and case 4 is inductive . sharpening the analysis of cases 1 and 2 yields an improved bound @xmath10 ; since @xmath228 , the value @xmath10 is not a bottleneck for cases 3 and 4 . details are given in appendix [ app:5/32 ] ; the running time remains @xmath13 . in this section we investigate better approximations for square packings . given a finite point set @xmath229 ^ 2 $ ] , perhaps the most natural greedy strategy for computing an anchored square packing of large area is the following . [ alg : sq]set @xmath230 and @xmath231 . while @xmath232 , repeat the following . for each point @xmath233 , compute a _ candidate _ square @xmath234 such that ( i ) @xmath235 ^ 2 $ ] is _ anchored _ at @xmath236 , ( ii ) @xmath234 is empty of points from @xmath20 in its interior , ( iii ) @xmath234 is interior - disjoint from all squares in @xmath237 , and ( iv ) @xmath234 has maximum area . then choose a largest candidate square @xmath234 , and a corresponding point @xmath236 , and set @xmath238 and @xmath239 . when @xmath240 , return the set of squares @xmath237 . [ [ remark.-1 ] ] remark . + + + + + + + let @xmath241 denote the approximation ratio of algorithm [ alg : sq ] , if it exists . the construction in fig . [ fig : rho2](a b ) shows that @xmath242 . for a small @xmath243 , consider the point set @xmath77 , where @xmath244 , @xmath245 , @xmath246 , and the rest of the points lie on the lower side of @xmath19 in the vicinity of @xmath101 , , @xmath247 and @xmath248 for @xmath249 . the packing generated by algorithm [ alg : sq ] consists of a single square of area @xmath250 , as shown in fig . [ fig : rho2](a ) , while the packing in fig . [ fig : rho2](b ) has an area larger than @xmath251 . by letting @xmath252 be arbitrarily small , we deduce that @xmath242 . [ fig : rho2 ] we first show that algorithm [ alg : sq ] achieves a ratio of @xmath253 ( theorem [ thm:1/6 ] ) using a charging scheme that compares the greedy packing with an optimal packing . we then refine our analysis and sharpen the approximation ratio to @xmath254 ( theorem [ thm:9/47 ] ) . [ [ charging - scheme - for - the - analysis - of - algorithmalgsq . ] ] charging scheme for the analysis of algorithm [ alg : sq ] . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + label the points in @xmath18 and the squares in @xmath255 in the order in which they are processed by algorithm [ alg : sq ] with @xmath256 and @xmath257 . let @xmath258 be the area of the greedy packing , and let @xmath259 denote an optimal packing with @xmath260 , where @xmath261 is the square anchored at @xmath3 . we employ a charging scheme , where we distribute the area of every optimal square @xmath261 with @xmath262 among some greedy squares ; and then show that the total area of the optimal squares charged to each greedy square @xmath263 is at most @xmath264 for all @xmath265 . ( degenerate optimal squares , , those with @xmath266 do not need to be charged ) . for each step @xmath267 of algorithm [ alg : sq ] , we shrink some of the squares @xmath268 , and charge the area - decrease to the greedy square @xmath263 . by the end ( after the @xmath6th step ) , each of the squares @xmath269 will be reduced to a single point . specifically in step @xmath144 , algorithm [ alg : sq ] chooses a square @xmath263 , and : ( 1 ) we shrink square @xmath270 to a single point ; and ( 2 ) we shrink every square @xmath261 , @xmath271 that intersects @xmath263 in its interior until it no longer does so . this procedure ensures that no square @xmath261 , with @xmath272 , intersects @xmath263 in its interior in step @xmath144 . refer to fig . [ fig : alg - sq](c ) . observe three important properties of the above iterative process : * after step @xmath144 , the squares @xmath273 have pairwise disjoint interiors . * after step @xmath144 , we have @xmath274 ( since @xmath270 was shrunk to a single point ) . * at the beginning of step @xmath144 , if @xmath261 intersects @xmath263 in its interior ( and so @xmath275 ) , then @xmath276 since @xmath263 is feasible for @xmath3 when @xmath270 is selected by algorithm [ alg : sq ] due to the greedy choice . [ lem : ratio ] suppose there exists a constant @xmath277 such that for every @xmath278 , square @xmath263 receives a charge of at most @xmath279 . then algorithm [ alg : sq ] computes an anchored square packing whose area @xmath280 is at least @xmath281 times the optimal . overall , each square @xmath263 receives a charge of at most @xmath282 from the squares in an optimal solution . consequently , @xmath283 and thus @xmath284 , as claimed . in the remainder of this section , we bound the charge received by one square @xmath263 , for @xmath285 . we distinguish two types of squares @xmath261 , @xmath271 , whose area is reduced by @xmath263 : * @xmath286 , the area of @xmath261 is reduced by @xmath263 , and @xmath287 contains no corner of @xmath288 , * @xmath289 , the area of @xmath261 is reduced by @xmath263 , and @xmath287 contains a corner of @xmath288 . it is clear that if the insertion of @xmath263 reduces the area of @xmath261 , @xmath271 , then @xmath261 is in either @xmath290 or @xmath291 . note that the area of @xmath270 is also reduced to 0 , but it is in neither @xmath290 nor @xmath291 . [ lem:6 ] each square @xmath263 receives a charge of at most @xmath264 . consider the squares in @xmath290 . assume that @xmath261 intersects the interior of @xmath263 , and it is shrunk to @xmath292 . the area - decrease @xmath293 is an l - shaped region , at least half of which lies inside @xmath263 ; see fig . [ fig : alg - sq ] . by property ( i ) , the l - shaped regions are pairwise interior - disjoint ; and hence the sum of their areas is at most @xmath294 . consequently , the area - decrease caused by @xmath263 in squares in @xmath290 is at most @xmath294 . consider the squares in @xmath291 . there are at most three squares @xmath261 , @xmath271 , that can contain a corner of @xmath263 since the anchor of @xmath263 is not contained in the interior of any square @xmath261 . since the area of each square in @xmath291 is at most @xmath295 by property ( iii ) , the area decrease is at most @xmath296 , and so is the charge received by @xmath263 from squares . finally , @xmath297 by property ( iii ) , and this is also charged to @xmath263 . overall @xmath263 receives a charge of at most @xmath264 . the combination of lemmas [ lem : ratio ] and [ lem:6 ] readily implies the following . [ thm:1/6 ] algorithm [ alg : sq ] computes an anchored square packing whose area is at least @xmath253 times the optimal . [ [ refined - analysis - of - the - charging - scheme . ] ] refined analysis of the charging scheme . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we next improve the upper bound for the charge received by @xmath263 ; we assume for convenience that @xmath298 ^ 2 $ ] . for the analysis , we use only a few properties of the optimal solution . specifically , assume that @xmath299 are interior - disjoint squares such that each @xmath261 : ( a ) intersects the interior of @xmath263 ; ( b ) has at least a corner in the exterior of @xmath263 ; ( c ) does not contain @xmath300 in its interior ; and ( d ) @xmath276 . the intersection of any square @xmath261 with @xmath178 is a polygonal line on the boundary @xmath178 , consisting of one or two segments . since the squares @xmath261 form a packing , these intersections are interior - disjoint . let @xmath301 denote the maximum area - decrease of a set of squares @xmath261 in @xmath290 , whose intersections with @xmath178 have total length @xmath69 . similarly , let @xmath302 denote the maximum area - decrease of a set of squares @xmath261 in @xmath291 , whose intersections with @xmath178 have total length @xmath69 . by adding suitable squares to @xmath290 , we can assume that @xmath303 is the total length of the intersections @xmath304 over squares in @xmath291 ( , the squares in @xmath305 cover the entire boundary of @xmath19 ) . consequently , the maximum total area - decrease is given by @xmath306 we now derive upper bounds for @xmath301 and @xmath302 independently , and then combine these bounds to optimize @xmath307 . since the total perimeter of @xmath19 is 4 , the domain of @xmath307 is @xmath308 . [ lem : delta1 ] the following inequalities hold : @xmath309 inequality was explained in the proof of theorem [ thm:1/6 ] . inequalities and follow from the fact that the side - length of each square @xmath261 is at most @xmath310 and from the fact that the area - decrease is at most the area ( of respective squares ) ; in addition , we use the inequality @xmath311 , for @xmath312 and @xmath313 , and the inequality @xmath314 , for @xmath315 , and @xmath316 . write @xmath317 where @xmath318 and @xmath319 denote the maximum area - decrease contained in @xmath19 and the complement of @xmath19 , respectively , of a set of squares in @xmath290 whose intersections with @xmath178 have total length @xmath69 , where @xmath320 . obviously , @xmath321 . we next show that @xmath322 and thereby establish inequality . consider a square @xmath261 of side - length @xmath323 in @xmath290 . let @xmath324 denote the length of the shorter side of the rectangle @xmath325 . the area - decrease outside @xmath19 equals @xmath326 and so it is bounded from above by @xmath327 ( equality is attained when @xmath328 ) . consequently , @xmath329 where the last equality follows from a standard weight - shifting argument , and equality is attained when @xmath69 is subdivided into @xmath330 unit length intervals and a remaining shorter interval of length @xmath331 . let @xmath332 be the number of squares @xmath261 in @xmath291 , where @xmath271 . we can assume that exactly @xmath189 squares @xmath261 , with @xmath333 , are in @xmath291 , one for each corner except the lower - left anchor corner of @xmath19 , that is , @xmath334 ; otherwise the proof of lemma [ lem:6 ] already yields an approximation ratio of @xmath335 . clearly , we have @xmath336 , for any @xmath69 . we first bring the squares in @xmath291 into _ canonical position _ : @xmath69 monotonically decreases , @xmath307 does not decrease , and properties ( a d ) listed earlier are maintained . specifically , we transform each square @xmath337 as follows ( refer to fig . [ fig : f7 ] ) : * move the anchor of @xmath261 to another corner if necessary so that one of its coordinates is contained in the interval @xmath338 ; * translate @xmath261 horizontally or vertically so that @xmath339 decreases to a skinny rectangle of width @xmath252 , for some small @xmath243 . = 4.7 in [ lem : delta2 ] the following inequality holds : @xmath340 assume that the squares in @xmath291 are in canonical position . let @xmath88 denote the side - length of @xmath261 , let @xmath86 denote the length of the longer side of the rectangle @xmath341 and @xmath324 denote the length of the shorter side of the rectangle @xmath325 , @xmath342 . since the squares in @xmath291 are in canonical position , we have @xmath343 , for @xmath342 . we also have @xmath344 . letting @xmath345 , we have @xmath346 . @xmath347 observe that the inequality @xmath348 , for every @xmath320 , is implied by the inequality . putting together the upper bounds in lemmas [ lem : delta1 ] and [ lem : delta2 ] yields lemma [ lem : delta ] : [ lem : delta ] the following inequality holds : @xmath349 from the opposite direction , @xmath350 holds even in a geometric setting , , as implied by several constructions with squares . the lower bound @xmath350 is implied by any of the two ( obviously different ! ) configurations shown in fig . [ fig : f8 ] . we now prove the upper bound . recall ; and by we obtain @xmath351 we partition the interval @xmath352 $ ] into several subintervals , and select the best upper bound we have in each case . we distinguish the following cases ( marked with @xmath353 ) : @xmath353 @xmath354 : by , @xmath355 ; since @xmath356 , we immediately get that @xmath357 @xmath353 @xmath358 : put @xmath359 . on this interval we have @xmath360 and so we can use the upper bound in on @xmath301 and obtain @xmath361 @xmath353 @xmath362 : put @xmath363 . on this interval we have @xmath364 and so we can use the upper bound in on @xmath301 and obtain @xmath365 @xmath353 @xmath366 : put @xmath367 . using the upper bound in on @xmath301 yields @xmath368 @xmath353 @xmath369 : put @xmath370 . using the upper bound in on @xmath301 yields @xmath371 @xmath353 @xmath372 : by , @xmath373 . all cases have been checked , and so the proof of the upper bound is complete . [ lem:47/9 ] each square @xmath263 receives a charge of at most @xmath374 . by lemma [ lem : delta ] , the area - decrease is at most @xmath375 , and so is the charge received by @xmath263 from squares in @xmath290 and from squares in @xmath291 with the exception of the case @xmath376 . adding this last charge yields a total charge of at most @xmath377 . the combination of lemmas [ lem : ratio ] and [ lem:47/9 ] now yields the following . [ thm:9/47 ] algorithm [ alg : sq ] computes an anchored square packing whose area is at least @xmath14 times the optimal . the following greedy algorithm , analogous to algorithm [ alg : sq ] , constructs a lower - left anchored square packing , given a finite point set @xmath229 ^ 2 $ ] . [ alg : llsq]set @xmath230 and @xmath231 . while @xmath232 , repeat the following . for each point @xmath233 , compute a _ candidate _ square @xmath234 such that ( i ) @xmath235 ^ 2 $ ] has @xmath236 as its _ lower - left _ anchor , ( ii ) @xmath234 is empty of points from @xmath20 in its interior , ( iii ) @xmath234 is interior - disjoint from all squares in @xmath237 , and ( iv ) @xmath234 has maximum area . then choose a largest candidate square @xmath234 , and a corresponding point @xmath236 , and set @xmath238 and @xmath239 . when @xmath240 , return the set of squares @xmath237 . [ [ remark.-2 ] ] remark . + + + + + + + let @xmath378 denote the approximation ratio of algorithm [ alg : llsq ] . the construction in fig . [ fig : greedy - lowerleft ] shows that @xmath379 . specifically , for @xmath243 , with @xmath380 , consider the point set @xmath381 . then the area of the packing in fig . [ fig : greedy - lowerleft](right ) is @xmath382 , but algorithm [ alg : llsq ] returns the packing shown in fig . [ fig : greedy - lowerleft](left ) of area @xmath383 . we next demonstrate that algorithm [ alg : llsq ] achieves approximation ratio @xmath15 . according to the above example , this is the best possible for this algorithm . [ thm:1/3 ] algorithm [ alg : llsq ] computes a lower - left anchored square packing whose area is at least @xmath15 times the optimal . label the points in @xmath18 and the squares in @xmath255 in the order in which they are processed by algorithm [ alg : llsq ] with @xmath256 and @xmath257 . let @xmath258 be the area of the greedy packing , and let @xmath259 denote an optimal packing with @xmath260 , where @xmath261 is the square anchored at @xmath3 . we charge the area of every optimal square @xmath261 to one or two greedy squares @xmath384 ; and then show that the total area charged to @xmath384 is at most @xmath385 for all @xmath386 . consider a square @xmath261 , @xmath387 , with @xmath262 . let @xmath388 be the minimum index such that @xmath263 intersects the interior of @xmath261 . let @xmath389 denote the candidate square associated to @xmath3 in step @xmath390 of algorithm [ alg : llsq ] . note that @xmath391 , thus @xmath392 . if @xmath393 , then let @xmath394 be the minimum index such that @xmath395 intersects the interior of @xmath389 . we can now describe our _ charging scheme _ : if @xmath261 contains the upper - left or lower - right corner of @xmath263 , then charge @xmath396 to @xmath263 ( fig . [ fig : ab ] , left ) . otherwise , charge @xmath397 to @xmath263 , and charge @xmath398 to @xmath395 ( fig . [ fig : ab ] , middle - right ) . we first argue that the charging scheme is well - defined , and the total area of @xmath261 is charged to one or two squares ( @xmath263 and possibly @xmath395 ) . indeed , if no square @xmath384 , @xmath399 , intersects the interior of @xmath261 , then @xmath400 , and @xmath401 ; and if @xmath402 and no square @xmath384 , @xmath403 , intersects the interior of @xmath389 , then @xmath404 and @xmath405 . = 2.5 in note that if @xmath396 is charged to @xmath263 , then @xmath276 . indeed , if @xmath406 , then @xmath261 is entirely free at step @xmath144 , so algorithm [ alg : llsq ] would choose a square at least as large as @xmath261 instead of @xmath263 , which is a contradiction . analogously , if @xmath398 is charged to @xmath395 , then @xmath407 . moreover , if @xmath398 is charged to @xmath395 , then the upper - left or lower - right corner of @xmath395 is on the boundary of @xmath389 , and so this corner is contained in @xmath261 ; refer to fig . [ fig : ab](right ) . fix @xmath408 . we show that the total area charged to @xmath384 is at most @xmath385 . if a square @xmath261 , @xmath22 , sends a positive charge to @xmath384 , then @xmath409 or @xmath410 . we distinguish two types of squares @xmath261 that send a positive charge to @xmath384 ; refer to fig . [ f5 + ] : 1 . @xmath261 contains the upper - left or lower - right corner of @xmath384 in its interior . @xmath261 contains neither the upper - left nor the lower - right corner of @xmath384 . since @xmath259 is a packing , at most one optimal square contains each corner of @xmath384 . consequently , there is at most two squares @xmath261 of type * t1. since @xmath411 , the charge received from the squares of type t1 * is at most @xmath412 . the following technical lemma is used for controlling the charge that a greedy square @xmath413 receives from squares of an optimal solution . [ lem : sides ] let @xmath413 be an axis - aligned square , and @xmath414 be interior - disjoint axis - aligned squares such that * each @xmath261 intersects the bottom or the left side of a square @xmath413 , but * the interior of @xmath261 does not contain any corner of @xmath413 . for @xmath415 , let @xmath389 be a maximal square contained in @xmath416 . then @xmath417 . for @xmath418 , let @xmath419 denote the side - length of @xmath261 , and define the _ depth _ @xmath420 of @xmath261 as the length of a shortest side of @xmath421 . with this notation , we have @xmath422 , the area of the @xmath423-shaped region @xmath424 is @xmath425 we apply a sequence of operations on the squares @xmath299 that maintain properties * p1 * and * p2 * , and monotonically increase @xmath426 . the operations successively transform the squares , eventually eliminate @xmath427 squares , and ensure that the last surviving square is @xmath428 . this implies @xmath429 , as required . = 6.3 in * operation 1 . * translate @xmath261 horizontally or vertically so that its depth increases until @xmath430 or @xmath261 is blocked by some other square . see fig . [ fig : op12 ] . perform operation 1 successively for every square @xmath261 in an arbitrary order . the operation can only increase the contribution @xmath431 of @xmath261 since @xmath419 is fixed and the depth @xmath420 can only increase . * operation 2 . * translate @xmath261 horizontally or vertically so that its depth remains fixed but @xmath261 moves closer to the lower - left corner @xmath172 until it reaches @xmath172 or is blocked by another square . see fig . [ fig : op12 ] . perform operation 2 successively for every square @xmath261 in the order determined by their distance from @xmath172 . this operation does not change the contribution of the squares @xmath261 . * operation 3 . * if @xmath261 intersects the bottom ( left ) side of @xmath413 and its right ( top ) side is not in contact with any other square or a corner of @xmath413 , then dilate @xmath261 from its upper - left ( lower - right ) corner until it is blocked by another square or the boundary of @xmath413 . see fig . [ fig : op12](right ) . since operation 3 increases @xmath419 and keeps @xmath420 fixed , the contribution @xmath432 of @xmath261 increases . note that if the lower - left corner of @xmath261 is @xmath172 originally and operation 3 is applied to @xmath261 , then its lower - left corner may move to the exterior of @xmath413 . = 6.3 in * operation 4 . * consider two squares , @xmath261 and @xmath270 , such that both intersect the bottom side of @xmath413 and they are adjacent . without loss of generality , assume @xmath261 is left of @xmath270 and @xmath433 . see fig . [ fig : op4 ] . we wish to increase @xmath419 and decrease @xmath434 while @xmath435 is fixed , and such that their depths does not increase . we distinguish two cases to ensure that @xmath261 intersects the bottom side of @xmath413 after the operation . if @xmath436 , then dilate @xmath261 from its upper - left corner and @xmath270 from its upper - right corner simultaneously until @xmath437 or @xmath261 is blocked by some other square or a corner of @xmath413 . in this case , @xmath420 and @xmath438 remain constant , and the contribution of @xmath261 and @xmath270 increases or remains the same . indeed , if @xmath419 increases and @xmath434 decreases by @xmath243 , then @xmath439 if @xmath437 , then dilate @xmath261 from its upper - left corner and @xmath270 from its lower - right corner simultaneously until @xmath440 or @xmath261 is blocked by some other square or a corner of @xmath413 . in this case , @xmath420 is constant , and @xmath437 decreases . the contribution of @xmath261 and @xmath270 increases or remains the same : @xmath441 we apply analogous operations to adjacent squares intersecting the left side of @xmath413 . = 4.5 in * operation 5 . * consider two adjacent squares , @xmath261 and @xmath270 , such that @xmath261 is incident to the lower - left corner of @xmath413 and intersects the bottom ( resp . , left ) side of @xmath413 , and @xmath270 intersects the left ( resp . , bottom ) side of @xmath413 . see fig . [ fig : op5 ] . note that @xmath442 , otherwise operation 2 could increase @xmath419 ; and if @xmath443 , then @xmath444 , otherwise operation 1 could increase @xmath438 . let @xmath123 be the axis - aligned bounding box of @xmath445 , which is incident to the lower - left corner of @xmath172 and @xmath446 . we replace @xmath261 and @xmath270 with one new square @xmath447 such that @xmath448 . it remains to show that operation 5 increases the total contribution of the squares . we distinguish two cases depending on which side of @xmath123 is longer . _ case 1 . _ if @xmath449 , then @xmath447 has side - length @xmath450 and depth @xmath451 . in case @xmath452 ( fig . [ fig : op4 ] , left ) , we have @xmath453 otherwise we have @xmath454 and @xmath437 ( fig . [ fig : op4 ] , right ) . thus , @xmath455 _ case 2 . _ if @xmath456 , then @xmath447 has side - length @xmath457 and depth @xmath458 . in this case , we have @xmath452 , otherwise operation 1 could increase @xmath438 . thus , @xmath459 after successively applying operation 5 , we obtain a single square @xmath428 , whose contribution is @xmath460 , as required . by lemma [ lem : sides ] , @xmath384 receives a charge of at most @xmath461 from squares of type * t2. it follows that each @xmath384 received a charge of at most @xmath385 . consequently , @xmath462 and thus @xmath463 . this completes the proof of theorem [ thm:1/3 ] . in this section , we show that there is a ptas ( resp . , qptas ) for the maximum anchored square ( resp . , rectangle ) problem using reductions to the minimum weight independent set ( mwis ) problem for axis - aligned squares ( resp . , rectangles ) in the plane . the current best approximation ratio for mwis with axis - aligned rectangles is @xmath464 @xcite , and @xmath465 in the unweighted case @xcite . there is a ptas for mwis with axis - aligned squares @xcite ( even a local search strategy works in the unweighted case @xcite ) ; and there is a qptas for mwis with axis - aligned rectangles @xcite . specifically , for @xmath6 axis - aligned squares an @xmath57-approximation for mwis which can be computed in time @xmath16 @xcite , and for @xmath6 axis - aligned rectangles in time @xmath466 @xcite . an anchored rectangle can be also viewed from the perspective of robotics applications such as _ a rectangular robotic arm _ anchored at the given point . in this context , an anchored packing of maximum total area represents the maximum collective _ reach _ of a collection of nonoverlapping rectangular robotic arms . the maximum area of a lower - left anchored rectangle ( resp . , square ) packing can be arbitrarily small , , with no constant lower - bound guarantee , when the points in @xmath20 are close to the top or right boundary of @xmath1 ^ 2 $ ] , or for square packings , when @xmath6 is large and the points are suitably placed , , along the diagonal of unit slope . however , we show below that a greedy packing still covers a constant fraction of the region that could potentially be covered by lower - left anchored rectangles ( resp . , squares ) . we then employ this lower bound in the design of a qptas ( resp . , ptas ) for lower - left anchored rectangles ( resp . , squares ) in section [ ssec : approx - schemes ] . [ [ the - reach - of - lower - left - anchored - rectangles . ] ] the reach of lower - left anchored rectangles . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for a set @xmath20 of @xmath6 points in @xmath19 , define the _ reach of lower - left anchored rectangles _ as @xmath467 , where @xmath468 is the axis - aligned rectangle spanned by @xmath172 and the upper - right corner of @xmath1 ^ 2 $ ] . clearly , @xmath469 is a simple orthogonal polygon in which every lower - left corner is an element in @xmath20 . vertical lines through the lower - left corners of @xmath469 subdivide @xmath469 into axis - aligned rectangles whose lower left - corners are in @xmath20 . each of these rectangles @xmath39 admits a lower - left anchored rectangle packing of area at least @xmath470 by the main result in @xcite , consequently @xmath471 . [ [ the - reach - of - lower - left - anchored - squares . ] ] the reach of lower - left anchored squares . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + similarly , we define the _ reach of lower - left anchored squares _ as @xmath472 , where @xmath473 is the ( unique ) maximal square contained in @xmath19 whose lower - left corner is @xmath172 and that is empty of points of @xmath20 in its interior . observe that @xmath474 is the union of candidate squares in the first iteration of algorithm [ alg : llsq ] . [ lem:3s - ll ] let @xmath475 be a point set with a lower - left anchored square packing @xmath476 returned by algorithm [ alg : llsq ] . then @xmath477 . we need to show that every point @xmath478 lies in @xmath479 . this claim clearly holds when @xmath236 is covered by one of the squares @xmath480 . let @xmath481 be a point left uncovered by the greedy square packing . by the definition of @xmath474 , point @xmath236 lies in some lower - left anchored square @xmath482 , @xmath483 , of maximum area contained in @xmath1 ^ 2 $ ] but empty of points of @xmath20 in its interior . note that the boundary of @xmath413 contains at least one point from @xmath20 , namely @xmath483 . let @xmath263 be the first square chosen by algorithm [ alg : llsq ] that intersects the interior of @xmath413 . we are guaranteed the existence of @xmath263 because if no previous greedy square intersects the interior of @xmath413 , then @xmath484 ( possibly @xmath376 ) . at the beginning of the step @xmath144 of algorithm [ alg : llsq ] , no square in @xmath237 intersects @xmath413 . by the greedy choice , we have @xmath485 . since @xmath486 and the side - length of @xmath413 is less than or equal to the side - length of @xmath263 , we have @xmath487 . [ thm:1/9-ll ] for every finite point set @xmath20 in @xmath1 ^ 2 $ ] , there is a lower - left anchored square packing of total area at least @xmath488 . given a point set @xmath489 ^ 2 $ ] and a greedy lower - left anchored square packing @xmath490 , lemma [ lem:3s - ll ] yields @xmath491 which implies @xmath492 , as claimed . [ [ the - reach - of - anchored - squares . ] ] the reach of anchored squares . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for a finite ( nonempty ) set @xmath229 ^ 2 $ ] , the _ reach of anchored squares _ @xmath493 is the union of all maximal squares anchored at a point in @xmath20 and contained in @xmath19 . obviously , we have @xmath494 by theorem [ thm:5/32 ] , and we suspect that @xmath495 for every @xmath20 . if true , this bound would be the best possible : choose a small @xmath243 and place all @xmath6 points in the @xmath252-neighborhood of @xmath496 on the lower side of @xmath19 . the side - length of any anchored square is at most @xmath497 , which is an upper bound for the distance of these points from the left and right side of @xmath1 ^ 2 $ ] . consequently all anchored squares lie below the horizontal line @xmath498 , and so the area of their reach is at most @xmath497 . [ [ anchored - rectangles . ] ] anchored rectangles . + + + + + + + + + + + + + + + + + + + + the anchored rectangle packing problem can be discretized @xcite , as one can specify a set of @xmath499 anchored rectangles that contains the optimum packing : consider the hanan grid @xcite induced by the points , , the union of @xmath178 and the @xmath500 unit horizontal and vertical unit segments incident to the @xmath6 points and contained in @xmath19 . then an optimal rectangle packing can be found in this grid ( with at most @xmath501 candidates for each anchor ) . to formulate the anchored rectangle packing problem as a maximum weight independent set ( mwis ) problem , the set of rectangles anchored at the same point are slightly enlarged so that they contain the common anchors in their interiors . for a set of points in general position , this procedure yields an qptas for finding a maximum area anchored rectangle packing for a set of @xmath6 points in @xmath19 . [ [ anchored - squares . ] ] anchored squares . + + + + + + + + + + + + + + + + + it is worth noting that no similar discretization is known for the anchored square packing problem . with an @xmath252-factor loss of the total area , however , we can construct @xmath502 squares for which mwis gives an anchored packing of area at least @xmath503 . after a similar transformation ensuring that the set of squares anchored at the same point contain the point in their interiors , and under the assumption of points in general position , we obtain a ptas for the maximum - area anchored square packing problem . [ thm : ptas - sq ] there is a ptas for the maximum area anchored square problem . for every @xmath243 , there is an algorithm that computes , for a set of @xmath6 points in general position in @xmath1 ^ 2 $ ] , an anchored square packing of area at least @xmath503 in time @xmath16 . first we show that the problem can be discretized if we are willing to lose a factor @xmath243 of the maximum area of a packing . let @xmath20 be a set of @xmath6 points in @xmath1 ^ 2 $ ] , and let @xmath259 be an anchored square packing of maximum area . for a given @xmath243 , drop all squares in @xmath259 of area less than @xmath504 , and shrink all remaining squares in @xmath259 so that its anchor remains the same , and its area decreases by a factor of at most @xmath505 to @xmath506 for some integer @xmath507 . denote by @xmath508 the resulting anchored square packing . we have dropped small squares of total area at most @xmath252 , and the area of any remaining square decreased by a factor of at most @xmath505 . consequently , @xmath509 . for anchored square packings , we have @xmath510 by theorem [ thm:5/32 ] , which yields @xmath511 let @xmath243 be given . for every set @xmath20 of @xmath6 points in general position in @xmath1 ^ 2 $ ] , we compute a set @xmath512 of @xmath513 _ candidate _ squares , and show that a mwis of @xmath512 ( where the weight of each square is its area ) has area at least @xmath514 . for each point @xmath171 , consider a set @xmath515 of empty squares anchored at @xmath413 and contained in @xmath19 whose areas are of the form @xmath516 , for @xmath517 . translate each square in @xmath515 , for all @xmath171 , by a sufficiently small @xmath518 so that each square contains its anchor in its interior , but does not cross any new horizontal or vertical line through the points in @xmath20 ; denote by @xmath519 the resulting set of squares . let @xmath520 . note that the squares in @xmath519 pairwise intersect for each @xmath171 , hence an mwis contains precisely one square ( possibly of 0 area ) for each point in @xmath20 . the number of squares is @xmath521 by , the area of a mwis for the candidate squares in @xmath512 is at least @xmath514 . using the ptas for mwis with axis - aligned squares due to chan @xcite : an @xmath57-approximation for a set of @xmath522 weighted axis - aligned squares can be computed in time @xmath523 . [ [ lower - left - anchored - packings . ] ] lower - left anchored packings . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + even though the area of _ lower - left _ anchored square packings is not bounded by any constant , and theorem [ thm:5/32 ] no longer holds , the ptas in theorem [ thm : ptas - sq ] can be extended for this variant . recall that for a set @xmath229 ^ 2 $ ] , the reach of lower - left anchored squares , denoted @xmath474 , is the union of all maximal squares in @xmath1 ^ 2 $ ] whose lower - left corner is in @xmath20 , and are empty from points of @xmath20 in their interior . it is clear that @xmath474 contains all squares of a lower - left anchored square packing for @xmath20 , and we have @xmath524 by theorem [ thm:1/9-ll ] . by adjusting the resolution of the candidate squares , we obtain a ptas for the maximum area lower - left anchored square packing problem . similarly , we obtain a qptas for the maximum area lower - left anchored rectangle problem . consequently , for @xmath6 anchors , an @xmath57-approximation for the lower - left anchored square packing can be computed in time @xmath16 ; and for the lower - left anchored rectangle packing in time @xmath466 . we conclude with a few open problems : 1 . is the problem of computing the maximum - area anchored rectangle ( respectively , square ) packing np - hard ? 2 . is there a polynomial - time approximation scheme for the problem of computing an anchored rectangle packing of maximum area ? 3 . what lower bound on @xmath525 can be obtained by extending lemma [ lem : rectangle ] concerning rectangles from @xmath121 to @xmath189 points ? is there a short proof of lemma [ lem : rectangle ] ? 4 . does algorithm [ alg : sq ] for computing an anchored square packing of maximum area achieve a ratio of @xmath526 ? by theorem [ thm:9/47 ] and the construction in fig . [ fig : alg - sq ] , the approximation ratio is between @xmath55 and @xmath526 . improvements beyond the @xmath335 ratio are particularly exciting . 5 . is the reach of anchored squares always at least @xmath59 ? ( does @xmath527 hold for every nonempty set @xmath20 ? ) 6 . is @xmath528 ? is @xmath529 ? what upper and lower bounds on @xmath525 and @xmath530 can be established in higher dimensions a natural variant of anchored squares is one where the anchors must be the centers of the squares . what approximation can be obtained in this case ? [ [ acknowledgment . ] ] acknowledgment . + + + + + + + + + + + + + + + the authors are thankful to ren sitters for constructive feedback on the problems discussed in this paper . in particular , the preliminary approximation ratio @xmath253 in theorem [ thm:1/6 ] incorporates one of his ideas . 99 a. k. abu - affash , p. carmi , m. j. katz , and g. morgenstern , multi cover of a polygon minimizing the sum of areas , _ int . j. comput . geometry & appl . _ 21(6 ) * ( 2011 ) , 685698 . a. adamaszek and a. wiese , approximation schemes for maximum weight independent set of rectangles , in _ proc . 54th sympos . on found . of comp . _ , ieee , 2013 . a. adamaszek and a. wiese , a quasi - ptas for the two - dimensional geometric knapsack problem , in _ proc . 26th acm - siam sympos . discrete algor . _ , siam , 2015 . k. agarwal and n. h. mustafa , independent set of intersection graphs of convex objects in 2d , _ comput . _ * 34(2 ) * ( 2006 ) , 8395 . a. aggarwal and s. suri , fast algorithms for computing the largest empty rectangle , in : _ proc . comput . geometry _ , acm , 1987 , pp . 278290 . m. ajtai , the solution of a problem of t. rado , _ bulletin de lacadmie polonaise des sciences , srie des sciences mathmatiques , astronomiques et physiques _ * 21 * ( 1973 ) , 6163 . s. aluru , quadtrees and octrees , ch . 19 in _ handbook of data structures and applications ( d. p. mehta and s. sahni , editors ) _ , chapman & hall / crc , 2005 . k. balas and cs . d. tth , on the number of anchored rectangle packings for a planar point set , in _ proc . _ , springer , 2015 , pp . 377389 . n. bansal and a. khan , improved approximation algorithm for two - dimensional bin packing , in _ proc . 25th acm - siam sympos . discrete algor . _ , siam , 2014 , pp . 1325 . s. bereg , a. dumitrescu , and m. jiang , maximum area independent set in disk intersection graphs , _ internat . j. comput . * 20 * ( 2010 ) , 105118 . s. bereg , a. dumitrescu and m. jiang , on covering problems of rado , _ algorithmica _ * 57 * ( 2010 ) , 538561 . s. bhowmick , k. varadarajan , and s .- k . xue , a constant - factor approximation for multi - covering with disks , in _ proc . 29th acm sympos . on comput . _ , acm , 2013 , pp . 243248 . s. bhowmick , k. varadarajan , and s .- k . xue , addendum to `` a constant - factor approximation for multi - covering with disks '' , manuscript , 2014 , available at http://homepage.cs.uiowa.edu/~kvaradar/papers.html . v. blo , i. caragiannis , c. kaklamanis , and p. kanellopoulos , geometric clustering to minimize the sum of cluster sizes , in _ proc . european sympos . _ , lncs 3669 , 2005 , pp . 460471 . p. chalermsook and j. chuzhoy , maximum independent set of rectangles , in _ proc . 20th acm - siam sympos . discrete algor . _ , siam , 2009 , pp . . t. m. chan , polynomial - time approximation schemes for packing and piercing fat objects , _ j. algorithms _ * 46 * ( 2003 ) , 178189 . t. m. chan and s. har - peled , approximation algorithms for maximum independent set of pseudodisks , _ discrete comput . geom . _ * 48(2 ) * ( 2012 ) , 373392 . k. l. clarkson , fast algorithms for the all nearest neighbors problem , in _ proc . 24th annu . ieee sympos . found . _ , ieee , 1983 , pp . 226232 b. chazelle , r.l . drysdale , and d.t . lee , computing the largest empty rectangle , _ siam j. comput . * 15(1 ) * ( 1986 ) , 300315 . a. dumitrescu and m. jiang , on the largest empty axis - parallel box amidst @xmath6 points , _ algorithmica _ * 66(2 ) * ( 2013 ) , 225248 . a. dumitrescu and m. jiang , computational geometry column 56 , _ sigact news bulletin _ * 44(2 ) * ( 2013 ) , 8087 . a. dumitrescu and cs . d. tth , packing anchored rectangles , _ combinatorica _ * 35(1 ) * ( 2015 ) , 3961 . t. erlebach , k. jansen , and e. seidel , polynomial - time approximation schemes for geometric intersection graphs , _ siam j. comput . _ * 34(6 ) * ( 2006 ) , 13021323 . m. formann and f. wagner , a packing problem with applications to lettering of maps , in _ proc . 7th sympos . comput . geometry _ , 1991 , acm press , pp . 281288 . m. hanan , on steiner s problem with rectilinear distance , _ siam j. appl . math . _ * 14 * ( 1966 ) , 255265 . c. iturriaga and a. lubiw , elastic labels around the perimeter of a map , _ j. algorithms _ * 47(1 ) * ( 2003 ) , 1439 . jung and k .- y . chwa , labeling points with given rectangles , _ inf . process . lett . _ * 89(3 ) * ( 2004 ) , 115121 . k. g. kakoulis and i. g. tollis , labeling algorithms , ch . 28 in _ handbook of graph drawing and visualization ( r. tamassia , ed . ) _ , crc press , 2013 . d. knuth and a. raghunathan , the problem of compatible representatives , _ siam j. disc . _ * 5 * ( 1992 ) , 3647 . a. koike , s .- nakano , t. nishizeki , t. tokuyama , and s. watanabe , labeling points with rectangles of various shapes , _ int . j. comput . geometry appl . _ * 12(6 ) * ( 2002 ) , 511528 . m. van kreveld , t. strijk , and a. wolff , point labeling with sliding labels , _ comput . * 13 * ( 1999 ) , 2147 . r. rado , some covering theorems ( i ) , _ proc . soc . _ * 51 * ( 1949 ) , 232264 . r. rado , some covering theorems ( ii ) , _ proc . * 53 * ( 1951 ) , 243267 . r. rado , some covering theorems ( iii ) , _ proc . * 42 * ( 1968 ) , 127130 . t. rado , sur un problme relatif un thorme de vitali , _ fund . _ , * 11 * ( 1928 ) , 228229 . w. tutte , _ recent progress in combinatorics : proc . 3rd waterloo conf . _ , may 1968 , academic press , new york , 1969 . p. winkler , packing rectangles , in _ mathematical mind - benders _ peters ltd . , wellesley , ma , 2007 , pp . 133134 . b. zhu and m. jiang , a combinatorial theorem on labeling squares with points and its application , _ j. comb . _ * 11(4 ) * ( 2006 ) , 411420 . we have @xmath122 with @xmath531 for @xmath532 , and @xmath533 . we can assume that @xmath534 by applying a reflection with respect to a vertical line if necessary . consider the grid induced by the two points in @xmath1 ^ 2 $ ] , consisting of six rectangles @xmath4 , @xmath535 , as shown in fig . [ fig:8packings ] . put @xmath536 , @xmath535 . as a shorthand notation , a union of rectangles , @xmath537 is denoted @xmath538 , and @xmath539 denotes its area . it is not difficult to show that every maximal rectangle packing for @xmath122 is one of the eight packings in fig . [ fig:8packings ] ( although , we do not use this fact in the remainder of the proof ) . specifically , @xmath82 is a lower - left or lower - right corner of a rectangle . if @xmath82 is the lower - right corner of a rectangle of height 1 , then @xmath101 can be the anchor of any of @xmath540 , @xmath540 , @xmath541 , and @xmath542 , producing the packings @xmath543 , @xmath544 , @xmath545 , and @xmath546 . if @xmath82 is the lower - left corner of a rectangle of height 1 , then @xmath101 can be the anchor of @xmath547 or @xmath542 , producing @xmath548 and @xmath549 . if @xmath82 is the lower - left corner of a rectangle of height @xmath550 , then @xmath101 can be the anchor of @xmath551 or @xmath547 , which yields the packings @xmath552 and @xmath553 . if @xmath82 is the lower - right corner of a rectangle of height @xmath550 , then @xmath101 must be the anchor of @xmath554 , and we obtain @xmath543 again . = 0.7 the area of each of the eight packing in fig . [ fig:8packings ] is a quadratic function in the variables @xmath555 over the domain @xmath556 we wish to find the minimum @xmath557 of the upper envelope of these eight 3-variate functions over the domain @xmath558 . observe that @xmath559 in particular , if @xmath560 holds , the desired lower bound , @xmath115 , follows . for the case analysis below , we also use the following fact . consider the quadratic equation @xmath561 . since its roots are @xmath562 and @xmath563 , we have @xmath564 for @xmath565 $ ] . _ [ [ case - analysis . ] ] case analysis . + + + + + + + + + + + + + + we distinguish six cases covering the range of @xmath566 : @xmath567 . _ case 1 : _ @xmath568 . if @xmath569 then @xmath570 as required . if @xmath571 , we distinguish two subcases : if @xmath572 , then select rectangle @xmath547 anchored at @xmath101 , and one of @xmath573 and @xmath574 anchored at @xmath82 . if follows that @xmath575 as required . if @xmath576 , then @xmath577 as required . _ case 2 : _ @xmath578 . if @xmath579 , we have @xmath580 as required . if @xmath571 , we have @xmath581 and the lower bound follows from . if @xmath582 , we distinguish two subcases : _ case 2.1 : _ @xmath572 . if @xmath583 , then @xmath584 and the lower bound follows from ( due to fact 1 ) . if @xmath585 , then @xmath586 as required . the last inequality above is equivalent to @xmath587 which holds for @xmath588 . _ case 2.2 : _ @xmath576 . if @xmath589 , then @xmath590 and the lower bound follows from . if @xmath591 , then @xmath592 as required . indeed , the last inequality above is equivalent to @xmath593 , which holds since @xmath594 $ ] and according to fact 1 . _ case 3 : _ @xmath595 . if @xmath596 , then @xmath597 and the lower bound follows from . if @xmath579 , we distinguish two subcases : _ case 3.1 : _ @xmath598 . then @xmath599 as required . _ case 3.2 : _ @xmath600 . then @xmath601 as required . _ case 4 : _ @xmath602 . observe that @xmath603 , and consequently , @xmath604 , as used below in case 2.4.2 . if @xmath605 , then @xmath606 and the lower bound follows from . if @xmath607 , we distinguish two subcases : _ case 4.1 : _ @xmath608 . then @xmath609 as required . _ case 4.2 : _ @xmath610 . then @xmath611 as required . _ case 5 : _ @xmath612 . if @xmath613 , then @xmath614 and the lower bound follows from . if @xmath615 , we distinguish two subcases : if @xmath616 , then @xmath617 as required . if @xmath618 and @xmath619 , then @xmath620 as required . if @xmath618 and @xmath621 , then select @xmath622 anchored at @xmath101 and a zero - area rectangle anchored at @xmath82 : @xmath623 as required . _ case 6 : _ @xmath624 . select @xmath573 and the largest of the @xmath199 empty rectangles anchored at @xmath101 ; the total area is @xmath625 as required . this completes the case analysis and thereby the proof of the lower bound . to see that this lower bound is the best possible , put @xmath626 and @xmath627 and verify that no system of anchored rectangles covers an area larger than @xmath115 . the same general idea applies to all cases ( as well as the notation in the figures ) : two adjacent axis - aligned rectangles are specified , one empty of points and the other containing points , that allows the selection of an empty square of a suitable size anchored at an extreme point of the nonemepty rectangle . the relevant subcases are illustrated in the figures ( nonempty squares or subsquares are shaded ) and the compact arguments are summarized in the tables . any missing case is either similar or obtained by reflection of a present case with respect to a line . in each case : * we consider a quadtree subdivision up to a certain level ; * in nonempty squares in the subdivision ( shaded in the figures ) , some anchored square packing covers @xmath628 times the area of corresponding squares by induction ; * we find new anchored squares of area @xmath526 , @xmath629 , @xmath630 , or other , ( in bold lines ) , which partially cover some empty and nonempty squares of the subdivision ; * the new anchored squares supersede the previous packing ( , obtained by induction ) in the nonempty squares of the subdivision that they intersect . [ obs : adjacent2 ] let @xmath631 be two axis - aligned interior - disjoint rectangles of sizes @xmath632 and @xmath633 , respectively , where @xmath634 , sharing a common edge of length @xmath635 . if @xmath636 and @xmath637 , then @xmath638 contains an anchored empty square whose area is at least @xmath639 . assume that @xmath640 lies left of @xmath641 and let @xmath642 be the rightmost point in @xmath640 . if @xmath172 lies in the lower half - rectangle of @xmath169 then the square of side @xmath643 whose lower - left anchor is @xmath172 is empty and has area @xmath639 . similarly , if @xmath172 lies in the higher half - rectangle of @xmath640 then the square of side @xmath643 whose upper - left anchor is @xmath172 is empty and has area @xmath639 . to specify anchor points , we use the following notation . let @xmath644 . let @xmath645 and @xmath646 denote the leftmost and rightmost points of @xmath647 , respectively ( these points coincide if @xmath648 ) . let @xmath649 and @xmath650 denote the lowest and highest points of @xmath647 , respectively ( these points coincide if @xmath648 ) . we next proceed with the details . in cases 1 and 2 , subdivide @xmath226 , @xmath197 ( in case 1 , @xmath651 , only ) into four congruent squares @xmath652 , @xmath653 , labeled counterclockwise around the center of @xmath226 according to the subquadrant containing the square . then partition @xmath200 , @xmath197 into four subsets @xmath654 , @xmath655 , @xmath656 ( ties are broken as specified earlier ) . we distinguish several subcases depending on the number of empty sets @xmath654 , @xmath657 , @xmath658 , in the second level of the quadtree subdivision . let @xmath660 denote the number of empty @xmath661 . the relevant subcases are illustrated in fig . [ f10 ] and summarized in table [ tab : case1 ] . for example , in the first two rows in the table there are two inductive instances of area @xmath526 and two inductive instances of area @xmath629 , contributing @xmath662 and @xmath663 , respectively , toward the total covered area .	for points @xmath0 in the unit square @xmath1 ^ 2 $ ] , an _ anchored rectangle packing _ consists of interior - disjoint axis - aligned empty rectangles @xmath2 ^ 2 $ ] such that point @xmath3 is a corner of the rectangle @xmath4 ( that is , @xmath4 is _ anchored _ at @xmath3 ) for @xmath5 . we show that for every set of @xmath6 points in @xmath1 ^ 2 $ ] , there is an anchored rectangle packing of area at least @xmath7 , and for every @xmath8 , there are point sets for which the area of every anchored rectangle packing is at most @xmath9 . the maximum area of an anchored _ square _ packing is always at least @xmath10 and sometimes at most @xmath11 . the above constructive lower bounds immediately yield constant - factor approximations , of @xmath12 for rectangles and @xmath10 for squares , for computing anchored packings of maximum area in @xmath13 time . we prove that a simple greedy strategy achieves a @xmath14-approximation for anchored square packings , and @xmath15 for lower - left anchored square packings . reductions to maximum weight independent set ( mwis ) yield a qptas and a ptas for anchored rectangle and square packings in @xmath16 and @xmath17 time , respectively . * keywords * : rectangle packing , anchored rectangle , greedy algorithm , charging scheme , approximation algorithm .
solar flares are the most extravagant examples of rapid energy release in the solar system , with the largest releasing around @xmath2 on a timescale of hours @xcite . this energy , imparted to the plasma confined along coronal magnetic loops of active regions , is distributed between kinetic , thermal , and radiative process in some way that may vary from flare to flare . while the ultimate source of this energy is likely stresses introduced by convective motion of the plasma at and below the photosphere , we believe the direct source is the conversion of free magnetic energy : magnetic energy in excess of the active region s potential magnetic field energy . as has been clear for many decades , active regions consist of bundles of flux tubes , concentrated prior to their emergence through the photosphere @xcite . the free energy builds up as the flux tubes forming an active region are stressed at the photospheric boundary , where plasma forces dominate field evolution ( plasma @xmath3 , with @xmath4 the gas pressure ) . moving outward from the solar surface into the corona , the plasma pressure rapidly diminishes and magnetic forces dominate , until a third regime is reached where plasma forces once again dominate . as noted by @xcite , even within an active region , the high @xmath5 portion of the upper corona may occur as low as @xmath6 above the solar surface . we are primarily concerned with lower laying loops and magnetic domains and so will assume a low @xmath5 regime . barring any reconfiguration of the coronal field , the active region s magnetic domains are pushed into a highly nonpotential state . relaxation towards a potential field configuration through magnetic reconnection then allows for the conversion of magnetic free energy into kinetic and thermal energy through , e.g. , field line shortening , shock formation , electron acceleration , or ( possibly ) ion acceleration @xcite . the number of quantitative estimates of this energy buildup using observations has recently increased , but results remain varied . nonlinear force free models ( nlff : * ? ? ? * ; * ? ? ? * ) have received much attention during the past decade , strongly driven by both increases in computing power and the arrival of vector magnetograms from space based telescopes onboard hinode ( sot / sp ) and sdo ( hmi ) . while these models are a promising avenue of research , they come with their own set of problems , as discussed in @xcite . the lower boundary conditions are , in general , incompatible with the force free assumption @xcite . several methods exist to overcome this difficulty @xcite , leading to different energy estimates for a single vector magnetogram , even when using a single extrapolation code@xcite . a further problem is that the models amount to a series of independent fields at consecutive timesteps . at each time , a new nlff field is generated from the boundary data , uninformed by the solution from the previous timestep . contrasting with this are flux transport and magneto frictional models , which do include a memory @xcite . these methods primarily focus on the global coronal response to active region emergence , destabilization , and eruption as opposed to the detailed analysis of processes within an active region , which is the topic of this investigation @xcite . one reason for this is that the coronal portion of these models evolve the large scale mean field using an induction equation with an effective magnetic diffusivity @xcite , so that the formation of fine scale current sheets is beneath their resolution . most dynamical simulations without magnetic diffusion show a tendency toward fine layers @xcite . we describe the coronal field using the magnetic charge topology ( mct ) model @xcite at each time . the system is described by a set of unipolar regions . the distribution of magnetic flux between each pair of oppositely signed regions defines the system s connectivity . as the active region evolves its connectivity will generally change . to relate each time with the next , we employ the minimum current corona model ( mcc : * ? ? ? * ; * ? ? ? . by itself , mct describes only potential fields , which contain no current . the mcc introduces currents , and the resulting energetics , into the mct model by asserting that the coronal field move through a series of flux constrained equilibria ( fce ) . in that case , the connectivity of the real field will be different from the potential field s connectivity . one shortcoming of the mcc method as currently used is its inability to account for violation of these flux constraints , which are the topological manifestations of reconnection and the resulting energy release . previous studies @xcite have therefore only reported the total free energy difference between the mcc and a potential field configurations . our goal here is to relax those flux constraints at any timestep , while also allowing the system to continue evolving thereafter . in this way , we may model multiple reconnection events for a single active region . we present here a method for identifying the magnetic domains activated in successive flares based on observations of flare ribbons in the aia 1600 channel . this allows us to separately calculate the free energy available to each successive , nonhomologous flare . if we further assume that all magnetic flux topologically capable of transferring during a reconnection event does transfer , then we may also estimate the actual energy release during a flare . in the following sections , we will describe the data used ( [ sec : data ] ) , our methods for modeling the photospheric and coronal fields ( [ sec : model ] ) , how one may estimate the mcc free energy based on those models ( [ sec : energy ] ) , and the use of observations of flare ribbons to determine those domains activated in successive flares ( [ sec : ribbons ] ) . we will conclude with a discussion of the results of our analysis ( [ sec : disc ] ) . to construct the mcc model of magnetic field evolution we use a series of 250 line of sight ( los ) magnetograms taken by the helioseismic and magnetic imager ( hmi : * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) onboard the solar dynamics observatory ( sdo ) . the data are at a 24 minute cadence between feb . 11 2011 08:10:12 ut and feb . 15 11:46:12 ut , and are taken from the jsoc ( level @xmath7 ) data series . the region considered , noaa ar11158 , produced the first goes x class flare of solar cycle 24 , and has therefore already been analyzed in a variety of ways by numerous authors ( see * ? ? ? * and references therein ) . in addition we have used images of flare ribbons observed with atmospheric imaging assembly ( sdo / aia ) in the 1600 channel @xcite . we obtained three sets of 1600 images via the ssw cutout service maintained by lockheed martin for @xmath8 during each flare with peak magnitude greater the m1.0 : an m6.6 flare peaking at feb 13 , 17:28 ; m2.2 peaking at feb 14 , 17:20 ; and x2.2 flare peaking at feb 15 , 01:44 . all aia data were prepared to level @xmath7 using the standard routine in solarsoftwareidl . we coalign each set of aia 1600 images to the magnetogram closest to the peak time of each flare . to do so , we rotate the first aia image in a sequence to the time of the magnetogram . it so happens that the 75 g contour of hmi los magnetograms ( after assuming a radial field and correcting for pixel foreshortening , discussed below ) outlines the bright network patches in the 1600 band . we shift the aia image , by eye , until the contour and bright network patches align . this could be automated by a cross correlation between the magnetogram contour and a corresponding contour in aia , but we have not yet implemented this procedure . the aia timesets are internally aligned , so we apply the same by eye offset to each image in the sequence , after shifting each by solar rotation to the time of the chosen magnetogram . the hmi magnetograms contain known ( but as yet unmodeled ) diurnal variations , due to the velocity of the spacecraft s orbit @xcite . the amplitude of the variations is around @xmath9 of the unsigned flux within an active region . we are concerned with flux emergence trends over the course of days , over which time the effects of these variations should largely cancel . we simply accept this as an additional source of error in our model . we apply the methods described in @xcite and references therein to generate the magnetic modeling for this series of events . the analysis splits into two sections , detailed below . first , we characterize the photospheric field by partitioning the observed magnetograms into a set of unipolar regions . pairs of oppositely signed regions @xmath10 may be linked through emergence when each region s flux increases between two timesteps .. algorithmically , there is no distinction between these processes . if any of a pole s flux change between two timesteps can not be paired with another region , it is formally paired with a source of opposite sign located at infinity . ] at each time @xmath11 , the amount of flux change of each pairing is recorded in a photospheric field change matrix @xmath12 . this set of matrices is therefore a time history of the flux with which each region emerged with every other region . in the second part of our analysis , we develop a topological model of coronal domains immediately prior to each major flare . our flux emergence matrix discussed above is the real flux in each coronal domain , which we may compare to the flux in each domain in a potential field extrapolation . the difference between the two is the nonpotentiality of each domain . the equilibrium with minimum magnetic energy that still includes this difference in domain fluxes , called the flux constrained equilibrium ( fce ) , contains current sheets on each of its separators @xcite . our topological model determines the location of all current sheets within the active region complex , the strength of each related to the nonpotentiality of its associated domains . finally , this provides us with an estimate of the energy in the fce , which is itself a lower bound on the magnetic free energy stored in the actual magnetic field . to characterize the photospheric field , we first convert each line of sight ( los ) pixel to vertical by assuming a radial field and account for reduced flux due to pixel foreshortening dividing the flux in each pixel by @xmath13 , where @xmath14 is the polar angle from disk center . we smooth our data by extrapolating the vertical field to a height of @xmath15 as detailed in @xcite , and reduce noise by ignoring any pixels below a @xmath16 threshold . pixels above this threshold are partitioned using the downhill tessellation algorithm of @xcite , creating a _ mask_. each pixel is assigned an integer , and contiguous groups of like signed pixels of the same integer compose a _ region_. pixels below our threshold belong to no region and are assigned a mask value of 0 . this tessellation scheme can generate thousands of small regions at each timestep , so adjacent regions of like polarity are merged when the saddle point value of the magnetic field between them is less than @xmath17 . finally , we exclude any region whose total flux is less than @xmath18 . this process is carried out at each timestep , independently . we call the resulting set of masks a _ mask array_. regions in one timestep are associated with those in the surrounding timestep first by a bidirectional association between timesteps , and then applying the and algorithms , described in detail in @xcite . the focus of these methods is to distinguish between regions of flux that emerge from below the photosphere at different times and in different places , and to keep track of these individual regions as they undergo shear motion on the solar surface after emergence . to this end , after the automatic algorithms listed above have run to convergence , we manually shift the boundaries between regions to ensure we consistently track flux emergence and migration over the entire timeseries . we allow individual regions to emerge , submerge , change shape , translate , split , and merge . to further reduce the number of regions , we exclude regions which have non zero flux for fewer than 5 timesteps ( @xmath19 ) . number of distinct positive , negative , and total mask regions at each time.,scaledwidth=48.0% ] figure [ fig : sources ] shows the number of mask regions of each polarity over the entire timeseries . in our final analysis , we track 118 regions for @xmath20 hrs . the number of distinct regions at any time varies between 10 and 35 , generally growing at a steady rate as the active region complex emerges between 2011 - 02 - 11t08:10:12 and 2011 - 02 - 14t16:10:12 , then dropping slightly as the fully emerged system continues to concentrate shows that for ar11158 , neither polarity is consistently or significantly more fragmented than the other . this contrasts with observations of most other active regions , where the leading polarity is substantially more concentrated than the trailing . this can show up in mct models as the leading polarity s flux distributed amongst fewer poles than that trailing polarity . this does not seem to be the case for ar11158 , which we find curious , although it does not affect our analysis . ] . four examples of the los magnetogram overlaid with masks are shown in figure [ fig : partition ] . the top four images are each a single frame from an animation of the full 250 timesteps viewable in supporting media in the online journal . arrows indicate the chronological sequence . below the magnetograms is a full disk integrated goes x ray curve @xmath21 over the course of the time series , with vertical lines showing the times of the four magnetograms . in preparation for the transition to an mct mcc analysis of the system , we represent each distinct region as a magnetic point source , or _ pole _ , in the local tangent plane at each timestep . for consistency from timestep to timestep , our point of tangency at each time is taken as the center of charge at the initial timestep , migrated through solar rotation to the present timestep . pole @xmath22 at time @xmath11 is defined by its associated region s total flux @xmath23 and flux weighted centroid @xmath24 : @xmath25 the vertical magnetic field @xmath26 accounts for both line of sight effects , by assuming a radial field , and pixel foreshortening , as described above . to reduce the effects of noise in our masking algorithms , we additionally smooth the flux in each region over the four day series by convolution with a 9 timestep ( 3.2 hour ) boxcar function , using edge truncation . for instance , pole @xmath22 s flux at time @xmath27 is averaged to @xmath28 . the resulting smoothed fluxes are shown in figure [ fig : sflux ] for regions that have @xmath29 at any timestep in the series . discontinuous jumps , for instance around 65 hrs , indicate merging , in this case p37 into p3 . this occurs when the distinction between separately emerged regions becomes ambiguous . smoothed flux in each region having at least @xmath30 at some time.,scaledwidth=50.0% ] it is evident from viewing the full animation associated with magnetograms in figure [ fig : partition ] that the active region complex emerges in several distinct episodes . the complex has two distinct primary locations of flux emergence : one to the north , which leads another to the south by about @xmath31 . the first emergence episode is ongoing at the beginning of our timeseries , at which point several smaller emerged regions are still consolidating ( eg . , n3/n5 in the south ; n1/n2/n6 , and p1/p6 in the north ) . the second episode begins around our 60th timestep ( 30 hrs from t=0 ) , on feb . 12th at 16:00ut , and continues very steadily until timestep 80 ( t=60 hrs , feb . 13th , 20:00 ) . at this time , the northern emergence ceases , while it appears that the southern emergence continues , possibly in repeated @xmath32 bursts . these bursts may be an artifact of the daily variations in hmi s reported flux , noted above . signed flux in the northern ( solid blue ) and southern ( solid black ) emergence zones . dashed lines show total signed flux in north ( black ) , south ( blue ) , and all ( red ) regions , as well as the combination of northern negative with southern positive regions . this readily shows the northern and southern emergence regions undergo two distinct emergence patterns.,scaledwidth=50.0% ] figure [ fig : f - imbal ] shows the evolution of signed flux in those regions belonging to the northern emergence ( blue ) , southern emergence ( black ) , and a set of surrounding quiet sun ( network ) regions that drifted above our thresholds at various times ( purple ) . of the 118 regions tracked over the timeseries , 54 belonged to the north ( 36 positive , 18 negative ) , 37 the south ( 13 positive , 24 negative ) , and 27 external to the active region complex ( 13 positive , 14 negative ) . dashed lines in the figure show the flux imbalance for each of the three sets , and also the flux imbalance of the total system ( green ) . note that qualitatively , the northern regions undergo a very different emergence evolution compared to the southern regions . this is easily seen with the dashed red line , which shows the total signed flux of the complex s central region : northern negative flux summed with southern positive flux . the system s connectivity is defined by the amount of flux connecting each pole to every other pole @xcite . this constitutes a _ graph _ , where each pole is a vertex and each domain an edge , with the weight of each edge given by the domain flux . the total flux of a single pole is the summed weight of all edges connected to it , and the total flux of the system of poles is the summed weight of all edges in the graph . if there is an overall imbalance of flux , the remainder must be connected to a source located ( formally ) at infinity . in general , a pair of vertices may be connected by more than one edge , and are then called multiply connected . we have found several such instances of multiply connected vertices in ar11158 , though we show below that , in this case , their effect on the system s energetics is negligible . in determining the distribution of flux emergence within the active region , we use the method of @xcite to define a connectivity graph for the flux difference between consecutive timesteps . this change must come in pairs , as positive and negative poles emerge and submerge together . the total flux change between times @xmath11 and @xmath33 for a single pole @xmath22 is given by eq . ( 4 ) of @xcite @xmath34 where @xmath35 describes any shift in the boundary between like signed region @xmath36 adjacent to @xmath22 and @xmath12 describes any change in the photospheric field itself . the former is a graph with edges connecting like signed regions of opposite flux change ( flux that one region loses , another gains ) , while the latter is a graph connecting opposite signed regions with same sensed flux change . the algorithms for determining these are fully described in @xcite . to accurately deal with the two regions of emergence , we first calculate the matrix @xmath37 for northern and southern regions separately , then combine the two resulting connectivity graphs . finally , we allow for connections between north and south using any remaining flux change . we may quantify our success at capturing the flux change processes by reconstructing the total flux of a region using its initial flux and elements of the surface change matrix . at time @xmath11 , we estimate a region @xmath22 s flux as @xmath38 . summing these reconstructed fluxes over a set of like signed regions connected by internal boundaries , say all the positive flux in the southern emergence zone , should represent the total emergence of the collected regions . we find that our method always underestimates this emergence . over the entire time series , we find a maximum discrepancy of between @xmath39 and @xmath40 , depending on the group we reconstruct ( northern positive , northern negative , southern positive , southern negative ) . we believe this conservative attribution of flux change to emergence or submergence processes stems from a greedy boundary change algorithm , asymmetries in the concentration of newly emerged positive and negative flux , and the diurnal variations due spacecraft motion , noted above . this forces the attribution of @xmath41 of flux change to emergence ( or submergence ) with sources formally at infinity . finally , we note that there is quite a bit of variation in our underestimate of emerging flux . our algorithm has the greatest underestimate when reconstructing the northern positive flux emergence regions . while at one point it is only able to pair up @xmath42 of the actual flux change , it spends fully half of all timesteps able to pair at least @xmath43 of the flux change . the flux change formally paired with sources at infinity generally rises over the timeseries , and peaks at @xmath44 of the total instantaneous unsigned flux 12 hours before the m6.6 flare , then varies between @xmath45 for the rest of the series , ending 10 hours after the x2.2 flare . this variation is consistent with the @xmath46 daily variation in unsigned flux found by @xcite . the flux change assigned to sources at infinity at the times of the m6.6 , m2.2 , and x2.2 flares are @xmath47 , @xmath48 , and @xmath49 of the instantaneous unsigned flux , respectively . having quantified the connectivity graph for flux change , we may use the topological methods of @xcite to calculate the free energy stored in coronal fields . at every timestep we determine the potential field connectivity matrix , @xmath50 , using the monte carlo method of @xcite . at the timesteps immediately preceding each m and x class flare , we calculate the system s potential field topology in terms of poles , nulls , and separators @xcite . in our analysis , deviations from a potential field configuration take the form of differences in the amount of flux ( either more or less ) in the real field s domains versus the potential field s domains . in the mcc model , the departure of a domain from a potential field configuration gives rise to currents in associated separators . every domain that is topologically linked by a separator contributes to that separator s current . to determine which domains each separator links we use the gauss linking number method of @xcite . completing the free energy estimate for each separator , we use the direct connection between currents flowing along separators and free magnetic energy given by @xcite . as shown in @xcite , the self flux of a separator ( current ribbon ) , generated by currents flowing along it , is equal to the difference between the linked domain fluxes in the real and potential fields : @xmath51 here , @xmath52 is the real domain flux of a domain @xmath53 at time @xmath11 , given by the initial potential field flux and the summation over time of the surface flux change matrix , defined above : @xmath54 the difference between the real and potential field fluxes at each time gives @xmath55 , the total amount of flux which may be redistributed between domains in a reconnection event . the sum of @xmath55 over all times up to @xmath11 , and over all linked domains @xmath53 , results in equation . we may extend this model to include reconnection by considering the effect of reconnection at some time @xmath56 on the connectivity matrices described above . the potential connectivities @xmath57 do not change , as they are uniquely determined by the photospheric boundary at any time . the only effect is to transfer flux between domains in the real field . we accomplish this by adding some flux transfer matrix @xmath58 to @xmath59 at time @xmath56 , so that @xmath60 the total flux through the photosphere does not change during a flare , so the potential field @xmath57 also does not change . according to equations [ eq : fce ] and [ eq : totfcr ] , then , there is an opposite assignment of flux in the redistribution matrix @xmath61 . we may model the effect of multiple reconnections by adding / subtracting flux transfer matrices @xmath62 as necessary . therefore , the separator self flux at any time @xmath11 , including all past reconnection events , is given by @xmath63 where @xmath64 is the heaviside step function . in the following section , we will propose a minimization scheme for estimating the reconnection matrix @xmath65 at the time of a flare . having thus determined each separator s self flux at any time , we follow @xcite to relate that self flux to the separator current by @xmath66 with @xmath67 allows us to represent the current in terms of the fluxes . finally , from @xcite equation ( 4 ) we can calculate the energy in the mcc model in excess of the potential field magnetic energy , @xmath68 which , via equation , is a function of the calculated separator fluxes @xmath69 . we determine the coronal topology for the m6.6 flare at 17:22 on feb . 13 , 6 minutes before flare onset , and 16 minutes before goes peak intensity . at this time , our model consists of 27 sources ( 16 negative , 12 positive ) and 26 nulls . following @xcite , these numbers satisfy both the 2d and 3d euler characteristics , so we believe we have found all nulls . every null is prone , and there are no coronal nulls . one null is asymptotic in the sense of @xcite , laying along the direction of the region s dipole moment computed about the center of unsigned flux , @xmath70 , at a distance @xmath71 , where @xmath72 is the net charge . this null s separatrix surface forms a boundary between the region s closed flux and surrounding open flux . 7 additional source null pairs are part of unbroken fans : p57/b22 , p52/b21 , n51/a20 , n45/a18 , n43/a16 , n42/a25 , and n38/a26 . using these values and the equation between ( 26 ) and ( 27 ) of @xcite , we expect to find 17 separators in the corona ( along with 17 mirror separators ) , which we do find . we therefore believe we have completely specified the system s topology on the eve of the flare . we perform a similar analysis just prior to the m2.2 and x2.2 flares . while we do not find every topological element in these later flares , we believe we find all that play a significant role in each case . for the 17 separators at the time of the m6.6 flare , we find 90 linked domains . two of the separators have the same endpoints , nulls a07/b01 , and therefore enclose multiply connected source pairs @xcite . these are known as redundant separators . in this case , the two separators lay nearly along the same path , implying a slight wrinkle in the intersecting fan surfaces . this creates one additional flux domain enclosed by the two separators . because the cross sectional area in this case is small , the enclosed flux is small relative to the total flux enclosed by each separator , and the corresponding energy due to the redundant separator is negligible . the monte carlo estimate of fluxes enclosed by the different separators used 500 field lines . there was no difference in the number linked by the separators , so we conclude that the fluxes they link are identical to a fraction of a percent . we use only one of the two in our calculation , and arrive at the same result independent of this choice . as stated above , one shortcoming of current mct / mcc analyses are their inability to account for violation of the flux constrains . as such , they have no way to account for reconnection . we here present a method for relaxing those constraints at any timestep , while allowing the system to continue evolving after reconnection . during reconnection , flux is exchanged across the field s separators . each separator lies at the boundary of four flux domains , and the separators involved in the flare identify the set of domains which exchange flux . some of these domains are flux superficient , containing more flux than in a potential field , and some deficient . not every separator needs to be involved in every flare , and not all flux is necessarily transferred in every flare , even within the subset of involved domains . reconnection does not simply involve the transfer of flux from superficient to deficient domains . two domains on opposite sides of a separator ( x point in 2d ) reconnect fieldlines , transferring flux to the remaining two domains . there is no physical reason why opposing domains must both have more ( or less ) flux than a potential field configuration . instead , the state of the current domain depends on the history of its poles : where and with whom they emerged , who they reconnected with in the past , and what their current geometric orientation is . the only requirement for reconnection is that the two flux donating domains contain some flux ( nonzero elements of @xmath59 in equation ) . in such cases , this poses the interesting question of whether the potential field configuration is always attainable through reconnection , or if there exists some local minima in configuration space . we briefly consider this below , but leave a more detailed analysis of the question for another investigation . we use a simple iterative minimization algorithm to model the redistribution of flux across a set of separators involved in a particular event . for each separator , we propose a small transfer of flux between the four surrounding domains . each domain is specified by the indices of its two poles in a connectivity matrix , and reconnection is described by the transfer of flux @xmath73 from two elements of our redistribution matrix to two others . supposing that the two flux donating domains are @xmath10 and @xmath74 , while the two receiving domains are @xmath75 and @xmath76 , then the 4 involved elements of redistribution matrix change as follows : @xmath77 . we then calculate the total energy drop over the entire system ( the summation of equation for every separator ) . we apply the small reconnection resulting in the largest total free energy drop . we then calculate the next set of energy drops for each small reconnection , and this process continues until one of two conditions is met : either 1 ) no flux transfer is possible because no pair of opposing domains both have nonzero elements of the connectivity matrix @xmath59 ; or 2 ) any small reconnection increases the total free energy . the sum of every small reconnection @xmath73 then fills out the flux transfer matrix @xmath58 for the flare at time @xmath56 . having determined how to model reconnection , we now turn our attention to determining when to apply a minimization . at present we have no model for the mechanism in the actual corona which initiates reconnection at a current sheet . all we can infer , from observations of actual flares , is that at some instant the reconnection does begin at certain separators . we therefore rely on observations of this kind to determine the separators undergoing reconnection , and when this reconnection occurs . for the present study , we perform a minimization for every flare associated with ar11158 with goes class of m1.0 or greater . this pares the number of separate minimizations down to a manageable amount for a proof of concept , while still allowing us to understand some of the large scale processes at work in the active region s evolution . we employ chromospheric data to select a subset of coronal domains involved in each flare . we associate the chromospheric flare ribbons observed in the 1600 channel of aia data with specific spine field lines of the topological skeleton . while the spine lines do not always geometrically match the ribbons , there is a topological correspondence @xcite . the spines are the photospheric footpoints of reconnecting loops , and therefore indicate which flux domains are involved in each flare . magnetic reconnection across a separator couples the flux redistribution in the corona to the photospheric spines of the separator s null points . the highlighting of spine lines by flare ribbons thus indicates those separators involved in each flare . to make use of this information , we relax flux constraints ( allow for reconnection between four domains ) using only those separators associated with the highlighted ribbons . log scaled aia 1600 image during the goes class m6.6 flare , with coordinates given in arcseconds from disk center . the grayscale saturates at 6000 counts , roughly half the pre flare maximum pixel value . the potential field skeleton is overlaid : pluses and crosses are positive and negative poles , respectively ; triangles are positive ( @xmath78 ) and negative ( @xmath79 ) nulls ; thin solid white lines depict spines . the energy calculation only attempts reconnection across those separators having two boxed nullpoints as footpoints . these seven separators displayed as thick solid blue lines . the remaining ten separators are displayed as green dashed lines.,scaledwidth=80.0% ] figure [ fig:143ribbon ] shows the aia 1600 data for a selected timestep during the m6.6 flare . the aia image is displayed in a logarithmic grayscale , and shows a relatively simple two ribbon flare . overplotted are contours of the magnetogram at + 75 g ( yellow ) and -75 g ( blue ) , as well as the topological skeleton ( see caption for details , and online material for an animation covering the time of the flare ) . clearly visible are the two primary flare ribbons , located on either side of the central polarity inversion line ( pil between southern emerged positive flux and northern negative flux ) . evident in the online animation are several other , smaller flare ribbons , located in the southern negative and northern positive regions . there are spine fieldlines associated with each ribbon . the northern primary ribbon corresponds to the spine lines of null a06 , between poles n2 and n26 , near @xmath80 . the southern ribbon is only morphologically similar the potential field mct model . we separate the more diffuse p59 region from the more concentrated p3 and p37 , which forces the creation of two null points ( b23 , b10 ) with associated spine lines , near @xmath81,-250)^{\prime\prime}$ ] , respectively . we believe the real field likely has a null directly between p3 and p37 , creating a more direct spine line between the two . the spines involved in this flare have the red boxed nullpoints in figure [ fig:143ribbon ] as their spine sources . this indicates that flux is transferred only across those separators connecting two of the boxed nulls . the projection onto the photosphere of seven such separators in this flare are shown as thick blue lines . the remaining separators are shown as dashed green lines . the free energy of these other separators may still change during the flare despite have no reconnection across them , provided the gauss linking number between the separator field line and any domain which does participate in reconnection is nonzero , as indicated by equation . figures [ fig:202ribbon ] and [ fig:224ribbon ] are similar to figure [ fig:143ribbon ] , during the m2.2 and x2.2 flares , respectively . we have left off the contours of the magnetogram in these figures for clarity . the x class flare in particular is more complex than previous flares , involving more and disparate parts of the active region complex . this increased activity is likely influenced by the creation of a coronal null point just prior to the m2.2 flare , whose fan surface effectively separates the northern and southern emergence zones . the energy buildup prior to the m6.6 flare is particularly dependent on the emergence of n26 in the north . this generates the null point in the north to which the four northern involved separators attach . in the 25 hours between n26 s emergence at 2011 - 02 - 12 16:00 ut and the m6.6 flare , we calculate an increase in free energy due to currents along these four separators of @xmath82 , about one third of the total mcc free energy in the system at this time . these separators link domains n26/p37 , n26/p59 , n26/p31 , n26/p39 , n26/p44 , n28/p31 , n37/p3 , and n37/p39 . same as figure [ fig:143ribbon ] , for the m2.2 flare . solid blue and green lines show separators connected to nulls with spines laying approximately along paths of flare ribbons observed in aia 1600 channel . locations of other separators are shown as dashed green lines.,scaledwidth=50.0% ] the 1600 flare ribbons indicate that 6 separators are involved in the m2.2 flare ( figure [ fig:202ribbon ] ) . four connect to null a14 between regions n25 and n56 , and 2 connect through the coronal null a33 ; their projections in the photospheric plane are shown as solid blue solids , with the remaining separators shown as dashed green lines . same as figure [ fig:143ribbon ] , for the x2.2 flare.,scaledwidth=50.0% ] figure [ fig : m66min ] illustrates the result of the free energy minimization for the m6.6 flare . thick orange lines show the projections of separators across which flux was transferred by the minimization process . note that this involves three separators of the six identified via flare ribbons in figure [ fig:143ribbon ] , which are themselves a subset of the 17 total separators at this time . the dashed lines overlaid on the magnetogram , mask , and topological footprint background show those domains that exchanged flux during minimization these are not fieldlines , just identifications of the involved domains . the amount of flux loss or gain is indicated by the colorbar , with white to green indicating increasing amounts of flux gain , and black to red increasing flux donation . it is immediately apparent that the minimization does not exactly match our expectation from the flare ribbons , despite our specification of `` involved separators . '' in particular , the four domains involving n2 , n26 , p37 , and p59 are essentially nonparticipants in modeled flare , whereas they are clearly the dominant players in the actual flare . in our model of emergence , these domains are not simply flux deficient relative to the potential field , but have _ zero _ initial flux . we will discuss this in more detail in [ sec : disc ] . in total for the m6.6 flare minimization , we find that @xmath83 of flux was exchanged between @xmath84 domains across @xmath85 separators . this exchange took @xmath86 iterative minimization steps , resulting in a total drop of @xmath87 , @xmath9 of the pre minimization mcc free energy ( @xmath88 ) and @xmath89 of the potential field energy ( @xmath90 ) . these results are summarized in table[tab : minimize ] , together with those for the m2.2 and x2.2 flares , and we discuss them in more detail in the next section . lccccccc flare & flux ( @xmath91 ) & domains & separators & steps & initial @xmath92 & @xmath93 & @xmath94 + m6.6 & 4.2 & 10 & 3 & 1922 & @xmath95 & @xmath96 & @xmath97 + m2.2 & 2.0 & 15 & 8 & 327 & @xmath98 & @xmath99 & @xmath100 + x2.2 & 21.0 & 17 & 10 & 29504 & @xmath101 & @xmath102 & @xmath103 + this investigation builds on @xcite in adding the observational history of an active region s flux evolution , in particular its flux emergence , to the mcc model . here we have relied on line of sight magnetograms provided by sdo / hmi and generated our flux histories by assuming a radial field . with the arrival of the hmi active region patches ( harps ) dataseries to jsoc , future investigations can use the actual vertical flux determined by hmi s vector magnetograms . we were fortunate in the present case to have hmi observe the entire history of ar11158 from its emergence around @xmath104 solar east on feb . 10th , 2011 , to its rotation off - disc some 9 days later . in the more common case where we do not observe the entire emergence of an active region , we could use a nlfff extrapolation to generate an initial connectivity state , which would then be updated in time via the methods of [ sec : mod - phot ] . we have gone several steps further than previous energy estimates using the mct / mcc framework , such as @xcite or @xcite . most importantly , we now allow for violation of the no reconnection flux constraints of the mcc . this enables us to not only consider energy storage due to currents along a subset of separators in a flare , but also allows for an estimate of flux transfer and energy conversion during a flare . further work may yield an interesting comparison between the flux transferred in our minimization to estimates of reconnected flux based on analysis of flare ribbons , as was done in @xcite . for each flare , we choose a subset of separators that are allowed to transfer flux during our minimization based on observations of chromospheric flare ribbons . each separator bounds 4 domains , though some of these domains are bounded by more than one separator . thus , the number of domains which may exchange flux via reconnection is typically less than 4 times the number of involved separators . it is interesting that the minimization scheme we have proposed does not utilize every allowed domain . as mentioned above , in order to undergo reconnection , two domains on opposite sides of a separator must both contain flux ( have nonzero elements in the connectivity matrix @xmath59 ) . focusing again on the m6.6 flare , the two separators that connect to null a06 , between poles n2 and n26 , and have pole p59 as a spine source of their b type nulls ( nulls b10 and b23 ) contain very strong currents and border flux domains with highly nonpotential connectivities . and yet , they do not participate in the minimization . in this case , there is no pathway of free energy loss that results in donatable flux on opposite sides of these separators . the lack of any flux in the p59/(n2 , n26 , n28 ) domains effectively cuts off any involvement of p37 in our model of this flare . between poles n2 , n25 , p37 , and p59 , flux transfer may occur in either of two directions . domains n2/p59 and n26/p37 may donate flux , with n2/p37 and n26/p59 receiving , or n2/p37 and n26/p59 may donate flux with n2/p59 and n26/p37 receiving . in either case , one of p59 s domains must donate flux . since both the p59/n2 and p59/n26 domains have _ no _ flux in our model ( not to be confused with having less than the potential field configuration ) , this reconnection can not occur . we do , however , observe the primary flare ribbons involving just these domains . so , where has our model of this active region gone awry ? first , we note that there are many small flare events smaller than m1.0 prior to the first flare we consider . any of these events may transfer flux into the domain necessary for the m6.6 event . second , it is likely that a great deal of reconnection takes place that is unassociated with any goes class flare . this may be a `` steady '' reconnection , as is observable in euv and x ray images of emerging magnetic regions . for instance , in our previous study @xcite of ar11112 from october 2011 , we saw that newly emerged flux steadily reconnected with surrounding preexisting flux , as evidenced by the encroachment of the bright xrt kernel into the surrounding flux . this is discussed in detail in @xcite . although the region did produce several flares later in its evolution , this steady reconnection occurred independently of any observable flares . we believe that a similar process is ongoing in ar11158 s evolution . in particular , we believe this type of steady reconnection occurs along the central pil , where southern emerged positive flux has collided with and sheared relative to northern emerged negative flux . such a steady , low level reconnection would populate these central domains with flux . the free energy of this portion would increase as these domains continued to shear , culminating in the series of explosive reconnections observed at later times . one may see evidence of this in euv images from aia , showing low laying loops that cross the central pil , together with very high loops apparently connecting the most westward positive flux to the most eastward negative flux concentrations . another interesting point is that , in our minimizations , no separator expelled all of its current and thereby reached the potential field state , as one might expect to be the case in an mcc model . instead , in each of the three reconnection events modeled here , the system reaches a state where reconnection across on separator reduces one domain s flux to zero . reconnection across a second separator can repopulate the zero flux domain , allowing further reconnection across the first separator . however , it often happens that the two separators each require flux donation from that same domain , and so no further reconnection is possible across either one . @xcite also found that the total free energy derived from the mcc model was greater than their calculated energy losses observed during the feb . 24 , 2004 x class flare . they attributed this to incomplete relaxation through reconnection , but had no way to assess why this might be the case , as we have developed here . the potential energy ( solid line ) and mcc free energies both before ( triangles ) and after ( diamonds ) each minimization , the times of which are indicated by vertical lines . there is no way to calculate a potential energy in the mct / mcc model , so the free energy has been plotted above the potential energy derived from a fourier transform method.,scaledwidth=50.0% ] we may compare the energies associated with each event in a variety of ways . the solid line in figure [ fig : eplot ] shows the potential energy calculated from each magnetogram using the fourier transform method of @xcite . this is similar to , but not directly comparable with , the potential energy shown in figure 4(c ) of ( * ? ? ? * ( hereafter s12 ) ) . when not stated explicitly in the text , we have determined approximate values for potential and free energies from their figure 4 . because s12 use the vertical flux determined from the hmi vector magnetograms whereas we use the los field , deprojected assuming a radial field at every pixel , we have different lower boundaries for determining the potential energy , so that our calculated values differ . our potential energy appears consistently lower than that of s12 , and also seems to have more substantial variations . these variations become more pronounced after @xmath105 , when the northern region ceases substantial emergence and the active region crosses disk center . the free energies of the mcc model @xmath92 are shown in figure [ fig : eplot ] as red triangles at the time of each flare and blue diamonds after minimization . we have added the mcc values to the calculated potential field energies at each time in order to emphasize that these are energies in excess of the potential field . however , this addition should be taken with caution , because the free and potential energies are calculated using two incompatible boundary conditions : the former using the point sources of the mct model with an imposed fce constraint , the later using continuously distributed magnetograms . the potential energy of a point source is infinite , and the mcc is only able to determine the difference between energies of the fce and potential fields . even given these issues , it is still a useful comparison to make . the initial free energy of the mcc increases from flare to flare , while the calculated energy drop due to flux exchange during minimization scales with the size of each flare . this is most easily seen in the @xmath93 column of table[tab : minimize ] . the general trend of the amount of flux exchange , and resulting energy drop , in each flare follows our expectations given the goes class . an order of magnitude more flux is exchanged for the x2.2 flare compared to the m flares , yet this leads to a two order of magnitude greater free energy drop . this highlights the important point that free energy does not scale linearly with flux difference from potential , even for a given separator . s12 made extensive use of hmi s vector magnetograms to present a detailed discussion of magnetic energy in ar11158 , so it is useful to compare the results of that nlfff model to ours . those authors extrapolate a nlff field at a 12 second cadence using the hmi vector magnetogram as a lower boundary . each extrapolation is independent of the others , maintaining no memory of previous magnetic flux or connectivity . it can therefore be difficult to ascribe any rise or drop in the free energy from one time to the next to any particular event . they state a spectropolarimetric noise of @xmath106 , but acknowledge that errors due to extrapolation are unknown and possibly greater . we note that a persistent change from before to after is more likely to be a manifestation of change in the actual solar field . for the x2.2 flare , s12 find an initial free energy of @xmath107 , and a persistent free energy drop from before to after the flare of @xmath108 . this may be compared to our model , which sets the initial free energy at @xmath109 and a pre to post minimization drop of @xmath110 . that both of these results are greater than those of s12 is a little surprising because mcc provides a lower bound on the free energy of linear force free fields evolving quasistatically @xcite . while it is not necessary that a lfff have greater free energy than a nlfff derived from the same boundary , one often assumes it to be the case . at the same time , the mct / mcc model does not involve an extrapolation , except to determine the topological structure of the region . the non potentiality of the region is determined simply by using observations emergence to fix the regions connectivity . our minimization scheme only determines the total amount of energy loss provided that all possible free energy minimizing reconnections take place . not only may the algorithm terminate in a local free energy minimum , as discussed above , there is no physical reason why all possible reconnections need to take place in a single event . it is much more difficult to make such a comparison between these two models for the m6.6 and m2.2 flares . from figure 4(d ) of s12 , we estimate initial free energies of @xmath111 and @xmath112 for these flares , respectively , compared to our results of @xmath113 and @xmath114 . our minimization gives pre to post reconnection drops of @xmath115 and @xmath116 in each case , which is the approximate level of the spectropolarimetic noise in the nlfff extrapolations . indeed , again looking at figure 4(d ) of s12 , neither the m6.6 or m2.2 flare appears cotemporal with a decrease in free energy , and certainly not with a persistent decrease , as is the case with the x2.2 flare . as a final energetic comparison , we calculate the energy loss due to radiation in each flare using the goes light curves via the method of @xcite . for the m6.6 , m2.2 , and x2.2 flares we find radiative losses of @xmath117 , @xmath118 , and @xmath119 , respectively . the above mentioned papers show that other energetic losses , such as thermal conduction and enthalpy flux , tend to dominate the radiative losses during flares . the total energetic loss is difficult to precisely quantify , but may exceed the goes estimated radiative loss by a factor of @xmath120 . given that , the results of our minimization for the m6.6 and m2.2 compare favorably with the goes estimates . however , our estimation for energy loss during the x class flare may be greater than the observed energetic losses by an order of magnitude . this is not surprising because the great extent of the x class flare ribbons encompassed more separators and more domains , ultimately allowing more pathways for energy minimizing reconnection . this allowed the minimization algorithm to exchange much more flux before halting in a local minimum . graham barnes provided code for producing the potential field connectivity matrices using a monte carlo algorithm with bayesian estimates , as described in @xcite . development of the code was supported by the air force office of scientific research under contract fa9550 - 06-c-0019 . we would like to acknowledge our use of nasa s astrophysics data system . this work is supported by nasa under contract sp02h3901r from lockheed martin to msu .	it is well known that photospheric flux emergence is an important process for stressing coronal fields and storing magnetic free energy , which may then be released during a flare . the _ helioseismic and magnetic imager _ ( hmi ) onboard the _ solar dynamics observatory _ ( sdo ) captured the entire emergence of noaa ar 11158 . this region emerged as two distinct bipoles , possibly connected underneath the photosphere , yet characterized by different photospheric field evolutions and fluxes . the combined active region complex produced 15 goes c class , 2 m class , and the x2.2 valentine s day flare during the four days after initial emergence on february 12th , 2011 . the m and x class flares are of particular interest because they are nonhomologous , involving different subregions of the active region . we use a magnetic charge topology together with the minimum current corona model of the coronal field to model field evolution of the complex . combining this with observations of flare ribbons in the 1600 channel of the _ atmospheric imaging assembly _ ( aia ) onboard sdo , we propose a minimization algorithm for estimating the amount of reconnected flux and resulting drop in magnetic free energy during a flare . for the m6.6 , m2.2 , and x2.2 flares , we find a flux exchange of @xmath0 , respectively , resulting in free energy drops of @xmath1 .
within the bosonization formalism the equal - time single - particle correlation function is @xmath108 the correlator appearing above is obtained from eqs . ( @xmath109 ) and ( @xmath110 ) of the main text and is given by @xmath111\end{aligned}\ ] ] where the derivative of eq . ( @xmath67 ) of the main text is @xmath112 \end{aligned}\ ] ] with @xmath113 the bessel function of the first kind . recall that @xmath31 is the typical lengthscale , @xmath45 the dimensionless momentum , and @xmath46 the dimensionless time . in order to obtain analytical approximations for the single - particle function @xmath114 , we need to approximate the integral @xmath115 . we identify three important regimes : @xmath116 , @xmath117 and @xmath118 ( note that @xmath119 and that we separate the regimes in @xmath4 at @xmath120 instead of @xmath121 as this corresponds to a splitting at @xmath122 in momentum space ) . these three regimes determine the size of the arguments of the bessel functions entering the integral @xmath115 . therefore , we split this integral into three parts @xmath123 with @xmath124 , @xmath125 , and @xmath126 using @xmath127 \times \\ & & \left[\left(j_{\frac 2 3}(s ) j_{- \frac 2 3}(s\tau^{\frac 3 2 } ) - j_{-\frac 2 3}(s ) j_{\frac 2 3}(s \tau^{\frac 3 2})\right)^2 + \left(j_{-\frac 1 3}(s ) j_{-\frac 2 3}(s \tau^{\frac 3 2 } ) + j_{\frac 1 3}(s ) j_{\frac 2 3}(s \tau^{\frac 3 2})\right)^2 \right ] . \nonumber\end{aligned}\ ] ] the bessel functions of the first kind can take simpler approximated forms ( using expressions taken from ref . ) in the two following regimes : @xmath128.\end{aligned}\ ] ] for the function @xmath129 which covers the regime @xmath130 , we get @xmath131 e^{-s \tilde{\alpha}}= \frac{1}{k(\tau ) } \left\ { e_1\left(\frac 2 3 \tilde{\alpha}\right ) - \frac 1 2 \left[e_1\left(\frac 2 3 ( \tilde{\alpha}+ i\xi)\right ) + e_1\left(\frac 2 3 ( \tilde{\alpha}- i\xi)\right)\right]\right\},\end{aligned}\ ] ] where @xmath132 and @xmath133 is the exponential integral function @xcite . for @xmath134 , we find @xmath135 while for @xmath136 , @xmath137\end{aligned}\ ] ] where @xmath138 is euler s constant . for the small momentum contribution , given by @xmath139 , we obtain the following expression @xmath140 \nonumber \\ & \simeq & \frac 1 { k_0 } \mathrm{cin}\left(\frac{2\xi}{3 } \frac{1}{\tau^{\frac 3 2 } } \right).\end{aligned}\ ] ] the most interesting contribution is the intermediate momentum one which is due to @xmath141 . this expression simplifies to @xmath142 where @xmath143~\sin^2(s\tau^{\frac 3 2}-\frac{\pi}{12 } ) , \\ \phi_2(\xi , \tau^{\frac 3 2 } ) & = & \int_{\frac{2}{3 } \tau^{-\frac 3 2}}^{\frac 2 3 } \frac{ds}{s^{\frac 2 3}}~[1-\cos(s\xi)]~\cos^2(s\tau^{\frac 3 2}+\frac{\pi}{12 } ) . \end{aligned}\ ] ] in the limit where @xmath144 and @xmath136 , this expression can be further simplified and we find @xmath145 whereas in the limit where @xmath144 and @xmath134 , we find @xmath146 the full single - particle correlation function is given by @xmath147 . we now use the expressions derived above to find approximations for the entire function . the fourier transform of a stretched exponential function of the form @xmath159 with @xmath160 is given by the expression @xmath161 which is a convergent sum . when @xmath162 , the sum can be further evaluated by using properties of the @xmath163 functions , @xmath164 , and definitions of the hypergeometric functions @xcite , this exercise yields the result @xmath165\end{aligned}\ ] ] where @xmath166 is the scorer function @xcite .	we investigate the response of a one - dimensional bose gas to a slow increase of its interaction strength . we focus on the rich dynamics of equal - time single - particle correlations treating the lieb - liniger model within a bosonization approach and the bose - hubbard model using the time - dependent density - matrix renormalization group method . for short distances , correlations follow a power - law with distance with an exponent given by the adiabatic approximation . in contrast , for long distances , correlations decay algebraically with an exponent understood within the sudden quench approximation . this long distance regime is separated from an intermediate distance one by a generalized lieb - robinson criterion . at long times , in this intermediate regime , bosonization predicts that single - particle correlations decay following a stretched exponential . this latter regime is unconventional as , for one - dimensional interacting systems , the decay of single - particle correlations is usually algebraic within the luttinger liquid picture . our results may serve as a benchmark for future experimental studies of breakdowns of conventional many - body dynamics in one - dimensional systems . _ introduction : _ recent advances in the development of fast probing and control techniques applicable to correlated systems have opened up the possibility to dynamically prepare complex quantum many - body states . for example , effective phase transitions have been induced through the application of external driving fields @xcite and states , such as a bell state of ions or a tonks - like state in a quantum gas , have been realized using tailored environments @xcite . in fact , the dynamical preparation of states promises to have an important impact in fields as diverse as condensed matter physics , quantum information , quantum optics and ultracold atomic physics . on the theoretical side , despite tremendous progress in recent years , many of the basic concepts behind the dynamical generation of states still remain to be understood . generally , the evolution of correlated systems involves intricate highly excited states and couplings to various dissipative channels . in this article , we focus on the preparation of unconventional states in isolated systems using slow parameter changes . considerable experimental efforts have been devoted to understand slow quench dynamics . for example , the evolution of interacting bose atoms under a slow variation of the depth of the confining optical lattice has been the topic of various experimental studies @xcite . in these works , as well as in many theoretical ones ( see ref . and references therein ) , the emphasis has been put on understanding how energy is absorbed and defects produced during a slow quench . in recent years , the focus has partially shifted towards the study of longer range correlation dynamics during a slow parameter quench @xcite . understanding the evolution of such correlations is paramount as the nature of many - body quantum states are typically characterized by longer range correlators . interestingly , light - cone - like spreading @xcite of parity correlations , both in space and time , has even been observed experimentally in an interacting one - dimensional bosonic gas after a sudden quench of the optical lattice depth @xcite . for slow quenches , a similar linear light - cone - like evolution of correlations has been predicted for density correlations in bosonic systems @xcite and for single - particle correlations in fermionic systems @xcite . in these studies , the functional form of the correlation decay was found to consist of different algebraic regimes , each one being characterized by a different exponent @xcite . we analyze here the correlation dynamics during a slow linear increase of the interaction strength in two paradigmatic one - dimensional interacting models : the lieb - liniger and bose - hubbard models . we show that a generalized lieb - robinson bound describes the evolution of single - particle correlations . outside the bound , these correlations decay algebraically with an exponent determined by the initial luttinger parameter . in contrast , inside the bound , the correlations present much more interesting dynamics . for short distances , the algebraic decay depends on the ramp time @xcite . while for larger distances and quench times , the correlations , within the lieb - liniger model , decay following a stretched exponential . this particular decay form is unexpected as , even for instantaneous quenches , an algebraic decay persists at all distances and times @xcite . a similar stretched exponential behavior was found in ref . . in the rest of the article , we analyze in detail the evolution of single - particle correlations , and highlight the different regimes both in position and momentum space . _ model : _ bosonic atoms in a one - dimensional wave guide can be described by the lieb - liniger ( ll ) model @xmath0\end{aligned}\ ] ] with @xmath1 the boson annihilation operator and @xmath2 the density . the interaction strength @xmath3 is related to the @xmath4-wave scattering length @xmath5 of the atoms and to the transverse trapping frequency @xmath6 by @xmath7 . we assume that the gas is initially prepared at a certain interaction strength @xmath8 and that for @xmath9 a linear variation of the interaction strength of the form @xmath10 is performed . experimentally this variation can be achieved , for example , by using a feshbach resonance or by varying the intensity of the transverse trapping @xcite . a similar interaction quench can be done by confining bosonic atoms to an optical lattice potential along the one - dimensional direction . the theoretical model describing this situation is the bose - hubbard model given by @xmath11 with @xmath12 the operator creating a boson at site @xmath13 and @xmath14 the local density operator . the first term of the hamiltonian corresponds to the kinetic energy of atoms with hopping amplitude @xmath15 while the second term is the potential energy with onsite interaction of strength @xmath16 . taking the continuum limit of the bose - hubbard model in the superfluid phase , this model can be mapped onto the ll hamiltonian @xcite . in this case , the linear interaction quench @xmath17 translates into a linear change of the interaction amplitude @xmath18 . for both models , in the superfluid phase , the low energy physics is well described by the tomonaga - luttinger liquid ( tll ) hamiltonian @xcite @xmath19\,\,\,\,\,\,\end{aligned}\ ] ] where @xmath20 and @xmath21 are conjugate fields satisfying the canonical commutation relation @xmath22 . we have set here @xmath23 and @xmath24 is a short distance cut - off . the sound velocity @xmath25 and the luttinger parameter @xmath26 are related to the parameters of the original hamiltonians . these parameters can , for example , be extracted from the bethe ansatz solution of eq . ( [ eq : lieb - liniger ] ) @xcite or through numerical approaches for the bose - hubbard hamiltonian @xcite . in the ll model , the galilean invariance ensures that the product @xmath27 remains unchanged upon varying the interaction parameters @xcite and thus @xmath28 . for small linear changes of these parameters , this translates , to first order in the variation , to a time - dependent ratio @xmath29 with @xmath30 and a typical lengthscale @xmath31 . this result is then used to obtain expressions for the time - dependent sound velocity @xmath32 and for the time - dependent luttinger parameter @xmath33 . these expressions are still valid for small parameter variations in the bose - hubbard model given the relation @xmath34 where @xmath35 is the lattice constant . a major distinctive feature of the ttl model is that its low energy excitations are collective modes ( density fluctuations ) instead of individual quasiparticles . hence , only quasi - long range order persists even down to zero temperature . this situation is exemplified by the anomalous ( non - integer ) power - law dependence of its correlation functions @xcite . moreover , the time - dependence does not introduce couplings between the different momentum modes of the ttl hamiltonian . this leads to momentum decoupled equations of motion for the fourier components of the fields of the form @xcite @xmath36 the solutions for these equations of motion can be written using bosonic quasiparticles with creation and annihilation operators @xmath37 and @xmath35 which diagonalize the hamiltonian at @xmath38 : @xmath39,\\ \theta(q , t)\ ! & = & \ ! \frac 1 { u(t ) k(t)q } \sqrt{\frac{\pi k_0}{2\|q\|}}\left[a_q \frac{d}{dt}f^ * + a^\dagger_{-q } \frac{d}{dt}f \right]\end{aligned}\ ] ] where @xmath40 is the solution of the equation @xmath41 with initial conditions @xmath42 , @xmath43 . this solution can be expressed in terms of bessel functions ( see eq . ( 9.1.51 ) of ref . ): @xmath44 \nonumber\end{aligned}\ ] ] where @xmath45 and @xmath46 are the dimensionless momentum and time , respectively @xcite . _ evolution of the single - particle correlation function : _ in the following , we survey the rich behavior of the equal - time single - particle correlation function @xmath47 during a slow interaction quench . in the bosonization representation , the equal - time single - particle correlation function takes the form @xmath48 where @xmath49 is a non - universal constant which depends on the underlying microscopic model . we introduced for convenience the dimensionless length @xmath50 and , correspondingly , the dimensionless short distance cut - off @xmath51 . the function @xmath52 of eq . ( [ eq : green-2 ] ) is then given by @xmath53 . \nonumber\end{aligned}\ ] ] from the equation above it immediately follows that eq . ( [ eq : green-2 ] ) only depends on the dimensionless variables @xmath54 , @xmath55 , @xmath56 and not separately on @xmath57 , @xmath58 , @xmath59 and @xmath24 . this implies that , for a given final value of the interaction strength , increasing the ramp velocity , @xmath60 , mainly enters the expressions through an increased rescaled length @xmath55 . and @xmath61 . comparison between results obtained using bosonization eq . ( [ eq : green-2 ] ) with luttinger liquid parameters @xmath62 and @xmath63 ( solid lines ) and using time - dependent density - matrix renormalization group ( t - dmrg ) for the bose - hubbard model ( circles ) for a quench from @xmath64 ( lattice length : @xmath65 , filling : @xmath66 , maximum number of bosons per site : @xmath67 ) . @xmath68 time evolution for different values of @xmath54 and for a fixed value of @xmath69 . the two dashed lines intersecting all @xmath54 data sets are the bounds : ( left ) @xmath70 and ( right ) @xmath71 . the colored dashed lines on the left of @xmath72 are proportional to the function @xmath73 ; while the dashed lines on the right of @xmath72 are proportional to the function @xmath74 . @xmath75 comparison between different ramp times @xmath61 for a fixed value of @xmath76 . the vertical dashed line is the bound @xmath71 . ] _ asymptotic expansion of the single - particle correlation function : _ the time evolution of single - particle correlations described by eqs . ( [ eq : green-2 ] ) and ( [ eq : theta - correlator ] ) is extremely rich . typical time evolutions of these correlations with distance are shown in panel @xmath68 of fig . [ fig : comparison ] for both the bose - hubbard model and the bosonization approach . for the chosen parameters , we found very good agreement between the two evolutions at longer distances , as long as an additional time - dependent prefactor is multiplied to the expression obtained using bosonization . this prefactor corrects for the short distance behavior which is not properly taken into account by the low energy theory . as expected , the bosonization description works best for slow and small parameter changes . in particular , deviations are observed when the mott - insulating phase of the bose - hubbard model is approached or when too many excitations are created . initially , before the slow quench begins ( at @xmath77 within our formalism ) , the correlation function decays algebraically with distance as @xmath78 . this behavior is typical of a luttinger liquid . then , as the interaction strength is slowly ramped up , the form of the correlation function evolves . for small @xmath55 and sufficiently short @xmath54 , changes are minimal as the correlation function still decays algebraically , but the exponent is now determined by the time - dependent luttinger parameter @xmath79 showing up in the exponent @xcite . this result implies that for short dimensionless distances , @xmath80 , the correlations react instantaneously to the slow interaction change and adjust to the ground state decay corresponding to the current interaction value ( see panel @xmath68 of fig . [ fig : comparison ] ) . the main contribution to this mechanism comes from quasiparticles with large momenta @xmath81 . this adiabatic regime spatially decreases with time and disappears completely when @xmath82 , where @xmath56 is the dimensionless short distance cut - off . for larger distances , the correlations deviate much more from their standard initial form and a dip appears . the formation of this dip is a clear signal of the non - equilibrium nature of the physics at play . for distances beyond this dip , the initial algebraic decay , @xmath74 , reappears as one can see in panel @xmath68 of fig . [ fig : comparison ] . the position of the dip coincides approximately with the correlation evolution front . the time - dependent position of this front can be understood by considering the propagation of quasiparticles . at any given time @xmath57 , the system hamiltonian is diagonal in its instantaneous quasiparticles as @xmath83 . assuming discrete time steps , this means that the action of the hamiltonian at time @xmath84 , diagonal in its own quasiparticles , has created ( and annihilated ) entangled quasiparticle pairs @xmath85 . these entangled quasiparticles , forming a pair , propagate with velocity @xmath86 in opposite direction and thereby carry correlations over a distance @xmath87 within a time interval @xmath88 . hence , for points separated by a distance @xmath55 larger than @xmath89 , the single - particle correlation decay is unaffected by the change in the interaction aside from an overall prefactor . for the system under study , @xmath90 and we find that @xmath91 . thus , the evolution front beyond which correlations still follow the initial algebraic decay is given by @xmath72 as evidenced in fig . [ fig : comparison ] . in particular , the position of the bound does not depend on the ramp velocity and time separately as can be seen in panel @xmath75 of fig [ fig : comparison ] . one clearly sees from there that , for a given @xmath54 , the position @xmath55 of the dip ( measured in units of @xmath92 ) is the same for different ramp times . the existence of such a propagation front is reminiscent of the light - cone - like evolution of correlations recently investigated in the context of instantaneous quenches @xcite . . exact evaluation of the bosonization expression , eq . ( [ eq : theta - correlator ] ) solid lines , is compared to the full approximate expression , eq . ( [ eq : stretched_exp_full ] ) dashed lines . for @xmath93 , we also compare the exact expression to the first exponential term of eq . ( [ eq : stretched_exp_full ] ) . in the large @xmath54 limit , if one adjusts the prefactor correctly , the stretched exponential provides a good description of the correlation decay before @xmath72 . the black dashed line indicates the position of the evolution front @xmath94 . used parameters : @xmath95 , @xmath96 ( the lower and upper cut - offs in eq . ( [ eq : theta - correlator ] ) ) and @xmath97 . ] for larger dimensionless times , as illustrated in fig . [ fig : stretched ] , an additional decay regime takes place at intermediate distances before the bound @xmath72 . this interesting behavior shows up in the bosonization approach and takes the form @xmath98 with @xmath99 a prefactor independent of @xmath55 . for intermediate @xmath54 , both exponential terms are required to adequately reproduce the behavior of eq . ( [ eq : theta - correlator ] ) as shown in fig . [ fig : stretched ] . however , for values of @xmath54 whose corresponding bound @xmath72 is located at sufficient large @xmath55 , only the first exponential term is important . in this case single - particle correlations decay with distance as a stretched exponential , a similar decay was found in ref . . such a functional form is unconventional for luttinger liquids as , typically , correlations decay algebraically in these systems . even for sudden interaction quenches in both bosonic and fermionic systems @xcite and for slow quenches in fermionic systems @xcite , only algebraic decay of correlations have been uncovered . the presence of such an unusual functional form is mainly due to the contribution of quasiparticles with momenta around @xmath100 @xcite . moreover , as the appearance of the stretched exponential decay is limited to large values of @xmath54 , this regime only occurs for relatively large parameter changes @xmath101 . it is still an open question , whether this stretched exponential decay regime arises within the bose - hubbard model . as this regime only occurs for large parameter changes , the tll model might not describe properly the dynamics of the bose - hubbard model and relaxation mechanisms not present in the tll model might dominate the evolution . a careful analysis of this last point would be extremely valuable but is left to further studies . _ experimental implementation and detection : _ one - dimensional interacting bosonic gases have been realized experimentally using a few different setups @xcite . the best scheme to implement time variations of the interaction strength depends on the actual experiment setup . one method relies on the use of a feshbach resonance by which the scattering length , and thereby the interaction strength , is directly varied . a second technique is based on the application of a weak optical lattice suppressing the tunneling amplitude in order to change the ratio of the interaction strength to the kinetic energy . finally , in one - dimensional wave guides , geometric confinement can be employed to tailor the interaction strength . the detection of the single - particle correlation function can also be carried out experimentally . using radio - frequency pulses , atoms can be outcoupled from the one - dimensional bose gas at two spatially separated positions and their interference is then observed after a free fall . this technique was successfully employed to measure the build - up of equal - time single - particle correlations in a bose - einstein condensate after a sudden decrease of its temperature and the critical correlations at the phase transition @xcite . another possible detection scheme relies on time - of - flight ( tof ) measurements which provide , in the far - field limit , access to the momentum distribution @xmath102 . the very long distance behavior of the single - particle correlation is dominated by the luttinger liquid power - law ; however , at a critical wavevector , @xmath103 , determined by the ballistic expansion condition , a crossover occurs and @xmath104 is dominated by the fourier transform of the stretched exponential . therefore , at @xmath105 ( with @xmath106 the atom mass ) a crossover should be visible in the tof measurements . we foresee that one of the main challenges towards the observation of the evolution of correlations will be the realization of a relatively homogeneous gas as inhomogeneities can cause mass transport and mask the internal evolution @xcite . _ conclusion : _ we carefully studied the dynamics of single - particle correlations arising during the slow interaction quench of a one - dimensional bose gas using both a bosonization approach and numerical simulations for the bose - hubbard model . we uncovered various regimes : for short distances , we found single - particle correlations to decay algebraically with an exponent which depends on the ramp time ; while for intermediate distances but at long times , the decay of correlations follows a stretched exponential . this unconventional regime is bounded by an evolution front , reminiscent of a lieb - robinson bound , whose dynamics is dictated by the counter - propagation of entangled quasiparticle pairs . the existence of this regime within the bose - hubbard model remains an open question . deviations between the bose - hubbard and luttinger models were in fact found for fast as well as for very long ramps . finally , beyond this evolution front , correlations decay algebraically with an exponent given by the sudden approximation . these results may serve as a basis for comparison with future experimental studies of unconventional time evolutions in many - body one - dimensional systems . we are grateful to t. giamarchi for helpful discussions and to g. roux for his insights on related works . we acknowledge support from anr ( famous ) , snfs under manep and divison ii , nserc of canada and cifar . 46 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , _ _ ( , , ) . , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , eds . , _ _ ( , , ) . equivalently , the solution can be expressed using airy functions , but the bessel function representation leads to a more straightforward derivation of asymptotics . , , , , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , * * , ( ) . supplementary material
omitting o and b - type stars in consideration of their short lifespans and strong ionizing winds ( e.g. , * ? ? ? * ; * ? ? ? * ) , as well as the transitory case of a - type stars , f - type main - sequence stars represent the hot limit of stars with a significant potential for providing circumstellar habitable environments ( e.g. , * ? ? * ; * ? ? ? generally , the investigation of habitability around different types of stars , particularly main - sequence stars , is considered a theme of pivotal interest to the thriving field of astrobiology ; see , e.g. , @xcite , @xcite , @xcite , @xcite , and @xcite for previous studies and reviews . for planetary host stars like any other star the stellar mass determines their lifetime ( shorter with larger mass ) , luminosity and effective temperature ( both higher with larger mass ) . main - sequence stars moderately more massive than the sun with masses between 1.1 and 1.6 @xmath0 , i.e. , f - type stars , are of particular interest as hosts to exosolar planets and exomoons in orbit about those planets . compared to stars of later spectral types , f - type stars are characterized by relatively large habitable zones , although from a general astrobiological point of view , they exhibit the adverse statistical property of being less frequent ( e.g. , * ? ? ? * ; * ? ? ? * ) . on the other hand , despite their reduced lifetimes compared to g - type stars , their lifetimes still exceed several billion years ( e.g. , * ? ? ? * ) , allowing the principle possibility of exobiology , potentially including advanced life forms . while habitable zones and their evolution can , in general , be well characterized in terms of the total amount of stellar irradiation , the spectral energy distribution ( including its portion of energetic radiation ) may nevertheless be significant as well for the facilitation of habitability ( e.g. , * ? ? ? the emergent radiation of f - type stars consists of significantly larger amounts of uv compared to the sun thus entailing potentially unfavorable effects on planetary climates and possible organisms ( e.g. , * ? ? ? previous studies showed that increased levels of uv , as well as the even more energetic euv radiation , can trigger a variety of chemical planetary atmospheric processes , including exoplanetary atmospheric evaporation ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? hence , the radiation output provided by f - stars places an additional constraint on circumstellar habitability , which needs to be considered as part of more comprehensive assessments ; see , e.g. , @xcite for previous results . it is the aim of the present study to consider some of these processes in an approximate manner while also taking into account the evolutionary status of the f - type host stars . in sect . 2 , we comment on the concept of habitability , which also includes a description of the climatological habitable zone . additionally , we discuss the governing equations of the present study involving both the dna action spectrum and planetary atmospheric attenuation . in sect . 3 , we describe the spectral energy distribution of the host stars based on sophisticated photospheric models and their spectral energy output computed by the phoenix code for the range of effective temperatures relevant to f - type stars . the specific effective temperature and total luminosity of each star for a given mass and age are derived from well - tested stellar evolution models , which also convey the time scales of the circumstellar conditions during stellar main - sequence evolution as well as the extents of the climatological habitable zones . results and discussion are given in sect . 4 , which focuses on habitability gauged by the damage inflicted upon dna . particular emphasis is placed on the relevance of the different types of uv ( i.e. , uv - a , uv - b , and uv - c ; see sect . 2.2 for definitions ) . our summary and conclusions are given in sect . a key aspect in the study of circumstellar habitability is the introduction of the climatological habitable zone , a concept , evaluated by @xcite . they utilized 1-d climate model to estimate the position and width of the habitable zone around solar - like stars as well as other types of main - sequence stars . the basic premise consists in assuming a earth - like planet with a co@xmath1/h@xmath1o / n@xmath1 atmosphere and , furthermore , that habitability requires the presence of water on the planetary surface . in their work they distinguished between the _ conservative _ habitable zone ( chz , with limits of 0.95 and 1.37 au ) and the _ general _ habitable zone ( ghz , with limits of 0.84 and 1.67 au ) ; subsequent work about the ghz has also been given by @xcite and others . the physical significance of the various kinds of hzs obtained by @xcite are given as follows : the ghz is defined as bordered by the runaway greenhouse effect ( inner limit ) and the maximum greenhouse effect ( outer limit ) . concerning the latter it is assumed that a cloud - free co@xmath1 atmosphere shall still be able to provide a surface temperature of 273 k. by contrast , the inner limit of the chz is defined by the onset of water loss . in this case , a wet stratosphere is assumed to exist where water is lost by photodissociation and subsequent hydrogen escape to space . furthermore , the outer limit of the chz is defined by the first co@xmath1 condensation attained by the onset of formation of co@xmath1 clouds at a temperature of 273 k ; see , e.g. , @xcite for additional details and applications . owing to the shape of the photospheric spectra and the total amount of the radiative energy fluxes , the limits of the habitable zones are known to depend both on the stellar effective temperatures @xmath2 and the luminosities @xmath3 . the results for the chzs and ghzs as well as the appropriate earth - equivalent positions for main - sequence stars between spectral type f0 and g0 are given in fig . 1 . for f0 v stars , the chz extends between 2.27 and 2.92 au and the ghz extends between 1.99 and 3.67 au . furthermore , for f8 v stars , the chz extends between 1.29 and 1.80 au and the ghz extends between 1.14 and 2.21 au . the corresponding stellar data are given in table 1 . they are those also adopted for the photospheric models computed with the phoenix code ( see sect . 3.1 ) , which have subsequently been used for our astrobiological studies . another aspect of habitability concerns the limits of the climatological habitable zones for the various types of stars . they have been calculated based on the formalism by @xcite . it provides a suitable polynomial fit and , furthermore , implements the required correction for the solar effective temperature in consideration of that @xcite used for an unusually low value of 5700 k instead of 5777 k as currently accepted . next we comment on the relevance of the dna action spectrum . the most fundamental radiometric technique to quantify radiative damage on biomolecules and microorganisms is spectroradiometry . biological effectiveness spectra can be derived from spectral data by multiplication with an action spectrum @xmath4 of a relevant photobiological reaction with the action spectrum typically given in relative units normalized to unity for , e.g. , @xmath5 nm . the biological effectiveness for a distinct range of the electromagnetic spectrum such as uv radiation is determined by @xmath6 where @xmath7 denotes the stellar irradiance ( ergs @xmath8 s@xmath9 nm@xmath9 ) , @xmath10 the wavelength ( nm ) , and @xmath11 the planetary atmospheric attenuation function ; see @xcite . here @xmath12 and @xmath13 are the limits of integration which in our computations are set as 200 nm and 400 nm , respectively . although a significant amount of stellar radiation exists beyond 400 nm , this portion of the spectrum is disregarded in the following owing to the minuscule values for the action spectrum @xmath4 in this regime . planetary atmospheric attenuation , sometimes also called extinction , results in a loss of intensity of the incident stellar radiation . in eq . ( 1 ) @xmath14 = 1 indicates no loss and @xmath14 = 0 indicates a complete loss ( see sect . 2.3 ) ; note that @xmath11 can be attained by various types of methods , which may also consider detailed atmospheric photochemical models . based on previous work , the uv region of the electromagnetic spectrum has been divided into three bands termed uv - a , uv - b , and uv - c . the subdivisions are somewhat arbitrary and differ slightly depending on the discipline involved . here we will use : uv - a , 400 - 320 nm ; uv - b , 320 - 290 nm ; and uv - c , 290 - 200 nm . following @xcite , the division between uv - b and uv - c is chosen as 290 nm since uv at shorter wavelengths is unlikely to be present in terrestrial sunlight , except at high altitudes @xcite . the choice of 320 nm as the division between uv - b and uv - a is suggested by the level of photobiological activity although subdivisions at 330 or 340 nm have previously also been advocated @xcite . in order to compute the irradiance toward targets in circumstellar environments , typically positioned in stellar habitable zones , a further equation is needed , which is @xmath15 where @xmath16 is the stellar radiative flux , @xmath17 is the stellar radius and @xmath18 is the distance between the target and the star . in the framework of this work , we will focus on planets at different positions in stellar hzs . for stellar uv radiation we will consider photospheric radiation only ( see sect . 3 ) because the chromospheric uv radiation from f - type stars is of minor importance @xcite . action spectra for dna , also to be viewed as weighting functions , have previously been utilized to quantify and assess damage due to uv radiation @xcite . besides dna , action spectra have also been derived for other biomolecules , for biostructures such as cellular components as well as for distinct species , especially extremophiles ( e.g. , * ? ? ? * ; * ? ? ? * and references therein ) . @xcite provides information on the dna action spectrum for the range from 285 nm ( uv - c ) to 400 nm ( uv - a ) . she found that between 400 and 300 nm , the action spectrum increases by almost four orders of magnitude ( see fig . 2 ) . the reason for this behavior is the wavelength - dependency of the absorption and ionization potential of uv radiation in this particular regime . a further significant increase in the dna action spectrum occurs between 300 and 200 nm ( see below ) . fortunately , however , the earth s ozone layer is very sufficient to filter out this type of lethal radiation ( e.g. , * ? ? ? * ; * ? ? ? * ) , which in our models is mathematically dealt with by considering an appropriate attenuation function @xmath11 ( see eq . 1 ) . the behavior of the dna action spectrum in the wavelengths regime between 200 and 290 nm has been given by @xcite . he points out that a significant reason for the susceptibility to uv - induced damage are @xmath19-electron systems , notably because of their @xmath19 to @xmath20 energy transitions . it is found that this type of interaction also accounts for protein damage and damage to enzymes of photosynthesis , which indicates that these types of biochemical reactions are of general importance . the increase in the relative biological damage is about a factor of 35 relative to the damage at the reference wavelengths of 300 nm ( see fig . 2 ) . a relevant ingredient to our study is the consideration of planetary atmospheric attenuation @xmath11 , which typically results in a notable reduction of the received stellar radiation . appropriate values for @xmath11 can be obtained through the analysis of theoretical exoplanetary models ( e.g. , * ? ? ? * ; * ? ? ? * ) , inspired by recent results from the _ kepler _ mission , or the usage of historic earth - based data ( e.g. , * ? ? ? within the scope of the present work that is mostly focused on the impact of photospheric radiation from stars of different spectral types , as well as on the role of active and inactive stellar chromospheres , we assume that @xmath11 is given by a parameterized attenuation function att defined as @xmath21 \ \ ; \ ] ] see fig . 3 for a depiction of different examples @xmath22 . here @xmath23 denotes the start - of - slope parameter , @xmath24 ( in nm ) the center parameter , and @xmath25 the maximum ( limited to unity ) of the distribution . for example , @xcite provided information about the ultraviolet irradiance reaching the surface of archean earth for various assumptions about earth s atmospheric composition ; the latter allow us to constrain the wavelength - dependent attenuation coefficients for earth 3.5 gyr ago . furthermore , there is a large array of recent studies about exosolar planetary atmospheres , including those for rocky planets . they encompass models regarding the detailed treatment of atmospheric photochemistry , including the build - up and destruction of ozone , as discussed by , e.g. , @xcite , @xcite and @xcite . these models provide information on different exoplanetary structures due to outside forcings , including variable stellar radiation , which in principle allow the derivation of detailed planetary atmospheric attenuation functions . in accordance to previous studies of our group , as given by @xcite and @xcite , which mostly focused on super - earth planets , we rely again on selected stellar evolution models that have been computed with the well - tested eggleton code ; see , e.g. , @xcite for a description of an adequate solar evolutionary model . the eggleton code allows us to take into account the changing properties of the host star during its evolution on the main - sequence and , subsequently , as a red giant . we use an advanced version of the eggleton code , including updated opacities and an improved equation of state as described by @xcite . besides other desirable characteristics , the adopted evolution code utilizes a self - adapting mesh and also permits treating overshooting " a concept of extra mixing , which has been thoroughly tested considering observational constraints . in particular , the two convection parameters , the `` mixing length '' and the `` overshoot length '' , have been calibrated by matching accurately the physical properties of various types of stars , including giants and supergiants of well - known masses , found in well - studied , eclipsing binary systems @xcite . for the abundance of heavy elements , which decisively affect the opacities , we use the near - solar value of @xmath26 . this choice is an appropriate representation of present - day samples of stars in the thin galactic disk , noting that they exhibit a relatively narrow distribution ( @xmath27 to 0.03 ) about this value . the adopted evolution code also considers a detailed and well - tested description of the stellar mass loss , which becomes important for the final stages of giant star evolution ; see @xcite and @xcite . regarding our models of planetary host stars , the principal input parameters are the total luminosity and effective temperature , which are found to change with time . obviously , the resulting total lifetime of the star is a quantity of high significance as well . while the effective temperature and luminosity of the host star already allow a good representation of circumstellar habitability , as done through the stellar climatological habitable zone ( see sect . 2.1 ) , the irradiation especially in the uv regime is of pivotal importance as well for arriving at a realistic evaluation of habitability . the main constraint arises from having sustainable conditions for biological organisms and biochemical processes , which provide the basis of life . here , damaging ultraviolet radiation must be of particular concern . its share regarding the total stellar luminosity critically depends on the stellar effective temperature and increases significantly from late ( f9 ) to early ( f0 ) f - type stars . we employ the necessary and accurate account of stellar radiation , including its spectral energy distribution , by utilizing a number of photospheric models computed by the phoenix code following @xcite ; see fig . the adopted range of models for the f - type stars are in response to effective temperatures of 7200 k for spectral type f0 , 7000 k for f1 , 6890 k for f2 , 6700 k for f3 , 6440 k for f5 , 6200 k for f8 , and 6050 k for g0 . the phoenix code iterates the principal physics and structure of a stellar atmosphere until a final model is obtained , which is in radiative and hydrostatic equilibrium ; see @xcite . as part of this procedure , energy transport by convection is also considered . the phoenix code solves the equation of state , including a very large number of atoms , ions , and molecules . with respect to radiation , the one - dimensional spherically symmetrical radiative transfer equation for expanding atmospheres is solved , including a treatment of special relativity ; see @xcite . opacities are sampled dynamically over about 80 million lines from atomic transitions and billions of molecular lines , in addition to background ( i.e. , continuous ) opacities , to provide an emerging spectral flux as reliable and realistic as possible . as part of our study , all spectra emergent from a stellar model atmosphere have first been calculated with a high resolution of @xmath28 , for a highly complete inclusion of the lines . the spectra were then binned down to a much lower resolution , which is more practical for our subsequent astrobiological analyses . main - sequence stars , including the sun , represent the slowest phase of stellar evolution , since at that stage the largest energy reservoir of the star is consumed : its central hydrogen is converted to helium ; see fig . 5 for examples of f - star evolutionary tracks . all later phases , where the star becomes a red giant twice , before and after the ignition of the central helium burning , are much faster and present much larger changes in stellar luminosity ; see @xcite for a detailed study of the implications for habitable super - earths . hence , as previously found , main - sequence stars are , in general , most promising in the context of astrobiology . though there are limits , since an increasing mass accelerates stellar evolution considerably . nonetheless , f - type stars corresponding to the mass - range of about 1.1 to 1.6 @xmath0 , provide stable lifetimes of 2 to 4 billion years , expected to be sufficient for the origin and evolution of life . however , regarding the evolution of effective temperatures and , in consequence , of the spectral distribution of the emergent radiation , f - type stars differ somewhat from the sun , especially the most massive ones : their effective temperatures rise to much higher values while still staying on the main - sequence , and there is a quick fall afterwards , and a rise again toward the end of this phase . the reason for this evolutionary behavior resides in the cores of f - stars : while the solar core is not convective but relies purely on radiative energy transport , f - stars employ the more efficient process of convection for transporting their high amounts of produced energy to the outer layers . inside the most massive f - stars , where core convection has gained full power , rising bubbles even overshoot the boundary set by the schwarzschild criterion . hence , as a by - product , the convective cores of f - stars benefit from an enhanced chemical mixing by gaining access to a larger hydrogen reservoir around them , within reach of the `` overshoot length '' . therefore , f - type stars still spend a relatively long time on the main - sequence , making good in part for their faster evolution in general . this phenomenon is most notable between 1.4 and 1.5 @xmath0 , where overshooting sets in . however , the evolutionary behavior of f - stars after the end of central hydrogen burning also differs a bit from that of the sun , which does neither experience overshooting nor any core convection . these particulars of stellar evolution determine both the stellar effective temperatures and luminosities ( see table 1 ) ; they also drive the changes of the spectral characteristics of the irradiation and , finally , the inner and outer limit of the respective habitable zones . compared to g - type and k - type stars , f - type stars move relatively fast through the final stages of evolution beyond the main - sequence . in those stages , very dramatic changes in luminosity ( increasing by over a factor 10@xmath29 ) , stellar radius ( increasing by over a factor 100 ) and effective temperature ( decreasing by over a factor 2 ) occur . the corresponding time scales are limited to a few hundred million years compared to the sun , which will take about 2 billion years for completing these steps . for these reasons , we will restrict the study of circumstellar habitability of f - stars to their phases on and near the main - sequence . next we focus on the evolution of f - star climatological habitable zones , and the subsequent identification of the inner and outer limits of the chzs and ghzs . as pointed out in sect . 3.2 , f - type stars , alike the sun , encounter a slow but consistent growth of luminosity during their main - sequence phase as well as characteristic changes of the stellar effective temperatures . this is relevant for all types of f - stars , i.e. , for the entire range of masses between about 1.2 and 1.5 @xmath0 as addressed in the following . these masses are identified as such at the zero - age main - sequence ( zams ) , and they experience little change prior to the departure of the star toward the red giant domain ; the latter is outside the scope of this study . an example is given as figure 6 . it shows the evolution of the climatological habitable zone for a star with an initial mass of 1.3 @xmath0 . it also depicts the development of the earth - equivalent position given as @xmath30 ( with @xmath3 as stellar luminosity in solar units ) , which is progressively moving outward . note that by the end of central hydrogen burning , the total stellar radiation output has almost doubled , entailing a considerable increase of both the inner and outer limits of the climatological habitable zone , identified either as chz or ghz . furthermore , the stellar spectral types change due to alterations of their effective temperatures during the course of stellar evolution , including main - sequence evolution . in the latter case , stars of initial masses between 1.2 and 1.5 @xmath0 stay within the f - type range ( see table 2 ) . although it is appropriate to explore uv - based habitability at any distance from a planetary host star , we rather select the inner and outer limits of the chz ( labelled as ic and oc , respectively ) , and the inner and outer limits of the ghz ( labelled as ig and og , respectively ) as venues of our study . results are given in table 3 and figure 7 . we especially consider the extreme positions of the ig , ic , oc , and og attained as consequence of the stellar main - sequence evolution , i.e. , as identified at the zams stage and when the stars depart from the main - sequence . our study indicates once again that when stars age the climatological habitable zones become broader and migrate outwardly . a special aspect of significance to astrobiology concerns the continuous domains of both the chzs and ghzs , also referred to as continuously habitable zones ( see fig . 7 ) , where the criteria for the chzs and ghzs ( see sect . 2.1 ) are fulfilled for a distinct period of time ( here : the stellar main - sequence stages ) . the continuously habitable zones have evidently much smaller widths than the habitable zones defined by the extremes for the same amounts of time . the continuous domains of habitability for our set of stars are particularly small for the chzs . for a 1.2 @xmath0 this domain extents from 1.366 to 1.688 au , whereas for a 1.5 @xmath0 it extents only from 2.535 to 2.566 au ; this aspect should be taken into account for any comprehensive habitability assessment . the aim of our study is the investigation of biological damage inflicted upon dna due to stellar uv radiation for objects around main - sequence f - type stars . our results will be given in terms of @xmath31 , a parameter defined as the ratio of damage for a given distance from the star for an object with or without atmospheric attenuation relative to the damage for an object at 1 au from a solar - like star without atmospheric attenuation present . we will focus on the following aspects : ( 1 ) influence of spectral types of the host stars , especially for f0 v , f2 v , f5 v , and f8 v , ( 2 ) relevance of planetary positions in the stellar habitable zones , including the general and conservative inner limits , earth - equivalent positions , and the general and conservative outer limits , ( 3 ) effects of planetary atmospheric attenuation approximated by analytic functions , and ( 4 ) the relative importance of uv - a , uv - b , and uv - c regarding the dna damage . the dna damage in the habitable zones around f0 v , f2 v , f5 v , and f8 v stars in cases without planetary atmospheric attenuation is found to be almost always more significant than for the reference object at earth distance from a solar - like star ( see figure 8) . the sole exception is for objects at the outer limit of the ghz around an f8 v star , which in the solar system would correspond to an orbit beyond mars ; here @xmath31 is given as 0.95 . planetary atmospheric attenuation effectively reduces the damage of dna , as expected ( see figure 9 ) . in this work , as a standard example adopted for tutorial reasons , the planetary attenuation parameters are chosen as @xmath32 , @xmath33 , and @xmath34 , except otherwise noted . in this case of attenuation , the damage on dna , for positions at the inner limit of the ghzs , is drastically reduced compared to the reference object without any attenuation . for other positions in the stellar habitable zones , the reductions are relatively high as well ( see figure 9 ) . it is particularly intriguing to compare the respective values of @xmath31 for f0 v , f2 v , f5 v , and f8 v stars for different positions in the stellar habitable zones as they indicate the role of the significantly higher uv fluxes of f - type stars compared to the photospheric radiation of a solar - like star . for the inner limits of the ghzs ( i.e. , the closest positions to the stars considered ) , @xmath31 for f0 v , f2 v , f5 v , and f8 v stars are identified as 9.8 , 7.3 , 4.2 , and 3.6 without atmospheric attenuation ; they decrease to 0.23 , 0.19 , 0.14 , and 0.14 , respectively , when our default choice of atmospheric attenuation is applied . at earth - equivalent positions ( which depend on the stellar spectral type much like the above - mentioned limits of habitability ) , @xmath31 for each star are given as 7.1 , 5.2 , 3.0 , and 2.5 , respectively , without atmospheric attenuation , and as 0.16 , 0.14 , 0.10 , 0.10 with attenuation . at the outer limits of ghzs , @xmath31 for each star is given as 2.9 , 2.1 , 1.1 , and 0.95 without atmospheric attenuation , and as 0.067 , 0.055 , 0.038 , and 0.037 with attenuation . intermediate values are obtained for the inner and outer limits of the chzs . in all cases dna damage due to uv - a , compared to uv - b and uv - c , is minuscule at best . @xmath31 due to uv - a at the inner limit of the ghz for an f0 v star is @xmath35 without atmospheric attenuation , which is the maximum value of all cases considered . this value is further reduced to @xmath36 in case of default atmospheric attenuation . of the three regimes of uv , the one most affected by our above choice of planetary attenuation is uv - c . for f0 v , f2 v , f5 v , and f8 v , the dna damage due to uv - c is 96% , 95% , 93% , and 91% , respectively , of the total for the uv regimes combined . if atmospheric attenuation is included , the dna damage due to uv - c is reduced to 1.7% , 1.9% , 2.3% , and 2.6% of the damage due to uv - c without planetary attenuation . it comprises 70% , 68% , 65% , and 61% of the total for each case of attenuation . the relative damage attributable to uv - c decreases with stellar types from f0 v to f8 v. the damage due to uv - a and uv - b also decreases with stellar types from f0 v to f5 v , but it slightly increases between f5 v and f8 v because of the shape of the stellar photospheric spectrum ( see figure 5 ) . since atmospheric attenuation is expected to impact preferably the uv regime , which in turn is most relevant to the dna damage , we also explored the effects of parameter choices for the attenuation function att ( see eq . 3 ) . for tutorial reasons , we focused on the f2 v star , and varied one of the parameters , i.e. , @xmath23 , @xmath24 , or @xmath25 , at a time . figure 10 depicts two cases out of several combinations of fixed parameters that were investigated . in the top panels of figure 10 , the parameter @xmath23 is varied from 0.01 to 1.0 , whereas the other parameters remain fixed at @xmath24 = 300 and @xmath25 = 0.5 . in the bottom panels , the parameter @xmath24 is varied from 250 to 350 and the two other parameters remain fixed at @xmath23 = 0.05 and @xmath25 = 0.5 . the two panels on the left show the relative impact of uv - a , uv - b , and uv - c on dna at an earth - equivalent position . note that the impact of uv - a is , however , completely unrecognizable in all panels . the two panels to the right show the dna damage at five selected positions in the habitable zone of the f2 v star , which are : the inner limits of the ghzs and chzs , the earth - equivalent position , and the outer limits of the chzs and ghzs . it is found that damage inflicted on dna is considerably larger for objects closer to the star . thus , in fig . 10 , the top two lines represent the inner limits of the chz and ghz , and the bottom two lines represent the outer limits of the chz and ghz . the parameter @xmath23 depicts the rate of change in att with increasing wavelength ( see fig . thus , a smaller value of @xmath23 corresponds to a gentler slope of att , but also to a considerably higher value of att at relatively short wavelength ( i.e. , near 200 nm ) , which implies a higher effect of attenuation where damage inflicted on dna is most severe ( see fig . 2 ) . the parameter @xmath24 describes the center wavelength of the attenuation function ; it also changes the response of each regime to varied values of @xmath23 . thus , the appearance of @xmath31 as function of the attenuation parameter @xmath23 heavily depends on the fixed value of parameter @xmath24 . for example , let us consider the case @xmath33 . here , the dna damage due to uv - a increases , while the damages due to uv - b and uv - c decrease as the parameter @xmath23 increases ( see fig . 10 ) . here @xmath31 decreases from 0.73 to 0.0074 for an earth - equivalent position with increasing @xmath23 . the largest changes occur between @xmath37 and 0.08 , where @xmath31 due to uv - c declines drastically . in these cases , especially for our standard choice of attenuation ( see sect . 4.1 ) , most of the dna damage is attributable to uv - c . but for @xmath38 , @xmath31 due to uv - b and due to uv - c are about the same , and for larger values of @xmath23 , @xmath31 due to uv - b becomes greater than @xmath31 due to uv - c . we also varied the parameter @xmath24 while keeping the values for @xmath23 and @xmath25 fixed . @xmath31 is identified as a decreasing function for variable values of @xmath24 . since parameter @xmath24 determines the center of the attenuation function , the effect of uv diminishes from the lowest wavelengths as @xmath24 is increased . in other words , uv - c is affected the most by varying @xmath24 . if @xmath32 and @xmath34 are used , @xmath31 decreases from 1.32 at @xmath39 to 0.14 at @xmath33 and to 0.0012 at @xmath40 for objects at earth - equivalent positions . this behavior is shown in figure 10 . if @xmath41 and @xmath34 are used , the appearance of each functional dependence is similar to the case of @xmath32 ; however , @xmath31 of uv - b exceeds that of uv - c at @xmath42 . the functional dependence for uv - c is steeper for @xmath41 than for @xmath32 . the total amount of @xmath31 decreases from 1.36 at @xmath39 to 0.048 at @xmath33 and to @xmath43 at @xmath40 for objects at the earth - equivalent position . dna damage is inversely proportional to the square of distance between a planet and the host star , and thus , the ratio of dna damage ( with or without the consideration of atmospheric attenuation ) at one position to another does not change with the selected combination of parameters . because of that , if @xmath31 , the ratio of dna damage at a position to the reference value , at different positions are compared , the differences are smaller for more effective combinations of parameters . for example , the dna damage at the inner limits of the ghzs and the chzs , and at the outer limits of the chzs and ghzs are 139% , 108% , 62% , and 40% of dna damage as measured by @xmath31 , respectively , as found for the f2 v star . if attenuation is considered , and if the attenuation parameters are chosen as @xmath44 , @xmath31 is found as 1.01 , 0.78 , 0.73 , 0.45 , and 0.29 at the inner limit of the ghz and chz , at the earth - equivalent position , and at the outer limit of the chz and ghz , respectively . by contrast , if the same parameters are set as @xmath45 , our standard choice ) , @xmath31 is identified as 0.0119 , 0.0092 , 0.0086 , 0.0053 and 0.0034 , respectively . we also studied the influence of stellar main - sequence evolution , resulting in pronounced changes in the stellar effective temperatures and luminosities regarding the damage inflicted on dna . again , we consider cases without and with planetary atmospheric attenuation ( see sect . 4.1 ) . specifically , we explore the change in @xmath31 at specific positions in the habitable zones for stars with masses of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 . the selected positions are the general outer limit at zams ( bottom dotted line in figures 11 and 12 ) , the conservative outer limit at zams ( bottom dashed line ) , the general inner limit at the end of main - sequence ( top dotted line ) , the conservative inner limit at the end of main - sequence ( top dashed line ) , the earth - equivalent positions at zams ( i.e. , minimum distance ; top solid line ) and the end of main - sequence ( i.e. , maximum distance value ; bottom solid line ) ; see sect . 3.3 . the distances for various positions are given in table 4 and figure 7 . furthermore , we also consider average ( i.e. , time - independent ) earth - equivalent positions , which are derived by interpolating the conservative outer limit at the zams and the conservative inner limit at the end of main - sequence evolution noting that these limits correspond to the continuous domains of the chzs . generally , the earth - equivalent positions are computed as weighted averages between the inner and outer limits of the chzs , which can be approximated as @xmath46 and @xmath47 ( with @xmath3 in solar units ) , respectively , with the earth - equivalent positions given as @xmath30 @xcite . for stars with masses of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 , the average earth - equivalent positions are obtained as 1.56 , 1.82 , 2.10 , and 2.54 au , respectively , which in the solar system would correspond to an approximate distance nearly between mars and ceres . moreover , we also consider evolving earth - equivalent positions , noting that the inner and outer limits of chzs change on evolutionary time scales ( see fig . 7 ) ; see discussion below . in principle , the damage inflicted on dna at any position within the stellar habitable zones at the zams increases as a function of the stellar mass ( see figures 11 and 12 ) . regarding the inner and outer limits of both the chzs and ghzs , the damage on dna first increases with evolutionary time , but then starts decreasing with time , while the stellar luminosity keeps increasing and both the chzs and ghzs continue to migrate outward . if no planetary attenuation is considered , the zams @xmath31 values at average earth - equivalent positions for stars of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 are found as 0.91 , 2.46 , 3.41 , and 5.02 , respectively . the maximum values for the dna damage , expressed as @xmath31 , at average earth - equivalent positions are given as 1.96 , 2.63 , 3.48 , and 5.07 , respectively . these values are obtained at stellar evolutionary times of 1.95 , 0.93 , 0.27 , and 0.14 gyr , respectively ; it means that the maximum amounts of damage inflicted on dna are attained much earlier for stars of higher mass ( and , by implication , of earlier spectral type and higher initial effective temperature ) . at the end of main - sequence evolution , the @xmath31 values at average earth - equivalent positions are then reduced to 0.96 , 1.04 , 1.07 , and 0.32 , respectively ; they are comparable to or smaller than those for a reference object orbiting a solar - like star . if default planetary atmospheric attenuation is assumed , corresponding to @xmath48 , the damage is drastically reduced , i.e. , by up to 96% , 97% , 97% , and 98% for stars of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 , respectively . the precise amount of reduction is a function of the stellar evolutionary status . the amount of reduction moreover depends on the type of star , which determines the shape of the emergent spectrum , therefore entailing different amounts of damage . as previously discussed , the damage is dominated by the uv - c regime ( see eq . 1 and fig . if standard attenuation is considered , the zams values for @xmath31 at average earth - equivalent positions for stars of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 are reduced to 0.038 , 0.082 , 0.098 , and 0.121 , respectively . the maximum @xmath31 values at those positions are given as 0.075 , 0.087 , 0.099 , and 0.122 , whereas at the end of the main - sequence stages , they are further reduced to 0.040 , 0.041 , 0.041 , and 0.014 , respectively . for specific age intervals , @xmath31 changes in different ways in the vicinities of stars of different masses . the following data refer to objects at average earth - equivalent positions , but they can also be converted for other star planet distances in a highly straightforward manner . we are particularly interested in the change of @xmath31 between 0.5 gyr and 2.5 gyr , i.e. , during the early stages of the systems , contemporaneous with the time when life originated on earth ( e.g. , * ? ? ? * and references therein ) , for a star with masses of 1.2 @xmath0 , @xmath31 increases from 1.75 to 1.96 until 1.95 gyr , and thereafter decreases to 1.92 at the average earth - equivalent position if planetary atmospheric attenuation is absent . for a star with masses of 1.3 @xmath0 , @xmath31 increases from 2.61 to 2.63 until 0.93 gyr , and then decreases to 1.86 . for stars with masses of 1.4 @xmath0 and 1.5 @xmath0 , @xmath31 decreases from 3.46 to 1.10 and decreases from 4.88 to 0.52 , respectively . thus , a planet in the habitable zone of a 1.2 @xmath0 mass star will experience the least change in the amount of dna damage , whereas that change will be greatest for a 1.5 @xmath0 mass star . adequate models should also take into account effects due to planetary atmospheric attenuation . in this case , even the climatological habitable zone of a 1.5 @xmath0 mass star may be able to offer a relatively well uv protected environment throughout the early 2 gyr period . the same statement is expected to apply for f - type stars of lower mass as well . if our default model of atmospheric attenuation is adopted , @xmath31 for a 1.2 @xmath0 mass star increases from 0.069 to 0.075 until 1.95 gyr , and then decreases to 0.074 ; note that these data refer to average earth - equivalent positions . for a 1.3 @xmath0 mass star , @xmath31 for the same setting increases from 0.086 to 0.087 until 0.93 gyr , and then decreases to 0.067 . in comparison , for stars with masses of 1.4 and 1.5 @xmath0 , @xmath31 decreases from the maximum value of 0.099 to 0.042 and from 0.118 to 0.021 , respectively . for other planetary positions , the respective values for @xmath31 can be obtained through appropriate scaling . moreover , we also computed the dna damage at evolving ( i.e. , time - dependent ) earth - equivalent positions , depicted as dashed lines in fig . 7 . for all cases from 1.2 to 1.5 @xmath0 , the earth - equivalent positions at zams are located very close to the stars . in fact , they are situated interior to the continuous domains of both the chzs and ghzs . as the stars age , the earth - equivalent positions migrate outward , crossing or even passing the continuous domains of the chzs , especially for relatively massive f - type stars . since the damage inflicted on dna is proportional to the square of distance from the star , the attained results reflect the behavior of the planetary positions . the maximum dna damage at evolving earth - equivalent positions without atmospheric attenuation for stars of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 are given as 2.55 , 3.56 , 4.74 , and 7.74 , respectively , occurring at very early stages of the stellar lifetimes , whereas the minimum dna damage are given as 0.92 , 0.99 , 1.01 , and 0.29 , respectively , obtained at the end of main - sequence evolution ( see fig . damage on dna is reduced by planetary atmospheric attenuation , as expected . for default attenuation , the maximum damage is found as 0.099 , 0.119 , 0.136 , and 0.186 , respectively , whereas the minimum damage is found as 0.038 , 0.040 , 0.039 , and 0.012 , respectively ( see fig . in this study , we investigated the general astrobiological significance of f - type main - sequence stars . dna has been taken as a proxy for carbon - based macromolecules following the paradigm that extraterrestrial biology may most likely be associated with hydrocarbon - based biochemistry . consequently , the dna action spectrum was utilized to describe the impact of the stellar uv radiation . we considered an array of important aspects , including ( 1 ) the role of stellar main - sequence evolution , ( 2 ) the situation for planets at different positions within the stellar habitable zones , and ( 3 ) the general influence of planetary atmospheric attenuation , which has been described based on a parameterized attenuation function . the damage on dna was described by the output parameter @xmath31 , defined as the ratio of damage for a given distance from the star for a general object relative to the damage for an object at 1 au from a solar - like star with attenuation absent . \(1 ) for average earth - equivalent planetary positions , located inside the continuous domains of the chzs , @xmath31 for stars of spectral type f0 v , f2 v , f5 v , and f8 v are obtained as 7.1 , 5.2 , 3.0 , and 2.5 , respectively . earth - equivalent planetary positions depend on the stellar spectral type , and are at greater distance for stars of higher effective temperature or , by implication , larger mass . these results are consistent with the earlier work by @xcite . \(2 ) for the inner and outer limits of the chzs as well as ghzs , the results for @xmath31 can be obtained by scaling . specifically , the damage inflicted on dna is considerably increased at the inner limits of the chzs and ghzs and considerably decreased at the outer limits of the chzs and ghzs relative to average earth - equivalent positions . for f2 v stars , the scaling factors for the inner limits are given as 139% and 108% and for the outer limits as 62% and 40% relative to that at average earth - equivalent positions . \(3 ) owing to the form of the dna action spectrum ( see fig . 2 ) , in the absence of significant planetary atmospheric attenuation , most of the damage on dna is because of uv - c . damage due to uv - b is significantly lower , and damage due to uv - a is virtually nonexistent . regarding the latter , the largest value in the context of this study was attained at the inner limit of the ghz for an f0 v star in the absence of planetary atmospheric attenuation , which is @xmath35 . \(4 ) planetary atmospheric attenuation , especially that associated with ozone layers ( e.g. , * ? ? ? * ) , is able to reduce damage inflicted on dna drastically . in consideration of realistic atmospheric attenuation functions , this aspect entails a drastic reduction of damage associated with uv - c . on a relative scale , it thus tends to increase the importance of uv - b . \(5 ) it is particularly intriguing to assess the behavior of @xmath31 during stellar main - sequence evolution , which has been evaluated in detail for stars of masses of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 . if no planetary attenuation is taken into account , the zams @xmath31 values at average earth - equivalent positions are identified as 0.91 , 2.46 , 3.41 , and 5.02 , respectively . the @xmath31 values at other positions in the stellar habitable zones can be obtained by scaling noting that the incident stellar radiation is diluted following the inverse square law . \(6 ) taking average earth - equivalent positions as reference , the @xmath31 values are found to change with time in response to changes in the stellar parameters owing to the stellar main - sequence evolution . @xmath31 first increases with time and reach maximum values of 1.96 , 2.63 , 3.48 , and 5.07 for stars with masses of 1.2 , 1.3 , 1.4 , and 1.5 @xmath0 , respectively ; they are attained at evolutionary times of 1.95 , 0.93 , 0.27 , and 0.14 gyr , respectively . thereafter , the @xmath31 values decline . they reach 0.96 , 1.04 , 1.07 , and 0.32 for the selected set of stars . these values are comparable to or smaller than that at an earth - equivalent position for a solar - like star . our study is a further contribution toward the exploration of the exobiological suitability of stars hotter and , by implication , more massive than the sun . although these stars are relatively rare compared to g - type solar - type stars , they possess significantly augmented habitable zones . on the other hand , their emergent photospheric uv fluxes are much larger ; fortunately , however , they can be diminished through planetary atmospheric attenuation . thus , at least in the outer portions of f - star habitable zones , uv radiation should not be viewed as an insurmountable hindrance to the existence and evolution of life . future studies for f - type stars should encompass ( 1 ) detailed chemical models of planetary atmospheres , aimed at constraining the attenuation parameters , ( 2 ) examples of specific star planet systems with information attained from observational constraints , and ( 3 ) cases of f - type stars that are members of binary ( or higher order ) systems . studies of the circumstellar habitability for those systems also encompassing analyses of planetary orbital stability have been given by , e.g. , @xcite and others . * acknowledgements . * this work has been supported by the department of physics , university of texas at arlington ( s. s. , m. c. ) , by a ugto / promep - funded project ( d. j. ) , and by a conacyt master student stipend ( c. m. g. o. ) . additionally , k. p. s. is grateful for conacyt support of his sabbatical year projects under application no . 207662 . guinan , e.f . & ribas , i. ( 2002 ) . in _ the evolving sun and its influence on planetary environments _ , proc . montesinos , b. , gimenez , a. and guinan , e.f . , pp . astr . soc . , san francisco . lammer , h. , bredehft , j.h . , coustenis , a. , khodachenko , m.l . , kaltenegger , l. , grasset , o. , prieur , d. , raulin , f. , ehrenfreund , p. , yamauchi , m. _ et al . _ _ * 17 * , 181249 . lcccccccccccc & & & & + 0.5 & 6178 & 1.770 & f8 @xmath49 k & 6444 & 2.644 & f5 & 6720 & 3.780 & f3 @xmath50 k & 7084 & 5.228 & f1 @xmath51 k + 1.0 & 6195 & 1.878 & f8 & 6447 & 2.839 & f5 & 6682 & 4.077 & f3 @xmath52 k & 6983 & 5.728 & f1 @xmath53 k + 1.5 & 6205 & 1.996 & f8 & 6432 & 3.036 & f5 & 6589 & 4.361 & f4 @xmath54 k & 6775 & 6.247 & f3 @xmath55 k + 2.0 & 6211 & 2.121 & f8 @xmath56 k & 6386 & 3.224 & f6 @xmath57 k & 6428 & 4.591 & f5 @xmath58 k & 6482 & 6.753 & f5 @xmath59 k + 2.5 & 6204 & 2.247 & f8 & 6303 & 3.383 & f7 @xmath60 k & 6218 & 4.755 & f8 @xmath61 k & 6117 & 7.047 & f9 + 3.0 & 6178 & 2.366 & f8 @xmath49 k & 6174 & 3.493 & f8 @xmath62 k & ... & ... & ... & ... & ... & ... + 3.5 & 6130 & 2.468 & f9 & ... & ... & ... & ... & ... & ... & ... & ... & ... + 4.0 & 6058 & 2.555 & g0 & ... & ... & ... & ... & ... & ... & ... & ... & ... +	we explore the general astrobiological significance of f - type main - sequence stars with masses between 1.2 and 1.5 @xmath0 . special consideration is given to stellar evolutionary aspects due to nuclear main - sequence evolution . dna is taken as a proxy for carbon - based macromolecules following the paradigm that extraterrestrial biology may be most likely based on hydrocarbons . consequently , the dna action spectrum is utilized to represent the impact of the stellar uv radiation . planetary atmospheric attenuation is taken into account based on parameterized attenuation functions . we found that the damage inflicted on dna for planets at earth - equivalent positions is between a factor of 2.5 and 7.1 higher than for solar - like stars , and there are intricate relations for the time - dependence of damage during stellar main - sequence evolution . if attenuation is considered , smaller factors of damage are obtained in alignment to the attenuation parameters . this work is motivated by earlier studies indicating that the uv environment of solar - type stars is one of the most decisive factors in determining the suitability of exosolar planets and exomoons for biological evolution and sustainability .
neutrino interactions with nucleons and nuclei have received a considerable attention in recent years stimulated by the needs in the analysis of neutrino oscillation experiments which give information about the oscillation probability @xmath8 . actually , new high quality data on cross sections are becoming available from miniboone and sciboone experiments @xcite . the neutrino energy that enters critically in the oscillation probability is not directly measurable but has to be reconstructed from the reaction products . in appearance experiments the initial and final neutrinos fluxes are compared to get the oscillation probability . in fact , miniboone was designed to measure the @xmath9 appearance signal in the energy range around @xmath3 gev , detecting electrons from the @xmath10 reaction via cerenkov radiation . in this case , neutral current @xmath11 events ( nc@xmath12 ) are an important source of background because the electron is not distinguished from a photon coming from @xmath13 decay if the other photon is missed . this would lead to fake @xmath1 signal in contrast with the true coming form the ccqe . however , another important source of single - photon background exists : the @xmath6 ( nc@xmath14 ) reaction , which can also be interpreted as a @xmath1 arriving signal . recent miniboone experiment reported , after various refinements , a persisting excess of electron - like events at low energy @xcite . in the recent nc@xmath14 cross section calculation and analysis from hill @xcite , it has been shown that process seems capable to provide enough photons to cover the excess found by miniboone . however , the implemented model is not completely _ consistent _ since the vertex and propagator used for the @xmath7 correspond to a different value of the parameter associated to contact transformations on the spin 3/2 field ( see @xcite and references therein ) . then , it is important to treat consistently and on the same footing both , the @xmath15 and photon production cross sections , to provide a trustily constraint of these of backgrounds . in the @xmath16 gev region relevant for miniboone and sciboone experiments , the process following in importance to the qe scattering is the excitation of the @xmath17 mev ) resonance . nuclei are used as neutrino detectors and require the inclusion of nuclear medium effects on the free nucleon cross sections . a consistent treatment should remedy the before mentioned resonance vertex - propagator correspondence and also keep the preservation of the electromagnetic gauge invariance in presence of finite widths . this program has been already developed with success for another reactions involving the @xmath7 resonance , which encourage us to develop here a model for the evaluation of the cross section of the @xmath6 reaction . we perform here the calculation of nc@xmath14 cross section by considering only the contribution of the resonant amplitude shown in fig . the total cross section is given by ( e_^cm)= _ 0^e_^+ de_^cm _ e_^-^e_^+ de_^cm _ -1^+1dcos_0 ^ 2d _ spin , pol\| m\|^2 , [ crossw]where @xmath18 , with the limits in the integrals estimated from kinematical conditions , and where @xmath19 . the resonant amplitude reads @xmath20 where @xmath21 , @xmath22 and @xmath23 are the neutrino , initial and final nucleon momenta , respectively , and @xmath24 is the @xmath25 boson momentum . @xmath26 reads ( @xmath27 is photon momentum and @xmath28 ) _ r^(p , p,q)= ^(p,p_,q_=p_-p)_^ * ig_(p_=p - q)w^(p_,q = p - p_,p ) , [ deltapole]some calculations involving the @xmath7 production take the simplest form for @xmath29 ( contact constant @xmath30 ) @xcite and , simultaneously , the simplest one for the vertex ( @xmath31 . to avoid this problem and get @xmath32 -independent amplitudes we use a set of reduced @xmath32-independent feynman rules @xcite . the expression for the corresponding bare @xmath7 propagator and weak vertex @xmath33 was defined in ref . finally , the @xmath34 decay vertex , from which @xmath35 is extracted through the cvc hypothesis , was defined in ref . the amplitude , should be multiplied by the new isospin factor @xmath36 for both processes with single photon production induced by neutral currents : @xmath37 and @xmath38 . we show in figure [ fig2 ] the cross section for nc@xmath12 and nc@xmath14 backgrounds to the ccqe process , for @xmath39 gev . the contribution of the radiative @xmath40 process is roughly one hundred times smaller than the neutral current one , since @xmath41 . in the same figure we compare our results with those from fig . 4 in ref . @xcite , obtained within an inconsistent model . we can observe that differences are not important at low energies but yes in the upper region where the @xmath7 resonance becomes relevant . this is also an expectable result originated by the fact we are considering the correct complete @xmath7 propagator ( @xmath42 ) in eq . . in summary , the results obtained within our present _ consistent _ formalism for @xmath40 process were compared with a previous determination that lacks in consistence in the treatment of the resonance field @xcite . our treatment gives similar results in the low energy region and increases the single photon cross section in the @xmath7 region . in view of these results , it would be interesting to reanalyze the excess of events counting contribution of this reaction done in ref . this finally answer the question if radiative corrections provides enough photons to cover the low energy excess found by miniboone .	the quasielastic charged current ( ccqe ) @xmath0 scattering is the dominant mechanism to detect appearance of a @xmath1 in an almost @xmath2 flux at the @xmath3 gev scale . actual experiments show a precision below @xmath4 and between less known background contributions , but necessary to constraint the event excess , we have the radiative corrections . a consistent model recently developed for the simultaneous description of elastic and radiative @xmath5 scattering , pion - photoproduction and single pion production processes , both for charged and neutral current neutrino - nucleon scattering , is extended for the evaluation of the radiative @xmath6 cross section . our results are similar to a previous ( but inconsistent ) theoretical evaluation in the low energy region , and show an increment in the upper region where the @xmath7 resonance becomes relevant . # 1#2#3#4#5#6#7#8#9 \ { ccc # 1&#2&#3 + # 4&#5&#6 + # 7&#8&#9 } # 1#2#3#4#5#6 \ { ccc # 1&#2&#3 + # 4&#5&#6 } # 1 # 1#2 ( c#1 + # 2 ) # 1#2 \ { l#1 + # 2 . # 1([#1 ] ) # 1#2#1 \| # 2 # 1#1\| # 1\|#1 # 1\|\|#1 # 1#1\|\| # 1#2 c#1 + # 2 # 1#2#3#4 c#1 + # 2 + # 3 + # 4 # 1#2 \ { l#1 + # 2 . # 1#2#3 \ { l#1 + # 2 + # 3 . address = departamento de fsica , universidad nacional de la plata , c. c. 67 , 1900 la plata , argentina address = departamento de fsica , universidad nacional de la plata , c. c. 67 , 1900 la plata , argentina
quantum entanglement not only defies our classical intuition but also finds its role in several practical applications devised to harness the power of quantum physics . the technological promise of entanglement has attracted lot of interest to develop a theory of its own , which deals with its characterization and quantification , optimal detection in theory and experiments , and the methods to reverse the inevitable process of decoherence @xcite . as entangled states are desired to generate and manipulate in experiments , therefore it is essential to study the effects of various environments on entanglement . in recent years , this study has received considerable attention and is currently an active area of research @xcite . the dynamics of entanglement under various environments were studied for both bipartite and multipartite systems @xcite . several works of entanglement dynamics considered bipartite aspects of entanglement of multipartite states @xcite , however this can only give partial results because entanglement in multipartite systems is different than the entanglement among different partitions . as theory of multipartite entanglement is still in progress , one can make statements on lower bounds of entanglement but not on its exact value @xcite . the exact value of multipartite entanglement was only calculated for specific model of decoherence and specific quantum states @xcite . in addition , in order to comment on the robustness or fragility of a state , one need to compare its dynamics with dynamics of random states . recently , we have addressed these issues and studied the robustness of several multipartite states by investigating an exact measure of genuine multipartite entanglement under markovian environments @xcite . in this work , we extend our study to dynamics of genuine entanglement under a specific type of non - markovian noise . although , the approximation of weak interaction and no back action of environment on principal system might be valid for certain circumstances , however in reality most systems are non - markovian . it is important to simulate the effects of non - markovian environments on genuine entanglement . recent progress in the theory of multipartite entanglement has enabled us to study decoherence effects on actual multipartite entanglement and not on entanglement among bipartitions . in particular , the ability to compute genuine negativity for multipartite systems has eased this task @xcite . we find that under non - markovian dephasing , ghz state appears to be resilient as it repeats the collapse and revival of genuine entanglement for long time . on the other hand , the w state turns out to be most fragile state . all other quantum states of four qubits exhibit behavior between these two extremes . we compare the dynamics of chosen quantum states of three and four qubits with dynamics of random pure states and weighted graph states . this paper is organized as follows . in section [ sec : model ] , we briefly discuss our model of interest . we review the concept of entanglement for multipartite systems in section [ sec : gme ] where we also describe the method to compute genuine negativity and discuss the quantum states which we study in this work . in section [ sec : res ] , we provide results and finally we conclude our work in section [ sec : conc ] . the model which we intend to study here is well known @xcite . recently , this model has also been studied for sudden change in dynamics of quantum discord @xcite . we consider a dephasing with colored noise with dynamics described by a master equation @xcite @xmath0 where @xmath1 is time - dependent integral operator whose action on a function is defined as @xmath2 with @xmath3 is a kernel which determines the type of environment memory . @xmath4 is density matrix of the principal system and @xmath5 is the lindblad super - operator which describes dynamics of principal system as a result of interaction with environment . note that in the absence of @xmath1 in eq.([eq : menm ] ) one usually get the master equation with markovian approximation . for a concrete example of a system , we may consider the time - dependent hamiltonian @xcite @xmath6 where @xmath7 are the pauli matrices and @xmath8 are the independent random variables which obey statistics of a random telegraph signal defined as @xmath9 . the random variable @xmath10 has a poisson distribution with a mean @xmath11 and @xmath12 is an independent random variable taking values @xmath13 . this model applies to any two - level quantum system interacting with an environment having random telegraph signal noise . as an example , this could describe a two - level atom subjected to a fluctuating laser field that has jump type random phase noise @xcite . using von neumann equation of motion @xmath14 $ ] , we can write the solution for density matrix of two level system as @xmath15\ , .\ ] ] substituting this equation back into von neumann equation and performing stochastic average , we obtain @xcite @xmath16\big ] \ , , \label{eq : menm2}\ ] ] where the correlation functions of random telegraph signal @xmath17 have been inserted . it turns out that the dynamical evolution generated by eq.([eq : menm2 ] ) is completely positive when two of the @xmath18 are zero . this would correspond to a physical situation where noise only acts in one direction . in particular if @xmath19 , and @xmath20 , then the dynamics of the system is that of a dephasing with colored noise . the kraus operators describing the dynamics of two - level system are given as @xcite @xmath21/2 } \ , i_2 \nonumber \\ k_2 & = & \sqrt{[1 - \lambda(\nu)]/2 } \ , \sigma_3 \ , , \end{aligned}\ ] ] where @xmath22 is @xmath23 identity matrix and @xmath24 $ ] , with @xmath25 and @xmath26 is the dimensionless time . the kraus operators satisfy the normalization condition @xmath27 . as we are interested in three and four qubit systems , the time evolution of an initial density matrix can be written as @xmath28 where @xmath29 are the kraus operators , satisfying the normalization condition @xmath30 . for three qubits , there are @xmath31 such operators , that is , @xmath32 , @xmath33,@xmath34 , @xmath35 . we have omitted the tensor product symbol between these operators . similarly , the respective @xmath36 operators for four qubits are @xmath37 , @xmath38,@xmath34,@xmath39 . the time evolved density matrix for a single qubit can directly be computed and it is given as @xmath40 where @xmath41 . \label{eq : gama}\ ] ] for more qubits , the calculation of density matrices is straightforward . in this section , we briefly review the concept of entanglement in multipartite systems and discuss the particular quantum states which we study in this article . we want to emphasize at this stage that material in this section is already known in the literature and we cite them appropriately . we review genuine multipartite entanglement by considering three particles @xmath42 , @xmath43 , and @xmath44 , as the generalization to more parties is straightforward . a state is called separable with respect to some bipartition , say , @xmath45 , if it is a mixture of product states with respect to this partition , that is , @xmath46 , where @xmath47 and @xmath48 . similarly , separable states for the two other bipartitions are @xmath49 and @xmath50 . a state is called biseparable if it can be written as a mixture of states which are separable with respect to different bipartitions , that is @xmath51 a state is genuinely multipartite entangled if it is not biseparable . in this paper , we study dynamics of this genuine multipartite entanglement . recently , it has been worked out to detect and characterize multipartite entanglement by using positive partial transpose mixtures ( ppt mixtures ) @xcite . we recall that a two - party state @xmath52 is ppt if its partially transposed matrix @xmath53 has no negative eigenvalues . it is known that separable states are always ppt @xcite . the set of separable states with respect to some partition is therefore contained in a larger set of states which has a positive partial transpose for that bipartition . the states which are ppt with respect to fixed bipartition may be called @xmath54 , @xmath55 , and @xmath56 . we ask whether a state can be written as @xmath57 such a mixing of ppt states is called a ppt mixture . the genuine multipartite entanglement of four or more particles can be detected and quantified in an analogous manner by considering all bipartitions ( like one party vs. @xmath58 parties , two parties vs. @xmath59 parties , etc . ) . as any biseparable state is a ppt mixture , therefore any state which is not a ppt mixture is guaranteed to be genuinely multipartite entangled . the prime advantage of considering ppt mixtures instead of biseparable states is that ppt mixtures can be fully characterized with the method of semidefinite programming ( sdp ) @xcite . in general , the set of ppt mixtures is a very good approximation to the set of biseparable states and delivers the best known separability criteria for many cases , nevertheless there are multipartite entangled states which are ppt mixtures @xcite . it is also interesting to note that there are biseparable states which may have negative partial transpose ( npt ) under each partition @xcite . it has been shown @xcite that a state is a ppt mixture iff the following optimization problem @xmath60 under the constraint that for all bipartitions @xmath61 @xmath62 has a positive solution . the constraints reflect that @xmath63 is a decomposable entanglement witness for any bipartition . if this minimum is negative then @xmath4 is not a ppt mixture and hence genuinely multipartite entangled . since this is a semidefinite program , the minimum can be efficiently computed and the optimality of the solution can be certified @xcite . we use the programs yalmip and sdpt3 @xcite to solve sdp . we also use implementation which is freely available @xcite . this approach can be used to quantify genuine entanglement as the absolute value of the minimization was shown to be an entanglement monotone for genuine multipartite entanglement @xcite . in the following , we will denote this measure by @xmath64 . for bipartite systems , this monotone is equivalent to _ negativity _ @xcite . for a system of qubits , this measure is bounded by @xmath65 @xcite . we are interested in several families of states in this work . two important families of states , namely the ghz states and the w states for @xmath66 qubits are given as @xmath67 ghz state has always maximum value of monotone , that is , @xmath68 , whereas for the w state , numerical value depends on the number of qubits . for three qubits @xmath69 and for four qubits @xmath70 . several interesting states for four qubits are dicke state @xmath71 , the singlet state @xmath72 , the cluster state @xmath73 and the so - called @xmath74-state @xmath75 , given as @xmath76 \ , , \nonumber \\ + \|1001 \rangle + \|1010\rangle ) ] \ , , \nonumber \\ respectively . all these states are maximally entangled with respect to multipartite negativity , @xmath77 . these states along with their properties are discussed in ref . @xcite . we describe the generation of random pure states . a state vector randomly distributed according to the haar measure can be generated as follows @xcite : first , we generate a vector such that both the real and the imaginary parts of the vector elements are gaussian distributed random numbers with a zero mean and unit variance . second we normalize the vector . it is easy to prove that the random vectors obtained this way are equally distributed on the unit sphere @xcite . we stress that we generate random pure states in the global hilbert space of three- and four qubits , so the unit sphere is not the bloch ball . another important family of multi - qubit states are weighted graph states which also includes states such as ghz and cluster states @xcite . these states have been studied for entanglement properties of spin gases @xcite . we consider a graph as a set of vertices and edges , where the vertices may denote the physical systems ( qubits ) and the edges represent the interactions among physical systems . we initially prepare all the qubits in the state @xmath78 . for any pair of qubits @xmath79 which are connected with an edge , we apply an interaction via hamiltonian @xmath80 the resulting unitary transformation is @xmath81 , where @xmath82 is the interaction time . the generated state after this process is called the weighted graph state given as @xmath83 parameters @xmath82 , which form only a small subset of all pure states described by @xmath84 parameters . however , many interesting belong to this class . we have chosen the interaction times @xmath85 $ ] uniformly distributed in the interval to generate these states . for the choice @xmath86 or @xmath87 , the usual graph states ( also containing the ghz and cluster states ) appear . we note that the temporal order of the interaction does not matter because the unitaries @xmath88 commute . some generalizations of these states have been investigated recently @xcite . we point out here that all random states generated by these two techniques are genuinely entangled . as described above , the weighted graph states may include ghz states and cluster states , however this does not mean that every random state generated is equivalent to only these two types of states . therefore , it is meaningful to compare the dynamics of specific states with dynamics of random states @xcite . in this section , we present our results for quantum states discussed in previous section . first we discuss the effect of non - markovian dephasing on genuine entanglement of three qubits . figure [ fig : e3qbnmd1 ] shows the multipartite negativity @xmath64 plotted against the dimensionless time @xmath89 . we have chosen @xmath90 and @xmath91 which are for non - markovian region @xcite . the solid line denotes the ghz state whereas dotted line is for the w state . it can be seen that the genuine entanglement of ghz state first goes to zero and then revives again with a lower peak than the previous one . this collapse and revival of genuine entanglement for ghz state repeats itself for sufficiently long time , however each time with a lower peak than the earlier ones . this suggest that ghz state is resilient state against non - markovian dephasing . this observation is similar to our recent studies where we have observed that ghz state is quite robust under markovian dephasing @xcite . we stress that in our previous study @xcite , we have used logarithmic derivative of genuine entanglement to claim the robustness of ghz state under markovian dephasing , whereas here we use the term `` resilience '' to this several time repetition of collapse and revival of genuine entanglement for ghz state . several authors have studied the robustness of entanglement @xcite . in @xcite , the authors have identified most robust states under local decoherence used the definition introduced before @xcite , whereas , the authors of ref.@xcite considered asymptotic long - time dynamics of initial states and identified two different classes of states , one class is fragile even there remains some coherence in the system and the second class as most robust states which become disentangled only when decoherence is perfect . however , as described above , our approach is different than these studies . in contrast , the w state is very fragile under non - markovian dephasing . the collapse and revival of genuine entanglement for w state disappear earlier . the revival of genuine entanglement for the w state also takes a little longer time than the ghz state . this observation is similar to markovian dephasing where the w state is found to be quite fragile @xcite . in addition , we plot the mean values of genuine negativity for random pure states ( mrs ) denoted by dashed line and weighted graph states ( mwgs ) denoted by dashed - dotted line for comparison purpose . it is interesting to note that genuine entanglement behavior of both type of random states almost resemble to the w state . at this stage , we want to remind the readers that our method to detect the genuine entanglement is not always perfect . if this measure is positive then the state is guaranteed to be genuinely entangled , however , if it is zero , then in general , we do not know whether the state is entangled or not . except for the ghz state which we discuss below in detail , we can not say with certainty that all zero values of genuine negativity in figure [ fig : e3qbnmd1 ] and all subsequent figures correspond to states with no entanglement of any type . an interesting property of the dynamical process is the fact that all zero elements of the initial density matrix remain zero . for ghz state the only non - zero density matrix elements are @xmath92 , @xmath93 , @xmath94 and @xmath95 . a recent result on the detection of genuine entanglement states that for biseparable states , the inequality @xmath96 is satisfied and the violation implies genuine entanglement @xcite . this criterion is a necessary and sufficient condition for ghz - diagonal states @xcite . we will not go into details of ghz - diagonal states here but for our purpose this criterion would imply that time evolved ghz state is genuinely entangled if and only if @xmath97 , that is , @xmath98 where @xmath99 is defined in eq.([eq : gama ] ) . in figure [ fig : ie ] , we plot the function @xmath100 scaled down to a factor of @xmath101 against parameter @xmath102 with @xmath103 for chosen values of @xmath104 and @xmath105 . a close comparison of figure [ fig : ie ] with figure [ fig : e3qbnmd1 ] reveals that for those points where @xmath100 is zero , the genuine entanglement for ghz state is also zero and vice versa . actually , at these points the off - diagonal elements of any arbitrary time - evolved density matrix disappear completely , so entanglement of any type disappear as well . this fact is evident in figure [ fig : e3qbnmd1 ] . it also explains the collapse and revivals of genuine entanglement of ghz state at these instances . it was shown @xcite that phase - damping channels are called locally entanglement - annihilating if and only if the off - diagonal elements disappear completely . this happens only at times when @xmath106 . therefore , only at these points , being entanglement - annihilating coincides with being entanglement - breaking . it is interesting to note that disappearance of genuine entanglement only coincides with these instances for ghz state . for other states , there are intervals where off - diagonal elements are not zero but our criterion fails to detect whether there is entanglement of any type as discussed before . the presence of exponential factor in the definition of @xmath99 also explains the gradually decaying peaks of genuine entanglement of all initial states . as every off - diagonal element of the time - evolved density matrix is being multiplied by a factor @xmath107 with @xmath108 , @xmath109 , or @xmath110 , therefore as @xmath100 is zero necessarily means @xmath99 is zero and @xmath99 is maximum when @xmath100 is maximum . this is the reason that all states have their peak value of genuine entanglement at instances when @xmath100 is maximum . this feature is also evident from figures [ fig : e3qbrsnmd ] and [ fig : e3qbwgsnmd ] . as the behavior of ghz state is very different than all other states , therefore in order to get some more insight , we explore the effects of decreasing the degree of non - markovianity on its dynamics . by fixing @xmath90 , the parameter @xmath111 denotes the non - markovian case , whereas @xmath112 corresponds to markovian case . figure [ fig : e3ghznmd ] shows genuine entanglement of ghz state for various values of parameter @xmath105 plotted against parameter @xmath102 . it can be observed that as we decrease the degree of non - markovianity , the collapses and revivals of genuine entanglement are also decreased and delayed . finally , for @xmath112 , there are no collapse and revival of genuine entanglement , as we expect from markovian dynamics . as we discussed above , for ghz state to be genuinely entangled , @xmath97 , and this value could only be non - zero if @xmath99 defined in eq.([eq : gama ] ) is non - zero . as @xmath99 depends on @xmath113 , which in turn depends on @xmath105 , that is , for @xmath111 , @xmath103 ( non - markovian case ) , for such a large value of @xmath113 , the sine and cosine functions are considerable and can not be ignored . hence these functions are responsible for collapses and revivals . on the other hand , for @xmath112 , we have @xmath114 ( markovian case ) , which is much smaller than the previous case . so for such a small argument , we can apply small angle approximation to sine and cosine functions . hence we can explain the corresponding disappearance of collapses and revivals of genuine entanglement for markovian case . figure [ fig : e3qbrsnmd ] shows multipartite negativity for @xmath115 random pure states plotted against @xmath102 . we also plot mean values of random pure states ( mrs ) denoted by dashed line and weighted graph states ( mwgs ) denoted by dashed - dotted line , however they almost overlap and may not be clearly visible . in figure [ fig : e3qbwgsnmd ] , we plot multipartite negativity for @xmath115 weighted graph states along with mean values of genuine entanglement for random pure states ( mrs ) and weighted graph states ( mwgs ) . let us now discuss the results for four qubits case . in figure [ fig : e4qbnmd ] , we plot genuine negativity for various states discussed in previous section . as for the case of three qubits the ghz state is resilient state whereas the w state is quite fragile . all other states including random pure states and weighted graph states exhibit a trend in between these two extremes as evident from the figure [ fig : e4qbnmd ] . as results for the four qubits are almost identical to the three qubits case , therefore we preferred to plot only mean value of genuine entanglement for random states only . we have studied the behavior of genuine multipartite entanglement under non - markovian dephasing . using a computable entanglement monotone for multipartite quantum systems , we have observed the collapses and revivals of genuine entanglement for various quantum states of three and four qubits . we have found that ghz state is resilient state as it repeats its revival for long time whereas all other states loose their revivals much earlier . we have examined and explained this different behavior of ghz state . we have found that the w state is the most fragile state similar to markovian environments . we have compared dynamics of chosen quantum states with dynamics of random pure states and weighted graph states so that we can make meaningful statements about their behavior under decoherence . we found that all random states and weighted graph states show a similar trend as the w state . we stress here that our conclusions are based on a criterion whose positive value for a given quantum state is guaranteed to be genuinely entangled , however , for states which are not detected by this criterion , we are not certain about their entanglement properties . for ghz state under current dynamics , this criterion provides a necessary and sufficient criterion to detect genuine entanglement . the author is grateful to both anonymous referees for their constructive comments , which brought much clarity and a new figure [ fig : e3ghznmd ] in the manuscript . 99 t. yu and j. h. eberly , phys . rev . b * 66 * , 193306 ( 2002 ) ; t. yu and j. h. eberly , phys . b * 68 * , 165322 ( 2003 ) ; t. yu and j. h. eberly , phys . . lett . * 93 * , 140404 ( 2004 ) ; j. h. eberly and t. yu , science * 316 * , 555 ( 2007 ) . a. r. p. rau , m. ali and g. alber , epl * 82 * , 40002 ( 2008 ) ; m. ali , g. alber , and a. r. p. rau , j. phys . b : at . mol . . phys . * 42 * , 025501 ( 2009 ) ; m. ali , j. phys . b : at . mol . . phys . * 43 * , 045504 ( 2010 ) ; m. ali , phys . rev . a * 81 * , 042303 ( 2010 ) . b. jungnitsch , t. moroder , and o. ghne , phys . rev . lett . * 106 * , 190502 ( 2011 ) ; l. novo , t. moroder , and o. ghne , phys . a * 88 * , 012305 ( 2013 ) ; m. hofmann , t. moroder , and o. ghne , j. phys . a : math . theor . * 47 * , 155301 ( 2014 ) .	we study dynamics of genuine entanglement for quantum states of three and four qubits under non - markovian dephasing . using a computable entanglement monotone for multipartite systems , we find that ghz state is quite resilient state whereas the w state is the most fragile . we compare dynamics of chosen quantum states with dynamics of random pure states and weighted graph states . genuine entanglement , non - markovian , random states 03.65.yz , 03.65.ud , 03.67.mn
the random velocities of disk stars in the solar neighborhood are known to increase with time since the 1940 s and this fact has been verified by recent hipparcos observations . the analysis of the hipparcos data @xcite reveals that the stellar velocity dispersions increase continuously ( rather than episodically as claimed by @xcite ) with time which is the well - known disk heating problem in our galaxy . the phenomenon of disk heating is not specific for our galaxy since the heating of stellar disks is also known to persist in external galaxies such as in ngc 2985 , ngc 2460 , and ngc 2775 @xcite . although mechanisms have been identified for the disk heating , a full understanding has yet to be achieved . it is known that disk heating is anisotropic ; that is , disk stars are preferentially heated along the plane in comparison to its perpendicular direction . thus , the main problem in disk heating is two - fold . the first is the requirement to explain the continual heating of disk stars especially in the case of isolated galaxies for which tidal heating or external perturbations can be ruled out . secondly , such a mechanism must be anisotropic in nature . taken together , disk heating remains a fundamental problem in galactic dynamics . historically , @xcite first showed that the secular increase in the stellar velocity dispersions could arise as a result of the scattering of disk stars with giant molecular clouds ( gmcs ) based on two dimensional calculations . subsequently , there has been considerable development and progress in understanding the disk heating problem in galaxies . for example , @xcite extended the previous theoretical work to a three dimensional disk and found that the velocity dispersions increase with time , @xmath0 , as @xmath1 where @xmath2 . in addition , it was demonstrated that transient stochastic spiral arms @xcite could also produce secular heating of the disk stars , however , such effects were only effective in the plane of the galaxy . the effect of giant molecular clouds in combination with the effect of stochastic spirals could scatter the stars off the plane @xcite to account for both the radial and vertical heating . alternatively , using an extensive orbit analysis of the stars subject to a 3d galactic bar potential , @xcite has shown that a large fraction of the phase space is chaotic and suggested that all the disk stars trapped in the chaotic phase space would be heated up in all directions . other possible candidate heating mechanisms include massive dark halo objects , e.g. black holes or dark clusters of mass @xmath3 m@xmath4 @xcite or a combination of giant molecular clouds and halo black holes @xcite . within the framework of standard cosmological models , a spectrum of subhalos and their substructures could heat up the disk @xcite . however , @xcite found that the subhalos are not efficient perturbers for disk heating because the orbits of the subhalos rarely take them near the disk . finally , satellite infall onto the galactic disk could produce an abrupt heating of the disk @xcite or moderate heating could arise due to sinking satellites on prograde orbits @xcite . the many possible heating mechanisms of the disk stars has been reviewed by @xcite . note that while most mechanisms produce radial heating of the disk stars they have not been shown to provide a satisfactory explanation of the detailed nature of vertical heating . in fact , studies focusing on the vertical heating are comparatively lacking . to be more precise , a generic and robust internal source remains to be identified which could be responsible for the continual vertical heating of the disk . here , we primarily concentrate on the issue of vertical disk heating . galactic disks often consist of various non - axisymmetric patterns such as bars , spirals , warps , corrugations , and rings . these patterns may be growing or transient and , in general , may be time - dependent . it is natural to ask whether such patterns contribute to the disk vertical heating . that is , what is the nature of vertical heating due , for example , to a growing bar ? how does a growing bar heat the disk stars in the vertical direction ? heating due to non - axisymmetric patterns are important for isolated galaxies as even for the milky way , there are prominent and characteristic signatures of non - axisymmetric patterns @xcite in the phase space in the solar neighbourhood . in this paper , we explore possible mechanisms giving rise to continuous heating of the disk stars in the vertical direction . in particular , we show that stellar bars that are formed nearly spontaneously in disk galaxies are capable of efficiently scattering stars in the vertical direction . since transient spiral arms are often associated with bars , they also contribute to the disk heating . in order to understand the nature of vertical heating of the disk stars , we have performed a large number ( statistically meaningful ) of n - body simulations of model disk galaxies constructed over a wide range of parameter space . the paper is organized as follows . section 2 describes the galaxy models used in the simulation . a description of the n - body simulation and the physical basis for the choice of our parameter space is presented in section 3 . the heating model used to interpret the numerical simulations is described in section 4 . section 5 describes the heating due to non - axisymmetric structures . the results on the correlation studies are presented in section 6 . the discussion and conclusions are presented in section 7 and 8 respectively . the construction of a very thin equilibrium disk model involves a non - trivial procedure in n - body simulation . however , using the self - consistent bulge - disk - halo model of ( * ? ? ? * hence kd95 ) we are able to construct extremely thin equilibrium model disks ( with the ratio of scale - height to scale length @xmath5 ) . their prescription provides nearly exact solution of the collisionless boltzmann and poisson equations which are suitable for studying disk stability related problems , allowing one to construct a wide range of initial models from a large parameter space . all the components in our models are active ( i.e. , the gravitational potential of each component can respond to an external or internal perturbation ) and , hence , provide a realistic representation for the evolution and structure of the galaxies . below we briefly describe each component of the model separately for the sake of completeness . for more details , the reader is referred to kd95 . a spherical live bulge is constructed from the king model @xcite and the corresponding distribution function ( df ) is given by @xmath6 here , the bulge is simulated using three parameters , namely the cut - off potential ( @xmath7 ) , central bulge density ( @xmath8 ) and @xmath9 governing the central bulge velocity dispersion . the depth of the potential well is measured by @xmath10 . an axisymmetric live dark matter halo is simulated using the distribution function of a lowered evans model and is given as @xmath11\\ \times ( e^{-e/\sigma_h^2 } -1 ) & \mbox{if } e<0,\\ 0 & \mbox{otherwise}. \end{array } \right.\ ] ] the velocity and density scales are given by @xmath12 and @xmath13 respectively . the halo core radius @xmath14 and the flattening parameter @xmath15 together with @xmath13 are contained in the parameters @xmath16 and @xmath17 . all the simulated halos are oblate in shape and kept at a constant value @xmath18 for simplicity . the disk distribution function is constructed using the approximate third integral given by @xmath19 , the energy of the vertical oscillations . this third integral is approximately conserved for orbits near the disk mid - plane . the radial density of the disk is approximately exponential with a truncation and the vertical density is chosen to depend exponentially on the vertical potential @xmath20 . the volume density of the axisymmetric disk is given by @xmath21 where @xmath22 governs the vertical structure of the disk , @xmath23 is the complementary error function . in the above equation , @xmath24 is the disk mass , @xmath25 is the scale length and @xmath26 is the scale height . there are @xmath27 parameters required to construct a particular galaxy model . the galactic disk produced in this manner remains in equilibrium as long as the simulation runs . the dark matter halo generated using the lowered evans model @xcite has a constant density core which is probably appropriate for the low surface brightness galaxies which we simulate . each model galaxy is constructed such that an initially almost flat rotation curve is produced . the disk outer radius ( @xmath28 ) is fixed at about @xmath29 and a truncation width @xmath30 is adopted within which the disk density smoothly decreases to zero at the outer radius . the primary aim of the present study is to achieve an understanding of the nature and origin of vertical heating of disk stars . in order to gain the necessary insight , we perform a large number of simulations of isolated galaxies . the models are evolved for a sufficiently long period of time so as to examine the nature of vertical heating in different spatial regions of the disk . using the kd95 method , we build the initial conditions for all the model galaxies . the method is not fully self - consistent because the disk distribution function is not known exactly . each model galaxy consists of a bulge , disk and dark matter halo as described above . since the initial conditions are constructed based on distribution functions , they are suitable for studying the long - term evolution of the non - axisymmetric patterns in the disk . building initial galaxy models with prescribed galaxy properties such as the toomre @xmath31 , ratio of dark - to - disk mass ( @xmath32 ) or ratio of velocity dispersions ( @xmath33 ) is not straightforward because the bulge and the halo models are not derived using their mass profiles but from their distribution functions . hence , the relation between the toomre @xmath34 parameter and @xmath32 or between @xmath33 and @xmath32 is not known from the outset . here , @xmath35 and @xmath36 denote the velocity dispersion in the radial and vertical direction , epicyclic frequency , and surface density of stars at a given radius respectively . the disk , bulge and halo parameters are chosen such that an almost flat rotation curve is produced in the outer region . we assume a scale length for the disk to be @xmath37 kpc . the radial and vertical forces are normalized such that it produces the circular velocity @xmath38 kms@xmath39 at about @xmath40 disk scale lengths . the unit of time is @xmath41 yr . the initial disk scale height ( @xmath26 ) varies from @xmath42 pc to @xmath43 pc in our sample of model galaxies . the softening length for the disk particles is @xmath44 pc , for the bulge particles @xmath42 pc and for the halo particles the softening length is @xmath45 pc . a total of @xmath46 to @xmath47 particles has been used to simulate the model galaxies . the dark matter halo mass varies from @xmath40 to @xmath48 which gives , on average , a mass resolution of @xmath49 . we have observed that with such a mass resolution the shot noise in the simulation reduces considerably and bending instabilities do not grow sufficiently to induce an outer disk warp during the evolution . each model galaxy in our simulation has been evolved with the gadget code @xcite which uses a variant of the leapfrog method for the time integration . the forces between the particles are calculated basically using the bh tree ( with some modification ) algorithm with a tolerance parameter @xmath50 . the integration time step is @xmath51 myr and each model galaxy is evolved for more than @xmath52 gyr in our simulation . the orbital time scale at the disk half mass radius is @xmath53 myr for the model mk97 ( see table . [ paratab ] ) . for convenience , we describe the parameter space for the simulations below . a wide range of parameters in the 3-d space spanned by [ @xmath34,@xmath33,@xmath32 ] is considered . the exact relationship between these parameters is dependent on how an active dark matter halo interacts with the other active disk and bulge components . however , we notice that if the central radial velocity dispersion is held constant , there appears to be a positive correlation between the toomre @xmath34 parameter and the dark halo - to - disk mass ratio ( @xmath32 ) as expected from a simple understanding of the disk dynamics with a rigid dark halo . the initial thickness of the disks ( @xmath55 ) in our sample of model galaxies ranges from thin ( @xmath56 or @xmath57 ) to superthin ( @xmath58 ) . to give an example , the value of @xmath59 in the superthin galaxy ugc 7321 @xcite . we explore a wide region of the parameter space spanned by @xmath60 ( see fig . [ fig : param ] ) and provide , here , the physical basis for our choice . each model galaxy has a distinct @xmath34 and @xmath33 radial profile ; we quote the values of @xmath34 and @xmath33 at the disk half mass radii ( see table [ paratab ] for representative models ) . specifically , most of the galactic disks are vertically cold , but radially hot because we primarily aim to simulate low surface brightness ( hence lsb ) galaxies where the rotation curve is normally dominated by the dark matter halo @xcite . in fact , many of the rotation curves in our sample ( e.g. , fig . [ fig : rotc ] ) are such that the dark matter dominates the disk rotation curve from the center of the galaxy . it is also known that the stellar disks of lsb galaxies are thinner than their high surface brightness ( hence hsb ) counterparts @xcite . lsb galaxies are known to be poor in star formation activities despite the presence of neutral hydrogen gas as in otherwise normal galaxies . this may indicate that these galaxies are dynamically hot ( with high @xmath34 as most of the galaxies are in our sample ) to prevent them from disk instabilities . even if the disks suffer from instabilities , the resulting non - axisymmetric features could be young . in fact , the low metallicities , high gas - to - star mass ratios , and blue colors of most lsb galaxies indicate that these systems are probably younger than their hsb counterparts @xcite . .initial conditions for the representative model galaxies ( arranged in ascending order of @xmath34 ) and the heating exponents computed at the end of each simulation . [ cols="^,^,^,^,^,^,^ " , ] [ paratab ] we use the following simple expression ( eq . 4 ) to fit all our simulation data , treating this functional form to quantify the effects of heating . it will be applied to the central region and to the outer region of the galaxy . @xmath61 in the above equation , @xmath62 and @xmath63 are independent of time , and @xmath64 is the orbital time scale measured at the disk half mass radius . @xmath65 is the heating exponent and is the logarithmic time derivative of the vertical velocity dispersion @xmath66 . we determine the time evolution of @xmath67 in the inner region ( @xmath68 ) and in the outer regime ( @xmath69 ) of the disk . the definitions for the two regions of the disk are maintained throughout the paper . the heating in each of these regions of the disk , characterized by apparently distinctive dynamical behaviors , is described by the resulting heating exponents given as @xmath70 and @xmath71 respectively . higher values of @xmath65 imply that the rate of heating is steeper as the galaxy evolves , whereas lower values indicate a gradual heating process . a thorough knowledge of these values are crucial for understanding the type of heating processes at work in the galaxy and possibly for identifying the mechanism responsible for the heating . the fitting is performed using a robust non - linear least squares fitting algorithm due to levenberg and marquardt @xcite . the best fit model parameters are chosen based on the minimum value of the chi - square ( @xmath72 ) . using the resulting parameters , an useful estimate for the time scale of heating can be obtained from the above formula and this can also be compared with the relaxation time scale for the model under consideration . let us consider that in time @xmath73 , the change in the vertical velocity dispersion @xmath74 at a particular radius in the disk . then differentiating eq.[4 ] , it can be shown that @xmath75^{1/\alpha } \tau_{orb}.\ ] ] we evaluate this time scale for different models presented here . in comparison , the energy relaxation time scale for the n - body model with a total number of particles @xmath76 , size @xmath77 and softening @xmath78 can be written following @xcite as @xmath79 where we have used the crossing time scale @xmath80 , assuming that the circular velocity @xmath81 remains almost flat beyond @xmath82 ; @xmath82 denotes the half - mass radius . considering @xmath83 , @xmath84 , @xmath85 and softening parameter @xmath86 , we have @xmath87 . previously , disk heating has been modelled as a diffusion process in the velocity space . if the diffusion coefficient @xmath88 remains constant in time , it can be shown that the vertical velocity dispersion evolves @xcite as @xmath89 note that the above formula is applicable to the radial dispersion also . the exponent @xmath90 denotes heating due to the halo black holes @xcite . in addition to eq . [ 4 ] , we also apply the diffusion model to fit our simulation data to some models for comparison and evaluate the corresponding diffusion coefficient @xmath88 . let us consider that @xmath91 is the time scale over which the change in the vertical velocity dispersion @xmath92 . then @xmath91 can be obtained as follows @xmath93 we denote @xmath91 as the diffusion time scale . in general , @xmath91 and @xmath94 are comparable . internal evolution of a disk galaxy is driven by various non - axisymmetric perturbations such as bars or spirals which redistribute the energy and angular momentum within the disk and/or between the halo and the disk . the time scale for this process is longer than the dynamical time scale ( @xmath95 ) and is usually known as the secular evolution time scale ( @xmath96 ) @xcite . @xmath96 could be a few to few @xmath97s of orbital time scales ( @xmath64 ) . in ordering these time scales , @xmath98 . we will use these definitions and compare the relevant time scales in the following to determine the relation between @xmath73 and the other time scales . -178.98pt -115pt -61.0pt -0pt it is well established that stars undergo continuous heating in both the radial and vertical directions in galaxies as they age . although there has been substantial progress in understanding the radial heating , there is little progress for the vertical heating . spiral structures have played a significant role in the studies of disk heating in the radial direction . we note , however , that a steadily rotating quasi - stationary spiral pattern heats the disk stars only negligibly @xcite . in this case , sufficient heating of disk stars requires the spiral structures to be stochastic in nature . transient , swing - amplified spiral density waves @xcite or multiple spiral density waves @xcite have recently been shown to be effective in the radial heating of disk stars . the primary mechanisms suggested for the vertical heating of the disk stars in early works range from a combination of stochastic spirals and giant molecular clouds @xcite to resonant heating of the trapped stars in the chaotic phase space in the presence of a 3d bar potential @xcite . more recently , other mechanisms have been suggested including a bending instability of the entire disk or of the bar @xcite and repeated impact of globular clusters @xcite , although it produces only a small change in the vertical velocity dispersion . disk galaxies are rich in non - axisymmetric structures , which may form either by means of mergers , tidal interaction or internal instabilities . examples include bars , transient spirals , rings ( e.g. ngc 1433 , ngc 6300 * ? ? ? * ) , lopsidedness @xcite , bending waves manifested in the form of a warp ( e.g. eso121-g6 , ngc 4565 * ? ? ? * ) or corrugation ( e.g. in ic 2233 * ? ? * ; * ? ? ? * in ngc 5907 ) of the disk mid - plane . these non - axisymmetric instabilities often drive the disk away from equilibrium and are generally time dependent . each of these features are , in principle , a potential source of heating of the disk stars . since our understanding of the interactions of these features with the surrounding live dark matter halo is not well understood , it is difficult to disentangle the signatures of each of these non - axisymmetric modes on the net heating of a group of stars . however , some progress has been made especially for our galaxy in which it has been shown based on the analysis of the hipparcos data that there exists a gap due to the @xmath99 resonance with the bar in the galactic uv plane @xcite . in the following , we study the nature of the vertical heating of the disk stars due to bar growth and transient spirals . we also briefly discuss the effect of halo noise and the corrugation and/or bending waves on the net heating process . we carry out a series of n - body simulations to first systematically identify the sources of disk heating in our model galaxies which are free from tidal interaction . about 70 simulations have been performed covering a wide range of the above mentioned simulation space and a statistical attempt to characterize and understand the nature of vertical heating in various regions of the disk has been carried out . -178.98pt -115pt -61.0pt -0pt ) of the galaxy model mk97 as a function of time . the solid line is the model fitted to the simulation data . @xmath100 is the initial value of the vertical dispersion at this radius . ] except for model mk112 . ] except for model mk112 . ] although most of the galaxies in our sample are radially hot , they form a bar at the center during its time evolution . some of the model galaxies form a bar within about @xmath101 to @xmath102 rotation time scales , while for others it takes more than about @xmath97 rotation time scales to reach nearly the same amplitude . it is found that a bar forms even for galaxies characterized by @xmath103 ( model mk55 ) in our simulation , implying that it is difficult for a rotating disk within a live dark matter halo to avoid forming bars at the center . hence , bar formation appears to be quite common and generic in disk galaxies and a live dark matter halo plays a significant role . as shown by @xcite a live dark matter halo supports , rather than suppresses , the bar instability in the disk . in contrast , bar formation in the disk could be suppressed in the presence of a non - responsive rigid dark matter halo @xcite . the growth of the bar depends on the initial condition and grows via non - linear processes as the disk stars interact with the live dark matter halo particles . the nature of this interaction is yet to be completely understood although recent n - body simulations by @xcite , @xcite , and @xcite have revealed insights on the nature of bar dynamics in disk galaxies . for our choice of parameters and initial conditions , we do not find any unique trends for the bar growth in our simulation . the sample of galaxies in our n - body simulations differs from typical galaxy models studied previously ( where @xmath34 is normally held constant throughout the disk or @xmath104 in general ) . overall , we observe that the growth of bars in our sample falls under two broad distinct categories . in the first category , the bar grows quickly to a peak amplitude within @xmath105 rotation time scales which is about an order of magnitude less than @xmath96 and nearly reaches saturation or starts growing again ( for convenience , we call these _ type - i bar _ ) . in the second category , the bar continues to grow but slowly as compared to _ type - i _ and does not show any tendency to saturate ( we call these _ type - ii bar _ ) . _ type - ii _ bars grow on a secular evolution time scale to reach the same amplitude as in _ type - i_. these two main features of bar growth are illustrated in the appropriate figures below , although there are some intermediate cases present in our simulation . in many cases , the bar undergoes the well known buckling instability ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * and references therein ) following which the bar takes the form of a peanut and/or x shaped bulge . the buckling instability driven by the anisotropy in the velocity dispersions is an important phase of the bar evolution as it leads to the formation of a pseudo - bulge @xcite directly influencing the secular evolution of disk galaxies . however , the condition for the onset of this instability and the exact underlying cause of this phase remains a matter of further investigation . for example , recent work by @xcite has shown that the buckling instability weakens the bar and the resulting peanut shaped phase lasts for several gyrs in their simulation . we also find that the peanut shaped phase of the bar is long lasting , typically surviving till the end of our simulation ( @xmath106 gyr ) producing boxy / peanut / x - shaped structure in the central region . the morphological evolution of the disk in model mk97 is depicted through fig . [ fig : mk97t0 ] and fig . [ fig : mk97tf ] . these snapshots are taken at time @xmath107 and @xmath108 gyr which denote the beginning and the end of the simulation respectively . the initially axisymmetric disk evolves to form a bar which eventually transforms into a peanut morphology . inspection of the evolution of the bar - amplitude reveals a diverse behavior while it is in the peanut phase ; in some cases the amplitude continually grows and in others , it saturates . while the bar strength continues to increase , the evolution of its pattern speed is not always correlated in the sense that the pattern speed does not always decrease . in particular , the bar pattern speed in some cases remains nearly unchanged with time while its amplitude grows ( as in mk17 , mk33 , mk104 ) . a similar anti - correlation between the bar growth and pattern speed evolution has been discussed by @xcite and more recently by @xcite . however , there are also the normal cases where the pattern speed decreases with time , while the bar grows to a higher amplitude , by losing angular momentum to the halo via dynamical friction . we notice that the bar growth is slow ( typically a factor of @xmath52 to @xmath97 less ) in cases where the pattern speed is observed to remain nearly constant in time . overall , the emerging diverse behaviour of the bar characteristics precludes detailed understanding for inferring gross physical properties ( e.g. bar size , velocity dispersions ) of the disk based on a few simulation results . as a consequence , the results of a large number of simulations have been used to approach the problem in a statistical sense . is the initial value of the vertical dispersion at this radius . ] for each of the model galaxies , we compute the @xmath109 fourier component ( @xmath110 ) of the particle distribution from each n - body snapshot using the following formula : @xmath111 in the above equation @xmath112 is the phase of the @xmath113 particle at position @xmath114 and time @xmath0 and @xmath115 represents the number of particles used . since @xmath116 is a complex number , the modulus of the @xmath117 fourier component is obtained by @xmath118 @xcite and the corresponding phase is given by @xmath119 , ( @xmath120 ) . in general , the radial variation of @xmath110 shows a pronounced peak corresponding to a bar in the central region of the disk and often a second peak indicating the presence of a spiral perturbation in the outer parts of the disk . the variation of the peak amplitude of @xmath110 in the inner region with time reveals the growth of a bar in the galaxy . the amplitude @xmath110 is normalized to the amplitude of the axisymmetric mode ( @xmath121 fourier component ) . in all the subsequent figures , we present the smoothed normalized @xmath110 . for a wide range of model bars , the following linear regression model is fit to the time evolution : @xmath122 where the slope @xmath123 is the logarithmic time derivative of the bar amplitude . the linear model in the log - log plot translates to the bar amplitude evolving as @xmath124 . in the following , @xmath125 represents the growth rate of a bar and @xmath126 represents the growth rate of a spiral ( in the outer region ) . the time evolution of the normalized bar amplitude in model mk97 is illustrated in fig . [ fig : bsmk97 ] . the bar amplitude nearly saturates through a series of high frequency oscillations while undergoing the buckling phase . such oscillations have also been reported in previous simulations by @xcite and @xcite . our estimate shows that the typical period of such oscillations is @xmath127 or less than @xmath64 in general . this roughly implies that the amplitude oscillates with a frequency close to @xmath128 , the radial epicyclic frequency . the vertical velocity dispersion ( @xmath67 ) of the stars in the inner region of the disk is plotted as a function of time in fig . [ fig : sigzinmk97 ] for model mk97 . from fig . [ fig : sigzinmk97 ] , it can be seen that @xmath67 changes its slope at @xmath129 12 rotation time scales , nearly following the overall amplitude evolution of the bar . the best fit parameters of the heating model ( eq.[4 ] ) are @xmath130 , @xmath131 and @xmath132 . using these parameters in eq.(5 ) , we obtain @xmath133 . the heating time scale due to the bar growth is found to lie in the range @xmath134 . we have also applied the diffusion model given by eq.(7 ) to the inner region of mk97 and find that the heating model given by eq.(4 ) provides a better fit than the diffusion model . in particular , the diffusion coefficient @xmath135 and @xmath136 are found to be absurdly large . the resulting time scale @xmath137 . interestingly , the temporal variation of @xmath67 does not reflect any such oscillation while the radial velocity dispersion undergoes oscillation with nearly the same frequency as the bar amplitude beyond about @xmath97 rotation time scales . the growth of a bar in model mk112 is similar to that in model mk97 , although the disk in model mk112 is hotter ( @xmath138 ) than mk97 . [ fig : bsmk112 ] shows the normalized bar amplitude and the time evolution of @xmath67 in the disk inner region is shown in fig . [ fig : sigzinmk112 ] . the best - fit parameters from model mk112 are found to be similar to mk97 . in the case of model mk107 , the bar does not reach saturation and continues to grow as shown in fig . [ fig : bsmk107 ] . the vertical velocity dispersion in the inner region of the disk grows more gradually ( fig . [ fig : sigzinmk107 ] ) following the evolution of the bar . note that the overall increase in the vertical velocity dispersion is less than that of model mk97 and mk112 . we find that the heating model given by eq.(4 ) better approximates the simulation data than the diffusion model . the resulting diffusion coefficient @xmath139 is similar to model mk97 and @xmath140 . for models mk97 , mk107 and mk112 , we find that , in general , the diffusion model overestimates the dispersion values in the inner region of the disk . in all our simulations , the vertical velocity dispersion increases following the growth of the bar . note that all the three galaxy models ( e.g. mk97 , mk107 and mk112 ) undergo a phase of buckling instability during their evolution , but _ the disk stars are continually vertically heated prior to and during / after the buckling instability phase . _ this fact suggests that the vertical heating of the disk stars in our simulations is not strictly related to the bar buckling phase ( e.g. the disk in model mk17 does not go through buckling instability within the simulation time period but still shows continuous heating in the vertical direction ) . however , we note that the rate of vertical heating during the bar buckling phase is generally higher than in phases without such buckling . in the presence of a rotating bar ( @xmath109 ) potential , the planar motion of the disk stars can be coupled with the vertical oscillation ( parametric resonance ) at the location of the vertical resonances @xmath141 , where @xmath142 is the vertical frequency , @xmath143 is the bar pattern speed , @xmath144 is the number of vertical oscillations . for @xmath109 perturbation , the @xmath145 represent the vertical lindblad resonances . for the @xmath146 resonances , the retrograde orbits in the inner region of the galaxy couple efficiently with the vertical motion through the binney instability strips @xcite . @xcite has shown that indeed these @xmath99 resonances inside the bar play a significant role in trapping stars , leading to rapid diffusion in the vertical direction . in the presence of a growing bar potential , the disk stars are subjected to a complex instability and the main effect of such an instability is to promote fast diffusion of the low angular momentum , high z amplitude and energetic stars to the nearby halo . the instability is shown to grow as the bar perturbation grows . basically , the presence of a growing bar potential ( i.e. a time dependent potential in the galaxy ) breaks the time invariance symmetry ( because bar growth is an irreversible proprocess ) and hence the jacobi integral which in turn enhances the ergodicity considerably allowing the stars to visit most of the chaotic phase space @xcite . this provides a basic understanding of the heating mechanism in the presence of a growing bar . of course , in n - body simulation the nature of bar growth is quite diverse as mentioned in the beginning of this section . the resonant heating of the disk stars in the presence of a rotating ( with changing pattern speed ) growing ( in amplitude ) bar potential is probably more complex , leading to shifting locations of the resonances in the inner part of the galaxy and at the same time promising in the context of vertical heating in the disk . it is natural to examine how the vertical heating of the disk stars is correlated with the growth of bar , the nature of the vertical heating due to a growing bar , and the typical exponent for the vertical heating . in section 6 , we seek the possible correlations between the bar growth rate ( @xmath125 ) and the inner heating exponent ( @xmath70 ) . ) of the galaxy model mk97 as a function of time . the solid line is the same model fitted to the simulation data . @xmath100 is the initial value of the vertical dispersion at this radius . ] except for galaxy model mk112 . ] ) of the galaxy model mk107 . solid line is the same model fitted to the simulation data . @xmath100 is the initial value of the vertical dispersion at this radius . ] strong two - armed spirals are difficult to excite and form in a radially hot disk galaxies . using the fourier decomposition of the disk surface density , however , many transient spirals ( often with no steady pattern speed ) are detected in our model galaxies . these transient spirals are often diffuse and generally weak in radially hot disks ( fig . [ fig : bsmk97 ] and fig . [ fig : bsmk112 ] ) as compared to the relatively strong spiral formed in the radially cold disk model mk107 ( see fig . [ fig : bsmk107 ] ) . these spirals are primarily confined to the outer region of the stellar disk , and we examine whether they play a role in the vertical heating of the disk stars in this region . in figs . [ fig : sigzoutmk97 ] , [ fig : sigzoutmk112 ] and [ fig : sigzoutmk107 ] , we present the time evolution of the vertical velocity dispersion in the outer region of the disk for models mk97 , mk112 and mk107 respectively . the slope of the heating curve in model mk107 ( concave ) is quite different from the other two models ( convex ) presented here . in the later stages of the evolution , @xmath67 in model mk107 increases faster than that in mk97 and mk112 ( where @xmath67 nearly saturates ) . using the best - fit parameters of our heating model fitted to the outer region of mk97 , we find that @xmath147 where @xmath148 , @xmath149 and @xmath150 . applying the diffusion model in the outer region of mk97 , we obtain a diffusion coefficient @xmath151 and @xmath152 and the corresponding time scale @xmath153 , which is similar to the relaxation time scale ( @xmath94 ) quoted in section 4 . in the case of model mk107 , the best - fit parameters from our heating model yields @xmath154 . the diffusion model in this case gives @xmath155 and @xmath156 . these time scales indicate that the process of vertical heating in the outer region of the disk is slower in comparison to the inner region . we find that in the presence of a relatively strong spiral in the outer region , disk stars are heated to a greater degree in the vertical direction and the heating rate is found to be faster . the actual heating time scale , @xmath73 , ( in gyr ) in mk107 is lower by a factor of @xmath157 in comparison to models mk97 or mk112 . although , the exact mechanism through which stars are vertically heated in the outer region is not clear , growing spirals definitely play a significant role . the outer region of the stellar disk is not just simply described by the transient spirals , as corrugation waves or mild warps of the disk midplane are often present . it is known that the transient spirals can heat the disk efficiently in the radial direction , but poorly in the vertical direction . however , gmcs could redistribute the random energy of the stars in the vertical as well as in the radial direction through scattering processes . in the absence of gmcs or massive objects ( such as halo black holes or dark clusters ) in the outer parts of the galaxy , one seeks a process whereby the heating due to spirals or some other feature in the disk is redistributed in the vertical direction . possible candidates , for example , include corrugation waves , large scale warps of the stellar disk or , in general , vertical motion of the disk stars coupled with the transient spirals . the growth and development of these bending waves or rather warps is known to a large extent to depend on the noise in the n - body system . we have verified that as the number of particles is increased ( say by a factor of @xmath97 ) in the system , the growth of these bending waves is significantly reduced because of the substantial decrement in the poisson noise arising from particle discreteness . with an average mass resolution of @xmath158 , only weak bending waves or mild warping of the stellar disk are present ; comparatively strong warps can arise with a poorer mass resolution @xmath159 . thus , it is unlikely that the bending waves alone can heat the disk stars in the outer region with higher mass resolution . however , the possibility exists that the non - linear coupling between the weak transient spirals and the weak bending waves @xcite could redistribute energy in the vertical direction . as mentioned earlier , an interesting behaviour about the bars in radially hot disks inside a live dark matter halo was noticed . it was found that the _ type - i _ and _ type - ii _ bars trigger transient spirals with distinctive characteristics in the outer disk . in the case of a _ type - ii _ bar , the transient spiral continues to grow in amplitude , and this process continues as the bar grows ( see fig . [ fig : bsmk107 ] ) . this is found to be the case even in one of our hottest disk models ( mk55 for which @xmath160 , see table [ paratab ] ) . because of the growing transient spirals , we find that the radial heating dominates over the vertical one throughout the disk . on the other hand , the amplitude of the spirals nearly saturates in the case of models mk97 and mk112 ( see fig . [ fig : bsmk97 ] and fig . [ fig : bsmk112 ] ) where the bar is of _ type - i_. note that in this case , the radial heating nearly saturates throughout the disk ( see fig . [ fig : heatcomp_rzin ] and fig . [ fig : heatcomp_rzout ] ) . overall , there is a parallel development prevailing in the inner and the outer regions of the disk whereby a bar and spiral grow respectively . an important aspect associated with the heating of disk stars is its anisotropic nature as can be inferred from the fact that the observed ratio of the vertical to radial velocity dispersion i.e. @xmath161 ( as in ngc 488 * ? ? ? * ) , as found to occur for the solar neighbourhood in the galaxy . this fact raises many questions . for example , why do disk galaxies have @xmath161 ? can the heating mechanism preserve an initial @xmath33 value ? what causes the stars to be heated anisotropically ? to gain some insight into these questions , we illustrate in fig . [ fig : heatcomp_rzin ] and fig . [ fig : heatcomp_rzout ] the evolution of the radial and vertical velocity dispersions in the inner and outer region of the model galaxy mk112 . it can be seen that the overall nature and the rate of heating is different for the radial and the vertical directions . in the inner region of mk112 , the vertical heating is more effective than the radial heating and in the outer regime the radial heating saturates while the vertical heating continues slowly . we find this trend to be the case for most of the galaxy models which are radially hot and in which the bar evolves to saturation . on the other hand , for model mk107 which is relatively cold ( @xmath162 ) , the radial heating dominates over the vertical heating throughout the disk due to the presence of a relatively strong spiral perturbation . it is clear that the non - axisymmetric structures ( such as bars or spirals ) are essential for disk heating . and @xmath163 are the values of initial dispersions at that radius . solid line denotes vertical heating and dashed line the radial heating . ] and @xmath163 are the values of initial dispersions at that radius . solid line denotes vertical and dashed line the radial heating . ] the role of a live or responsive dark matter halo has been emphasized by @xcite in the context of bar formation in n - body simulations . unlike the case of a rigid dark matter halo , a live halo facilitates the growth of the bar in an n - body disk , likely through resonant interactions with the disk particles . on the other hand , the finite number of particles employed in the n - body simulation can be an important source of disk heating as well . because of the finite number of particles , two - body relaxation processes @xcite can be effective in heating the disk stars efficiently . it is highly desirable to eliminate or , at least , minimize the poisson fluctuation in the halo particle distribution by increasing the number of halo particles employed in the simulation as permitted by the available computational resources . to quantify this effect , we performed four additional simulations on the galaxy model dma ( see table [ paratab ] ) with different number of halo particles ( @xmath164 ) . there is an order of magnitude variation in the number of halo particles between models with the poorest resolution ( 0.5 million ( m ) particles ) and the highest resolution . the intermediate two simulations use @xmath40 and @xmath101 million halo particles . in fig . [ fig : hnoise_in ] and fig . [ fig : hnoise_out ] , we show the four vertical heating curves in the inner region and in the outer region respectively . it is found that the vertical heating curves nearly converge for models with @xmath165 . the convergence on the number of particles in our n - body simulation is in accordance with the conclusions reached in the recent work by @xcite in the context of bar formation and evolution . comparison of fig . [ fig : hnoise_in ] and fig . [ fig : hnoise_out ] reveals that the effect of halo noise is more important in the central region than in the outer region . in order to quantify this trend , the time scales @xmath73 in the inner and the outer region were determined . we find that an order of magnitude increase in the halo particle number increases @xmath73 by at least a factor of @xmath97 in the inner region and @xmath166 in the outer region . in contrast , we find different behaviour when the temporal variation of the radial velocity dispersion in the disk was examined . because these models are dominated by spiral arms in the outer region and we find that by increasing the number of halo particles ( from 0.5 m to 5 m ) the spiral arms became even stronger . this resulted in a lower radial dispersion in models with lower number of halo particles compared to the models with higher number of halo particles . in the case of model mk112 , the spiral arms are rather weak and the radial heating of the disk stars could have resulted from the halo noise . however , we find that the radial dispersion remains nearly constant over more than @xmath97 rotation time scales ( see fig . [ fig : heatcomp_rzin ] and fig . [ fig : heatcomp_rzout ] ) . based on these facts , it is possible to conclude that the noise due to the halo ( with @xmath167 few million particles ) is neither effective nor the dominant source of vertical heating in the outer region of the disk . ) of the galaxy models with different number of dark matter halo particles . solid line is for 0.5 m , dashed line for 2 m , dash - dot line for 3 m , and dotted line for 5.0 m particles ] in summary , we find a positive correlation between the growth of the bar at the galactic center and the vertical heating taking place in the central few kpc region . the formation of a bar is almost unavoidable in a stellar disk embedded in a live dark matter halo under the wide range of physical parameter space that has been explored . bar formation occurs even in a very hot galaxy with its time for onset delayed at higher numerical resolution . in addition , analysis based on the four simulations on model dma reveals that the strength of the spiral arms also increases with higher mass resolution . it has been noted that the vertical heating exponent is generally larger in the outer regions of a galaxy in the presence of strong spirals . hence , the transient spirals in the 3d disk which are common in our simulations not only play a significant role in radial heating , but also the vertical heating of disk stars . ) in the inner region ( @xmath168 ) for all the models . dependence on the bar growth rate ( @xmath169 ) . ] in the absence of a clear understanding of the physical processes responsible for the vertical heating of disk stars , we approach the problem by seeking possible correlations between the average physical properties ( e.g. , bar strength versus vertical heating rate ) from our fairly large sample of n - body galaxies . to obtain a measure of the strength and slope of possible relations , we carry out a statistical correlation , measuring it between two variables x and y by the pearson s product - moment correlation coefficient defined as follows : @xmath170 in the above equation , @xmath171 represents the covariance and @xmath172 denotes the standard deviation of the particular variable in the subscript . the summation is carried over the sample size @xmath173 . according to the cauchy - schwarz inequality the correlation can not exceed 1 . if the value of the correlation is positive , it indicates an increasing linear relationship and in the case of negative value , it denotes a decreasing linear relationship . in fig . [ fig : sigzinmk97 ] , fig . [ fig : sigzinmk112 ] , and fig . [ fig : sigzinmk107 ] the time evolution of the vertical velocity dispersion in the inner region of the disk is illustrated . in comparison with the evolution of the bar , it appears that the vertical velocity dispersion of the stars roughly follows the time evolution of the bar . thus , for individual galaxies , there is a trend for the vertical heating , and it is interlinked with the growth of the bar . however , the exact relationship between the two and the reason for the vertical heating of the disk stars closely following the growth of the bar is unknown . in the absence of a definitive answer , we seek to determine whether the trend found in individual galaxies is generic . for example , does it depend on the initial condition , and is this trend robust when examined on a large sample of galaxies with different initial conditions ? we note that these are difficult questions to answer because galaxies with varying initial conditions evolve quite differently . as pointed out recently by @xcite a live dark matter halo interacts with the disk in a more complicated way than previously envisioned . in fig . [ fig : corr_barsigz ] , the dependence of the vertical heating exponent ( @xmath70 ) in the inner region of the galaxy on the growth rate of the bar ( @xmath125 ) is illustrated for our sample of model n - body galaxies . the figure clearly shows that the heating exponent in the inner region of the disk is strongly positively correlated with growth rate of the bar with a value of @xmath174 . as pointed out earlier , when examined on a case by case basis , greater vertical heating results in galaxies which host stronger bars ( e.g. , models mk56a , mk52 , mk112 ) . our analysis suggests that as the strength of the bar increases , the disk stars continue to be heated vertically , irrespective of the bar s pattern speed evolution . the heating exponents tend to be higher in models with comparatively less massive dark matter halos . however , for more massive halos , the vertical heating exponent ( @xmath70 ) tends close to unity . on the other hand , in the absence of a strong bar , the heating exponent is lower , @xmath175 . ) in the outer region ( @xmath69 ) for all the models . dependence on the growth rate of spirals ( @xmath176 ) . ] a weak correlation is found between the growth of spirals and the heating exponent in the outer region as can be seen from fig . [ fig : corr_spiralsigz ] corresponding to a value of @xmath177 . its low value indicates that there may be more than one mechanism operating . although the correlation between the heating exponents and the growth exponents for the spirals is not strong , higher exponents are associated with the presence of stronger spirals ( e.g. , model mk43 , mk55 , mk64 ) and the overall vertical heating is low when the spiral arms are more diffuse and weak ( e.g. , models mk5 , mk51a , mk48 ) . the vertical heating in the outer region is less extensive than the central region . the heating exponents in the outer region in many models are very different from @xmath178 indicating that the disk stars are not heated purely by massive objects in the halo or by the transient spirals , however , the exact source(s ) which are responsible for the vertical heating in the outer region have yet to be identified . on the other hand , we find that there is a relatively strong correlation ( see fig . [ fig : corr_barspiral ] ) between the strength of the bars and the strength of the spirals in the disk with a value of @xmath179 . this correlation indicates that in the presence of a bar there is always at least a weak spiral present in the outer disk . this may also suggest that such weak transient spirals are triggered by the presence of a bar in the disk central region . however , the vertical heating in the inner region and the outer region is not well correlated . as the correlation between the two is very weak @xmath180 , the source of vertical heating in the inner and the outer region of the disk is likely to differ . ) and spirals ( @xmath176 ) for all the model galaxies . the figure shows that atleast weak spirals are formed in the disk whenever there is a bar . ] we have studied the nature of vertical heating of the disk stars in a large sample of radially hot galaxies as a function of halo mass and @xmath33 spanning a wide range . for an individual galaxy , we find that the stars in the inner region are heated vertically as long as a bar continues to grow . such correlation is quite generic and found to be robust when examined on a larger sample of model galaxies constructed from different initial conditions . bars form in galaxies with diverse initial conditions but do not evolve in a comparable fashion likely reflecting their non - linear interaction with the surrounding dark matter distribution . however , we find that on a broad category there are two kinds of bar : _ type - i _ ( growing rapidly and approaching saturation ) and _ type - ii _ ( growing slowly with no evidence of saturation ) as mentioned in section 5.1 . heating due to the growth of a bar differs from that due to a diffusion process since it occurs on a time scale of a few rotations in the disk central region . the diffusion time scale in a stellar system is similar to the relaxation time scale because diffusion mainly causes the stars velocity to change via two body encounters . thus from our analysis , it can be said that the diffusion approximation is likely to be inappropriate for the vertical heating of disk stars in the central region . the boxy / peanut bulges are commonly believed to originate from the galactic bar via the vertical heating ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * and references therein ) and references therein ) . the growth of a bar in the n - body simulation with a live halo is a complex process because of the non - linearity in the growing process and its interaction with the dark matter halo particles . the growing bars in our simulation are , at times , associated with a nearly constant pattern speed and , at other times , with a decreasing pattern speed . such a growing bar is likely to drive the inner disk to a non - equilibrium state and depending on its strength and pattern speed , the inner stellar disk may evolve on @xmath129 a few dynamical time scales . fundamentally , the growth of the bar facilitates the capturing of the disk stars into its various vertical resonances , e.g. the 1:1 ; 2:1 ; 4:1 vertical ilrs ( inner lindblad resonances ) etc . which are often present in the bar region of the disk @xcite . as shown by @xcite , the 4:1 vertical resonance becomes stronger as the bar grows , capturing disk stars to form the boxy bulge seen in many n body - simulations , although such studies do not explicitly include the interaction of the disk stars with a live dark matter halo . furthermore , the presence of weakly dissipative orbits in a barred galaxy potential is shown to produce appreciable z - amplification as resonances are crossed , hence facilitating the growth of boxy bulges @xcite . for example , a simple hamiltonian model of the growth of a bar perturbation @xcite illustrated the effect of resonant capture of the stars into the 2:1 vertical ilr , lifting them out of the disk plane to explain the growth of the boxy / peanut bulge . one common result amongst these studies is that the vertical resonances ( when present ) are efficient in capturing disk stars , facilitating their motion perpendicular to the disk plane , essential for the growth of a boxy bulge . we note that , in this context , the disk stars in our simulation are not only vertically heated at the resonance locations but throughout the bar region of the disk and often with similar magnitude . this could imply either the resonant heating is not the only mechanism responsible for the overall vertical heating or broad resonances reflecting strong chaos are present throughout the bar region @xcite . although a growing bar with / without a decreasing pattern speed could promote resonance sweeping through the bar region , it appears unlikely for the disk to accommodate the later scenario making it unclear how the non - resonant stars are vertically heated . in any case , it would be useful to have further insight into how the non - resonant stars might be contributing to the growth of the bulge . the disks of many galaxies in our sample are superthin and are dark matter dominated . from the galaxy formation point of view , it is extremely important to understand how these superthin galaxies are formed and how they evolve . can they preserve their initial superthinness ? in the hierarchical structure formation scenario , a galaxy grows via a large number of mergers and as a result , the n - body simulations of galaxy formation underestimate the number of thin galaxies . so it remains a puzzle how superthin galaxies remain superthin . assuming hydrostatic equilibrium , it can be shown that the disk half - thickness @xmath181 ; where @xmath182 is the mid - plane volume density . because of various unavoidable heating processes @xmath67 would always increase with time and hence the disk thickness ( @xmath26 ) . thus , either superthin galaxies somehow maintained their initial thickness or they evolved from an even thinner state to the present day superthin state . in the absence of environmental influences , the evolution of these galaxies would depend on the non - axisymmetric structures produced in the disk through internal instabilities . our simulation shows that the non - axisymmetric structures such as bars or spirals are spontaneously formed in the disk and heat the disk stars in their respective regime of dominance . we briefly mention here that normally in radially hot and vertically very cold galaxies ( e.g. mk25 , mk27 , mk48 for which initial @xmath183 ) , the disk thickness remains in the superthin regime with the final half thickness @xmath184 after @xmath52 to @xmath157 gyr of evolution . most of these galaxies form a _ type - ii _ bar but there are exceptions . if the initial thickness is greater than @xmath185 or @xmath186 , then the final thickness evolves to be in the thin ( @xmath187 ) regime . the vertical heating in the outer region is comparatively more complicated because of the presence of spiral arms and , at times , mild warps or corrugations . the presence of these perturbations complicate the identification of the heating mechanism since more than one may operate simultaneously . it is known that a stationary spiral pattern does not lead to heating ( except at resonance locations ) and transient spirals heat more effectively in the radial direction . however , our analysis indicates that the transient spiral arms also grow ( e.g. , model mk107 ) i.e. they are time dependent and , hence , a similar situation may prevail in the outer disk as in the central region . in both regions , the disk stars respond to the time - dependent potentials . thus , a common mechanism where the growing bars and spirals interact resonantly with the dark matter halo particles and contribute to the vertical heating in their respective regions of dominance may be operating . our investigation of the 70 n - body models of disk galaxies reveals a positive correlation between the growth of a bar and the vertical heating of the disk stars in the central region . a growing bar seems to contribute significantly to the vertical heating of the disk stars . overall , the heating exponent @xmath188 for the various galaxy models studied . the disk stars in the central region are generally heated in the vertical direction by a factor @xmath189 to @xmath102 above their initial values over @xmath52 to @xmath157 gyr . we find that the transient spirals are always present whenever a growing bar is present at the center . most of these spirals are weak and diffuse , and it is likely due to the fact that the disks in our simulations are relatively radially hot . the numerical results show that the amount of vertical heating in the outer region is lower compared to the inner region with the disk stars in the outer region generally heated vertically by a factor of @xmath166 above their initial values over a time period of @xmath190 gyr ( if there is a growing spiral present ) or more ( otherwise ) . from the analysis of our simulations , we find that , in general , in radially hot galaxies the vertical heating in the central region dominates over radial heating and in the outer region the relative importance is reversed . in contrast , radial heating dominates over the vertical heating throughout the disk for radially cold galaxies . we conclude that heating due to non - axisymmetric structures appears to be most promising in the context of disk heating problem in general . our simulation results suggest that there is likely a common physical process through which the disk stars are heated vertically , which is active throughout the disk from inner to the outer region . such a process should be investigated in detail by studying the vertical motion of the disk stars in the presence of a time dependent perturbing potential which could arise due to a growing bar or spiral arms inside a live dark matter halo . all the simulation and analysis presented in this paper are carried out in 3 clusters namely the asiaa computer cluster , tiara cluster and the academia sinica grid computing ( asgc ) center . the authors would like to express their thanks for the generous computing support from all of them . the authors also thank the referee , daniel pfenniger , for encouraging and useful comments which improved the presentation of the paper . ks would like to thank yao - huan tseng , sam tseng , and jason shih for helping with the computer resources at various stages . ks would like to thank the alexander von humboldt foundation for its financial support during which the the paper was written .	the nature of vertical heating of disk stars in the inner as well as the outer region of disk galaxies is studied . the galactic bar ( which is the strongest non - axisymmetric pattern in the disk ) is shown to be a potential source of vertical heating of the disk stars in the inner region . using a nearly self - consistent high - resolution n - body simulation of disk galaxies , the growth rate of the bar potential is found to be positively correlated with the vertical heating exponent in the inner region of galaxies . we also characterize the vertical heating in the outer region where the disk dynamics is often dominated by the presence of transient spiral waves and mild bending waves . our simulation results suggest that the non - axisymmetric structures are capable of producing the anisotropic heating of the disk stars .
studies of the atomic interstellar medium ( ism ) , through observations of the 21-cm line of atomic hydrogen ( hi ) , have proven to be critical for our understanding of the processes leading to star formation , the dynamics and structure of the ism , and the ( dark ) matter distribution ; thus touching on major issues related to galaxy formation . in the last decades remarkable progress has been made in these areas . however , the lack of high - resolution hi observations in a representative sample of nearbly galaxies precludes a systematic study of the physical characteristics of the atomic ism . here we present first maps of things , ` the hi nearby galaxy survey ' , the largest program ever undertaken at the vla to perform 21-cm atomic hydrogen observations of the highest quality of nearby galaxies . the goal of things is to obtain high - quality observations of the atomic ism ( through observations of the 21-cm line of neutral atomic hydrogen ) of a substantial sample of nearby galaxies , covering a wide range of hubble types , star formation rates , absolute luminosities , evolutionary stages , and metallicities . the database will have homogeneous sensitivity and spatial and velocity resolution at the limit of what can currently be achieved with the very large array . the angular resolution will be @xmath3 , and the spectral resolution will be @xmath1 km s@xmath2 . all galaxies will be observed with the vla b , c and d arrays . a more complete description of things , survey parameters and observational and technical details will be presented in walter et al . . a list of sample targets is given in table 1 . as an example of the high quality of the things data , figs . 1 and 2 show high - resolution integrated hi column density maps of four galaxies from the sample . data from things will complement sings , the spitzer infrared nearby galaxies survey @xcite ; hence high - quality observations from the x - ray through the radio will be available at comparable resolution for each galaxy . data from things can be used to investigate issues such as the small - scale and three - dimensional structure of the ism , the ( dark ) matter distribution , and the processes leading to star formation . furthermore , things will enable studies of the variation of each of these properties as a function of galaxy environment . things is designed to address numerous scientific topics , including , but not limited to , the following : * * dust atomic ism . * the resolution of the things hi data is similar to that obtained for the intermediate wavelength bands on - board spitzer . one of the main goals of things is therefore a comparison between the spatial distributions of the hi and dust as measured by spitzer . * * interplay ism star formation . * things allows studies of the interplay between star formation and the ambient ism at 100 - 300 pc resolution over a range of hubble types . the location and energy input of regions of recent star formation and the impact they have on the structure and dynamics of the hi will be investigated . for example , for the first time a census will be possible of supergiant shells as a function of hubble type . things data products will enable studies of how these structures form and how they , in turn , might trigger secondary star formation . using the multi wavelength data from sings a complete energy budget of the ism can be derived . * * global mass distribution . * the things data products will enable studies of the dark matter distribution in galaxies at high angular resolution . high resolution data near the centres of galaxies are especially important as this is the regime where differences between the cuspy and constant - density haloes show themselves most clearly . the data from spitzer and additional data at other wavelengths will deliver more reliable estimates for the stellar mass to light ratio of the disk , obviating the need to resort to the `` canonical '' maximum disk assumption in order to derive the properties of the dark matter halo . this study will ultimately shed more light on the validity of the cdm paradigm on scales of individual galaxies and will measure possible trends of the dark matter properties as a function of hubble type . * * star formation threshold . * using things , studies will be performed at high spatial resolution to investigate whether or not there is a ` universal ' star formation threshold and how , or even if , this is a function of galaxy type . combinations of things hi surface density maps with co data will allow one to create total gas surface density maps . amongst other things , with the velocity dispersion maps and the derived rotation curves , this information can be used to calculate spatially resolved `` toomre@xmath4 '' parameter maps for each galaxy ( the @xmath4 parameter is a measure for the local gravitational balance , @xcite ) as a function of hubble type . issues that can be addressed are , e.g. , in which regime does ` @xmath4 ' break down ? what is the importance of local ( disk or cloud instability ) versus global effects ( spiral density waves , tidal forces ) ? obervations for things are currently ongoing at the vla ( total observing time : @xmath5 hours , including archival data ) . the data acquisition will be completed at the end of 2005 . final data products of things will include moment maps ( integrated hi distribution , velocity fields , velocity dispersion ) as well as the hi data cubes , and will be made available to the community one year after the observations have completed . lllllrrcrrrr name & alias & type & ra ( 2000.0 ) & dec ( 2000.0 ) & v@xmath6 & w@xmath7 & size & incl + & & & h m s.s & @xmath8 @xmath9 @xmath10&kms@xmath2&kms@xmath2 & @xmath11 & @xmath8 + ngc628 & m74 & sc & 01 36 41.7 & + 15 46 59 & 656 & 77 & 10.5 @xmath12 9.5 & 35 + ngc925 & & sbcd & 02 27 16.9 & + 33 34 45 & 553 & 217 & 10.5 @xmath12 5.9 & 61 + ngc1569 & & ibm & 04 30 49.0 & + 64 50 53 & 90 & 83 & 3.6 @xmath12 1.8 & 68 + ngc2366 & & irr & 07 28 47.6 & + 69 11 39 & 99 & 114 & 8.1 @xmath12 3.3 & 90 + ngc2403 & & sbc & 07 36 51.4 & + 65 36 09 & 130 & 241 & 21.9 @xmath12 12.3 & 60 + holmbergii & ugc4305 & irr & 08 19 04.0 & + 70 43 09 & 157 & 72 & 7.9 @xmath12 6.3 & 47 + m81dwa & pgc23521 & irr & 08 23 56.0 & + 71 01 45 & 113 & 33 & 1.3 @xmath12 0.7 & + ddo53 & ugc4459 & irr & 08 34 07.2 & + 66 10 54 & 19 & & 1.5 @xmath12 1.3 & 45 + ngc2841 & & sb & 09 22 02.6 & + 50 58 35 & 638 & 607 & 8.1 @xmath12 3.5 & 68 + ngc2903 & & sbbc & 09 32 10.1 & + 21 30 04 & 555 & 384 & 12.6 @xmath12 6.0 & 56 + holmbergi & ugc5139 & irr & 09 40 32.3 & + 71 10 56 & 137 & 44 & 3.6 @xmath12 3.0 & 37 + ngc2976 & & sc & 09 47 15.3 & + 67 55 00 & 3 & 135 & 5.9 @xmath12 2.7 & 61 + ngc3031 & m81 & sab & 09 55 33.2 & + 69 03 55 & 35 & 442 & 26.9 @xmath12 14.1 & 59 + ngc3077 & & sd & 10 03 20.6 & + 68 44 04 & 13 & 90 & 5.4 @xmath12 4.5 & 41 + m81dwb & ugc5423 & irr & 10 05 30.6 & + 70 21 52 & 343 & 61 & 0.9 @xmath12 0.6 & 67 + ngc3184 & & sbc & 10 18 16.9 & + 41 25 28 & 591 & 151 & 7.4 @xmath12 6.9 & 24 + ngc3198 & & sbc & 10 19 54.9 & + 45 32 59 & 662 & 321 & 8.5 @xmath12 3.3 & 70 + ic2574 & ugc5666 & sbm & 10 28 21.2 & + 68 24 43 & 48 & 109 & 13.2 @xmath12 5.4 & 75 + ngc3351 & m95 & sbb & 10 43 57.8 & + 11 42 14 & 778 & 280 & 7.4 @xmath12 5.0 & 42 + ngc3521 & & sbbc & 11 05 48.6 & 00 02 09 & 809 & 462 & 11.0 @xmath12 5.1 & 66 + ngc3621 & & sbcd & 11 18 16.0 & 32 48 42 & 726 & 278 & 12.3 @xmath12 7.1 & 66 + ngc3627 & m66 & sbb & 11 20 15.0 & + 12 59 30 & 726 & 377 & 9.1 @xmath12 4.2 & 57 + ngc4214 & & irr & 12 15 38.9 & + 36 19 40 & 290 & 86 & 8.5 @xmath12 6.6 & 42 + ngc4449 & & irr & 12 28 11.2 & + 44 05 36 & 202 & 143 & 6.2 @xmath12 4.4 & 56 + ngc4736 & m94 & sab & 12 50 53.0 & + 41 07 14 & 309 & 232 & 11.2 @xmath12 9.1 & 35 + ddo154 & & irr & 12 54 05.2 & + 27 08 55 & 373 & 93 & 3.0 @xmath12 2.2 & 44 + ngc4826 & m64 & sab & 12 56 43.7 & + 21 40 52 & 406 & 315 & 10.0 @xmath12 5.4 & 60 + ngc5055 & m63 & sbc & 13 15 49.2 & + 42 01 49 & 502 & 400 & 12.6 @xmath12 7.2 & 56 + ngc5194 & m51a & sbc & 13 29 52.7 & + 47 11 43 & 463 & 199 & 11.2 @xmath12 6.9 & 30 + ngc5236 & m83 & sbc & 13 37 00.8 & 29 51 59 & 515 & 281 & 12.9 @xmath12 11.5 & 46 + ngc5457 & m101 & sbc & 14 03 12.5 & + 54 20 55 & 241 & 188 & 28.8 @xmath12 26.9 & 22 + ngc6946 & & sbc & 20 34 52.3 & + 60 09 14 & 51 & 235 & 11.5 @xmath12 9.8 & 31 + ngc7331 & & sab & 22 37 04.1 & + 34 24 56 & 816 & 520 & 10.5 @xmath12 3.7 & 71 + ngc7793 & & scd & 23 57 49.7 & 32 35 30 & 229 & 192 & 9.3 @xmath12 6.3 & 53 +	we describe _ the hi nearby galaxy survey ( things ) _ , the largest program ever undertaken at the vla to perform 21-cm hi observations of the highest quality ( @xmath0 , @xmath1 km s@xmath2 resolution ) of nearby galaxies . the goal of things is to investigate key characteristics related to galaxy morphology , star formation and mass distribution across the hubble sequence . a sample of 34 objects with distances between 3 and 10 mpc will be observed , covering a wide range of evolutionary stages and properties . data from things will complement sings , the spitzer infrared nearby galaxy survey . for the things sample , high - quality observations at comparable resolution will thus be available from the x - ray regime through to the radio part of the spectrum . things data can be used to investigate issues such as the small - scale structure of the ism , its three - dimensional structure , the ( dark ) matter distribution and processes leading to star formation . to demonstrate the quality of the things data products , we present some prelimary hi maps here of four galaxies from the things sample .
a central quantity for many algorithms operating on linear recurrences and their solutions is the dispersion . the _ dispersion set _ of two polynomials @xmath3 and @xmath4 in @xmath5 $ ] is the set of positive integer roots of the resultant @xmath6 . when this set is not empty , its maximal element is called the _ dispersion _ of @xmath3 and @xmath4 . thus , the dispersion is the largest integer difference between roots of @xmath3 and @xmath4 . as shown by the simple example @xmath7 with @xmath8 , the dispersion can be exponentially large in the _ bit _ size of the input polynomials . it can not get much worse : when the polynomials have integer coefficients whose absolute value is bounded by @xmath9 , their dispersion is bounded by @xmath10 ( * ? ? ? * fact 7.11 ) . this exponential size yields the dominant term in the worst - case complexity of many algorithms computing or operating on solutions of linear recurrences . for instance , the computation of a gosper - petkovek form produces a polynomial whose degree @xmath0 can be linear in the dispersion of its input and thus exponential in its bit size . if this polynomial is expanded it has @xmath11 coefficients ; over the integers , its total bit size is @xmath12 . this form is used in the first step of gosper s summation algorithm and of abramov s algorithm for computing rational solutions of linear recurrences . thus , it makes an important contribution to the complexity of these algorithms . once this form is computed , these algorithms search for polynomial solutions of an associated linear recurrence . this is done by linear algebra using a bound on the possible degree of solutions which is at least as large as @xmath0 , leading again to a more than quadratic complexity , even when no nonzero solution exists . in turn , a parameterized variant of gosper s algorithm forms the basis of zeilberger s definite summation algorithm which inherits this costly behaviour . by contrast , we provide a probabilistic algorithm that detects that no nonzero rational solution of a homogeneous linear recurrence exists in @xmath13 bit operations and a deterministic algorithm that gives a compact representation of all solutions in @xmath14 bit operations . all the algorithms in the present work eventually rely on the computation of polynomial solutions of linear recurrences . in a previous work @xcite , we dealt with the analogous problem in the linear differential case , by exploiting the linear recurrence satisfied by the coefficients of power series solutions and reducing the computation to that of _ matrix factorials_. for the latter operation , there exist fast probabilistic and deterministic algorithms ( see @xcite and the references in @xcite ) . in the case of linear recurrences , it is not true that the coefficients of polynomial solutions satisfy a linear recurrence in general ; however , it becomes true if the polynomials are expanded in a binomial basis ( * ? ? ? xiii , art . 5 ) . this is the basis of a simple quadratic algorithm @xcite to compute polynomial solutions . in section [ sec1 ] , we show how this conversion is performed , we recall the basic results on matrix factorials and apply them to get the announced complexities . from there , in section [ sec2 ] , we proceed in three steps : _ ( i ) _ we slightly modify the computation of the gosper - petkovek form so that it does not expand the potentially large polynomial but instead computes a first - order , moderately - sized recurrence for it ; _ ( ii ) _ we show that this first - order recurrence can be used to compute a linear recurrence satisfied by the numerators of rational solutions , in a complexity that is only logarithmic in @xmath0 , both in the homogeneous and nonhomogeneous cases ; _ ( iii ) _ we then compute the numerators as polynomial solutions via matrix factorials . the close relation between abramov s and gosper s algorithm makes it possible to transfer these results to gosper s algorithm . then in section [ zeilberger ] , we show how this machinery can be adapted to the parameterized variant needed in zeilberger s algorithm . finally , we describe experimental results in section [ experiments ] . all along this text , @xmath15 denotes a linear difference operator with coefficients in @xmath16 $ ] . we view it as a polynomial in the non - commutative ring @xmath17 , where @xmath18 is the shift operator @xmath19 . similarly , @xmath20 , @xmath21 , and @xmath22 denote the shifts with respect to @xmath23 , @xmath24 , and @xmath25 . to any difference operator @xmath15 is attached a homogeneous linear recurrence equation @xmath26 . we view the solution @xmath27 either as a sequence @xmath28 ( also denoted @xmath29 ) , or as a function @xmath30 ( the cases of particular interest being polynomial and rational functions ) . for our complexity analyses , the measure we use is the bit ( or boolean ) complexity . for this purpose , our complexity model is the multi - tape turing machine , see for instance @xcite . we use the _ number of bit operations _ to express time complexities in this model . we call _ bit size _ ( or simply _ size _ ) of an integer @xmath31 the number @xmath32 ( @xmath33 denotes the logarithm of @xmath23 in base 2 ) . by convention , we assume that @xmath34 . the bit size of a matrix or vector is the sum of the bit sizes of its entries . polynomials given as input to our algorithms are stored in a dense representation ; a measure of their bit size is given by the sum of the bit sizes of their coefficients , including the zero coefficients . similarly , the bit size of a linear recurrence equation ( lre ) is the sum of the bit sizes of its polynomial coefficients . to simplify complexity estimates , we assume that the product of two integers of bit size @xmath35 can be computed within @xmath36 bit operations using fast fourier transform @xcite . to keep the notation compact , we sometimes write @xmath37 ; the tilde indicates that the factors polynomial in @xmath38 or smaller have been omitted . for any prime number @xmath39 , the bit complexities of the operations @xmath40 in the finite field @xmath41 are in @xmath42 . we assume that over the rings we use , the product of two polynomials of degree at most @xmath35 can be computed within @xmath43 base ring operations ( each ring operation being counted at unit cost ) and that @xmath44 ( * ? ? ? * ch . 2 ) . for computations in @xmath45 $ ] , the bit complexity is bounded by multiplying the arithmetic cost estimates by the bit complexity of the basic operations in @xmath46 . in all our algorithms we are interested in reducing the complexity with respect to a potentially exponential parameter @xmath23 ( related to a dispersion or to a root of an indicial polynomial ) . thus we consider as having cost @xmath47 any operation whose complexity is polynomial in the bit size of the input recurrence or polynomials , and concentrate on the dependency of the complexity in @xmath23 . in order to provide the code with an actual bound on the size of primes that need to be used so that the bound on probability of error is guaranteed , we have to perform a much more precise complexity analysis taking into account all parameters ( order , degree of coefficients ) ( as in the proof of ( * ? ? ? * thm . 3 ) ) . such a detailed analysis will appear in @xcite . in symbolic summation and in the resolution of linear recurrences , all the known algorithms ultimately require polynomial solutions of linear recurrence equations . in this section , we give algorithms for computing descriptions of the @xmath48-vector space of solutions of a linear recurrence operator @xmath15 with coefficients in @xmath16 $ ] : @xmath49 we focus on two types of solutions of such recurrences : solutions with finite support and polynomial solutions . in what follows , we make the hypothesis that _ @xmath50 is an ordinary point of the recurrence_. this means that the leading coefficient @xmath51 does not vanish at any of the integers @xmath52 ; in other words , when unwinding the recurrence , no division by zero is encountered . this condition is ensured after a generic translation @xmath53 . under our complexity assumptions , a proper @xmath54 and the corresponding translation can be computed in a polynomial number of bit operations , so that there is no loss of generality for the problems we consider . the general case ( when 0 is not ordinary ) is technically more demanding but does not change the complexity estimates we give here . it will be presented in @xcite . let @xmath55 denote the vector space of solutions @xmath56 of . in the case of an arbitrary @xmath57 , the dimension of @xmath55 as a @xmath48-vector space may be different from @xmath58 ( both larger or smaller ) . however , when 0 is an ordinary point of @xmath15 , @xmath55 has dimension exactly @xmath58 and a basis is given by the sequences @xmath59 , @xmath60 , satisfying @xmath15 and having initial conditions @xmath61 , for @xmath62 , where @xmath63 is kronecker s @xmath64 symbol ( @xmath65 , @xmath66 if @xmath67 ) . in [ compact ] , we describe the compact representation that forms the basic data structure of our algorithms . then , in [ nthterm ] we recall classical results that allow for the efficient computation of the @xmath0th element of a solution of @xmath15 . in [ rectorec ] we describe the reduction from the problem of searching for polynomial solutions to that of finding solutions with finite support . next , we give in [ polysols ] algorithms to compute finitely supported and polynomial solutions of recurrences . we conclude this section by showing in [ compactoperations ] how the evaluation of a polynomial and its finite differences can be performed efficiently in the compact representation . classically , a polynomial solution @xmath30 of is represented by its coefficients in the monomial basis @xmath68 . we use an alternative data structure for @xmath30 , which is motivated by the observation that its coefficients @xmath69 in the binomial basis @xmath70 obey a recurrence with polynomial coefficients . [ ex : compact ] the recurrence @xmath71 has a unique nontrivial monic polynomial solution @xmath72 . to write down its coefficients in the monomial basis at least @xmath73 bits are needed . in contrast , @xmath30 can be represented by the recurrence @xmath74 on the coefficients @xmath69 of @xmath30 in the binomial basis @xmath70 ; the bit size of this new representation is only linear in @xmath75 , most of the size being in the initial condition @xmath76 . the _ compact representation _ of a polynomial solution of is the data of a linear recurrence and initial conditions for its coefficients in the binomial basis , together with an upper bound on its degree . our aim in this article is to demonstrate that this representation of polynomial solutions of recurrences can be carried through different algorithms from indefinite and definite hypergeometric summation and that it is beneficial from the complexity point of view . the reason why this representation deserves the name `` compact '' appears in [ polysols ] below . let @xmath28 be a sequence satisfying . the recurrence @xmath57 can be rewritten as a first - order matrix recurrence @xmath77 , where @xmath78 is the vector @xmath79 and @xmath80 is an @xmath81 matrix with rational function entries . the problem of computing a selected term @xmath82 reduces to that of computing @xmath83 and the _ matrix factorial _ @xmath84 . this makes sense since under our hypothesis the leading term of the initial recurrence does not vanish at @xmath85 . the numerator and denominator of the matrix factorial can be computed efficiently , either in @xmath86 using a _ binary splitting _ algorithm , or modulo a prime @xmath39 using a _ baby - step / giant - step _ algorithm . these algorithms are described in ( * ? ? ? * , 3.1 ) , see the references therein . for further use , we extract from @xcite the following result . [ thm : reccomplexity ] let @xmath87 be a sequence of vectors of rational numbers that satisfies a recurrence @xmath88 , with @xmath89 an @xmath90 matrix with rational function entries in @xmath91 . let @xmath39 be a prime number such that the denominator of @xmath92 does not vanish mod @xmath39 at @xmath85 . then , as @xmath93 : 1 . @xmath94 and @xmath95 have bit size @xmath96 ; their values can be computed using @xmath97 bit operations . 2 . @xmath98 and @xmath99 can be computed using + @xmath100 bit operations . 3 . the rank of @xmath101 can be computed in @xmath102 bit operations using a probabilistic monte carlo algorithm . for completeness , we recall here an algorithm from @xcite that we call to perform the conversion from a recurrence with polynomial coefficients to the recurrence satisfied by the coefficients of series solutions in the binomial basis . earlier ( and slightly more complicated ) algorithms have been given in ( * ? ? ? * chapter xiii ) and ( * ? ? ? * section 4.2 ) . the starting point are the following two identities : @xmath103 if @xmath104 , then applying these identities to rewrite @xmath105 and @xmath106 and extracting coefficients of @xmath107 shows that the ring morphism @xmath108\rightarrow{{\mathbb{q}}}[k , s_k , s_k^{-1}]$ ] defined by @xmath109 and @xmath110 sends a homogeneous lre satisfied by @xmath30 to another one satisfied by @xmath69 . the image of is a lre of the form @xmath111 where the leading term is exactly that of and the trailing term may involve a negative shift ( by convention , @xmath112 when @xmath113 ) . in particular , if @xmath50 is an ordinary point for @xmath57 , so is it for @xmath114 . the resulting algorithm is as follows . its complexity is clearly polynomial in the bit size of @xmath57 . we consider here the problem of computing a basis of solutions _ with finite support _ , that is , whose terms beyond a certain index are all zero . the _ degree _ of a solution with finite support @xmath27 is , by definition , the unique integer @xmath115 such that @xmath116 and @xmath117 , for all @xmath118 a universal bound @xmath0 on the degrees of all solutions with finite support of the input recurrence @xmath119 is given by the largest positive integer root of the trailing coefficient @xmath120 of @xmath57 . note that @xmath0 is generally not bounded polynomially in the bit size of @xmath57 . recall that @xmath55 has dimension @xmath58 , with a basis @xmath121 formed by the sequences @xmath59 , with @xmath122 , given by the initial conditions @xmath123 , for @xmath62 . thus , a finitely supported solution @xmath27 is ( an unknown ) linear combination @xmath124 such that the elements in the slice @xmath125 all vanish . this yields linear constraints on the initial conditions @xmath126 . to determine these constraints , it is sufficient to compute the values at indices @xmath127 of all the elements in @xmath121 using thm . [ thm : reccomplexity ] . the rank of the resulting @xmath81 matrix gives the dimension of the vector space of solutions with finite support . since the entries of this matrix have bit size @xmath96 , the desired @xmath126 s , which are determined by a kernel computation , also have bit size @xmath96 . putting together these considerations , we get the following result . [ th : finitesuppsols ] there exists a basis @xmath128 ( @xmath129 ) of solutions of recurrence ( [ rec ] ) with finite support , where each @xmath130 is uniquely specified by the data of initial conditions of bit size @xmath96 , with @xmath0 a bound on the integer roots of @xmath131 . the dimension @xmath35 as well as the degrees of the @xmath130 s can be computed by a probabilistic algorithm using @xmath132 bit operations . the initial conditions of the @xmath130 s and their maximal degree @xmath133 can be computed deterministically in @xmath134 bit operations . [ th : finitesuppsols ] is the basis for using the name `` compact representation '' : it shows that the compact representation has a size of the same order as the initial conditions , while the expanded polynomials have size @xmath12 . in general , this latter bound is reached . using the results in [ rectorec ] , thm . [ th : finitesuppsols ] carries over literally to the compact representation of a basis of polynomial solutions of the recurrence @xmath119 . the corresponding statement requires a bound on the degree of polynomial solutions that is given by the roots of the _ indicial polynomial_. the _ indicial polynomial _ of @xmath57 at infinity is the trailing coefficient of @xmath135 . [ cor : finitesuppsols ] the statement of thm . [ th : finitesuppsols ] holds for polynomial solutions of @xmath119 , with @xmath0 the largest integer root of the indicial polynomial of @xmath57 at infinity . * nonhomogeneous equations . * we now consider the equation @xmath136 , with coefficients in @xmath16 $ ] and right - hand side of degree @xmath25 . applying and expanding @xmath137 in the binomial basis , the initial problem boils down to the search of finitely supported solutions of a nonhomogeneous equation @xmath138 , where @xmath139 is a sequence with finite support , @xmath140 for @xmath141 . in matrix notation , we have @xmath142 , where @xmath143 is the vector @xmath144 and @xmath145 is the vector @xmath146 . then the vector of initial conditions @xmath83 satisfies the affine constraint @xmath147 , where @xmath148 , @xmath149 , @xmath150 , for @xmath151 and @xmath152 . for large @xmath0 , using thm . [ thm : reccomplexity ] , the matrices @xmath153 and @xmath9 can be computed efficiently . the bit size and the computational cost of @xmath154 is @xmath47 . thus , solving the affine system of size @xmath155 yields the finitely supported solutions of @xmath156 and the polynomial solutions of @xmath157 and we get the following . let @xmath158 be a polynomial . then the statement of thm . [ th : finitesuppsols ] holds for polynomial solutions of nonhomogeneous equations @xmath159 as the largest integer root @xmath0 of the indicial polynomial of @xmath57 at infinity becomes large . the compact representation is not only a data structure for intermediate computations . it can actually be exploited further . in particular , we now detail the evaluation at an algebraic number @xmath54 of a polynomial @xmath160 and an iterated difference @xmath161 ( where @xmath162 and @xmath163 is potentially large ) . the polynomial @xmath27 is given by its degree @xmath0 and the recurrence @xmath164 satisfied by its coefficients @xmath69 in the binomial basis @xmath165 , together with initial conditions . the basic idea is embodied in the following . if @xmath166 and @xmath167 are solutions of linear difference equations with polynomial coefficients , then so is the sequence @xmath168 defined by @xmath169 . this lemma can be applied to the sequences @xmath170 and @xmath171 . evaluating the resulting sequence at @xmath0 using thm . [ thm : reccomplexity ] gives @xmath172 for @xmath173 bit operations , when @xmath0 is large . using pascal s formula @xmath174 , we deduce that @xmath175 the recurrence satisfied by the sequence @xmath176 is obtained by shifting by @xmath163 the coefficients of the recurrence of @xmath170 . this new recurrence has bit size @xmath177 and initial conditions can be determined by binary splitting in @xmath178 bit operations . here , our asymptotic bound involves the two parameters @xmath0 and @xmath163 , as both are potentially exponential in the input size . as above , the compact representation of the recurrence satisfied by @xmath179 can be determined efficiently , as well as its @xmath0th term @xmath180 . the classical gosper - petkovek normal form @xcite of a reduced rational function @xmath181 in @xmath182 consists of three polynomials @xmath183 in @xmath5 $ ] such that @xmath184 with the constraints @xmath185 the degree @xmath0 of the polynomial @xmath186 is potentially exponentially large . thus , in our algorithm below , we modify the usual algorithm ( e.g. , in @xcite ) slightly so that the polynomial @xmath186 is not expanded . similar ideas appear in @xcite in the context of indefinite rational summation . @xmath187 . note that the input is an ordered pair @xmath188 and not a rational function @xmath181 . the output of the algorithm changes if @xmath188 is replaced by @xmath189 for @xmath190 $ ] . this will be necessary for our treatment of rational solutions below . on the other hand , the output @xmath153 , @xmath9 , and @xmath191 s also satisfy ( [ gp - ext ] ) whenever @xmath3 and @xmath4 have no common factor , so that the gosper - petkovek normal form of a rational function in @xmath182 given in reduced form @xmath181 is obtained by @xmath192 . as an outcome of this algorithm , the rational function @xmath193 ( @xmath194 ) is easily obtained as @xmath195 for large @xmath0 and @xmath196 , it has `` small '' numerator and denominator of degrees bounded by @xmath197 times those of @xmath3 and @xmath4 . this equation for @xmath198 is a homogeneous lre that plays the role of a compact representation of @xmath92 . the initial value @xmath199 ( more generally @xmath200 where @xmath201 ) has size @xmath202 and can be computed by thm . [ thm : reccomplexity ] within @xmath203 bit operations . in the next sections , we use this to design `` compact '' variants of abramov s and gosper s algorithms . [ logh ] algorithm is correct . for @xmath188 with rational coefficients , it has deterministic polynomial bit complexity in the bit size of @xmath188 . the correctness is that of the classical algorithm since the only difference is that we do not expand @xmath92 . step 1 is dealt with by a deterministic algorithm due to loos @xcite ( cf . @xcite for faster probabilistic algorithms ) . step 3 is performed at most @xmath204 times , and each step is polynomial by the classical algorithms as found in @xcite . we now consider rational solutions of the lre @xmath159 , with @xmath158 a polynomial in @xmath5 $ ] . our starting point is the following result of abramov @xcite . [ abramov - lemma ] the polynomial @xmath186 of the + gosper - petkovek form of @xmath205 is a multiple of the denominator of all rational solutions of @xmath159 . abramov s algorithm first computes @xmath186 , then performs the change of variable @xmath206 , leading to @xmath207 whose polynomial solutions @xmath208 are then sought . in the homogeneous case ( @xmath209 ) , using reduces this equation to an equation of polynomial size . this is described in algorithm homcompactratsols ( see figure ) . in step 2 , the `` normalize '' operation consists in expanding @xmath193 using and taking the numerator of the resulting expression . also , if necessary , we change @xmath115 into @xmath210 with @xmath211 , so that 0 is not a singular point in step 3 . this can be detected and changed at a cost of @xmath212 operations . in step 4 , the output is given by the compact forms of the numerators and @xmath92 is given by the output of . in the nonhomogeneous case , reducing to the same denominator would lead to an equation whose right - hand side has a potentially exponential degree . instead , we consider the homogeneous operator @xmath213 , whose bit size is polynomial in that of @xmath159 and that can be treated by the algorithm above . if @xmath29 is a rational solution of @xmath214 , then @xmath215 is a rational solution of @xmath216 . this implies that @xmath217 for all @xmath115 larger than the largest root of @xmath158 and since @xmath218 is rational , also for all other values of @xmath115 . thus fixing @xmath219 so that @xmath220 for any @xmath24 such that @xmath221 concludes the computation . this is the basis of the following algorithm . in step 2 , just iterating @xmath222 till a point where @xmath158 is found to be nonzero is sufficient for our purpose . if @xmath0 is a bound on the degree of the numerators and denominator computed in step 1 , then the values of the @xmath223 s in step 3 have size @xmath202 and can be computed by binary splitting . from there , it follows that the affine equation in step 3 has coefficients of size @xmath202 , which is then also a bound on the size of its solutions . these solutions can be computed in the form of a point and a basis of a vector space within @xmath212 bit operations by standard linear algebra . the same complexity is sufficient for the products of initial conditions in step 4 . the results of this section are summarized as follows . let @xmath0 be the sum of the largest nonnegative integer root of the indicial polynomial of @xmath224 at infinity and the degree of the polynomial @xmath186 of with @xmath225 and @xmath226 . the dimension of the affine space of rational solutions of @xmath159 can be computed probabilistically using @xmath227 bit operations . a compact representation of the solutions can be computed deterministically in @xmath212 bit operations . the largest integer root of the indicial polynomial of @xmath224 at infinity is a bound on the valuations of power series solutions of @xmath119 at infinity , including the valuation of @xmath228 . adding the degree of @xmath92 gives the announced bound on the degree of polynomial @xmath229 s . from there , the theorem follows from cor . [ cor : finitesuppsols ] . given a hypergeometric term @xmath230 , i.e. , such that @xmath231 , gosper s algorithm @xcite finds its indefinite hypergeometric sum , if it exists . such a sum is necessarily of the form @xmath232 for some @xmath233 . thus , the problem is reduced to finding rational solutions of @xmath234 . this can be solved by . a further optimization is present in gosper s algorithm : if @xmath235 in reduced form , the polynomial @xmath236 of satisfies , so that it divides the numerator of @xmath105 . ( this can be generalized to detect factors of numerators in arbitrary lre s ) . this does not affect the expression of the complexity result , which is as follows . let @xmath230 be a hypergeometric term such that @xmath237 , with @xmath238 . let @xmath0 be a bound on the degree of @xmath92 in and on the largest positive integer root of the indicial polynomial of @xmath239 at infinity . then the existence of an indefinite hypergeometric sum of @xmath230 can be determined by a probabilistic algorithm using @xmath240 bit operations , a compact representation of it can be computed deterministically using @xmath212 bit operations . note that in the special case of rational summation ( i.e. , @xmath241 ) , it is actually possible to decide the existence of a rational sum in only _ polynomial _ complexity , see @xcite . a bivariate hypergeometric term @xmath242 is such that both @xmath243 and @xmath244 belong to @xmath245 . given such a term , zeilberger s algorithm @xcite computes a lre satisfied by @xmath246 . the idea is to synthesize a _ telescoping recurrence _ , i.e. , a rational function @xmath247 and a linear operator @xmath248 such that @xmath249 indeed , summing over @xmath115 and granted boundary conditions known as `` natural boundaries '' , we obtain @xmath250 . if @xmath3 was known , then gosper s algorithm would find the left - hand side . this is the basis of zeilberger s algorithm ( see figure ) . termination is guaranteed only if such a lre exists . this occurs in the so - called `` proper - hypergeometric '' case @xcite and a general criterion has been given by abramov @xcite . note that knowing @xmath27 permits to check the output operator @xmath3 by simple rational function manipulations , which is why the rational function @xmath27 is called `` certificate '' in @xcite . zeilberger s algorithm is based on a refinement of gosper s algorithm for steps 2 and 3 . it reduces the computation in step 2 to solving a system that is linear simultaneously in the @xmath251 s and in another set of @xmath11 variables , where @xmath0 is potentially exponential in the bit size of @xmath252 , see e.g. @xcite . an equivalent linear system in a small number of variables can be computed by algorithm ( see figure ) . the important point is linearity : not all solutions of @xmath253 are linear in the @xmath251 s , but this property is ensured when the initial conditions satisfy @xmath254 . indeed , in step 2 , by lemma [ abramov - lemma ] , @xmath92 does not depend on the @xmath251 s . then , by induction on @xmath115 , starting from @xmath255 , the factor @xmath137 of the leading coefficient in @xmath214 cancels out and thus the solution @xmath208 is linear in the @xmath251 . this property is then preserved by the linearity of . the final system @xmath256 has solutions if and only if @xmath252 has rational solutions . the description of is geared towards the use of compact representations and matrix factorials in intermediate steps . this is straightforward for steps 15 . in step 6 , we can not make direct use of the factorial of the matrix associated to @xmath257 : this matrix involves the @xmath251 s rationally and its factorial has too large a size for our target complexity . instead , we exploit the linearity in the @xmath251 s by constructing the vector @xmath258 using matrix factorials for @xmath219 a vector of 0 s with a 1 in @xmath259th position for @xmath260 and setting the initial condition to 0 , which gives the coefficients @xmath261 . from there we derive our compact version of zeilberger s algorithm given in . in step 2 , the whole construction can be performed by matrix factorials with integer entries , within the complexities of thm . [ thm : reccomplexity ] . if a rational solution @xmath262 exists , then the system @xmath256 has the corresponding @xmath263 for solutions . thus if @xmath256 does not have a nonzero solution , @xmath252 does not have a rational one . this gives a fast probabilistic test . then , in step 5 , the algorithm is used again with matrices that are polynomial in the variable @xmath25 . in that case , the system @xmath256 can be computed by binary splitting with @xmath264 _ arithmetic _ operations . the final system has coefficients of degree @xmath265 with coefficients of bit size @xmath202 each and this is also the size of the @xmath251 s to be found . at the same time , we find @xmath266 , which gives us a compact representation of the certificate . an optimization is obtained by using the values of the @xmath267 s to compute the value @xmath268 of the degree of the corresponding sequence . with high probability this is the actual degree in @xmath115 of the numerator of @xmath247 , which can be much smaller than @xmath0 , thus saving a lot of computation in step 5 . the following theorem summarizes this section . [ thm : zeilberger]let @xmath242 be hypergeometric over @xmath48 . let @xmath0 be the maximal number of variables in the linear system solved in the classical version of zeilberger s algorithm . then it is possible to detect probabilistically that this system does not have any nonzero solution in @xmath227 bit operations . if it does have a solution , it is possible to compute the corresponding @xmath251 s of degree @xmath265 and total bit size @xmath12 , as well as a compact representation of the certificate , in @xmath269 bit operations . for the sake of comparison , a crude analysis by unrolling the triangular system of dimension @xmath270 and taking into account coefficient growth leads to a @xmath271 bit complexity estimate for the classical algorithm , which can be reduced to @xmath272 by using the binomial basis . we consider two families of linear recurrences : @xmath273 the first one ( @xmath274 ) does not have any rational solution , while the second one ( @xmath275 ) has @xmath276 as a solution . in both cases , when @xmath0 is a large integer , a large dispersion has to be considered . in table [ table - abramov ] , we give a comparison of the timings obtained by our maple prototype ( denoted compact ) and that of the command ` ratpolysols ` of maple s package ` lretools ` ( denoted classical ) . this table illustrates the `` nonexponential '' character of the compact versions of the algorithms . in the first case , both output are identical ( no solution ) . in the second case , however , we return a compact representation of the output . for instance , with @xmath277 we get ( in 0.04s ) the denominator @xmath278 ( in expanded form ) and for the numerator the recurrence @xmath279 satisfied by its coefficients in the binomial basis , together with initial conditions @xmath280 , @xmath281 , where the coefficients @xmath282 are 200 bit long integers . .timings ( in sec . ) for classical and compact versions of abramov s algorithm[table - abramov ] [ cols="^,^,^,^,^ " , ] in table [ timings ] , we give a comparison of the timings obtained by our prototype implementation in maple ( denoted `` compact '' ) and those obtained by maple s ` zeilberger ` command in the package ` sumtools :- hypergeometric ` ( denoted `` classical '' ) . the indication `` @xmath283 gb '' means that the computation had to be stopped after two gigabytes of memory had been exhausted . the first part of the table ( classical ) suggests that the implementation does not behave well for large @xmath0 : the observed behaviour is exponential instead of polynomial . even then , it is still much better than our implementation . indeed , we have implemented only the case with rational values of @xmath25 and for small @xmath0 it often takes longer for our implementation to compute the result with this value than for the classical method to find the result with a formal @xmath25 . however , things change as @xmath0 gets larger : the predicted behaviour is well observed . when @xmath0 is multiplied by 2 , the time is multiplied by slightly more than 2 . had we implemented the baby - step / giant - step version of binary splitting , the timings in the columns for random @xmath25 would have been much better , since the time should be multiplied by slightly more than @xmath284 from one line to the next . our experiments with symbolic @xmath25 show that so far , our complexity result is more of a theoretical nature : although the degrees of the coefficients of the equations grow like @xmath265 , the constant in front of the @xmath285 term is about 18 in this example , and a massive cancellation takes place in the final linear solving . the result has degrees that also grow like @xmath265 , but with a much smaller constant , so that a direct resolution in @xmath271 is much faster in this range than our @xmath286 . this work was supported in part by the french national agency for research ( anr gecko ) . comments of the referees on the first version of this article have been very useful . a. bostan , p. gaudry , and . linear recurrences with polynomial coefficients and computation of the cartier - manin operator on hyperelliptic curves . in _ international conference on finite fields and applications ( toulouse , 2003 ) _ , volume 2948 of _ lecture notes in computer science _ , pages 4058 . springer verlag , 2004 .	we consider two kinds of problems : the computation of polynomial and rational solutions of linear recurrences with coefficients that are polynomials with integer coefficients ; indefinite and definite summation of sequences that are hypergeometric over the rational numbers . the algorithms for these tasks all involve as an intermediate quantity an integer @xmath0 ( dispersion or root of an indicial polynomial ) that is potentially exponential in the bit size of their input . previous algorithms have a bit complexity that is at least quadratic in @xmath0 . we revisit them and propose variants that exploit the structure of solutions and avoid expanding polynomials of degree @xmath0 . we give two algorithms : a probabilistic one that detects the existence or absence of nonzero polynomial and rational solutions in @xmath1 bit operations ; a deterministic one that computes a compact representation of the solution in @xmath2 bit operations . similar speed - ups are obtained in indefinite and definite hypergeometric summation . we describe the results of an implementation . * categories and subject descriptors : * i.1.2 [ * symbolic and algebraic manipulation * ] : algorithms * general terms : * algorithms , experimentation , theory * keywords : * computer algebra , polynomial and rational solutions , linear recurrences , summation , creative telescoping , complexity
the riemann hypothesis states that the non - trivial zeros of the riemann function , @xmath0 , have the form @xmath1 , where the @xmath2 are all real . a fascinating approach to treating the riemann problem is to consider the @xmath2 to be eigenvalues of a self - adjoint operator , since if such an operator could be identified the @xmath2 would necessarily be real . this idea , known as the plya and hilbert conjecture , is supported by numerous evidence ( see @xcite for reviews ) , notably that the zeros appear to closely follow the gue statistics @xcite of random matrix theory and quantum chaos . this unexpected connection between number theory and physics has inspired many suggestions for finding physical implementations of the riemann function , such as evaluating the fourier transform of a suitable wave using opto - mechanical means @xcite or far - field diffraction @xcite , or by measuring entanglement in quantum systems @xcite . in this work we propose a very different approach , in which a time - periodic driving potential is used to modify the dynamics of a quantum system . as the system is periodically - driven its dynamics is not described by energy eigenvalues , but by a generalization of these quantities termed `` quasienergies '' . our central result is the construction of a driving field for which the corresponding quasienergy spectrum is given by the riemann @xmath3 function @xcite , or by a smoothed version of this function , @xmath4 , introduced by plya . this last has the appealing feature that the appropriate driving field can be expressed in a simple closed form . in both cases the zeros of the @xmath3 function correspond to degeneracies , or crossings , of the quasienergies . these are of particular physical significance because they correspond to the phenomenon known as `` coherent destruction of tunneling '' ( cdt ) in which the dynamics of the system is frozen @xcite . as an example we show how the zeros can be seen directly in cold atom experiments by measuring the expansion rate of a condensate held in a driven optical lattice . this technique thus represents a new and powerful way of finding a physical realization of the riemann function . we begin by considering a standard two - level system , driven by a time - periodic function @xmath5 @xmath6 where @xmath7 is the tunneling between the two levels . as @xmath8 is time - periodic , the natural framework to treat the problem is given by floquet theory @xcite . in this approach one seeks the eigensystem of the floquet operator @xmath9 where we have set @xmath10 . henceforth we shall also measure all energies ( and frequencies ) in units of @xmath7 . the eigenstates of @xmath11 are @xmath12-periodic functions called floquet states , and their associated eigenvalues , which play an analogous role to energy eigenvalues for the case of a static hamiltonian , are called quasienergies . the floquet states provide a complete basis , and expanding the wavefunction in these states provides similar advantages to the normal procedure of expressing a state in energy eigenstates in the undriven case . in general it is difficult to obtain analytical expressions for the floquet states . in the strong - driving limit , however , when the frequency @xmath13 is the dominant energy scale , it is possible to make an expansion by first solving just for the time - dependent component of @xmath14 , and applying the static part as perturbation @xcite . in this way one obtains a perturbative series in orders of @xmath7 . truncating at first - order gives the simple result @xmath15 , where the effective tunneling @xmath16 is given by @xmath17 and @xmath18 . the quasienergies are given simply by the eigenvalues of @xmath19 , namely @xmath20 . for the smoothed function @xmath21 ( eq.[final ] ) . for small values of @xmath22 the function is quite featureless , but progressively increases in amplitude and develops more oscillations as @xmath22 increases . the corresponding driving function for @xmath23 behaves similarly ( see fig.[surface_plot].),scaledwidth=45.0% ] from eq.[integral_eff ] it is thus straightforward to calculate the behavior of the quasienergies ( or equivalently , of the effective tunneling ) for a given driving potential @xmath8 . for the case @xmath24 , for example , this yields the well - known bessel function renormalization @xcite of tunneling @xmath25 . at zeros of the bessel function , @xmath26 , the effective tunneling vanishes , producing cdt . this effect has been measured experimentally @xcite in the dynamics of driven ultracold atoms . we need , however , to solve the _ inverse problem _ ; to find an @xmath8 that produces a given behavior of the quasienergies . we shall first explain our technique using plya s function , @xmath21 , as this gives a convenient closed - form for the solution , and then go on to consider the more complicated case of the true riemann @xmath3 function . plya s function is given by @xcite @xmath27 where @xmath28 is the modified @xmath29-bessel function , @xmath30 and @xmath31 . this is a smoothed version of the riemann @xmath3 function @xmath32 whose zeros coincide with the non - trivial zeros of @xmath0 . although as titchmarsh pointed out @xcite plya s zeta function can not truly be regarded as an approximation to @xmath23 in the most obvious sense , they do share many properties . most importantly @xmath21 has the same average distribution of zeros @xcite , following the smooth term of the riemann - mangoldt formula , and so nonetheless represents an interesting application of our method . plya further proved that the zeros of @xmath21 are real for any value of the constant @xmath33 . the spectrum of the @xmath34-type @xcite and dirac hamiltonian @xcite is given , for example , by the zeros of ( [ zeta ] ) with @xmath35 . the modified bessel function can be conveniently expressed as the integral identity @xmath36 and thus @xmath37 \nonumber \\ & = & 8 \pi^2 \int_0^\infty dt \cosh \left ( a t \right ) e^{-2 \pi \cosh t } \cos \left ( e t / 2 \right ) . \label{bessel_useful}\end{aligned}\ ] ] we are aiming to obtain the result @xmath38 \propto { \xi^{\ast}}(e)$ ] . combining eqs.[integral_eff ] and [ bessel_useful ] reveals that this requires @xmath39 to obey the relation @xmath40 where @xmath41 is an arbitrary constant . as the integrand on the right - hand side decays rapidly with @xmath42 , we can replace the upper limit of integration with @xmath12 , as long as we take @xmath12 to be sufficiently large . in this case we can then write @xcite @xmath43 \ . \label{define_f}\ ] ] the value of @xmath41 is now fixed by noting that we require @xmath44 , and thus @xmath45 . the driving field is then given by @xmath46 , yielding the final result @xmath47 where @xmath48 . in fig.[drive_function ] we show the form of @xmath8 as @xmath22 is varied . little structure is visible for @xmath49 . the reason is the rapid decrease of @xmath50 with @xmath42 , arising from the exponential term @xmath51 . indeed @xmath52 and @xmath53 . accordingly we now set the period of the driving , and thus the cut - off in the integration in eq.[requirement ] , to be @xmath54 . we shall use this value throughout the rest of the paper . in fig.[driving]a we show the full periodic driving field , obtained by periodically repeating cycles of @xmath8 . although this form of @xmath8 indeed satisfies ( [ requirement ] ) , the driving can be made more effective by imposing a further set of conditions on it : _ i ) _ to avoid heating in the cold atom model ( see section [ bec ] ) , the average of @xmath8 over one period should vanish , _ ii ) _ discontinuities should be avoided , and _ iii ) _ for the quasienergy crossings to be well - defined , the floquet states must be from different parity classes . if condition _ ( iii ) _ is not fulfilled , the von neumann - wigner theorem implies that the quasienergies can not cross as @xmath22 is varied , and will instead form a broad avoided crossing . as the system is periodically - driven , the appropriate generalized parity operator involves both inversion and time - translation @xmath55 . these three conditions can be satisfied by joining four copies of the fundamental waveform ( [ final ] ) as shown in fig.[driving]b , to create what amounts to two pulses of opposite sign , with a total period of @xmath56 . , is extended to be a periodic function by periodically repeating its behavior over the interval @xmath57 . this produces , however , a function with rather poor performance . ( b ) construction of a more efficient driving potential . four copies of @xmath8 are joined together , following some reflection transformations , to create a single continuous function , resembling two localized pulses of opposite sign . the vertical dashed lines indicate where the segments are joined . this function is continuous , has a time - average of zero , and @xmath58 where @xmath56 is the total period of the signal , satisfying the parity requirement given in the text.,scaledwidth=45.0% ] we now turn to the more important case of the riemann function itself . this can be written @xmath59 where @xmath60 , as defined in @xcite . we now follow the same procedure as before , seeking a driving potential such that @xmath38 \propto \xi(e)$ ] . this requires finding the solution of the equation @xmath61 where @xmath41 is a constant . we show the behavior of @xmath62 in fig.[phi ] . since @xmath62 decreases rapidly with @xmath42 , being fitted reasonably well by a simple gaussian function @xcite , we can substitute the upper limit of integration on the right - hand side of eq.[require_true ] to be @xmath12 , as long as @xmath12 is sufficiently large for the integral to converge . as before , we take @xmath63 . we can then write @xmath64 \ . \label{define_f_true}\ ] ] imposing the boundary condition @xmath65 requires @xmath41 to take the value @xmath66 . as @xmath67 is the global maximum of the function , this guarantees that @xmath68 for all values of @xmath42 , and thus @xmath39 is real and well - defined . and its derivative . solid black line : @xmath62 has a global maximum at @xmath69 , for which @xmath70 , and decays rapidly as @xmath42 increases . dashed red line : the derivative , @xmath71 , evaluated from eq.[def_deriv ] . the series expansions for both @xmath62 and @xmath72 were truncated at 30 terms.,scaledwidth=45.0% ] having obtained @xmath39 it is now straightforward to calculate @xmath46 . this yields the final result @xmath73 } { \sqrt { \phi^2(0 ) - \left [ \phi(t ) \cos \left ( e t / 2 \right ) \right]^2 } } \ . \label{def_f}\ ] ] the derivative of @xmath62 can be evaluated by differentiating its series expansion term by term , to give @xmath74 which we also show in fig.[phi ] . we show the full behavior of @xmath8 as a function of time and the parameter @xmath22 in fig.[surface_plot ] . despite the complicated form of ( [ def_f ] ) , this plot shows a striking similarity to fig.[drive_function ] , in which the @xmath8 giving rise to @xmath21 is shown . this should not be unexpected as @xmath21 is simply a smoothed version of the riemann function , and consequently the main features of the driving functions must be the same . ( eq.[def_f ] ) , which produces a quasienergy spectrum proportional to the riemann @xmath3 function . the behaviour resembles strongly the driving function for the smoothed riemann function , shown in fig.[drive_function].,scaledwidth=45.0% ] just as for the case of the smoothed riemann function , a more effective driving potential is obtained by joining four copies of @xmath8 , to produce a continuous function with a definite parity and zero time average . in fig.[compare ] we show the form of the driving function we obtain in this way , and compare it with the driving function that produces @xmath21 . examining the two curves in detail , reveals that although the functions share the same general form ( as seen in the similarity of the two surface plots fig.[drive_function ] and fig.[surface_plot ] ) , there are nonetheless significant small differences . these minor differences lead to the different location of zeros for @xmath23 and @xmath21 , demonstrating the well - known analytical sensitivity of the zeta function . , that produce @xmath21 and @xmath23 look superficially very similar , they differ in details . we show here the driving functions for @xmath75 for the two riemann functions we consider.,scaledwidth=45.0% ] the expression for the effective tunneling , @xmath16 , that we use in this work comes from a first - order perturbation theory calculation of the driven system . in this perturbation theory , the `` small parameter '' is @xmath76 , and so this result is only valid in the high - frequency limit @xmath77 . this would require the driving period , @xmath78 , to be small ; however this contradicts the requirement that @xmath12 be as large as possible so that the integrals ( [ requirement ] ) and ( [ require_true ] ) are well - converged . to be able to satisfy both these requirement , we therefore scale the driving as @xmath79 , where @xmath80 , while keeping @xmath81 constant . note that this is not a trivial rescaling of time , as its effect is to make the pulses shown in fig.[driving]b narrower and taller , while keeping their spacing constant at @xmath12 . we show the effect of increasing @xmath82 on the pulse - shape in fig.[pulse]a . it can be clearly seen that the pulses become progressively more localized and of higher amplitude . as the pulses become shorter and more intense , the quasienergy spectrum of the system approaches that of the high - frequency limit , as we show in fig.[pulse]b . in particular we can see that the locations of the quasienergy crossings converge toward the zeros of the riemann function . this evolution of the quasienergy spectrum as the driving frequency increases is a general feature of periodically - driven systems , including for example , sinusoidal , squarewave , and triangular driving @xcite . is increased . the pulses become progressively narrower and taller , while their spacing remains the same . ( b ) as @xmath82 increases , the quasienergy spectrum asymptotically approaches that of the infinite frequency limit , and the quasienergy crossings move towards the zeros of the riemann function . here we show the quasienergies for a system driven to reproduce the smoothed riemann function , @xmath21 . note how the first crossing asymptotically approaches the first zero of @xmath21 at @xmath83 as @xmath82 increases.,title="fig:",scaledwidth=45.0% ] is increased . the pulses become progressively narrower and taller , while their spacing remains the same . ( b ) as @xmath82 increases , the quasienergy spectrum asymptotically approaches that of the infinite frequency limit , and the quasienergy crossings move towards the zeros of the riemann function . here we show the quasienergies for a system driven to reproduce the smoothed riemann function , @xmath21 . note how the first crossing asymptotically approaches the first zero of @xmath21 at @xmath83 as @xmath82 increases.,title="fig:",scaledwidth=45.0% ] in principle @xmath82 should be made as large as possible , to ensure that the system is well within the high - frequency regime . we can see from fig.[pulse]b that good precision is obtained for @xmath84 , and the results we present below use @xmath85 . increasing @xmath82 beyond this value was found to introduce instabilities in the numerical integration of the system s time - evolution . in fig.[quasi]a we compare the quasienergies , obtained by the direct integration of the equation of motion ( [ 2level ] ) under the driving potential given in eq.[final ] , with the exact behavior of @xmath21 . the agreement is seen to be excellent . similarly in fig.[quasi]b we compare the riemann @xmath3 function ( [ true_zeta ] ) with the quasienergies resulting from driving the system with the potential given by eq.[def_f ] , and again see essentially perfect agreement . as the @xmath3 functions decay roughly exponentially with @xmath22 @xcite , we show in fig.[quasi]c the same data plotted logarithmically . the cusps visible in this plot correspond to zeros of the @xmath3 functions , and thus to crossings of the quasienergies at which @xmath16 vanishes . we see that the quasienergies accurately reproduce the behavior of the @xmath3 functions over at least six orders of magnitude , although eventually precision effects do lead to deviations at large values of @xmath22 . given by eq.[final ] , the dashed ( red ) line is the smoothed riemann function @xmath21 . the quasienergies are normalized with respect to the quasienergies at @xmath86 . the agreement between the two is excellent . ( b ) as above , but for the true riemann @xmath3 function , with the driving potential given by eq.[def_f ] . again , the results agree perfectly . ( c ) comparison on a logarithmic scale . the first three zeros of @xmath21 , corresponding to degeneracies of the quasienergies , are clearly visible as the downward - pointing cusps ( black circles , red dashed line ) , as are the first two zeros of @xmath23 ( blue diamonds , blue dashed line ) at @xmath87 and @xmath88 . the quasienergies reproduce the behavior of the @xmath3 functions to a high degree of accuracy.,scaledwidth=45.0% ] a way of directly measuring @xmath16 in experiment is to observe the expansion of a gas of cold atoms @xcite . if the atoms are held in an optical lattice potential they can be described well by a tight - binding model @xmath89 where @xmath90 labels the lattice site , @xmath91 are nearest - neighbors , and @xmath92 is a trap potential . by `` shaking '' the optical lattice @xcite it is possible to introduce a time - periodic driving potential @xmath93 which generalizes eq.[2level ] from a two - level model describing two sites to the case of @xmath94 lattice sites . for a parabolic trap potential , the initial state of the system will be gaussian . if the trap potential is then released , this gaussian wavepacket will undergo free expansion at a rate governed by @xmath95 @xcite . in fig.[expand]a we show the spread of the wavepacket with time , @xmath96 , under the periodic driving corresponding to @xmath21 . for @xmath86 the wavepacket expands rapidly , and soon enters the ballistic regime in which @xmath97 . for @xmath75 the expansion is slower since @xmath16 is smaller , and for @xmath98 the wavepacket barely expands at all , indicating that this value of @xmath22 is close to a zero of @xmath21 . in principle one would expect only partial destruction of tunneling , even when @xmath22 is tuned exactly to a zero of @xmath4 , due to the presence of longer - ranged hopping elements . such an effect is seen in sinusoidally - driven systems @xcite where the band strongly narrows but does not collapse totally . for the case of the drivings we consider , the longer - ranged hoppings are suppressed considerably more than for the sinusoidal case , however , meaning their effect is essentially negligible . sites under the periodic driving potential @xmath8 ( [ final ] ) , such that @xmath99 . for @xmath86 the wavepacket spreads quickly , but for @xmath75 the wavepacket spreads less rapidly , indicating that @xmath16 is smaller . for @xmath98 the wavepacket hardly expands at all , indicating that cdt is occurring . ( b ) the solid black curve shows @xmath100 , and the black circles the values of @xmath95 measured from the expansion curves . the agreement between the two is excellent . similarly the dotted red line shows @xmath101 , and the red squares the measured values of @xmath102 for a system driven by the potential ( [ def_f ] ) . again the agreement is perfect.,scaledwidth=45.0% ] these estimations can be made quantitative . in fig.[expand]b we show the values of @xmath102 obtained by measuring the spread of the wavepacket after an expansion time of @xmath103 and fitting it to the result @xmath104 , where @xmath105 is the initial width of the condensate . both of the driving potentials show excellent agreement with the exact @xmath3 functions , demonstrating that measuring the expansion of a trapped cold atom system is a viable method to experimentally determine the zeros of the riemann functions . we propose a novel approach for a physical realization of the riemann zeros . the main idea is to use a time - dependent driving potential to modify the dynamics of a quantum system , so that its quasienergy spectrum mimics the desired riemann function . we provide a systematic scheme for calculating the appropriate potential , and suggest a physical realization using driven cold atoms . although the observation of high riemann zeros is hindered by the rapid decay of the @xmath3 function , this could be compensated in experiment by increasing the tunneling strength ( which depends exponentially on the optical lattice depth ) , or by using a form of the zeta function which decays more slowly , such as @xmath106 @xcite . an intriguing possibility for future study would be to apply this technique to more general trigonometric integrals of the function @xmath62 , which could thereby lead to the measurement of a new bound on the de bruijn - newman constant , the value of which is related to the riemann hypothesis @xcite . the authors thank michael berry for stimulating discussions . cec was supported by the spanish mineco through grant nos . fis2010 - 21372 and fis2013 - 41716-p , and gs by grant no . fis2012 - 33642 , quitemad , and the severo ochoa programme under grant sev-2012 - 0249 .	the riemann hypothesis , which states that the non - trivial zeros of the riemann zeta function all lie on a certain line in the complex plane , is one of the most important unresolved problems in mathematics . we propose here a new approach to finding a physical system to study the riemann zeros , which in contrast to previous examples , is based on applying a time - periodic driving field . this driving allows us to tune the quasienergies of the system ( the analogue of the eigenenergies for static systems ) , so that they are directly governed by the zeta function . we further show by numerical simulations that this allows the riemann zeros to be measured in currently accessible cold atom experiments .
the goal of low - energy nuclear physics is to understand nuclei and how they react . this fascinating science problem is relevant to other fields and to a gamut of societal applications . new vistas have been opened by experimental advances in the production of rare isotopes @xcite and new theoretical approaches @xcite backed by unprecedented computing power @xcite . the rapid experimental developments have resulted in a wealth of unique data from previously unexplored regions of the nuclear landscape . this situation poses a serious challenge to models of nuclear structure and calls for their improved reliability and better - controlled extrapolability . theorists seek to formulate a coherent framework for nuclear structure and reactions based on a well - founded microscopic theory that would deliver maximum predictive power with well - quantified uncertainties . to this end , the steady increase in computing power , currently crossing the petaflop barrier , has been beneficial . a paradigm for the new mode of nuclear theory is the scidac universal nuclear energy density functional ( unedf ) project @xcite , an example of the close alignment of the physics research with the necessary applied mathematics and computer science research . this study is the fruit of such a partnership , under unedf , in which physicists collaborate with mathematicians and computer scientists on a specific science challenge . our long - term goal in unedf is to develop a spectroscopic - quality theoretical framework rooted in the nuclear density functional theory ( dft ) @xcite . in the first phase of the project , we have developed efficient dft solvers for the self - consistent hartree - fock - bogoliubov ( hfb ) problem . various improvements that we have implemented to carry out large - scale dft calculations have been recently presented in @xcite . these improvements enable comprehensive mass - table calculations , including all even - even nuclei and many different configurations in odd - even and odd - odd nuclei , in less than a day @xcite . the second phase of the project concerns the development and optimization of the nuclear energy density functional ( edf ) . since standard functionals are clearly too restrictive when one is aiming at a quantitative description @xcite , the form of edf needs to be improved . novel functionals can be constructed from two- and three - nucleon interactions by using effective field theory and the density matrix expansion technique @xcite and by using constraints from ab initio calculations for very light nuclei and nuclear matter . they can also be obtained by enriching density dependence and adding higher gradient terms in a systematic way @xcite . having determined the form of the edf , one must still optimize the coupling constants of the underlying energy density ( ed ) . indeed , all energy functionals , irrespective of their theoretical foundations , rely on parameters that must be directly fitted to experimental data . it has been realized recently that high - performance computing can positively impact the optimization strategy . historically , most nuclear ed parameterizations , such as skyrme or gogny , were obtained by a direct fit to selected experimental data from finite nuclei and various nuclear matter properties ( nmps ) . observables commonly included in the fit are binding energies , proton radii , surface thickness , and/or single - particle ( s.p . ) energies of doubly closed - shell nuclei as well as nmps ( pseudo - observables ) such as energy per particle of infinite and semi - infinite nuclear matter , saturation density , or incompressibility . this is the case , for example , for the sly4 parameterization of the skyrme functional of @xcite , which we take ( somewhat arbitrarily ) as a reference point in our study . the d1 and d1s parameterizations of the gogny interaction have also been obtained in such a framework @xcite . we refer to @xcite for a more thorough discussion of various fitting strategies and protocols . in fact , very few examples of eds are fitted to other types of data . for skyrme edfs only , we mention the early attempt of the skm * parameterization @xcite , which was adjusted semi - classically to account for the fission barrier of @xmath0pu . the brussels - montreal set of edfs has been optimized to data on deformed nuclei , although the actual fit is always performed with a spherical code by using a multistep procedure . for example , in the early versions msk1-msk6 , the deformation energy of the ground - state configuration was used to renormalize nuclear masses so that the optimization could proceed in spherical symmetry @xcite . similarly , while in the later version hfb14 , data on fission barriers were used , the core part of the fitting procedure was carried out in spherical geometry @xcite . for sly4 itself , several parameters were fixed at values empirically expected to yield a correct description of giant resonance energy centroids in random - phase approximation ( rpa ) calculations , although no such calculation was included in the fit nor any quantitative check performed a posteriori . the choice to restrict the data set of observables to those pertaining to nuclear matter and spherical nuclei has almost always been dictated by practical considerations : the cost of performing huge numbers of deformed hfb calculations was deemed too high . it was also rightly argued that the driving terms of the edf could be pinned down by considering spherical nuclei only . with the need for more precision , however , the limitation to spherical nuclei and nmp is clearly not sufficient . the advent of supercomputers makes it possible to free ourselves from this restriction . specifically , the availability of supercomputers has two consequences . first , one can now include in the set of fit observables data corresponding to deformed nuclei , odd - mass systems , excited states , and so forth . more comprehensive data sets should better constrain the various channels of the energy functionals , for example , its deformation or spin - polarization properties . it might soon become possible to directly optimize symmetry - restored edfs @xcite , either in a single - reference @xcite or a multireference @xcite framework . in addition , in our quest for improved edfs , a key step is to understand various constraints imposed by experimental data on ed parameters and the resulting uncertainty margins . early attempts to use statistical methods of linear - regression and error analysis @xcite have been revived recently and applied to determine the correlations between ed parameters , parameter uncertainties , and the errors of calculated observables @xcite . this approach is essential for providing predictive capability and extrapolability and for estimating the theoretical uncertainties . the purpose of this work is to revisit the problem of skyrme ed optimization by ( i ) removing some of the previous limitations with the help of modern computational resources , and ( ii ) applying regression diagnostics methods on the resulting parameterization . to these ends , we perform functional optimization with a model - based method that is particularly adapted to costly function evaluations , such as when the objective function contains the result of hundreds of symmetry - unrestricted hfb calculations . in our model study , we focus on nuclear masses and radii , with a bias toward heavy nuclei . the final ed parameterization is subjected to a fully fledged correlation and sensitivity analysis . while we do not claim to have found an end - all parameterization of the skyrme edf , we believe that the set of techniques we have applied in this study can pave the way to a universal nuclear edf of spectroscopic quality . the paper is organized as follows . in sec . [ sec - framework ] we briefly present the dft framework used , in particular various parameterizations of the skyrme edf and their relations to nuclear matter properties . we also discuss the choice of experimental observables . section [ sec - optimization ] presents the specific model - based algorithm used in this work and contains all the technical information related to large - scale hfb calculations . results are discussed in sec . [ sec - results ] . section [ conclusions ] contains the conclusions of this work . this section recalls the features of the skyrme - dft theory that are relevant to the optimization problem . a detailed presentation of the theory itself can be found in , for example , @xcite and references therein . the main focus of the following discussion is on various parameterizations of the skyrme edf and the selection of experimental observables chosen to constrain ed parameters . in nuclear dft , the total energy of the nucleus is given by @xmath1 where @xmath2 is the local energy density that is supposed to be a real , scalar , time - even , and isoscalar function of local densities and their derivatives . the skyrme ed can be decomposed into the kinetic term , interaction ed @xmath3 , pairing ed , coulomb term , and additional corrections , such as the center - of - mass term . for the kinetic energy term , we set @xmath4=20.73553mev@xmath5 . the coulomb hartree term is calculated exactly , while the exchange term is computed by the slater approximation . the contribution from the center - of - mass correction has the same structure as the kinetic term and leads to a renormalization of the nucleon mass @xmath6 $ ] . all these prescriptions follow the sly4 parameterization . the interaction ed can be further decomposed into @xmath7 , with @xmath8 where the isospin index @xmath9 labels isoscalar ( @xmath9=0 ) and isovector ( @xmath9=1 ) densities . since in this work we limit the discussion to even - even nuclei , the terms involving spin , spin - kinetic , and current densities @xcite are absent . the coupling constants @xmath10 contain an additional dependence on the isoscalar density of the form @xmath11 the standard skyrme interaction ed therefore contains 13 independent parameters : @xmath12 when dealing with the skyrme interaction edf ( i.e. , the functional that originates from the skyrme interaction ) , the coupling constants ( [ c - parameters ] ) are uniquely related to the well - known @xmath13-parameterization of the skyrme interaction @xmath14 the equations connecting the @xmath15- and @xmath13-parameterization can be found , for example , in @xcite . in this study , nucleonic superconductivity is described by the pairing ed : @xmath16\breve\rho^2({\boldsymbol{r } } ) , \label{vpair}\ ] ] where @xmath17 is the local pairing density and @xmath18=0.16@xmath19 ( mixed - pairing prescription @xcite ) . the @xmath13 and @xmath15-representations are natural parameterizations of the skyrme edf , the former in terms of an effective , density - dependent two - body interaction and the latter as a general functional of the density . however , these representations do not provide a straightforward connection to physical observables ; hence , it is not immediately obvious what the search range for these parameters should be . it is therefore advantageous to relate them to fundamental properties of symmetric and asymmetric homogeneous nuclear matter , which have a clear physical interpretation and the range of which is known @xcite . the starting point in the discussion of nmps is the equation of state ( eos ) of the infinite homogeneous nuclear matter : @xmath20 . the coulomb energy is disregarded , all gradient terms vanish , and the kinetic energy density is replaced by its thomas - fermi expression . assuming an unpolarized system , one can also ignore terms involving time - odd spin densities and currents . the expansion of @xmath21 around the equilibrium density @xmath22 and @xmath23 can be written as @xmath24 where @xmath25 is the relative neutron excess , @xmath26 , @xmath27 , @xmath28 and @xmath29 in these equations , @xmath30 stands for the total energy per nucleon at equilibrium , @xmath31 represents the nucleonic pressure , @xmath32 is the nuclear matter incompressibility , @xmath33 is the symmetry energy coefficient , @xmath34 represents the density dependence of the symmetry energy , and @xmath35 is a correction to the incompressibility . in the regime of symmetric nuclear matter ( snm ) , @xmath36 and @xmath37=0 , which eliminates all isovector terms . the isoscalar kinetic energy density is @xmath38 the nuclear matter saturation curve @xmath39 is expected to have the following properties : @xmath40 the value of the incompressibility modulus is related to the centroid energies of giant isoscalar monopole resonances in isospin - symmetric nuclei @xcite and is expected to be @xcite @xmath41 with a strong preference for 230mev @xcite . another important nmp , entering the snm eos indirectly , is the isoscalar effective mass @xmath42 which quantifies the momentum - dependence of the mean field and drives the density of the s.p . an appropriate value for a fit to experimental s.p . energies is @xmath43 @xcite , while ab initio calculations performed at the brueckner - hartree - fock level in inm suggest a slightly lower value for the landau ( fermi - level ) effective mass extracted from the on - shell s.p . spectrum @xcite . mass fits also seem to favor a value close to unity , although significant freedom exists @xcite . the snm eos expressed in terms of the coupling constants of the skyrme edf is @xmath44 computing the quantities ( [ symnmexpbegin]-[symnmexpend ] ) using ( [ wc ] ) allows us to express the coupling constants @xmath45 , @xmath46 , @xmath47 and the power @xmath48 in terms of @xmath30 , @xmath49 , @xmath32 and @xmath50 . the resulting expressions are @xcite as follows : @xmath51 \tau_c \right . \nonumber \\ & & \left . + 3\left(1 + \gamma\right ) \tfrac{{e^\text{nm}}}{a}\right\ } , \\ c_{0d}^{\rho\rho } & = & \frac{1}{3 \gamma { \rho_\text{c}}^{1 + \gamma } } \left [ \tfrac{\hbar^2}{2 m } \left(3 - 2m_s^ { * -1}\right ) \tau_c -3\tfrac{{e^\text{nm}}}{a } \right],\\ c_0^{\rho\tau } & = & \frac{\hbar^2}{2 m } \left(m_s^ { * -1}-1\right)\frac{1}{{\rho_\text{c } } } , \\ \gamma & = & \frac{\tfrac{\hbar^2}{2 m } \left(4 m_s^ { * -1 } - 3\right ) \tau_c - { k^\text{nm}}- 9 \frac{{e^\text{nm}}}{a}}{\frac{\hbar^2}{2 m } \left(6 m_s^ { * -1}-9\right ) \tau_c + 9 \tfrac{{e^\text{nm}}}{a } } , \label{cfromsnm}\end{aligned}\ ] ] where @xmath52 . in asymmetric nuclear matter ( anm ) , neutron and proton densities are different , and isovector terms are nonzero . the local and kinetic energy densities are @xmath53.\end{aligned}\ ] ] the nuclear matter eos @xmath54 now depends on the relative neutron excess @xmath37 . the most important parameter characterizing the isospin dependence of the anm eos is the symmetry energy at saturation density , @xmath55 the value of @xmath56 varies from 28 to 36 mev among edfs extrapolated to nuclear matter @xcite . it is understood @xcite that nuclear masses constrain a combination of the symmetry- and surface - symmetry energy parameters in a given edf , and this fact explains the large spread of values . the variation of the density - dependent symmetry energy @xmath57 with @xmath58 is usually parameterized through @xmath59 the value of which appears correlated with the thickness of neutron skins in asymmetric nuclei ( see @xcite and references therein ) . an empirical determination of this parameter yields @xmath60mev @xcite . one now introduces @xmath61 which affects the incompressibility of the anm and thus the isoscalar monopole resonance energies in neutron - rich nuclei @xcite . for the sly4 edf , the values of the last two parameters were determined by the fit to the neutron matter eos . we let these quantities be constrained by our experimental data set . we will see whether these data leave enough freedom to apply additional constraints in the regression analysis . the momentum dependence of the mean field is also affected by isospin : neutron and proton effective masses are different in asymmetric matter @xcite , an effect quantified by the isovector effective mass @xmath62 the eos of homogeneous asymmetric nuclear matter can be written as @xmath63 \rho_0 . \label{wca}\end{gathered}\ ] ] just as for snm , we compute the quantities ( [ asymnmexpbegin]-[asymnmexpend ] ) from ( [ wca ] ) and obtain an expression for @xmath64 , @xmath65 , and @xmath66 @xcite : @xmath67 , \\ c_{1d}^{\rho\rho } & = & \left . \frac{1}{27 \gamma { \rho_\text{c}}^{\gamma+1 } } \right[-27 { a_\text{sym}^\text{nm}}+ 9 { l_\text{sym}^\text{nm}}\nonumber \\ & & + \left . 5 \left(\frac{\hbar^2}{2 m } - 2 { \rho_\text{c}}\left(c_0^\tau + 3 c_1^\tau\right)\right ) \tau_c \right ] . \label{cfromanm}\end{aligned}\ ] ] using relations ( [ cfromsnm])-([cfromanm ] ) , we express 7 of the original 13 parameters ( [ c - parameters ] ) of the skyrme edf as functions of nuclear matter properties . the remaining 6 are not known exactly and should therefore not be used as rigid constraints @xcite . however , the expected values of all these nmps are sufficient to provide well - defined intervals of variation during the optimization process . the 6 remaining coupling constants are the isoscalar and isovector @xmath68 , spin - orbit @xmath69 , and tensor @xmath70 terms . consequently , the skyrme edf depends on the following 13 parameters : @xmath71 to calibrate the edf , we selected a pool of fit observables that constitute the unedf experimental database @xcite . the purpose of the database is to provide a standard and comprehensive set of experimental data that can be used to systematically optimize edfs . since we wish to provide , together with the optimized set of parameters , a measure of its intrinsic quality via the error and sensitivity analysis , for every observable an error bar should also be defined . we organized our database into three major categories spherical , deformed , and symmetry - unrestricted which reflect the level of symmetry - breaking of the underlying edf and thereby the complexity of its numerical implementation . more details can be found in @xcite . the focus of this work is on a well - controlled optimization methodology , and the emphasis is on global nuclear properties such as masses and proton radii . our functional is therefore restricted to time - even densities , and only spherical or axially deformed nuclei are considered . the chosen observables embrace data for 72 nuclei , which are proven to allow a reasonable dft description . the selected experimental data set is presented in fig . [ fig - data ] . as can be seen , the emphasis is on the heavy nuclei . indeed , there are only 11 nuclei with @xmath7266 in our data set . below , we give a detailed description of the set of fit observables used in this work . 66 and 61 nuclei with @xmath73106 . ] .nuclear binding energies ( in mev ; the electronic energy correction has been subtracted ) @xcite for the 44 deformed nuclei selected in this work . the column marked `` # '' is the data point number . [ cols="^,^,^,^,^,^,^,^",options="header " , ] the covariance matrix depends on the scaling of the parameters ; hence , we will work with the standard correlation coefficient , @xmath74 which captures the ( positive or negative ) correlation between parameters @xmath75 and @xmath76 . tables [ tab : cor]-[tab : corb ] provide the approximate @xmath77 correlation matrix @xmath78 calculated when @xmath79 is chosen to be @xmath80 the size of the interval of interest of parameter @xmath75 , for the solutions unedfpre and unedfnb , respectively . overall , tables [ tab : cor]-[tab : corb ] show that most parameters are interdependent , although the number of significant correlations with @xmath81 is small . for unedfnb , where all parameters are free , we note that two pairs of nm parameters are well correlated : @xmath32 is 87% correlated with @xmath22 @xcite , while @xmath33 is 97% correlated with @xmath34 @xcite . the value of the ( inverse of the ) effective mass appears well correlated with both pairing strengths . we also notice strong correlation between the pairing strengths and the isoscalar spin - orbit coupling constants . both observations reflect the interplay between single - particle level density and pairing discussed in sec . [ subsec - pairing ] . we also notice that the proton pairing strength is significantly correlated with the neutron pairing strength . in the case of the unedfpre parameterization , @xmath32 and @xmath82 are removed from the sensitivity analysis . nevertheless , we note that the various correlations between parameters overall remain , even if they are attenuated compared to the no - bound case . ( color online ) sensitivity of the parameters of unedfpre to different data types entering @xmath83 . the edf parameters are labeled as in table [ table : ci95 ] . ] next , we illustrate how sensitive the parameters @xmath75 are to the different data types entering @xmath83 : masses , proton radii , and oes . here we focus only on the unedfpre parameterization , as it is more realistic . we define the @xmath84 jacobian matrix @xmath85 of the residuals as @xmath86 , that is , the matrix formed as the juxtaposition of the @xmath87 column - vectors @xmath88 ( @xmath87 being the number of data , @xmath89 the number of parameters ) . the @xmath84 sensitivity matrix @xmath90 is @xmath91^{-1}j({\mathbf{x}}).\ ] ] for each line in the sensitivity matrix ( each parameter ) , we can compute the partial sums over each of the three types of data . this computation gives us a measure of the change of the parameter under a global change of all the data of a given type . figure [ fig : sens_bar ] shows the relative change of parameter @xmath75 when such an _ average _ datum of an observable is changed . for example , for @xmath92= 1 ( masses ) , it shows the change in @xmath75 under a variation of all experimental masses . all of the bars in fig . [ fig : sens_bar ] have been renormalized to unity , and only relative strengths between mass , radii , and oes data are shown . a large percentage contribution from data type @xmath92 means that @xmath75 is very sensitive to changes in @xmath92 , and other data types have little impact on it at the convergence point . as expected , pairing strengths ( parameters 9 and 10 ) are primarily affected by oes data . it is worth noting the very similar sensitivity of the spin - orbit coupling constants ( parameters 11 and 12 ) on all 3 types of data . also , nuclear matter parameters appear to be significantly more dependent on the proton radius than other coupling constants . this is not surprising , considering the relation between the saturation density and the wigner seitz radius . the integrated information contained in fig . [ fig : sens_bar ] can not assess the impact of a particular data piece on model parameters ; hence , a more detailed analysis is needed . to this end , for each experimental observable @xmath93 ( masses , radii , oes ) , we compute the global change in @xmath94 as individual data @xmath93 change ( one - by - one ) by @xmath95 , namely , 200 kev for masses , 0.002 fm for proton radii , and 50 kev for oes . in this way we can , for instance , evaluate the possible importance of some new experimental observable on a given model @xcite . figure [ fig : sens ] shows the quantity @xmath96 with @xmath97 being the change in the value of the parameter @xmath98 under a change of the data @xmath93 , for all @xmath99=108 data points . this is nothing but the norm of the total change in units of the standard deviation @xmath100 , defined as before by @xmath101 . large changes in @xmath94 mean that the parameter values are highly sensitive to the particular value of @xmath93 . ( [ eq : change_normed ] ) for the unedfnb and unedfpre parameter sets when data @xmath93 change by @xmath95 one - by - one . the labeling of data points is consistent with tables [ table : edef]-[table : oes ] . ] in principle , the sensitivity analysis can be performed at any point @xmath102 of the @xmath89-dimensional parameter space : for a given scalar function @xmath103 of the type ( [ eq : chi2 ] ) , the sensitivity at point @xmath102 is only based on the local gradient . in particular , it is totally independent of the procedure that leads to the specific selection of @xmath102 . it only depends on the degrees of freedom , i.e. , free parameters , retained in @xmath103 . the question of degrees of freedom is highly relevant in the context of the unedfpre parameter set , where two parameters are actively constrained at the solution . these constraints have been directly implemented by restricting the domain where the @xmath104 function is evaluated , and not by modifying the function by adding a penalty . the net result of imposing the bound constraints is that only 10 parameters out of twelve are allowed to change near the end of the optimization process . one has , therefore , two options as far as the sensitivity analysis is concerned : ( i ) remove these two parameters from the set of active parameters and calculate the jacobian @xmath85 and the sensitivity matrix @xmath105 with only @xmath106=10 parameters ; ( ii ) keep all @xmath89=12 parameters in the calculation of the jacobian and doing a tangent plan approximation to obtain the relevant covariance and sensitivity matrices . in this way , the uncertainties of the other 10 parameters are affected by the local fluctuations of the surface induced by these two actively constrained parameters . the alternative ( i ) boils down to computing the gradient at point @xmath107 in a 10-parameter subspace of the original 12-parameter space . in this subspace , @xmath107 is the stable point of the @xmath104 function . the curve labeled unedfpre@xmath108 in fig . [ fig : sens ] corresponds to this approach . this sensitivity response is compared with that performed at the free minimum @xmath109 of @xmath104 in the full 12-parameter space of unedfnb . since both sets correspond to ( unconstrained ) minima in their respective spaces , the overall changes in @xmath94 are of the same order of magnitude and they are very small , @xmath110 . this indicates that the set of fit observables has been chosen very consistently . indeed the mass of deformed @xmath111fm is the single observable that yields the noticeable parameter variations around the unedfnb minimum while in the case of unedfpre@xmath108 no sensitivity to a single piece of data can be noticed . by contrast , within the alternative ( ii ) , @xmath107 is not an unconstrained minimizer in the full 12-parameter space . in such an approach , we find the overall sensitivity to be about 2 orders of magnitude larger than what is depicted in fig . [ fig : sens ] . this reflects the fact that @xmath112 and @xmath113 are far away from the unconstrained minimum in that space , at the same time being strongly correlated with other parameters . in this case , the sensitivities depend strongly on the actual value of @xmath107 and the way the domain is constrained . for this reason , the option ( ii ) is of no practical interest in the comparison with the no - bounds results . one of the major challenges for the low - energy nuclear theory is to construct the global nuclear energy density functional of spectroscopic quality , rooted in microscopic theory . an important element of this program is to optimize the parameters of the functional on a set of experimental observables and selected theoretical pseudo - data . this work shows how such an optimization can be done by using modern optimization algorithms and nonlinear regression analysis . the purpose of this study was to optimize the standard skyrme functional based on a set of experimental data ( masses , charge radii , and odd - even mass differences ) pertaining to 72 spherical and deformed nuclei amenable to a mean - field description . the new model - based optimization algorithm pounders was compared with other standard derivative - free optimization methods such as nelder - mead and was found to be significantly better in terms of speed , accuracy , and precision . the optimization was carried out at the fully self - consistent , deformed hartree - fock - bogoliubov level . here , we took advantage of the efficient dft solver hfbtho optimized in the first phase of the project . we have implemented various improvements that enable us to quickly compute global self - consistent mass tables . this capability is essential for optimization . as a result of the twelve - parameter optimization of skyrme edf , we arrived at two solutions . the first one corresponds to a minimum ( stable point ) in the considered parameter space . the corresponding functional unedfnb describes well the assumed set of fit observables , but its incompressibility parameter is too large , as this property has not been well constrained by our data set . the second optimization was carried out assuming hard bounds on the nuclear matter parameters . for the bound - constrained solution , the nuclear incompressibility and scalar effective mass appear at their respective bounds . the resulting parameter set unedfpre gives good agreement with experimental masses , radii , and deformations and seems to be free of finite - size instabilities . in particular , for two - neutron separation energies and masses of even - even heavy nuclei with @xmath114 , unedfpre yields the rms deviation of 0.45mev and 1.2mev , respectively , which is a satisfying result . we emphasize that the original skyrme edfs seem to be inherently limited in this respect , as demonstrated in @xcite , unless specific corrections are introduced . our result is therefore in line with the best expectations one could have for such edfs . nevertheless , the lack of specific constraints on the shell structure in our data set implies that single - particle levels of light nuclei are not well reproduced . for that reason , unedfpre may not yet be recommended for truly global applications across the chart of the nuclides . however , this functional is expected to work well for heavy nuclei and should be considered as a reference against which more advanced edfs will be benchmarked . we have also applied full - fledged regression diagnostics on unedfnb and unedfpre , focusing on statistical correlations between ed parameters and the sensitivity of parameters to variations in fit observables . to this end , we computed and analyzed the correlation and sensitivity matrices at the optimal parameter set . this kind of nonlinear regression analysis is expected to be helpful when designing next generation edfs . moreover , the statistical tools presented in this study can be used to pinpoint specific nuclear observables that are expected to strongly affect the developments of the nuclear universal density functional . we thank peter klpfel for his help with the experimental database used in this work . this work was supported by the office of nuclear physics , u.s . department of energy under contract nos . de - fc02 - 09er41583 ( unedf scidac collaboration ) , de - fg02 - 96er40963 and de - fg02 - 07er41529 ( university of tennessee ) , de - fg0587er40361 ( joint institute for heavy ion research ) , and de - ac0z-06ca11357 ( argonne national laboratory ) . computational resources were provided through an incite award computational nuclear structure " by the national center for computational sciences ( nccs ) and national institute for computational sciences ( nics ) at oak ridge national laboratory , and through an award by the laboratory computing resource center at argonne national laboratory . j. mor t. munson s. benson , l. curfman mcinnes and j. sarich . tao user manual ( revision 1.9 ) . technical report anm / mcs - tm-242 , mathematics and computer science division , argonne national laboratory , 2007 .	we carry out state - of - the - art optimization of a nuclear energy density of skyrme type in the framework of the hartree - fock - bogoliubov ( hfb ) theory . the particle - hole and particle - particle channels are optimized simultaneously , and the experimental data set includes both spherical and deformed nuclei . the new model - based , derivative - free optimization algorithm used in this work has been found to be significantly better than standard optimization methods in terms of reliability , speed , accuracy , and precision . the resulting parameter set unedfpre results in good agreement with experimental masses , radii , and deformations and seems to be free of finite - size instabilities . an estimate of the reliability of the obtained parameterization is given , based on standard statistical methods . we discuss new physics insights offered by the advanced covariance analysis . = # 1
let @xmath0 be a prime . in this paper , by a _ @xmath0-group _ we mean a finite group whose order is a power of @xmath0 . recall that every @xmath0-group is nilpotent , and in particular solvable . given a property @xmath2 of groups , a group @xmath3 is called _ residually @xmath2 _ if for every @xmath4 with @xmath5 there exists a morphism @xmath6 onto a group @xmath7 enjoying property @xmath2 such that @xmath8 . for example , a group @xmath3 being residually a @xmath0-group ( residually @xmath0 , for short ) simply means that the intersection of all @xmath0-power index normal subgroups of @xmath3 is trivial . being residually @xmath0 is a very strong property : e.g. , every finite subgroup of a residually @xmath0 group is a @xmath0-group , and every non - abelian subgroup of a residually @xmath0 group has a quotient isomorphic to @xmath9 . free groups and finitely generated torsion free nilpotent groups are well - known to be residually @xmath0 for all @xmath0 . in this paper we study @xmath1-manifold groups ( i.e. , groups which are fundamental groups of @xmath1-manifolds ) . throughout this paper we assume that the manifolds are connected and compact , but unless otherwise stated , we make no assumptions on orientability or on the type of boundary . compactness is not a serious restriction : every finitely generated group which is the fundamental group of a non - compact @xmath1-manifold also appears as the fundamental group of a compact @xmath1-manifold ( and hence is finitely presentable ) , cf . it is well - known that every finitely presentable group can be realized as the fundamental group of a compact , closed , connected , smooth manifold of dimension @xmath10 ( or higher ) . however , there are severe restrictions on which groups may occur as fundamental groups of @xmath1-dimensional manifolds ; for example , it is well - known that every @xmath1-manifold group is residually finite . ( this was first shown for haken manifolds @xcite , and since perelman s proof of the geometrization conjecture , the proof now extends to include all @xmath1-manifolds . ) residual finiteness of @xmath1-manifold groups leads to easy proofs for the decidability of the word problem in @xmath1-manifold groups , and for the decidability of knot triviality . in general , however , @xmath1-manifold groups tend to be far from being residually @xmath0 or even just being residually finite solvable . indeed , let @xmath11 be a knot and @xmath12 , where @xmath13 is a tubular neighborhood of @xmath14 in @xmath15 . it follows from dehn s lemma that @xmath16 if and only if @xmath14 is the trivial knot . but it is a consequence of stallings theorem ( * ? ? ? * theorem 3.4 ) that any map from @xmath3 to a nilpotent group necessarily factors through the abelianization @xmath17\cong\z$ ] . furthermore , if @xmath14 is a knot with trivial alexander polynomial , then in fact any map from @xmath3 to a solvable group necessarily factors through @xmath18 . the purpose of this paper is to show that virtually the picture is very different . one says that a group @xmath3 is _ virtually @xmath2 _ if there exists a finite index subgroup of @xmath3 which is @xmath2 . ( subgroups of residually @xmath0 groups are again residually @xmath0 , so virtually residually @xmath0 is equivalent to residually @xmath0-by - finite . ) we call a finitely generated group _ linear _ if it admits an embedding into @xmath19 for some @xmath20 and some field @xmath14 of characteristic zero ( which may always chosen to be @xmath21 ) . many linear groups are not residually solvable ( for example , @xmath22 for @xmath23 , since these groups are perfect ) . on the other hand , it is well - known that linear groups are , for all but finitely many @xmath0 , virtually residually @xmath0 ; see , e.g. , ( * ? ? ? * theorem 4.7 ( i ) ) or ( * ? ? ? * window 7 , proposition 9 ) . ( this is essentially due to mal@xmath24cev @xcite ; however , he only showed the weaker statement that linear groups are residually finite . ) therefore , each @xmath1-manifold group which is linear is virtually residually @xmath0 for all but finitely many @xmath0 . as is well - known , the fundamental group of each hyperbolic @xmath1-manifold and of each seifert fibered manifold is linear ( cf . sections [ sec : hyperbolic ] and [ sec : seifert ] below ) , but it is not known whether this is true for all @xmath1-manifold groups . our main result now is the following . let @xmath25 be a @xmath1-manifold . then for all but finitely many primes @xmath0 the group @xmath26 is virtually residually @xmath0 . our proof of this theorem is substantially more involved than the proof of residual finiteness in @xcite . indeed , in order to show that the fundamental group @xmath27 of a given @xmath1-manifold @xmath25 is residually finite , one first easily reduces to the case that @xmath25 is closed , orientable and prime . then @xmath25 admits a jsj decomposition , hence may be viewed as the topological realization of a finite graph of based cw - complexes whose vertex spaces @xmath28 are either seifert fibered or hyperbolic @xmath1-manifolds , and whose edge spaces @xmath29 are tori . by virtue of the seifert - van kampen theorem , this gives rise to a description of @xmath3 as the fundamental group @xmath30 of a ( finite ) graph of groups @xmath31 whose vertex groups @xmath32 are linear groups , and whose edge groups @xmath33 are free abelian of rank @xmath34 . ( see section [ sec : graphs of groups ] below for a definition of a graph of groups . ) as remarked , finitely generated linear groups are residually finite ( in fact , virtually residually @xmath0 for all but finitely many @xmath0 ) . so one may assume that the jsj decomposition of @xmath25 is non - trivial . let now @xmath4 , @xmath5 . by choosing suitable filtrations of the vertex groups @xmath35 , one then constructs a group morphism @xmath36 where @xmath37 is the fundamental group of a graph @xmath38 of _ finite _ groups ( with the same underlying graph as @xmath31 ) , such that @xmath8 . it now simply remains to prove that such a graph of finite groups has residually finite fundamental group ( * ? ? ? * proof of theorem 3.1 ) . ( for the special cases of amalgamated products or hnn extensions , this was known much earlier @xcite . ) in contrast to this , amalgamated products and hnn extensions of @xmath0-groups , although always residually finite , are rarely residually @xmath0 @xcite , making it necessary to deviate from the flow of the argument in @xcite for the proof of our main theorem . another subtle difficulty is posed by the fact that for an infinite group @xmath3 , the intersection of all @xmath0-power index subgroups of @xmath3 being trivial does not guarantee that the intersection of all @xmath0-power index _ normal _ subgroups of @xmath3 is trivial ( i.e. , that @xmath3 is residually @xmath0 ) : it is easy to see that for every fibered @xmath1-manifold @xmath25 the intersection of all @xmath0-power index subgroups of the fundamental group @xmath3 of @xmath25 is trivial ( cf . proposition [ prop : fibered implies weakly residually p ] ) ; but there are fibered @xmath1-manifolds which are not residually nilpotent ( e.g. , as mentioned above , the fundamental group of a non - trivial fibered knot complement ) . we outline the overall strategy of our proof after discussing a few applications , and some commonalities between @xmath1-manifold groups and linear groups . our theorem may be used to significantly simplify the proof @xcite that twisted alexander polynomials detect fibered knots . more precisely , our main theorem replaces the ad hoc argument given in sections 5 and 6 of @xcite . this is explained in detail in @xcite . here , we restrict ourselves to discussing two more immediate applications . first , combining the theorem above with the fact ( proved in @xcite ) that the automorphism group of every finitely generated virtually residually @xmath0 group is also virtually residually @xmath0 , we obtain : [ cor : corollary 1 ] the automorphism group @xmath39 of every @xmath1-manifold group @xmath3 is virtually residually @xmath0 for all but finitely many @xmath0 , and hence virtually torsion free . therefore , every torsion subgroup of @xmath39 is finite , and there is an upper bound on the orders of the finite subgroups of @xmath39 . this corollary complements @xcite where it was shown that if @xmath25 is a haken @xmath1-manifold which is not seifert fibered , then there is an upper bound on the orders of the finite subgroups of the group @xmath40 of outer automorphisms of @xmath27 . however , since not every finite subgroup of @xmath40 lifts to a finite subgroup of @xmath39 ( see @xcite ) , this result is not immediately subsumed by corollary 1 . it seems not to be known whether the group of outer automorphisms of a @xmath1-manifold group is virtually residually @xmath0 . suppose @xmath25 is a closed aspherical @xmath1-manifold . then there is a bound on the size of finite groups of self - homeomorphisms of @xmath25 having a common fixed point . this follows immediately from corollary 1 and the fact that under the assumptions made here , for every effective action of a finite group @xmath41 on @xmath25 with a fixed point , the natural morphism @xmath42 is injective , cf . ( questions similar to the one addressed in corollary 2 are posed in kirby s list @xcite , see problem 3.39 . ) the main theorem of this paper stated above can also been seen as further evidence to the conjecture ( due to thurston ) that @xmath1-manifold groups are linear . ( this is problem 3.33 ( a ) in @xcite . ) it is known that every linear group @xmath3 has the following group - theoretic properties . ( recall from above that in our parlance , `` linear '' includes the condition of being finitely generated . ) @xmath3 is virtually residually @xmath0 for all but finitely many primes @xmath0 ( in particular , @xmath3 is residually finite and virtually torsion free ) . @xmath3 is either virtually solvable of finite rank ( hence has polynomial subgroup growth ) or the subgroup growth of @xmath3 is at least of type @xmath43 . @xmath3 satisfies the tits alternative : either @xmath3 is virtually solvable or @xmath3 contains a non - abelian free group . 4 . @xmath3 satisfies the lubotzky alternative : either @xmath3 is virtually solvable , or @xmath3 has `` large finite quotients . '' ( see ( * ? ? ? * window 9 ) for a precise statement . ) @xmath3 has property @xmath44 : if @xmath14 is a non - nilpotent normal subgroup ( not necessarily finitely generated ) of @xmath3 , then there is a normal subgroup @xmath45 of @xmath3 such that @xmath46 and @xmath47 is finite and non - nilpotent . we refer to @xcite , @xcite and ( * ? ? * windows 7 and 9 ) for proofs of statements ( 1)(5 ) above . property @xmath44 was isolated and studied in @xcite . examples for groups with property @xmath44 ( besides all linear groups ) include all polycyclic - by - finite groups . this class of groups is of interest since for a finitely presented group with property @xmath44 , one can algorithmically decide whether a given finitely generated normal subgroup is nilpotent . moreover , every group @xmath3 with property @xmath44 is an fgh ( ` frattini - gaschtz - hall ' ) group in the sense of @xcite , meaning that various important characteristic subgroups of @xmath3 ( among them the fitting subgroup of @xmath3 and the hirsch - plotkin radical of @xmath3 ) agree and are nilpotent . it is a consequence of the proof of the geometrization conjecture and the work of various authors that statements ( 1)(4 ) also hold for all @xmath1-manifold groups @xmath3 : property ( 1 ) is our main theorem . ( virtual torsion freeness also follows from the fact that irreducible @xmath1-manifolds with infinite fundamental group @xmath3 are eilenberg - mac lane spaces of type @xmath48 . ) property ( 1 ) also yields property ( 2 ) by ( * ? ? ? * theorem 8.3 ) . however , a stronger estimate is proved in ( * ? ? ? * theorem 1.1 ) : if @xmath3 is a @xmath1-manifold group which is not virtually solvable , then the subgroup growth of @xmath3 is at least of type @xmath49 . the fact that ( 3 ) holds for @xmath1-manifold groups is well - known to the experts . it is also known ( and implicit in ( * ? ? ? * theorem 4.1 ) ) that every @xmath1-manifold group @xmath3 satisfies a ` weak lubotzky alternative ' : either @xmath3 is virtually solvable or @xmath50 for every prime @xmath0 . here , given a ring @xmath51 the virtual first @xmath51-betti number @xmath52 of a group @xmath3 is the defined as the supremum ( possibly @xmath53 ) of all first @xmath51-betti numbers of normal , finite index subgroups of @xmath3 . ( the weak lubotzky alternative for linear groups is a consequence of the lubotzky alternative , cf . * theorem 1.3 ) . ) property ( 5 ) appears not to be known for arbitrary @xmath1-manifold groups ; in a future publication we plan address this issue . in @xcite , lubotzky gave a group - theoretic characterization of linear groups among finitely generated virtually residually @xmath0 groups , and it is tempting to employ it in combination with the methods in the present paper to prove the linearity of @xmath1-manifold groups . however , in general it seems to be difficult to verify that a given group satisfies this criterion . ( for some rare cases where this has been done successfully see @xcite . ) let @xmath25 be a @xmath1-manifold . in order to show that the finitely presentable group @xmath27 is virtually residually @xmath0 for all but finitely many @xmath0 , we proceed in five steps . we first show that it suffices to consider only @xmath1-manifolds which are closed , orientable and prime . suppose @xmath25 satisfies these requirements . then @xmath25 admits a jsj decomposition , so @xmath54 for a graph of groups @xmath31 whose vertex groups @xmath32 are fundamental groups of hyperbolic or seifert fibered @xmath1-manifolds ( and hence are linear groups ) , and whose edge groups @xmath33 are free abelian of rank @xmath34 . ( below we identify the fundamental group @xmath55 of every boundary torus @xmath29 of a jsj component @xmath28 of @xmath25 with a subgroup of @xmath32 in the natural way . ) since finitely generated linear groups are virtually residually @xmath0 for all but finitely many @xmath0 , we may assume from now on that the jsj decomposition of @xmath25 is non - trivial . we say that @xmath25 is _ seifert simple _ if every seifert fibered component in its jsj decomposition is of the form @xmath56 for some orientable surface @xmath57 which is either closed or has at least two boundary components . after passing to a finite cover of @xmath25 , we can further assume that @xmath25 is seifert simple . ( this step of the proof is analogous to the first part of the proof of residual finiteness in @xcite . ) next , we show that with the exception of finitely many primes @xmath0 , after passing to another finite cover of @xmath25 , we can construct what we call a _ complete boundary compatible @xmath0-filtration @xmath58 _ on each vertex group @xmath32 . this is a descending sequence of normal subgroups @xmath59 of @xmath60 such that @xmath61 is an elementary abelian @xmath0-group , which exhausts @xmath35 ( i.e. , @xmath62 ) , subject to further conditions , which we refrain from spelling out completely here ( see chapter [ ch : proof of the main results ] for the precise definition ) ; for example , we require that @xmath63 separates the fundamental group @xmath33 of every boundary torus @xmath29 of @xmath28 ( i.e. , for each @xmath64 there is some @xmath65 with @xmath66 ) , and that it intersects to the lower central @xmath0-filtration of @xmath55 ( i.e. , @xmath67 for all @xmath65 ) . the construction of these filtrations is inspired by the proof of the fact that linear groups are virtually residually @xmath0 for all but finitely many @xmath0 , but needs further ingredients : a localization theorem for finitely generated integral domains ( which we prove by using a combination of elementary commutative algebra and model theory ) and some facts about congruence subgroups . from now on until the end of this discussion we assume that on each @xmath35 we are given such a complete boundary compatible @xmath0-filtration @xmath63 . ( in some cases , e.g. , for graph manifolds , we can arrange for the lower central @xmath0-filtration @xmath68 of @xmath35 to be a complete boundary compatible @xmath0-filtration . ) a general criterion for the fundamental group of a graph of groups to be residually @xmath0 ( proposition [ hempel criterion residually p ] ) reduces our task to showing that for all @xmath65 , the graph of groups @xmath69 , obtained by replacing each vertex group @xmath35 and each edge group @xmath55 by their ( @xmath0-power ) quotients @xmath70 respectively @xmath71 , has residually @xmath0 fundamental group . the case @xmath72 plays an important role . if @xmath73 for each vertex @xmath74 , then @xmath75 is the `` mod @xmath0 homology graph of @xmath31 '' : its vertex and edge groups are the first homology groups with @xmath76-coefficients @xmath77 and @xmath78 of the vertex respectively edge groups of @xmath31 . let @xmath79 be a maximal subtree of the graph underlying the graph of groups @xmath31 . the fundamental group @xmath80 of @xmath75 can be described as an iterated hnn extension of the fundamental group @xmath81 of the restriction @xmath82 of @xmath75 to @xmath79 . replacing @xmath81 in this iterated hnn extension by the fiber sum @xmath83 ( colimit in the category of abelian groups ) of the tree @xmath82 of elementary abelian @xmath0-groups , we obtain a `` partial abelianization '' of @xmath80 along @xmath79 , which we denote by @xmath84 . in this step we show that , possibly excluding finitely many more primes , if @xmath84 is residually @xmath0 , then all the fundamental groups of the quotient graphs @xmath69 are residually @xmath0 ( and hence @xmath27 is residually @xmath0 by step 3 ) ; cf . theorem [ thm : reduction theorem ] . this requires two main ingredients : an amalgamation theorem for _ filtered _ @xmath0-groups ( theorem [ thm : higman ] ) which sharpens a result of higman @xcite , the proof of which requires some facts about wreath products and group rings ; and analogous criteria for iterated hnn extensions of @xmath0-groups to be residually @xmath0 ( corollary [ cor : chatzidakis ] ) , related to work of chatzidakis @xcite . a sufficient condition for @xmath84 to be residually @xmath0 is the existence of a group morphism @xmath85 to a finite abelian @xmath0-group @xmath86 which is injective on each vertex groups of @xmath75 . hence , in order to finish the proof , we show that after passing to a suitable finite cover of @xmath25 , we can achieve exactly this . now , the identification morphisms between the fundamental groups of boundary tori yield a collection of isomorphisms @xmath87 between subgroups of @xmath88 which provide obstructions to the existence of such a morphism . extending each @xmath87 to an automorphism of @xmath88 yields a morphism @xmath89 which is trivial on each vertex group of @xmath31 . to obtain the desired finite cover of @xmath25 , we introduce a construction , which we call the _ unfolding of @xmath31 along @xmath90 , _ which produces a morphism of graphs of groups @xmath91 and complete boundary compatible @xmath0-filtrations on the vertex groups of @xmath38 ( which give rise to the associated quotient graph of elementary abelian @xmath0-groups @xmath92 ) such that the induced morphism @xmath93 is injective with image @xmath94 , and such that for some maximal subtree @xmath95 of the underlying graph of @xmath38 , we do have a morphism @xmath96 to a @xmath0-group which is injective on the vertex groups of @xmath92 as required . it is legitimate to wonder whether our main theorem is but the echo of a more general fact . for example , one may ask : _ is every finitely presented residually finite torsion free group , for all but finitely many @xmath0 , virtually residually @xmath0 ? _ if we replace `` finitely presented '' by `` finitely generated '' and drop the condition `` torsion free , '' then the answer to this question is `` no '' : if @xmath3 is a group which is virtually residually @xmath0 for two distinct primes @xmath0 , then @xmath3 is virtually torsion free ; however , for every prime @xmath0 there are examples of @xmath0-torsion groups which are infinite , finitely generated , and residually finite , constructed by golod @xcite . ( on the other hand , it seems to be unknown whether there exist infinite _ finitely presented _ periodic groups . ) if we replace `` finitely presented '' by `` finitely generated , '' then the answer is also negative : it is a well - known theorem of grigorchuk @xcite ( see also ( * ? ? ? * interlude e ) ) that if @xmath3 is a finitely generated residually @xmath0 group which has word growth @xmath97 , then @xmath3 is virtually nilpotent ( and hence has polynomial growth ) . grigorchuk @xcite also constructed a finitely generated residually finite torsion free group @xmath98 whose word growth is intermediate ( i.e. , between polynomial and exponential ) and which is therefore not virtually residually @xmath0 for _ any _ @xmath0 . ( for a while , the question whether every finitely generated group of intermediate growth is residually finite was open ( * ? ? ? * vi.64 ) , but this turned out to be false @xcite . ) we also note that it might be difficult to find a @xmath1-manifold @xmath25 for which the set of exceptional primes in our main theorem is non - empty : it has been conjectured ( by thurston ) that every irreducible @xmath1-manifold which is not a graph manifold has a finite cover which is fibered ; on the other hand , our techniques easily yield : let @xmath25 be a fibered @xmath1-manifold . then for _ every _ prime @xmath0 the group @xmath26 is virtually residually @xmath0 . topics of this kind are also discussed , employing other arguments , in @xcite . the method for proving residual properties of @xmath1-manifold groups introduced in this paper is quite flexible . for example , it yields the following improvement of our main theorem for graph manifolds : let @xmath25 be a graph manifold . then for _ every _ prime @xmath0 the group @xmath26 is virtually residually @xmath0 . in @xcite , perron and shalen introduce the notion of a ` weakly residually @xmath0-nilpotent ' group , and show that all graph manifold groups have a finite index subgroup which is weakly residually @xmath0-nilpotent for every @xmath0 ( * ? ? ? * proposition 0.3 ) . however , the proof for that claim has a gap @xcite : in fact , one may show that a group is weakly residually @xmath0-nilpotent if and only if it is residually @xmath0 , and if @xmath25 is a sol - manifold , then @xmath26 does _ not _ have a finite index subgroup which is residually @xmath0 for every @xmath0 . these matters will be discussed further in @xcite . proposition 2 may be refined to show that virtually , the pro-@xmath0 topology on the fundamental group of a graph manifold @xmath25 interacts well with the structure of the graph of groups arising from the jsj decomposition of @xmath25 . to make this precise , suppose we are interested in a property @xmath2 of groups . the pro-@xmath2 topology on a group @xmath3 is the group topology with fundamental system of neighborhoods of the identity given by the normal subgroups @xmath99 such that @xmath100 is a @xmath2-group . inspired by a notion introduced in @xcite , we say that a @xmath1-manifold @xmath25 is _ if @xmath25 is closed prime orientable ( so the jsj decomposition of @xmath25 exists ) , its fundamental group @xmath27 is residually @xmath2 , and for each jsj component @xmath28 of @xmath25 and each component @xmath29 of the boundary of @xmath28 , the fundamental groups @xmath32 and @xmath33 , identified with subgroups of @xmath3 in the natural way , the following conditions hold : 1 . @xmath35 and @xmath55 are closed in the pro-@xmath2 topology on @xmath3 ; 2 . the pro-@xmath2 topology on @xmath3 induces the pro-@xmath2 topology on @xmath35 and on @xmath55 . wilton and zalesskii @xcite showed that all closed prime orientable @xmath1-manifolds are @xmath2-efficient , for @xmath2 the property of being finite . we show : let @xmath25 be a closed graph manifold . then for every @xmath0 , @xmath25 has a finite cover which is @xmath0-efficient . it seems plausible that this result might play the same role in studying the ( virtual ) conjugacy @xmath0-separability of graph manifold groups as finite efficiency does in the proof of their conjugacy separability , which was also established in @xcite . we do not pursue this issue further in the present paper ; however , we note that proposition 3 also implies that virtually , the fundamental group of a closed graph manifold group and its pro-@xmath0 completion have the same mod @xmath0 cohomology in a certain sense , explained in section [ sec : coh p - completeness ] . we do not know whether every closed prime @xmath1-manifold is , for all but finitely many @xmath0 ( or even just for infinitely many @xmath0 ) , virtually @xmath0-efficient . this question will be discussed further in section [ sec : p - efficiency in general ] . we note here that if @xmath25 is a @xmath0-efficient @xmath1-manifold with fundamental group @xmath3 , then condition ( 2 ) of @xmath0-efficiency implies that there exists some @xmath20 such that for every jsj component @xmath28 of @xmath25 with fundamental group @xmath35 ( identified in the natural way with a subgroup of @xmath3 ) , we have @xmath101 . as an indication that every closed prime @xmath1-manifold might indeed be virtually @xmath0-efficient for all but finitely many @xmath0 , we show that we can achieve this condition on @xmath102 in a finite cover . in fact , we may take @xmath72 , and so our result may be phrased as a statement about the mod @xmath0 homology of finite covers of @xmath1-manifolds as follows : every closed prime @xmath1-manifold @xmath25 has , for all but finitely many @xmath0 , a finite cover @xmath103 such that for every jsj component @xmath104 of @xmath103 , the natural morphism @xmath105 is injective . as should be apparent from the outline of the strategy given above , the proof of the main result of this paper involves going through several technical prerequisites ; some of those can be assumed as black boxes in a first reading . we recommend that the reader should first only skim the preliminary chapter [ ch : preliminaries ] , and take the main results of chapter [ ch : embedding theorems ] for granted before immediately moving to chapters [ ch : residual properties of graphs of groups ] and [ ch : proof of the main results ] , which contain the essence of steps 35 in the proof strategy indicated above . the various subsections labeled `` digression '' may also be skipped upon a first reading of the paper . in the final chapter [ ch : graph manifolds ] we give proofs of propositions 24 from above . throughout this paper , @xmath106 , @xmath20 range over the set @xmath107 of natural numbers . unless otherwise noted , @xmath0 denotes a prime number . by a @xmath0-group we mean a finite group whose order is a power of @xmath0 . the fields institute in toronto provided kind hospitality to the first - named author while a part of the work on this paper was done . he was also partially supported by national science foundation grant dms-0556197 . the second - named author was partially supported by a crm - ism fellowship and by cirget . we would like to thank steve boyer for many insightful conversations . we are also grateful to ian agol , christian haesemeyer , marc lackenby , yehuda shalom , and stefano vidussi for helpful comments . we also thank the referee for a careful reading of the manuscript . we begin by introducing basic terminology and fundamental facts concerning filtrations of groups . facts stated without proof or reference are standard ; for more information on filtrations ( in particular the connection to lie algebras , not explicitly exploited here ) , see @xcite . throughout this section we let @xmath3 be a group . we call a descending sequence @xmath108 of subgroups of @xmath3 a _ filtration _ of @xmath3 . we say that a filtration @xmath109 as above is _ complete _ if @xmath110 . we call a filtration @xmath111 of @xmath3 _ normal _ if each subgroup @xmath112 is normal in @xmath3 ; in this case , we write @xmath113 for the @xmath20th `` layer '' of @xmath111 . a _ filtered group _ is a pair @xmath114 where @xmath3 is a group and @xmath111 is a filtration of @xmath3 . the class of filtered groups are the objects of a category , where a morphism of filtered groups @xmath115 is a group morphism @xmath116 such that @xmath117 for all @xmath20 . note that if @xmath118 is a morphism between normal filtered groups , then for each @xmath20 , @xmath119 induces an injective group morphism @xmath120 . the category of filtered groups has direct products : if @xmath121 and @xmath122 are filtrations of @xmath3 respectively @xmath7 , then @xmath123 is a filtration of @xmath124 ( the _ product _ of the filtrations @xmath111 and @xmath125 ) . if both @xmath111 and @xmath125 are normal , then so is @xmath126 , with @xmath127 for all @xmath20 . let @xmath121 be a filtration of @xmath3 , and let @xmath128 . we call the filtration @xmath129 of @xmath14 the _ intersection _ of @xmath111 with @xmath14 . if @xmath130 is a filtration of a subgroup of @xmath3 with @xmath131 , or equivalently , if the natural inclusion @xmath132 is a morphism of filtered groups @xmath133 , then we say that @xmath111 _ intersects to @xmath134 _ on @xmath14 . we say that filtrations @xmath121 and @xmath135 of @xmath3 are _ equivalent _ if @xmath136 notation : @xmath137 . clearly @xmath138 is an equivalence relation on filtrations of @xmath3 . we say that @xmath111 _ induces _ a filtration @xmath134 on a subgroup @xmath14 of @xmath3 if @xmath139 . the following is obvious : [ lem : refine ] let @xmath14 be a common subgroup of @xmath3 and another group @xmath7 . let @xmath111 be a filtration of @xmath3 and @xmath125 be a filtration of @xmath7 with @xmath140 . we say that @xmath111 , @xmath125 _ induce the same filtration on @xmath14_. then there are filtrations @xmath141 of @xmath3 and @xmath142 of @xmath7 with @xmath143 . let @xmath121 and @xmath135 be filtrations of @xmath3 . we say that _ @xmath144 refines @xmath111 _ ( and @xmath144 is called a _ refinement _ of @xmath111 ) if there exists a strictly increasing map @xmath145 such that @xmath146 and @xmath147 for all @xmath65 . if in addition @xmath148 then we say that _ @xmath144 stretches @xmath111 _ ( and @xmath144 is called a _ stretching _ of @xmath111 ) . note that in this case @xmath141 , and the layers of @xmath149 are given by @xmath150 given a collection @xmath151 of groups , and for each @xmath152 a filtration @xmath153 of @xmath3 and a stretching @xmath149 of @xmath111 , we say that the @xmath149 ( where @xmath3 ranges over @xmath151 ) are _ compatible _ if @xmath154 can be chosen uniformly for all pairs @xmath155 . given @xmath156 , we say that a filtration @xmath121 of @xmath3 has _ length @xmath157 _ if @xmath158 ; the smallest such @xmath20 is called the length of @xmath111 , and if no such @xmath20 exists , then @xmath111 is said to have infinite length . we sometimes write a finite - length filtration of @xmath3 simply in the form @xmath159 with the understanding that @xmath160 for all @xmath161 . given a subgroup @xmath7 of @xmath3 we say that a filtration @xmath121 of @xmath3 _ separates @xmath7 _ if @xmath162 , and we simply say that @xmath111 is _ separating _ if it separates the trivial subgroup @xmath163 of @xmath3 . we say that a filtration of @xmath3 separates a collection of subgroups of @xmath3 if it separates each member . given two filtrations @xmath121 and @xmath135 of @xmath3 , we say that _ @xmath111 descends faster than @xmath144 _ ( in symbols : @xmath164 ) if @xmath165 for all @xmath20 . let @xmath121 be a complete filtration of @xmath3 . we say that @xmath111 is a _ central filtration _ of @xmath3 if @xmath166\leq g_{n+1}$ ] for all @xmath20 . every such central filtration @xmath111 of @xmath3 is automatically normal , and the successive quotients @xmath167 are central , that is , @xmath168 for all @xmath20 . in particular , the groups @xmath169 are abelian , and every refinement of a central filtration is central . the _ lower central filtration _ of @xmath3 is the fastest descending central filtration @xmath170 of @xmath3 , defined by @xmath171\text { for $ n\geq 1$.}\ ] ] the @xmath172 are totally invariant subgroups of @xmath3 . the first quotient @xmath173 $ ] is the abelianization @xmath174 of @xmath3 . for each @xmath65 , the commutator map @xmath175\colon \gamma_n(g)\times g\to\gamma_{n+1}(g)\ ] ] yields a surjective morphism @xmath176 thus if @xmath177 is finitely generated ( in particular , if @xmath3 is finitely generated ) , then so is each quotient @xmath178 . ( see , e.g. , ( * ? ? ? * ( 4.6 ) ) . ) given @xmath20 , there exists a central separating filtration of @xmath3 of length @xmath157 if and only if @xmath179 ; in this case , @xmath3 is said to be nilpotent of class @xmath157 . moreover , the group @xmath3 is residually nilpotent if and only if there exists a separating central filtration of @xmath3 , and in this case the lower central filtration of @xmath3 is separating . a _ strongly central filtration _ of @xmath3 is a complete filtration @xmath180 of @xmath3 with the property that @xmath181\leq g_{n+m}$ ] for all @xmath20 and @xmath106 . ( this is called an `` @xmath25-series '' in @xcite , and sometimes called a `` lazard filtration . '' ) the lower central filtration of @xmath3 is indeed strongly central . let @xmath121 be a filtration of @xmath3 . we say that @xmath111 is a _ @xmath0-filtration _ if @xmath182 for all @xmath65 . thus if @xmath111 is a normal @xmath0-filtration of @xmath3 then each non - trivial element of the quotient @xmath167 has order @xmath0 . a filtration of @xmath3 equivalent to a @xmath0-filtration of @xmath3 is itself a @xmath0-filtration . similarly , the intersection of a @xmath0-filtration of @xmath3 with any subgroup @xmath14 of @xmath3 is a @xmath0-filtration of @xmath14 , and the product of two @xmath0-filtrations is a @xmath0-filtration . ( all these statements , of course , also hold with `` central filtration '' respectively `` strongly central filtration '' in place of `` @xmath0-filtration '' everywhere . ) an important role in this paper is played by central @xmath0-filtrations : @xmath111 is a central @xmath0-filtration of @xmath3 precisely if @xmath183 ( g_n)^p \leq g_{n+1}\text { for all $ n\geq 1$.}\ ] ] in this case , each successive quotient @xmath184 has a natural structure of an @xmath76-linear space . the _ lower central @xmath0-filtration _ @xmath185 of @xmath3 is defined by @xmath186 \gamma^p_n(g)^p\text { for $ n\geq 1$.}\ ] ] the @xmath187 are totally invariant : if @xmath188 is a group morphism , then @xmath189 for all @xmath65 . we have @xmath190 and if @xmath3 is a @xmath0-group , then @xmath191\ , g^p$ ] is the frattini subgroup of @xmath3 . the lower central @xmath0-filtration is the fastest descending central @xmath0-filtration of @xmath3 . it was introduced in @xcite , and some of its basic properties are described in detail in ( * ? ? ? * chapter viii ) ( under the name `` @xmath192-series '' ) , for example : [ prop : lower p ] for all @xmath193 : 1 . @xmath194\leq \gamma^p_{m+n}(g)$ ] ; 2 . @xmath195 . for a proof see ( * ? ? ? * chapter viii , theorem 1.5 ) . [ lem : canonical filtration ] let @xmath196 . suppose there exists a central @xmath0-filtration @xmath121 of @xmath3 which intersects to the lower central @xmath0-filtration on @xmath79 . then every central @xmath0-filtration @xmath144 of @xmath3 with @xmath197 intersects to the lower central @xmath0-filtration of @xmath79 . in particular , the lower central @xmath0-filtration of @xmath3 intersects to the lower central @xmath0-filtration of @xmath79 . suppose @xmath144 is a central @xmath0-filtration of @xmath3 with @xmath197 . then @xmath198 . on the other hand , @xmath199 is a central @xmath0-filtration of @xmath79 , hence @xmath200 . the following is shown by induction on @xmath106 ( simultaneously for all @xmath20 ) ; we leave the details to the reader . [ lem : iterated gamma ] let @xmath121 be a central @xmath0-filtration of @xmath3 . then @xmath201 if there exists a central separating @xmath0-filtration of @xmath3 of length @xmath157 , then we have @xmath202 , and in this case we say that @xmath3 has _ lower @xmath0-length @xmath157 ; _ the smallest such @xmath20 ( if it exists ) is the _ lower @xmath0-length _ of @xmath3 . so for example , @xmath3 has lower @xmath0-length @xmath163 if and only if @xmath3 is a non - trivial elementary abelian @xmath0-group . in general , a finite group is a @xmath0-group if and only if it has finite lower @xmath0-length . a finitely generated group @xmath3 is residually @xmath0 if and only if there exists a separating central @xmath0-filtration of @xmath3 ; in this case the lower central @xmath0-filtration of @xmath3 is separating . a strongly central filtration @xmath180 of @xmath3 is called a _ dimensional @xmath0-filtration _ if @xmath203 for all @xmath65 . ( so every dimensional @xmath0-filtration is indeed a @xmath0-filtration . ) the fastest descending dimensional @xmath0-filtration @xmath204 of @xmath3 is given by @xmath205\quad \text{for $ n>1$.}\ ] ] the @xmath206 are called the _ dimension subgroups _ of @xmath3 in characteristic @xmath0 . if we want to stress the dependence of @xmath206 on @xmath3 and @xmath0 , we write @xmath207 . we call @xmath204 the _ lower dimensional @xmath0-filtration _ of @xmath3 . ( this is also variously called the zassenhaus , jennings or lazard series of @xmath3 in the literature . dimensional @xmath0-filtrations are called `` @xmath208-series '' in @xcite . ) its length will be called the _ lower dimensional @xmath0-length _ of the group @xmath3 . note that the @xmath206 are totally invariant , and clearly @xmath209 for all @xmath65 . a closed formula for @xmath210 is due to lazard ( cf . * theorem 11.2 ) ) : @xmath211 ( the groups @xmath206 will only appear in chapter [ ch : embedding theorems ] . ) we say that a complete normal finite - length filtration @xmath121 of @xmath3 is a _ chief filtration _ of @xmath3 if each quotient @xmath167 is either trivial or a minimal non - trivial subgroup of @xmath212 . every chief filtration of @xmath3 induces a chief filtration on any subgroup of @xmath3 . the quotients @xmath184 in a chief filtration of @xmath3 are called the _ chief factors _ of the filtration . every finite group has a chief filtration of finite length . every normal filtration of finite length of a finite group may be refined to a chief filtration of finite length . in particular , if @xmath3 admits a finite - length chief filtration and @xmath25 is any normal subgroup of @xmath3 , then one can always find a finite - length chief filtration in which @xmath25 is one of the terms . every chief factor of a finite nilpotent group is central , and every chief factor of a @xmath0-group has order @xmath0 . ( hence every chief filtration of a @xmath0-group is a central @xmath0-filtration . ) let @xmath213 be an operation which associates to a group @xmath214 a totally invariant subgroup @xmath215 of @xmath214 , i.e. , every group morphism @xmath216 restricts to a group morphism @xmath217 . recall that a subgroup @xmath7 of @xmath3 is called a _ retract _ of @xmath3 if the natural inclusion @xmath218 left splits . note that if @xmath7 is a free factor of @xmath3 , i.e. , if there exists a subgroup @xmath219 of @xmath3 such that @xmath3 is ( isomorphic to ) the free product @xmath220 , then @xmath7 is a retract of @xmath3 . [ lem : lower central for semidirect products ] if @xmath221 is a retract of @xmath3 , then @xmath222 if @xmath223 where @xmath224 and @xmath221 , then @xmath225 let @xmath221 . we have @xmath226 , and if @xmath6 is a morphism which is the identity on @xmath7 , then @xmath227 . this shows the first statement . for the second statement , assume @xmath223 , where @xmath221 , and let @xmath228 be the natural morphism . then @xmath229 , thus @xmath230 as required . this lemma applies in particular to the operation @xmath88 which , for given @xmath65 , yields the @xmath20th term in the lower central filtration @xmath231 , lower central @xmath0-filtration @xmath232 , or lower dimensional @xmath0-filtration @xmath210 , respectively . we also note : [ lem : retract ] let @xmath221 be a retract of @xmath3 , and suppose @xmath3 is residually @xmath0 . then @xmath233 separates @xmath7 . let @xmath6 be a morphism which is the identity on @xmath7 . let @xmath234 . let @xmath65 , and take @xmath235 with @xmath236 . then @xmath237 and hence @xmath238 . since this holds for every @xmath65 , and @xmath239 , we obtain @xmath240 as required . for every @xmath65 we write @xmath241 . recall that each @xmath242 is in a natural way an @xmath76-linear space . if @xmath177 is finitely generated , then each of the layers @xmath242 is finite ( cf . ii ) ; see also @xcite ) . note that if @xmath3 is abelian and written additively , then @xmath243 and so @xmath244 , for all @xmath20 . if @xmath7 denotes the abelianization @xmath245 $ ] of @xmath3 , with natural morphism @xmath228 , then @xmath246 for every @xmath65 , and the restriction of @xmath247 to @xmath187 induces a surjective morphism @xmath248 which is an isomorphism for @xmath249 . in the rest of this subsection we develop some further basic facts about the groups @xmath242 . the main tool is a precise version of the hall - petrescu formula ( * ? ? ? * theorem 1.1.30 ) . in the theorem and its corollaries below , we let @xmath250 , @xmath251 range over @xmath3 . [ thm : hp ] for all @xmath193 , @xmath252^{m\choose 2}\ , r_{mn}\mod \gamma_{n+1}(g)\ ] ] where @xmath253 is a product of powers @xmath254 with @xmath255 and @xmath256 is a @xmath18-linear combination of @xmath257 , for some @xmath258 with @xmath259 . this yields a sharpening of ( * ? ? * lemma 11.9 ) : [ cor : hp ] let @xmath260 . then @xmath261^{2^{m-1 } } \mod \gamma_2(g)^{2^m}\gamma_3(g)^{2^{m-1}}\prod_{d=2}^m \gamma_{2^d}(g)^{2^{m - d}},\ ] ] and if @xmath0 is odd , then @xmath262 we apply the previous theorem with @xmath263 and @xmath264 in place of @xmath106 . if @xmath0 is odd then @xmath265 and hence @xmath266^{p^m\choose 2}\in\gamma_2(g)^{p^m}$ ] . for @xmath267 note that @xmath266^{2^m\choose 2}\equiv [ x , y]^{2^{m-1}}\bmod\gamma_2(g)^{2^m}$ ] since @xmath268^{2^m\choose 2 } = [ y , x]^{2^{2m-1}-2^{m-1}}=[y , x]^{2^{2m-1}}[x , y]^{2^{m-1}}.\ ] ] it is well - known that if @xmath269 where @xmath270 , @xmath271 not divisible by @xmath0 , then @xmath272 is divisible by @xmath273 . therefore every @xmath18-linear combination of the binomial coefficients @xmath274 is divisible by @xmath275 where @xmath276 . so each of the powers @xmath254 in theorem [ thm : hp ] lies in @xmath277 . therefore each @xmath254 lies in @xmath278 , since @xmath279 . we might have @xmath280 , but in this case @xmath254 actually lies in @xmath281 , since @xmath282 and so @xmath283 . similarly , we have that if @xmath267 and @xmath284 , then @xmath285 . therefore the corollary follows from theorem [ thm : hp ] . we note some consequences of this corollary : [ cor : hp , 2 ] let @xmath260 . then @xmath261^{2^{m-1 } } \mod \gamma^2_{m+2}(g),\ ] ] and if @xmath0 is odd , then @xmath286 we have @xmath287 as well as @xmath288 and @xmath289 if @xmath290 and @xmath291 , or if @xmath267 and @xmath292 , then we have @xmath293 and thus @xmath294 . hence the corollary follows from corollary [ cor : hp ] . if @xmath3 is abelian ( written additively ) , then @xmath244 for every @xmath65 , and multiplication by @xmath0 on @xmath3 induces a group morphism @xmath295 . ( if in addition @xmath3 does not have @xmath0-torsion , then for every @xmath65 , this morphism is actually an isomorphism . ) more generally , we have : [ cor : phi_n ] let @xmath121 be a central @xmath0-filtration of @xmath3 , and let @xmath65 . suppose @xmath296 or @xmath297\leq g_{n+2}$ ] . then for every @xmath260 the map @xmath298 is a group morphism whose kernel contains @xmath299 . this follows from corollary [ cor : hp , 2 ] ( applied to @xmath112 in place of @xmath3 ) and lemma [ lem : iterated gamma ] . for @xmath300 , the previous corollary shows that unless @xmath267 , @xmath249 , and @xmath301 , the map in induces group morphisms @xmath302 in general , the morphisms @xmath303 are not injective . however , we do have : [ lem : phin injective ] suppose @xmath296 , @xmath304 , or @xmath305 . also assume that for @xmath306 , the group @xmath307 is @xmath0-torsion free or equivalently , the sections @xmath308 are @xmath0-torsion free for @xmath309 . then the group morphisms @xmath303 @xmath260 are injective . this lemma is a consequence of the following fact , elucidating the structure of the elementary abelian @xmath0-groups @xmath242 . we set @xmath310 for @xmath65 . for every @xmath65 the map @xmath311 where @xmath312 and @xmath313 , is a surjection , and a group morphism unless @xmath0 is odd and @xmath249 . if for @xmath306 , the group @xmath307 is @xmath0-torsion free , then this map is also injective . for a proof of this theorem see ( * * chapter viii , theorems 1.8 and 1.9 ) . ( theorem 1.9 of loc . cit . has the more restrictive assumption that @xmath314 is torsion free for @xmath306 , but inspection of the proof shows that only @xmath0-torsion needs to be excluded . ) note that this theorem in particular implies that if @xmath177 is finitely generated , then for each central @xmath0-filtration @xmath111 of @xmath3 , all layers @xmath184 are finite . in the situation of lemma [ lem : phin injective ] , we now have a commutative diagram @xmath315 \ar[r ] & l^p_n(g ) \ar[d]_{\phi_{n , m } } \\ \gamma_1\times\cdots\times\gamma_n \times\gamma_{n+1}\times\cdots\times\gamma_m \ar[r ] & l^p_{n+m}(g ) } \ ] ] where the top horizontal map is a bijection , and the left vertical map is given by @xmath316 and hence injective . this yields lemma [ lem : phin injective ] . suppose that @xmath0 is odd or @xmath317 . then for each @xmath260 we have a group morphism @xmath318 which is given by @xmath319 if the abelian group @xmath320 does not have @xmath0-torsion , then for each @xmath260 , the morphism @xmath321 is injective . let now @xmath121 be an arbitrary central @xmath0-filtration of @xmath3 . in corollary [ cor : phi_n ] we saw that if @xmath0 is odd or @xmath322\leq g_3 $ ] , then for each @xmath65 , exponentiation to the power @xmath323 on @xmath3 induces a group morphism @xmath324 , and above we gave a criterion , in the case @xmath300 , for these morphisms to be injective . the following definition captures this phenomenon : we say that @xmath111 is _ @xmath0-potent _ if for every @xmath65 , the map @xmath325 is a group morphism whose kernel _ equals _ @xmath326 . in this case , we have an induced injective group morphism @xmath327 , which we denote by @xmath328 . it will be useful to have terminology for a few variants of this definition available : we say that @xmath111 is _ strongly @xmath0-potent _ if for every @xmath65 , the map @xmath329 is a morphism with kernel @xmath299 . in this case we have induced injective morphisms @xmath330 . we call @xmath111 _ uniformly @xmath0-potent _ if @xmath111 is strongly @xmath0-potent and each @xmath331 ( @xmath65 ) is an isomorphism . we simply say that @xmath3 is _ @xmath0-potent _ if @xmath233 is @xmath0-potent . ( similarly for strongly @xmath0-potent respectively uniformly @xmath0-potent . ) clearly uniformly @xmath0-potent @xmath332 strongly @xmath0-potent @xmath332 @xmath0-potent . 1 . if @xmath3 is abelian @xmath0-torsion free , then @xmath3 is uniformly @xmath0-potent . 2 . every uniformly powerful pro-@xmath0 group is ( almost by definition ) uniformly @xmath0-potent ; see ( * ? ? ? * chapter 4 ) . ( this explains our choice of terminology . ) examples may be obtained by considering the congruence subgroups of @xmath333 . for simplicity assume @xmath0 is odd , and for each @xmath65 define @xmath334 then the groups @xmath335 form a uniformly @xmath0-potent separating filtration of the pro-@xmath0 group @xmath336 , and in fact @xmath337 for every @xmath65 ; see ( * ? ? ? * theorem 5.2 ) , and also section [ sec : csg ] below . we say that a central @xmath0-filtration @xmath111 of @xmath3 has finite rank if there exists an integer @xmath338 such that @xmath339 for all @xmath65 . 1 . if @xmath3 admits a separating strongly @xmath0-potent filtration , then @xmath3 is torsion free . 2 . if @xmath121 is a strongly @xmath0-potent filtration of @xmath3 having finite rank , then for some @xmath260 , the filtration @xmath340 of @xmath341 is uniformly @xmath0-potent . our chief interest in uniformly @xmath0-potent filtrations is in the case of example ( 1 ) above . for sake of completeness , we note : [ lem : linearity ] suppose @xmath3 is finitely generated . if @xmath3 admits a separating strongly @xmath0-potent filtration of finite rank , then @xmath3 is necessarily linear over @xmath342 . conversely , if @xmath3 is linear , then for all but finitely many @xmath0 , some finite index normal subgroup of @xmath3 admits a separating uniformly @xmath0-potent filtration . this is essentially lubotzky s linearity criterion @xcite ( see also interlude b in @xcite ) : suppose @xmath111 is a separating strongly @xmath0-potent filtration of finite rank ; then @xmath111 is a @xmath0-congruence system of finite rank in the sense of ( * ? ? ? * definition b.2 ) , hence @xmath3 is linear over @xmath342 by proposition b.3 of loc . conversely , suppose @xmath3 is linear . then for all but finitely many @xmath0 , the group @xmath3 admits an embedding into @xmath333 for some @xmath343 ( * ? ? ? * lemma b.4 ) . assuming now that @xmath0 is odd and @xmath344 , the filtration @xmath121 of @xmath3 given by @xmath345 is separating and , as filtration of @xmath346 , strongly @xmath0-potent of finite rank . now use the preceding remark ( 2 ) . every normal filtration @xmath121 of @xmath3 gives rise to a topology on @xmath3 making @xmath3 into a topological group with a fundamental system of neighborhoods of the identity given by the groups @xmath112 . we denote by @xmath347 the associated completion of @xmath3 , that is , the topological group @xmath348 we have a natural continuous morphism @xmath349 , which is injective precisely if @xmath111 is separating . if all @xmath112 are of finite index in @xmath3 , then @xmath347 is a profinite group . in particular , if @xmath177 is finitely generated and @xmath300 , then we obtain the pro-@xmath0 topology on @xmath3 , and @xmath350 is the pro-@xmath0 completion of @xmath3 . the proof of ( * ? ? ? * proposition b.3 ) shows that if @xmath111 is a strongly @xmath0-potent filtration of @xmath3 having finite rank , then @xmath347 is isomorphic ( qua topological group ) to a @xmath0-adic analytic group . ( this was originally shown in ( * ? ? ? * thorme 3.1.7 ) . ) if @xmath177 is @xmath0-torsion free , then @xmath3 is @xmath0-potent for all odd primes @xmath0 ( by corollary [ cor : phi_n ] ) , but not in general for @xmath267 . however , assuming that both @xmath351 and @xmath352 are @xmath34-torsion free ensures that @xmath3 always has a fully characteristic subgroup ( of @xmath34-power - index if @xmath3 is finitely generated ) carrying a @xmath34-potent filtration . more precisely , and more generally : [ lem:2-potent ] let @xmath353 , where @xmath354 , and for @xmath65 define @xmath355 . then @xmath356 is a strongly central @xmath0-filtration of @xmath357 , and for every @xmath65 , the map @xmath358 induces a group morphism @xmath359 . if @xmath360 is @xmath0-torsion free for @xmath361 , then @xmath144 is @xmath0-potent . if @xmath360 is @xmath0-torsion free for every @xmath65 , then @xmath144 is strongly @xmath0-potent . it is clear that @xmath144 is a strongly central @xmath0-filtration of @xmath357 ( cf . proposition [ prop : lower p ] ) . by corollary [ cor : phi_n ] , @xmath358 induces a group morphism @xmath359 . the rest follows from lemma [ lem : phin injective ] . we are mainly interested in this lemma when @xmath3 is free , in which case @xmath360 actually is torsion free for every @xmath65 . we note that the assumption that @xmath360 is torsion free for every @xmath65 is also satisfied in other interesting situations , e.g. , for primitive @xmath163-relator groups @xcite , such as surface groups , and primitive link groups ( * ? ? ? * theorems 2 and 1 ) . in this section we summarize some basic definitions and facts concerning graphs of groups and their fundamental groups . we refer to @xcite for missing details . a _ graph _ @xmath362 consists of a set @xmath363 of _ vertices _ and a set @xmath364 of _ edges _ , and two maps @xmath365 and @xmath366 , subject to the following condition : for each @xmath367 we have @xmath368 , @xmath369 , and @xmath370 . we call @xmath370 and @xmath371 the _ origin _ respectively the _ terminus _ of @xmath372 . the edge @xmath373 is called the _ inverse edge _ of @xmath372 . we call the set @xmath374 a _ topological edge _ of @xmath362 . given a graph @xmath362 as above , a _ subgraph of @xmath362 _ is a graph @xmath375 whose vertex and edge sets @xmath376 and @xmath377 are subsets of @xmath378 respectively @xmath379 such that @xmath380 and @xmath381 for all @xmath382 , and such that the maps @xmath383 , @xmath384 and @xmath385 on @xmath375 are the restrictions of the corresponding maps from @xmath379 to @xmath377 . if @xmath386 is a vertex of @xmath362 we denote by @xmath387 the subgraph of @xmath362 with vertex set @xmath388 and edge set @xmath389 . similarly , if @xmath390 we denote by @xmath391 the subgraph of @xmath362 with vertex set @xmath378 and edge set @xmath392 . given graphs @xmath362 , @xmath375 , a morphism @xmath393 consists of two maps @xmath394 and @xmath395 , both also denoted by @xmath396 , with the property that @xmath397 we say that a morphism @xmath398 is an embedding if the corresponding maps @xmath399 and @xmath395 are injective . if @xmath362 is a subgraph of @xmath375 , then the natural inclusions @xmath399 and @xmath395 form an embedding @xmath398 . let @xmath362 be a graph , and let @xmath379 denote a subset of the edge set @xmath400 of @xmath362 which is closed under taking inverses . an _ orientation _ of @xmath379 is a subset @xmath401 of @xmath379 such that @xmath379 is the disjoint union of @xmath401 and @xmath402 . by an orientation of @xmath362 we mean an orientation of @xmath400 . an _ oriented graph _ is a graph together with one of its orientations . given two sets @xmath378 and @xmath401 and a map @xmath403 there exists ( up to isomorphism ) a unique graph @xmath362 with vertex set @xmath378 such that @xmath401 is an orientation of @xmath362 and such that the origin and terminus maps on @xmath401 are given by @xmath383 and @xmath384 respectively . ( the set of edges of @xmath362 is the disjoint union @xmath404 where @xmath402 is a copy of @xmath401 . ) note that given graphs @xmath362 and @xmath375 and an orientation @xmath401 of @xmath362 , in order to specify a morphism @xmath398 it suffices to give maps @xmath399 and @xmath405 , both here denoted by @xmath396 , such that @xmath406 and @xmath407 for all @xmath408 . let @xmath409 denote the graph with vertex set @xmath410 and orientation given by @xmath411 with @xmath412 , @xmath413 . a _ path _ of length @xmath20 in the graph @xmath362 from a vertex @xmath414 to a vertex @xmath415 is a morphism @xmath416 such that @xmath417 and @xmath418 . such a path @xmath0 is determined by the sequence @xmath419 of edges @xmath420 with @xmath421 , @xmath422 for @xmath423 , and @xmath424 ; we also write @xmath425 . a _ circuit _ of length @xmath20 in @xmath362 ( where @xmath426 ) is a path @xmath427 such that the vertices @xmath428 are pairwise distinct , and @xmath429 . a _ tree _ is a graph which is connected ( for all vertices @xmath74 , @xmath430 there is a path from @xmath74 to @xmath430 ) and without circuits . if @xmath362 is a tree , then for any pair @xmath431 of vertices of @xmath362 there exists a unique path of shortest length from @xmath74 to @xmath430 , called the _ geodesic _ from @xmath74 to @xmath430 in @xmath362 . _ in the rest of this paper , unless otherwise noted , we assume that all graphs are finite i.e. , their vertex sets and edge sets are finite and connected . _ let @xmath362 be a graph . a _ graph @xmath151 of groups based on @xmath362 _ consists of families @xmath432 and @xmath433 of groups satisfying @xmath434 for every @xmath435 , together with a family @xmath436 of group embeddings @xmath437 ( @xmath435 ) . ( we also call @xmath362 the _ underlying graph _ of @xmath31 . ) let @xmath31 be a graph of groups based on a graph @xmath362 . we recall the construction of the fundamental group @xmath54 of @xmath31 from ( * ? ? ? * i.5.1 ) . first , consider the group @xmath438 ( the _ path group _ of @xmath31 ) generated by the groups @xmath35 ( @xmath439 ) and the elements @xmath435 subject to the relations @xmath440 by a _ path _ of length @xmath20 in @xmath31 from a vertex @xmath414 to a vertex @xmath415 we mean a sequence @xmath441 where @xmath419 is a path of length @xmath20 in @xmath362 from @xmath74 to @xmath430 and where @xmath442 for @xmath443 , with @xmath444 and @xmath445 for @xmath446 . if @xmath447 for every @xmath258 then @xmath231 is called an _ edge path_. we say that the path @xmath231 _ represents _ the element @xmath448 of @xmath438 . such a path @xmath231 in @xmath31 is _ reduced _ if it satisfies the following conditions : 1 . if @xmath449 then @xmath450 ; 2 . if @xmath426 then @xmath451 for each index @xmath258 such that @xmath452 . it is easy to see that every @xmath5 in @xmath438 is represented by a reduced path in @xmath31 ( namely , a path of minimal length representing @xmath453 ) . conversely , if @xmath453 is represented by a reduced path in @xmath31 , then @xmath5 ( * ? ? ? * i.5.2 , theorem 11 ) . in particular , the natural morphisms @xmath454 are injective ( * ? ? ? * i.5.2 , corollary 1 ) . let now @xmath386 be a fixed vertex of @xmath362 . the fundamental group @xmath455 of @xmath31 ( with base point @xmath386 ) is defined to be the subgroup of @xmath438 consisting of the elements represented by paths in @xmath31 from @xmath386 back to itself . if @xmath456 is another base point , and @xmath453 is an element of @xmath438 represented by a path from @xmath386 to @xmath457 , then @xmath458 is an isomorphism ; by abuse of notation we write @xmath30 to denote @xmath455 if the particular choice of base point is irrelevant . given a subgraph @xmath375 of @xmath362 , we denote by @xmath459 the restriction of the graph of groups @xmath31 to @xmath375 , defined in the obvious way ; given a base point @xmath460 , the natural inclusion @xmath461 induces an injective morphism @xmath462 ( * ? ? ? * corollary 1.14 ) . an alternative description of @xmath455 , which is often useful , is as follows : let @xmath79 be a maximal subtree of @xmath362 . the fundamental group @xmath463 of @xmath31 relative to @xmath79 is defined by contracting @xmath79 to a point : @xmath464 the natural projection @xmath465 restricts to an isomorphism @xmath466 , cf . * i.5.1 , proposition 20 ) . identifying @xmath455 with @xmath463 via this isomorphism , for each @xmath414 we obtain a group morphism @xmath467 . this group morphism is injective and given by @xmath468 where @xmath469 is represented by the reduced edge path in @xmath470 from @xmath386 to @xmath74 . we often tacitly identify the vertex groups @xmath35 with subgroups of @xmath455 in this way . this depends on the choice of maximal subtree @xmath79 of @xmath362 , but for a different choice of @xmath79 we obtain a conjugate subgroup of @xmath455 , so this dependence is generally harmless . for every @xmath435 let @xmath471 and let @xmath472 . fix an orientation @xmath401 of @xmath379 . the identification of @xmath473 with @xmath463 means that @xmath3 may be described as the iterated hnn extension @xmath474 of @xmath475 , where the latter group is itself an iterated amalgamated product ( of the vertex groups of @xmath470 , with identifications @xmath476 for @xmath477 , @xmath478 ) . let @xmath362 be a graph , @xmath363 , @xmath364 . a _ graph of based cw - complexes _ @xmath479 with underlying graph @xmath362 consists of a collection of connected cw - complexes @xmath480 and @xmath481 such that @xmath482 for every @xmath367 together with a collection @xmath483 of cellular embeddings . furthermore we require a choice of base points for each @xmath484 ( @xmath367 ) and @xmath485 ( @xmath414 ) , with @xmath484 and @xmath486 being assigned the same base point , and for each @xmath367 a choice of path in @xmath487 connecting the base points of @xmath488 and @xmath487 . finally we demand that the following hold : for each @xmath367 the map @xmath489 induces an embedding @xmath490 of groups ( henceforth also denoted by @xmath489 ) ; given @xmath414 the images of the maps @xmath491 with @xmath492 are disjoint ; and given @xmath414 the set @xmath493 is non - empty . ( conditions ( 2 ) and ( 3 ) can always be achieved by attaching products , without changing the homotopy types of the complexes @xmath485 and @xmath484 ) . let @xmath479 be a graph of based cw - complexes with underlying graph @xmath362 . the _ topological realization _ of @xmath479 is the following cw - complex : @xmath494 \right ) \bigg/ \sim \qquad \text{(disjoint union)}\ ] ] where for @xmath495 we set @xmath496 and for @xmath497 $ ] we set @xmath498 $ ] . note that we have canonical inclusion maps of @xmath485 and @xmath499 into @xmath500 . we identify @xmath485 and @xmath501 with their images in @xmath500 . the collections @xmath502 and @xmath503 together with the group embeddings @xmath504 form a graph of groups based on @xmath362 , whose fundamental group ( as defined above ) is naturally isomorphic to the fundamental group of the topological space @xmath505 ; see ( * ? ? ? * theorem 2.1 ) . conversely , every graph of groups @xmath31 based on @xmath362 gives rise to a graph @xmath479 of based cw - complexes with underlying graph @xmath362 such that @xmath506 , @xmath507 , and with the embeddings @xmath508 inducing the edge morphisms @xmath437 of @xmath31 . ( by choosing for each @xmath485 and @xmath484 classifying spaces @xmath509 respectively @xmath510 . ) let now @xmath31 and @xmath511 be graphs of groups , based on the graphs @xmath362 respectively @xmath375 . a _ morphism _ @xmath512 of graphs of groups consists of the following data : a morphism @xmath398 of graphs , also denoted by @xmath396 ; families of group morphisms @xmath513 and families of elements @xmath514 and @xmath515 of @xmath516 , satisfying the following conditions , with @xmath74 ranging over @xmath517 and @xmath372 over @xmath400 : 1 . @xmath518 ; 2 . @xmath519 ; 3 . for @xmath520 we have @xmath521 , and 4 . the diagram @xmath522^{{\operatorname{ad}}(\delta_e)\circ f'_{\phi(e ) } } & g'_{\phi(v ) } \\ g_e \ar[u]^{\phi_e } \ar[r]^{f_e } & g_{v } \ar[u]^{\phi_{v } } } \end{split}\ ] ] commutes . here , given a group @xmath3 and @xmath523 , @xmath524 denotes the inner automorphism @xmath525 of @xmath3 . every such morphism of graphs of groups as above induces a group morphism @xmath526 between path groups satisfying @xmath527 given a base point @xmath528 for @xmath529 and taking the base point @xmath530 for @xmath531 , this morphism restricts to a morphism @xmath532 . * proposition 2.4 ) . ) given an integer @xmath533 we say that @xmath396 has degree @xmath343 if @xmath534=d$ ] . every finite index subgroup of the fundamental group of a graph of groups arises in this way : [ thm : bass - serre ] let @xmath31 be a graph of groups with underlying graph @xmath362 , and let @xmath7 be a finite index subgroup of @xmath54 . then there exists a graph of groups @xmath535 and a morphism @xmath536 such that @xmath537 is injective with image @xmath7 . this may be seen as a consequence of the main structure theorem of the bass - serre theory of groups acting on trees ( cf . * section 4 ) ) . alternatively , one may employ the topological interpretation of @xmath30 sketched in section [ sec : morphisms of graphs of groups ] ( cf . @xcite ) : if @xmath479 is a graph of based cw - complexes based on a graph @xmath362 , with topological realization @xmath538 , and @xmath539 is a finite covering map , then @xmath540 is in a natural way the topological realization of a graph of based cw - complexes @xmath541 : the vertex spaces of @xmath541 are given by the components of @xmath542 , @xmath439 , and the edge spaces of @xmath541 are given by the components of @xmath543 , @xmath435 . let @xmath544 be the underlying graph of @xmath541 ; the covering map @xmath545 induces a morphism of graphs @xmath546 which we also denote by @xmath247 . the edge spaces and vertex spaces of @xmath541 are then endowed with base points covering the base points of @xmath479 ; and for @xmath547 , @xmath548 , we pick a path connecting the base points of @xmath549 and @xmath550 . this data gives rise to a morphism @xmath551 between the graphs of groups @xmath552 and @xmath31 associated to @xmath553 respectively @xmath479 such that @xmath554 agrees with the group morphism @xmath555 induced by @xmath247 . in the context of the previous theorem , one can choose the vertex groups of @xmath535 to be @xmath556 , where @xmath74 ranges over the vertices of @xmath362 and @xmath453 over a suitable set of @xmath557-double coset representatives , and the edge groups to be @xmath558 , where @xmath372 ranges over the edges of @xmath362 and @xmath453 over a suitable set of @xmath559-double coset representatives ( * ? ? ? * section 8.5 , theorem 27 ) . in particular , if @xmath7 is a _ normal _ subgroup of @xmath3 , then one can choose @xmath535 such that its vertex groups are of the form @xmath560 , and its edge groups of the form @xmath561 . let @xmath31 , @xmath511 be graphs of groups based on the graphs @xmath362 , @xmath375 , respectively . then @xmath511 is a _ subgraph of subgroups _ of @xmath31 if 1 . @xmath375 is a subgraph of @xmath362 ; 2 . @xmath562 for each @xmath563 ; 3 . @xmath564 and @xmath565 for each @xmath382 ; 4 . @xmath566 for all @xmath382 , @xmath520 . in this case , the natural inclusions @xmath567 ( @xmath563 ) and @xmath568 ( @xmath382 ) form a morphism @xmath569 of graphs of groups , and given @xmath460 , the induced morphism @xmath570 is injective ( * ? ? ? * corollary 1.14 ) . we say that a subgraph of subgroups @xmath511 of @xmath31 is _ normal _ if @xmath571 for each @xmath563 . ( then @xmath572 for each @xmath382 by condition ( 4 ) . ) for example , given a subgraph @xmath375 of @xmath362 , the graph of groups @xmath459 is a normal subgraph of subgroups of @xmath31 . a subgraph of subgroups @xmath511 of @xmath31 is called a _ graph of subgroups _ of @xmath31 if the underlying graph @xmath375 of @xmath511 is the same as the underlying graph @xmath362 of @xmath31 . let @xmath511 be a normal graph of subgroups of @xmath31 . then we obtain a graph of groups based on the graph @xmath362 , which we denote by @xmath573 , with vertex groups @xmath574 ( @xmath439 ) , edge groups @xmath575 and edge morphisms @xmath576 induced by @xmath489 @xmath577 ) . the natural surjections @xmath578 and @xmath579 yield a morphism @xmath580 of graphs of groups , which in turn yields a group morphism @xmath581 , which we denote by @xmath582 . a graph of subgroups of @xmath31 is completely determined by its family of vertex groups : suppose we are given a family @xmath583 of subgroups @xmath562 ( @xmath439 ) which is _ compatible _ in the sense that @xmath584 then there is a unique graph of subgroups of @xmath31 with vertex groups @xmath585 , which we continue to denote by the same symbol @xmath511 . ( in particular , we may write @xmath586 . ) let @xmath7 be a normal subgroup of @xmath54 . then the collection @xmath587 of normal subgroups is compatible ; moreover , there is a unique morphism @xmath588 such that the diagram @xmath589_{\pi_{\scrh}}\ar[dr]^{\text{natural projection } } \\ \pi_1(\scrg/\scrh ) \ar[rr ] & & g / h}\ ] ] commutes . [ prop : commoncover ] let @xmath590 be a compatible collection of finite index normal subgroups of @xmath31 . then there exists a graph of groups @xmath552 based on a graph @xmath544 and a morphism @xmath551 with the following properties : 1 . @xmath396 has finite degree ; 2 . for each @xmath591 , @xmath592 is an isomorphism @xmath593 ; 3 . for each @xmath547 , @xmath594 is an isomorphism @xmath595 . this is sketched in ( * ? ? ? * theorem 2.2 ) ; for the convenience of the reader , we give a proof . we employ covering space theory . thus , let @xmath479 be a graph of based cw - complexes based on @xmath362 , with topological realization @xmath505 , such that @xmath596 , as described in section [ sec : graphs of cw - complexes ] . the collections @xmath597 and @xmath598 of subgroups of the vertex respectively edge groups of @xmath31 give rise to collections @xmath599 of coverings of the vertex respectively edge spaces of @xmath479 which are also compatible in the following sense : for each @xmath435 we have @xmath600 , and for each @xmath435 and for each component @xmath601 of @xmath602 there exists an embedding @xmath603 with image @xmath601 which covers @xmath491 . we will show that the various covering spaces can be glued together to give a connected covering @xmath539 such that for each @xmath439 and each component @xmath604 of @xmath542 the covering @xmath605 corresponds to @xmath606 , and for each @xmath435 and each component @xmath607 of @xmath543 the covering @xmath608 corresponds to @xmath609 . then every morphism of graph of groups @xmath610 associated to @xmath247 as in section [ sec : graphs of cw - complexes ] has the required properties . we proceed by induction on the number of topological edges of @xmath362 . the case where @xmath362 has no topological edge ( and hence only one vertex ) being obvious , let us first consider the case that the graph @xmath362 has a single topological edge @xmath374 and hence either one or two vertices . first assume that @xmath362 has a single vertex @xmath74 . in that case @xmath611 and @xmath612 have the same number @xmath20 of components . matching up these components in pairs and connecting them via @xmath20 copies of @xmath607 , the resulting complex @xmath540 has the desired property . now assume @xmath362 has two distinct vertices @xmath457 and @xmath613 such that @xmath614 and @xmath615 . let @xmath616 be the number of components of @xmath617 , for @xmath618 . we get the cover @xmath619 by taking @xmath620 copies of @xmath621 and @xmath622 copies of @xmath623 and by connecting each copy of a component of @xmath624 with a copy of a component of @xmath625 via a copy of @xmath607 . we now turn to the case of graphs with more than one topological edge . since @xmath362 is connected , either there is an edge @xmath372 such that the graph @xmath626 obtained by deleting the edge @xmath372 from @xmath362 is connected , or there is an edge @xmath372 such that one of the components of @xmath627 consists of a single vertex with no topological edge . first consider the case that we have an edge @xmath372 such that @xmath628 is connected . we denote by @xmath629 the subgraph of based cw - complexes of @xmath479 with underlying graph @xmath375 , and write @xmath630 . we identify @xmath631 with a topological subspace of @xmath632 in the natural way . by inductive hypothesis there exists a covering @xmath633 with the desired properties . we now consider the graph of cw - complexes @xmath634 whose underlying graph has a single vertex @xmath430 and a single topological edge @xmath635 , vertex space @xmath636 , edge spaces @xmath637 , and edge maps @xmath638 , @xmath639 given by @xmath640 respectively @xmath641 . writing @xmath642 we note that @xmath643 . clearly the finite index subgroup @xmath644 of @xmath645 forms a compatible collection of subgroups ( of the graphs of groups associated to @xmath634 ) . since we already showed that the conclusion of the proposition holds for graphs with one edge we now get a cover @xmath646 which has the desired properties . a similar argument deals with the case that we have an edge @xmath372 such that one component of @xmath647 consists of a single vertex with no topological edge . 1 assume that the indices of the normal subgroups @xmath648 are powers of @xmath0 . in that case the indices of the normal subgroups @xmath649 are powers of @xmath0 . with the notation introduced in the proof above we have @xmath650}{[g_e : h_e ] } \text { and } n_i=\frac{[g_{v_i}:h_{v_i}]}{[g_e : h_e ] } \text { for $ i=1,2$.}\ ] ] in particular @xmath20 , @xmath651 and @xmath652 are powers of @xmath0 . it is now clear that the degree of the cover @xmath539 ( and hence the degree of the associated morphism @xmath610 of graphs of groups ) constructed in the proof is a power of @xmath0 . 2 . there is no guarantee that the resulting cover constructed in the proof will be regular ( i.e. , that @xmath653 will be a normal subgroup of @xmath30 ) . let @xmath31 be a graph of groups with underlying graph @xmath362 . we say that a collection @xmath654 of filtrations @xmath655 of @xmath35 ( one for each vertex @xmath74 ) is a _ filtration _ of @xmath31 if for each @xmath65 , the collection @xmath656 of subgroups of the vertex groups is compatible . let @xmath654 be a filtration of @xmath31 . then for each @xmath435 we obtain a filtration @xmath657 of @xmath55 where @xmath658 for every @xmath65 . note that @xmath489 is a morphism @xmath659 of filtered groups , and @xmath660 for all @xmath435 . suppose that @xmath111 is normal ( i.e. , each @xmath63 is normal ) . then for each @xmath65 we have three graphs of groups with same underlying graph @xmath362 associated to @xmath31 : 1 . the graph of subgroups @xmath661 of @xmath31 ; 2 . the quotient graph of groups @xmath662 associated to @xmath661 , with vertex groups @xmath70 , edge groups @xmath663 and edge morphisms @xmath664 induced by @xmath489 ; and 3 . the graph @xmath665 , whose vertex and edge groups are the @xmath20th layers @xmath666 of @xmath63 respectively @xmath667 of @xmath668 , and with the morphisms @xmath669 induced by @xmath489 as the edge morphisms . we say that another filtration @xmath670 of @xmath31 is a _ compatible stretching _ of @xmath111 if the filtrations @xmath671 , as @xmath74 ranges over @xmath517 , form a compatible collection of stretchings of the @xmath63 as defined in section [ subsec : filtrations ] above . note that in this situation , for each @xmath65 , either every vertex group of @xmath672 is trivial , or @xmath673 for some @xmath260 . given a property @xmath2 of filtrations , we say that @xmath111 has the property @xmath2 if @xmath63 has @xmath2 for each @xmath439 . for example , we say that @xmath111 is separating if for each @xmath439 , the filtration @xmath63 of @xmath35 is separating : @xmath674 . we also say that _ @xmath111 separates the edge groups of @xmath31 _ if for each @xmath435 and @xmath520 , the filtration @xmath63 of @xmath35 separates @xmath675 : @xmath676 . in the next lemmas , let @xmath677 be a morphism of graph of groups . [ lem : morphisms and filtrations , 1 ] let @xmath678 be a compatible collection of normal subgroups of @xmath511 . then @xmath679 is a compatible collection of normal subgroups of @xmath31 . the proof of this lemma is a routine verification , using the diagram ( left to the reader ) . in the rest of this section let @xmath680 be a normal filtration of @xmath511 . then @xmath681 is a normal filtration of @xmath31 , by lemma [ lem : morphisms and filtrations , 1 ] , and if @xmath680 is central , a @xmath0-filtration , or @xmath0-potent , then clearly @xmath682 also has the respective property . [ lem : separating , 1 ] suppose each @xmath683 is injective . then 1 . if @xmath680 is separating , then @xmath682 is separating ; 2 . if each @xmath684 is bijective and @xmath680 separates the edge groups of @xmath511 , then @xmath682 separates the edge groups of @xmath31 ; 3 . if each @xmath684 maps @xmath55 isomorphically onto the first group @xmath685 in the filtration @xmath686 of @xmath687 , and @xmath680 separates the edge groups of @xmath511 , then @xmath682 separates the edge groups of @xmath31 . if @xmath680 is separating , then @xmath688 for each @xmath74 , hence @xmath682 is separating . for each @xmath372 and @xmath520 , denoting by @xmath689 the inverse of @xmath683 , by we also have @xmath690 assume now that each @xmath684 is bijective and @xmath680 separates the edge groups of @xmath511 . then for all @xmath372 and @xmath520 , we have @xmath691 hence @xmath682 separates the edge groups of @xmath31 . this shows ( 2 ) . part ( 3 ) is proved in a similar way . in the next lemmas we fix @xmath435 , and write @xmath520 , @xmath692 , @xmath693 . [ lem : separating , 2 ] suppose that for each @xmath65 , the morphism @xmath684 restricts to an isomorphism @xmath694 , and @xmath695 is a uniformly @xmath0-potent filtration of @xmath696 . then @xmath697 is a uniformly @xmath0-potent filtration of @xmath675 . note that the hypothesis that @xmath695 be uniformly @xmath0-potent is equivalent to the filtration @xmath698 of @xmath699 being uniformly @xmath0-potent , and that it is enough to show that @xmath700 is a uniformly @xmath0-potent filtration of @xmath55 . now by assumption , @xmath684 induces isomorphisms @xmath701 , for each @xmath65 . hence for each @xmath65 , @xmath358 yields a surjective group morphism @xmath702 whose kernel equals @xmath703 , as required . the following two lemmas are not used until section [ sec : mod p homology graph ] , and may be skipped upon first reading . [ lem : separating , 3 ] suppose @xmath683 is injective and @xmath684 is bijective , and @xmath260 such that @xmath704 then @xmath705 note that @xmath706 and , since @xmath684 is bijective : @xmath707 hence for @xmath708 : @xmath709 since @xmath683 is injective , this yields @xmath710 the lemma follows , since @xmath711 is the fastest descending central @xmath0-filtration of @xmath675 . [ lem : separating , 4 ] suppose @xmath55 is abelian , @xmath683 is injective , @xmath684 maps @xmath55 isomorphically onto the first group @xmath712 in the filtration @xmath713 of @xmath699 , and @xmath714 such that @xmath715 then @xmath716 as in the proof of the previous lemma , it suffices to show that @xmath717 let @xmath65 . by hypothesis @xmath718 since @xmath699 is abelian and @xmath719 , we have @xmath720 and thus @xmath721 as required . let @xmath31 be a graph of groups with fundamental group @xmath54 . let @xmath401 be an orientation of the underlying graph @xmath362 of @xmath31 . then for each @xmath3-module @xmath722 , there is an exact sequence @xmath723 where each morphism @xmath724 occurring in this sequence is induced by the inclusion @xmath725 , and each composition @xmath726 where @xmath727 denotes the natural projection onto @xmath728 , is given by @xmath729 ( respectively @xmath730 ) if @xmath520 ( respectively @xmath731 ) , and is @xmath732 otherwise . @xcite or ( * ? ? ? * section ii.2.8 ) . ) dually , there is also a mayer - vietoris sequence in cohomology , which we wo nt formulate explicitly here . in this chapter we consider two types of embedding theorems for @xmath0-groups : one for amalgamating ( filtered ) @xmath0-groups , and one for extending partial automorphisms between subgroups of a given @xmath0-group to inner automorphisms of a larger @xmath0-group . these results will be basic for establishing in the next chapter that certain fundamental groups of graphs of groups are ( virtually ) residually @xmath0 . an _ amalgam _ of groups consists of a pair @xmath733 of groups with a common subgroup @xmath734 ; notation : @xmath735 . an _ embedding _ of an amalgam @xmath735 into a group @xmath736 is a pair @xmath737 consisting of injective group morphisms @xmath738 , @xmath739 such that @xmath740 . ( figure [ fig : embedding ] . ) @xmath741^{\alpha } & & h \ar@{.>}[ul]_{\beta}\\ & \ar[ul ] u \ar[ur ] & } \ ] ] such an embedding of @xmath735 into @xmath736 is said to be _ strong _ if in addition @xmath742 . the following is a variant of a theorem of higman @xcite . the equivalences of ( 1)(4 ) may also be deduced from ( * ? ? ? * theorem 2 ) ; for our application in section [ sec : a reduction theorem ] , we are mostly interested in the addendum . [ thm : higman ] let @xmath735 be an amalgam of @xmath0-groups . the following are equivalent : 1 . @xmath735 is strongly embeddable into a @xmath0-group . 2 . @xmath735 is embeddable into a @xmath0-group . 3 . there exist finite - length chief filtrations of @xmath3 and @xmath7 inducing the same filtration of @xmath734 . 4 . there exist finite - length central @xmath0-filtrations @xmath111 of @xmath3 and @xmath125 of @xmath7 inducing the same filtration of @xmath734 . moreover , if we are given @xmath111 and @xmath125 as in , then there exists a strong embedding @xmath737 of @xmath735 into a @xmath0-group @xmath736 and a central @xmath0-filtration @xmath743 of @xmath736 of finite length which induces @xmath744 on @xmath745 and @xmath746 on @xmath747 . note that the implication `` ( 1 ) @xmath332 ( 2 ) '' is trivial . if @xmath736 is a @xmath0-group containing both @xmath3 and @xmath7 as subgroups , then any chief filtration of @xmath736 induces a chief filtration on @xmath3 and a chief filtration on @xmath7 . this shows `` ( 2 ) @xmath332 ( 3 ) '' , and `` ( 3 ) @xmath332 ( 4 ) '' follows from the fact that each chief filtration of a @xmath0-group is a central @xmath0-filtration . hence it remains to give the proof of `` ( 4 ) @xmath332 ( 1 ) '' in the stronger formulation given in the `` moreover '' part . the class of _ elementary abelian _ @xmath0-groups has the ( strong ) amalgamation property , i.e. : every amalgam of elementary abelian @xmath0-groups @xmath735 always admits a strong embedding into an elementary abelian @xmath0-group @xmath736 . somewhat more generally , let @xmath3 and @xmath7 be abelian groups ( written additively ) , and let @xmath748 , @xmath749 be injective morphisms from a group @xmath734 to @xmath3 respectively @xmath7 . then @xmath750 is a subgroup of @xmath751 , and @xmath752 is called the fiber sum of @xmath3 and @xmath7 over @xmath734 ( via @xmath119 , @xmath90 ) . it is easy to check that the natural morphisms @xmath753 and @xmath754 are injective . the fiber sum @xmath755 is the solution to an universal problem : it is the push - out , in the category of abelian groups , of the diagram @xmath756 . ( in connection with this construction it is interesting to note that no varieties of groups other than the variety of all groups and varieties of abelian groups are known to have the amalgamation property ; cf . @xcite . ) let now @xmath735 be an amalgam of @xmath0-groups and @xmath757 and @xmath758 be central @xmath0-filtrations of @xmath3 respectively @xmath7 as in . for showing the rest of theorem [ thm : higman ] , after refining @xmath111 and @xmath125 suitably , we may assume that @xmath759 . ( lemma [ lem : refine ] . ) in this case , we actually have a more precise result : [ prop : higman , more precise ] let @xmath735 be an amalgam of @xmath0-groups , and suppose @xmath111 and @xmath125 are finite - length central @xmath0-filtrations of @xmath3 respectively @xmath7 which intersect to the same filtration of @xmath734 . then there exists a @xmath0-group @xmath736 containing both @xmath3 and @xmath7 as subgroups , with @xmath760 , a finite - length central @xmath0-filtration @xmath761 of @xmath736 , as well as compatible stretchings @xmath149 of @xmath111 respectively @xmath762 of @xmath125 , with the following properties : * @xmath763 and @xmath764 ; * for each @xmath765 , identifying @xmath766 and @xmath767 with subgroups of @xmath768 in the natural way , and similarly @xmath769 with subgroups of @xmath766 and @xmath767 , where @xmath770 , we have @xmath771 hence @xmath768 embeds @xmath772 in a natural way . the remainder of this section is occupied by the proof of this proposition . the main idea of the argument is to apply the fiber sum construction inductively to the layers of the given filtrations @xmath111 and @xmath125 . before we can carry this out , we need some preliminary observations on group algebras , wreath products , and a construction of central filtrations on semidirect products . let @xmath51 be a commutative ring and @xmath3 be a group , and let @xmath773 $ ] be the group algebra of @xmath3 over @xmath51 . ( as a general reference for group algebras we use @xcite . ) the _ augmentation ideal _ of @xmath773 $ ] is the kernel @xmath774)$ ] of the @xmath51-algebra morphism @xmath773\to r$ ] given by @xmath775 for every @xmath4 . ( here and below , `` ideal '' will always mean `` two - sided ideal '' unless noted otherwise . ) it is easy to see ( cf . * lemma 3.1.1 ) ) that @xmath776(1-g).\ ] ] note that if @xmath51 has characteristic @xmath0 , then @xmath773=r\otimes_{\f_p}\f_p[g]$ ] and hence @xmath777)^n = r\otimes_{\f_p}\omega(\f_p[g])^n$ ] for every @xmath20 . given an ideal @xmath778 of @xmath773 $ ] we say that a descending sequence @xmath779 of ideals of @xmath773 $ ] satisfying @xmath780 and @xmath781 for all @xmath106 , @xmath20 is a _ filtration _ of @xmath778 . every strongly central filtration @xmath782 of @xmath3 gives rise to a filtration @xmath783 of the ideal @xmath784 as follows : for each @xmath4 define @xmath785 to be the largest @xmath786 for which @xmath787 , where we set @xmath788 . for each @xmath65 let @xmath789 be the left @xmath51-submodule of @xmath773 $ ] spanned by all products of the form @xmath790 for some @xmath106 with @xmath791 . it is easily seen that @xmath783 is a filtration of @xmath784 ; cf . conversely , let @xmath792 be a filtration of @xmath784 . then we associate to @xmath783 a filtration @xmath180 of @xmath3 by setting @xmath793 if @xmath51 has characteristic @xmath0 , then @xmath180 is a dimensional @xmath0-filtration of @xmath3 ; cf . * lemma 3.3.2 ) . in particular , taking @xmath794 for every @xmath20 , we obtain a corresponding dimensional @xmath0-filtration of @xmath3 . in fact , this filtration turns out to be exactly the lower dimensional @xmath0-filtration @xmath204 of @xmath3 : [ thm : jennings ] suppose @xmath51 is of characteristic @xmath0 and @xmath3 is a @xmath0-group . then @xmath795 moreover , the filtration of @xmath784 determined by @xmath204 is precisely @xmath796 , the powers of the augmentation ideal . the ideal @xmath784 is nilpotent of nilpotency class @xmath797 this theorem is due to jennings @xcite ; for a proof see ( * ? ? ? * theorems 3.3.6 and 3.3.7 ) . the _ nilpotency class _ of an ideal @xmath778 of @xmath773 $ ] is the smallest @xmath533 , if it exists , such that @xmath798 . by the theorem , the nilpotency class of @xmath784 is an upper bound on the lower @xmath0-length ( and hence the nilpotency class ) of @xmath3 . we denote by @xmath799 the left annihilator of @xmath784 ( a left ideal of @xmath773 $ ] ) . note that if @xmath800 then @xmath801 contains @xmath802 . if @xmath3 is finite we define @xmath803.\ ] ] note that @xmath804 for all @xmath4 , hence @xmath805 is central in @xmath773 $ ] . it is easy to see that if @xmath3 is finite , then @xmath806 ; cf . * lemma 3.1.2 ) . the following observation is standard ( cf . @xcite ) : suppose @xmath3 is a @xmath0-group , and @xmath51 is a field of characteristic @xmath0 . then every non - zero ideal @xmath778 of @xmath773 $ ] contains @xmath805 . let @xmath807 be an ideal of @xmath773 $ ] . the ideal @xmath784 is nilpotent , so we can choose an integer @xmath808 maximal with @xmath809 . then clearly @xmath810 and thus @xmath811 since @xmath51 is a field . applying this lemma to the prime field @xmath76 in place of @xmath51 and using that @xmath773=r\otimes_{\f_p } \f_p[g]$ ] yields : [ cor : hatg ] suppose @xmath3 is a @xmath0-group and @xmath812 . then @xmath813 where @xmath343 denotes the nilpotency class of @xmath784 . wreath products play an important role in the proof of the main result of @xcite , and so naturally also in the proof of theorem [ thm : higman ] . we recall the definition : let @xmath538 , @xmath7 be groups . we turn the set @xmath814 of maps @xmath815 into a group under the coordinate - wise operations . we have a right action @xmath816 of @xmath7 on @xmath814 , given by @xmath817 the ( unrestricted ) _ wreath product _ of @xmath538 and @xmath7 is a semidirect product @xmath818 ; its underlying set is @xmath819 , with group operation @xmath820 the group @xmath814 is called the _ base group _ of @xmath821 . both @xmath814 and the group @xmath7 can be naturally identified with subgroups of @xmath821 . ( note that then for every @xmath235 and @xmath822 the conjugate @xmath823 of @xmath824 by @xmath825 is conveniently given by @xmath826 . ) if @xmath14 is a subgroup ( respectively normal subgroup ) of @xmath7 , then @xmath827 is a subgroup ( respectively normal subgroup ) of @xmath821 . note also that if @xmath538 is a subgroup of a group @xmath362 , then @xmath821 is a subgroup of @xmath828 . the wreath product @xmath821 contains a copy of every extension of @xmath7 by @xmath538 ( theorem of kaluzin - krasner ) : in fact , given a group morphism @xmath829 ( not necessarily surjective ) with kernel @xmath538 there always exists a map @xmath830 such that the map @xmath831 is an injective group morphism ( cf . the map @xmath832 is obtained as follows : one first chooses a _ countermap _ to @xmath833 , that is , a map @xmath834 ( not necessarily a group morphism ) such that @xmath835 or , in other words : @xmath836 such countermaps always exist : choose a system @xmath837 of right coset representatives of the subgroup @xmath838 of @xmath7 , and for @xmath235 denote the unique @xmath839 such that @xmath840 by @xmath841 . for every @xmath235 choose @xmath842 such that @xmath843 , and for each @xmath839 choose an arbitrary @xmath844 . then @xmath845 defines a countermap to @xmath833 . ( so for example , if @xmath833 is surjective then every right inverse of @xmath833 is a countermap to @xmath833 . ) now set @xmath846 note that @xmath847 indeed lies in @xmath848 , so @xmath849 is an element of @xmath814 . if we want to stress the dependence of @xmath832 on the choice of countermap @xmath850 , we write @xmath851 instead of @xmath849 . we call an embedding @xmath852 of the form @xmath853 a _ standard embedding _ of @xmath214 into @xmath821 . if in addition @xmath538 is given as a subgroup of a group @xmath362 , we also speak of the composition @xmath854 of a standard embedding @xmath852 with the natural inclusion @xmath855 as a _ standard embedding _ of @xmath214 into @xmath828 . in the following let @xmath856 be the wreath product of the groups @xmath538 and @xmath7 . it is well - known ( cf . @xcite ) that if @xmath857 and @xmath7 is finite , then @xmath736 is nilpotent if and only if there is a prime @xmath0 such that @xmath538 is a nilpotent of finite @xmath0-power exponent and @xmath7 is a @xmath0-group . we are mostly interested in the case where the group @xmath7 is a @xmath0-group , and @xmath538 is an elementary abelian @xmath0-group , and hence may be construed as the additive group of a finite - dimensional @xmath76-algebra . somewhat more generally , _ from now on we assume that @xmath538 is the additive group of a commutative ring @xmath51 and @xmath7 is finite . _ below we simply denote @xmath538 by @xmath51 . we now can and will canonically identify the additive group of the group algebra @xmath858 $ ] of @xmath7 over @xmath51 with the base group @xmath859 of the wreath product @xmath860 . after this identification , the action of @xmath7 on @xmath859 is related to the action of @xmath7 by left multiplication on the group ring @xmath858 $ ] via @xmath861 $ , $ h\in h$.}\ ] ] ( to avoid confusion between the group operation in @xmath862 and the multiplication in the ring @xmath858 $ ] , we will always use a dot @xmath863 to denote the latter . we also reserve @xmath824 , possibly with decorations , to denote elements of @xmath858 $ ] and @xmath825 for the elements of the subgroup @xmath7 of @xmath736 . ) we therefore have the useful identity @xmath864 } [ f , h ] = ( h^{-1}-1)\cdot f \qquad ( f\in r[h ] , h\in h),\ ] ] where the commutator @xmath865=f^{-1}h^{-1}fh$ ] is computed in @xmath736 and the right - hand side in the group algebra @xmath858 $ ] . we let @xmath247 be the natural map @xmath866 $ ] , which is not a group morphism in general : in fact , for elements @xmath867 and @xmath868 of @xmath736 ( where @xmath869 , @xmath870 $ ] ) , we have @xmath871 for @xmath872 $ ] , @xmath235 , @xmath873 we obtain @xmath874 and for later use we also record that if in addition @xmath875 $ ] then @xmath876 multiplication @xmath877 in @xmath858 $ ] by a fixed element @xmath878 extends uniquely to an endomorphism of @xmath736 which is the identity on @xmath7 . we also note : [ lem : fully inv ] let @xmath3 be a fully invariant subgroup of @xmath736 . then @xmath879 $ ] is a left ideal of @xmath858 $ ] , with @xmath879=\pi(g)$ ] . moreover , if @xmath837 is a generating set for @xmath3 , then @xmath880 generates the left ideal @xmath879 $ ] . clearly @xmath879 $ ] is an additive subgroup of @xmath858 $ ] contained in @xmath881 . the subgroup @xmath3 of @xmath736 being fully invariant implies that @xmath882)$ ] and thus @xmath879=\pi(g)$ ] . since @xmath3 is fully invariant , @xmath879 $ ] is also closed under multiplication by elements from @xmath51 , and if @xmath883 $ ] and @xmath235 , then @xmath884 since @xmath3 is normal . hence @xmath879 $ ] is a left ideal of @xmath858 $ ] . now let @xmath837 be a generating set for @xmath3 . then @xmath881 clearly contains the left ideal @xmath885\,\pi(s)$ ] of @xmath858 $ ] generated by @xmath880 , and the identities and yield @xmath886 . the following theorem due to buckley @xcite shows that there is a close relationship between the lower central series of the wreath product @xmath860 and the powers of the ideal @xmath887)$ ] of the group ring @xmath858 $ ] . for every @xmath20 we have @xmath888 = \omega^n.\ ] ] since the proof of the theorem above given in @xcite is somewhat roundabout ( it is deduced from properties of polynomial functions on groups ) , we give a direct proof of a refinement valid in the situation of interest to us , namely , where @xmath812 . _ from now until the end of this subsection we therefore assume that @xmath51 has characteristic @xmath0 . _ [ prop : buckley ] for all @xmath20 : @xmath889 = \gamma^p_{n+1}(w)\cap r[h ] = \gamma_{n+1}(w)\cap r[h ] = \omega^n.\ ] ] since @xmath890 it is enough to show that @xmath891 \leq \omega^n \leq \gamma_{n+1}(w)\cap r[h].\ ] ] we proceed by induction on @xmath20 , the case @xmath449 holding vacuously . suppose we have established for some @xmath20 . then for every @xmath892 and @xmath235 , by we have @xmath893\in [ \gamma_{n+1}(w),w]\leq\gamma_{n+2}(w).\ ] ] lemma [ lem : fully inv ] now yields @xmath894 $ ] . for the inclusion on the left in , recall first that the subgroup @xmath895\ ] ] of @xmath736 is fully invariant . thus by lemma [ lem : fully inv ] it suffices to show the following claim , where we abbreviate @xmath896 : suppose @xmath897 and @xmath898 $ ] are such that @xmath899 then @xmath900 \big ) \in \omega^{n+1}.\ ] ] for the first part , we note that , writing @xmath901 , we have @xmath902 $ ] by lemma [ lem : fully inv ] , and hence @xmath903 by inductive hypothesis . this yields @xmath904 and thus @xmath905 . we obtain @xmath906 using theorem [ thm : jennings ] . for the second part , by standard commutator formulas ( cf . , e.g. , ( * ? ? ? * section 0.1 ) ) we have @xmath907 = [ h_i , h_jf_j]^{f_i}[f_i , h_if_j ] = [ h_i , f_j]^{f_i } [ h_i , h_j]^{f_jf_i } [ f_i , h_j]^{f_j } \\ = [ h_i , f_j ] [ h_i , h_j]^{f_jf_i } [ f_i , h_j],\end{gathered}\ ] ] using that @xmath908=1 $ ] as well as @xmath909^{f_i}=[h_i , f_j]$ ] and @xmath910^{f_j}=[f_i , h_j]$ ] ( since @xmath909,[f_i , h_j]\in r[h]$ ] ) . by this yields @xmath911 \big ) = ( 1-[h_j , h_i])\cdot ( f_i+f_j ) + [ h_j , h_i]\cdot ( 1-h_i^{-1})\cdot f_j - ( 1-h_j^{-1})\cdot f_i.\ ] ] by theorem [ thm : jennings ] we have @xmath912 , @xmath913 , and @xmath914\in\omega^{n+2}$ ] ; and by inductive hypothesis @xmath915 and @xmath916 . this implies the second part of our claim . combining this proposition with lemma [ lem : lower central for semidirect products ] and jennings theorem ( theorem [ thm : jennings ] above ) we obtain : suppose @xmath7 is a @xmath0-group . then @xmath860 is nilpotent , and the nilpotency class , lower @xmath0-length , and lower dimensional @xmath0-length of @xmath736 all equal @xmath917 in the proof of the main result of this section we need : [ cor : buckley ] let @xmath214 be a group containing @xmath51 as a central subgroup , let @xmath829 be a group morphism to a @xmath0-group @xmath7 not necessarily surjective with kernel @xmath51 , and let @xmath918 be a corresponding standard embedding . then @xmath919=r\widehat{h}\ ] ] where @xmath343 is the nilpotency class of @xmath887)$ ] . let @xmath850 be a countermap to @xmath833 such that @xmath920 for all @xmath921 . since @xmath922 we have for all @xmath878 , @xmath235 : @xmath923 and hence @xmath924 $ ] . now use corollary [ cor : hatg ] and proposition [ prop : buckley ] . assume that the group @xmath223 is a semidirect product of its subgroups @xmath925 and @xmath7 , with @xmath224 . the following is easy to verify : [ lem : filtration of semidirect product ] let @xmath926 be a filtration of @xmath7 . then @xmath927 defines a filtration of @xmath223 . if @xmath926 is normal , then so is @xmath928 , and for every @xmath929 the natural morphism @xmath930 is an isomorphism . so if @xmath926 is a central filtration , or a central @xmath0-filtration , then @xmath928 also has the respective property . let now @xmath180 be a central filtration of @xmath3 , and put @xmath931 for each @xmath20 . note that @xmath932\leq b$ ] , hence @xmath933\leq b_{n+1}$ ] for each @xmath20 . assume also that a central finite - length filtration @xmath934 of @xmath7 is given . we can then combine the central filtration @xmath935 of @xmath925 with the central filtration @xmath928 of @xmath3 to a central filtration @xmath936 of @xmath3 . ( this construction of filtrations on semidirect products is used in the next subsection . ) we need yet another lemma , indicating how an embedding of an amalgam into a wreath product can be constructed inductively . for this , let @xmath735 be an amalgam of groups . an amalgam @xmath937 is called a _ subamalgam _ of @xmath735 if @xmath938 , @xmath939 and @xmath940 . the subamalgam @xmath937 of @xmath735 is called _ normal _ if @xmath538 is normal in @xmath3 and @xmath362 normal in @xmath7 . in this case , identifying @xmath941 with @xmath942 in the natural way , we can form the _ factor amalgam _ @xmath943 . the first part of the following lemma is ( * ? ? * lemma 2 ) : let @xmath937 be a normal subamalgam of the amalgam @xmath735 , and suppose @xmath378 is central in both @xmath3 and @xmath7 . if @xmath937 embeds into a group @xmath79 and @xmath943 embeds into a group @xmath14 , then @xmath735 embeds into @xmath944 . an analogous statement holds for strong embeddings in place of embeddings . suppose @xmath79 is a group containing both @xmath538 and @xmath362 as subgroups , and @xmath945 , @xmath946 are group morphisms with @xmath848 , @xmath947 and @xmath948 . then by ( * ? ? ? * lemma 2 ) there are standard embeddings @xmath949 and @xmath950 which form an embedding of @xmath735 into @xmath944 . this is already the first part of the lemma . however , for the statement concerning strong embeddings , we need a more detailed description of how @xmath951 and @xmath952 are constructed . first one chooses a countermap @xmath953 to @xmath954 . this yields a standard embedding @xmath955 one can show ( cf . * lemma 1 ) ) that there exists a countermap @xmath956 to @xmath945 such that the map @xmath957 defined by @xmath958 is constant on each right coset of @xmath959 in @xmath14 . similarly , there exists a countermap @xmath960 to @xmath946 such that the map @xmath961 defined by @xmath962 is constant on each right coset of @xmath959 in @xmath14 . this yields standard embeddings @xmath963 and @xmath964 in the manner described above . note that for @xmath965 and @xmath966 we have @xmath967 since @xmath378 is central in @xmath3 . similarly we have @xmath968 for all @xmath965 . hence the diagram in figure [ img : embeddings ] is commutative . @xmath969 & & y\wr k\ar[ul ] \\ & \ar[ul ] v\wr k\ar[ur ] & \\ g \ar[uu ] & & h \ar[uu]\\ & \ar[ul ] u \ar[uu ] \ar[ur ] & } \ ] ] to see the statement about strong embeddings , suppose that @xmath970 as subgroups of @xmath79 and @xmath971 . let @xmath4 , @xmath235 such that @xmath972 . then @xmath973 , hence @xmath974 and @xmath975 for all @xmath966 , and @xmath976 , hence @xmath977 for all @xmath966 . since @xmath378 is central in @xmath3 , this yields @xmath978 for all @xmath966 , and similarly @xmath979 this yields @xmath980 . with these tools at hand , we can now give the proof of proposition [ prop : higman , more precise ] . given a filtration @xmath981 of a group , we call the number of distinct non - trivial groups occurring among @xmath982 the _ essential length _ of @xmath111 . note that the essential length of @xmath111 is an invariant of the equivalence class of @xmath111 . the length of @xmath111 is never smaller than its essential length . let @xmath735 be an amalgam of @xmath0-groups , and let @xmath757 and @xmath758 be finite - length central @xmath0-filtrations of @xmath3 respectively @xmath7 with @xmath759 . we proceed by the maximum @xmath983 of the essential lengths of @xmath111 and @xmath125 to show that there exists a strong embedding of @xmath735 into a @xmath0-group @xmath736 and a central @xmath0-filtration of @xmath736 as well as compatible stretchings @xmath144 of @xmath111 and @xmath984 of @xmath125 satisfying conditions ( a1 ) and ( a2 ) in proposition [ prop : higman , more precise ] ( after identifying @xmath3 and @xmath7 with their images under the embedding of @xmath735 into @xmath736 ) . if @xmath985 , then there is nothing to show , so assume @xmath986 . let @xmath20 be the maximum of the lengths of @xmath111 and @xmath125 . then @xmath987 is central in @xmath3 and @xmath988 is central in @xmath7 , and both groups are elementary abelian @xmath0-groups ; their fiber sum @xmath79 over @xmath989 is an elementary abelian @xmath0-group , and @xmath970 after identifying @xmath538 and @xmath362 in the natural way with subgroups of @xmath79 . by induction and after replacing @xmath111 and @xmath125 by compatible stretchings if necessary , we may assume that we have group morphisms @xmath945 and @xmath946 such that @xmath990 as well as a central @xmath0-filtration @xmath991 of @xmath14 such that 1 . @xmath992 and @xmath993 for @xmath994 ; 2 . @xmath995 and @xmath996 for @xmath997 ; 3 . @xmath998 ( as subgroups of @xmath999 ) for @xmath1000 , where @xmath1001 . put @xmath1002 , and following the proof of the lemma above take standard embeddings @xmath1003 and @xmath1004 which form a strong embedding of @xmath735 into @xmath736 . now for every @xmath1005 , where @xmath1006 , define the normal subgroup @xmath1007 of @xmath736 , and for @xmath1008 , where @xmath1009 , set @xmath1010 then @xmath1011 is a central @xmath0-filtration of @xmath736 . ( see the remark following lemma [ lem : filtration of semidirect product ] . ) it remains to show that there are compatible stretchings @xmath144 of @xmath111 and @xmath984 of @xmath125 which , together with @xmath1012 , satisfy conditions ( a1 ) and ( a2 ) in proposition [ prop : higman , more precise ] . clearly @xmath1013 on the other hand , by the construction of @xmath1014 in the proof of the preceding lemma , the image of @xmath1014 is contained in the subgroup @xmath1015 of @xmath736 , and by corollary [ cor : buckley ] , @xmath1016 = x\widehat{k } = \alpha(x),\ ] ] hence @xmath1017 . therefore @xmath1018 thus the stretching @xmath1019 of @xmath111 defined by @xmath1020 has the property that @xmath1021 . similarly one argues that the stretching @xmath1022 of @xmath125 defined by @xmath1023 satisfies @xmath1024 ; note that @xmath144 and @xmath984 are compatible . this shows condition ( a1 ) . we also have , with @xmath1025)$ ] : @xmath1026 moreover @xmath1027 and similarly for @xmath1028 . property ( a2 ) follows . throughout this section we let @xmath3 be a @xmath0-group , and we let @xmath1029 ( @xmath1030 ) be partial automorphisms of @xmath3 , i.e. , isomorphisms between subgroups of @xmath3 . in this section we want to discuss the following question : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ under which conditions on the @xmath1031 is there a @xmath0-group @xmath7 which contains @xmath3 as a subgroup such that each @xmath1031 extends to an inner automorphism of @xmath7 ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this issue is of interest to us since the existence of @xmath7 as in the question above is equivalent to the iterated hnn extension @xmath1032 of @xmath3 to be residually @xmath0 . ( see lemma [ residual p criterion ] below . ) it is clear that in order to find such @xmath7 it suffices to produce a @xmath0-group @xmath357 which contains @xmath3 as a subgroup and for each @xmath258 an extension @xmath1033 of @xmath1031 to an automorphism of @xmath357 such that the subgroup @xmath214 of @xmath1034 generated by @xmath1035 has @xmath0-power order : in this case , the semidirect product @xmath1036 has the right property . note however that it is not enough to have , for each @xmath258 individually , an extension of @xmath1031 to a @xmath0-automorphism of some @xmath0-group containing @xmath3 as a subgroup : suppose @xmath1037 , and consider the automorphisms @xmath1038 and @xmath1039 of @xmath3 . then @xmath119 and @xmath90 both have order @xmath0 ; however @xmath119 and @xmath90 do not commute . since every @xmath0-subgroup of @xmath1040 is conjugate to a subgroup of the group @xmath1041 of upper unitriangular matrices ( see , e.g. , @xcite ) , and @xmath1042 is abelian , there are no extensions of @xmath119 , @xmath90 to automorphisms of a @xmath0-group containing @xmath3 as a subgroup and which generate a @xmath0-group . given a partial automorphism @xmath1043 of @xmath3 , we say that a filtration @xmath109 of @xmath3 is _ @xmath119-invariant _ if @xmath1044 for all @xmath20 ; in this case , if in addition @xmath111 is normal , then for each @xmath20 , @xmath119 induces an automorphism between the canonical images @xmath1045 and @xmath1046 of @xmath214 respectively @xmath925 in @xmath1047 , which we denote by @xmath1048 . the following proposition was shown in ( * ? ? ? * lemma 1.2 ) ( and later rediscovered in @xcite ) . it may be seen as an analogue for hnn extensions of higman s theorem @xcite discussed in the previous section ; the proof given in @xcite also employs wreath products . let @xmath1043 be a partial automorphism of @xmath3 . suppose that @xmath3 admits a @xmath119-invariant chief filtration @xmath121 such that @xmath1049 then @xmath3 embeds into a @xmath0-group @xmath7 such that @xmath119 extends to an inner automorphism of @xmath7 and @xmath111 is induced by a chief filtration of @xmath7 . this fact easily yields the following corollary , which generalizes a well - known characterization of unipotent subgroups of the general linear group over a field of characteristic @xmath0 ; cf . @xcite : [ cor : chatzidakis ] the following are equivalent : 1 . there exists a @xmath0-group @xmath7 which contains @xmath3 as a subgroup such that each @xmath1031 extends to an inner automorphism of @xmath7 . 2 . there exists a @xmath0-group @xmath357 which contains @xmath3 as a subgroup and extensions of @xmath1031 to automorphisms of @xmath357 which generate a @xmath0-subgroup of @xmath1034 . 3 . there exists a chief filtration @xmath1050 of @xmath3 which is @xmath1031-invariant for each @xmath258 and such that @xmath1051 the equivalence of ( 1 ) and ( 2 ) was already noted above . the implication ( 3 ) @xmath332 ( 1 ) follows from the proposition above by induction on @xmath338 . for the converse note that every chief filtration of a @xmath0-group containing @xmath3 as a subgroup induces a chief filtration of @xmath3 , and that every central filtration is invariant under inner automorphisms . criterion ( 3 ) may actually be relaxed slightly : [ lem : hnn ] suppose there exists a finite - length central filtration @xmath1052 of @xmath3 which is @xmath1031-invariant for each @xmath258 and such that for each @xmath1053 there is a @xmath0-group @xmath1054 containing @xmath1055 as a subgroup and for each @xmath258 an extension of the partial automorphism @xmath1056 of @xmath1055 induced by @xmath1031 to an inner automorphism of @xmath1054 . then there exists a @xmath0-group @xmath7 which contains @xmath3 as a subgroup such that each @xmath1031 extends to an inner automorphism of @xmath7 . for each @xmath1053 there is a chief filtration @xmath1057 of the @xmath1053th layer @xmath1055 of @xmath111 which is @xmath1058-invariant for each @xmath258 and such that @xmath1059 for each @xmath1053 let @xmath1060 be the natural epimorphism . then @xmath1061 lifts to a complete filtration @xmath1062 of @xmath1063 . the @xmath1064 combine to a complete filtration @xmath144 of @xmath3 : @xmath1065 this filtration refines @xmath111 and hence is central . moreover , for @xmath1066 , the @xmath1067th layer of @xmath1068 is @xmath1069 , whereas for @xmath1070 the @xmath1067th term of @xmath1068 is @xmath1071 , and @xmath1072 hence @xmath144 is a chief filtration satisfying condition ( 3 ) in corollary [ cor : chatzidakis ] . as an application of this lemma we obtain : [ cor : generate hnn ] suppose @xmath3 is abelian , and let @xmath7 be a subgroup of @xmath3 containing all @xmath1073 and @xmath1074 @xmath1030 . then the following are equivalent : 1 . there are a @xmath0-group @xmath1075 and extensions of each @xmath1031 to an inner automorphism of @xmath1076 . there are a @xmath0-group @xmath1077 and extensions of each @xmath1031 to an inner automorphism of @xmath357 . clear ( 2 ) implies ( 1 ) . suppose conversely that ( 1 ) holds . then @xmath1078 is a central filtration of @xmath3 satisfying the conditions of the previous lemma , which yields ( 2 ) . this corollary immediately implies : [ cor : generate hnn , 2 ] suppose @xmath3 is abelian , and let @xmath1079 @xmath1030 be partial automorphisms of an abelian @xmath0-group @xmath1080 . let @xmath1081 be a morphism which , for @xmath1030 , restricts to morphisms @xmath1082 and @xmath1083 making @xmath1084^{\varphi_i}\ar[d]^{\phi } & b_i \ar[d]^{\phi } \\ a_i ' \ar[r]^{\varphi_i ' } & b_i ' } \ ] ] commutative . if @xmath1085 is injective and @xmath1032 is residually @xmath0 , then @xmath1086 is residually @xmath0 . this fact will be used in the next chapter , however only in the trivial case where @xmath3 and @xmath1080 are elementary abelian @xmath0-groups . we would like to take the opportunity to stress that the analogue of the last corollary without the requirement that @xmath1085 be injective is false , as a simple example based on permutation matrices shows : let @xmath1087 , and consider @xmath1088 and its subgroups @xmath1089 , @xmath1090 . let @xmath1091 be the automorphism of @xmath3 given by @xmath1092 . then @xmath1091 has order @xmath1 , and @xmath1091 restricts to an isomorphism @xmath1043 . hence the hnn extension @xmath1093 of @xmath3 with associated subgroups @xmath214 , @xmath925 identified via @xmath119 is residually @xmath1 . now consider the morphism @xmath1094 given by @xmath1095 . then @xmath1085 restricts to isomorphisms @xmath1096 and @xmath1097 , and if we let @xmath1098 be the automorphism of @xmath1080 given by @xmath1099 , then @xmath1100 . hence @xmath1085 gives rise to a surjective morphism @xmath1101 however , @xmath219 is not residually @xmath1 , since @xmath1098 has order @xmath34 . this phenomenon forced us to include the condition that the morphisms @xmath328 arising in the definition of a @xmath0-potent filtration of a group ( cf . section [ sec : p - potent ] ) to be _ injective . _ we say that a filtration @xmath1102 of a graph of groups @xmath31 with underlying graph @xmath362 is _ @xmath0-excellent _ if 1 . @xmath111 is normal with @xmath1103<\infty$ ] for each @xmath439 and @xmath65 ; 2 . @xmath111 , construed as a complete filtration of @xmath1104 , is @xmath0-potent ; 3 . @xmath111 is separating ; 4 . @xmath111 separates the edge groups of @xmath31 ; and 5 . for each @xmath435 , the filtration @xmath1105 intersects to a uniformly @xmath0-potent filtration on the subgroup @xmath675 of @xmath1106 . the main result of this chapter is : [ thm : reduction theorem , 2 ] let @xmath31 be a graph of finitely generated groups admitting a @xmath0-excellent filtration . then @xmath30 is virtually residually @xmath0 . we prove this result in the last section of this chapter . after some general facts concerning residual properties of fundamental graphs of groups ( section [ sec : residual properties of graphs of groups ] ) we first establish a useful criterion for fundamental groups to be residually @xmath0 ( section [ sec : a reduction theorem ] ) and introduce an unfolding procedure on graphs of groups ( section [ sec : unfolding a graph of groups ] ) . following @xcite , we say that a property @xmath2 of groups is a _ root property _ if 1 . every subgroup of a group with property @xmath2 has property @xmath2 ; 2 . every direct product of two groups with property @xmath2 has property @xmath2 ; 3 . if @xmath3 is a group and @xmath1107 are such that @xmath100 , @xmath1108 have property @xmath2 , then @xmath14 contains a subgroup @xmath45 , _ normal in @xmath3 , _ such that @xmath1109 is @xmath2 . important root properties are : solvability , finiteness , and having order a power of a given prime @xmath0 ( see @xcite ) . ( nilpotence does not satisfy ( 3 ) and hence is not a root property . ) this notion is tailor - made for the proof of the following lemma to go through : ( * ? ? ? * lemma 1.5)[lem : gruenberg ] let @xmath41 be a group and @xmath1110 , and let @xmath2 be a root property . if the group @xmath25 is residually @xmath2 and @xmath1111 has property @xmath2 , then @xmath41 is residually @xmath2 . in the rest of this section we let @xmath2 be a root property , we let @xmath31 be a graph of groups based on a graph @xmath362 , and write @xmath54 . we are interested in conditions which force @xmath3 to be residually @xmath2 . in the rest of this section we assume that @xmath2 is residually satisfied by every finitely generated free group . _ of course , the case where @xmath2 is the property of having @xmath0-power order is the one of most interest for us , but the extra generality comes at little additional cost , and might be useful for further applications . probably the most interesting example for a root property @xmath2 of the type considered here which does not include the condition of finiteness is the property of being a solvable group . in connection with this it is worth noting that there is a distinction between `` residually solvable '' and `` residually finite solvable '' : there do exist finitely presentable solvable groups which are not residually finite , cf . @xcite . if each vertex group of @xmath31 is a @xmath0-group , then @xmath3 always has a free ( but not necessarily normal ) subgroup whose index is the least common multiple of the orders of the vertex groups ( hence a power of @xmath0 ) , cf . * ii.2.6 , lemmas 8 and 10 ) ; to have a free _ normal _ subgroup of @xmath0-power index is equivalent to @xmath3 being residually @xmath0 . more generally : [ residual p criterion ] suppose that each vertex group of @xmath31 is a finite @xmath2-group . then the following are equivalent : 1 . @xmath3 has a free normal subgroup @xmath25 such that @xmath1112 has @xmath2 . @xmath3 is residually @xmath2 . 3 . there is a morphism @xmath1113 to a @xmath2-group @xmath86 such that @xmath1114 is injective for every @xmath439 . the implication `` ( 1 ) @xmath332 ( 2 ) '' follows from lemma [ lem : gruenberg ] and the assumption that free groups are residually @xmath2 . suppose @xmath3 is residually @xmath2 . then for every @xmath439 and @xmath1115 there is a normal subgroup @xmath1116 of @xmath3 with @xmath1117 satisfying @xmath2 and @xmath1118 . let @xmath25 be the intersection of all @xmath1116 as @xmath453 ranges over @xmath1119 . then @xmath25 is a normal subgroup of @xmath3 with @xmath1112 satisfying @xmath2 , and the natural morphism @xmath1120 is injective on each @xmath35 . this shows `` ( 2 ) @xmath332 ( 3 ) . '' finally , given @xmath1113 as in ( 3 ) , the kernel @xmath25 of @xmath90 is a normal subgroup of @xmath3 with @xmath1112 satisfying @xmath2 , and @xmath25 is free by ( * ? ? ? * ii.2.6 , lemma 8) . this shows `` ( 3 ) @xmath332 ( 1 ) . '' the proof of the implications `` ( 3 ) @xmath332 ( 1 ) @xmath332 ( 2 ) '' did not use the finiteness of the @xmath35 , and the proof of `` ( 2 ) @xmath332 ( 3 ) @xmath332 ( 1 ) '' is valid without assuming that free groups are residually @xmath2 . [ cor : residually p quotient ] let @xmath7 be a normal subgroup of @xmath3 such that @xmath100 satisfies @xmath2 . then @xmath1121 is residually @xmath2 , where @xmath587 . the unique morphism @xmath588 making commutative is injective on each vertex group @xmath1122 of @xmath1123 , so @xmath1121 is residually @xmath2 by the implication `` ( 3 ) @xmath332 ( 2 ) '' in lemma [ residual p criterion ] ( the proof of which did nt need finiteness of the @xmath1122 ) . suppose the underlying graph @xmath362 of @xmath31 is a tree , and fix a vertex @xmath386 of @xmath362 . assume that for each edge @xmath372 of @xmath362 , the morphism @xmath1124 extends to an isomorphism @xmath1125 , such that @xmath1126 for all @xmath372 . for each vertex @xmath74 of @xmath362 we then have an isomorphism @xmath1127 given as the composition @xmath1128 where @xmath419 is the geodesic in @xmath362 from @xmath74 to @xmath386 . in particular , @xmath1129 , and @xmath1130 for every edge @xmath372 of @xmath362 . this yields a simple but useful fact about the iterated amalgamated product @xmath54 : [ lem : special amalgam ] there exists a unique morphism @xmath1131 such that for each vertex @xmath74 of @xmath362 , the diagram @xmath1132^{\varphi } & g_{v_0 } \\ g_v\ar[u ] \ar[ur]_{\varphi_v } & } \ ] ] commutes , where @xmath725 is the natural morphism . the following application ( not used later ) of lemmas [ residual p criterion ] and [ lem : special amalgam ] is implicit in @xcite . a finite group @xmath41 is homogeneous ( in the sense of model theory ) if every isomorphism between subgroups of @xmath41 is induced by an automorphism of @xmath41 . finite homogeneous groups have been completely classified @xcite ; for example , given a prime @xmath0 , the group of the form @xmath1133 ( for some @xmath1067 ) are homogeneous , and if @xmath0 is odd , then these are the only homogeneous @xmath0-groups . ( there are two non - abelian homogeneous @xmath34-groups . ) suppose @xmath362 is a tree , every vertex group @xmath35 is a finite homogeneous @xmath2-group , and @xmath1134 for all @xmath1135 . then @xmath3 has a free normal subgroup @xmath25 with @xmath1136 , hence @xmath3 is residually @xmath2 . the following is a generalization of the criterion for residual finiteness of fundamental groups of graphs of groups found in ( * ? ? ? * theorem 3.1 ) and , with a different proof , in @xcite ( see also ( * ? ? ? * section ii.2.6 ) ) . it unifies and generalizes several criteria for groups to be residually @xmath0 in the literature . ( for example , it can be used to easily deduce ( * ? ? ? * corollary 3.5 , theorems 4.2 and 4.3 ) . ) [ hempel criterion residually p ] suppose that the following hold : * for each @xmath528 and each @xmath1137 there is a compatible collection @xmath1138 of normal subgroups such that @xmath1139 and @xmath1121 is residually @xmath2 . * for all @xmath1140 and @xmath1141 , @xmath309 there exists a compatible collection @xmath1138 of normal subgroups such that @xmath1142 for @xmath309 and @xmath1121 is residually @xmath2 . then @xmath3 is residually @xmath2 . in this proof @xmath535 ranges over all compatible collections of normal subgroups such that @xmath1121 is residually @xmath2 . the natural morphisms @xmath1143 combine to a morphism @xmath1144 since @xmath357 is residually @xmath2 , it is enough to show that @xmath247 is injective . let @xmath5 be an element of @xmath3 , and represent @xmath453 by a path @xmath1145 in @xmath31 as in section [ sec : pi1 ] . then @xmath1146 is represented by the path @xmath1147 in @xmath1123 . moreover , if @xmath231 is reduced , then assumptions ( a ) and ( b ) yield that for some @xmath535 , the path @xmath1148 is reduced as well , in particular @xmath1149 . therefore we have @xmath1150 . we do not know whether in general we can simplify condition ( b ) by requiring this condition to hold only for a single edge at a time ( i.e. , for @xmath249 ) . next , we indicate a few situations where this is possible . the first one concerns the case where we can choose all subgroups in @xmath1151 to have finite index : [ cor : vertex and edge groups closed ] suppose that the following hold : * for each @xmath528 and each @xmath1137 there is a compatible collection @xmath1138 of finite index normal subgroups such that @xmath1139 and @xmath1121 is residually @xmath2 . * for all @xmath1152 and @xmath1153 there exists a compatible collection @xmath1138 of finite index normal subgroups such that @xmath1154 and @xmath1121 is residually @xmath2 . then @xmath3 is residually @xmath2 . moreover , every vertex and edge group of @xmath31 identified with a subgroup of @xmath3 in the natural way is closed in the pro-@xmath2 topology on @xmath3 . recall here that the pro-@xmath2 topology on a group @xmath41 is the topology with fundamental system of neighborhoods of @xmath163 given by the normal subgroups @xmath1110 with @xmath1111 a @xmath2-group . the first statement in this corollary is a consequence of proposition [ hempel criterion residually p ] and the following lemma , for which we let @xmath1155 @xmath618 be compatible collections of normal subgroups , and set @xmath1156 then @xmath1157 is a compatible collection of normal subgroups , and we have a natural group morphism @xmath1158 which is injective on vertex groups . [ lem : intersect h ] suppose @xmath1159 is finite for @xmath618 and every vertex @xmath74 . if @xmath1160 and @xmath1161 are residually @xmath2 , then so is @xmath1162 . suppose @xmath1160 and @xmath1161 are residually @xmath2 . since we have @xmath1163<\infty$ ] for @xmath618 and all @xmath74 , by lemma [ residual p criterion ] there exist group morphisms @xmath1164 to @xmath2-groups @xmath1165 such that @xmath1166 is injective , for @xmath618 and all @xmath74 . composing with the morphism , we obtain a group morphism @xmath1167 which is injective on vertex groups , and this shows ( again by lemma [ residual p criterion ] ) that @xmath1162 is residually @xmath2 . for showing the second statement in corollary [ cor : vertex and edge groups closed ] we use the following simple observation ( the proof of which we skip ) : [ lem : separate finite subgroups ] suppose @xmath41 is a residually @xmath2 group , and let @xmath1168 be a finite subgroup of @xmath41 . then for every @xmath1169 with @xmath1170 there exists a morphism @xmath1171 to a @xmath2-group such that @xmath1172 . as remarked , the first statement in this corollary follows from proposition [ hempel criterion residually p ] and lemma [ lem : intersect h ] . let @xmath1173 , @xmath528 . by the proof of proposition [ hempel criterion residually p ] there exists a compatible collection @xmath535 of finite index normal subgroups such that @xmath1121 is residually @xmath2 and @xmath1174 . since @xmath1175 is finite , there exists a morphism @xmath1176 to a @xmath2-group such that @xmath1177 , by lemma [ lem : separate finite subgroups ] . hence @xmath1178 is closed in @xmath3 . similarly ones sees that each edge group of @xmath31 is closed in @xmath3 . if all the edge groups @xmath55 of @xmath31 are trivial , then every collection @xmath590 of subgroups @xmath1179 of the vertex groups is compatible , and if @xmath1155 ( @xmath618 ) are collections of normal subgroups , then the morphism is injective ; hence if in addition the fundamental group of @xmath1180 is residually @xmath2 for @xmath618 , then so is the fundamental group of @xmath1181 . so in this special case , proposition [ hempel criterion residually p ] also simplifies : suppose that @xmath1182 for every @xmath435 . if @xmath31 satisfies condition in the previous proposition , then @xmath3 is residually @xmath2 . we record two ( presumably well - known ) consequences : [ cor : trivial edge groups ] the fundamental group of every graph of groups whose vertex groups are residually @xmath2 and whose edge groups are trivial is residually @xmath2 . by the previous corollary , it is enough to show that if @xmath528 is such that @xmath1178 is a @xmath2-group and @xmath1183 for all @xmath439 with @xmath1184 , and @xmath1182 for all @xmath435 , then @xmath3 is residually @xmath2 . however , this is immediate by the equivalence of ( 3 ) and ( 2 ) in lemma [ residual p criterion ] . in particular , the previous corollary shows that the free product of finitely many residually @xmath0 groups is residually @xmath0 , and ( combined with theorem [ thm : bass - serre ] ) that the free product of finitely many virtually residually @xmath0 groups is virtually residually @xmath0 . [ cor : crucial , preparation ] assume that all vertex groups of @xmath31 are residually @xmath2 , and suppose there exists a morphism @xmath1185 to a finite @xmath2-group @xmath86 which is injective when restricted to each subgroup @xmath675 of @xmath3 . then @xmath3 is residually @xmath2 . by theorem [ thm : bass - serre ] , the kernel @xmath25 of @xmath90 is the fundamental group of a graph of groups whose vertex groups are of the form @xmath1186 , and whose edge groups are of the form @xmath1187 . however , since @xmath1188 is injective , we have @xmath1189 for each @xmath372 . thus @xmath25 is the fundamental group of a graph of groups with residually @xmath2 vertex groups and trivial edge groups . hence @xmath25 is residually @xmath2 by corollary [ cor : trivial edge groups ] , and so @xmath3 is residually @xmath2 by lemma [ lem : gruenberg ] . finally , we note that in later sections , we will apply proposition [ hempel criterion residually p ] with the @xmath535 coming from a filtration of @xmath31 ; in this case , this proposition reads as follows : [ cor : hempel , cor ] let @xmath111 be a normal separating filtration of @xmath31 which separates the edge groups of @xmath31 , such that @xmath1190 is residually @xmath2 for every @xmath65 . then @xmath3 is residually @xmath2 . ( the material in the rest of this section is not used later . ) it is interesting to ask whether the conditions ( a ) and ( b ) in the previous proposition are necessary for @xmath3 to be residually @xmath2 . from corollary [ cor : residually p quotient ] we immediately obtain : [ lem : a ] if @xmath3 is residually @xmath2 , then @xmath31 satisfies . the situation for condition ( b ) is slightly less clear . here we can only show : [ lem : b ] suppose for each edge @xmath435 , either 1 . the subgroup @xmath675 of @xmath1106 is maximal with respect to being verbal with respect to a collection @xmath1191 of words ; or 2 . @xmath1192 is abelian and @xmath675 , @xmath1193 are proper subgroups of @xmath1106 , @xmath1192 , respectively . if @xmath3 is residually @xmath2 , then @xmath31 satisfies . in this context , recall the following definition : let @xmath736 be a collection of elements of a free group @xmath57 ( `` words '' ) . a group @xmath41 is _ verbal _ with respect to @xmath736 if every morphism @xmath1194 maps each @xmath1195 to @xmath163 . for example , a group @xmath41 is verbal with respect to @xmath1196 if and only if @xmath41 is abelian . we have the following generalization of a lemma from @xcite : [ lem : verbal ] suppose @xmath41 is a group and @xmath7 is a subgroup of @xmath41 . assume that @xmath41 is residually @xmath2 and @xmath7 is verbal with respect to a collection of words @xmath736 . then @xmath1197 is also verbal with respect to @xmath736 . for a normal subgroup @xmath14 of @xmath41 , every substitution of elements of @xmath1198 into a word from @xmath736 yields an element of @xmath14 ( since such an element maps to @xmath163 under @xmath1199 ) . since @xmath41 is residually @xmath2 , the intersection of all @xmath1200 such that @xmath1201 has property @xmath2 is trivial . hence @xmath1076 is verbal with respect to @xmath736 . we now prove lemma [ lem : b ] . so suppose @xmath54 is residually @xmath2 . let @xmath372 be an edge of @xmath362 , and @xmath1202 , where @xmath520 . in order to establish ( b ) it suffices to show : there is @xmath1203 such that @xmath1204 is a @xmath2-group and @xmath1205 . to see this note that if @xmath1206 are such that @xmath1207 is a @xmath2-group , @xmath618 , then @xmath1208 is also a @xmath2-group , and use corollary [ cor : residually p quotient ] assume first that @xmath675 is a maximal subgroup of @xmath1209 with respect to being verbal with respect to some @xmath1191 . considering @xmath1210 as a subgroup of @xmath3 , we see that @xmath1076 as defined in lemma [ lem : verbal ] is verbal with respect to @xmath736 , hence @xmath1211 . thus there is some @xmath1203 such that @xmath1204 has property @xmath2 and @xmath1212 . suppose next that @xmath1213 ( @xmath1214 ) is abelian and @xmath675 , @xmath1193 are proper subgroups of @xmath1209 , @xmath1215 , respectively . choose @xmath1216 . we may assume that our base point @xmath386 for @xmath529 agrees with @xmath430 . then @xmath1217=eg^{-1}\overline{e } h^{-1 } e g \overline{e } h$ ] is an element of @xmath3 which is not the identity . hence there is some @xmath1203 such that @xmath1204 has property @xmath2 and @xmath1218 . this yields @xmath1205 : otherwise @xmath1219 for some @xmath1220 , so @xmath1221 and hence @xmath1222 \equiv 1 \bmod k$ ] since @xmath1213 is abelian , a contradiction . in both cases we found @xmath1203 such that @xmath1204 is a @xmath2-group and @xmath1205 as required . given an orientable @xmath1-manifold @xmath25 with incompressible toroidal boundary , the fundamental groups of the boundary components of @xmath25 are maximal abelian subgroups of @xmath26 ; cf . lemma [ lem : maximal abelian ] below . this fact in combination with lemmas [ lem : a ] and [ lem : b ] , ( 1 ) shows that in proving that the fundamental group of a @xmath1-manifold is residually @xmath2 , we essentially have to use proposition [ hempel criterion residually p ] : if @xmath25 is @xmath1-manifold which is closed , orientable , and prime , with non - trivial jsj decomposition , then @xmath26 is residually @xmath2 if and only if the graph of groups @xmath31 associated to the jsj decomposition of @xmath25 satisfies conditions and . ( see section [ sec : jsj ] for the definition of the graph of groups @xmath31 . ) we now would like to apply the results of the previous chapter to establish a criterion for fundamental groups of certain graphs of groups to be residually @xmath0 . this is theorem [ thm : reduction theorem ] at the end of this section . before we can formulate this theorem , we need to introduce a generalization of the fiber sum construction , which is then used to define the `` partial abelianization '' of a fundamental group of a graph of abelian groups . let @xmath151 be a graph of abelian groups , and let @xmath401 be an orientation of its underlying graph @xmath362 . the injective morphisms @xmath1223 combine to a natural morphism @xmath1224 whose cokernel we call the _ colimit _ @xmath1225 of @xmath31 . ( indeed , @xmath1225 may be seen as the colimit , in the sense of category theory , of a certain diagram of abelian groups . ) the colimit of @xmath31 is determined , up to isomorphism , by the following universal property : for every abelian group @xmath7 and every collection @xmath1226 of morphisms with the property that for each edge @xmath408 , the diagram in figure [ fig : colimit ] commutes , there exists a unique morphism @xmath1227 such that @xmath1228 for all @xmath439 , where @xmath1229 is the natural morphism . also note that @xmath1225 does not depend on the choice of orientation @xmath401 of @xmath362 . @xmath1230^\psi & \\ g_{t(e ) } \ar[ur]_{\iota_{t(e ) } } \ar@/^/[uur]^{\psi_{t(e ) } } & & g_{t(\ol{e } ) } \ar[ul]^{\iota_{t(\ol{e } ) } } \ar@/_/[uul]_{\psi_{t(\ol{e } ) } } \\ & \ar[ul]^{f_e } g_e\ar[ur]_{f_{\ol{e } } } & } \ ] ] if @xmath362 is a tree , we also speak of the _ fiber sum _ @xmath1225 of @xmath31 . if @xmath362 has exactly two vertices @xmath386 , @xmath457 and one topological edge @xmath374 , then @xmath1225 is indeed the fiber sum of @xmath1178 and @xmath1231 over @xmath55 as introduced earlier . more generally , if @xmath386 is a terminal vertex of @xmath362 ( i.e. , there is only one edge @xmath372 with @xmath1232 ) , then in a natural way @xmath1233 where @xmath372 is the edge of @xmath362 with @xmath1232 , and @xmath1234 . by induction on the number of vertices of @xmath362 , this yields in particular : [ lem : injection into fiber sum ] suppose @xmath362 is a tree . then for each @xmath439 , the natural morphism @xmath1235 is injective . if @xmath362 is a tree , then the fiber sum of @xmath31 may also be identified with the abelianization of the fundamental group @xmath3 of @xmath31 ; more generally : [ lem : homology and fiber sums ] let @xmath79 be a maximal subtree of @xmath362 , and @xmath472 . suppose @xmath1236 for all @xmath367 , where @xmath1237 , @xmath439 , is the natural morphism . then there is a natural isomorphism @xmath1238 if for some prime @xmath0 , all @xmath35 are elementary abelian @xmath0-groups , then we have @xmath1239 this follows easily from the mayer - vietoris sequence associated to a graph of groups ( cf . section [ sec : mv ] ) . note that under the assumption of the previous lemma , for each vertex @xmath74 , the natural morphism @xmath725 induces an injective morphism @xmath1240 . this conclusion can also be achieved under a weaker hypothesis on @xmath31 : it is enough to require that there exists a morphism @xmath1241 to an abelian group which is injective on the vertex groups of @xmath31 ( since such a morphism factors through @xmath1242 ) . similarly , if there exists a morphism @xmath1241 to an elementary abelian @xmath0-group which is injective on the vertex groups of @xmath31 , then the induced morphisms @xmath1243 are injective . let @xmath31 be a graph of groups based on the graph @xmath362 . let @xmath79 be a maximal subtree of @xmath362 , let @xmath1244 , and let @xmath401 be an orientation of @xmath379 . from section [ sec : pi1 ] recall that the fundamental group @xmath3 of @xmath31 may be identified with the iterated hnn extension @xmath1245 of @xmath475 , where @xmath1246 suppose now that all vertex groups of @xmath31 are abelian . we let @xmath1247 the natural morphism @xmath1248 extends to a morphism @xmath1249 , which is injective on the vertex groups of @xmath31 . the natural morphism @xmath1250 factors through @xmath1251 . for this reason we call @xmath1252 the _ partial abelianization _ of @xmath3 along @xmath79 . note that although @xmath463 is independent ( up to isomorphism ) of the choice of @xmath79 , simple examples show that the group @xmath1252 in general depends on @xmath79 . in the following let @xmath2 be a root property such that free groups are residually @xmath2 . by lemma [ residual p criterion ] , we have : [ lem : residually p for partial abelianization ] suppose all vertex groups of @xmath31 are finite abelian groups . if @xmath1252 is residually @xmath2 , then @xmath1253 is also residually @xmath2 . we do not know whether the converse of the implication stated in the previous lemma holds in general ; however , we do have : [ lem : crucial ] suppose there exists a morphism @xmath1254 to a finite abelian @xmath2-group @xmath86 , which is injective when restricted to each vertex group @xmath35 of @xmath31 . then @xmath1252 is residually @xmath2 . this is a consequence of corollary [ cor : crucial , preparation ] and the following observation : let @xmath1113 be a morphism to an abelian group @xmath86 . then there is a morphism @xmath1255 which agrees with @xmath90 on each vertex group of @xmath31 and such that @xmath1256 for @xmath408 . by the universal property of the fiber sum , the restriction of @xmath90 to a morphism @xmath1257 factors through a morphism @xmath1258 . since @xmath86 is abelian , this morphism extends to a morphism @xmath1255 by @xmath1256 for @xmath408 which has the required properties . let @xmath31 be a graph of finitely generated groups with underlying graph @xmath362 , and let @xmath111 be a central @xmath0-filtration of @xmath31 . recall that the @xmath20th layer @xmath184 of @xmath111 is the graph of groups whose underlying graph is also @xmath362 and whose vertex and edge groups are the elementary abelian @xmath0-groups @xmath1259 and @xmath667 , respectively , with edge morphisms @xmath1260 . fix a maximal subtree @xmath79 of @xmath362 . for every @xmath65 we can form the partial abelianization @xmath1261 of the fundamental group of @xmath184 along @xmath79 . the following important theorem is the culmination of the results obtained so far in this paper . essentially , it reduces the question whether @xmath30 is residually @xmath0 to finding a filtration @xmath111 such that the partially abelianized fundamental group of the first layer @xmath1262 of @xmath111 is residually @xmath0 . [ thm : reduction theorem ] suppose the following hold : 1 . @xmath111 is @xmath0-potent ; 2 . for each @xmath372 , @xmath1105 intersects to a uniformly @xmath0-potent filtration on @xmath675 ; 3 . @xmath1263 is residually @xmath0 . then @xmath1190 is residually @xmath0 for each @xmath65 . if in addition @xmath111 is separating and separates the edge groups of @xmath31 , then @xmath54 is residually @xmath0 , and the vertex and edge groups of @xmath31 are closed in the pro-@xmath0 topology on @xmath30 . for the proof of this theorem we first show that its hypotheses imply that the partial abelianization of each layer of @xmath111 is residually @xmath0 ( not only the first , as stipulated by condition ( 3 ) ) . below , we continue to use the notation introduced in section [ sec : partial ab ] above . for every edge @xmath1264 we equip @xmath1265 and @xmath1266 with filtrations @xmath1267 respectively @xmath1268 , and let @xmath1269 be the isomorphism induced by @xmath1270 . [ lem : reduction theorem ] suppose @xmath111 satisfies conditions in theorem [ thm : reduction theorem ] . then for each @xmath65 , the group @xmath1261 is residually @xmath0 . by ( 1 ) and ( 2 ) we can take injective @xmath76-linear maps @xmath1271 such that for each edge @xmath372 with @xmath492 , @xmath1272 restricts to isomorphisms @xmath1273 and @xmath1274 , and the diagram @xmath1275^{l_{n+1}(\varphi_{e } ) } & & l_{n+1}(\mathbf b_e)\\ l_1(\mathbf a_e ) \ar[u]^\cong\ar[rr]^{l_1(\varphi_{e } ) } & & l_1(\mathbf b_e ) \ar[u]^\cong } \ ] ] commutes . the @xmath1276 give rise to an _ injective _ @xmath76-linear map @xmath1277 which restricts to isomorphisms @xmath1273 and @xmath1274 ( after identifying @xmath1278 and @xmath1279 with subgroups of @xmath1280 as usual ) . this morphism extends to a morphism @xmath1281 by the identity on @xmath401 . therefore , since the group @xmath1263 is residually @xmath0 ( by assumption ( 3 ) ) , so is @xmath1261 for each @xmath1282 , by corollary [ cor : generate hnn , 2 ] . we will deduce theorem [ thm : reduction theorem ] from the following proposition : [ prop : reduction ] assume that each vertex group of @xmath31 is a @xmath0-group and @xmath111 is of finite length . suppose moreover that @xmath1261 is residually @xmath0 for every @xmath65 . then @xmath3 is residually @xmath0 . let us see how this proposition yields theorem [ thm : reduction theorem ] . so suppose conditions ( 1)(3 ) hold . by lemma [ lem : reduction theorem ] , the groups @xmath1261 are residually @xmath0 , for each @xmath65 . by the proposition applied to @xmath662 in place of @xmath31 and the filtration of @xmath662 of length @xmath1283 given on vertex groups by @xmath1284 we obtain that @xmath1190 is residually @xmath0 , for each @xmath65 . this is the first conclusion of theorem [ thm : reduction theorem ] . the rest of the theorem follows from this and corollaries [ cor : hempel , cor ] and [ cor : vertex and edge groups closed ] of proposition [ hempel criterion residually p ] . it remains to give a proof of proposition [ prop : reduction ] . for this , we need a consequence of the main results of section [ sec : an embedding theorem ] : [ lem : higman , 2 ] suppose @xmath362 is a tree , all vertex groups of @xmath31 are @xmath0-groups , and @xmath111 has finite length . then there exists a @xmath0-group @xmath86 , a finite - length central @xmath0-filtration @xmath1285 of @xmath86 , and for each @xmath439 a group embedding @xmath1286 and a compatible stretching @xmath1287 of @xmath111 such that 1 . @xmath1288 for every @xmath435 ; 2 . @xmath1289 is a morphism @xmath1290 of filtered groups , for every @xmath439 ; 3 . for each @xmath65 , after naturally identifying each vertex group @xmath1291 of the tree @xmath672 of elementary abelian @xmath0-groups with its image in @xmath1292 , the latter group contains the internal fiber sum @xmath1293 of @xmath672 . in particular , @xmath54 is residually @xmath0 . the proof of the first statements proceeds by an obvious induction on the number of vertices of @xmath31 , using proposition [ prop : higman , more precise ] in the inductive step . the existence of the family of group embeddings @xmath1286 satisfying ( 1 ) together with lemma [ residual p criterion ] yields that @xmath30 is residually @xmath0 . with this lemma in hand , we now turn to : note that the hypotheses of the proposition are preserved under passage from @xmath111 to a compatible stretching . ( since free groups are residually @xmath0 . ) thus , by lemma [ lem : higman , 2 ] applied to the tree of groups @xmath470 in place of @xmath31 , we may assume that there exists a morphism @xmath1294 to a @xmath0-group @xmath86 and a finite - length central @xmath0-filtration @xmath1285 of @xmath86 such that 1 . @xmath1295 for every @xmath477 ; 2 . @xmath90 restricts to an injective morphism @xmath1296 of filtered groups , for every @xmath439 ; and 3 . @xmath1297 for each @xmath65 ( after naturally identifying each vertex group @xmath1259 of @xmath1298 with its image in @xmath1292 under @xmath1299 ) . for @xmath367 define @xmath1300 the filtration @xmath1285 of @xmath86 is @xmath1301-invariant , for each @xmath367 . for each @xmath65 , the iterated hnn extension @xmath1302 of @xmath1303 is residually @xmath0 , by assumption , hence by corollary [ cor : generate hnn ] ( or simply because @xmath1303 is a direct summand of the @xmath76-linear space @xmath409 ) , there is a @xmath0-group containing @xmath409 and an extension of each @xmath1304 , @xmath408 , to an inner automorphism . hence by lemma [ lem : hnn ] there is a @xmath0-group @xmath1305 and for each @xmath408 an element @xmath1306 such that @xmath1307 for all @xmath408 and @xmath1308 . hence @xmath90 extends to a morphism @xmath1309 by @xmath1310 for all @xmath408 and so by lemma [ residual p criterion ] , @xmath3 is residually @xmath0 . in section [ sec : virtual res p ] we use the theorem above to prove a criterion for the fundamental group of a graph of groups to be _ virtually _ residually @xmath0 . a last missing ingredient is a procedure on graphs of groups introduced in the next section . let @xmath31 be a graph of groups with underlying graph @xmath362 and fundamental group @xmath54 . suppose @xmath1311 is a surjective group morphism to a finite group @xmath214 which is trivial on each vertex group @xmath35 of @xmath31 . by theorem [ thm : bass - serre ] , the ( finite index , normal ) subgroup @xmath1312 of @xmath3 can also be realized as the fundamental group of a certain graph of groups whose vertex and edge groups are given by the vertex respectively edge groups of @xmath31 . we need an explicit description of such a graph of groups @xmath38 , which may be thought of as a kind of `` unfolding '' of @xmath31 along @xmath90 . we begin with a description of the underlying graph @xmath1313 of @xmath38 . let @xmath362 be a graph ; we write @xmath363 and @xmath364 . we denote by @xmath1314 the fundamental group of @xmath362 , i.e. , @xmath1315 where @xmath1316 is the graph of groups based on @xmath362 with trivial vertex groups . we fix a maximal subtree @xmath79 of @xmath362 , and identify @xmath1314 with @xmath1317 as indicated in . let @xmath214 be a finite group and @xmath1318 a surjective group morphism . note that then @xmath90 extends uniquely to a group morphism @xmath1319 . ( this depends on the choice of @xmath79 . ) here @xmath1320 is the path group of @xmath1316 , i.e. , the group generated by @xmath379 subject to the relations @xmath1321 for @xmath367 . we have @xmath1322 for all @xmath477 . let now @xmath544 be the finite graph with vertex set @xmath1323 , edge set @xmath1324 , and @xmath1325 for all @xmath1326 . it is easy to see that @xmath544 is connected since @xmath1327 generates the group @xmath214 . we have a natural morphism @xmath1328 of graphs given by @xmath1329 we call @xmath396 the _ unfolding of the graph @xmath362 along @xmath90 . _ ostensibly , @xmath396 seems to depend on the choice of @xmath79 . however , it is not hard to see ( e.g. , using covering spaces and proposition [ prop : phistar ] below ) that in fact , given another choice @xmath1330 of maximal subtree of @xmath362 and @xmath1331 constructed as above , there exists an isomorphism @xmath1332 of graphs such that @xmath1333 . if @xmath362 is a tree ( so @xmath1334 , then the group @xmath1314 and hence @xmath214 is trivial , and @xmath544 is naturally isomorphic to the original graph @xmath362 . there is an obvious connection between unfolding of graphs and cayley graphs of finite groups , as explained in the following example ( not used later ) : recall that given a generating set @xmath837 of a group @xmath98 , the cayley graph of @xmath98 with respect to @xmath837 is the _ oriented _ ( possibly infinite ) graph @xmath1335 with vertex set @xmath1336 , orientation @xmath1337 , and @xmath1338 let @xmath401 be an orientation of @xmath362 ; then @xmath1339 generates @xmath214 , and @xmath1340 is an orientation of @xmath544 . thus the natural surjections @xmath1341 and @xmath1342 define a morphism @xmath1343 of graphs . in particular , if @xmath362 has only a single vertex , and @xmath1344 is injective , then @xmath544 is nothing but the cayley graph of @xmath214 with respect to its generating set @xmath1339 . we continue to use the notation introduced in the last subsection , but assume in addition that we are given a graph of groups @xmath31 based on @xmath362 . we define @xmath38 as the graph of groups with underlying graph @xmath544 , vertex respectively edge groups @xmath1345 and edge morphisms @xmath1346 there is a natural morphism @xmath610 of graphs of groups with underlying morphism of graphs @xmath1347 , which is the identity on each vertex group @xmath1348 and on each edge group @xmath1349 . this morphism , also denoted by @xmath396 , yields a surjective morphism @xmath1350 , which restricts to a morphism @xmath1351 between fundamental groups . we call @xmath396 the _ unfolding of @xmath31 along @xmath90 . _ in the proposition below we continue to denote the extension of @xmath1352 to a morphism @xmath1353 with @xmath775 for @xmath1354 , @xmath414 , by the same symbol . the following proposition may also be proved easily using the topological interpretation of graphs of groups ; we give a purely algebraic proof . [ prop : phistar ] the morphism @xmath554 is injective , with image @xmath1312 . let @xmath1355 be a base point of @xmath3 , that is , @xmath529 . we choose @xmath1356 as our base point for @xmath1357 . let @xmath1358 be an element of @xmath1359 : @xmath1360 where @xmath1361 , and @xmath1362 and @xmath1363 note that then @xmath1364 by and @xmath1365 by , so @xmath1366 conversely , given an element @xmath1367 of @xmath14 , defining @xmath1368 as in and yields an element @xmath1358 of @xmath1359 as above with @xmath1369 . this shows that @xmath1370 . now suppose @xmath1371 , and the representation of @xmath1358 above is reduced , that is , either 1 . @xmath449 and @xmath450 ; or 2 . @xmath426 and @xmath1372 for each @xmath258 such that @xmath1373 . since @xmath1374 , the image @xmath1375 of @xmath1358 under @xmath554 is also represented by a reduced path . thus @xmath554 is injective . the following simple observation is used in section [ sec : virtual res p ] . [ lem : unfolding and quotients ] let @xmath1318 be a morphism onto a finite group , and let @xmath91 be the unfolding of @xmath31 along @xmath90 . let @xmath1376 be a compatible collection of normal subgroups of @xmath31 , and let @xmath1377 be the unfolding of @xmath1123 along @xmath90 . then @xmath1378 as graphs of groups , where @xmath1379 . we leave the proof to the reader . it might be instructive for the reader to see the unfolding procedure at work in a concrete simple example . ( the material in this subsection is not used later . ) suppose the underlying graph @xmath362 of @xmath31 has only one vertex @xmath386 ; we write @xmath1380 . the fundamental group of @xmath31 is the iterated hnn extension @xmath1381 of @xmath88 , where @xmath401 is an orientation of @xmath362 . in this case , the definition of the unfolding @xmath38 of @xmath31 along @xmath90 given above simplifies somewhat : we may simply describe its underlying graph @xmath544 by @xmath1382 , @xmath1324 , and @xmath1383 and we have @xmath1384 for @xmath1326 . figure [ fig : loop ] illustrates this for the case where @xmath362 has only one topological edge @xmath374 , and @xmath1385 is cyclic , with @xmath1386 . @xmath1387&\quad } \xygraph { ! { < 0cm,0cm>;<1cm,0cm>:<0cm,1 cm > : : } ! { ( 0,0);a(0)*{}?(2.0)}+{\sigma\times 0}="g0 " ! { ( 0,0);a(36)*{}?(2.0)}+{\sigma\times 1}="g1 " ! { ( 0,0);a(72)*{}?(2.0)}+{\sigma\times 2}="g2 " ! { ( 0,0);a(108)*{}?(2.0)}+{}="g3 " ! { ( 0,0);a(144)*{}?(2.0)}+{}="gi-2 " ! { ( 0,0);a(180)*{}?(2.0)}+{\sigma\times i}="gi-1 " ! { ( 0,0);a(216)*{}?(2.0)}+{\sigma\times ( i+1)}="gi " ! { ( 0,0);a(252)*{}?(2.0)}+{}="gi+1 " ! { ( 0,0);a(288)*{}?(2.0)}+{}="gs-2 " ! { ( 0,0);a(324)*{}?(2.0)}+{\sigma\times ( s-1)}="gs-1 " " g0":@/_0.5em/"g1"_{\varphi } " g1":@/_0.5em/"g2"_{\varphi } " g2":@/_0.5em/"g3 " " g3":@{.}@/_0.2em/"gi-2 " " gi-2":@/_0.5em/"gi-1 " " gi-1":@/_0.5em/"gi"_{\varphi } " gi":@/_0.5em/"gi+1 " " gi+1":@{.}@/_0.2em/"gs-2 " " gs-2":@/_0.5em/"gs-1 " " gs-1":@/_0.5em/"g0"_{\varphi } } \ ] ] suppose now that @xmath214 is a subgroup of @xmath1388 , and for each edge @xmath372 of @xmath362 , the automorphism @xmath1389 extends the partial automorphism @xmath1124 of @xmath88 . let @xmath95 be a maximal subtree of @xmath544 , and identify each vertex group @xmath1390 with a subgroup of @xmath1391 as usual . in this situation unfolding has the following desirable property : [ lem : unfolding hnn ] there exists a morphism @xmath1392 which is bijective on each vertex group @xmath1390 of @xmath38 . for all @xmath1326 , the isomorphism @xmath1393 extends @xmath1394 , and the composition of the @xmath1393 along the geodesic @xmath1395 from the vertex @xmath1014 to @xmath163 in @xmath95 is the isomorphism @xmath1396 given by @xmath1397 . hence by lemma [ lem : special amalgam ] there is a morphism @xmath1398 which , for each @xmath1399 , extends @xmath1400 ( and hence is bijective on @xmath1390 ) . now @xmath1401 , and therefore @xmath1402 ( see figure [ fig : unfolding hnn ] . ) this allows us to extend @xmath1091 to a morphism @xmath1403 by @xmath1404 for all edges @xmath1405 . an analogue of this fact will be established for general graphs of elementary abelian @xmath0-groups in proposition [ prop : unfolded ] . @xmath1406 we now continue where section [ sec : a reduction theorem ] left off , and put ourselves back in the setting of section [ subsec : reduction theorem ] . that is , we let @xmath31 be a graph of finitely generated groups with underlying graph @xmath362 . we recall the statement of the main theorem of this chapter : [ thm : reduction theorem , 3 ] if @xmath31 admits a @xmath0-excellent filtration , then the group @xmath54 is virtually residually @xmath0 . throughout the rest of this section , we fix a maximal subtree @xmath79 of the underlying graph @xmath362 of @xmath31 . we identify @xmath3 with @xmath463 and in this way each vertex group @xmath35 of @xmath31 with a subgroup of @xmath3 . theorem [ thm : reduction theorem , 3 ] will be obtained by applying the reduction theorem ( theorem [ thm : reduction theorem ] ) to a suitable finite index subgroup of @xmath30 ; the heart of the construction of such a subgroup is contained in the following proposition : [ prop : unfolded ] suppose all vertex groups @xmath35 of @xmath31 are elementary abelian @xmath0-groups . then there exists a morphism @xmath1318 onto a finite group @xmath214 , with corresponding unfolding @xmath610 of @xmath31 , and a morphism @xmath1407 which is injective when restricted to each vertex group of @xmath38 . we exploit the homogeneity of the elementary abelian @xmath0-group @xmath1408 : choose an orientation @xmath401 of @xmath1244 , i.e. , @xmath379 is the disjoint union @xmath1409 ; and for every @xmath408 choose an extension of the partial automorphism @xmath1410 of @xmath88 to an automorphism @xmath1411 of @xmath88 . we denote the free group on the generators @xmath401 by @xmath57 . note that @xmath1412 . we let @xmath1413 be the morphism with @xmath1414 ( @xmath408 ) , and we consider the semidirect product @xmath1415 of @xmath57 with @xmath88 via @xmath90 : @xmath1416 we have a surjective morphism of the group @xmath1417 onto @xmath1415 which is the identity on @xmath88 and on @xmath401 . composing with the natural morphism @xmath1418 we thus obtain a morphism @xmath1419 which is injective on the vertex groups of @xmath31 . let @xmath214 be the subgroup of @xmath1388 generated by the @xmath1411 ( @xmath408 ) . writing @xmath1420 , we have a short exact sequence @xmath1421 note that @xmath98 acts trivially on @xmath88 , in particular , @xmath1422 . as in section [ sec : unfolding a graph of groups ] we continue to denote the extension of @xmath1352 to a morphism @xmath1353 which is the trivial morphism on each vertex group @xmath35 by the same symbol @xmath90 . then the following diagram commutes : @xmath1423^\psi \ar[dr]^\psi & f\ltimes\sigma \ar[d ] \\ & a. } \ ] ] we let @xmath91 be the unfolding of @xmath31 along @xmath1424 as constructed in section [ subsec : unfolding ] above . then by proposition [ prop : phistar ] @xmath1425 and thus @xmath1426 so there is a morphism @xmath1427 making the diagram @xmath1428_{\phi _ * } \ar[r]^{\widetilde{\psi } } & \gamma\ltimes\sigma \ar[d ] \\ \pi_1(\scrg ) \ar[r]^\psi & f\ltimes\sigma } \ ] ] commute . clearly @xmath1429 is injective on the vertex groups of @xmath38 . let @xmath247 be the natural projection @xmath1430 . then @xmath1431 is injective on the vertex groups of @xmath38 . for the following corollary of proposition [ prop : unfolded ] , let @xmath1376 be a compatible collection of normal subgroups of @xmath31 with @xmath1432 for all @xmath439 . ( later this will be applied to @xmath1433 where @xmath111 is a complete @xmath0-excellent filtration of @xmath31 . ) then @xmath1123 is a graph of elementary abelian @xmath0-groups based on @xmath362 . given a morphism @xmath91 we denote by @xmath1434 the compatible collection of normal subgroups of @xmath552 defined in lemma [ lem : morphisms and filtrations , 1 ] . note that then @xmath1435 for all vertices @xmath1436 of the underlying graph @xmath544 of @xmath38 . [ cor : unfolded , 1 ] there exists a morphism @xmath1437 of graphs of groups with the following properties : 1 . each group morphism @xmath1438 and @xmath1439 is bijective ; 2 . the induced group morphism @xmath1440 is injective ; 3 . its image @xmath653 is a finite index normal subgroup of @xmath3 ; and 4 . for every maximal subtree @xmath95 of the graph underlying the graph of groups @xmath38 , the group @xmath1441 is residually @xmath0 . applying proposition [ prop : unfolded ] with @xmath1123 in place of @xmath31 , we obtain a morphism @xmath1442 to a finite group , with corresponding unfolding @xmath1377 of @xmath1123 , and a morphism @xmath1443 to an elementary abelian @xmath0-group which is injective when restricted to each vertex group of @xmath1444 . let @xmath91 be the unfolding of @xmath31 along @xmath90 . then ( 1)(3 ) hold by proposition [ prop : phistar ] . by lemma [ lem : unfolding and quotients ] we have @xmath1445 , hence ( 4 ) follows from lemma [ lem : crucial ] . we can now prove theorem [ thm : reduction theorem , 3 ] . suppose first that @xmath111 is a _ complete _ @xmath0-excellent filtration of @xmath31 . apply corollary [ cor : unfolded , 1 ] with @xmath1446 , and let @xmath91 have properties ( 1)(4 ) in that corollary . by ( 1 ) and lemmas [ lem : separating , 1 ] and [ lem : separating , 2 ] , the filtration @xmath1447 of @xmath38 remains @xmath0-excellent . this together with ( 4 ) shows that @xmath38 satisfies the hypotheses of theorem [ thm : reduction theorem ] . hence @xmath1359 is residually @xmath0 , and thus by ( 2 ) and ( 3 ) , the group @xmath30 is virtually residually @xmath0 . if @xmath111 is an arbitrary @xmath0-excellent filtration of @xmath31 , we consider a morphism of graphs of groups satisfying ( 1)(3 ) in proposition [ prop : commoncover ] applied to the compatible collection of normal subgroups @xmath1448 of @xmath31 . the pullback of @xmath111 under such a morphism is complete and , by lemmas [ lem : separating , 1 ] and [ lem : separating , 2 ] , remains @xmath0-excellent . in this way we reduce to the case of complete @xmath111 treated above . in this chapter we first define ( in section [ sec : p - compatible ] ) what we mean by a filtration of a group to be @xmath0-compatible with a given collection of subgroups , and then show ( in section [ sec : p - compatible linear ] ) a general theorem which allows to construct filtrations of finitely generated linear groups which are @xmath0-compatible with certain abelian subgroups . in section [ sec : proof of the main theorem ] we then apply this theorem to give a proof of our main theorem from the introduction . section [ sec : approx ] fills in the proof of a commutative algebraic fact used earlier ( in section [ sec : p - compatible linear ] ) , and section [ sec : fibered ] discusses the main theorem in the special case of fibered @xmath1-manifolds . we make the following somewhat ad - hoc definition ( which is related to the `` @xmath0-congruence systems '' of ( * ? ? ? * interlude b ) ) : let @xmath3 be a group , let @xmath1449 be a collection of subgroups of @xmath3 , and let @xmath714 be an integer . we say that a filtration @xmath1450 of @xmath3 is _ @xmath0-compatible of level @xmath1451 with @xmath1449 _ if the following hold : 1 . @xmath111 is a normal filtration of @xmath3 with @xmath1452<\infty$ ] for each @xmath65 ; 2 . @xmath111 , construed as a complete filtration of @xmath346 , is @xmath0-potent ; 3 . @xmath111 is separating ( i.e. , @xmath1453 ) ; 4 . @xmath111 separates @xmath1449 ( i.e. , @xmath1454 for every @xmath1455 ) ; and 5 . for every @xmath1455 the filtration @xmath111 intersects to the lower central @xmath0-filtration @xmath1456 of @xmath79 , shifted by @xmath1451 terms to the left , i.e. , @xmath1457 . a filtration of @xmath3 which is @xmath0-compatible of level @xmath732 with @xmath1449 will simply said to be @xmath0-compatible with @xmath1449 . ( we will only encounter @xmath0-compatible filtrations of level @xmath732 or @xmath163 below . ) if the group @xmath3 admits a filtration which is @xmath0-compatible with some collection of subgroups , then @xmath3 is virtually residually @xmath0 . we introduced the notion of @xmath0-compatible filtration because it furnishes a `` local '' version of the concept of @xmath0-excellent filtration of a graph of groups considered in the previous chapter : let @xmath714 be an integer ; then , given a graph of groups @xmath31 and for each vertex @xmath74 a filtration @xmath63 of @xmath35 which is @xmath0-compatible of level @xmath1451 with the subgroups @xmath675 of @xmath35 , where @xmath372 ranges over all edges with @xmath492 , we obtain a filtration @xmath1458 of @xmath31 . we say that a filtration @xmath111 of @xmath31 arising in this way is _ @xmath0-compatible of level @xmath1451 . _ such a @xmath0-compatible filtration @xmath111 of level @xmath1451 of @xmath31 automatically satisfies conditions ( 1)(4 ) in the definition of @xmath0-excellency ; if in addition the lower central @xmath0-series of the edge groups of @xmath31 , shifted by @xmath1451 terms to the left , is uniformly @xmath0-potent ( e.g. , if the edge groups of @xmath31 are abelian @xmath0-torsion free , see section [ sec : p - potent ] ) , then every @xmath0-compatible filtration of level @xmath1451 is @xmath0-excellent . from the main result ( theorem [ thm : reduction theorem , 2 ] ) of the preceding chapter we therefore immediately obtain : [ thm : virt res p for graphs of linear groups , 1 ] let @xmath31 be a graph of finitely generated groups , and suppose all edge groups @xmath55 are abelian @xmath0-torsion free . suppose for some @xmath714 , @xmath31 admits a @xmath0-compatible filtration of level @xmath1451 . then @xmath30 is virtually residually @xmath0 . in the next section we construct filtrations of linear groups which are @xmath0-compatible with certain abelian subgroups . in section [ sec : proof of the main theorem ] we ll apply this construction together with the criterion in theorem [ thm : virt res p for graphs of linear groups , 1 ] above in the case of graphs of groups arising from @xmath1-manifolds . in the rest of this section we collect a few more observations on @xmath0-compatible filtrations , and record an application of theorem [ thm : virt res p for graphs of linear groups , 1 ] . first we note that under natural conditions , if there is a complete @xmath0-compatible filtration at all , then there is a canonical one , by lemma [ lem : canonical filtration ] : [ lem : canonical filtration , 2 ] let @xmath3 be a group and let @xmath1449 be a collection of subgroups of @xmath3 . if @xmath111 is a central @xmath0-filtration of @xmath3 such that @xmath1459 for every @xmath1455 , then @xmath1460 for every @xmath1455 . therefore , assuming that @xmath233 is @xmath0-potent and @xmath177 is finitely generated , if there exists a complete filtration of @xmath3 which is @xmath0-compatible with @xmath1449 , then @xmath233 is @xmath0-compatible with @xmath1449 . examples of @xmath0-compatible filtrations of free groups may be obtained via the following : [ lem : p - compatible for free groups ] let @xmath3 be a finitely generated residually @xmath0 group with @xmath0-torsion free quotients @xmath1461 , for @xmath1462 , and let @xmath1449 be a collection of retracts of @xmath3 . if @xmath0 is odd , then the complete filtration @xmath1463 of @xmath3 is @xmath0-compatible with @xmath1449 . for arbitrary @xmath0 , the filtration @xmath1464 is @xmath0-compatible of level @xmath163 with @xmath1449 . by lemma [ lem : retract ] , @xmath233 separates @xmath79 , and by lemma [ lem : lower central for semidirect products ] we have @xmath1465 . this shows that @xmath233 satisfies conditions ( 1 ) and ( 3)(5 ) in the definition of @xmath0-compatibility . if @xmath0 is odd , then @xmath233 is @xmath0-potent by corollary [ cor : phi_n ] , hence @xmath233 is @xmath0-compatible with @xmath1449 . for general @xmath0 , shifting @xmath233 by @xmath163 to the left yields a @xmath0-potent filtration of @xmath1466 , by lemma [ lem:2-potent ] . [ retract ] by virtue of the previous lemma , theorem [ thm : virt res p for graphs of linear groups , 1 ] applies to graphs of finitely generated residually @xmath0 groups with abelian @xmath0-torsion free edge groups , such that for each edge @xmath372 , the image of the edge group @xmath55 under the edge morphism @xmath489 is a retract of @xmath1106 . this leads to a refinement of a result of wise : let @xmath31 be a graph of free groups of finite rank with cyclic edge groups , and @xmath54 . suppose that @xmath3 is _ balanced , _ i.e. , there are no @xmath4 , @xmath5 , and non - zero integers @xmath106 , @xmath20 with @xmath1467 such that @xmath1468 and @xmath1469 are conjugate . then for every @xmath0 , @xmath3 is virtually residually @xmath0 . wise @xcite showed that the balancedness condition formulated in this corollary implies that there exists a morphism @xmath1470 of graphs of groups having finite degree and such that @xmath1471 is _ clean _ , that is , each of the ( cyclic ) edge groups of @xmath1471 maps to a free factor of the target ( free ) vertex group under the edge morphism . he then proved that fundamental groups of clean graphs of free groups are residually finite ; in fact , much better , they are subgroup separable . our theorem [ thm : virt res p for graphs of linear groups , 1 ] shows that they are , for each @xmath0 , virtually residually @xmath0 . note that if @xmath3 is a balanced group , then @xmath3 does not contain any baumslag - solitar group . hsu and wise @xcite recently showed that if @xmath31 is a graph of free groups with cyclic edge groups and @xmath54 contains no baumslag - solitar group , then @xmath3 virtually embeds into a right - angled artin group , and hence has a finite - index subgroup which is residually @xmath0 for every @xmath0 ( and is linear over @xmath18 ) ; this also implies the corollary above in a stronger form . given a commutative ring @xmath51 , we denote by @xmath1472 the group of upper unitriangular @xmath1473-matrices with entries in @xmath51 ( a subgroup of @xmath1474 ) . a matrix in @xmath1475 is called unipotent if it is conjugate to an element of @xmath1472 , and a subgroup of @xmath1475 is called unipotent if it is conjugate to a subgroup of @xmath1472 . ( if @xmath14 is a field , then every subgroup of @xmath19 consisting entirely of unipotent matrices is unipotent , cf . * corollary 1.21 ) . ) a subgroup @xmath79 of a group @xmath1476 is said to be maximal unipotent if @xmath79 is unipotent and not contained in a strictly larger unipotent subgroup of @xmath3 . the main result of this section is : [ thm : virt res p for graphs of linear groups , 2 ] let @xmath3 be a finitely generated linear group , and let @xmath1449 be a finite collection of finitely generated abelian unipotent subgroups of @xmath3 , such that for each @xmath1455 , 1 . @xmath79 is maximal abelian or maximal unipotent ; and 2 . the image of @xmath79 in @xmath1477 has rank @xmath1478 . then for all but finitely many @xmath0 , there exists a filtration of @xmath3 which is @xmath0-compatible of level @xmath163 with @xmath1449 . the proof of this theorem is given in section [ sec : proof of thm : virt res p for graphs of linear groups , 2 ] below . theorem [ thm : virt res p for graphs of linear groups , 2 ] together with theorem [ thm : virt res p for graphs of linear groups , 1 ] immediately yields a criterion for the fundamental group of certain graphs of linear groups to be virtually residually @xmath0 ( which refines ( * ? ? ? * theorem 5.1 ) ) : [ cor : virt res p for graphs of linear groups ] let @xmath31 be a graph of finitely generated linear groups . suppose that each edge group @xmath55 maps to a finitely generated abelian unipotent subgroup of @xmath35 where @xmath520 which is maximal abelian or maximal unipotent in @xmath35 and whose image in @xmath1479 has rank @xmath1480 . then for all but finitely many @xmath0 , the group @xmath30 is virtually residually @xmath0 . the hypothesis ( 2 ) on @xmath79 in the theorem above is satisfied if @xmath79 is free abelian of rank @xmath34 and maps to an infinite subgroup of @xmath1477 ; this is the situation of interest in our later applications . hypothesis ( 2 ) , however , also holds if @xmath79 is cyclic ; hence from the previous corollary we obtain : let @xmath31 be a graph of finitely generated linear groups with cyclic unipotent edge groups . suppose that for each edge @xmath372 , the image of @xmath55 in @xmath1209 , where @xmath520 , is maximal abelian . then for all but finitely many @xmath0 , the group @xmath30 is virtually residually @xmath0 . before we turn to the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] we present an instructive example : consider the subgroup @xmath1481 of @xmath1482 , which has a presentation @xmath1483 ( so @xmath3 is the baumslag - solitar group @xmath1484 ) . the group @xmath3 is metabelian , and its ( unique ) maximal unipotent subgroup is the normal subgroup @xmath1485 \\ 0 & 1\end{smallmatrix } \right)$ ] of @xmath3 ( isomorphic to the additive group of @xmath1486 $ ] , in particular , not finitely generated ) . this shows that in general , the assumption in our theorem that @xmath3 be finitely generated does not force its abelian maximal unipotent subgroups to be finitely generated . ( however , if @xmath3 is discrete , then each abelian maximal unipotent subgroup of @xmath3 is finitely generated , so the assumption in our theorem that each member of @xmath1449 be finitely generated may be dropped . ) this example can also be used to show that the hypothesis in corollary [ cor : virt res p for graphs of linear groups ] that each edge group maps onto a maximal abelian or maximal unipotent subgroup of the target vertex group can not be omitted : let @xmath1487 , a unipotent abelian subgroup of @xmath3 ( which , however , is clearly neither maximal abelian nor maximal unipotent ) . it was shown by higman @xcite that the amalgamated product @xmath1488 of @xmath3 with itself via the identity @xmath1489 is non - hopfian and hence not residually finite . of course , examples for this phenomenon ( in the form of hnn extensions of @xmath18 ) are also provided by baumslag - solitar groups : as is well - known , for integers @xmath106 , @xmath20 with @xmath1490 and @xmath1491 , the group @xmath1492 is not residually finite . a key algebraic ingredient for the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] in full generality is the following fact . let @xmath51 be a finitely generated subring of @xmath1493 . it is well - known that for all but finitely many @xmath0 there exists a maximal ideal @xmath1494 of @xmath51 such that @xmath1495 . ( see , e.g. , proposition [ non - trivial and reduced ] below . ) evidently the characteristic of @xmath1496 then divides @xmath1497 , for every @xmath765 . but in general we do not have @xmath1498 . ( consider , e.g. , @xmath1499 $ ] , @xmath1500 and @xmath267 . ) the following theorem implies that by judicious choice of @xmath0 we can exclude this phenomenon : [ thm : proprmi ] for all but finitely many primes @xmath0 there exists a maximal ideal @xmath1501 of @xmath51 such that the localization @xmath1502 of @xmath51 at @xmath1501 is regular unramified of residue characteristic @xmath0 . we refer to section [ sec : approx ] below for an explanation of the concepts from commutative algebra mentioned in the conclusion of this theorem . for its application in the present section , the reader only needs to be aware of one aspect of this conclusion , namely : [ lem : dividing by p ] let @xmath1503 be a regular unramified local ring of residue characteristic @xmath0 . then for all @xmath1504 and @xmath1505 : @xmath1506 here @xmath1507 . this follows immediately from lemma [ criterion ] below . in particular , given a prime @xmath0 and a maximal ideal @xmath1501 as in theorem [ thm : proprmi ] , we do have @xmath1508 for every @xmath765 . it may be worth noting that our theorem thus implies a more precise version of ( * ? ? ? * proposition 2 ) , which states that except for a finite number of primes @xmath0 , for every @xmath765 there exists a morphism @xmath1509 to some finite ring @xmath1510 such that @xmath1511 . we postpone the proof of theorem [ thm : proprmi ] to section [ sec : approx ] . we do note here , however , that in the case where @xmath51 is a ring of integers in a number field , this fact is quite easy to show : suppose @xmath51 is the ring of integers in a number field @xmath14 , and let @xmath1512 be the discriminant of @xmath14 ( a non - zero integer ) . for every @xmath0 there exists a prime ( hence maximal ) ideal @xmath1494 of @xmath51 , and @xmath1513 is a dvr ( hence regular ) of residue characteristic @xmath0 . moreover , if @xmath0 does not divide @xmath1512 , then for every such @xmath1494 , @xmath1513 is unramified . ( see any standard text in algebraic number theory , e.g. , ( * ? ? ? * chapter iii , section 2 ) . ) this allows us to give a proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] in the case relevant to our applications to @xmath1-manifolds ( namely , where @xmath3 is a lift of a finite covolume torsion free kleinian group ) which bypasses theorem [ thm : proprmi ] ; this point will be addressed in section [ sec : kleinians ] . before we now turn the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] , we make some observations on congruence subgroups . let @xmath1514 be a complete noetherian local ring whose residue field @xmath1515 is finite of characteristic @xmath0 . for every @xmath765 the natural surjective ring morphism @xmath1516 induces a surjective group morphism @xmath1517 whose kernel is called the _ @xmath258-th congruence subgroup _ of @xmath1474 , and denoted by @xmath1518 . note that @xmath1496 and hence @xmath1519 are finite , for any @xmath765 . in the following we abbreviate @xmath1520 , and set @xmath1521 . if @xmath51 is an integral domain , then @xmath1522 by the krull intersection theorem , hence the normal filtration @xmath1523 of @xmath1474 is separating ( and thus @xmath1474 is residually finite ) . proofs of the following properties of congruence subgroups can be found in @xcite ( see also ( * ? ? ? * chapter 13 ) ) . given an ideal @xmath1524 of @xmath51 , we write @xmath1525 for the @xmath258-th cartesian power of the set @xmath1524 ( not to be confused with the @xmath258-th power @xmath1526 of the ideal @xmath1524 ) . [ prop : ls94 ] for all integers @xmath1527 : 1 . @xmath1528 an elementary abelian @xmath0-group ; 2 . @xmath1529 \subseteq g_{i+j}$ ] ; 3 . @xmath1530 if @xmath51 has equal characteristic @xmath0 . in particular , by ( 1 ) and ( 2 ) , @xmath1523 is a ( strongly ) central @xmath0-filtration of @xmath1531 , hence @xmath1532 for every @xmath765 . if @xmath51 has equal characteristic @xmath0 , then @xmath1523 is also dimensional , by ( 3 ) , so @xmath1533 for every @xmath765 . under additional assumptions , these statements can be strengthened : ( * ? ? ? * proposition 3.2 ) if @xmath1534 or @xmath1535 , then @xmath1536 and if in addition @xmath51 is of equal characteristic @xmath0 , then @xmath1537 the following lemma combined with the previous proposition shows that for odd @xmath0 and well - behaved @xmath51 , the filtration @xmath1523 of @xmath3 is strongly @xmath0-potent : suppose that @xmath0 is odd and @xmath1514 is regular unramified . then for every @xmath765 , the morphism @xmath1538 induced by @xmath1539 is injective . let @xmath1540 , @xmath765 , such that @xmath1541 ; we need to show @xmath1542 . write @xmath1543 where @xmath1544 . then @xmath1545 we have @xmath1546 since @xmath296 , and @xmath1547 . moreover , for @xmath1548 , every binomial coefficient @xmath1549 is divisible by @xmath0 . therefore @xmath1550 . since @xmath1514 is regular unramified of residue characteristic @xmath0 , this yields @xmath1551 as required ( lemma [ lem : dividing by p ] ) . let @xmath51 be a commutative ring . given an @xmath1473-matrix @xmath1552 with entries in @xmath51 and @xmath1553 , we call the @xmath1554-tuple @xmath1555 the @xmath1067th codiagonal of @xmath25 . so the @xmath732th codiagonal is the main diagonal of @xmath25 ; an upper triangular matrix whose main diagonal is zero is called strictly upper triangular . if @xmath25 is strictly upper triangular and the @xmath258th codiagonal of @xmath25 vanishes for @xmath1556 , then @xmath1557 . the following lemma is obvious if @xmath72 ( the case of interest in the situation arising from @xmath1-dimensional topology considered in section [ sec : proof of the main theorem ] ) . [ lem : order ] suppose @xmath51 has characteristic @xmath1558 where @xmath291 and @xmath1559 . then @xmath1472 has exponent @xmath1558 . if @xmath25 is a non - zero strictly upper triangular @xmath1473-matrix over @xmath51 and some entry in the first non - zero codiagonal of @xmath25 is a unit in @xmath51 , then @xmath1560 has order @xmath1558 . note that every element @xmath722 of @xmath1561 has the form @xmath1543 where @xmath25 is strictly upper triangular ( in particular , @xmath1562 ) ; hence @xmath1563 since for @xmath1564 , the binomial coefficient @xmath1565 is divisible by @xmath1558 . now suppose in addition that @xmath1566 and some entry in the first non - zero codiagonal of @xmath25 is a unit in @xmath51 . say this entry occurs in the @xmath1067th codiagonal of @xmath25 , where @xmath1567 . then all entries in the @xmath1067th codiagonal of @xmath1568 are zero . since for every @xmath765 , @xmath1569 and a unit of @xmath51 has order @xmath1558 as element of the additive group of @xmath51 , we see that @xmath1570 has order @xmath1558 . let now @xmath3 be a finitely generated linear group , and suppose @xmath1571 is a collection of finitely generated abelian unipotent subgroups of @xmath3 , such that each @xmath1572 is maximal abelian or maximal unipotent , and @xmath1573 we write @xmath1572 additively ; also , the natural morphism @xmath1574 is an isomorphism , and in the following we identify these two groups using this map . we pick a family @xmath1575 of elements of @xmath1572 which form a basis for @xmath1576 . in the case that @xmath1577 we choose @xmath1578 to lie in @xmath1579 . in order to prove the theorem we may and shall assume that @xmath1580 for some @xmath1581 . take @xmath1582 such that @xmath1583 . since @xmath3 is finitely generated there is a finitely generated subring @xmath51 of @xmath1584 which contains all the entries of each element of @xmath3 , as well as all entries of @xmath1585 for @xmath1030 and the multiplicative inverses of the non - zero entries of @xmath1586 , for @xmath1030 . by theorem [ thm : proprmi ] there exists a finite set @xmath86 of primes such that for any prime @xmath1587 there exists a maximal ideal @xmath1494 of @xmath51 such that @xmath1588 is regular unramified of residue characteristic @xmath0 . we write @xmath1589 . as a consequence of the elementary divisors theorem for subgroups of finitely generated free abelian groups ( * ? ? ? * theorem iii.7.8 ) , after enlarging @xmath86 suitably if necessary , we may assume that for any @xmath1587 , any @xmath1590 and any @xmath1591 we have @xmath1592 in the case that @xmath1577 , the element @xmath1593 maps to zero under the map @xmath1594 . again , after enlarging @xmath86 if necessary , we may also assume that @xmath86 contains all primes @xmath157 . let now @xmath0 be a prime with @xmath1595 . let @xmath1494 be a maximal ideal of @xmath51 such that the local ring @xmath1588 , whose completion we denote by @xmath1596 , is regular unramified of residue characteristic @xmath0 . the natural ring morphisms @xmath1597 are injective , hence give rise to a group embedding @xmath1598 . we denote the map @xmath1599 by @xmath1600 , and the map @xmath1601 by @xmath833 . given @xmath1591 we denote the map @xmath1602 by @xmath1603 , and we denote the map @xmath1604 by @xmath1605 . for @xmath1591 we now let @xmath1606 we claim that the filtration @xmath1607 of @xmath3 is @xmath0-compatible with @xmath1449 of level @xmath163 . by the results in section [ sec : csg ] , @xmath111 is a separating normal @xmath0-filtration of @xmath3 with each @xmath1608 of finite index in @xmath3 , and considered as a complete filtration of @xmath346 , @xmath111 is central and ( strongly ) @xmath0-potent . to see that @xmath111 separates @xmath1449 , let @xmath1609 and @xmath1610 . if @xmath1611 is maximal unipotent , we have @xmath1612 ; since @xmath1613 there exists @xmath1591 such that @xmath1614 . if @xmath1572 is maximal abelian , then there is some @xmath1615 with @xmath1616\neq 1 $ ] ; as before , there is some @xmath1067 such that @xmath1617 ) = [ \ol{\rho}(g),\ol{\rho}(t)]$ ] in @xmath1618 . in both cases we obtain @xmath1619 , i.e. , @xmath1620 . it remains to show that @xmath1621 since @xmath1572 ( being unipotent ) is torsion free , we have @xmath1622 ; we need to show that @xmath1623 recall that @xmath1624 note that by choice of @xmath51 and @xmath1501 , lemma [ lem : order ] implies that the matrix @xmath1625 has order @xmath1626 in @xmath1618 . so in the case that @xmath1627 we clearly have @xmath1628 as desired . if @xmath1577 then @xmath1629 , whereas @xmath1625 has order @xmath1630 ; thus also in this case @xmath1628 . suppose in the proof of the theorem , in addition to @xmath1595 we also require that @xmath0 is such that @xmath352 is @xmath0-torsion free . then the filtration @xmath1631 of @xmath1466 is @xmath0-potent , by lemma [ lem:2-potent ] . moreover , by proposition [ prop : lower p ] , ( 2 ) we have @xmath1632 with equality for @xmath1633 . hence ( identifying @xmath3 with its image in @xmath1634 under @xmath1600 ) , the filtration @xmath1635 of @xmath3 defined by @xmath1636 is also @xmath0-compatible with @xmath1449 of level @xmath163 , cf . lemma [ lem : canonical filtration ] . in this section we give a proof of the main theorem as stated in the introduction , by applying the results of the previous section . we begin by first discussing the special cases of hyperbolic @xmath1-manifolds and seifert fibered manifolds . it is a well - known fact that the fundamental group of every hyperbolic @xmath1-manifold is linear and hence virtually residually @xmath0 , for all but finitely many @xmath0 . we need a more precise formulation of this fact : let @xmath25 be a @xmath1-manifold which is either closed or has incompressible boundary . let @xmath1449 be the collection of fundamental groups of the boundary components of @xmath25 , considered as subgroups of @xmath27 in the natural way . given a prime @xmath0 and an integer @xmath714 we say that a filtration @xmath111 of @xmath3 is a _ boundary compatible @xmath0-filtration of level @xmath1451 _ for @xmath25 if @xmath111 is @xmath0-compatible of level @xmath1451 with the collection of subgroups @xmath1637 , @xmath1455 . [ prop : virtphyp ] let @xmath25 be an orientable @xmath1-manifold with empty or toroidal and incompressible boundary such that the interior of @xmath25 has a complete hyperbolic structure . then for all but finitely many @xmath0 , there exists a boundary compatible @xmath0-filtration of level @xmath163 for @xmath25 . we need a topological lemma sketched in @xcite : [ lem : maximal abelian ] let @xmath25 be an orientable @xmath1-manifold with toroidal incompressible boundary , and let @xmath79 be a component of @xmath1638 . then @xmath1637 is a maximal abelian subgroup of @xmath26 , i.e. , no element of @xmath1639 commutes with @xmath1637 . suppose @xmath1640 and @xmath1637 generate an abelian subgroup @xmath214 of @xmath26 which properly contains @xmath1637 . since @xmath25 is not a @xmath1-torus , @xmath1641 by ( * ? ? * theorem 9.13 ) . denote by @xmath1642 the covering of @xmath25 corresponding to @xmath214 . there exists a compact submanifold @xmath722 of @xmath103 such that @xmath1643 is an isomorphism ( * ? ? ? * theorem 8.6 ) . we can assume that @xmath722 contains a homeomorphic lifting @xmath837 of @xmath79 in its boundary . since @xmath1637 is of finite index in @xmath214 and since @xmath25 is orientable , we can apply ( * ? ? ? * theorem 10.5 ) to conclude that @xmath722 is a product @xmath1644 $ ] with possibly some @xmath1-balls removed . in particular @xmath1645 is an isomorphism , which implies that @xmath1646 is an isomorphism , contradicting our assumption that @xmath1647 . now let @xmath25 be an orientable @xmath1-manifold with empty or toroidal and incompressible boundary . suppose moreover that @xmath25 is hyperbolic , so @xmath27 has a faithful representation to a discrete torsion free subgroup of @xmath1648 . thurston showed ( cf . , e.g. , ( * ? ? ? * proposition 3.1.1 ) or @xcite ) that this representation lifts to a discrete and faithful representation @xmath1649 ; we identify @xmath3 with its image under this embedding . let now @xmath79 be one of the ( toroidal ) components of @xmath1638 . standard poincar duality arguments yield that the natural image of @xmath1650 in @xmath1477 has rank @xmath1651 . every discrete abelian subgroup of @xmath1652 of rank @xmath34 is unipotent ( cf . hence @xmath1637 is unipotent ; in fact , by the previous lemma , @xmath1637 is a maximal unipotent subgroup of @xmath3 . proposition [ prop : virtphyp ] now is a consequence of theorem [ thm : virt res p for graphs of linear groups , 2 ] . in the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] in the previous section , given a finitely generated subgroup @xmath3 of @xmath1653 , we chose a certain finitely generated subring @xmath51 of @xmath1584 so that @xmath1654 , and we appealed to theorem [ thm : proprmi ] to obtain , for all but finitely many @xmath0 , a maximal ideal @xmath1501 of @xmath51 such that @xmath1655 for every @xmath1591 . above , this result has been applied in the context where @xmath3 is a lift from @xmath1648 of a finite covolume torsion free kleinian group . thurston observed that then the mostow rigidity theorem implies that @xmath3 is conjugate ( in @xmath1652 ) to a subgroup of @xmath1656 where @xmath14 is a number field . ( in fact , @xmath14 can be taken to be a quadratic extension of the trace field @xmath1657 of @xmath3 , see ( * ? ? ? * corollary 3.2.4 ) . ) hence in this case , the ring @xmath51 introduced in the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] may be assumed to be a subring of a ring @xmath1658 of @xmath837-integers of a number field @xmath14 . the ring @xmath1658 is not finitely generated in general . however , all but finitely many @xmath0 are unramified in @xmath14 and do not become units in @xmath1658 , and for those @xmath0 , if we choose the prime ideal @xmath1659 of @xmath1658 which contains @xmath0 , then @xmath1659 is maximal and @xmath1660 is regular unramified with finite residue field of characteristic @xmath0 . we may now carry out the rest of the argument in the proof above with @xmath51 and @xmath1494 replaced by @xmath1658 respectively @xmath1659 . thus , in this application of theorem [ thm : virt res p for graphs of linear groups , 2 ] one may avoid the use of theorem [ thm : proprmi ] at the cost of appealing to a non - trivial fact from hyperbolic geometry . in general , however , theorem [ thm : virt res p for graphs of linear groups , 2 ] can not simply be reduced to the number field case : there do exist finitely presented linear groups which admit no faithful representation as linear groups over global fields whatsoever @xcite . in the case of seifert fibered spaces we have : [ prop : seifert res p ] let @xmath25 be a seifert fibered space . then @xmath26 has a finite index subgroup which is residually @xmath0 for every @xmath0 . rhemtulla @xcite showed that every group which is residually @xmath0 for infinitely many @xmath0 is bi - orderable . hence the previous proposition also yields another proof of the result ( obtained in ( * ? ? ? * corollary 1.6 ) ) that fundamental groups of seifert fibered spaces are virtually bi - orderable . by similar arguments as the ones used below in the proof of this proposition one may also show the folklore result that the fundamental group of a seifert fibered space is linear over @xmath18 ; we will not pursue this here . [ lem : s1-bundles res p ] suppose @xmath1661 is an @xmath1662-bundle where @xmath57 is an orientable surface with @xmath1663 . then @xmath26 has a finite index subgroup which is residually @xmath0 for each @xmath0 . if @xmath25 has non - empty boundary , then @xmath57 also has non - empty boundary and we obtain @xmath1664 , hence the euler class of the @xmath1662-bundle @xmath1661 is trivial . we therefore conclude that @xmath25 is a trivial @xmath1662-bundle , i.e. , @xmath1665 . but then @xmath1666 is the direct product of @xmath18 with a free group , hence @xmath26 is residually @xmath0 for each @xmath0 . now assume that @xmath25 and hence @xmath57 is closed . the subgroup @xmath1667 of @xmath26 generated by a regular fiber @xmath384 is normal and infinite cyclic , and we have a short exact sequence @xmath1668 let @xmath1669 be the euler class of @xmath57 . a presentation for @xmath1670 is given by @xmath1671=t^e,\ \text{$t$ central}\right\rangle.\ ] ] let @xmath1672=u,\ \text{$u$ central}\right\rangle.\ ] ] one verifies easily that the finitely presented group @xmath7 is torsion free , @xmath1673 , and @xmath1674 ( hence @xmath7 is nilpotent ) . thus @xmath7 is residually @xmath0 for every @xmath0 ( * ? ? ? * theorem 2.1 ( i ) ) . let @xmath55 be the subgroup of @xmath3 generated by the @xmath1675 , @xmath1676 and @xmath1677 . the assignment @xmath1678 yields a morphism @xmath1679 . let also @xmath1680 be the composition @xmath1681 then @xmath1682 is an embedding . it is well - known that orientable surface groups , such as @xmath1683 , are residually free ( cf . @xcite ) and hence residually @xmath0 for each @xmath0 . this shows that the finite index subgroup @xmath55 of @xmath3 is residually @xmath0 for each @xmath0 . clearly we only have to consider the case that the fundamental group of @xmath25 is infinite . in that case it is well - known that @xmath25 is finitely covered by a @xmath1-manifold which is an @xmath1662-bundle over an orientable surface @xmath57 with @xmath1663 ( cf . now use lemma [ lem : s1-bundles res p ] . analogous to proposition [ prop : virtphyp ] we also have , for certain trivial @xmath1662-bundles : [ prop : seifert simple lower p - central ] suppose @xmath1665 where @xmath57 is an orientable surface with at least two boundary components , and @xmath27 . then for any @xmath0 there exists a boundary compatible @xmath0-filtration of level @xmath163 for @xmath25 . this follows immediately from lemma [ lem : p - compatible for free groups ] , observing that since @xmath57 has at least two boundary components , its fundamental group @xmath1684 is free , and for each @xmath1685 representing a boundary curve of @xmath57 , the subgroup @xmath1686 of @xmath98 generated by @xmath1687 is a free factor of @xmath98 . let @xmath57 be an orientable surface . it is a classical fact ( fricke ) that @xmath1688 has a faithful representation as a fuchsian group such that each boundary curve @xmath1687 of @xmath57 is represented by a parabolic transformation @xcite . that is , there is a faithful representation @xmath1689 where each @xmath1687 is represented by a unipotent matrix in @xmath1690 . every such @xmath1691 lifts to a representation @xmath1692 . in fact , one may choose @xmath1691 to have image in @xmath1693 ; this follows from teichmller theory ( * ? ? ? * theorem 1 ) ( see also ( * ? ? ? * section 4.9 ) ) . one may therefore also try to apply theorem [ thm : virt res p for graphs of linear groups , 2 ] to prove the proposition above ( even if @xmath1694 ) . however , this only yields the existence of a boundary compatible @xmath0-filtration of level @xmath163 for @xmath1665 for cofinitely many @xmath0 . let @xmath25 be a @xmath1-manifold . we have to show that for all but finitely many @xmath0 the group @xmath27 is virtually residually @xmath0 . [ [ reduction - to - the - case - of - orientable - and - closed - prime - manifolds ] ] reduction to the case of orientable and closed prime manifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + by gluing in @xmath1-balls if necessary we can always assume , without changing @xmath3 , that @xmath1638 has no spherical components . in this situation we have : [ lem : incomp ] suppose @xmath25 has no spherical boundary components . then there exist @xmath1-manifolds @xmath1695 whose boundaries are incompressible and without spherical components , and a free group @xmath57 such that @xmath1696 . let @xmath1697 be a compressible component . by dehn s lemma there exists a properly embedded disk @xmath1698 such that @xmath1699 is homologically essential . let @xmath1700 be the result of capping off the spherical boundary components of @xmath1701 by @xmath1-balls . if @xmath1700 is connected , then @xmath1702 , otherwise @xmath1703 where @xmath1704 , @xmath1705 are the two components of @xmath1700 . the lemma now follows by induction on the lexicographically ordered pair @xmath1706 since we have either @xmath1707 ( in the case that @xmath88 is not a torus ) , or @xmath1708 and @xmath1709 ( in the case that @xmath88 is a torus ) . combining lemma [ lem : incomp ] with the remark following corollary [ cor : trivial edge groups ] we see that it suffices to consider @xmath1-manifolds with incompressible boundary . suppose @xmath25 has non - empty incompressible boundary . then we consider the double @xmath1710 , which is a closed @xmath1-manifold . then the natural morphism @xmath1711 is injective , and since the property of being virtually residually @xmath0 restricts to subgroups we can from now on assume that @xmath25 is in fact closed . by going to a double cover , if necessary , we can and will also assume that @xmath25 is orientable . finally , by the remarks after corollary [ cor : trivial edge groups ] again it suffices to show that prime @xmath1-manifold groups are virtually residually @xmath0 for all but finitely many @xmath0 . [ [ sec : jsj ] ] the setup for the case of a closed orientable prime @xmath1-manifold + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + now suppose @xmath25 is a closed orientable prime @xmath1-manifold . if @xmath25 is hyperbolic or seifert fibered , it follows immediately from propositions [ prop : virtphyp ] respectively [ prop : seifert res p ] that @xmath3 is virtually residually @xmath0 for all but finitely many @xmath0 . assume that @xmath25 has a non - trivial jsj decomposition ; that is , there are pairwise distinct , incompressible embedded tori @xmath1712 ( @xmath1713 ) such that the components @xmath1714 of @xmath25 cut along @xmath1715 are either seifert fibered or algebraically atoroidal ( and thus , by geometrization , hyperbolic ) . we associate a graph @xmath362 to this decomposition of @xmath25 as follows : take a set @xmath1716 of @xmath1717 distinct vertices and an orientation @xmath1718 of its edge set ( with @xmath1719 ) , and for every @xmath258 choose distinct vertices @xmath1720 and @xmath1721 such that @xmath1722 and @xmath1723 and set @xmath1724 , @xmath1725 . we may describe @xmath25 as the topological realization of a graph @xmath1726 of based cw - complexes with underlying graph @xmath362 , whose vertex spaces are the @xmath1727 and whose edge spaces are the tori @xmath1572 . we let @xmath31 be the associated graph of groups with underlying graph @xmath362 ( cf . section [ sec : graphs of cw - complexes ] ) ; so @xmath1728 . [ [ sec : reduction to seifert simple ] ] reduction to seifert simple manifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the following we say that a closed orientable prime @xmath1-manifold is _ seifert simple _ if any seifert fibered component in its jsj decomposition is of the form @xmath56 for some orientable surface @xmath57 , where @xmath57 is either closed or has at least two boundary components . we now show that there exists a finite cover @xmath1642 such that @xmath103 is seifert simple . we first prove the following lemma which is a slight variation on well - known lemmas ( cf . , e.g. , @xcite , ( * ? ? ? * lemma 6 ) , or ( * ? ? ? * proposition 2.5 ) ) . [ lem : sfsq ] let @xmath722 be a seifert fibered manifold with non - empty incompressible boundary . let @xmath1729 be a prime with @xmath1730 . then there exists a finite index subgroup @xmath1731 of @xmath1732 with the following properties : for any torus @xmath1733 the group @xmath1734 is the unique characteristic subgroup of @xmath1637 of index @xmath1735 ; and the cover @xmath1736 corresponding to @xmath1731 is of the form @xmath56 for some orientable surface @xmath57 with @xmath1737 . let @xmath1729 be a prime with @xmath1730 . denote by @xmath1738 the base orbifold of @xmath722 and denote the boundary curves by @xmath1739 ( @xmath1591 ) . we denote by @xmath1740 the @xmath34-orbifold with underlying space @xmath1741 and cone angle @xmath1742 and finally we denote by @xmath1743 copies of @xmath1740 . we let @xmath1744 be the closed orbifold given by gluing to each boundary component @xmath1745 of @xmath1738 the @xmath34-orbifold @xmath1746 . note that @xmath1747 in particular the images of @xmath1739 are now elements of order @xmath1729 in the orbifold fundamental group @xmath1748 . since @xmath1749 we see that @xmath1744 is a good orbifold ( see ( * ? ? ? * theorem 13.3.6 ) ) , i.e. , there exists an epimorphism @xmath1750 to a finite group @xmath3 such that the corresponding finite cover @xmath1751 has the property that @xmath1752 is a surface . note that we can and will assume that @xmath1753 . the preimages of @xmath1746 are disks , and this implies that the image of @xmath1754 has order @xmath1729 . since @xmath1755 it follows that @xmath1756 has order @xmath1729 . also note that the preimage of @xmath1757 under @xmath453 is a surface @xmath57 . since @xmath1758 is surjective we see that @xmath1759 is connected . note that @xmath1760 we now return to the study of @xmath722 . first note that since @xmath722 has non - empty incompressible boundary we know that @xmath1761 is infinite and hence the subgroup @xmath1762 generated by a regular fiber is normal and infinite cyclic ( cf . note that there exists a canonical isomorphism @xmath1763 . we denote by @xmath1764 the map @xmath1765 defined as the composition @xmath1766 now let @xmath1767 be the cover corresponding to @xmath1764 . then @xmath1768 is a seifert fibered space with base orbifold the surface @xmath57 , i.e. , @xmath1768 is a @xmath1662-bundle over the surface @xmath57 ( see also ( * ? ? ? * theorem 11.10 ) ) . as in the proof of lemma [ lem : s1-bundles res p ] we conclude that @xmath1768 is in fact a product @xmath1769 . the boundary components of @xmath722 are given by @xmath1770 ( @xmath1771 ) , in particular @xmath1772 is the free abelian group generated by @xmath384 and @xmath1745 . also note that @xmath1773 is generated by @xmath384 and @xmath1774 . finally note that @xmath1775 . it is easy to see that the cover of @xmath1768 given by @xmath1776 gives rise to a cover of @xmath722 with the desired properties . we now let @xmath1730 be a prime such that all hyperbolic manifolds @xmath1727 admit a subgroup @xmath1777 such that for any boundary component @xmath1778 of @xmath1727 , @xmath1779 is the unique characteristic subgroup of @xmath1780 of index @xmath1735 . we can find such @xmath1729 and such subgroups by proposition [ prop : virtphyp ] . furthermore , given any @xmath1781 such that @xmath1727 is seifert fibered we now pick a subgroup @xmath1777 as in the above lemma . the collection of subgroups @xmath1782 is clearly a compatible collection of finite index subgroups of the graph of groups @xmath31 . we now let @xmath1783 be the finite cover arising from ( the proof of ) proposition [ prop : commoncover ] . recall that the manifold @xmath103 is in a canonical way the topological realization of a graph @xmath1784 of based cw - complexes where the vertex spaces are the components of @xmath1785 , @xmath1786 and the edge spaces are the components of @xmath1787 , @xmath1030 . it now follows that @xmath103 is seifert simple . hence , after replacing @xmath25 by @xmath103 if necessary , we may assume that @xmath25 is seifert simple . [ [ sec : conclusion ] ] conclusion of the proof + + + + + + + + + + + + + + + + + + + + + + + by the discussion above it remains to prove the main theorem in the case where @xmath25 is a closed orientable prime @xmath1-manifold which has a non - trivial jsj decomposition and is seifert simple . in this situation , by propositions [ prop : virtphyp ] and [ prop : seifert simple lower p - central ] , for all but finitely many @xmath0 , each jsj component of @xmath25 admits a boundary compatible @xmath0-filtration of level @xmath163 , and so by theorem [ thm : virt res p for graphs of linear groups , 1 ] , the group @xmath54 is virtually residually @xmath0 . in this section we finally give the proof of theorem [ thm : proprmi ] . for the reader s convenience we repeat the statement . in the following , `` ring '' will mean `` commutative noetherian ring with unit @xmath163 , '' and @xmath51 will always denote a ring with @xmath1788 . [ main - theorem ] [ mt ] suppose @xmath51 is a finitely generated integral domain of characteristic zero . then for all but finitely many @xmath0 there exists a maximal ideal @xmath1494 of @xmath51 such that @xmath1588 is regular unramified of residue characteristic @xmath0 . we first recall the concepts mentioned in the conclusion of the theorem . a sequence @xmath1789 ( where @xmath426 ) of elements of @xmath51 is said to be _ regular _ if @xmath1790 and for @xmath1791 , the canonical image of @xmath1792 is not a zero divisor of the ring @xmath1793 . for example , the sequence @xmath1794 of indeterminates in a polynomial ring @xmath1795 $ ] over a ring @xmath837 is a regular sequence . in general , regular sequences in @xmath51 do behave very much like indeterminates in a polynomial ring : if @xmath1789 is a regular sequence in @xmath51 , then the associated graded ring of the ideal @xmath1796 , @xmath1797 is a polynomial ring over @xmath1798 ; more precisely , the morphism of graded @xmath1798-algebras @xmath1799 \to \operatorname{gr}_i(r)\qquad\text{with $ x_i\mapsto x_i+i^2 $ for every $ i$}\ ] ] is an isomorphism ; in particular , @xmath1800 for every @xmath258 and @xmath343 ( * ? ? ? * theorem 16.2 ) . if @xmath1801 is a local ring , then @xmath51 is called regular if its maximal ideal @xmath1494 can be generated by a regular sequence ; in general , the ring @xmath51 is called regular if each localization @xmath1802 of @xmath51 at one of its prime ideals @xmath1803 is a regular local ring . if @xmath1801 is local of residue characteristic @xmath0 _ unramified _ if @xmath1804 . in general , we need a criterion to determine whether the localization @xmath1513 of @xmath51 at one of its maximal ideals @xmath1494 is regular unramified . this is provided by the following lemma ( a special case of theorem 14.2 in @xcite ) : [ criterion ] let @xmath1801 be a regular local ring , and let @xmath1805 , @xmath1806 . then the following are equivalent : 1 . @xmath1494 is generated by a regular sequence containing @xmath1807 ; 2 . @xmath1808 ; 3 . the local ring @xmath1809 is also regular . we equip , as usual , the set @xmath1810 of prime ideals of the ring @xmath51 with its zariski topology , with sub - basis given by the open sets of the form @xmath1811 recall that @xmath1812 for all @xmath1813 ; in particular any open subset of @xmath1810 contains an open set of the form @xmath1814 , @xmath1815 . also , @xmath1816 if and only if @xmath824 is nilpotent . we denote by @xmath1817 the regular locus of @xmath51 , that is , the set of all @xmath1818 such that the localization @xmath1802 is regular . it is well - known that @xmath1817 is open if @xmath51 is a finitely generated algebra over a field ( zariski ) or over a dedekind domain ( e.g. , @xmath18 ) of characteristic zero ( nagata ) ; see ( 6.12.5 ) respectively ( 6.12.6 ) of @xcite . recall that @xmath51 is said to be _ reduced _ if @xmath51 contains no non - zero nilpotent elements ; equivalently , if the intersection of all prime ideals of @xmath51 is trivial . clearly a domain is in particular reduced . we also need the notion of a _ jacobson ring _ ( also sometimes called a _ hilbert ring _ ) . this is a ring each of whose prime ideals is the intersection of maximal ideals . examples include all fields as well as every principal ideal domain with infinitely many pairwise non - associated primes ( like the ring of integers ) . if @xmath51 is a jacobson ring , then so is every finitely generated @xmath51-algebra @xmath837 , and the pullback of each maximal ideal of @xmath837 is a maximal ideal of @xmath51 . ( see ( * ? ? ? * iv.3.4 ) . ) [ regular maximal ideal ] suppose @xmath51 is a reduced ring whose regular locus is open . then @xmath1817 is non - empty , and if @xmath51 is jacobson , then @xmath1817 contains a maximal ideal . since @xmath51 is reduced , for every minimal prime ideal @xmath1803 of @xmath51 , the localization @xmath1802 is a field ( isomorphic to the fraction field of @xmath1819 ) , hence @xmath1820 . in particular , @xmath1817 is a non - empty open subset of @xmath1821 , and therefore contains an open subset @xmath1814 , for some non - zero @xmath1815 . if @xmath51 is jacobson , then the intersection of all maximal ideals of @xmath51 is zero ; hence there exists a maximal ideal @xmath1494 of @xmath51 not containing @xmath824 , and thus @xmath1822 . our final ingredient is the following fact , whose proof we postpone : [ non - trivial and reduced ] suppose @xmath51 is finitely generated . then @xmath51 has characteristic zero if and only if for all but finitely many primes @xmath0 , we have @xmath1823 . moreover , if @xmath51 is reduced then for all but finitely many primes @xmath0 the ring @xmath1824 is reduced . 1 . the converse of the second statement in the proposition is false : if @xmath1825/(4x)$ ] , then @xmath51 is not reduced , yet @xmath1824 is reduced for every prime @xmath1534 . note also that the analogous statement for the property of being an integral domain is also false , since there are polynomials in @xmath1826 $ ] ( e.g. , @xmath1827 ) which are irreducible over @xmath1828 but reducible modulo every prime . ( in fact , for each composite natural number @xmath20 there is a monic polynomial in @xmath1826 $ ] of degree @xmath20 with this property , cf . @xcite . ) 2 . after a first version of this manuscript had been completed , pascal adjamagbo informed us that in the course of investigations around the dixmier conjecture on weyl algebras , he also established a version of proposition [ non - trivial and reduced ] . assuming this proposition , let us outline how theorem [ main - theorem ] follows : suppose @xmath51 is a finitely generated integral domain of characteristic zero . by lemma [ regular maximal ideal ] and the remark preceding it we know that @xmath1817 is open and non - empty . in particular there exists a non - zero @xmath1815 such that @xmath1829 . the prime ideals in @xmath1814 are in one - to - one correspondence with the prime ideals of @xmath1830 $ ] , and the corresponding localizations of @xmath51 and @xmath837 are isomorphic . hence @xmath837 is regular . thus , replacing @xmath51 by @xmath837 , we may assume from the outset that @xmath51 is regular . let now @xmath0 be such that the ring @xmath1831 is non - trivial and reduced , as in the proposition above . let @xmath1494 be a maximal ideal of @xmath51 containing @xmath0 whose image @xmath1832 under the natural morphism @xmath1833 is in @xmath1834 . ( such @xmath1494 exists by lemma [ regular maximal ideal ] applied to the finitely generated @xmath1835-algebra @xmath1836 . ) note that @xmath1837 . therefore @xmath1588 is unramified regular of residue characteristic @xmath0 , by lemma [ criterion ] . in the proof of proposition [ non - trivial and reduced ] , we use results about dependence on parameters for certain properties of ideals in polynomial rings , formulated in lemmas [ member ] and [ radical ] below . the elements of the boolean algebra of subsets of @xmath1810 generated by the sets @xmath1814 with @xmath1815 , i.e. , the smallest collection of subsets of @xmath1810 which contains the sub - basis of @xmath1810 and which is stable under finite union and complementation ( and hence , also under finite intersection ) are called _ constructible . _ it is easy to see that each constructible subset of @xmath1810 is a finite union of sets of the form @xmath1838 where @xmath1839 . see ( * ? ? ? * chapter 2 ) for more on constructible sets , and ( * ? ? ? * section 12.2 ) on the connection with `` constructible properties '' exploited below . fix a tuple @xmath1840 of pairwise distinct indeterminates , and let @xmath1841 be another pairwise distinct tuple of indeterminates ( thought of as placeholders for parameters ) distinct from @xmath1842 . given a field @xmath14 and @xmath1843 we let @xmath1844 denote the kernel of the ring morphism @xmath1845\to k$ ] with @xmath1846 for every @xmath258 . note that for @xmath1847 $ ] we have @xmath1848 if and only if @xmath1849 in @xmath14 . more generally , a constructible subset @xmath88 of @xmath1850)$ ] expressed in the form contains @xmath1844 if and only if @xmath1687 satisfies the corresponding system of inequalities and equalities @xmath1851 for @xmath1852 and a prime number @xmath0 we set @xmath1853 , and we extend this notation to @xmath722-tuples @xmath1854 in the natural way : @xmath1855 then @xmath1848 if and only if @xmath1856 for all but finitely many prime numbers @xmath0 , and @xmath1857 if and only if @xmath1858 for all ( equivalently , all but finitely many ) primes @xmath0 . thus , we obtain the following principle for the reduction of constructible properties from characteristic zero to finite characteristic : [ reduction mod p ] let @xmath88 be a constructible subset of @xmath1850)$ ] , and let @xmath1859 . then @xmath1860 if and only if for all but finitely many prime numbers @xmath0 we have @xmath1861 . the following is a theorem by chevalley , also discovered by tarski ( and known among logicians as `` quantifier elimination for algebraically closed fields '' @xcite ) . [ chevalley ] let @xmath88 be a constructible subset of @xmath1862)$ ] , where @xmath210 is a new indeterminate . then there exists a constructible subset @xmath1863 of @xmath1864)$ ] such that for every field @xmath14 with algebraic closure @xmath1865 and every @xmath1866 , @xmath1867 see ( * ? ? ? * theorem 12.2.10 ) for a proof . ( the proof given there in fact shows how to construct , in an effective way , a description of the constructible set @xmath1863 from a description of @xmath88 . ) let @xmath1868 $ ] . for a tuple @xmath1869 we denote by @xmath1870 the ideal of @xmath1871 $ ] generated by the polynomials @xmath1872 . the next lemma ( applied with @xmath1873 ) says in particular that the property of @xmath1870 being a proper ideal is a constructible property of the parameters @xmath1687 : [ member ] there is a constructible subset @xmath88 of @xmath1850)$ ] such that for every field @xmath14 and @xmath1866 , we have @xmath1874 for the following lemma recall that a field @xmath14 with @xmath1875 is called _ perfect _ if @xmath1876 . note that @xmath76 is perfect . a field of characteristic zero is always referred to as perfect . we now have the following lemma . [ radical ] there is a constructible subset @xmath1877 of @xmath1850)$ ] such that for every perfect field @xmath14 and @xmath1866 , @xmath1878 let us first see how proposition [ non - trivial and reduced ] follows from the previous lemmas . suppose @xmath51 is finitely generated . we may assume @xmath1825/i$ ] for an ideal @xmath1879 of @xmath1826=\z[x_1,\dots , x_n]$ ] , for some @xmath25 . let @xmath1859 be the tuple of coefficients of the polynomials @xmath1880 ( in some order ) ; so we may write @xmath1881 where @xmath1882 $ ] ( with each non - zero coefficient of @xmath1883 equal to @xmath163 ) . now @xmath51 has characteristic zero if and only if @xmath1884 , if and only if @xmath1885=i(c , x)$ ] . moreover , @xmath51 is reduced if and only if @xmath778 is radical ; we have @xmath1886 = \sqrt{i\q[x]}$ ] , hence if @xmath778 is radical then so is @xmath1887 $ ] . similarly , for a prime @xmath0 , we have @xmath1823 if and only if the ideal @xmath1888 of @xmath1889 $ ] is proper , and @xmath1824 is reduced if and only if @xmath1888 is radical . proposition [ non - trivial and reduced ] is now a consequence of lemma [ reduction mod p ] applied to the constructible sets @xmath88 and @xmath1877 associated to the polynomials @xmath1890 and @xmath1891 in lemmas [ member ] and [ radical ] , respectively . the first part of proposition [ non - trivial and reduced ] ( which is related to grothendieck s ` generic freeness lemma ' ( * ? ? ? * theorem 24.1 ) ) may also be proved in different ways . for example , by the noether normalization lemma , there exists a non - zero integer @xmath1512 and an injective morphism @xmath1892[x ] \to r[\frac{1}{\delta}]$ ] , for some tuple @xmath1840 of indeterminates , such that @xmath1893 $ ] is a finite module over the image @xmath837 of @xmath154 . suppose @xmath0 does not divide @xmath1512 . then the natural morphism @xmath1894 induces an isomorphism @xmath1895 . moreover , @xmath0 generates a prime ideal of @xmath837 ; let @xmath1803 be a prime ideal of @xmath1896 with @xmath1897 . then the natural inclusion @xmath1898 yields an inclusion @xmath1889\cong s / ps \to r'/\pp$ ] ; in particular , @xmath1899 and thus @xmath1900 . another ( algorithmic ) proof of this part of proposition [ non - trivial and reduced ] , employing grbner basis theory , can be deduced from @xcite . it is also an immediate consequence of the non - trivial fact that the unit group of a finitely generated integral domain is finitely generated @xcite . we now turn to the proof of lemmas [ member ] and [ radical ] . these can probably be deduced from the results in of @xcite ; however , we prefer to follow a slightly less sophisticated approach , based on : [ vddries - schmidt ] there exists an integer @xmath1014 depending on the @xmath1883 with the following property : for every field @xmath14 and @xmath1866 , we have @xmath1901 if and only if there are @xmath1902 $ ] of degree at most @xmath1014 with @xmath1903 moreover , there exists an integer @xmath1904 such that for every field @xmath14 and @xmath1866 , if the implication @xmath1905 holds for all @xmath1906 $ ] of degree at most @xmath1904 , then the ideal @xmath1870 is radical . for a simple model - theoretic proof of this fact see @xcite . it is known , essentially since hermann s work @xcite ( see also @xcite ) , that @xmath1907 suffices , where @xmath343 is the maximal degree of @xmath1908 in @xmath538 ; in the important case @xmath1873 , by @xcite we may take @xmath1909 where @xmath1910 for @xmath1911 and @xmath1912 for @xmath1913 . analyzing one of the common algorithms to construct the radical of an ideal in a polynomial ring ( see , e.g. , @xcite ) also would allow one , in principle , to explicitly compute an integer @xmath1904 ( depending on @xmath25 and @xmath343 ) with the required property . by theorem [ vddries - schmidt ] , @xmath1901 if and only if a certain ( very large ) system of linear equations over @xmath14 , obtained ( uniformly in @xmath14 and @xmath1687 ) from the equation by comparing the coefficients corresponding to equal monomials in @xmath538 on both sides , has a solution in @xmath14 ( namely , the coefficient tuple of the polynomials @xmath1914 of degree at most @xmath1014 ) . that is , there exists a @xmath1915-matrix @xmath1916 with entries in @xmath1845 $ ] and a column vector @xmath1917^k$ ] ( for some @xmath1067 , @xmath1918 ) such that for every field @xmath14 and @xmath1866 , the ideal @xmath1870 contains @xmath1919 if and only if the system @xmath1920 of linear equations has a solution in @xmath14 , that is , if and only if @xmath1921 the rank of a matrix over a field is the least @xmath1922 such that all of its @xmath1923-minors vanish . it is now routine to construct a finite family @xmath1924 of systems as in such that for given @xmath14 and @xmath1687 , the equality holds if and only if @xmath1687 satisfies one of the @xmath1925 . using the `` rabinovich trick , '' lemma [ member ] yields : [ member - cor ] there is a constructible subset @xmath1926 of @xmath1850)$ ] such that for every field @xmath14 and @xmath1866 , we have @xmath1927 let @xmath362 be an indeterminate distinct from @xmath1842 . then for every @xmath14 and @xmath1687 we have @xmath1928 if and only if the ideal @xmath1929 of @xmath1930 $ ] contains the constant polynomial @xmath163 . we now turn to : given a field extension @xmath1931 of a perfect field @xmath14 , an ideal @xmath1932 of @xmath1933 $ ] is radical if and only if @xmath1934 $ ] is radical ( * ? ? ? * chapitre v , 15 , proposition 5 ) . hence it is enough to show the existence of a constructible subset @xmath1877 of @xmath1864)$ ] satisfying for every @xmath722-tuple @xmath1687 of elements of an _ algebraically closed _ field @xmath14 . let @xmath1935 $ ] be a polynomial of degree at most @xmath1904 with indeterminate coefficients ( where @xmath210 is a tuple of new indeterminates ) . using lemma [ member ] and corollary [ member - cor ] , let @xmath1936 , @xmath1937 be constructible subsets of @xmath1862)$ ] such that for every field @xmath14 and tuples @xmath1687 , @xmath343 of elements of @xmath14 of the appropriate lengths , @xmath1938 and @xmath1939 consider the constructible subset @xmath1940)\setminus\sigma_1\big ) \cup \sigma_2\ ] ] of @xmath1862)$ ] : by theorem [ vddries - schmidt ] , the ideal @xmath1870 is radical if and only if for every tuple @xmath343 of elements of @xmath14 we have @xmath1941 . one now obtains @xmath1877 from @xmath88 by applying theorem [ chevalley ] . we show proposition 1 in a more general form : [ prop : fibered ] let @xmath31 be a graph of finitely generated virtually residually @xmath0 groups such that for each edge @xmath372 of its underlying graph , the morphism @xmath489 is bijective . then the group @xmath30 is virtually residually @xmath0 . the _ mapping torus _ of a group endomorphism @xmath1942 is the group @xmath1943 more generally , given group endomorphisms @xmath1944 , the mapping torus of @xmath1945 is the group @xmath1946 here and below , @xmath1947 denotes the free group on @xmath338 generators . the proposition above immediately yields : [ cor : fibered ] the mapping torus of finitely many automorphisms of a finitely generated virtually residually @xmath0 group is virtually residually @xmath0 . this corollary readily implies proposition 1 : suppose @xmath25 is a fibered @xmath1-manifold , so @xmath1948 where @xmath1949 is the fundamental group of a surface @xmath57 . thus @xmath57 is either the free group @xmath1947 on @xmath338 generators ( for some @xmath338 ) or a surface group ( that is , the fundamental group of a closed surface ) . free groups are residually @xmath0 for all @xmath0 , and orientable surface groups are residually free @xcite and hence residually @xmath0 for all @xmath0 . every non - orientable surface group of genus @xmath453 contains a characteristic subgroup of index @xmath34 which is isomorphic to the orientable surface group of genus @xmath1950 . in any case , the group @xmath3 is virtually residually @xmath0 , for every @xmath0 , and hence so is @xmath26 by corollary [ cor : fibered ] . corollary [ cor : fibered ] can be seen as a more general and precise version of the result @xcite that the mapping torus of an automorphism of a free group of finite rank is residually finite . ( in fact , the mapping torus of an _ endo_morphism of a linear group is always residually finite , as shown by borisov and sapir @xcite ; they also showed that the mapping torus of an endomorphism of a finite - rank free group is virtually residually @xmath0 for all but finitely many @xmath0 @xcite . ) as remarked in @xcite , the linearity of the braid group @xmath1951 @xcite implies that the mapping torus of an automorphism of @xmath1952 is a linear group ; in fact , the holomorph of @xmath1952 is linear @xcite . it is unknown whether a mapping torus of an automorphism of @xmath1947 is always linear for @xmath1953 , cf . * problem 5 ) . for the proof of proposition [ prop : fibered ] we first note that it suffices to treat the special case of mapping tori : given a graph of groups @xmath31 such that for each edge @xmath372 of the underlying graph @xmath362 of @xmath31 , the morphism @xmath489 is bijective , for every maximal subtree @xmath79 of @xmath362 and every @xmath439 we have @xmath1954 and thus @xmath1955 , where @xmath1956 . so suppose now that @xmath1957 is the mapping torus of the automorphisms @xmath1945 of the finitely generated group @xmath3 . every finite index subgroup of the finitely generated group @xmath3 contains a finite index subgroup of @xmath3 which is fully characteristic ( even verbal ) in @xmath3 . thus , if @xmath3 is virtually residually @xmath0 , then @xmath3 contains a finite index residually @xmath0 subgroup which is fully characteristic . using proposition [ prop : commoncover ] we thus reduce to proving that `` @xmath3 residually @xmath0 '' implies `` @xmath357 virtually residually @xmath0 . '' the next lemma shows that our best chance for showing that @xmath357 is residually @xmath0 is by showing that the @xmath1031 act unipotently on @xmath1958 : [ lem : unipotent on homology ] let @xmath1959 be the mapping torus of the automorphisms @xmath1960 of @xmath1961 induced by @xmath1945 . then @xmath1076 is residually @xmath0 if and only if the automorphisms @xmath1960 generate a @xmath0-subgroup of @xmath1962 . if @xmath3 and @xmath1076 are residually @xmath0 , then so is @xmath357 . the first statement follows from the equivalence of ( 1 ) and ( 2 ) in corollary [ cor : chatzidakis ] . suppose @xmath3 and @xmath1076 are residually @xmath0 . to show that @xmath357 is residually @xmath0 , it is enough to show , by proposition [ hempel criterion residually p ] and lemma [ lem : hnn ] , that for each @xmath65 , the mapping torus of the automorphisms of @xmath242 induced by the @xmath1031 is residually @xmath0 . this follows from the first statement and lemma [ lem : induced automorphisms ] below . for every @xmath65 , by restriction we obtain a natural group morphism @xmath1963 . if @xmath1964 induces the identity on @xmath1965 , then @xmath1680 induces the identity on @xmath242 for each @xmath65 ( * ? ? ? * chapter viii , theorem 1.7 ) . therefore : [ lem : induced automorphisms ] if the image of a subgroup @xmath214 of @xmath39 in @xmath1966 is a @xmath0-group , then the image of @xmath214 in @xmath1967 is a @xmath0-group for every @xmath65 . suppose now that @xmath3 is residually @xmath0 . then by corollary [ cor : unfolded , 1 ] we see that the mapping torus @xmath357 has a finite index normal subgroup which is itself ( isomorphic to ) the mapping torus of finitely many automorphisms of @xmath3 , and such that the mapping torus of the corresponding induced automorphisms of @xmath1968 is residually @xmath0 . hence by lemma [ lem : unipotent on homology ] , @xmath357 is virtually residually @xmath0 . this finishes the proof of proposition [ prop : fibered ] . it may be instructive to contrast corollary [ cor : fibered ] with the following proposition , which shows that fundamental groups of fibered @xmath1-manifolds always satisfy , without passing to a finite index subgroup , a very weak variant of being residually @xmath0 . ( this proposition also fills in the details for a claim made in the introduction . ) [ prop : fibered implies weakly residually p ] let @xmath98 be the fundamental group of a fibered @xmath1-manifold . then for every @xmath1969 with @xmath1970 there is a subgroup @xmath1731 of @xmath98 of @xmath0-power index such that @xmath1971 . as above we write @xmath1972 where @xmath3 is a free group of finite rank or a surface group . then @xmath3 is residually @xmath0 . now , for any non - zero integer @xmath1067 we can consider the subgroup of @xmath1973 generated by @xmath1974 and @xmath3 , which we denote by @xmath1975 ; this subgroup is normal in @xmath98 of index @xmath1976 . for any @xmath65 we may also consider the subgroup of @xmath41 generated by @xmath384 and @xmath187 , denoted by @xmath1977 ; note that @xmath1977 is of @xmath0-power index in @xmath41 , and this subgroup is not normal in general . now suppose @xmath1978 where @xmath1979 , @xmath4 . if @xmath1980 , then there exists an @xmath20 such that @xmath1981 . if @xmath1982 and @xmath5 , then there exists an @xmath65 such that @xmath1983 . it is not surprising that in the special case of graph manifolds , our methods yield more refined results . we first restate and prove proposition 2 from the introduction : if @xmath25 is seifert fibered , then @xmath26 has a finite index subgroup which is residually @xmath0 for every @xmath0 , by proposition [ prop : seifert res p ] . so we may assume from now on that @xmath25 is not seifert fibered . the rest of the argument is as in sections [ sec : reduction to seifert simple][sec : conclusion ] in the proof of the main theorem : after passing to a finite cover of @xmath25 if necessary , we may assume that @xmath25 is seifert simple . then , by proposition [ prop : seifert simple lower p - central ] , for all @xmath0 , each jsj component of @xmath25 admits a boundary compatible @xmath0-filtration of level @xmath163 , and so @xmath3 is virtually residually @xmath0 by theorem [ thm : virt res p for graphs of linear groups , 1 ] . in the rest of this chapter we prove proposition 3 from the introduction . we also discuss @xmath0-cohomological completeness and obstacles towards a proof of virtual @xmath0-efficiency for arbitrary @xmath1-manifolds , and establish proposition 4 . 1 . @xmath54 is residually @xmath0 ; 2 . for all vertices @xmath74 and edges @xmath372 , the subgroups @xmath35 and @xmath55 of @xmath3 are closed in the pro-@xmath0 topology on @xmath3 ; 3 . for all vertices @xmath74 and edges @xmath372 , the pro-@xmath0 topology on @xmath3 induces the pro-@xmath0 topology on @xmath35 and on @xmath55 . so a closed orientable prime @xmath1-manifold @xmath25 is @xmath0-efficient precisely if its associated graph of groups is . for the proof of proposition [ prop : prop 2 restated ] , we first need to establish a criterion for @xmath0-efficiency of a graph of groups , which is corollary [ cor : p - efficiency criterion ] of the following sequence of lemmas . given a filtration @xmath111 of a group @xmath3 by @xmath0-power index normal subgroups , we say that @xmath111 induces the pro-@xmath0 topology on @xmath3 if the group topology on @xmath3 with fundamental system of neighborhoods of the identity given by @xmath111 agrees with the pro-@xmath0 topology on @xmath3 ; equivalently , for every @xmath0-power index normal subgroup @xmath7 of @xmath3 there is some @xmath260 with @xmath1984 . ( if @xmath177 is finitely generated , it suffices to require this for @xmath7 of the form @xmath1985 , @xmath65 . ) 1 . for each vertex @xmath74 , the filtration @xmath63 induces the pro-@xmath0 topology on @xmath35 ; 2 . for each edge @xmath372 , the filtration @xmath668 induces the pro-@xmath0 topology on @xmath55 ; and 3 . for every @xmath65 , the group @xmath1190 is residually @xmath0 . then the pro-@xmath0 topology on @xmath54 induces the pro-@xmath0 topology on each vertex and edge group of @xmath31 . if in addition @xmath111 is separating and separates the edge groups of @xmath31 , then @xmath31 is @xmath0-efficient . by condition ( 2 ) , the vertex groups induce the pro-@xmath0 topology on the edge groups , so it is enough to show that @xmath3 induces the pro-@xmath0 topology on each vertex group . fix a vertex @xmath386 and a normal subgroup @xmath1986 of @xmath1178 of @xmath0-power index ; we need to find a normal subgroup @xmath7 of @xmath3 of @xmath0-power index with @xmath1987 . by hypothesis ( 1 ) there is some @xmath65 such that @xmath1988 . denote the natural morphism @xmath1989 by @xmath247 . by ( 3 ) and since @xmath1990 is finite , there exists a normal subgroup @xmath14 of @xmath1190 of @xmath0-power index with @xmath1991 . then @xmath1992 is a @xmath0-power index normal subgroup of @xmath3 with @xmath1993 as desired . the last statement now follows from the first in combination with corollaries [ cor : hempel , cor ] and [ cor : vertex and edge groups closed ] . 1 . for each vertex @xmath74 , the filtration @xmath63 induces the pro-@xmath0 topology on @xmath1994 ; 2 . for each edge @xmath372 , the filtration @xmath668 induces the pro-@xmath0 topology on @xmath1995 . we argue as in the proof of theorem [ thm : reduction theorem , 3 ] . first we apply proposition [ prop : commoncover ] to @xmath1448 in order to reduce to the case that @xmath111 is complete . from now on we assume this to be the case . next , we apply corollary [ cor : unfolded , 1 ] with @xmath1996 and let @xmath91 have properties ( 1)(4 ) in that corollary . let @xmath1997 . then by theorem [ thm : reduction theorem ] , the group @xmath37 is residually @xmath0 , the vertex and edge groups of @xmath38 are closed in the pro-@xmath0 topology on @xmath1998 , and for every @xmath65 , the group @xmath1999 is residually @xmath0 . moreover , by condition ( 1 ) in corollary [ cor : unfolded , 1 ] , for each vertex @xmath1436 of the graph underlying @xmath38 , the filtration @xmath2000 induces the pro-@xmath0 topology on @xmath2001 , and for each edge @xmath2002 , @xmath2003 induces the pro-@xmath0 topology on @xmath2004 . thus by lemma [ lem : p - efficient ] , @xmath38 is @xmath0-efficient . [ cor : p - efficiency criterion ] suppose that all edge groups @xmath55 are abelian @xmath0-torsion free . assume moreover that @xmath111 is @xmath0-compatible of some level @xmath714 and for each vertex @xmath74 , the filtration @xmath63 induces the pro-@xmath0 topology on the subgroup @xmath1994 of @xmath35 . then there is a finite degree morphism @xmath610 of graphs of groups such that @xmath38 is @xmath0-efficient . we can now prove proposition [ prop : prop 2 restated ] . so let @xmath25 be a closed prime graph manifold . if @xmath25 seifert fibered , then there is nothing to do except to apply proposition [ prop : seifert res p ] . suppose @xmath25 is not seifert fibered . then , after passing to a finite cover of @xmath25 if necessary , we may assume that @xmath25 is seifert simple . in the proof of proposition [ prop : seifert simple lower p - central ] we showed that for every jsj component @xmath28 of @xmath25 , the filtration @xmath2005 of @xmath32 is a boundary compatible @xmath0-filtration of level @xmath163 for @xmath28 . clearly @xmath63 induces the pro-@xmath0 topology on @xmath2006 . hence by the previous corollary , @xmath25 admits a finite @xmath0-efficient cover . the usefulness of the notion of @xmath0-efficiency is to be found in connection with a description of the pro-@xmath0 completion of the fundamental group of a @xmath0-efficient graph of groups , which we give now . for simplicity we restrict our attention to graphs of finitely generated groups . for a ( discrete ) group @xmath3 we denote by @xmath2007 the pro-@xmath0 completion of @xmath3 , that is , the inverse limit of the system of @xmath0-group quotients of @xmath3 . the group @xmath2007 comes equipped with a topology making it a compact totally disconnected topological group , and we have a natural morphism @xmath2008 which is injective precisely if @xmath3 is residually @xmath0 . let first @xmath31 be a graph of finitely generated pro-@xmath0 groups , i.e. , a graph of groups all of whose vertex and edge groups are ( topologically ) finitely generated pro-@xmath0 groups , and all of whose edge morphisms are continuous . in this situation , the is defined to be the pro-@xmath0 completion of @xmath54 : @xmath2009 . we have a natural continuous group morphism @xmath2010 , were @xmath3 is equipped with the pro-@xmath0 topology . unlike the case of the ordinary fundamental group , the natural morphisms @xmath2011 are usually not injective ( cf . * example 9.2.9 ) ) . one says that @xmath31 is _ proper _ if all morphisms @xmath2011 are injective . now let @xmath31 be an arbitrary graph of ( abstractly ) finitely generated groups . suppose the pro-@xmath0 topology on each vertex group @xmath35 induces the pro-@xmath0 topology on @xmath675 for each edge @xmath372 with @xmath492 . ( this will be satisfied if @xmath31 is @xmath0-efficient . ) then the induced continuous group morphisms @xmath2012 between the pro-@xmath0 completions are injective , hence the graph of groups @xmath2013 with same underlying graph as @xmath31 , vertex groups @xmath2014 , edge groups @xmath2015 , and edge morphisms @xmath2016 , is a graph of finitely generated pro-@xmath0 groups as above . we have a natural morphism @xmath2017 of graphs of groups and hence a group morphism @xmath2018 . [ lem : p - efficient implies proper ] suppose @xmath31 is a @xmath0-efficient graph of finitely generated groups . then applying the pro-@xmath0 completion functor to the morphism @xmath2019 yields a continuous isomorphism @xmath2020 , and @xmath2013 is proper . this is well - known in the profinite case ; we only sketch the proof . first note that @xmath0-efficiency implies that the natural morphism @xmath2021 is injective . using the commutative diagram one sees that @xmath2022 induces the pro-@xmath0 topology on @xmath54 , and @xmath3 is dense in @xmath2022 . hence the continuous morphism @xmath2020 remains injective and has dense image in @xmath2023 . since @xmath2007 is compact and @xmath2023 is hausdorff , this image in fact equals @xmath2023 . hence @xmath2020 is an isomorphism . since @xmath3 induces the pro-@xmath0 topology on each @xmath35 , the natural injective morphisms @xmath725 give rise to injective morphisms @xmath2024 . thus @xmath2013 is proper . following @xcite we call a group @xmath3 _ cohomologically @xmath0-complete _ if the natural morphism @xmath2025 induces an isomorphism @xmath2026 for each @xmath65 . here , @xmath2027 is the usual ( discrete ) cohomology of @xmath3 , and @xmath2028 is the continuous cohomology ( sometimes called galois cohomology @xcite ) of the pro-@xmath0 group @xmath2007 . in both cases the action on @xmath76 is assumed to be trivial . cohomological @xmath0-completeness can be characterized in different ways : 1 . @xmath3 is cohomologically @xmath0-complete ; 2 . for every finite @xmath0-primary @xmath3-module @xmath722 and @xmath65 , the natural morphism @xmath2025 induces an isomorphism @xmath2029 ; 3 . for every finite @xmath0-primary @xmath3-module @xmath722 and @xmath65 , @xmath2030 4 . for all @xmath65 , @xmath2031 for the implication ( 1 ) @xmath332 ( 2 ) see ( * ? ? ? * lemma 4.7 ) . the equivalences ( 1 ) @xmath2032 ( 4 ) and ( 2 ) @xmath2032 ( 3 ) follow from ( * ? ? ? * exercise 1 , section i.2.6 ) . ( this exercise is stated in terms of profinite completion , but extends easily to the pro-@xmath0 case . ) see @xcite for more on this notion . a group which is cohomologically @xmath0-complete for every @xmath0 is said to be _ cohomologically complete . _ ( this class of groups is of interest in connection with atiyah s conjecture on the integrality of @xmath2033-betti numbers of compact manifolds with torsion free fundamental group @xcite . ) free groups are cohomologically complete , as are right - angled artin groups @xcite and primitive link groups @xcite . ( the argument in ( * ? ? ? * theorem 1.2.1 ) claiming that all link groups are cohomologically complete seems to have a gap , cf . @xcite . ) by induction on the nilpotency class , the following fact implies that finitely generated nilpotent groups are cohomologically complete : [ prop : seifert virtually cc ] let @xmath25 be a seifert fibered space . then @xmath26 is virtually cohomologically complete . if @xmath1665 for some orientable surface @xmath57 , then @xmath26 is cohomologically complete . as in the proof of proposition [ prop : seifert res p ] , after passing to a finite index subgroup if necessary , we may assume that @xmath25 is an @xmath1662-bundle over an orientable surface @xmath57 . if @xmath1665 for some orientable surface @xmath57 , then @xmath25 is already of this form . in any case , @xmath26 fits into an exact sequence @xmath2036 with ( the image of ) @xmath18 central in @xmath26 . proof of lemma [ lem : s1-bundles res p ] . ) now @xmath1683 is either free or an orientable surface group ; either way , @xmath1683 is a primitive one - relator group and hence cohomologically complete by ( * ? ? ? * exemple ( 2 ) , p. 144 ) . thus @xmath26 is also cohomologically complete , by proposition [ prop : cc and central extensions ] . on the other hand , it does not seem to be known which fundamental groups of hyperbolic @xmath1-manifolds are cohomologically @xmath0-complete . in @xcite ( with generalizations and different proofs in @xcite ) it is shown that if @xmath3 is the fundamental group of a closed hyperbolic @xmath1-manifold which is not virtually haken ( i.e. , a hypothetical counterexample to thurston s virtually haken conjecture ) , whose pro-@xmath0 completion is infinite , then @xmath3 is cohomologically @xmath0-complete . we have : this complements a result by wilton and zalesskii @xcite who showed that every graph manifold group @xmath3 is _ good _ in the sense of serre ( * ? ? ? * section 2.6 , exercises ) , that is , with @xmath805 denoting the profinite completion of @xmath3 , for every finite @xmath3-module @xmath722 and @xmath65 , the natural morphism @xmath2037 induces an isomorphism @xmath2038 . for the proof of proposition [ prop : virt cohomological p - completeness for graph manifolds ] we use the following well - known fact , a consequence of lemma [ lem : p - efficient implies proper ] ( cf . * proof of proposition 4.3 ) ) : suppose @xmath31 is a @xmath0-efficient graph of finitely generated groups with underlying graph @xmath362 and fundamental group @xmath54 . let @xmath401 be an orientation of @xmath362 and @xmath363 . then for every @xmath0-primary @xmath3-module there is a commutative diagram @xmath2039 & h^{n}(g;m ) \ar[r ] & \bigoplus\limits_{v\in v } h^n(g_v;m ) \ar[r ] & \bigoplus\limits_{e\in e_+ } h^n(g_e;m ) \ar[r ] & h^{n+1}(g;m ) \ar[r]&\cdots \\ \cdots\ar[r ] & h^{n}(\widehat{g};m ) \ar[r]\ar[u ] & \bigoplus\limits_{v\in v } h^n(\widehat{g_v};m ) \ar[r]\ar[u ] & \bigoplus\limits_{e\in e_+ } h^n(\widehat{g_e};m ) \ar[r]\ar[u ] & h^{n+1}(\widehat{g};m)\ar[u ] \ar[r]&\cdots}\ ] ] with exact rows . here the hats indicate pro-@xmath0 completion , the groups in the bottom row are continuous cohomology groups , and the vertical maps are induced by the natural embeddings of the groups into their pro-@xmath0 completions . suppose @xmath31 is a @xmath0-efficient graph of finitely generated groups . suppose all vertex and edge groups of @xmath31 are cohomologically @xmath0-complete . then @xmath54 is cohomologically @xmath0-complete . the results for graph manifolds in the previous sections naturally raise the question whether _ every _ closed prime @xmath1-manifold is , for all but finitely many @xmath0 , virtually @xmath0-efficient . ( if the answer was `` yes , '' then every closed prime @xmath1-manifold also would be , for all but finitely many @xmath0 , virtually cohomologically @xmath0-complete . ) now , by corollary [ cor : p - efficiency criterion ] , a positive answer to this question is implied by a positive answer to the following : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath25 be an orientable hyperbolic @xmath1-manifold with non - empty toroidal incompressible boundary . does @xmath25 , for all but finitely many @xmath0 , admit a boundary compatible @xmath0-filtration @xmath121 of some level which induces the pro-@xmath0 topology on @xmath2041 ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to investigate this issue more closely , let us return to the setting of the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] , in the case where @xmath72 and @xmath27 is the fundamental group of a @xmath1-manifold @xmath25 as in the question above , identified with a subgroup of @xmath1652 . we employ the modifications of this proof outlined in section [ sec : kleinians ] , in order to take advantage of the special properties of @xmath3 : after conjugating @xmath3 suitably in @xmath1652 we may and shall assume that @xmath3 is defined over a ring @xmath51 of @xmath837-integers of a number field . suppose @xmath1494 is a maximal ideal of @xmath51 such that @xmath1513 is regular unramified with finite residue field of characteristic @xmath0 . for all but finitely many @xmath0 there will be such @xmath1494 . then @xmath1596 is a finite ring extension of @xmath342 . let @xmath86 be the finite set of exceptional primes as defined in the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] . in particular , @xmath1589 is @xmath0-torsion free for @xmath1595 . we note that long and reid @xcite showed , as a consequence of chebotarev s density theorem and the availability of a complete list of all subgroups of @xmath2042 ( cf . * section 3.6 ) ) : in the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] , given @xmath1595 , we constructed a boundary compatible @xmath0-filtration @xmath2046 of level @xmath163 for @xmath25 by setting @xmath2047 in particular we have @xmath2048 now the following three possibilities for answering the question posed above present themselves : 1 . _ show that @xmath111 induces the pro-@xmath0 topology on @xmath346 . _ ( if so , we would be done . ) 2 . _ show that @xmath3 induces the pro-@xmath0 topology on @xmath346 . _ ( if so , by remark ( 2 ) following the proof of theorem [ thm : virt res p for graphs of linear groups , 2 ] , the filtration @xmath2049 of @xmath3 defined by @xmath2050 would be a boundary compatible @xmath0-filtration of level @xmath163 for @xmath25 which clearly induces the pro-@xmath0 topology on @xmath2051 . ) _ show that @xmath2052 is @xmath0-torsion free . _ ( if so , @xmath346 would be @xmath0-potent , hence by lemma [ lem : canonical filtration , 2 ] , @xmath2053 would be a boundary compatible @xmath0-filtration of level @xmath163 for @xmath25 . ) this follows from an important result of lubotzky ( * ? ? ? * theorem 2.3 ( b ) ) : if @xmath98 is a finitely presentable group whose @xmath0-adic completion is analytic and @xmath2056 , then its deficiency @xmath2057 ( number of generators @xmath2058 number of relations ) satisfies the inequality @xmath2059 where @xmath2060 . on the other hand , it is well - known ( cf . @xcite ) that @xmath2055 has deficiency @xmath163 . thus @xmath2055 is not @xmath0-adic analytic . suppose @xmath2064 . let @xmath2065 denote the finite regular cover corresponding to @xmath2066 , and let @xmath79 be a boundary component of @xmath1638 . then @xmath2067}{[t : g_1\cap t ] } \geq \frac{p^2(p^2 - 1)}{p^2}\geq 3.\ ] ] hence @xmath2068 by the previous lemma . if @xmath111 induced the pro-@xmath0 topology on @xmath346 , then the pro-@xmath0 completion of @xmath2069 would be @xmath0-adic analytic ( cf . the remark following the proof of lemma [ lem : linearity ] ) , contradicting theorem [ thm : lubotzky ] . let us now turn to the second approach . in order to show that @xmath3 induces the pro-@xmath0 topology on @xmath346 , it is convenient to refine the normal series @xmath2070 : @xmath2071 note that @xmath346 has @xmath0-power index in @xmath2072 and @xmath2072 is residually @xmath0 ; hence @xmath2072 induces the pro-@xmath0 topology on @xmath346 , by the following well - known fact ( cf . * lemma 3.1.4 ( a ) ) ) : let @xmath2073 be a group extension where @xmath214 is finitely generated , @xmath41 is residually @xmath0 , and @xmath2035 is a @xmath0-group . then @xmath41 induces the pro-@xmath0 topology on @xmath214 . thus it remains to show that @xmath3 induces the pro-@xmath0 topology on @xmath2072 . we do nt know whether this is the case . we do , however , have the following general fact ( inspired by ( * ? ? ? * proposition 4.23 ) ) , which might be of general interest . for this proposition and the lemmas used in its proof below , we fix a group extension @xmath2074 as a matter of convenience we shall assume that @xmath2075 and @xmath154 is just the natural inclusion map . note that @xmath41 acts on @xmath214 by conjugation , and this action descends to an action of @xmath2035 on @xmath2076 and hence on @xmath2077 . we say that @xmath2035 acts unipotently on @xmath2077 if the image of the associated morphism @xmath2078 is a @xmath0-group . [ prop : pro - p induced ] suppose @xmath214 is finitely generated , @xmath2035 is finite and perfect , and @xmath2079 is coprime to @xmath0 . suppose @xmath2035 acts unipotently on @xmath2077 . then @xmath41 induces the pro-@xmath0 topology on @xmath214 . here @xmath2080 denotes the schur multiplier of a finite group @xmath2035 . this proposition applies to extensions of @xmath2081 : if the sylow @xmath0-subgroups of a finite group @xmath2035 are cyclic for @xmath296 and each sylow @xmath34-subgroup of @xmath2035 is a generalized quaternion @xmath34-group , then @xmath2082 ( * ? ? ? * section 2.9 ) . this holds for @xmath2081 , for every @xmath0 ( cf . * section 3.6 ) ) . also recall that if @xmath2083 then the only proper normal subgroup of @xmath2042 is its center , hence @xmath2042 is perfect . we need to show that there exists a normal subgroup @xmath7 of @xmath41 of @xmath0-power index with @xmath2085 . we consider @xmath214 as a trivial @xmath2035-module . since @xmath2035 is perfect , we have @xmath2086 . now @xmath214 is a @xmath0-group and @xmath2079 is not divisible by @xmath0 , hence @xmath2087 . thus @xmath2088 , and this yields the existence of @xmath7 as desired . let @xmath2089 be the morphism arising from the action of @xmath41 on @xmath214 , and let @xmath14 be its kernel . then lemma [ lem : induced automorphisms ] and the assumption that @xmath2035 acts unipotently on @xmath2077 shows that @xmath14 has @xmath0-power index in @xmath41 . the normal subgroup @xmath2090 of @xmath2035 has @xmath0-power index , hence equals @xmath2035 since the latter group is perfect . thus we have a central group extension @xmath2091 by the previous lemma there exists a normal subgroup @xmath7 of @xmath14 of @xmath0-power index such that @xmath2092 . since @xmath41 is finitely generated , after replacing @xmath7 by its normal core in @xmath41 we may achieve that @xmath7 is normal in @xmath41 . we have to show : for every normal @xmath0-power index subgroup @xmath925 of @xmath214 , there is a normal subgroup @xmath7 of @xmath41 of @xmath0-power index with @xmath2093 . by replacing @xmath925 by its normal core in @xmath41 we may reduce to the case that @xmath925 is normal in @xmath41 . we thus obtain a group extension @xmath2094 where @xmath2095 is a finite @xmath0-group and @xmath2035 acts unipotently on @xmath2096 . by lemma [ lem : pro - p unipotent ] applied to this group extension there exists a @xmath0-power index subgroup @xmath2097 of @xmath2098 with @xmath2099 . then the preimage @xmath7 of @xmath2097 under the natural surjection @xmath2100 has the desired property . with @xmath2064 as in proposition [ prop : long - reid ] , @xmath3 and @xmath2072 fit into a short exact sequence @xmath2101 in order to be able to apply our proposition [ prop : pro - p induced ] , we would need @xmath2035 to act unipotently , and hence trivially , on @xmath2102 . however , this is not the case : [ prop : q acts non - trivially ] let @xmath722 be an orientable hyperbolic @xmath1-manifold with non - empty boundary , and suppose @xmath2103 be the cover corresponding to a surjective morphism @xmath2104 , where @xmath2105 . then @xmath2035 acts non - trivially on @xmath2106 . for the proof of this proposition we use the well - known fact ( cf . * chapter iv , corollary 6.8 ) and @xcite ) that for odd @xmath0 , the mod @xmath0 cohomology ring of @xmath2081 has the form @xmath2107/(y^{2}),\ ] ] equipped with the grading given by @xmath2108 , @xmath2109 . hence @xmath2110 we assume for a contradiction that @xmath2035 acts trivially on @xmath2106 . we have @xmath2111 as @xmath2035-modules . also , the natural morphism @xmath2112 is a surjective morphism of @xmath2035-modules , hence @xmath2035 acts trivially on @xmath2113 and thus , by poincar - lefschetz duality , on @xmath2114 . therefore the hochschild - serre spectral sequence of the short exact sequence @xmath2115 looks like @xmath2116 where both @xmath2117 and @xmath2118 are trivial @xmath2035-modules . using the description of @xmath2119 above we now see that since @xmath2105 , the @xmath76 with the coordinates @xmath2120 , @xmath2121 , in the bottom row of the second sheet @xmath2122 only get hit by the zero differentials , hence survive in @xmath2123 , contradicting the fact that @xmath2124 is finite - dimensional . this proposition shows that in order to successfully complete approach ( 2 ) to answering our question , i.e. , to show that @xmath3 induces the pro-@xmath0 topology on @xmath346 , one would have to use other methods than proposition [ prop : pro - p induced ] . as to approach ( 3 ) , that is , to prove that @xmath2052 is @xmath0-torsion free , it may be useful to again refine @xmath2070 , this time as @xmath2125 this leads to two subquestions , given @xmath25 as above . 1 . are there only finitely many @xmath0 such that for the cover @xmath2126 corresponding to the natural group morphism @xmath2127 , the abelian group @xmath2128 has non - trivial @xmath0-torsion ? are there only finitely many @xmath0 such that for the cover @xmath2129 corresponding to the group morphism @xmath2130 , @xmath2131 has non - trivial @xmath0-torsion ? concerning subquestion ( 1 ) : currently available technology allows one to extract information about the prime - to-@xmath0-torsion in the homology of abelian @xmath0-power covers from algebraic data associated to @xmath3 , cf . , e.g. , @xcite ; however , little seems to be known about @xmath0-torsion in the homology of such covers . it is known that if @xmath722 is a knot complement in @xmath15 and @xmath2132 is a @xmath0-power cyclic cover of @xmath722 , then @xmath2133 has trivial @xmath0-torsion @xcite . moreover , by ( * ? ? ? * theorem 5.11 ) , if @xmath722 is the complement of a @xmath34-component link @xmath2134 in @xmath15 and @xmath0 does not divide the linking number of @xmath2135 and @xmath2136 , then for every abelian @xmath0-power index cover @xmath2132 of @xmath722 , again @xmath2133 has only trivial @xmath0-torsion . this leads us to conjecture that at least for link complements , subquestion ( 1 ) has a positive answer . as to ( 2 ) , we do not know the answer . we note that reznikov ( * ? ? ? * theorem 9.2 ) proved a result pointing to a negative answer ; however his result only holds for certain covers of _ closed _ @xmath1-manifolds : if @xmath722 is a closed @xmath1-manifold , and @xmath2137 the cover corresponding to a surjective group morphism @xmath2138 where @xmath2105 , and @xmath2139 , then @xmath2140 has non - trivial @xmath0-torsion . [ prop : injective on mod p homology ] every closed orientable prime @xmath1-manifold @xmath25 has , for all but finitely many @xmath0 , a finite cover @xmath103 such that for every jsj component @xmath104 of @xmath103 , the natural map @xmath105 is injective . in this subsection we let @xmath31 be a graph of finitely generated groups based on a graph @xmath362 , and we let @xmath79 be a maximal subtree of @xmath362 . we assume that for a given @xmath0 , we have @xmath2142 then the collection @xmath2143 of normal subgroups is compatible , and for each edge @xmath372 of @xmath362 , the edge morphism @xmath2144 ( where @xmath520 ) induces an injective morphism @xmath2145 we call the graph of groups @xmath2146 the _ mod @xmath0 homology graph _ of @xmath31 , and denote it by @xmath2141 . it has the same underlying graph @xmath362 as @xmath31 , vertex and edge groups @xmath2147 respectively @xmath2148 , and edge morphisms @xmath2149 . the fundamental groups of @xmath31 and of @xmath2141 have the same first mod @xmath0 homology : the natural surjections @xmath2150 and @xmath2151 define a morphism @xmath2152 of graphs of groups , hence yield a surjective morphism @xmath2153 which induces an isomorphism in homology mod @xmath0 , by the following lemma : [ lem : iso in homology ] let @xmath722 be a trivial @xmath3-module , and let @xmath1138 be a compatible collection of normal subgroups of @xmath31 such that for each @xmath439 and @xmath435 , the natural surjections @xmath2154 induce isomorphisms @xmath2155 then the natural surjective morphism @xmath2156 induces an isomorphism @xmath2157 this is shown by an easy application of the five - lemma in combination with the mayer - vietoris sequence of @xmath31 , cf . section [ sec : mv ] this fact together with the remark following lemma [ lem : homology and fiber sums ] yields : suppose there exists a morphism @xmath2158 to an elementary abelian @xmath0-group which is injective on the vertex groups of @xmath31 . then the natural inclusions @xmath725 yield injective morphisms @xmath2159 . [ lem : injective on homology ] suppose @xmath31 satisfies . then there exists a morphism of graphs of groups @xmath610 of finite degree such that for each vertex @xmath591 , the natural inclusion @xmath2160 yields an embedding @xmath2161 . we apply the same argument as in the proof of corollary [ cor : unfolded , 1 ] with @xmath2162 . we obtain a morphism @xmath1442 to a finite group , with corresponding unfolding @xmath1377 of @xmath1123 , and a morphism @xmath1443 to an elementary abelian @xmath0-group which is injective when restricted to each vertex group of @xmath1444 . let @xmath91 be the unfolding to @xmath31 along @xmath90 . since @xmath2163 , the graphs of groups @xmath1444 and @xmath2164 are naturally isomorphic , and by lemma [ lem : separating , 3 ] , @xmath38 also satisfies . thus the previous corollary applies to @xmath38 in place of @xmath31 . suppose all edge groups of @xmath31 are abelian @xmath0-torsion free , and suppose for some @xmath714 , @xmath31 admits a @xmath0-compatible filtration of level @xmath1451 . then there exists a morphism @xmath610 of graphs of groups such that the image of @xmath37 in @xmath30 has finite index , and for each vertex @xmath591 , the natural inclusion @xmath2160 yields an embedding @xmath2161 . let @xmath111 be a @xmath0-compatible filtration of @xmath31 of level @xmath1451 ; then @xmath111 is @xmath0-excellent . arguing as in the proof of theorem [ thm : reduction theorem , 3 ] , we let @xmath91 be a morphism of graphs of groups obtained by applying proposition [ prop : commoncover ] to the compatible collection of normal subgroups @xmath2165 of @xmath31 . then the filtration @xmath1997 of @xmath38 is complete ; in fact , @xmath2166 is a central @xmath0-filtration of @xmath38 . now , in the proof of theorem [ thm : reduction theorem , 3 ] we simply applied lemma [ lem : separating , 2 ] to show that @xmath2166 intersects to a uniformly @xmath0-potent filtration on the image of each edge group of @xmath38 under the edge morphisms . however , the somewhat more complicated lemma [ lem : separating , 4 ] shows that in our situation , these intersections are precisely the lower central @xmath0-filtrations of the images of the edge groups under the edge morphisms . by lemma [ lem : canonical filtration , 2 ] , the lower central @xmath0-filtration @xmath2167 of @xmath38 also has this property , i.e. , @xmath38 satisfies . so may now apply lemma [ lem : injective on homology ] to @xmath38 in place of @xmath31 to complete the proof . let @xmath25 be a closed orientable prime @xmath1-manifold @xmath25 . clearly we may assume that @xmath25 has a non - trivial jsj decomposition . as shown in section [ sec : reduction to seifert simple ] , after replacing @xmath25 by a finite cover if necessary , we may then reduce to the case that @xmath25 is seifert simple . then , by propositions [ prop : virtphyp ] and [ prop : seifert simple lower p - central ] , for all but finitely many @xmath0 , each jsj component of @xmath25 admits a boundary compatible @xmath0-filtration of level @xmath163 , and so by the previous lemma , the proposition follows . h. bass and a. lubotzky , _ linear - central filtrations on groups , _ in : w. abikoff et al . ( eds . ) , _ the mathematical legacy of wilhelm magnus : groups , geometry and special functions , _ pp . 4598 , contemp . 169 , amer . soc . , providence , ri , 1994 . p. e. conner and f. raymond , _ manifolds with few periodic homeomorphisms , _ in : h. t. ku et al . , _ proceedings of the second conference on compact transformation groups univ . massachusetts , amherst , mass . , 1971 , _ part ii , pp . 175 , lecture notes in math . 299 , springer - verlag , berlin , 1972 . l. van den dries , _ big cohen - macaulay modules in equal characteristic @xmath732 _ , london mathematical society lecture note series , vol . 145 , ch . 12 , pp . 221284 , cambridge university press , cambridge , 1990 . g. a. fernndez - alcober , i. v. kazachkov , v. n. remeslennikov , and p. symonds , _ comparison of the discrete and continuous cohomology groups of a pro-@xmath0 group _ , st . petersburg math . j. * 19 * ( 2008 ) , 961973 . s. friedl and s. vidussi , _ twisted alexander polynomials and fibered @xmath1-manifolds _ , in : m. usher ( ed . ) , _ low - dimensional and symplectic topology _ , proceedings of symposia in pure mathematics , vol . 82 , pp . 111130 , american mathematical society , providence , ri , 2011 . j. hempel , _ residual finiteness for @xmath1-manifolds _ , in : s. m. gersten and j. r. stallings ( eds . ) , _ combinatorial group theory and topology alta , utah , 1984 _ , pp . 379396 , ann . 111 , princeton univ . press , princeton , nj , 1987 . j. hillman , d. matei , and m. morishita , _ pro-@xmath0 link groups and @xmath0-homology groups , _ in : t. kohno and m. morishita ( eds . ) , _ primes and knots , _ pp . 121136 , contemp . 416 , amer . soc . , providence , ri , 2006 . r. kirby , _ problems in low - dimensional topology , _ in : w. h. kazez ( ed . ) , _ geometric topology , _ pp . 35473 , ams / ip studies in advanced mathematics , vol . 2.2 , american mathematical society , providence , ri ; international press , cambridge , ma , 1997 . scott and t. wall , _ topological methods in group theory , _ in : c. t. c. wall ( ed . ) , _ homological group theory proc . sympos . , durham , 1977 , _ pp . 137203 , london math . lecture note ser . 36 , cambridge univ . press , cambridge - new york , 1979 .	given a prime @xmath0 , a group is called residually @xmath0 if the intersection of its @xmath0-power index normal subgroups is trivial . a group is called virtually residually @xmath0 if it has a finite index subgroup which is residually @xmath0 . it is well - known that finitely generated linear groups over fields of characteristic zero are virtually residually @xmath0 for all but finitely many @xmath0 . in particular , fundamental groups of hyperbolic @xmath1-manifolds are virtually residually @xmath0 . it is also well - known that fundamental groups of @xmath1-manifolds are residually finite . in this paper we prove a common generalization of these results : every @xmath1-manifold group is virtually residually @xmath0 for all but finitely many @xmath0 . this gives evidence for the conjecture ( thurston ) that fundamental groups of @xmath1-manifolds are linear groups . dedicated to the memory of + sr . m. benigna fehlner ( 19222008 ) + who , as their teacher in kindergarten , first introduced the authors to each other .
beneath the scattered point sources that speckle the x - ray sky should lie a subtler , more continuous background of x - rays emanating from the hot gas that fills the spaces between galaxies . many , if not most , of the universe s baryons inhabit intergalactic space and are heated to temperatures of @xmath2 k by gravitationally - driven shocks ( e.g. , cen & ostriker 1999 ; dav 2000 ) . once heated , these baryons tend to settle into gravitational potential wells and assume the characteristic temperature of the dark matter that confines them . because the emissivity of these baryons depends on how severely they are compressed , the mean intensity of the continuous x - ray background reflects the amount of non - gravitational energy injected into intergalactic space . large amounts of energy injection by supernovae and active galactic nuclei can inhibit compression of the intergalactic medium , lowering the mean level of the continuous background . current estimates place the point - source contribution to the 0.5 - 2 kev background at @xmath3% ( hasinger 1998 ; mushotzky 2000 ; giacconi 2000 ) , meaning that hot intergalactic gas contributes less than 20% , which amounts to @xmath4 . a sizeable fraction of this residual emission comes from clusters of galaxies . integrating over the observed luminosity function of clusters , assuming no luminosity evolution , places the background from clusters hotter than 5 kev at @xmath5 . however , these clusters cover only @xmath6% of the sky . the remainder is covered by a confused patchwork of groups and intercluster filaments whose properties depend critically on non - gravitational processes . if gravitational processes alone were responsible for establishing the entropy distribution of intergalactic gas , emission from groups and filaments would vastly overproduce the remainder of the 0.5 - 2 kev background ( pen 1999 ; wu , fabian , & nulsen 2000 ) . somehow , the lowest - entropy , most compressible gas has been eliminated . non - gravitational heating has been the most widely studied way of establishing this entropy floor , but it is also possible that low - entropy gas has been removed by radiative cooling and subsequent condensation ( e.g. , bryan 2000 ) . analyzing individual groups and filaments to determine the impact of energy injection and cooling will not be easy . several factors complicate the task of compiling unbiased samples of x - ray emitting groups : groups are low surface - brightness objects , some apparent groups are chance superpositions of galaxies , and the potential wells of individual galaxies can strongly affect a group s x - ray properties ( see mulchaey 2000 for a review ) . projection effects further complicate matters . virialized objects with @xmath7 kev cover over a third of the sky , making individual filaments very difficult to isolate and creating a significant probability that one group will overlap another somewhere along the same line of sight ( voit , evrard , & bryan 2000 ) . for example , about three higher - redshift @xmath8 kev groups are expected to lie within the projected virial radius of a 0.5 kev group at @xmath9 . because of these projection effects , statistical analyses of the 0.5 - 2 kev surface brightness distribution should be explored as an alternative way to characterize the lowest surface - brightness structures in the x - ray sky and perhaps to gauge the impact of non - gravitational energy injection and cooling . we have begun to investigate this brightness distribution using both hydrodynamical simulations and semi - analytical techniques and report some of our early results in this letter . section 2 briefly describes our simulations of the x - ray surface brightness distribution and shows how the high - brightness end can be approximated with a semi - analytical model . section 3 demonstrates that non - gravitational processes strongly influence this distribution function , and 4 summarizes our results . our analysis of the x - ray background owing to intergalactic gas centers on the quantity @xmath10 , the probability that a given line of sight through the universe will have a 0.5 - 2 kev surface brightness @xmath11 . the ideal computation of @xmath10 would involve a hydrodynamical simulation encompassing the entire observable universe , but this is currently infeasible . instead , we have chosen to simulate a box large enough to contain a fair sample of clusters and groups , yet small enough to resolve the cores of these virialized objects . a companion paper ( bryan & voit 2001 ) describes in more detail these simulations , which employ an adaptive mesh refinement technique for the hydrodynamics ( bryan 1999 ; bryan & norman 1997 ; norman & bryan 1999 ) . here we will focus on simulations of a @xmath12 box with @xmath13 grid points and mesh refinement down to a minimum cell size of @xmath14 . we reconstruct @xmath10 for lines of sight from @xmath15 by computing @xmath16 for the box alone at a number of discrete redshift points to determine how @xmath16 varies with redshift for a given comoving box size . at any given redshift , the comoving size of our simulation box corresponds to a redshift interval @xmath17 , so we can reconstruct @xmath16 for the entire line of sight by appropriately convolving the individual distributions corresponding to each redshift interval ( see bryan & voit 2001 for more details ) . this procedure yields the distribution shown in figure [ sbplt_ps ] for a @xmath18cdm cosmology ( @xmath19 , @xmath20 , @xmath21 ) with a baryon fraction @xmath22 and without non - gravitational energy injection . because of the limited box size , we can not capture brightness enhancements owing to correlated structure on @xmath23 scales , but we anticipate that their effect on @xmath10 will be small . several features of this distribution are worth noting . first , the mean surface brightness is @xmath24 , about five times higher than allowed by observations , verifying the expectations of pen ( 1999 ) and wu ( 2000 ) . in addition , our numerical experiments reveal that @xmath25 depends on numerical resolution , implying that @xmath25 in an optimally resolved simulation would be even higher ( bryan & voit 2001 ) . second , @xmath26 at @xmath27 indicating that the median value of @xmath28 is close to the maximum allowed by observations . third , the quantity @xmath29 shows a broad peak between @xmath30 and @xmath31 , indicating that most lines of sight have a surface brightness broadly distributed in this range . finally , we show for comparison two @xmath10 distributions derived from the power - law fits to rosat and chandra point - source counts of hasinger ( 1998 ) and giacconi ( 2000 ) assuming gaussian point - spread functions with full - width at half - max of 1 arcsec and 10 arcsec . in both cases , diffuse hot gas dominates the surface - brightness distribution from @xmath30 to @xmath32 . an early hydrodynamical computation of x - ray surface brightness was performed by scaramella , cen , & ostriker ( 1993 ) , who computed the mean and variance of the specific intensity at 1 and 2 kev for a standard cdm cosmology using simulations of somewhat lower effective resolution . because of the different cosmological model and resolution limit , their results are difficult to compare directly with our own . however , we do verify their conclusions regarding the brightness distribution at intermediate brightness levels . scaramella ( 1993 ) found that the pixel distribution of specific intensity ( @xmath33 ) varies like @xmath34 , equivalent to @xmath35 . figure [ sbplt_ps ] illustrates this power - law slope , which is quite similar to that of our model in the neighborhood of @xmath36 to @xmath31 . this scaling stems from the @xmath0-model surface brightness distribution typical of clusters of galaxies . most clusters are adequately fit by the law @xmath37^{-3\beta+1/2}$ ] , where @xmath38 is the cluster s core radius . ( cavaliere & fusco - femiano 1978 ) . for an individual cluster we therefore have @xmath39}$ ] , which reduces to @xmath40 for @xmath41 , quite close to the slope found by scaramella ( 1993 ) . as figure [ sbplt_ps ] shows , this relation is also a good approximation of @xmath10 in the intermediate range . because the simulations produce clusters and groups that are well fit by @xmath0-models , we can successfully reproduce our simulated @xmath10 through semi - analytical means . following voit ( 2000 ) , we have taken the cluster catalog from the @xmath18cdm hubble volume simulation performed by the virgo consortium ( evrard 1999 ; macfarland 1998 ; frenk 2000 ) and have computed @xmath10 assuming @xmath41 surface - brightness profiles and the core radius - luminosity relation from jones ( 1998 ) . however , instead of using the observed @xmath42-@xmath43 relation , we assume @xmath44 , correct @xmath42 to the 0.5 - 2.0 kev band as described in bryan & norman ( 1998 ) for a metallicity of 0.3 solar , and normalize the relation to fit our simulated clusters at @xmath6 kev . figure 2 compares the resulting @xmath10 distribution with the hydrodynamical model and the @xmath10 derived from the hubble volume catalog using the _ rosat _ @xmath42-@xmath43 relation observed by markevitch ( 1998 ) instead of the one derived from the simulation . apparently , the contribution to @xmath10 from the virialized regions of clusters and groups can be adequately modeled by combining the cluster mass function with appropriate analytical equations relating luminosity to mass and to a cluster s surface - brightness profile . the excellent agreement between the simulated and semi - analytical @xmath10 distributions also simplifies the task of identifying the major contributors at each level of @xmath28 . the steep slope of @xmath10 at the surface - brightness levels of cluster cores ( @xmath45 ) echoes the steep slope of the cluster mass function . between @xmath46 the @xmath0-model outskirts of clusters and groups dominate the background , and below @xmath47 is the realm of true intercluster emission . our model without energy injection is illuminating but clearly does not represent reality , primarily because @xmath25 is far too high . applying the observed @xmath42-@xmath43 relation to the hubble volume clusters significantly shifts the @xmath10 distribution to lower @xmath28 ( see figure [ sbcomp ] ) with @xmath48 , on the verge of being disallowed by observations . however , this model relies on an uncertain extrapolation of the cluster @xmath42-@xmath43 relation down to group scales . in fact , the @xmath42-@xmath43 relation may steepen below @xmath49 , perhaps because supernova energy injection becomes comparable to the gravitational energy of the intragroup gas ( e.g. , heldson & ponman 2000 ) . in order to explore the effect of non - gravitational energy injection on @xmath10 , we have run a hydrodynamical model with a very simple prescription for preheating : we instantaneously add @xmath50 of energy per baryon at @xmath51 , similar to the level needed to explain the observed @xmath42-@xmath43 relation for clusters ( e.g. , ponman , cannon , & navarro 1999 ) . figure [ sbfbck ] shows the resulting @xmath10 in terms of the quantity @xmath52 , which peaks in the neighborhood of @xmath28 values that contribute most to the mean . the distribution has indeed shifted to lower @xmath28 , relative to the no - preheating case , and @xmath52 has also flattened , indicating that a larger range of @xmath28 contributes significantly to the mean . yet , the mean surface brightness , @xmath53 , still exceeds observational limits . furthermore , our numerical experiments indicate that even this value may be an underestimate ( bryan & voit 2001 ) . another crude but inexpensive way to way to explore the effects of non - gravitational processes is to apply an ad hoc entropy floor to the simulation results . the heat input required to explain the scaling properties of clusters corresponds to a minimum entropy level @xmath54 ( e.g. , ponman 1999 ) . interestingly , the same critical entropy level also emerges from cooling considerations : @xmath49 gas will cool and condense within a hubble time if its specific entropy level is @xmath55 . in either case , establishing an entropy floor lowers the mean gas density in the cores of groups and clusters . thus , we have chosen to mimic the effect of an entropy floor by recalculating @xmath10 after substituting the quantity @xmath56^{-3/2}$ ] for the original gas density . figure [ sbfbck ] shows the resulting @xmath52 distributions . for @xmath57 , we obtain @xmath58 , just barely consistent with current observations , and for @xmath59 , we obtain @xmath60 . croft ( 2000 ) have recently performed a similar computation using a smooth - particle hydrodynamics code that includes cooling and a prescription for supernova feedback and find @xmath61.cdm cosmology with @xmath62 . their value for the @xmath63 background from diffuse gas , @xmath64 , is even lower , partly because of the lower baryon fraction . ] because the simulation analyzed by croft includes multiple non - gravitational processes that ours do not , it is difficult to pinpoint the process most responsible for lowering their value of @xmath25 . quite possibly , removal of low - entropy gas by cooling and condensation could be just as important as supernova heating in establishing @xmath25 and the @xmath42-@xmath43 relation of groups and clusters ( see also bryan 2000 ) . it also remains possible that @xmath25 would be larger in a higher - resolution sph computation . additional simulations will be needed to clarify the roles of various non - gravitational processes and to relate the thermal history of intergalactic baryons to the @xmath10 distribution . given the difficulties in compiling an unbiased @xmath42-@xmath43 relation for groups , we suggest that observations of the @xmath10 distribution at 0.5 - 2 kev be pursued as an alternative way to constrain preheating and radiative cooling . the quantity @xmath52 is clearly sensitive to non - gravitational processes like energy injection and cooling , particularly around @xmath65 , and the background from hot gas is competitive with the point - source background in _ chandra _ and _ xmm - newton _ observations over the interesting range of @xmath28 ( @xmath66 ) . identification of extended sources and follow - up redshift surveys are unnecessary . however , the trick will be to distinguish the true astronomical background from non - astronomical detector events . we have derived the 0.5 - 2 kev surface - brightness distribution function @xmath10 from hydrodynamical simulations of structure formation in a @xmath18cdm cosmology . the bright end of this distribution is well reproduced by combining the cluster catalog from the @xmath18cdm hubble volume simulation with appropriate @xmath42-@xmath43 and @xmath38-@xmath42 relations and a @xmath67 surface - brightness law . without non - gravitational heating or cooling , @xmath25 exceeds observed limits by a factor of several . another simulation that adds @xmath68 of energy at @xmath51 substantially lowers @xmath25 , but not by enough . an entropy floor @xmath69 appears necessary to reconcile our simulations with observational limits . observations of @xmath10 with _ chandra _ and _ xmm - newton _ appear to be a promising way to constrain the thermodynamical history of this intergalactic gas . glb is supported by nasa through hubble fellowship grant hf-01104.01 - 98a from the space telescope science institute , which is operated under nasa contract nas6 - 26555 . this research has used data products made available by the virgo consortium . norman , m. l. , & bryan , g. l. 1999 , in numerical astrophysics : proceedings of the international conference on numerical astrophysics 1998 , eds . s. m. miyama , k. tomisaka , & t. hanawa ( boston : kluwer ) , p. 19	hot intergalactic gas in clusters , groups , and filaments emanates a continuous background of 0.5 - 2.0 kev x - rays that ought to be detectable with the new generation of x - ray observatories . here we present selected results from a program to simulate the surface - brightness distribution of this background with an adaptive - mesh cosmological hydrodynamics code . we show that the bright end of this distribution is well approximated by combining the cluster temperature function with a @xmath0-model for surface brightness and appropriate luminosity - temperature and core radius - luminosity relations . our simulations verify that the x - ray background from hot gas vastly exceeds observational limits if non - gravitational processes do not modify the intergalactic entropy distribution . an entropy floor @xmath1 , which could be established by either heating or cooling , appears necessary to reconcile the simulated background with observations . because the x - ray background distribution is so sensitive to the effects of non - gravitational processes , it offers a way to constrain the thermal history of the intergalactic medium that is independent of the uncertainties associated with surveys of clusters and groups .
two computational difficulties in simulating the emission of gravitational waves from black hole interactions are the existence of a singularities inside the black holes and a consistent treatment of the outer boundary so that the waveform of the gravitational radiation can be accurately extracted . fortunately , the observable waves emanate only from the region outside the black holes and can be simulated , without loss of physical content , in a domain from which the singularities have been excised by introducing inner boundaries inside the black holes . however , the treatment of this domain by cauchy evolution has had limited success . we present here a quasi - linear model excision problem whose mathematical analysis is simple enough to reveal some underlying computational pitfalls and the means to avoid them . in particular , we show that an accurate simulation of this model excision problem can be obtained by ( i ) blending together two evolution algorithms , one of which is well - posed near the outer boundary and the other well - posed near the inner boundary , where a superluminal evolution is employed , and ( ii ) incorporating discrete conservation laws which control spurious exponential growth . the results are aimed at numerical formulations of the einstein equation as second order quasi - linear wave equations , such as in recent harmonic approaches @xcite . the problem can be illustrated by considering the propagation of scalar waves on the fixed background geometry of a spherically symmetric schwarzschild black hole . a choice of coordinates commonly adopted in numerical work for excising the singular region are ingoing eddington - finkelstein coordinates @xmath0 , in which the schwarzschild spacetime metric @xmath1 is given by @xmath2 here , the spacetime is foliated by the spacelike cauchy hypersurfaces @xmath3 , which are used to carry out the numerical evolution . the radial coordinate extends from the location of the singularity at @xmath4 to the asymptotic region @xmath5 , to which the radiation propagates . the black hole is located at @xmath6 . in the excision strategy , the simulation is carried out in a region @xmath7 , where the inner boundary satisfies @xmath8 and the outer boundary ( introduced for the purpose of a finite coordinate range ) satisfies @xmath9 . the covariant wave equation for a scalar field @xmath10 is @xmath11 where @xmath12 and @xmath13 is the inverse metric . in the above coordinates , it takes the explicit form @xmath14 although this wave equation is non - singular and can be reduced to symmetric hyperbolic form in the region @xmath7 , it has some notable features which complicate a numerical treatment : * the appearance of the mixed @xmath15 derivative . this is an effect of the eddington - finkelstein coordinates in which the @xmath16 observers are moving relative to the time slicing . in the standard adm @xcite formalism , this gauge effect is described as a non - vanishing _ shift _ , described by the shift vector @xmath17 where @xmath18 * the change in sign of the coefficient of @xmath19 in passing across the black hole . this change in sign does not affect the hyperbolicity of the wave equation but it creates a horizon at @xmath6 across which waves can not pass to the exterior . for @xmath20 , the @xmath16-observers move on timelike worldlines ; for @xmath6 , these worldlines are lightlike ; and for @xmath21 , they are spacelike . thus , inside the horizon the @xmath22 direction becomes spacelike , i.e. an evolution based upon grid points with constant @xmath23 is superluminal inside the horizon . * the energy density on the cauchy hypersurface @xmath3 associated with a timelike vector @xmath24 is @xmath25 , where @xmath26 is the unit timelike normal to the cauchy hypersurface and @xmath27 is the energy momentum tensor of the scalar field @xmath10 . in this paper we use the conserved energy associated with the choice @xmath28 , defined by @xmath29 , whose flow leaves the schwarzschild metric invariant . however , the corresponding energy density @xmath30 is positive - definite and provides a norm for the scalar field only for @xmath20 , where @xmath31 is timelike . inside the horizon , @xmath31 is spacelike and @xmath32 represents momentum rather than energy density . this complicates the use of the energy method for establishing a stable algorithm in the interior of the horizon . the non - conserved energy associated with the choice @xmath33 does provide a global norm but we will not pursue that direction here . instead , we use mode analysis to treat the inner region . these features are further complicated by the consideration of proper boundary conditions for a numerical treatment . they are not peculiar to a second differential order treatment . in particular , all these features are prominent in a model excision problem based upon a first order symmetric hyperbolic treatment of the linear wave equation ( [ eq : swave ] ) carried out in cartesian coordinates @xcite . they are also prominent in another second order discussion of the excision problem @xcite , in which the energy norm associated with the choice @xmath33 is used to treat the stability of the evolution algorithm ( [ eq : hlinwave ] ) which we adopt for the superluminal case . we study the numerical simulation of the excision problem by introducing a model 1-dimensional system which retains all the geometric features of the full problem . in the domain @xmath34 , we consider the quasi - linear wave equation @xmath35 where the coefficients are smooth and satisfy @xmath36 and @xmath37 for @xmath38 and @xmath39 for @xmath40 . these conditions ensure that the @xmath41-foliation is spacelike , that the metric determinant satisfies the lorentzian requirement @xmath42 , that the outer boundary @xmath40 is timelike and that the inner boundary @xmath38 is spacelike . the shift vector has component @xmath43 . we also require , analogous to ( [ eq : shift ] ) , that @xmath44 at @xmath38 , which ensures that the inner boundary is oriented so that all characteristics leave the domain . there is an event horizon at @xmath45 across which waves can not pass in the outer direction . for initial data satisfying @xmath46 , the cauchy problem for this equation is well - posed . the non - linearity is introduced into ( [ eq : nlwave ] ) in order to model another computational difficulty related to the full einstein equations . when harmonic coordinates are introduced to fix the gauge freedom , the einstein equations reduce to quasi - linear equations for the components of the metric whose principle part is identical to ( [ eq : gnl ] ) . in these harmonic coordinates , there exists pure gauge solutions ( flat metrics ) @xmath47 where @xmath48 which describing traveling waves with profile @xmath49 . the simulation of ( [ eq : shgw ] ) is one of the standardized tests proposed in the `` apples with apples '' project @xcite , whose purpose is to compare different numerical treatments of the einstein equations . the accuracy of the simulation is complicated by the existence of exponentially growing gauge waves @xcite @xmath50 where @xmath51 which solve the same equations and can not be damped by dissipative techniques . for @xmath52 , the cauchy data for ( [ eq : shlgw ] ) approximate the cauchy data for the test gauge wave ( [ eq : shgw ] ) . the excitation of this exponential error mode can dominate the simulation after a relatively short time . an accurate long term simulation of ( [ eq : shgw ] ) requires controlling the excitation of ( [ eq : shlgw ] ) by some global method . excellent long term accuracy has been obtained for the gauge wave test ( with periodic boundaries ) using a code @xcite based upon a semi - discrete energy estimate derived at the linearized level by the _ summation by parts _ method @xcite . comparable results for the gauge wave test can also be obtained using analogs of the semi - discrete monopole conservation laws applied in sec . [ sec : qlinexcis ] to the quasi - linear wave equation ( [ eq : nlwave ] ) . see @xcite for details . this quasi - linear wave equation closely models the gauge wave problem . solutions of ( [ eq : nlwave ] ) can be used to construct gauge waves via ( [ eq : shgw ] ) . the traveling waves ( [ eq : tphi ] ) and ( [ eq : tlphi ] ) are examples . thus the accurate long term simulation of solutions to ( [ eq : nlwave ] ) can be complicated by the existence of neighboring solutions which grow exponentially in time , in precise analogy with the gauge wave problem . we analyze the quasi - linear excision problem in terms of its elementary pieces . a stable algorithm for an initial - boundary value problem must also be stable for the corresponding linearized problem with frozen coefficients . this problem is considered in sec . [ sec : lin ] . we consider three separate regions : ( i ) behavior of the interior evolution , ( ii ) behavior at a time - like boundary and ( iii ) behavior at a spacelike boundary . treatment of a time - like boundary in general relativity introduces yet another computational difficulty related to the constraints . in the harmonic formulation of the initial - boundary value problem , each component of the metric satisfies a quasi - linear wave equation whose principle part is identical to ( [ eq : wave ] ) . the only well - posed version of the initial boundary value problem which is known at present to preserve the harmonic constraints requires a dirichlet boundary condition for some metric components and a neumann boundary condition for the others @xcite . for the purpose of future application of our results to harmonic evolution in numerical relativity , we concentrate here on such dirichlet and neumann boundary conditions , although the treatment extends to the general maximally dissipative boundary condition . the spatial discretization of the wave equation in second differential order form has many advantages over first order formulations , as discussed in @xcite . these advantages have been recognized in recent second order harmonic approaches to numerical relativity @xcite . for the case of zero shift , a comprehensive treatment of the wave equation in second order form has been given for both dirichlet @xcite and neumann boundary conditions @xcite . a major challenge of the excision problem is not just dealing with a shift but with a shift that is superluminal . this introduces complications even in the absence of boundaries . standard implicit @xcite and explicit @xcite evolution algorithms for the second order wave equation with shift are unstable in the superluminal case . we treat this cauchy problem in sec . [ sec : cauchy ] . in sec . [ sec : tbound ] , we give a finite difference algorithm for the subluminal case which is stable for both dirichlet and neumann boundary conditions . we give a detailed analysis , since there is little treatment of the second order wave equation with shift in the literature . whereas treatment of the dirichlet condition carries over unchanged , a non - zero shift forces a revision in the treatment of the neumann condition . the algorithm for the subluminal case ( the outer algorithm ) is well suited for treating the outer boundary of the excision problem but becomes unstable inside the horizon , where the shift is superluminal . in sec . [ sec : sbound ] , we give an alternative finite difference algorithm ( the horizon algorithm ) which is stable in this region and admits a stable spacelike excision boundary . we establish stability by using the method of lines to reduce the wave equation to ordinary differential equations in time on a spatial grid and then applying two standard techniques : mode analysis and semi - discrete energy estimates supplied by the summation by parts technique @xcite . the energy method is important because it extends directly to the wave equation with non - constant coefficients in a region with inner and outer boundaries . we use the energy associated with the @xmath41-direction along which the spatial grid propagates . for the region inside the horizon , where this energy does not provide a positive - definite norm , we use mode analysis to establish a stable excision algorithm at the inner spacelike boundary . in sec . [ sec : linexcis ] we treat the linearized excision problem . this involves blending together the outer and horizon algorithms obtained in sec . [ sec : lin ] . the additional ingredient is the necessity of non - constant coefficients in order to model a domain with spacelike inner boundary and timelike outer boundary . the homogeneous dirichlet and neumann boundary conditions are constructed so that the energy of the semi - discrete system is conserved when the coefficients of the wave operator are time independent . in addition , the algorithm is constructed so that it conserves a monopole moment associated with the scalar field in the case of time - dependent coefficients . the algorithm is implemented as a numerical code using a fourth order runge - kutta time integration . two code tests of the linear excision problem are presented in sec . [ sec : ltests ] : one where the data describe a pulse completely inside the horizon ; the other where the data consist of an ingoing wave entering the outer boundary . the tests confirm the accuracy and stability of the finite difference algorithm . in sec . [ sec : qlinexcis ] , we treat the excision problem for the quasi - linear wave equation . although semi - discrete energy conservation is not considered in the nonlinear case , we construct the algorithm so that the scalar monopole moment remains conserved . studies of the gauge wave problem @xcite show that either energy conservation or monopole conservation are sufficient to suppress the exponential mode ( [ eq : tlphi ] ) . this is confirmed in the numerical tests of the quasi - linear excision problem presented in sec . [ sec : qltests ] . in sec . [ sec : disc ] , we discuss the application to the harmonic einstein equations of the algorithms developed here for the model excision problem . we consider the wave equation @xmath53 obtained from ( [ eq : nlwave ] ) by freezing the coefficients and making the linear approximation @xmath54 . ( where no confusion arises , we use the abbreviated notation @xmath55 , etc . ) we treat three separate problems : 1 . the interior cauchy problem with both signs of @xmath56 . the initial - boundary value problem in the domain @xmath57 with a timelike boundary at @xmath38 , i.e. @xmath58 . the initial - boundary value problem in the domain @xmath57 and a spacelike boundary at @xmath38 , i.e. @xmath59 . if we can construct stable difference approximations for these three problems then the combined difference approximation for the full linear excision problem is also stable and the rate of growth is bounded independently of the step size . the reduction of the initial boundary value problem to cauchy and halfplane problems is discussed in @xcite and in ch . 12 of @xcite . the cauchy problem for ( [ eq : linwave ] ) is well - posed for every fixed value of @xmath60 and @xmath61 . the fourier mode @xmath62 satisfies @xmath63 with solution @xmath64 where @xmath65 thus @xmath66 and there are no exponentially growing modes on a uniform spatial grid @xmath67 , consider the simplest second - order accurate finite difference approximation to ( [ eq : linwave ] ) , @xmath68 where @xmath69 , @xmath70 and @xmath71 . the discrete fourier mode @xmath72 satisfies @xmath73 with solution @xmath64 where @xmath74 for @xmath75 it follows that @xmath66 but for @xmath76 there are exponentially growing modes , e.g. @xmath77 for the shortest wavelength mode @xmath78 . in order to remedy this problem , we consider the alternative second order accurate finite difference approximation @xmath79 for which the discrete fourier mode satisfies @xmath80 with solution @xmath64 where @xmath81 now @xmath66 for all values of the coefficients , subject to @xmath61 . unfortunately , because the evolution algorithm ( [ eq : hlinwave ] ) involves a wider stencil than ( [ eq : olinwave ] ) it is difficult to analyze its behavior and design a clean algorithm near a timelike boundary . accordingly , we will use the _ outer algorithm _ ( [ eq : olinwave ] ) near a timelike boundary and the _ horizon algorithm _ ( [ eq : hlinwave ] ) in a region extending from the excision boundary to the exterior of the horizon . we consider the initial - boundary value problem for ( [ eq : linwave ] ) with @xmath82 in the domain @xmath83 . without restriction we can assume @xmath84 . thus we consider the system @xmath85 we can derive an energy estimate . for real functions , let @xmath86 denote the @xmath87-scalar product and norm . the energy of the system is @xmath88 integration by parts gives @xmath89 we obtain energy conservation @xmath90 if the boundary condition at @xmath38 is given by the dirichlet condition @xmath91 or the neumann condition @xmath92 note that the neumann condition ( [ eq : cneum ] ) involves the normal derivative @xmath93 in the rest frame intrinsic to the boundary . we now discuss the difference approximation @xmath94 corresponding to ( [ eq : cnorm ] ) we use the discrete scalar product @xmath95 using lemma 11.1.1 of @xcite , one can easily derive the summation by parts rules @xmath96 from ( [ eq : wnuconst ] ) , @xmath97 using ( [ eq : wdov ] ) with @xmath98 replaced by @xmath99 , we obtain @xmath100 using ( [ eq : wd - v ] ) with @xmath98 replaced by @xmath99 and @xmath101 , respectively , we obtain @xmath102 i.e. @xmath103 thus we obtain discrete energy conservation if we use the boundary condition @xmath104 or @xmath105 corresponding to second order accurate approximations to the dirichlet condition ( [ eq : cdir ] ) or the neumann condition ( [ eq : cneum ] ) . the neumann condition involves the _ ghost point _ @xmath106 , which we eliminate by means of ( [ eq : wnuconst ] ) to obtain @xmath107 we use ( [ eq : wpneum ] ) to update the boundary point via @xmath108 in the continuum problem , a homogeneous neumann boundary condition also leads to conservation of the monopole quantity @xmath109 this carries over to the conservation of the semi - discrete monopole quantity @xmath110 i.e. @xmath111 , when ( [ eq : imneum0 ] ) is satisfied . this semi - discrete monopole conservation is extended to the non - linear problem in sec . [ sec : qlinexcis ] , where it has proved effective at suppressing the long wavelength slowly growing mode ( [ eq : tlphi ] ) . we now treat the initial - boundary value problem for ( [ eq : linwave ] ) in the halfspace @xmath113 for the case @xmath114 with @xmath44 , the condition @xmath114 implies that the boundary @xmath38 is spacelike and the condition @xmath44 implies that the boundary is oriented so that the characteristics leave the halfspace . thus no boundary conditions are necessary in the continuum problem . since there is no energy estimate of type ( [ eq : etconst ] ) , we use mode analysis with frozen coefficients to formulate a stable discretization near the boundary . consider first the continuum system . we study bounded solutions of the form @xmath115 introduction of this ansatz into ( [ eq : linwave ] ) gives the ordinary differential equation @xmath116 thus @xmath10 has the form @xmath117 where @xmath118 are the solutions of the characteristic equation @xmath119 i.e. @xmath120 since @xmath121 all roots satisfy @xmath122 for @xmath123 and there are no bounded solutions . thus the kreiss condition ( see @xcite , ch . 10 ) is trivially satisfied and the problem is well posed . as difference approximation for the halfplane problem we use the horizon evolution algorithm ( [ eq : hlinwave ] ) , i.e. @xmath124 for @xmath125 . we need two extra boundary conditions to determine @xmath126 and @xmath127 . we use the extrapolation conditions @xmath128 now we study bounded solutions of type @xmath129 where @xmath130 are solutions of the characteristic equation @xmath131 * lemma * \(i ) ( [ eq : tilde ] ) has no solutions with @xmath132 \(ii ) there are exactly two solutions @xmath133 with @xmath134 \(iii ) for @xmath135 , @xmath136 _ proof . _ part ( i ) follows from the stability of the cauchy problem , established in sec . [ sec : cauchy ] . for @xmath137 real , as @xmath138 the solutions of ( [ eq : tilde ] ) with @xmath139 converge to the solutions of @xmath140 correspondingly , the solutions of ( [ eq : tilde ] ) with @xmath141 converge to the solutions of @xmath142 since the four roots @xmath143 of ( [ eq : tilde ] ) are continuous functions of @xmath137 and @xmath144 for @xmath145 there are exactly two roots @xmath146 with @xmath147 for all @xmath137 with @xmath148 , thus establishing ( ii ) . for @xmath149 we have @xmath150 i.e. @xmath151 therefore , @xmath152 a simple perturbation analysis shows that @xmath153 for @xmath154 @xmath155 , i.e. @xmath156 correspond to the solution of ( [ eq : smodech ] ) . since @xmath157 we have @xmath158 for @xmath135 and ( [ eq : zerlim ] ) follows . this proves the lemma . the lemma shows that @xmath159 to account for possible double roots we write ( [ eq : extrmode ] ) as @xmath160 which becomes @xmath161 here @xmath162 are determined by the boundary conditions ( [ eq : uextrap ] ) , i.e. @xmath163 the determinant of this system is @xmath164 we can now prove : * theorem. ( [ eq : doub ] ) , ( [ eq : sigma ] ) has only the trivial solution @xmath165 for @xmath166 . therefore the difference approximation ( [ eq : hlinwave2 ] ) , ( [ eq : uextrap ] ) is stable . _ by ( [ eq : det ] ) , we have @xmath167 or @xmath168 then , by ( [ eq : tilde ] ) , @xmath169 and , by ( [ eq : zerlim ] ) , @xmath170 which is a contradiction . therefore the kreiss condition is satisfied . in ch . 12 of @xcite , it is proved that the kreiss condition implies stability . * the ryabenkii - godonov condition states that there are no bounded solutions of type ( [ eq : doub ] ) for @xmath171 the kreiss condition requires that there are no bounded solutions for @xmath172 the ryabenkii - godonov condition is a necessary condition while the kreiss condition is a sufficient condition for stability . 12 of @xcite . ) * remark 2 . * mode analysis is closely connected with the laplace transform . therefore the stability estimates are expressed in terms of @xmath173 as explained in ch . 12 of @xcite , one can use these estimates to obtain standard energy estimates . * remark 3 . * the stability theory in @xcite and in ch . 12 of @xcite is only developed for first order systems . however , it generalizes directly to second order equations . also in @xcite , it is indicated how to prove that the principle of frozen coefficients holds . the approximation with smooth coefficients is stable if the system is strictly hyperbolic and the kreiss condition holds . the results in sec s ( [ sec : tbound ] ) and ( [ sec : sbound ] ) prove that the outer algorithm gives a stable difference approximation for the halfplane problem with timelike boundary and that the horizon algorithm gives a stable difference approximation for the halfplane problem with a spacelike boundary . in sec . ( [ sec : linexcis ] ) , these results will be combined to give a stable difference approximation for the strip problem with spacelike inner boundary and timelike outer boundary . for the time discretization , we use the method of lines . the spatial discretization reduces the problem to a large system of stable ode s @xmath174 introducing @xmath175 we obtain the first order system @xmath176 we solve the system numerically using a 4th order runge - kutta time integrator . the time step is limited by the cfl condition determined from the cauchy problem . we model a pulse of energy propagating into a horizon using the wave equation @xmath177 on the interval @xmath34 , where we set @xmath178 with @xmath179 providing a smooth monotonic transition between the inner and outer regions . the underlying metric is lorentzian and ( [ eq : lxwave ] ) is hyperbolic . the inner boundary is spacelike , the outer boundary is timelike and there is an event horizon at @xmath180 . the solution to ( [ eq : lxwave ] ) in the region @xmath181 has the exact form @xmath182 consisting of an ingoing wave @xmath183 and an outgoing wave @xmath184 . near the inner boundary all waves are ingoing , as can be see by freezing the coefficient @xmath60 to obtain the solution @xmath185 which is valid in the short wavelength limit . our goal is to simulate a wave @xmath183 incident on the boundary at @xmath40 which propagates across the horizon , while being partially backscattered . we also check that a pulse of compact support inside the horizon does not propagate into the exterior region . first we must modify the evolution - boundary algorithm to account for non - constant coefficients , as required for the excision problem . we consider the equation @xmath186 where @xmath187 and @xmath188 . the energy associated with @xmath189 is @xmath190 note that for @xmath191 , @xmath192 is not necessarily positive . for the case @xmath193 , energy conservation holds in the form @xmath194 where the boundary flux is @xmath195 with this generalization , the homogeneous dirichlet condition ( [ eq : cdir ] ) remains @xmath196 but the homogeneous neumann condition ( [ eq : cneum ] ) must be modified to @xmath197 we modify the outer algorithm to @xmath198 where @xmath199 the last two terms in ( [ eq : onclinwave ] ) are equal because of the identity @xmath200 but it is advantageous to express @xmath201 in a form which is manifestly reflection invariant . consider now semi - discrete energy conservation for the case @xmath193 . we modify the semi - discrete version of the energy to @xmath202 where @xmath203 it follows that @xmath204 following the procedure leading to ( [ eq : econs ] ) , the summation of this expression gives the semi - discrete flux conservation law @xmath205 we express the neumann condition ( [ eq : nchomneum ] ) in the second order accurate form @xmath206 where @xmath207 . at the boundaries , @xmath208 and @xmath209 contain ghost points outside the computational domain , which we eliminate via the relations @xmath210 then , requiring that the wave equation be satisfied at the boundary points , i.e. that @xmath211 , the conservation law ( [ eq : ncecons ] ) implies that the neumann boundary conditions @xmath212 are dissipative . energy conservation holds in the continuum theory only when the coefficients of the wave operator are time independent . however , in the continuum problem , conservation of the monopole quantity ( [ eq : monint ] ) holds for general time - dependent coefficients in the case of homogeneous neumann boundary conditions . this carries over to the conservation of the semi - discrete quantity @xmath213 when the neumann condition ( [ eq : phomneum ] ) has the correction @xmath214 which preserves second order accuracy . as a result of this correction , ( [ eq : ncimneumn ] ) and ( [ eq : ncimneum0 ] ) are modified to @xmath215 setting @xmath216 for @xmath217 , a straightforward calculation then gives @xmath218 thus , again requiring that @xmath211 , the boundary conditions @xmath219 imply @xmath111 . thus we have established that @xmath111 in the general case and @xmath220 when the coefficients are frozen in time . note that when the coefficients are not frozen in time , an energy estimate still applies provided @xmath221 are bounded functions , with @xmath222 . for simplicity , consider the case with periodic boundaries . then for the continuum problem ( [ eq : lcwave ] ) and ( [ eq : uenergy ] ) imply @xmath223 the inequality @xmath224 gives @xmath225 where the constant @xmath226 satisfies @xmath227 ; and @xmath228 where the constant @xmath229 satisfies @xmath230 . integration of ( [ eq : et ] ) then gives @xmath231 so that @xmath232 where @xmath233 is the initial energy . a straightforward calculation based upon the summation @xmath234 leads to a semi - discrete energy estimate analogous to ( [ eq : eest ] ) . finally , in the inner region , we modify the horizon algorithm ( [ eq : hlinwave ] ) to the non - constant coefficient form @xmath235 where @xmath236 . for the purpose of excision , we need a prescription for switching from the outer algorithm @xmath237 to the horizon algorithm @xmath238 . we introduce a smooth , monotonic blending function @xmath239 , satisfying @xmath240 for @xmath241 and @xmath242 for @xmath243 . referring to ( [ eq : onclinwave ] ) and ( [ eq : nchlinwave ] ) , we then use the blended evolution algorithm @xmath244 , where @xmath245 which reduces to @xmath201 near the outer boundary and to @xmath246 inside the horizon . when @xmath247 , the blended algorithm satisfies a semi - discrete energy conservation law . for periodic boundary conditions ( identifying the points @xmath248 and @xmath249 ) , this takes the form @xmath250 where @xmath251 with @xmath252 ( here , as before , @xmath32 is not necessarily positive inside the horizon . ) for the boundary conditions of the excision problem , the energy ( [ eq : ncde ] ) is modified in the same way , @xmath253 since the support of the blending function @xmath183 is isolated from the outer timelike boundary , the semi - discrete energy flux through the outer boundary remains given by the first term in ( [ eq : ncecons ] ) . in order to validate the linear algorithm a set of test - runs was performed where the outer boundary condition at @xmath40 was either dirichlet or neumann , while at @xmath38 the extrapolation conditions ( [ eq : uextrap ] ) were applied . the gridstep was chosen to be @xmath254 , @xmath255 with the time - step set to @xmath256 . the background coefficients were set according to ( [ eq : exc_background ] ) with @xmath257 so that @xmath258 differentiability results . in the first set of runs the initial data at @xmath259 were a @xmath260 pulse of compact support inside the horizon with @xmath261 and @xmath262 ^ 4 \quad \mbox{for } \quad x_0 \leq x \leq x_0 + 0.2 \label{eq : cpulse } \\ u(0,x ) & = & 0 \quad \mbox{elsewhere } , \nonumber\end{aligned}\ ] ] with @xmath263 . homogeneous dirichlet data @xmath264 was given at the outer boundary . at the initial location of the pulse inside the horizon , both characteristics point to the left . the condition @xmath261 implies that the initial data is the superposition of two pulses , whose amplitudes have opposite signs , which propagate along these two characteristics . this is clearly seen in fig . [ fig : lin.in.slides ] where at first the `` faster '' mode ( with small negative amplitude ) falls into the excised region , while the `` slower '' mode ( with positive amplitude ) propagates in the same direction . we also monitor the convergence factor for the error modes that have propagated outside the horizon , defined in terms of the @xmath87 norm over the interval @xmath265 as @xmath266 second order accuracy would imply @xmath267 since the analytic solution vanishes outside the horizon . in the region outside the horizon , the simulation consists purely of short wavelength error . at first there is a high convergence rate but after about 6 crossing times the short wavelength error is unresolved on the grid . the addition of dissipation to the evolution scheme greatly extends the time for which convergence holds in the outer region , as shown in fig . [ fig : lin.in.conv ] . dissipation is added by modifying ( [ eq : fo ] ) according to @xmath268 where we set @xmath269 and @xmath270 is a smooth function with @xmath271 in the interior and @xmath272 near both boundaries . .,title="fig:",width=302 ] .,title="fig:",width=302 ] as a function of time for the test run with an initial pulse inside the horizon . it is evident that the addition of dissipation is necessary for long term convergence.,width=321 ] in the second set of runs , the initial cauchy data were set to zero , @xmath273 , and a wave was introduced through the outer boundary @xmath40 by prescribing inhomogeneous neumann data @xmath274 figure [ fig : lin.out.slides ] shows snapshots of the evolution for a wave generated by boundary data @xmath275 consisting of a single pulse . the pulse enters the outer boundary in the incoming mode of ( [ eq : lsol ] ) with characteristic velocity @xmath276 , propagates across the blending region and leaves the grid at the excision boundary . the figure also shows the remnant signal when no dissipation is added to the code . after the initial pulse has entered the outer boundary , the neumann data is homogeneous and reflects any signal propagating to the right but signals propagating to the left can leave the system across the excision boundary . at @xmath277 , short wavelength noise fills the region inside the horizon . by @xmath278 , the short wavelength error has either propagated through the excision boundary or has been converted into longer wavelengths . at @xmath279 , no visible signal remains . the short wavelength noise at @xmath277 can be effectively eliminated by a small amount of dissipation . we checked long term global convergence by prescribing a periodic wave entering though the outer boundary with the inhomogeneous neumann data @xmath280 \right ) \big ] \right\|_{x=1 } , \mbox{for } \quad t\ge 1/150 . \label{eq : lin.test.periodicpulse}\end{aligned}\ ] ] again the initial cauchy data was set to zero . the @xmath260 boundary data ( [ eq : lin.test.periodicpulse ] ) is initially set to zero to provide @xmath281 consistency with the cauchy data . no artificial dissipation was added in this test . we measured cauchy convergence of the numerical solution by monitoring the convergence factor @xmath282 for the given data and grid sizes , the code displays second order accuracy to within less than one percent . figure [ fig : lin.long ] plots the convergence factor as well as a snapshot of the residual finite difference error at the end of the simulation , at @xmath283 . vs. time indicates good second order convergence up to the end of the simulation at @xmath283 . the quality of the long term performance is reinforced by the right graph which shows that , at the end of the simulation , the rescaled error profiles are in nearly perfect agreement . no artificial dissipation was introduced.,title="fig:",width=321 ] [ lb]@xmath284 [ l]@xmath285 vs. time indicates good second order convergence up to the end of the simulation at @xmath283 . the quality of the long term performance is reinforced by the right graph which shows that , at the end of the simulation , the rescaled error profiles are in nearly perfect agreement . no artificial dissipation was introduced.,title="fig:",width=321 ] we extend our algorithms to the quasi - linear wave equation ( [ eq : nlwave ] ) in such a way that the outer algorithm continues to obey a semi - discrete monopole conservation law in the case of homogeneous neumann boundary conditions . we again set @xmath236 and treat the case of time - dependent , non - constant coefficients . we generalize the outer algorithm ( [ eq : onclinwave ] ) to the finite - difference form @xmath286 and generalize the horizon algorithm ( [ eq : nchlinwave ] ) to @xmath287 we blend these algorithms using the quasi - linear version of ( [ eq : blend ] ) , i.e @xmath244 with @xmath288 as the boundary condition for the horizon algorithm at the spacelike excision boundary , we use the extrapolation conditions ( [ eq : uextrap ] ) , now written as @xmath289 similarly , as homogeneous dirichlet condition for the outer algorithm at a timelike outer boundary we retain the linear form @xmath290 . we formulate the homogeneous neumann condition for the outer algorithm to establish semi - discrete monopole conservation corresponding to the continuum conservation law @xmath291 we proceed as follows . for periodic boundary conditions , ( [ eq : oncnlinwave ] ) implies the semi - discrete conservation law @xmath292 where @xmath293 this conservation law can be extended to a homogeneous neumann boundary condition by adding boundary terms to ( [ eq : qper ] ) in the form @xmath294 we generalize the neumann condition ( [ eq : homneum ] ) to the non - linear form @xmath295 as a result of the nonlinearities , ( [ eq : nctimneumn ] ) and ( [ eq : nctimneum0 ] ) are modified to @xmath296 setting @xmath216 for @xmath217 , a straightforward calculation leads to the nonlinear version of ( [ eq : fluxcons ] ) , i.e. @xmath297 we implement the homogeneous neumann condition in the form @xmath298 then the requirement that @xmath211 leads to the semi - discrete monopole conservation @xmath111 . the excision tests performed with the quasi linear code were based upon the ones performed with the linear code , with the ( time independent ) coefficients of the wave operator having spatial dependence determined by ( [ eq : exc_background ] ) and with the same cauchy data or boundary data now prescribed for @xmath299 as were prescribed for the linearized field @xmath10 in sec . [ sec : ltests ] . the initial data for the first set of runs consisted of a pulse of compact support inside the horizon with homogeneous dirichlet data at the outer boundary . the initial peak amplitude of the pulse ( [ eq : cpulse ] ) was @xmath300 , putting it in the nonlinear regime . since the characteristic speeds determined by the background metric are the same as in the linearized case , the snapshots of the quasi - linear evolution shown in fig . [ fig : nonlin.in.slides ] are qualitatively similar to the linearized ones in fig . [ fig : lin.in.slides ] . there is no visible propagation of the signal into the region outside the horizon . the convergence factor @xmath301 for the error modes that have propagated outside the horizon is defined by replacing @xmath10 by @xmath299 in ( [ eq : convout ] ) . as in the linear case , the short wavelength error in the exterior region is unresolved on the grid . the plots in fig . [ fig : nonlin.in.conv ] show that dissipation must be added to the code to obtain long term convergence . .,title="fig:",width=302 ] .,title="fig:",width=302 ] vs. time . for the non - linear test run for an initial pulse inside the horizon . addition of dissipation is necessary for long term convergence.,width=321 ] in the second set of non - linear runs , the initial cauchy data were set to zero and a wave introduced through the outer boundary by prescribing inhomogeneous neumann data . figure [ fig : nonlin.out.slides ] shows snapshots of the evolution for a wave consisting of a single pulse entering the outer boundary . no artificial dissipation was used . the main signal behaves qualitatively similar to the linearized case in figure [ fig : lin.out.slides ] but the error modes behave differently . at @xmath277 the short wavelength error again fills the region inside the horizon but now long wavelength error has also been excited which extends to the outer boundary . this corresponds to mixing in the exponential modes @xmath302 which are exact solutions ( for any @xmath303 ) to the quasi - linear wave equation ( [ eq : nlwave ] ) in the region @xmath304 . these modes are consistent with the homogeneous neumann boundary condition ( [ eq : modneum ] ) in effect after the pulse has entered the system and @xmath275 vanishes . however , the mode ( [ eq : expmode ] ) is not excited with positive @xmath303 and decays by @xmath278 . no visible signal remains at @xmath279 . thus the discrete monopole conservation built into the quasi - linear system suppress exponentially growing long wavelength modes . . the left graph shows the propagation of the signal itself , while the right graph illustrates the behavior of the residual modes . the long wavelength mode , which is apparent at @xmath277 , decays by @xmath278 and no visible signal remains at @xmath279.,title="fig:",width=321 ] . the left graph shows the propagation of the signal itself , while the right graph illustrates the behavior of the residual modes . the long wavelength mode , which is apparent at @xmath277 , decays by @xmath278 and no visible signal remains at @xmath279.,title="fig:",width=321 ] a further test of long term performance is the introduction of a periodic wave through the outer boundary by prescribing the inhomogeneous neumann data ( [ eq : lin.test.periodicpulse ] ) . we measured cauchy convergence by monitoring the convergence factor @xmath305 ( [ eq : conv ] ) applied to the numerical solution for @xmath306 . figure [ fig : nonlin.long ] plots @xmath307 as well as a snapshot of the residual finite difference error at @xmath283 , when the simulation was ended . the code displays second order accuracy to better than 1 percent . no artificial dissipation was used in the test . indicates good second order convergence . this result is reinforced by the right graph where , at the end of the simulation at @xmath283 , the rescaled error profiles show a nearly perfect agreement . no artificial dissipation was used in this test.,title="fig:",width=321 ] [ lb]@xmath308 [ l]@xmath309 indicates good second order convergence . this result is reinforced by the right graph where , at the end of the simulation at @xmath283 , the rescaled error profiles show a nearly perfect agreement . no artificial dissipation was used in this test.,title="fig:",width=321 ] we have based successful simulations of the model excision problem for a 1-dimensional quasi - linear wave equation on evolution and boundary algorithms which were proved to be stable in the linear case . the quasi - linear algorithm incorporates a conserved scalar monopole moment which suppresses the excitation of long wavelength exponentially growing modes , such as ( [ eq : expmode ] ) , which are latent in the system when neumann boundary conditions are used . the global excision strategy involves matching an exterior evolution algorithm to a horizon algorithm for evolution of the interior region . the exterior algorithm admits simple implementation of an outer boundary condition and it obeys a discrete version of flux - energy conservation which guarantees stability . inside the horizon , this energy is not positive - definite and the exterior algorithm is unstable . the horizon algorithm is stable in the interior region and admits a stable extrapolation algorithm to update the spacelike excision boundary . the choice of horizon algorithm presented here is based upon centered differencing . other choices are possible in the model 1-dimensional problem . in particular , a horizon algorithm based completely upon one - sided differencing is stable and requires no extrapolation at the inner boundary . this would be a very attractive feature if it could be extended to the higher dimensional problem . preliminary investigations indicate that the computational algorithms presented here can be taken over in a fairly straightforward manner to the harmonic formulation of the full 3-dimensional gravitational problem . in that case the einstein equations reduce to to the quasi - linear form @xmath310 where @xmath311 consists of lower differential order terms which do not contribute to the principle part . these equations have a well - posed initial - boundary value problem when the boundary data is explicitly prescribed @xcite . the principle part of ( [ eq : he ] ) can be treated in the same way as a quasi - linear wave equation ( [ eq : wave ] ) for each component of the metric . from the point of view of designing a code , the two main differences from our model excision problem are ( i ) the the 3-dimensional nature of the full problem and ( ii ) the constraints which enter into the boundary conditions . the evolution algorithms readily extend to 3-dimensions . when @xmath311 is neglected , the flux conservative form of ( [ eq : he ] ) can be carried over to the 3 dimensional discretized system to obtain discrete analogues of the scalar wave monopole conservation . there results a conserved quantity representing a spatially averaged rate of growth for each metric component , which has monopole symmetry for the @xmath312 component , dipole symmetry for the @xmath313 components and quadrupole symmetry for the @xmath314 components . the exact form of the conserved quantities depends upon how the principle part is split off in conservative form , with ( [ eq : he ] ) just one possibility . see the discussion in @xcite for how a splitting analogous to the logarithmic form ( [ eq : nlwave ] ) for the quasi - linear scalar problem leads to accurate long term simulation of the gauge wave problem ; it remains to be seen if this splitting is effective in more general spacetimes . when @xmath311 is neglected and the metric coefficients in the wave operator are frozen , semi - discrete energy conservation also results for each metric component in the case of periodic boundary conditions . with a cubic grid boundary , the results for the quasi - linear wave equation generalize to dissipative dirichlet and neumann boundary conditions for the metric components . the chief complication is the application of neumann conditions at the edges and corners of the cube . here among possible approaches are the use of the _ summation by parts _ technique , as in @xcite , or the introduction of a smooth boundary which is treated by interpolation , as in @xcite . the constraints on the boundary data present another complication . the constraints on the system ( [ eq : he ] ) , which guarantee that the solutions satisfy einstein s equations , are the harmonic coordinate conditions @xmath315 where @xmath316 are explicit harmonic driving functions . these constraints are satisfied by the solutions of ( [ eq : he ] ) if a certain mixture of dirichlet and neumann boundary conditions are used for the metric components @xcite . in that case , the constraints are satisfied when homogeneous boundary data are given and the initial - boundary problem for einstein s equations is well - posed . in the case of inhomogeneous constraint - preserving boundary data it is not known whether the system remains well - posed . it should be emphasized that , in addition to the computational difficulty , there are analytic and geometric problems in treating black holes . there is the possibility of exponentially growing perturbations to the analytic problem in the inner region near the excision boundary . there is the nonlocal nature of the _ true _ ( null ) event horizon . although the spacelike apparent horizon can be traced out as the evolution proceeds , it is impossible to locate the null event horizon until the exterior evolution is complete . the matching of an inner horizon algorithm to an outer algorithm must be carried out across an artificial horizon which results from a given choice of shift , as in our model problem . in the black hole excision problem , such an artificial horizon could be defined in terms of the hypersurface across which the evolution goes from superluminal to subluminal . pretorius @xcite has recently obtained promising results using a second order harmonic code to simulate a black hole by means of excision . we are now in the process of applying the new techniques presented here to this problem .	we analyze the excision strategy for simulating black holes . the problem is modeled by the propagation of quasi - linear waves in a 1-dimensional spatial region with timelike outer boundary , spacelike inner boundary and a horizon in between . proofs of well - posed evolution and boundary algorithms for a second differential order treatment of the system are given for the separate pieces underlying the finite difference problem . these are implemented in a numerical code which gives accurate long term simulations of the quasi - linear excision problem . excitation of long wavelength exponential modes , which are latent in the problem , are suppressed using conservation laws for the discretized system . the techniques are designed to apply directly to recent codes for the einstein equations based upon the harmonic formulation .
characteristics of flare - productive sunspot groups have been studied by many authors ( zirin & tanaka 1973 ; hagyard et al . 1984 ; kurokawa 1987 ; zirin & liggett 1987 ; tanaka 1991 ; and kurokawa 1991 , 1996 ) . in these studies , the authors pointed out that sheared configuration of magnetic field and @xmath2-type configuration of sunspot group are essential characteristics for strong flare activities . from statistical studies , however , shi & wang ( 1994 ) found that 23 % of @xmath2-type sunspot groups produced x - class flares , and chen et al . ( 1994 ) found that even strong magnetic shear of more than 80-degree shear angle is not a sufficient condition to produce flares . on the other hand , tanaka ( 1991 ) suggested that an emergence of a twisted magnetic flux rope produces strong flares . kurokawa ( 1987 , 1991 , 1996 ) also concluded that strong flare activities occur whenever a magnetic shear is rapidly developed by a successive emergence of twisted magnetic loops . zirin & liggett ( 1987 ) showed that big flares occurred in @xmath2-groups which emerged as continents . these studies suggest that flare - productivity of a sunspot group depends on formation process of the @xmath2-type configuration or the magnetic shear . this processes can be examined by measuring of sunspot proper motions . a large active region observed in march 1989 ( noaa 5395 ) was a highly flare - productive sunspot group with @xmath2-type configuration . the region produced about two hundred flares including ten x - class flares . all x - class flares were observed in three particular regions . wang et al . ( 1991 ) studied this region using magnetograms , dopplergrams and monochromatic images mainly obtained at big bear solar observatory ( bbso ) and zhang ( 1995 ) also studied using magnetograms obtained at huairou solar observing station ( hso ) . we measured proper motions of sunspots in noaa 5395 by using h@xmath3 images obtained with the 60-cm domeless solar telescope ( dst ) at hida observatory , kyoto university in order to study the formation process of @xmath2-type configuration and magnetic shear in the region . in this study we demonstrate that twisted magnetic flux tubes successively emerged at the preceding edge of the great sunspot group and that they played an essential role in the production of strong flare activities of noaa 5395 . we describe the procedure of data analysis in sec . 2 and show the obtained results of sunspot proper motions in sec . 3 . in sec . 4 we discuss the cause of peculiar sunspot motions found at the leading edge of the active region and propose a morphological model of an emerging twisted magnetic flux bundle which is essential to the strong flare activity . section 5 is devoted for our conclusion . the active region noaa 5395 turned into the visible hemisphere on march 6 and turned off behind the western limb on march 19 . during this period , except on march 7 , 13 and 17 , we continuously observed the evolution of the region with dst . many h@xmath3 monochromatic images were sequentially obtained with the zeiss lyot filter of 0.25 @xmath1 passband and a nikon motor - drive film camera of dst in 10 wavelength ( _ i.e. _ h@xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 @xmath1 and @xmath9 ) . in this study we mainly used h@xmath10 @xmath1 images to measure sunspot proper motions . we digitized the film densities of the h@xmath10 @xmath1 images by a film scanner and analyzed them by using idl ( interactive data language ) software on a personal computer and unix . times and numbers of measured images are listed in table 1 . one example of the h@xmath10 @xmath1 images is shown in fig . 1 . we carefully identified individual sunspot umbrae of the group in successive images and gave numbers to them as shown in fig . 1b , where the signs f and f stand for the _ following _ magnetic polarity , p and p , the _ preceding _ polarity and the sign : means uncertainness . we used the magnetograph data supplied by okayama astrophysical observatory ( oao ) , bbso and hso for the determination of magnetic polarities of the sunspots . the field of view of the h@xmath3 filtergraph of dst is a circle of about 390 @xmath11 diameter . the heliocentric coordinate of the center of the circle is recorded for each observation . referring to the center of the circle , we determined the heliocentric coordinate of individual sunspot umbra in idl pictures . the umbra position was defined as the center of gravity position of the umbra , which is determined by drawing the density contour of each umbra in an idl picture . by means of coordinate transformation we finally obtained heliographic latitude and carrington longitude of each umbra . during this observation the pointing accuracy of dst was about 10 @xmath11 , and this limits the accuracy of absolute positions of umbrae . then we defined the center of gravity of a triangle made by three umbrae f1 , f3 and p1 as a reference point and determined the relative positions of all sunspot umbrae referred to this point . the mean heliographic latitude of the reference point averaged over all measured images is @xmath12 n and its standard deviation is 1.0 degree . by excluding the images whose deviation exceeds @xmath13 , we get @xmath14 n as the average latitude of the reference point . the longitudinal positions are significantly influenced by the differential rotation . tang(1981 ) gave the differential rotation rate for the sunspot groups at high ( @xmath15 ) latitude by the formula @xmath16 where @xmath17 is the sidereal rotation rate . we would like to know the synodic rotation rate . according to zirin ( 1989 ) , the synodic rotation rate ( @xmath18 ) and the sidereal rate ( @xmath17 ) differ by the earth s orbital velocity of 0.9865 deg / day ; @xmath19 and we find that the synodic rotation rate at the latitude of @xmath20 n is 12.59 deg / day . on the other hand , the heliographic longitude of the disk center ( @xmath21 ) decreased by the rate of 13.18 deg / day . considering the difference of above two values ( i.e. 12.59 deg / day and 13.18 deg / day ) , the carrington longitude of the reference point should change by the rate of 0.59 deg / day due to the differential rotation . we first determined the heliographic latitude and carrington longitude of the reference point for the image of 00:00 ut of mrach 11 , which are @xmath20 n ( latitude ) and @xmath22 l ( longitude ) respectively . then for all other images we calculated the carrington longitude of the reference point by correcting the value of 0.59 deg / day . for each sunspot umbra , we adapted a different correction rate for the differential rotation corresponding to its latitude . after these corrections for the differential rotation , we calculated velocities of proper motions of sunspot umbrae . the term of umbra velocity used hereafter in this paper means a velocity of proper motion . we noticed that the motion of f1 is quite different from those of other big umbrae ( f3 , f4 and p1 ) ; the umbra f1 moved southward or southwestward , while the others moved eastward or northwestward . the umbrae f3 , f4 and p1 moved toward east at velocities of @xmath23 km / s . the umbra f1 moved southwestward at a velocity of about 0.05 km / s and it suddenly accelerated after 12 march . the velocity of f1 reached at 0.11 km / s on march 14 . the other three big umbrae moved eastward at velocities of @xmath23 km / s . these velocities are approximately the same as those of previous works , in which sunspot proper motions were measured in various active regions ( van driel - gesztelyi & petrovay 1990 , van driel - gesztelyi _ et al . _ 1993 , van driel - gesztelyi _ et al . _ 1994 , schmieder _ et al . _ 1994 , and herdiwijaya , makita , & anwar 1997 ) . according to herdiwijaya et al . ( 1997 ) , the longitudinal velocities of sunspots varies with their zrich classes ; the average velocity of ghj classes ( i.e. decaying phase ) is smaller than that of ab classes ( i.e. emerging phase ) . the _ following _ polarity sunspots of zrich ab classes show the average longitudinal velocity of 88 m / s . on the other hand , the average longitudinal velocities of ef classes ( i.e. well - developed phase ) and ghj classes are 18 m / s and 21 m / s respectively . the average longitudinal velocity of _ following _ polarity spot f1 is 57 m / s for the period from 9 to 15 march . this means that the magnetic flux tube forming sunspot f1 is still emerging from the subphotospheric layer in our observation period . we also noticed conspicuous motions of small sunspot umbrae around the leading sunspot f1 . they successively appeared at the south and southeast edge of f1 and moved eastward . we examined their motions in details . by calculating the displacements of these small umbrae between successive frames of images in each day , we determined their velocities of proper motions . we also calculated their velocities relative to f1 ( relative velocities , hereafter ) . the results obtained only for images of good seeing are presented in fig . 2 , where arrows indicate the relative velocities of the small umbrae . on march 9 ( fig . 2a ) , small umbrae near f1 ( p1 and p2 ) moved southward and others ( p3 , p4 and p5 ) moved eastward . the relative velocities of these umbrae were all 0.15 km / s . on march 10 ( fig . 2b ) , small umbrae ( p:6 , f:1 ) near f1 moved southward at relative velocities of @xmath24 km / s , respectively . a @xmath25-polarity umbra ( p7 ) moved southeastward at a relative velocity of @xmath26 km / s . a @xmath25-polarity umbra ( p9 ) , which moved toward p1 , had a relative velocity of 0.20 km / s . a @xmath25-polarity umbra ( p8 ) moved eastward at a relative velocity of 0.10 km / s . small umbrae at the north side of f1 moved eastward along the channel between f1 and f2 . on march 11 ( fig . 2c ) , small umbrae at the east and the southeast sides of f1 moved eastward or northeastward at relative velocities of @xmath27 km / s . small umbrae at the north side and at the west side of f1 ( @xmath28-polarities ) also moved eastward or northeastward . on march 12 ( fig . 2d ) , small umbrae at the east and southeast side of f1 also moved eastward at much higher relative velocities of @xmath29 km / s . small umbrae ( @xmath28-polarities ) at the north side and at the northwest side of f1 also moved east or northeast along the north edge of f1 . we also calculated the relative velocities of proper motions of the small umbrae by using all available images for each day . the average velocities obtained in each day are given in fig . it is found that the two results of fig . 2 and fig . 3 agree with each other . from these results we summarize main characteristics of proper motions of small umbrae as follows . at the south and southeast edges of f1 , small umbrae successively emerged and moved toward p1 through the east side of f1 . appearing near f1 , they first moved southward , turned to the east and finally moved northeastward to approach p1 . it means they moved along curved trajectory like an outflowing vortex . such emergences and motions of small sunspots continued with their speed accelerated from 9 through 12 of march . spot p1 became bigger in its umbra area from 10 through 12 , march . wang et al . ( 1991 ) also found that @xmath25-polarity umbrae at the east side of f1 merged into p1 . we found that spot p2 newly formed on 14 mar at the region where small emerging sunspots or emerging magnetic flux successively converged from 12 through 14 march ( see fig . 1b ) . some small umbrae emerging at the south west edge of f1 ( i.e. f6 , f10 , f11 , f13 , f14 , and f15 ) first moved northward at the west side of f1 and moved eastward or northeastward along the north edge of f1 from mar 10 through 12 . they converged into a few larger umbrae at the northeast side of f1 on march 12 . by march 14 the northeast part of f1 has been separated from its leading part and moved northeastward to form f5 umbra . the most remarkable result found in the previous section is the vortex - like motions of small sunspot umbrae around the leading sunspot f1 . pairs of small umbrae of different magnetic polarities successively emerged at the leading edge of f1 ( i.e. at the south or southwest edges of f1 ) from 9 through 12 , march . these umbrae moved clockwise toward the southwest of p1 , where growth of p1 was observed from 10 through 12 , march and p2 was formed on march 14 . at the west and north sides of f1 , small umbrae moved anticlockwise and formed f5 by merging with some parts of decaying f1 . these motions are schematically summarized in fig . 4a . at the east side of f1 , small umbrae , most of which are @xmath25-polarities , moved clockwise and formed p2 . at the west and north sides of f1 , small umbrae , all @xmath28-polarities , moved anticlockwise and formed f5 by merging with some parts of decaying f1 . notice vortex - like lines and structures of small sunspot umbrae and penumbrae surrounding f1 spot in fig . 1a . wang et al . ( 1991 ) also found the same vortex - like motions of magnetic features in this region using magnetograph data . for the explanation of these peculiar vortex - like motions , we propose a schematic model of emerging twisted flux tubes given in fig . the model is characterized by the successive emergence of twisted flux tubes coiling around the main trunk of flux tube f1 . the bundle of flux tubes is twisted and each flux tube is also twisted . as parker ( 1955 ) pointed out , the magnetic buoyancy raises up the system of twisted magnetic - flux bundle . the successive emergence of the inclined and twisted flux bundle can explain the vortex - like motions of small sunspots both at the southeast and the northwest sides of f1 as shown in fig . 4a . in fig . 4b planes @xmath30 show the positions of the photosphere relative to the emerging flux tubes for the different days . the cross sections of flux tubes on each photospheric plane correspond to sunspots . as the coiling loops successively emerge , the small @xmath25-polarity and @xmath28-polarity sunspots can be seen to move clockwise and anticlockwise , at the southeast side and at the northwest side of f1 respectively . notice that small @xmath25-polarity sunspots seen on the planes @xmath31 are all converged into the sunspots p2 on @xmath32 and that @xmath28-polarity ones are converged into the sunspot f5 . these are very consistent with the observations summarized at the beginning of this section . in addition , it is interesting to notice another evolutional feature of f1 . as studied in section 3 , the southwestward motion of the umbra f1 suddenly accelerated after 12 march . we observed the drastic change of the shape of f1 at the same time ( see fig 1b ) ; f1 suddenly elongated and started to decay after 12 march . we also observed the rapid growth of p2 and f5 umbrae between 12 and 14 march . this time coincidence between the decay of f1 and the growth of p2 and f5 strongly suggests a close causal relation between them : the umbrae f1 was stable until 12 march because it was bound by the coiling flux tubes . when the coiling flux tubes already emerged above the photosphere after 12 march , however , the trunk of flux tube f1 suddenly expanded , or elongated and started to decay . furthermore , it must be worthwhile here to mention about a differential emergence of magnetic loops . suppose the gas pressure and magnetic field strength in the tube are @xmath33 and @xmath34 and the surrounding external pressure at the same height is @xmath35 . then the lateral total pressure balance implies @xmath36 if the temperature ( @xmath37 ) is uniform and the corresponding densities are @xmath38 , equation ( 3 ) becomes @xmath39 where we take the perfect gas law ; @xmath40 ( @xmath41 is the boltzmann constant and @xmath42 is the mean particle mass ) . the plasma in the tube feels a resultant buoyancy force of @xmath43 per unit volume , which tends to make the tube rise . the trunk and coiling loops in fig 4b are considered to have different magnetic pressures ( @xmath34 ) and different densities ( @xmath44 ) . different densities give different buoyancy force , so the system rises differentially . difference of the shapes of flux tubes ( i.e. different curvatures of flux tubes ) may also make the system rise differentially . in spite of this consideration , the schematic drawing like fig . 4b may give the impression that the whole system emerges at the same speed , because we do not draw how differentially the loops rise . we need more observational data of better temporal- and spatial- resolutions before we can discuss more realistic structures of successively emerging flux loops . and this is an important subject to be attacked in future works . figure 4b , however , well shows the essential characteristics of our model , i.e. an emergence of a twisted magnetic flux bundle . it does not influence our conclusion how differentially the flux loops rise . according to solar geophysical data ( sgd ) , about two hundred flares occurred in this active region . some example of flares observed with dst at hida observatory are shown in fig . the positions of the flares given in sgd were transformed to our coordinate system ( heliographic latitudes and carrington longitude ) and these of the flares stronger than c - class in soft x - ray importance were plotted in fig . 6 , where the positions of the flares are given for each day ( from 15:00 ut of the previous day to 15:00 ut of the day ) . the relative velocities of the small sunspot umbra around f1 are also shown in fig . notice that almost all flares occurred around f1 from 9 through 12 march . on 15 march , before which the twisted flux bundle has already emerged out , however , no strong flare occurred around f1 ( fig . 6 ) . on the contrary , some strong flares occurred along the sheared neutral line formed between p2 and f5 on 14 and 15 march . especially noticeable in these two figures ( fig . 5 and 6 ) is that from 9 through 12 , march , many m - class and c - class flares occurred at the leading edge and at the east of f1 where the vortex - like motions of the small umbrae were found . the north side of f1 is also flare - productive especially on 12 march . these results strongly suggest that the vortex motions of small spots around f1 or the successive emergence of twisted magnetic flux tubes should be sources of the strong flare activities in this region . the vortex - like motions of small satellite spots around f1 were explained with emergences of the twisted magnetic flux bundle(sec . 4.1 ) , and strong flare activities preferentially occurred around this emerging flux region ( sec . these evidence strongly suggest the existence of causal relations between the emergences of twisted magnetic flux tubes and the strong flare activities . with successive emergence of twisted flux tubes , very complex magnetic field structures , which consist of many current loops , were formed around f1 and at the south of p1 . notice the twisted dark loops at the east side of f1 in the h@xmath3 images of fig . they have different orientations with each other , and they are considered to reconnect with each other and sometimes to produce flares as shown in fig . 5 . in the convection zone , magnetic flux tubes are intensified by a dynamo action and twisted by convection in a rotating fluid . when the twisted flux tubes emerge to the photosphere , the twists should get loose as shown in fig . 4 with decrement of the gas pressure . since the twisted magnetic tubes store much energy , the energy could be released in the violent flare activities in the f1 region of noaa 5395 . the idea that twisted magnetic flux tubes store the energy for flares is first proposed by piddington ( 1974 ) , and several observational studies developing this idea have been published by tanaka ( 1991 ) , kurokawa ( 1987 , 1991 , 1996 ) , wang ( 1994 ) and wang et al . our study presents another clear evidence of the emergence of a strongly twisted and flare - productive magnetic flux bundle . especially important findings in this study are the outflowing vortex - like motions of small sunspots and extremely hot flare activity around the leading sunspot f1 . such conspicuous motions of sunspots and associated hot flare activities have been hardly reported before , though kurokawa et al . ( 1987 ) noticed a conspicuous growth of the leading part of @xmath2-type sunspot just before a great flare of april 25 , 1984 in noaa 4474 . leka et al . ( 1994 , 1996 ) also reported some vortex - like motions of small sunspots in an emerging flux region , but the region is not so active in flare production . more examinations of the relation between the vortex motions of sunspot structures and flare activity are necessary in future observations and measurements of existing data . sunspot proper motions and flares of a super active region noaa 5395 , which is one of the biggest and the most flare - active regions in the 22nd sunspot cycle , were analyzed in details . we measured the sunspot proper motions by using the h@xmath3 images obtained with dst and found some peculiar vortex - like motions of small satellite spots successively emerging from the leading sunspot f1 . for the explanation of these motions of small sunspots , we proposed a schematic model of the successive emergence of twisted and winding magnetic flux loops coiling around a trunk of magnetic flux tube ( fig 4 ) . we also found most of flares preferentially occurred in this emerging region around f1 . these results are consistent with our previous conclusion that the flare productive magnetic shear is produced by the emergence of twisted magnetic flux tubes ( kurokawa 1987 , 1991 , 1996 ) . this study strongly suggests that the magnetic energy of a flare - productive sunspot group is stored in a twisted flux bundle which is originally formed in the convection zone . the energy is released as flares in the course of the emergence of the twisted flux bundle above the photosphere . studies of the typical flux tube geometry which has enough energy for big flares should be useful for forecasting flare occurrence . it is still unclear , however , which type of emergence of twisted flux tube always produces strong flare activity . we need more observational studies of the process of magnetic shear developments in more active regions by using proper motions of sunspots and evolutional changes of h@xmath3 fine structures and photospheric vector magnetic field . we would like to thank the anonymous referee for the useful comments which improved our paper . we also thank drs . t. sakurai , f. tang and h. zhang for their providing magnetograms . we are grateful to drs . r. kitai and y. funakoshi for their helps in the observations at hida observatory . h. hirashita contributed helpful discussions and advice . one of the authors ( ttt ) acknowledges the research fellow of the japan society for promotion of science for young scientists . 1. the h@xmath10 @xmath1 image obtained with dst at 23:19 on march 11 . notice the vortex - like lines of small umbrae and penumbra structures surrounding a large sunspot umbra f1 . ( b ) daily evolutional changes of noaa 5395 . north is up , west is to right . an interval of two hair lines corresponds to 100 @xmath11 . sunspots measured their proper motions are given numbers . the signs f and f stand for the _ following _ polarity and the p and p , the _ preceding _ polarity . the sign : means uncertainness . this region was elongated toward southwest and northeast . 2. velocities of small umbrae relative to f1 measured for two images of good seeing . an arrow in the upper right corner of each panel indicates the velocity scale of 100 m / s . vortex - like motions of small satellite spots are found . ( a ) from 23:31 ut on march 8 to 05:10 ut on march 9 . ( b ) from 00:37 ut to 05:46 ut on march 10 . ( c ) from 23:24 ut on march 10 to 01:28 ut on march 11 . ( d ) from 23:18 ut on march 11 to 02:48 ut on march 12 . fig . 4. ( a ) schematic summary of observed vortex - like motions of small umbrae surrounding f1 . pairs of small umbrae successively emerged at the leading edge of f1 . at the east side of f1 , small umbrae , mainly @xmath25-polarities , moved clockwise and formed p2 . at the west and north sides of f1 , small umbrae , all @xmath28-polarities , moved anticlockwise and formed f5 by merging with some parts of decaying f1 . ( b ) a schematic model of emerging flux tubes . the bundle of flux tubes is twisted and each flux tube is also twisted . planes @xmath31 and @xmath32 show the positions of the photosphere relative to the emerging flux tubes for the different days . the cross section of flux tubes on each photospheric plane correspond to sunspots . 6. flare positions and relative velocities of small umbrae to f1 . almost all flares occurred around f1 from 9 through 12 march . on march 9 and 11 , many m - class and c - class flares occurred at the east side of f1 . on 15 march , before which the twisted magnetic flux bundle has already emerged out , no strong flare occurred around f1 . 7. h@xmath4 @xmath1 images obtained at ( a ) 02:31 ut 11 march ( b ) 00:14 ut 12 march . notice the twisted dark loops at the east side of f1 . they have different orientations with each other , and they sometimes reconnect with each other to produce flares as shown in fig . 5 .	sunspot proper motions and flares of a super active region noaa 5395 , which was the biggest and the most flare - active region in the 22nd sunspot cycle , were analyzed in details . we measured sunspot proper motions by using the h@xmath0 @xmath1 images obtained with the 60-cm domeless solar telescope ( dst ) at hida observatory , kyoto university and found some peculiar vortex - like motions of small satellite spots successively emerged from the leading edge of this sunspot group . to explain these motions of small sunspots , we proposed a schematic model of the successive emergence of twisted and winding magnetic flux loops coiling around a trunk of magnetic flux tube . the location of the strongest flare activity was found to coincide with very the site of the vortex - like motions of sunspots . we conclude that the flare - productive magnetic shear is produced by the emergence of the twisted magnetic flux bundle . magnetic energy is stored in the twisted flux bundle which is originally formed in the convection zone and released as flares in the course of the emergence of the twisted flux bundle above the photosphere . e - mail : [email protected] revised november 28th 1997
the field of computational astrophysics is entering an exciting and challenging era . the large amount of observational data involving general relativistic phenomena requires the integration of numerical relativity with the traditional tools of astrophysics , such as hydrodynamics , magneto - hydrodynamics , nuclear astrophysics , and radiation transport . general relativistic astrophysics astrophysics involving gravitational fields so strong and dynamical that the full einstein field equations are required for its accurate description , is quickly becoming a promising area of research . @xmath0 correspondence should be addressed to mark miller . as a first step in our study of `` computational general relativistic astrophysics '' , our collaboration ( the ncsa / potsdam / wash u numerical relativity collaboration ) is building a code called `` cactus '' for solving the full set of einstein field equations coupled to a perfect fluid source . such a code will have many applications for astrophysical processes involving neutron stars and black holes . in this paper we present the formulation and methods of the 3d general relativistic hydrodynamic part of the code , and its coupling to the spacetime part of the code . we also present various tests for the validation of the code . the complimentary presentation on the ( vacuum ) spacetime evolution part of the code has been given in @xcite . in the following we begin by discussing the background of our code development effort . two of the major directions of astronomy in the next century are high energy astrophysics ( @xmath1-ray and @xmath2-ray astronomy ) and gravitational wave astronomy . the former is driven by advanced x - ray and @xmath2-ray satellite observations , e.g. , cgro , axaf , glast @xcite , xmm , integral , that are either current or planned in the next few years . high energy radiation is often emitted by highly relativistic events in regions of strong gravitational fields , e.g. , near black holes ( bhs ) and neutron stars ( nss ) . one of the biggest mysteries of modern astronomy , @xmath2-ray bursts , is likely related to processes involving interactions of compact binaries ( bh / ns or ns / ns ) or highly explosive collapse to a black hole ( `` hypernova '' ) ( see , e.g. , @xcite and references therein ) . such high energy astrophysical events often involve highly dynamical gravitational fields , strong gravitational wave emissions , and ejecta moving at ultrarelativistic speeds with relativistic lorentz factors up to @xmath3 . the modeling of such events can only be achieved by means of hydrodynamical simulations in the full theory of general relativity . the second major direction , gravitational wave astronomy , involves the dynamical nature of spacetime in einstein s theory of gravity . the tremendous recent interest in this frontier is driven by the gravitational wave observatories presently being built or planned in the us , europe , and outer space , e.g. , ligo , virgo , geo600 , lisa , lagos @xcite , and the lunar outpost astrophysics program @xcite . the american ligo and its european counterparts virgo and geo600 are scheduled to be on line in a few years @xcite , making gravitational wave astronomy a reality . the space detector lisa has been selected as one of the three `` cornerstone missions '' of the european space agency @xcite . these observatories provide a completely new window on the universe : existing observations are mainly provided by the electromagnetic spectrum , emitted by individual electrons , atoms , or molecules , and are easily absorbed , scattered , and dispersed . gravitational waves are produced by the coherent bulk motion of matter and travel nearly unscathed through space , coming to us carrying the information of the strong field regions where they were originally generated @xcite . this new window will provide very different information about our universe that is either difficult or impossible to obtain by traditional means . the numerical ( theoretical ) determination of gravitational waveforms is crucial for gravitational wave astronomy . physical information in the data is to be extracted through template matching techniques @xcite , which _ presupposes _ that reliable waveforms are known . accurate waveform detections are important both as probes of the fundamental nature of gravity and for the unique physical and astronomical information they carry , ranging from nuclear physics ( the equation of state of nss @xcite ) to cosmology ( direct determination of the hubble constant without going through the `` cosmic distance ladder '' @xcite ) . in most situations , the waveform can not be calculated without a numerical simulation based on the full theory of general relativity . this need for waveform templates is an important motivation of our effort . in short , both of these frontiers of astronomy call for computational general relativistic astrophysics , i.e. , the integration of numerical relativity with traditional tools of computational astrophysics , e.g. , computational hydrodynamics , radiation transport , nuclear astrophysics , and magneto - hydrodynamics . if we are to fully understand the observational data generated by the non - linear and dynamical gravitational fields , detailed modeling taking dynamic general relativity into full account must be carried out . we begin by briefly reviewing some of the significant existing investigations in the field of numerical general relativistic hydrodynamics ( gr - hydro in the following ) to set the stage for the description of our own work . while there has been much effort in the study of relativistic hydrodynamics in _ pre - determined _ ( fixed , or with its time evolution specified ) background spacetimes , we focus on studies that are most relevant to _ dynamical _ spacetimes with the matter flows acting as sources to the einstein equations . the pioneering work dates back to the one - dimensional supernova core - collapse code by may and white @xcite . it was based on a lagrangian ( i.e. , coordinates co - moving with the fluid ) finite difference scheme with artificial viscosity terms included in the equations to damp the spurious numerical oscillations caused by the presence of shock waves in the flow solution . numerous astrophysical simulations were based on this approach . one drawback is that the lagrangian character of the code makes it difficult to be extended to the multidimensional case . the pioneering eulerian ( i.e. , coordinates not co - moving with the fluid ) finite difference gr - hydro code was developed by wilson @xcite in the early 70 s . it used a combination of artificial viscosity ( av ) and upwind techniques . it became the kernel of a large number of codes developed in the 80 s . many different astrophysical scenarios were investigated with these codes , ranging from axisymmetric stellar core - collapse @xcite , to accretion onto compact objects @xcite , and to numerical cosmology @xcite . in the following , we give a short overview of this large body of work , paying more attention to the numerical methods used than to the physical results obtained . while there are a large number of numerical investigations in pre - determined background spacetimes based on the av approach ( e.g. , @xcite ) , we focus on those using a fully self - consistent treatment evolving the spacetime dynamically with the einstein equations coupled to a hydrodynamic source . although there is much recent interest in this direction , only the spherically symmetric case ( 1d ) can be considered essentially solved @xcite . in axisymmetry , i.e. 2d , only a few attempts have been made , with most of them devoted to the study of the gravitational collapse and bounce of rotating stellar cores and the subsequent emission of gravitational radiation @xcite . @xcite was the first to calculate a general relativistic stellar core collapse . the computation succeeded in tracking the evolution of matter and the formation of a black hole but the numerical scheme was not accurate enough to compute the emitted gravitational radiation . the code in @xcite used a radial gauge and a mixture of polar and maximal slicing . the gr - hydro equations were solved with standard finite difference methods with av terms . in @xcite the numerical scheme for the matter fields was more sophisticated , using monotonic upwind reconstruction procedures and flux limiters , with discontinuous solutions handled by adding av terms in the equations . in @xcite , a numerical study of the stability of star clusters in axisymmetry was performed . in this investigation , the source of the gravitational field was assumed to be a configuration of collisionless ( dust ) particles , which reduces the hydrodynamic computation to a straightforward integration of the geodesic equations . three - dimensional extensions of these av based gr - hydro treatments have been attempted over the last few years . wilson s original scheme has been applied to the study of ns binary coalescence in @xcite under the assumption of a conformally flat spacetime , which leads to a considerable simplification of the gravitational field equations . a code employing the _ full _ set of einstein equations and self - gravitating matter fields is currently being developed @xcite . in this work the complete set of the equations , spacetime and hydrodynamics , are finite differenced in a uniform cartesian grid using van leer s scheme @xcite with total variation diminishing ( tvd ) flux limiters ( see , e.g. , @xcite for definitions ) . shock waves are spread out using a tensor av algorithm . with this code they have studied the gravitational collapse of a rotating polytrope to a black hole ( comparing to the original axisymmetric computation of ref . @xcite ) and the coalescence of a binary ns system . further work to achieve longer term stability is under way @xcite . the success of the artificial viscosity approach is well - known . however , it has inherent difficulties in handling the ultrarelativistic regime @xcite . in wilson s formulation of the gr - hydro equations , there are explicit spacetime derivatives of the pressure in the source terms . this breaks the conservative character of the system and introduces complications into the numerical treatment . this motivated , in recent years , the effort of extending to relativistic hydrodynamics high - resolution shock - capturing schemes ( hrsc ) originally developed in classical ( newtonian ) computational fluid dynamics . such schemes are based on the solution of local riemann problems , exploiting the hyperbolicity of the hydrodynamic equations . to use such numerical treatments , the hydrodynamic equations are first cast into a first order ( hyperbolic ) system of conservation ( or balance ) laws . the characteristic fields of the system are then determined which allows the construction of numerical schemes which propagate the information along the fluid characteristics . we refer the reader to @xcite for a review of these methods for general hyperbolic systems of conservation laws . hrsc schemes were first introduced into gr - hydro in @xcite , and applied in ( spherical ) dynamical spacetimes in @xcite and @xcite . the latter investigation focussed on , among other problems , the study of supernova core collapse ( including the infall epoch , bounce , and shock propagation ) . the numerical code was based on the radial - gauge and polar - slicing coordinate conditions @xcite . in @xcite the gr - hydro equations were analyzed in the `` 3 + 1 '' formalism and the theoretical building blocks to construct a hrsc scheme in multidimensions were presented . axisymmetric studies using hrsc schemes are currently being carried out in @xcite . this investigation focussed on the study of accretion phenomena onto ( dynamic ) rotating black holes and the associated emission of gravitational radiation induced by the presence of the matter fields . axisymmetric studies will also provide useful test beds " in forthcoming investigations with the present 3d code discussed in this paper . as will be discussed in later sections of this paper , our present code is based on the same hrsc algorithmic machinery as in the aforementioned works . we extend the treatment to 3d , and develop a code that makes no assumptions on the nature of the spacetime , the form of the metric , or the slicing and spatial coordinates . we refer the interested reader to the above references for a first understanding of the numerical schemes used in our work . we also want to mention a completely different approach for gr - hydro based on pseudospectral methods @xcite . these methods are well known for having extraordinary accuracy in smooth regions of the solution . the numerical error is evanescent , i.e. , it decreases as @xmath4 with @xmath5 being the number of coefficients in the spectral expansion . the main drawback of pseudospectral methods has been , traditionally , the inaccurate modeling of discontinuous solutions due to the appearance of the so - called gibbs phenomenon . in the presence of discontinuities , the numerical approximation of the solution does not converge at the discontinuity and spurious oscillations appear . recently , however , an innovative pseudospectral method based on a multidomain decomposition has been developed @xcite which circumvents the gibbs phenomenon . this new approach has already been shown to work remarkably well in the 3d numerical construction of mclaurin and roche equilibrium models . in this subsection we discuss the main issues we considered in choosing our approach , building on the existing work discussed above . the main aim of our program is to study violent and highly - energetic astrophysical processes like ns / ns coalescence within the framework of general relativity . these scenarios involve strong gravitational fields , matter motion with ( ultra ) relativistic speeds and/or strong shock waves . these features make the numerical integration of the hydrodynamic equations a very demanding task . the difficulty is exacerbated by the intrinsic multidimensional character of these astrophysical systems , and by the inherent complexities in einstein theory of gravity , e.g. , coordinate degrees of freedom and the possible formation of curvature singularities ( e.g. , collapse of matter configurations to black holes ) . these complications call for the use of advanced numerical methodology , a flexible code construction which allows for the use of different treatments , and a large amount of careful testbed studies . in the following we discuss these issues in more detail . two major issues in gr - hydro which are purely hydrodynamical in origin are the numerical modeling of flows with large lorentz factors and strong shock waves . in @xcite it was shown that the av based schemes have difficulties in handling ultrarelativistic velocity flows with lorentz factors @xmath6 . as a result , @xcite proposed using implicit finite difference schemes to handle the gr - hydro equations in the ultrarelativistic regime . however , investigations during the last decade have provided increasing evidence that the most appropriate schemes to deal with ultra - relativistic flow with strong shocks are those based on ( approximate or exact ) riemann solvers , i.e. , hrsc schemes . these methods have high accuracy ( second order or more ) in regions where the flow solution is smooth , and at the same time are able to resolve discontinuities in the solution ( e.g. , shock waves ) with little smearing . they have been extensively tested and found to be applicable in the ultra - relativistic regime ( see , e.g. , @xcite for a recent review ) . while we believe hrsc schemes may be capable of providing the technology for treating the hydrodynamic part of the evolution , the field of computational gr - hydro still contains many issues that are as yet unexplored , especially for cases where the relativistic fluid is coupled to a dynamical spacetime . for a fully dynamical spacetime , one major issue is the handling of the gauge degrees of freedom . this problem is exacerbated in 3d simulations without symmetry assumptions . in a general 3d problem , there is no preferred choice of gauge to restrict the metric functions as in lower dimension simulations ( e.g. , radial gauge and polar slicing in spherically symmetric simulations ) . lagrangian coordinate systems are inappropriate for complicated 3d flows . the inevitable lower resolution in 3d simulations also makes the problem more acute . even in vacuum spacetime studies , the choice and implementation of appropriate gauge conditions for a general dynamical evolution is a largely unexplored territory @xcite . how will the gauge choices be affected by the presence of relativistic fluid flows or by the existence of strong shocks which create sharp features in the sources of the metric evolution ? for example , what will be a useful gauge condition for a process like the inspiral of a ns / ns binary ? these are completely open issues . in order to provide the capability to investigate these problems , the code we construct here is designed to allow arbitrary gauge conditions , making no assumptions on the lapse function or the shift vector . another class of problems involves the connection of the numerical integration of the hydrodynamic equations to that of the spacetime equations . what is the best set of variables to use , locally measured quantities , coordinate variables , densitized quantities or some combination ? with the spacetime metric an evolved variable , there are many choices . what is the best way to connect the hydrodynamics and the spacetime finite differencing steps to achieve not only a second order accurate scheme in both space and time , but also in a way that is suitable for long term evolutions ? even in newtonian strong field evolutions , coupling the hydrodynamic integration to the gravitational potential calculation in different ways can yield different long term behavior @xcite . as a consequence of the different character of the equations governing the geometry of the spacetime and the evolution of the matter fields , the numerical methods to handle them are drastically different . what are the effects of combining different methods , and is there a best combination for a particular class of problems ? with the recent development of hyperbolic formulations in gr , an interesting possibility would be to consider all of the dynamical variables , both spacetime and matter fields , to be members of one master state vector . the entire system of equations could then be written as a single ( vector ) conservation ( or balance ) equation . one could then apply the same hrsc schemes to the entire system . what advantages would this bring ? these are some of the issues that we have in mind in choosing our approach in developing the code as will be discussed next . our overall goal is to develop an efficient , flexible , computational tool for general relativistic astrophysics . specifically for this paper , the aims are : ( 1 ) to establish the formulation , including the spectral decomposition of the gr - hydro equations , on which our code is based , ( 2 ) to validate the numerical code we constructed for solving the gr - hydro equations , and ( 3 ) to compare the different numerical schemes we used . the set of differential equations we are attempting to solve consists of very complicated , coupled partial differential equations involving thousands of terms . considering the complexity and generality of the code , along with the fact that the solution space of the differential equations is largely unexplored , it is essential that any physical result produced by a 3d gr - hydro code be preceded by a series of tests such as the ones we report here , in order to insure the fidelity of the discretization to the original differential equations . in fact , we consider the tests presented here to be a minimal set : _ any _ 3d gr - hydro code should be able to reproduce these results . further tests , especially those related to the long term stability of the code and detailed comparisons of 3d and 1d results will be presented in a forthcoming paper . in exploring the very complex system of the gr - hydro equations , it is also essential to have the capability to compare results based on different mathematical formulations and different numerical schemes . our code is currently set up to allow two different formulations of the einstein equations : the standard arnowitt - deser - misner ( adm ) formulation @xcite and the bona - mass ( bm ) hyperbolic formulation @xcite ( other hyperbolic formulations will be included and reported later ) . the code allows for two different choices for finite differencing the adm equations : a standard leapfrog scheme and an iterative crank - nicholson scheme . the bm equations are finite differenced using a strang split to separate the source and flux updates . the latter are performed using a maccormack method . as for the numerical treatment of the hydrodynamic equations , the code has the capability of using three different hrsc schemes : the first one is the flux split method , mainly chosen for its simplicity . the second is roe s method @xcite ( note that contrary to @xcite , we do not use roe s averaging but instead employ arithmetic averaging ( see section [ sec : discrete ] below for details ) ) . the third scheme we use is the recently developed marquina s method @xcite . all three schemes are coupled to the spacetime evolution solver in a way which is second order accurate in both space and time . in this code we also allow for arbitrary spacetime coordinate conditions . as mentioned previously , this enables the investigation of gauge choices in gr - hydro and allows the use of different coordinate systems for different astrophysical simulations . this capability is built into our development , and we have carried out tests with non - trivial lapses and shifts in this paper . however , more investigation is needed in this direction . as the aim of our program is to study _ realistic _ astrophysical systems , which often require full 3d simulations and involve many different time and length scales , it is important that the computer code we develop be capable of carrying out large scale simulations . this requires the use of massively parallel supercomputers . the `` cactus '' code was built with this in mind . here we give a brief overview of the computational infrastructure of the code and its performance . for a more extensive review , see @xcite . the cactus code achieves parallelism through the mpi message passing interface @xcite . this allows high performance portable parallelism using a distributed memory model . all major high performance parallel architectures , including the sgi origin 2000 , cray t3e , hp / convex exemplar , and ibm sp-2 support this programming model . the mpi layer of cactus also allows computing on clusters of networked workstations and pc s . parallelism in cactus is based on a generic domain decomposition , distributing uniform grid functions across multiple processors and providing ghost - zone based communications for a variety of stencil widths and grid staggerings . the code can also compile without mpi , allowing the same source code to be run on a single processor workstation and on massively parallel supercomputers . the platforms currently supported and tested include : the sgi origin 2000 ( up to 256 nodes ) , the cray t3e ( up to 1024 nodes ) , sgi o2 clusters , nt clusters , dec alphas , and sgi workstations . we have recently benchmarked a version of the code ( the `` gr3d '' version , constructed for the nasa neutron star grand challenge project , see http://wugrav.wustl.edu/relativ/nsgc.html ) on a 1024 node t3e-1200 , achieving over 140 gflop / sec and a scaling efficiency of over 95% ( for details of the benchmark , see http://wugrav.wustl.edu/codes/gr3d/ ) . besides the floating point and scaling efficiency , it is also noteworthy that a relatively large grid size ( 644 x 644 x 1284 grid points for 32 bit accuracy , and 500 x 500 x 996 grid points for 64 bit accuracy ) were used for the benchmarked run on the t3e-1200 . this is made possible by the efficient memory usage of the code . with the full set of the einstein equations coupled to the relativistic hydrodynamics equations , a large number of 3d arrays are required to evolve the system . in order to have reasonable resolutions for realistic simulations , it is essential that the code make efficient use of available memory . it is also essential that the code be highly optimized in order for these large simulations to be carried out in a reasonable time . during the code development , special attention was also given to software engineering problems , such as collaborative code development , maintenance , and management . the code was developed to be shared by the entire community for the investigation of general relativistic astrophysics . to minimize barriers associated with collaborative development , the code was constructed to have : 1 . a modular code structure that allows new code modules to easily plug in as `` thorns '' to the core part of the code ( the `` flesh '' ) . the `` flesh '' contains the parallel domain decomposition software , i / o , and various utilities . 2 . a consistency test suite library to make sure that new thorns will not conflict with other parts of the code . 3 . various code development tools , such as : documentation , elliptic solvers , and visualization tools , which provide a complete environment for code development , and testing . for detailed discussions of these and other features of collaborative infrastructure of the code , see @xcite these computational features of the code significantly enhance our effort in constructing a multi - purpose code for general relativistic astrophysics . the organization of the paper is as follows : the formulation of the differential equations are given in section [ sec : formulation ] . a spectral decomposition of the gr - hydro equations suitable for a general non - diagonal spatial metric is presented . the details of the discretization of the equations and of the coupling of the spacetime and hydrodynamics are given in section [ sec : discrete ] . shock tube tests are performed in section [ sec : shocktube ] for shocks along the coordinate axes and along the diagonal . these test the hydrodynamic part of the code , with the background geometry held flat . we then go on to test the coupling of the hydrodynamics to curved and dynamical spacetimes . section [ sec : frw ] is on tests using friedmann - robertson - walker cosmologies with dust . section [ sec : tov ] contains tests using static spherical star solutions with a polytropic equation of state . we present a practical procedure which gives stable evolution of the surface region of the star . section [ sec : boost ] contains tests using the spherical star solutions described in section [ sec : tov ] but now relativistically boosted along the diagonal @xmath7 . this is a strong test of the fully coupled spacetime and hydrodynamics system , with all possible terms in the equations activated and with a non - trivial lapse and shift . finally , section viii contains a brief summary . all tests presented in sections [ sec : frw]-[sec : boost ] contain convergence studies performed in the following way : errors are obtained by subtracting the exact solution at a specific time from the computed solution for a number of dynamical variables . these errors are produced at three different resolutions , @xmath8 , @xmath9 , and @xmath10 . to demonstrate they have the correct convergence properties for a second order accurate discretization we check that each error function decreases by a factor of four for each factor of two increase in resolution . this is demonstrated by plotting the various error functions along 1-d lines . these convergence tests are an essential part in validating the code . in this subsection we present the hydrodynamic equations for a general curved spacetime in a form suitable for advanced numerical treatment . the equations for the evolution of the spacetime , including the hydrodynamic source , will be presented in a later subsection . the general relativistic hydrodynamic equations , written in the standard covariant form , consist of the local conservation laws of the stress - energy , @xmath11 , and the matter current density , @xmath12 @xmath13 @xmath14 where @xmath15 , @xmath16 is the rest mass density and @xmath17 the 4-velocity of the fluid . @xmath18 stands for the covariant derivative with respect to the 4-metric of the underlying spacetime . throughout this paper we are using , unless otherwise stated , natural units ( @xmath19 ) . greek ( latin ) indices run from 0 to 3 ( 1 to 3 ) . in what follows we will neglect viscous effects , assuming the stress - energy tensor to be that of a perfect fluid @xmath20 where @xmath21 is the fluid pressure and @xmath22 is the 4-metric describing the spacetime . in addition , the relativistic specific enthalpy , @xmath23 , is given by @xmath24 where @xmath25 is the rest frame specific internal energy density of the fluid . the equations written in this covariant form are _ not _ suitable for the use of advanced numerical schemes . in order to carry out numerical hydrodynamic evolutions , and in particular to take advantage of the benefits of hrsc methods , the hydrodynamic equations after the 3 + 1 split must be written as a hyperbolic system of first order flux conservative equations . we introduce coordinates @xmath26 and write eqs . ( [ eq : stressenergycons ] ) and ( [ eq : masscons ] ) in terms of coordinate derivatives . we project eq.([eq : stressenergycons ] ) and eq.([eq : masscons ] ) onto the basis @xmath27 , with @xmath28 being a timelike vector normal to a given hypersurface . a straightforward calculation yields the set of equations in the desired form @xmath29 where @xmath30 denotes a partial derivative with respect to time and @xmath31 indicates a partial derivative with respect to the spatial coordinate @xmath32 . the evolved state vector @xmath33 is given , in terms of the primitive variables @xmath34 , as @xmath35 = \left [ \begin{array}{c } \sqrt{\gamma } w \rho \\ \sqrt{\gamma } \rho h w^2 v_j \\ \sqrt{\gamma } ( \rho h w^2 - p - w \rho ) \\ \end{array } \right ] , \label{eq : evolvedvar}\ ] ] where @xmath2 is the determinant of the 3-metric @xmath36 , @xmath37 is the fluid 3-velocity , and @xmath38 is the lorentz factor , @xmath39 . notice that the spatial components of the 4-velocity @xmath40 are related to the 3-velocity by the following formula : @xmath41 where @xmath42 and @xmath43 are , respectively , the lapse function and the shift vector of the spacetime . also notice that we are using a slightly different set of variables as those used in @xcite . we are now densitizing " the evolved quantities , @xmath44 , @xmath45 and @xmath46 , with the factor @xmath47 . the three flux vectors @xmath48 are given by @xmath49.\ ] ] finally , the source vector @xmath50 is given by @xmath51,\ ] ] where @xmath52 is the 4-christoffel symbol @xmath53 a technical point must be included here . while the numerical code updates the state vector @xmath33 forward in time it makes use , internally , of the set of primitive variables defined above , @xmath34 . those are used throughout , e.g. , in the computation of the characteristic fields ( see below ) . these variables can not be obtained from the evolved ones in a closed functional form . instead , they must be recovered through some appropriate root - finding procedure ( an example of this can be found in @xcite ) . the use of hrsc schemes , as will be presented in detail in the next section , depends crucially on the knowledge of the spectral decomposition of the jacobian matrix of the system @xmath54 the characteristic speeds ( eigenvalues ) and fields ( eigenvectors ) are the key ingredients of any hrsc scheme . the spectral decomposition of the jacobian matrices of the general relativistic hydrodynamic equations with general equation of state was first reported in @xcite ( for polytropic eos see @xcite ) . however , we have found that the eigenvectors reported in @xcite are correct only in the case of a diagonal spatial metric . in this section we display the full spectral decomposition valid for a generic spatial metric . we focus on the @xmath55-direction , hence presenting the spectral decomposition of @xmath56 , as the other two directions can be found by simple permutation of indices . we start by considering an equation of state in which the pressure @xmath21 is a function of @xmath16 and @xmath25 , @xmath57 . the relativistic speed of sound in the fluid @xmath58 is given by ( see , e.g. , @xcite ) @xmath59 where @xmath60 , @xmath61 , @xmath62 is the entropy per particle , and @xmath63 is the total rest energy density which in our case is @xmath64 . we require a complete set of eigenvectors @xmath65 $ ] and corresponding eigenvalues @xmath66 along the @xmath55-direction , i.e. @xmath67 [ \vec{r}_i ] = { \lambda}_i [ \vec{r}_i ] , \;\;\ ; i=1,\cdots,5.\ ] ] the solution contains a triply degenerate eigenvalue @xmath68 a set of linearly independent vectors that span this degenerate space is given by @xmath69}^t,\ ] ] @xmath70}^t,\ ] ] @xmath71}^t.\ ] ] the superscript @xmath72 denotes transpose . the other two eigenvalues are given by @xmath73 } \right \ } - { \beta}^x,\ ] ] with corresponding eigenvectors @xmath74 } ^t.\ ] ] the dynamics of the gravitational field in general relativity theory is described by einstein s field equations @xmath75 which relate the ( ten ) metric components @xmath76 of the spacetime to the stress energy tensor @xmath77 . here , @xmath78 is the einstein tensor which involves second derivatives , in both space and time , of the dependent variables @xmath79 . a formulation of the einstein equations suitable for numerical evolutions has been known for more than three decades @xcite . in recent years , many new formulations have been proposed ( for a review , see @xcite and references therein ) seeking to expose the hyperbolicity of the evolution components of the einstein field equations . in the present paper , we discuss the mathematical and algorithmic issues related to the coupling of the hydrodynamic equations to two different formulations of the einstein equations . we start with the more commonly used adm formulation @xcite . then we discuss the bm hyperbolic formulation of the einstein equations @xcite . in the adm formulation @xcite , spacetime is foliated into a set of non - intersecting spacelike hypersurfaces . there are two kinematic variables which describe the evolution between these surfaces : the lapse function @xmath42 , which describes the rate of advance of time along a timelike unit vector @xmath80 normal to a surface , and the spacelike shift vector @xmath43 that describes the motion of coordinates . the line element is written as @xmath81 the adm formulation casts the einstein equations into a first order ( in time ) quasi - linear @xcite system of equations . the dependent variables are the 3-metric @xmath36 and the extrinsic curvature @xmath82 . the evolution equations read : @xmath83 \nonumber \\ & \ & + \beta^m \nabla_m k_{ij}+k_{im } \nabla_j \beta^m+k_{mj } \nabla_i \beta^m , \label{dtkij}\end{aligned}\ ] ] where @xmath84 denotes a covariant derivative with respect to the 3-metric @xmath36 and @xmath85 is the ricci curvature of the 3-metric . in addition to the evolution equations , @xmath86 and @xmath87 must satisfy the hamiltonian constraint @xmath88 and the momentum constraints @xmath89 here , @xmath90 are the components of the stress energy tensor projected onto the 3d hypersurface ( for a more detailed discussion , see @xcite ) . in this paper we use the stress - energy tensor of a perfect fluid ( eq . ( [ perf_fluid ] ) ) . hence , explicitly in terms of the primitive hydrodynamic variables , @xmath91 in the bm hyperbolic formulation @xcite , the evolution equations are written as a first order balance law with , formally , the same mathematical structure as the hydrodynamic equations , eq . ( [ balance ] ) . now , the state vector , containing the evolved quantities for the spacetime , has the following @xmath92 components : @xmath93 where @xmath94 accordingly , the fluxes are given , using a self - explained notation , by : @xmath95 \ ; , \nonumber \\ f^k_-d_{kij } & = & -\beta^r d_{rij } + \alpha\;(k_{ij}-s_{ij } ) \ ; , \label{fluxd } \\ f^k_-a_k & = & -\beta^r a_r + \alpha\;q \ ; , \label{fluxa } \\ f^k_-v_i & = & -\beta^k v_i + b^k_{\;i } - b_i^{\;k } \ ; , \label{fluxv}\end{aligned}\ ] ] where @xmath96 is calculated from the user supplied shift vector . finally , the source terms read : @xmath97 \nonumber \\ & & + n/4\;\alpha \gamma_{ij}\;[\ ; -d_k^{\;rs}\gamma^k_{\;rs } + d_{kr}^{\;\;r}d^{ks}_{\;\;s } -2\,v^ka_k \nonumber \\ & & \;\;\;\ ; + k^{rs}k_{rs}-(tr\,k)^2 \ ; ] \ ; , \nonumber \\ s_-d_{kij } & = & 0 \ ; , \label{sourced } \\ s_-a_{k } & = & 0 \ ; , \label{sourcea } \\ s_-v_i & = & \alpha\;[8 \pi j_i + a_r\;(k^r_{\;i}-tr\,k\;\delta^r_i ) \nonumber \\ & & + k^r_{\;s}(d_{ir}^{\;\;s}-2d_{ri}^{\;\;s } ) - k^r_{\;i}(d_{rs}^{\;\;s}-2d_{sr}^{\;\;s } ) ] \label{sourcev } \\ & & + 2(b_i^{\;r } - \delta_i^r\;tr\,b)\;v_r + 2(d_{ri}^{\;\;s}-\delta^s_i\;d^j_{\;jr})b^r_{\;s } \ ; , \nonumber\end{aligned}\ ] ] where @xmath98 is used to simplify the equations . the free parameter @xmath99 allows one to select a specific evolution system ( it is zero for the `` ricci '' system and one for the `` einstein '' system ) as discussed in @xcite . as stated previously , the main aim of this work is to confirm the _ consistency _ of the coded finite difference equations with the partial differential equations and the _ convergence _ of three independent discretizations of the gr - hydro equations . all three approaches are based on finite - difference schemes employing hrsc schemes to account , explicitly , for the characteristic information of the equations . the methods considered are a flux split method , roe s approximate riemann solver @xcite , and marquina s recently developed scheme @xcite . to simplify the discussion , let us examine the update for the state vector @xmath100 for a flux in the @xmath55-direction : @xmath101 the discretization of this equation takes the form @xmath102 where @xmath103 is the numerical flux " function calculated at the interfaces @xmath104 of the spatial cell @xmath105 . the different methods we are using simply differ in the way the numerical fluxes are calculated . the way the source terms are integrated is explained later in this section . the first scheme is a flux split method , where the flux is decomposed into the part contributing to the eigenfields with positive eigenvalues ( fields moving to the right ) and the part with negative eigenvalues ( fields moving to the left ) . these fluxes are then discretized with one sided derivatives whose side depends on the sign of the particular eigenvalue . for the flux split method , one makes the assumption that @xmath106 for any constant @xmath107 . this is only true for the fluxes of eq . ( [ hydroflux ] ) if the equation of state has the following functional form @xmath108 for some function @xmath109 . for the flux split method we therefore assume the equation of state to be in the following form @xmath110 with @xmath111 being the ( constant ) adiabatic index of the fluid . it is easy to show that , under the above assumptions , the flux vector @xmath112 can be written @xmath113 using the spectral decomposition of section [ sec : spectral ] one can express the jacobian matrix @xmath114 as @xmath115 where m is the matrix whose columns are the right - eigenvectors of the system and @xmath116 is a diagonal matrix constructed from the corresponding eigenvalues ( see section [ sec : spectral ] ) . given the characteristic information , we can now split the flux into the part that is moving to the right and the part that is moving to the left : @xmath117 where @xmath118 , and @xmath119 . the numerical flux which corresponds to an upwind method ( first order in space ) is then simply @xmath120 one could attempt to construct a numerical flux based on one - sided derivatives that were second order accurate in space : @xmath121 however , the method would not have the ( numerically desirable ) total variation diminishing ( tvd ) property ( see , e.g. @xcite for definition ) unless flux limiters are used in front of the second order correction terms : @xmath122 with @xmath123 where we are using van leer flux limiters @xcite @xmath124 the second scheme we use to integrate the hydrodynamic equations makes use of roe s approximate riemann solver @xcite . this is undoubtedly the most established method for the accurate integration of non - linear hyperbolic systems of conservation laws . the suitability of roe s method for the relativistic hydrodynamic equations have been shown in @xcite . this method makes no assumption on the equation of state , and , in this respect , is more flexible than the flux split methods . as the method is well documented in the literature ( see , e.g. , @xcite ) it will only be briefly outlined here . as mentioned in the introduction , all simulations reported in this paper using roe s scheme are performed employing arithmetic averages . for the use of roe s averaging in gr - hydro see @xcite . a monotonic piecewise - constant ( piecewise - linear ) reconstruction of the cell centered values of the primitive variables to the cell interfaces provides first - order ( second - order ) accuracy in space @xcite . in order to get second - order convergence we have implemented a standard minmod piecewise - linear reconstruction algorithm @xcite . the numerical fluxes across interfaces are calculated according to @xmath125 , \label{roeflux}\ ] ] where @xmath126 and @xmath127 indicate the right and left sides of a cell interface . in addition , @xmath128 are , respectively , the eigenvalues and right - eigenvectors of the jacobian matrix of the system calculated at the cell interfaces as the arithmetic mean of the left and right reconstructed ( interpolated ) primitive variables . averaged quantities in eq . ( [ roeflux ] ) are identified by a tilde " . finally , the quantities @xmath129 , the jumps of the characteristic variables across each characteristic field , are obtained from @xmath130 in @xcite donat and marquina proposed a new flux formula to compute the numerical flux at a cell interface . the new flux formula has a clear flux splitting structure , and leads to an upstream scheme . the novelty of marquina s approach lies in the extension of shu and osher s entropy - satisfying numerical flux @xcite to systems of hyperbolic conservation laws . in this scheme there are no artificial intermediate states constructed at each cell interface . this implies that there are no riemann solutions involved ( either exact or approximate ) ; moreover , the scheme has been proven to alleviate several numerical pathologies associated to the introduction of an averaged state ( as roe s method does ) in the local diagonalization procedure ( see @xcite ) . to compute the numerical flux at a given interface , separating the states @xmath131 and @xmath132 , we compute first the sided local characteristic variables and fluxes : @xmath133 for @xmath134 . here @xmath135 , @xmath136 , are the ( normalized ) left eigenvectors of the jacobian matrices of the system . let @xmath137 and @xmath138 be their corresponding eigenvalues . for @xmath139 the procedure is the following : * if @xmath140 does not change sign in @xmath141 $ ] , then the scheme is upwind if @xmath142 then @xmath143 else @xmath144 endif * otherwise , the scheme is switched to the more viscous , entropy - satisfying , local - lax - friedrichs scheme @xmath145 @xmath146 is a curve in phase space connecting @xmath147 and @xmath148 . in addition , @xmath149 can be determined as @xmath150 where , @xmath151 , @xmath152 , are the right ( normalized ) eigenvectors of the system . for further technical information about this solver we refer the reader to @xcite . the suitability of this scheme for the accurate integration of the hydrodynamic equations and many of its desirable properties can be found in @xcite ( newtonian hydrodynamics ) and @xcite ( relativistic hydrodynamics ) . in this section we outline the discretization techniques used in the vacuum spacetime part of the code . for a more detailed discussion we refer the reader to @xcite . here we give the essential formulae for completeness and discuss in detail only the issues relevant to its coupling to hydrodynamics described in the next subsection . the bm system uses the so - called strang splitting @xcite to separate eq . ( [ balance ] ) into two evolution steps . in the first step , only the source terms are used to update the variables @xmath153 while in the second step , only the flux terms are used for the update @xmath154 to ensure second order accuracy in both space and time , this is done by first evolving the source terms forward in time half a time step , then evolving with only the flux terms a full time step , and finally evolving with only the source terms another half time step . the source terms are evolved forward using a second order accurate predictor - corrector method , while the flux terms are evolved using a second order accurate maccormack scheme . specific details of these methods are discussed in @xcite . the adm system supports the use of several different numerical schemes . currently , a leapfrog ( non - staggered in time ) and iterative crank - nicholson scheme have been coupled to the hydrodynamic solver . the leapfrog method assumes that all variables exist on both the current time step @xmath99 and the previous time step @xmath155 . variables are updated from @xmath155 to @xmath156 ( future time ) evaluating all terms in the evolution equations on the current time step @xmath99 . the iterative crank - nicholson solver first evolves the data from the current time step @xmath99 to the future time step @xmath156 using a forward in time , centered in space ( ftcs ) first order method . the solution at steps @xmath99 and @xmath156 are then averaged to obtain the solution on the half time step @xmath157 . this solution at the half time step @xmath157 is then used in a leapfrog step to re - update the solution at the final time step @xmath156 . this process is then iterated . the error is defined as the difference between the current and previous solutions on the half time step @xmath157 . this error is summed over all grid points and all evolved variables . this process is repeated until some desired tolerance is reached . care is taken to make sure that at least two iterations are taken to make the process second order accurate . our code evolves the spacetime geometry and the matter fields separately . this allows different methods to be used for each system ( spacetime and hydrodynamics ) . the coupling of those different evolution algorithms in a way that is second order accurate in both space and time is highly method dependent . we will therefore discuss the coupling of each system , adm or bm , with hydrodynamics , separately . a summary of the different combined schemes appears in table [ table : discrete_names ] . [ tb ] .this table summarizes the abbreviations used for the various methods used for the spacetime and hydrodynamical evolutions . [ cols="^,^,^",options="header " , ] in this paper we present a new three - dimensional , eulerian , general relativistic hydrodynamical code constructed for general relativistic astrophysics . this code is capable of evolving the coupled system of the einstein and hydrodynamic equations . the code is constructed for a completely general spacetime metric based on a cartesian coordinate system , with arbitrarily specifiable lapse and shift conditions . this paper discussed the general relativistic hydrodynamics part of the code , and its coupling to the spacetime code , in parallel to the presentation of the spacetime ( vacuum ) part of the code in @xcite . we have derived a spectral decomposition for the gr - hydro equations valid for general spatial metrics , generalizing the results of @xcite which were only valid for the case of a diagonal metric . based on this spectral decomposition , three different approximate ( linearized ) riemann solvers , flux - split , roe and marquina , were used to integrate the relativistic hydrodynamic equations . we tested these methods individually and compared the results against one another . while we found all methods converging to second order in the discretization parameter , we also compared the absolute values of errors of the different methods . which method produced the smallest absolute error , and whether the spacetime or hydrodynamical evolution was the dominant source of error , depends on the initial data being evolved . for the shocktube problem , only the hydrodynamical evolution was relevant since the evolution took place on a flat background metric . for an evolution along a coordinate axis , the roe and marquina methods were superior to the flux split method . for an evolution where the shockfront is along the diagonal , the flux split method was slightly more accurate than both the roe and marquina method . for the frw evolutions , the spacetime evolution is the main source of error . the bm system tends to be more accurate than the adm system . for the tov tests , we find that the roe and marquina methods are more accurate than the flux split method , and the bm system is more accurate than the adm system . for the boosted tov test , the roe and marquina methods are again superior to flux split . we caution that these statements could depend on the resolution used and the duration of evolution . the hydrodynamic evolution is coupled to the spacetime evolution in a manner which is second order accurate in _ both _ space and time . the coupled code was subjected to a series of convergence tests , with different combinations of the spacetime and hydrodynamics finite differencing schemes , demonstrating the consistency of the discrete equations with the differential equations @xcite . the extensive convergence tests performed are important not only for the validation of the code , but have also been important debugging tools during the code development process . we consider the tests presented to be a minimal set that any 3d gr - hydro code should pass before actual applications . the test - beds that we report on in this paper include : special relativistic shock tubes , friedmann - robertson - walker cosmology tests , evolution of equilibrium configurations of compact stars ( solutions to the tolman - oppenheimer - volkoff equations ) , and the evolution of relativistically boosted tov stars transversing diagonally across the computational domain . the degree of complexity presented in these tests increases from purely special relativistic flows in flat backgrounds to fully general relativistic flows in dynamical spacetimes . in particular , the last test - bed ( the boosted star ) involves _ all _ possible terms in the coupled set of gr - hydro evolution equations and were carried out with a non - trivial lapse and shift vector . we found a simple , yet effective treatment for handling the surface region of a general relativistic self - gravitating compact object . the key idea is to replace the energy equation update by the condition of adaiabatic flow in regions of low density . while the surface region is not changing the overall dynamics of the star , numerical instabilities there could halt the numerical evolution if uncontrolled . the capability to handle the surface region in a stable fashion is important for the application of the code to the study of neutron star astrophysics . we have demonstrated this capability in the equilibrium and boosted star test - beds . refinement of this treatment for long term stability is presently being investigated . additional code calibrations that are underway include long - term stability analysis of single neutron stars , comparisons of waveforms from perturbed neutron stars , and comparisons with one - dimensional and axisymmetric ( 2d ) independent gr - hydro codes that we ( together with our collaborators ) constructed @xcite . those will be reported in later papers in this series . the formulation of the coupled set of equations and the numerical code reported in this paper were used for the construction of the milestone code `` gr3d '' for the nasa neutron star grand challenge project ( for a description of the project , see http://wugrav.wustl.edu/relativ/nsgc.html ) . the goal of this project is to develop a code for general relativistic astrophysics , and in particular , one that is capable of simulating the inspiral coalescence of a neutron star binary system . the coalescences of neutron star binaries are expected to be important sources of gravitational waves for interferometric detectors . the strongest signal will come from the highly dynamic `` plunge '' during the final phase of the inspiral ; a fully general relativistic code provides the only way to calculate this portion of the waveform . a version of the code which passed the milestone requirement of the nasa grand challenge project , has recently been released to the community @xcite . this code has been benchmarked at over 140 gflop / sec on a @xmath158 node cray t3e with a scaling efficiency of over 95% , showing the potential for large scale 3d simulations of realistic astrophysical systems . further development of our general relativistic code , and its application to the specific study of the neutron star coalescence scenario , will be described in later papers in this series . to summarize , this paper presents the first ( and necessary ) steps towards constructing an accurate and reliable tool for the numerical study of astrophysical phenomena involving matter at relativistic speeds and strong gravitational fields . the general relativistic hydrodynamical module mahc " presented and studied in this paper is coupled to the cactus " code for the spacetime evolution . the cactus code is being developed by an international collaboration , with a major contribution coming from the albert einstein institute in potsdam ( germany ) , and with significant contribution from the relativity group ( wugrav ) at washington university in st . louis , missouri , and from colleagues at the national center for supercomputing applications in urbana , illinois . further development has been carried out at the university of the balearic islands in mallorca ( spain ) , the university of valencia ( spain ) , and elsewhere . the hydrodynamical module was developed mainly at wugrav , with significant contributions from the potsdam group , and has benefited from interactions with the hydro group at the university of valencia . we would like to thank miguel alcubierre , gabrielle allen , pete anninos , toni arbona , carles bona , steve brandt , bernd brgmann , dan bullok , tom clune , teepanis chachiyo , ming c. chu , greg comer , thomas dramlitsch , comer duncan , ed evans , ian foster , tom goodale , carsten gundlach , philip gressman , philip hughes , jos mara ibez , sai iyer , gerd lanferman , joan mass , peter miller , philippos papadopoulos , manish parashar , bo qin , k.v . rao , paul saylor , bernard schutz , edward seidel , john shalf , hisa - aki shinkai , joan stela , doug swesty , ryoji takahashi , robert young , paul walker , ed wang , and william wu for useful discussions and various help with the code development . this work is supported by nsf grants phy 96 - 00507 and 96 - 00049 , nsf nrac allocation grant no . mca93s025 , nasa grant nasa - nccs5 - 153 , the albert einstein institute , and the institutes of mathematical sciences , chinese university of hong kong . one of us ( j.a.f ) acknowledges financial support from the tmr program of the european union ( contract number erbfmbict971902 ) . c. bona , j. i. nez , j. mart , and j. mass , in _ gravitation and general relativity : rotating bodies and other topics _ , vol . 423 of _ lecture notes in physics _ , edited by f. chinea ( springer - verlag , new york , 1993 ) , chap . shock capturing methods in 1d numerical relativity . the source and documentation of the released code can be downloaded at http://wugrav.wustl.edu /codes / gr3d . for credit of the code development , see the document http://wugrav.wustl.edu /codes / gr3d / nasa_ms2.ps .	this is the first in a series of papers on the construction and validation of a three - dimensional code for general relativistic hydrodynamics , and its application to general relativistic astrophysics . this paper studies the consistency and convergence of our general relativistic hydrodynamic treatment and its coupling to the spacetime evolutions described by the full set of einstein equations with a perfect fluid source , complimenting a similar study of the ( vacuum ) spacetime part of the code @xcite . the numerical treatment of the general relativistic hydrodynamic equations is based on high resolution shock capturing schemes , specifically designed to solve non - linear hyperbolic systems of conservation laws . these schemes rely on the characteristic information of the system . a spectral decomposition for general relativistic hydrodynamics suitable for a general spacetime metric is presented . evolutions based on different approximate riemann solvers ( flux - splitting , roe , and marquina ) are studied and compared . the coupling between the hydrodynamics and the spacetime ( the right and left hand side of the einstein equations ) is carried out in a treatment which is second order accurate in _ both _ space and time . the spacetime evolution allows for a choice of different formulations of the einstein equations , and different numerical methods for each formulation . together with the different hydrodynamical methods , there are twelve different combinations of spacetime and hydrodynamical evolutions . convergence tests for all twelve combinations with a variety of test beds are studied , showing consistency with the differential equations and correct convergence properties . the test - beds examined include shocktubes , friedmann - robertson - walker cosmology tests , evolutions of self - gravitating compact ( tov ) stars , and evolutions of relativistically boosted tov stars . special attention is paid to the numerical evolution of strongly gravitating objects , e.g. , neutron stars , in the full theory of general relativity , including a simple , yet effective treatment for the surface region of the star ( where the rest mass density is abruptly dropping to zero ) . the code has been optimized for massively parallel computation , and has demonstrated linear scaling up to 1024 nodes on a cray t3e .
the scaling of the cmos transistor has been the driving force behind the tremendous increase in microprocessor performance observed during the last decades . while the problems of the past were dominated by manufactural aspects , one now faces the first fundamental physical limitations as structures in logical devices approach atomic dimensions . as a rule of thumb , the thickness of the gate - oxide has to be directly proportional to the channel length in a mosfet ( metal oxide silicon field effect transistor ) device . in the course of the ongoing miniaturization , also the thickness of this insulating layer is being continually reduced . by the year 2007 , gate - oxides in a transistor will approach a thickness of 1.5nm@xcite , which corresponds to about ten typical atomic distances . the quantum mechanical tunneling currents through such a thin oxide are intolerable , and cause increased power consumption and deteriorated switching characteristics of the transistor . replacing sio@xmath0-based oxides , nowadays employed as gate dielectric , is one of the `` key challenges '' to the semiconductor industry@xcite , which has to find a solution within the next 4 to 5 years . employing high - k ( @xmath1 large dielectric constant ) oxides would allow for a greater physical thickness while preserving the electrical properties . possible candidates have to meet an extensive list of requirements@xcite such as sufficiently large band offsets , thermodynamical stability or interface quality . sio@xmath0 system meets all these requirements in an unparallelized way@xcite . first attempts to grow alternative oxides on si(001 ) did not yield satisfactory results for a variety of reasons often originating in the interface region . in the course of changing to a new gate material , one also considers to now grow crystalline oxide layers . it has been a paradigm that the gate oxide must be amorphous in order to avoid dislocations and grain boundaries , which provide natural pathways for leakage currents and atom diffusion . however , in the very small devices used in the near future , the probability for such a defect may be sufficiently small . on the other hand , epitaxial oxides bear the promise of very low defect concentrations at the interface to silicon , which is crucial for efficient device operation . the first account of heteroepitaxial growth of a transition metal oxide has been presented by a seminal work of mckee@xcite who succeeded in growing srtio@xmath2 on si(001 ) . for the divalent elements detailed studies have been performed@xcite . as similar understanding does not yet exist for the technological relevant early transition metals such as zirconium . theoretical studies on zr are limited bulk interface calculations of silicates@xcite . the deposition and formation of an interface between one of the major contenders for high - k oxides , namely zirconia ( zro@xmath0 ) , and silicon has not yet been investigated with ab - initio simulations . as a first step we performed state - of - the - art electronic structure calculations of the deposition of zr on si(001 ) up to a coverage of two monolayers ( ml ) . we performed first - principles calculations within the framework of density functional theory@xcite using the gradient corrected density functional of perdew , burke and enzerhof@xcite . the electronic structure problem has been solved with the projector augmented waves method@xcite , which uses augmented plane waves to describe the full wave - functions and densities without shape approximation . the paw method as implemented in the cp - paw code employs the car - parrinello approach@xcite to minimize the total energy functional . the core electrons are described within the frozen core approximation with the semi - core ( 4@xmath3 and 4@xmath4 ) shells of the zr atoms treated as valence electrons . plane wave cutoffs of 30 and 60ry for the wave functions and the density have been used . for metallic system we minimized the mermin functional with respect to the occupation numbers which yields the fermi - dirac distribution for the electrons . an electronic temperature of 1000k was used . the zero - kelvin result has been extrapolated using the method suggested by @xcite . the reconstructed si(001 ) surface is the template for our growth studies . it is modeled by a slab consisting of 5 silicon layers , where the bottom layer is saturated with two hydrogen atoms per silicon . the position of the hydrogen atoms and the lowest silicon layer has been frozen . supercells with 16 atoms per layer have been used for the calculations of zr in the dilute limit , 8 atoms per layer for all higher coverages . in the first case , the distance between two periodic images of an adatom is 15.36 . for all surface calculations the k - mesh density corresponds to 64 lateral k - points per @xmath5 surface unit cell . fig . [ fig1 ] . shows the structure of the reconstructed si(001 ) surface . excellent accounts covering the full complexity of the surface reconstruction can be found in literature@xcite . the resulting structure can , however , be explained by the following considerations : while every bulk silicon is 4-fold coordinated , the surface atoms lack their upper bonding partners which leaves 2 dangling bonds per atom , each occupied by one electron . in a first step , two neighboring silicon atoms form a dimer bond initially parallel to the surface which leaves one dangling bond per atom . now both electrons are transferred into one dangling bond to reduce the number of unpaired electrons to zero . this results in the buckling as the silicon with the filled dangling bond prefers a tetrahedral bonding arrangement , while the other prefers a planar sp@xmath6 configuration . from the electrostatic point of view , the alternating buckling behavior is favorable . our calculations readily reproduce previously reported results such as the @xmath7 reconstruction and the difference in @xmath8-coordinate of two silicon atoms within a dimer@xcite . reconstruction . the figure represents one supercell of the slab as used as a template in our calculations.,scaledwidth=45.0% ] the energy surface of a zr adatom on si(001 ) , has been obtained by freezing the lateral position of the adsorbed atom relative to the slab backplane , while all other degrees of freedom were fully relaxed . we used a grid of twelve zr positions in the irreducible zone of the reconstructed @xmath9 silicon surface . the resulting total energy surface for an isolated zr on top of si(001 ) is shown in fig . two nearly degenerate positions can be identified . one is located in the valley right in the middle of two dimers of neighboring rows and the other on top of a dimer row between to adjacent dimers . a third , local minimum , is located in the valley and has an energy 0.30ev higher relative to the most favorable positions . the diffusion is quasi one - dimensional with barriers of 0.70ev for diffusion parallel to the dimer rows and 1.63ev from the valley to the row . in order to investigate the formation of a continuous film , we increased the coverage . the structures with coverages of 0.25 ml and 0.5 ml have similar energies per zr atom as the dilute limit ( see fig . [ fig3 ] ) . a wealth of complex structures has been found . here , however , we only summarize the main trends of the chemical binding . and bulk si.,scaledwidth=45.0% ] to first approximation zr prefers a formal 4 + charge state the projected density of states of the zr @xmath10-states is located well above the fermi level . it is well known that for transition metal cations the @xmath3-and @xmath4-electrons are located above the , and have only a minor effect on the occupied states . the silicon dimers at the surface accept up to two electrons in their dangling bonds . a clear structural indication that the dangling bonds are filled is the disappearance of the dimer buckling . further electrons that are supplied at increased coverage of zr , occupy the antibonding states of the dimer bond . as a consequence the silicon dimers break up . this happens at a coverage of half a ml and above . it should be noted , however , that metallic states accept some of the electrons so that the number of broken dimers does not directly correlate with the number of zr adatoms in the ratio one to one . as a result of the interplay between breaking dimers and metallic states we find fairly complex reconstructions of the surface structure for intermediate coverages . upon increase of the coverage to a full ml , the energy drops by 0.64 ev per adatom ( see fig . [ fig3 ] ) . the stable structure with ml coverage is shown in fig . all dimer bonds of the surface layer are broken , and the zr atoms occupy the centers of the square array of surface silicon atoms in the resulting @xmath5 reconstruction . such a layer is one structural element of bulk zrsi@xmath0 . the energy gain is , however , not due to the surface geometry but can be attributed to a dimer reconstruction of the silicon subsurface . a metastable state without this reconstruction is even higher in energy than adsorbed zr at lower coverages . during the reconstruction , the silicon atoms in the layer underneath the zrsi surface layer form dimers analogous to the bare silicon surface . in contrast to the dimer row reconstruction of the silicon surface , these subsurface dimers are not buckled and are arranged in a checkerboard instead of a row pattern . at a coverage of two mls a second zrsi layer is formed . this configuration is shown in fig . the second layer is nearly identical to the first , but is shifted laterally . the double layer is again a structural element of bulk zrsi@xmath0 , the structure of which is shown in fig . the zrsi double layer consists of one - dimensional si zig - zag chains separated by zr atoms . the zr atoms lie approximately in the plane of the upper and lower atoms of the si chain . as for ml coverage the first silicon layer underneath the zrsi double layer exhibits a dimer reconstruction . ( compare to fig . note the dimer reconstruction of the silicon layer underneath the zrsi double layer . , scaledwidth=45.0% ] bulk . it consists of pure silicon layers separated by a zrsi double layers . this double layer exhibits si zig - zag chains separated by zr atoms . , scaledwidth=45.0% ] despite an additional energy gain from the ml coverage to a coverage of two mls of 0.25ev per adatom , even this structure is still 0.67 ev higher in energy than bulk zrsi@xmath0 . while the atomic process has not yet been resolved in every detail , our findings give strong indications for the nucleation of the silicide . we presented the results of state of the art ab - inito electronic structure calculations aiming at understanding the deposition of zr atoms on a si(001 ) surface as it is the case in an mbe reaction chamber . our results are summarized in fig . [ fig3 ] , which shows the energy per zr adatom as a function of the coverage . for coverages below one monolayer the the energy is nearly independent of the coverage . at a coverage of 1 ml we observe a sharp drop in energy by 0.64ev , followed by further drops in energy for higher coverages . all structures are less stable than bulk silicide . our findings suggest that islands with a local coverage of 1 ml or higher are formed even at low coverages . the islands contain structural elements of bulk zrsi@xmath0 , which is more stable than any surface structure . therefore a likely scenario is the formation of bulk silicide grains , that disrupt the surface morphology , and are detrimental for epitaxial growth . modification of growth conditions , such as exposing the surface to a oxygen containing ambient , may bypass silicide formation during the first growth steps . this work has been funded by the european commission in the project ( integration of very high - k dielectrics with cmos technology ) ist-2000 - 28495 and by the aurora project ( ) of the austrian science fond . this work has benefited from the collaborations within the esf programme on electronic structure calculations for elucidating the complex atomistic behavior of solids and surfaces. r.a . mckee , f.j . walker and m.f . chisholm , science * 293 * 468 ( 2001 ) . x. yao et al . b * 59 * 5115 ( 1999 ) . k. ojima , m. yoshimura , and k. ueda , surf . * 491 * 169 ( 2001 ) . a. herrera - gmez et al . , phys . b * 61 * 12988 ( 2000 ) . r. z. bakhtizin et al . , j. vac . b * 14 * 1000 ( 1996 ) . w. c. fan , n. j. wu , and a. ignatiev , phys . b * 42 * 1254 ( 1990 ) . j. wang et al . , phys . b. * 60 * 4968 ( 1999 ) . a. kawamoto et al . , j. appl . phys . * 90 * 1333 ( 2001 ) . p. hohenberg and w. kohn , phys . rev . * 136 * , b864 ( 1964 ) . w. kohn and l.j . sham , phys . rev . * 140 * , a1133 ( 1965 ) . perdew and y. wang , phys . b * 45 * 13244 ( 1992 ) . blchl , phys . b * 50 * , 17953 ( 1994 ) . r. car and m. parrinello , phys . * 55 * , 2471 ( 1985 ) . gillan , j.phys . : condens . matter 1 , 689 ( 1989 )	metal deposition of zr an a si(001 ) surface has been studied by state - of - the - art electronic structure calculations . the energy per zr adatom as a function of the coverage shows , that zr forms silicide islands even at low coverages . adsorbed zr is thermodynamically unstable against the formation of bulk silicide zrsi@xmath0 . the observation that the islands consist of structural elements of the bulk silicide is an indication that silicide grains will form spontaneously .
since the discovery of cmb anisotropies by the cobe team ( smoot 1992 ) there has been rapid progress on the observational front , culminating with publication earlier this year of evidence for multiple acoustic peaks in the cmb power spectrum ( netterfield 2001 ; lee 2001 ; halverson 2001 ) . it has long been known that accurate measurements of the cmb anisotropies can be used to estimate parameters characterising the primordial fluctuations , the geometry of the universe and its matter content ( _ e.g. _ jungman etal 1996 ; bond , efstathiou & tegmark 1997 ; zaldarriaga , spergel & seljak 1997 ) . in fact , the most recent measurements paint a gratifyingly consistent picture compatible with the simplest models of inflation ( _ i.e. _ a spatially flat universe with scale - invariant adiabatic fluctuations ) . furthermore , the derived value of the baryon density @xmath2 appears to be consistent with the value @xmath3 inferred from primordial nucleosynthesis and deuterium abundance measurements from quasar absorption line spectra ( burles , nollett & turner 2001 , and references therein ) . it has also long been known that there are significant degeneracies amongst cosmological parameters estimated from cmb anisotropies , _ i.e. _ parameter combinations exist that produce nearly identical cmb power spectra ( bond 1994 ; efstathiou & bond 1999 , hereafter eb99 ) . the best known is the _ geometrical degeneracy _ between the matter and vacuum energy densities , @xmath4 and @xmath5 , and the curvature @xmath6 . this degeneracy is almost exact and precludes reliable estimates of either @xmath5 or the hubble parameter @xmath7 from measurements of the cmb anisotropies alone . the existence of parameter degeneracies means that the best fitting parameters and their errors can be extremely sensitive to the chosen parameter set ( _ e.g. _ whether the universe is assumed to be spatially flat ) and to adopted ` prior distributions ' ( _ e.g. _ observational constraints on the hubble constant ) . this complicates the interpretation of cmb anisotropy parameter studies and the intercomparison of limits determined by different authors . almost all cmb parameter analyses ( with the noteable exception of the work of tegmark and collaborators , ( tegmark 1999 ; tegmark , zaldarriaga & hamilton , 2001 ; wang , tegmark & zaldarriaga , 2001 ) have ignored a tensor component ( see for example , lange 2001 ; jaffe 2001 ; de bernardis 2001 ; pryke 2001 ; stompor 2001 ) . this special case is certainly interesting because a wide class of inflationary models predict a negligible contribution from tensor modes . indeed it has been argued persuasively that the absence of tensor modes is generic to any model in which the inflaton potential is related to the higgs sector of a grand unified theory ( lyth 1997 ) . however , so little is known about inflation ( if indeed inflation occured ) that it may be dangerous to neglect a tensor mode , particularly if some cosmological parameters of interest are sensitive to tensor modes . in fact , in single - field inflation models the relative amplitudes of the tensor and scalar tensor modes ( @xmath8 ) and their spectral indices ( @xmath9 and @xmath10 ) are related to the inflaton potential and its first two derivatives ( _ e.g. _ see hoffman & turner 2001 for a recent discussion ) . in this class of model the relation between @xmath10 and @xmath9 is model dependent while the relation between @xmath9 and @xmath8 depends only on the validity of the slow - roll approximation . however , even the latter relation can be violated in multi - field inflationary models . these relations may also be violated in some superstring inspired models ( see _ e.g. _ lidsey , wands & copeland 2000 , and references therein ) . examples of the latter include the pre - big bang model of veneziano and collaborators ( _ e.g. _ buonnono , damour & veneziano , 1999 ) which it is argued can lead to ` blue ' scalar spectral indices ( @xmath11 ) , and the ekpyrotic scenario ( khoury , ovrut , steinhardt & turok 2001 ) which produces a strongly blue tensor mode spectrum . the point of view taken in this paper is to define a minimal parameter set on the assumption that the primordial fluctuations are gaussian , adiabatic and featureless ( _ i.e. _ defined by power - law spectral indices ) . thus the minimal model is specified by @xmath12 parameters : four parameters specifying the amplitudes and spectral indices of scalar and tensor components ( @xmath13 , @xmath14 , @xmath10 and @xmath9 , see section 2 for more precise definitions ) , four parameters defining the matter content and curvature of the universe ( @xmath2 , @xmath15 , @xmath5 , @xmath16 ) and a single parameter @xmath17 quantifying the optical depth to thomson scattering since the universe was reionized . however , as pointed out by eb99 , including a tensor component with no assumed constraints between the parameters @xmath10 , @xmath9 and @xmath14 introduces an new major degeneracy between cosmological parameters that we will call the _ tensor degeneracy _ in this paper . this degeneracy has a dramatic effect on the permitted ranges of some parameters , in particular the baryon and cold dark matter densities @xmath2 and @xmath15 . one approach to these degeneracies is to apply brute - force maximum likelihood analysis to a large parameter set ( see the papers by tegmark and collaborators ) . here we show that the observational data has now improved to the point that a simple principal component analysis of the fisher matrix defined by the data identifies the major degeneracy directions ( geometrical and tensor ) and provides a useful first approximation to the likelihood function . using the principal components it is easy to analyse the correlations between physical parameters introduced by the tensor degeneracy and to assess the effects of introducing external ( non - cmb ) constraints on the parameters . we use the compilation of band power estimates @xmath18 and their covariance matrix @xmath19 computed by wang ( 2001 , hereafter wtz01 ) from @xmath20 cmb anisotropy measurements . these band power estimates include a model for calibration and beam errors ( see wtz01 for further details ) . each band power estimate is related to the power spectrum @xmath21 of the cmb anisotropies by @xmath22 where @xmath23 is the window function for each band power ( also computed by wtz01 ) . these band - power estimates are plotted in fig . 1 . also plotted in this figure is a fiducial model with the following parameters : @xmath24 , @xmath25 , @xmath26 ( @xmath27 , @xmath28 ) , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 . these parameters provide an extremely good fit to the observations and are very close to the concordance values determined by wtz01 and the author from a full likelihood analysis . the addition of a small tensor component has little effect on the fit shown in fig . 1 , but is introduced to regularize the fisher matrix of equation ( [ 2.2 ] ) . 3.0 truein 3.7 truein given the covariance matrix of the observations , we can form the fisher matrix for the ( mean subtracted ) parameter set @xmath34 : @xmath35 in defining the parameter set @xmath36 , we use @xmath37 and @xmath38 rather than @xmath2 and @xmath15 and @xmath39 rather than @xmath16 . the latter expresses the geometrical degeneracy ( see eb99 ) , since if all parameters other than @xmath16 and @xmath5 are held fixed , the condition @xmath40 preserves the positions of the acoustic peaks for small variations of the parameters about those of the fiducial model . the normalization parameter @xmath13 is defined following eb99 so that @xmath41 is the mean band power ( @xmath42 ) of a model relative to that of the fiducial model plotted in fig . 1 . with this definition , @xmath13 is observationally well constrained and is largely decoupled from variations in cosmology and the optical depth @xmath17 ( unlike measures of the amplitude related to low multipoles ) . the tensor to scalar ratio @xmath14 is defined so that @xmath43 . with these definitions , we can compute the fisher matrix ( [ 2.2 ] ) from derivatives of the power spectra @xmath44 and @xmath45 , as in standard analyses of parameter forecasting ( jungman 1996 , bond 1997 ) . alternatively , we could compute the hessian matrix by estimating second derivatives of the likelihood function around the maximum likelihood value either directly or by summing over first and second derivatives of @xmath46 . in most situationa using either the fisher and hessian matrixes should give closely similar results , and in fact , the second derivatives of @xmath46 are often ignored in order to stabilise numerical evaluations of the hessian ( see _ e.g. _ press 1992 ) . having computed the fisher matrix , we diagonalise it , @xmath47 the matrix @xmath48 defines a set of principal components , @xmath49 , _ i.e. _ orthogonal linear combinations of the original parameters , @xmath50 such that the variance of the component @xmath51 is equal to @xmath52 . ( see be99 for an application of principal component analysis to the map and planck satellites . * table 1 : principal components for data of fig . 1 * ccccccccccc & @xmath53 & @xmath54 & ln @xmath15 & @xmath10 & @xmath13 & @xmath17 & @xmath5 & @xmath55 & @xmath9 & @xmath14 @xmath56 & @xmath578.4923e-03 & @xmath578.0062e-02 & -1.8131e-01 & @xmath572.0836e-02 & -1.5648e-01 & -6.8989e-03 & -3.0498e-03 & @xmath579.6731e-01 & @xmath573.1160e-03 & -6.4147e-03 @xmath58 & 1.6231e-02 & @xmath572.9071e-02 & -3.1679e-02 & -3.7260e-01 & @xmath579.0306e-01 & @xmath571.0592e-01 & @xmath576.3355e-02 & @xmath571.4733e-01 & -4.5002e-02 & @xmath576.9623e-02 @xmath59 & 3.1626e-02 & -1.5949e-01 & -1.0387e-01 & @xmath578.0645e-01 & @xmath573.9347e-01 & -2.9357e-01 & -1.8037e-01 & @xmath573.5921e-02 & @xmath571.1322e-01 & -1.6053e-01 @xmath60 & 7.1340e-02 & -7.3973e-01 & @xmath571.7199e-01 & @xmath571.7869e-01 & -1.1711e-02 & @xmath574.2868e-01 & @xmath573.4928e-01 & @xmath579.3853e-02 & -1.7909e-01 & @xmath572.1096e-01 @xmath61 & 1.3033e-01 & @xmath575.4011e-01 & @xmath576.8276e-01 & @xmath573.2525e-01 & @xmath577.0219e-02 & @xmath572.4081e-01 & @xmath572.1371e-01 & @xmath579.0979e-02 & -1.0073e-01 & @xmath579.6515e-02 @xmath62 & 2.8274e-01 & -2.1034e-01 & @xmath574.2130e-01 & -1.2216e-01 & -9.8231e-03 & -4.0134e-01 & -5.1452e-01 & @xmath579.6988e-02 & -9.6036e-02 & @xmath575.6463e-01 @xmath63 & 8.9800e-01 & @xmath572.1023e-01 & -3.8577e-01 & @xmath571.4257e-01 & -1.7712e-03 & -3.3061e-01 & @xmath574.5135e-01 & -8.9152e-02 & -4.5136e-01 & @xmath575.1182e-01 @xmath64 & 1.5724e+00 & @xmath571.9488e-01 & -3.5449e-01 & @xmath571.8691e-01 & @xmath573.5187e-03 & @xmath575.7374e-01 & -3.6421e-01 & -8.0544e-02 & @xmath572.6539e-01 & @xmath575.1250e-01 @xmath65 & 3.8197e+00 & -4.4314e-02 & @xmath578.7287e-02 & -4.9695e-02 & -1.3832e-03 & -2.4747e-01 & @xmath574.4122e-01 & @xmath571.9666e-02 & @xmath578.1212e-01 & @xmath572.6851e-01 @xmath66 & & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & @xmath75 the eigenvalues and components of @xmath76 for our chosen set of variables and fiducial model are listed in table 1 . the principal components have been ordered by their expected variance so that @xmath56 is the best determined parameter and @xmath65 is the worst . the last line in the table lists the predicted variances of the parameters @xmath36 , ( @xmath77 ) . as can be seen from table 1 , the present data can be used to constrain three parameter combinations well and three extremely poorly , with the remaining constrained at intermediate levels of accuracy . the best and worst determined principal components have a straightforward interpretation . @xmath56 has a high weight from @xmath55 , and so , in effect , measures the positions of the acoustic peaks ; @xmath58 has a high weight from the overall amplitude of the spectrum @xmath13 ; @xmath59 has high weights from the scalar spectral index @xmath10 and @xmath17 and provides a measure of the shape of the fluctuation spectrum . the component @xmath65 accounts for almost all of the variance of @xmath5 , @xmath9 and @xmath14 , and describes the geometrical degeneracy and the extremely poor constraints that the present data place on tensor modes . the component @xmath64 describes the tensor degeneracy and accounts for the almost all of the variance of the parameters @xmath37 , @xmath38 , @xmath10 and @xmath78 . together , the components @xmath65 and @xmath64 accounts for almost all of the variance of all of the parameters with the exception of the amplitude @xmath13 . the principal component analysis therefore suggests that acceptable models lie on a plane within the @xmath12 dimensional parameter space . this is illustrated in fig . 2a which shows likelihood contours in projections chosen to illustrate the geometrical and tensor degeneracies . the upper panels show the geometrical degeneracy in the parameter pairs @xmath79 and @xmath80 . the dashed lines show the degeneracy directions defined by the component @xmath65 . the solid line in the @xmath81 plane shows the constraint @xmath82 . the solid line in the @xmath80 plane is computed from the constraint equation @xmath83 with @xmath2 , @xmath15 and @xmath55 set to the values of the target model . the dotted lines in the lower two panels show the degeneracy directions defined by the component @xmath64 . the fisher matrix and the associated principal component analysis is approximate and it is not obvious _ a priori _ how well they decribe degeneracies especially for parameter values that are quite a long way from those of the target model . fig . 2b shows the analogous contours computed from a full likelihood analysis of the data of fig . 1 ( see the companion paper of efstathiou 2001 for a detailed discussion ) . the agreement is surprisingly good . the general directions of the geometrical and tensor degeneracies follow those of the fisher matrix analysis . there are some differences however . the low order cmb multipoles offer some discrimination of models with high values of @xmath5 via the integrated sachs - wolfe effect ( see eb99 ) , thus models with @xmath84 are excluded by the data . the maximum likelihood analysis shows that the tensor degeneracy allows high values of @xmath2 but that the likelihood function falls sharply for @xmath85 ( interestingly only just below the value favoured from primordial nucleosynthesis deuterium constraints ) . 3.4 truein the three filled circles in fig . 2 show the parameter values for the target model and for two nearly degenerate models with extreme values of @xmath2 chosen to lie along the tensor degeneracy direction . ( the device used here , of varying low order principal components to produce degenerate pairs of models , was used by eb99 to investigate parameter degeneracies for the map and planck satellites . ) one model has @xmath86 and the other has @xmath87 . the models were intentionally chosen to have similar values of @xmath5 , consequently the high baryon density model has a low value of @xmath7 which one might argue is incompatible with independent measurements of the hubble constant ( freedman 2001 ) . however , our purpose here is to show how the constaints on @xmath2 derived from cmb anisotropies _ alone _ are weakened when tensor modes are included . the high and low baryon density models are compared to the data and to the fiducial model in fig . despite the very different parameter values , the models produce almost identical cmb power spectra by construction . the exact likelihood analysis ( fig . 2b ) shows that the model with the high baryon density is compatible with the data at about the @xmath88 level . the low baryon density model is formally excluded by the data at a high level of significance ( @xmath89 ) because it fails to match the height of the first acoustic peak . this is a characteristic feature of models with a low baryon density nevertheless , the diagram is interesting because it shows that that the lower limits on @xmath2 are extremely sensitive to any residual systematic errors that might affect the peak heights ( see _ e.g. _ de bernardis 2001 , figure 1 ) . the reader might question the usefulness of the results presented in the previous section . firstly , the principal component analysis provides only an approximate description of the parameter degeneracies , whereas they emerge precisely from a brute force maximum likelihood analysis . secondly , models at the extreme ends of the ranges allowed by the tensor degeneracy have unusual parameters ( for example , the model with @xmath90 in fig . 3b has a low hubble constant and a high value of the scalar spectral index ) and so are surely excluded by other observational constraints . we discuss each of these points in turn . _ ( i ) effective dimensionality : _ the main use of the principal component analysis is to assess the effective dimensionality of the space of acceptable models within the multidimensional space defined by the physical parameters @xmath36 and to assess whether this effective dimensionality is sensitive to changes of the parameter set . for example , let us assume that we are interested in the values of the parameters @xmath2 and @xmath15 . the principal component analysis tells us that most of the variance of these parameters is contributed by only two poorly constrained principal components . the values of these parameters are therefore affected by major parameter degeneracies which can only be removed by imposing external constraints or by performing a fundamentally different type of cmb exeperiment . for example , the effects of the tensor degeneracy on @xmath2 and @xmath15 can be broken by extending the cmb measurents to much higher multipoles ( see eb99 ) and/or by setting limits on a tensor component from an analysis of a b - type polarization pattern in the cmb ( see _ e.g. _ kamionkowski & jaffe , 2000 , and references therein ) . _ ( ii ) complementary information : _ cmb parameter degeneracies can be broken by invoking complementary information . a well known example is the combination of type ia supernovae measurements with the cmb to break the geometrical degeneracy ( see _ e.g. _ de bernardis 2001 for a recent analysis ) . wtz01 break the tensor degeneracy by combining the cmb data with a number of constraints including estimates of the power spectrum on small scales from observations of the ly@xmath91 forest ( croft 2001 ) and limits on the hubble constant from hst hubble key project ( freedman 2001 ) . the problem here is that it becomes progressively more difficult to assess whether parameter values are affected by systematic errors as more external constraints are applied , particularly if the external constraints involve complex observations and assumptions . ( for example , how can we test empirically whether density fluctuations in the inter - galactic medium as traced by the ly@xmath91 forest match those of the dark matter ? ) . even if the parameter values are shown to remain consistent as external constraints are applied , there is no guarantee that the final combined likelihood distribution will be accurate . ideally , we would like to apply as small a number of external constraints as possible using data sets with well controlled errors . the fisher matrix analysis of the previous section offers a guide as to which external constraints will be most effective at breaking degeneracies . the effects of external constraints can easily be assessed using the fisher matrix : let @xmath92 be the covariance matrix of the parameters @xmath36 from external constraints , then the covariance matrix after combining with the cmb is @xmath93 as menstioned in section 2 , the models plotted in fig . 3 were chosen by construction to have similar values of @xmath5 , so it is clear that combining the cmb results with constraints on @xmath5 from type ia supernovae observations will have little effect on the parameters @xmath2 and @xmath15 . constraints on the hubble constant will be more effective in narrowing the range of allowed values of @xmath2 and @xmath15 . however , from fig . 2 we can see that the tensor degeneracy exhibits a strong correlation between @xmath2 , @xmath15 , and the scalar spectral index @xmath10 . even a relatively poor constraint on the scalar spectral index will lead to a sharp tightening of the @xmath2 and @xmath15 error contours . ( a constraint @xmath94 on the spectral index will narrow the error on the baryon density to @xmath95 , _ i.e. _ @xmath96 . ) this suggests that the recent estimates of the galaxy power spectrum from the 2df galaxy redshift survey ( percival 2001 ) will be highly effective at breaking the tensor degeneracy , leading to tight constraints on the matter content of the universe . this is borne out by a detailed analysis presented in a companion paper ( efstathiou 2001 ) . * aknowledgments : * i thank caltech for the award of a moore scholarship . i thank andrew liddle for discussions of inflationary models and various members of the 2df galaxy survey team for their comments on an earlier draft .	a principal component analysis of cosmic microwave background ( cmb ) anisotropy measurements is used to investigate degeneracies among cosmological parameters . the results show that a degeneracy with tensor modes the ` tensor degeneracy ' dominates uncertainties in estimates of the baryon and cold dark matter densities , @xmath0 , @xmath1 , from an analysis of cmb anisotropies alone . the principal component analysis agrees well with a maximum likelihood analysis of the observations , identifying the main degeneracy directions and providing an impression of the effective dimensionality of the parameter space . 0.2truein * key words : * cosmic microwave background- cosmology : miscellaneous . 0.3 truein
classical local thermodynamical equilibrium ( lte ) , plane - parallel , horizontally homogeneous , hydrostatic equilibrium model stellar atmospheres have been providing a basic tool for stellar spectroscopy for several decades , but are now known to suffer from various weaknesses when applied to accurate observations of individual stars . several lines of research have been adopted to improve the models . for example , in the context of late - type stars , we mention here three - dimensional hydrodynamical simulations of surface convection ( e.g. stein & nordlund 1998 , asplund et al . 2000 ) ; improvements in molecular opacities ( e.g. hauschildt et al . 1999a ; tsuji 2002 ) ; the adoption of spherical geometry ( e.g. plez , brett , & nordlund 1992 ; hauschildt et al . 1999b ) ; and semi - empirical modeling ( e.g. allende prieto et al . 2000 ) . lte has been one of the strongest hypotheses invoked extensively in solving the coupled problems of radiative transfer and structure in stellar atmospheres . this assumption enormously simplifies the complexity of the calculations , since the populations of all ions and molecules involved follow from the boltzmann and saha formulae . yet , lte has been recognized to be inappropriate in many areas of the hertzsprung - russell diagram . models that take departures from lte into account are traditionally labeled non - lte ( or nlte ) models . the main area where the nlte models have been applied are hot stars ( types o , b , a , hot white dwarfs and subdwarfs , and other objects with high - temperature atmospheres ; see , e.g. , hubeny & lanz 1995 ) . however , the advent of large telescopes with high - dispersion spectrographs , and the subsequent refinement of observations of late - type stars , call for an assessment of the possible departures from lte on the model atmospheres for these stars . non - lte calculations require detailed radiative and collisional rates of all involved transitions in order to solve the statistical equilibrium equations . the role of photoionization cross - sections should be emphasized , because changes of the degree of ionization ( with respect to the lte ionization equilibrium ) are a major non - lte effect in stellar atmospheres . not all of the required energies and cross - sections for the atomic / molecular processes that play a role in a stellar atmosphere are available . nonetheless , the available data has been expanded in recent years . large collaborative projects , such as the opacity project ( see , e.g. , cunto et al . 1993 ) , the iron project ( see , e.g. butler 1998 ) , and sam ( see , e.g. brage , judge , & brekke 1996 ) have been enormously helpful . long - term efforts on compilation of data , such as chianti ( e.g , dere et al . 1997 ) , vald ( kupka et al . 1999 ) , and the atomic spectroscopic database ( asd ) at the national institute of standards and technology ( nist ) , suplemented by numerous contributions from individuals and small groups , are to be noted . the model atoms presented here are built for use in the model stellar atmosphere code tlusty ( hubeny 1988 , hubeny & lanz 1995 ) . a family of model atoms and ions , primarily in the context of hot stellar atmospheres , has been made publicly available with the code . these models are briefly described in 2 . section 3 is devoted to the new set of model atoms , describing the numerical details of the smoothing process applied to the photoionization cross - sections , and the construction of model atoms . some comments on the limitations and practicalities for using these data , as well as a summary , are outlined in 4 . in the context of hot stellar atmospheres , a collection of model atoms for , , , , , , , , , and was prepared , ranging from very simple to quite complete structures . these models are essentially based on data extracted from topbase . to manipulate these large sets of atomic data , an interactive , idl - based tool , was developed : modion ( varosi et al . 1995 ; see also lanz et al . this interface program works with three files for each ion , containing the level energies , the line oscillator strengths , and the photoionization cross - sections , as extracted from topbase . the energy files have been updated with observed energies extracted from asd / nist ( martin 1997 ; wiese 1997 ) , so that all transitions will be calculated at their actual frequency . modion allows model atoms of various sophistication to be built by displaying a grotrian diagram from which the explicit nlte levels are interactively selected . although it might seem desirable to include as many levels as possible , this is often impracticable due to computing time and memory limitations . to overcome these limitations , some individual levels are merged into superlevels ( see hubeny & lanz 1995 ) . low - excitation levels are generally considered as individual levels , while levels of higher excitation are merged . the most excited levels are often not treated explicitly , but are assumed to be in lte with respect to the ground state of the next ion . model atoms and ions have been built for different purposes : simple model atoms ( typically less than 15 explicit levels ) are often quite appropriate for inclusion in photospheric structure calculations , while very detailed models ( over 40 explicit levels ) may be required for detailed line formation studies . after the level selection step , modion builds a list of bound - bound and bound - free transitions . oscillator strengths from topbase are assigned to bound - bound transitions between individual levels . modion takes care of the appropriate summing and averaging for transitions between superlevels , supplementing topbase data with hydrogenic values when data are missing . the necessary sums are similarly performed for photoionization cross - sections . tlusty is able to deal with detailed and complex representations of the cross - sections , but in most cases we use approximate , yet adequate , representations . using modion , we display the cross - sections in a @xmath0 plot of the photoionization cross - section vs. frequency , and hand - pick some points with a cursor . the approximate cross - sections are selected to be the lower envelopes of the detailed cross - sections , thus the autoionization resonances are mostly neglected . finally , when the selections have been completed , modion writes out a data file in the format required by tlusty . some limited upgrades have been made in a later stage to incorporate stark profiles for strong resonance lines or to introduce fine - structure in strong resonance doublets like @xmath1 or @xmath2 . this new set of model atoms is , like the modion models , based on data from topbase and nist . they treat , however , the photoionization cross - sections differently , and in more detail . as they are new and have never been used in calculations of radiative transfer and model atmospheres before , we describe them in depth . these models are , at the time of this writing , available for neutral na and s , singly ionized ne , and neutral and singly - ionized li , be , b , c , n , o , f , mg , al , si and ca . it has been acknowledged that errors in the computed energies of the atomic levels may result in spurious shifts of the frequencies at which strong photoionization resonances are predicted by theoretical calculations . consequently , several authors have suggested different procedures to smooth the opacity project data ( see , e.g. , verner et al . 1996 ; bautista , romano , & pradhan 1998 ) . in addition , smoothing the cross - sections reduces sometimes the number of data points to a more manageable level . bautista et al ( 1998 ) suggest a recipe for a convenient gaussian smoothing . they recommend a value for the width of the gaussian of @xmath3 , where @xmath4 is the energy of the ionizing photon . they apply such a smoothing to the photoionization cross - sections of ground levels of all atoms and ions included in topbase and to new data on @xmath5 provided by the iron project , producing resonance - averaged photoionization ( rap ) cross - sections . while in low - density environments , such as gaseous nebulae , virtually all atoms and ions are in ground states , this is not the case for stellar atmospheres . extending the calculations of rap cross - sections to the rest of the levels is therefore desirable . we deal with neutral and singly - ionized species of the elements included in topbase that have been described by @xmath6 coupling in asd / nist , excluding h and fe . in fact , the list covers most of the atomic species of relevance for studying late - type stellar atmospheres . we have closely followed the procedure described by bautista et al . the rap cross - section at a given energy , @xmath7 , is computed from the topbase photoionization cross - section , @xmath8 , through the integral @xmath9 } dx}{\displaystyle \int_{e_0}^{\infty } { \displaystyle \exp \left [ \frac{-(x - e)^2}{2 ( \delta e)^2 } \right ] } dx}\ ] ] where @xmath10 is the ionization threshold energy , and @xmath11 . @xmath10 is slightly lower than the energy difference between the level and the continuum , due to line merging near the series limit . to speed up the calculations , without losing accuracy , we restricted the limits of the integral ( 1 ) to @xmath12 . in order to avoid unwanted systematic effects at the minimum ( threshold ) and maximum energies in the topbase calculations , we performed a linear interpolation over the energies for which less than five points were available within @xmath13 . the rap cross - sections were derived at energies separated by the smallest steps allowed by the sampling theorem , using the recurrence formula @xmath14 from @xmath15 to @xmath16 . figure 1 shows two examples corresponding to the ground levels of , and : the thin solid line corresponds to the original cross - sections , and the filled dots represent the rap cross - sections . the following sections describe how radiative and collisional processes are accounted for in the the rap model atoms . only collisions with electrons are included . collisions with neutral h atoms may be important in late - type stars , but lacking term - dependent or species - dependent data , they are not considered in these models . the phenomenological approximations usually found in the literature are best implemented directly in the nlte codes , rather than in the input data files . the opacity project provides energies and oscillator strengths for atomic levels and radiative transitions among them . however , even though this database provides a large amount of atomic data of sufficient quality for many purposes , theoretical calculations can not , with few exceptions , predict the energies of the levels with accuracy . as a result , the wavelengths of the lines can not be predicted to the precision available to laboratory spectroscopists . there are also measurements of transition probabilities that may be more accurate than those in the op database . we have chosen to use the atomic energy levels and the spectral lines included in asd / nist , version 2.0 , to build models for the same atoms and singly - ionized ions for which the have derived rap cross - sections from topbase data . the photoionization cross - sections are given as a function of the energy from the threshold , and therefore they are simply shifted to the observed energy levels . topbase ignores fine structure , and so have we , by grouping the split levels into a single one with an averaged energy , using @xmath17 as weights ( where j is the quantum number that represents the total angular momentum ) , according to the relative populations of the levels expected in thermodynamic equilibrium . transitions between fine - structure states of a given level are ignored . a scaled hydrogenic photoionization cross - section has been assigned to the levels whose photoionization cross - sections were not listed in topbase , assuming that the ionization takes place to the ground state of the next ion : @xmath18 where @xmath19 is planck s constant , @xmath20 is the quantum principal number of the last electron in the configuration , @xmath21 the frequency , @xmath22 the charge of the ion ( i.e. , @xmath23 for neutrals , @xmath24 for once ionized , etc . ) , and @xmath25 the gaunt factor expressed as @xmath26 with the coefficients @xmath27 given by mihalas , heasley , & auer ( 1975 ) . units are used throughout the paper . the radiative transitions between fine - structure states ( @xmath28 , @xmath29 ) of two levels ( @xmath30 and @xmath31 ) have been summed to a single one with a @xmath32-value : @xmath33 where @xmath34 represents the statistical weight of a fine structure level . allowed and forbidden radiative transitions whose @xmath35-values are listed in the nist database have been included in the models . also , allowed radiative transitions not listed in nist have been included , assuming a scaled hydrogenic transition probability @xmath36 when more than one level in topbase with separate photoionization cross - sections was listed with identical energies in the nist lists , they were included in the models with their respective rap cross - sections . the states described in the nist database with a coupling other than @xmath6 have not been included , but states with configurations , terms , or energies listed as uncertain are included when matched by the topbase states within 0.08 ryd . only levels with energies lower than the series limit are kept . as an example , figure 2 displays the grotrian diagrams for and . table 1 lists the number of levels and radiative transitions considered for each model . we have formatted the described data to be read by the program tlusty . the code is designed to build non - lte model atmospheres , and it will be the tool we shall employ in future analyses of stellar spectra . there are several documents explaining the data format in detail ( hubeny 1988 ; hubeny & lanz 1997 ) , hence we shall not describe it here . in the files , a series of _ options _ that apply to the solution of the statistical equilibrium equations have been chosen . these options set several input parameters ( e.g. collisional ionization and excitation , depth - dependency of the line absorption profiles , etc . ) , which are briefly described below , but they can be changed at ease by the user ( see hubeny 1988 ) . free - free absorption for metals is accounted for by a hydrogenic cross - section with the gaunt factor set to one , while the exact gaunt factors are used for h and he ii . the line absorption profiles are assumed gaussian and depth - independent . all levels are connected with each other and with the continuum by collisions with electrons . how these collisional processes are treated is described in more detail below . mostly , but not all , applies as well to the modion models . the collisional rate for the transition @xmath37 is given by @xmath38 , where @xmath39 is the number density of electrons and @xmath40 is the excitation rate coefficient . excitations induced by electron collisions between states connected by radiatively permitted transitions are accounted for using van regemorter s formula ( van regemorter 1962 , mihalas 1972 ) , with the excitation rate coefficient as : @xmath41 where @xmath42 is the bohr radius , @xmath43 is boltzmann s constant , @xmath44 is the mass of the electron , @xmath45 is the threshold ionization energy for hydrogen , @xmath46 is the electron temperature , @xmath47 , and for ions @xmath48\ ] ] with @xmath49 ( transitions between levels with the same principal quantum number ) or @xmath50 ( otherwise ) ( mihalas 1978 ) , @xmath51 is the first - order exponential integral , but for neutral atoms @xmath52 ( auer & mihalas 1973 ) . collisional excitation between levels linked by forbidden radiative transitions is considered by means of the eissner - seaton formula ( seaton 1962 ) : @xmath53 adopting , @xmath54 for all cases . pradhan & peng ( 1995 ) and pradhan & zhang ( 2000 ) review some of the detailed calculations and scaling laws of effective collisional strengths that are available for particular ions . those data should update the approximate values adopted here . we also used an approximate formula ( seaton 1962 ) to evaluate collisional ionizations : @xmath55 with the photo - ionization cross - section at the threshold ( @xmath56 ) from topbase cross - sections ( when available ) , or the hydrogenic approximation ( with the gaunt factor equal to one ) @xmath57 where @xmath20 is again the principal quantum number of the last electron and @xmath22 the charge of the ion . collisional recombinations and de - excitations follow from the principle of detailed balance . detailed calculations should replace these approximate values where available ( see , e.g. , nahar & pradhan 1997 and nahar 1999 ) . we have smoothed the photoionization cross - sections of the opacity project for neutral na and s , singly ionized ne , and neutral and singly - ionized li , be , b , c , n , o , f , mg , al , si and ca , using a gaussian profile with @xmath58 , where @xmath4 is the energy of the ionizing photon . this procedure follows bautista et al . the smoothed cross - sections have been merged with the energy levels and line transition probabilities listed in the atomic and spectroscopic database at nist to derive model atoms and ions suitable for non - lte calculations of model atmospheres and line formation . this new collection of model atoms complements and , for the species in common , supersedes a previous dataset , the modion model atoms , developed for early - type stellar atmospheres . as an example , we show the results of two calculations involving the rap models . figure 3 shows the relative importance of neutral c , n , o , and mg on the continuum absorption in the near - uv spectrum of the sun ( @xmath59 k ; @xmath60 dex ; [ fe / h]=0.0 dex ) and procyon ( @xmath61 k ; @xmath62 dex ; [ fe / h]=0.0 dex ) . the spectral synthesis assumes lte and makes use of kurucz s atmospheric structures ( kurucz 1993 ) . although the models are based on observed energy levels and lines , together with detailed calculations of photoionization cross - sections , judging their appropriateness for non - lte calculations in stellar atmospheres is beyond the scope of this paper . several approximations and simplifications are implicit in the models and , therefore , caution is required . only detailed comparison with high quality observed spectra will determine the models usefulness . more complex ions not yet included in topbase are of interest . iron is at the head of this group , and new calculations of photoionization cross - sections are available ( bautista 1997 ) but not as part of topbase at the time of this writing . many other sources of data , although with a significant heterogeneity , exist and should be used to complement , check , and improve these models . such tasks and a detailed comparison with high - quality observations for a series of standard stars are in progress and will be reported in the future . we thank the people and institutions involved in creating and maintaining the asd / nist and topbase databases , which have been extensively used in this work . this research has been supported by nsf through grant ast-0086321 and by the robert a. welch foundation of houston , texas . crrcrr & 26 & 153 & & 37 & 143 + & 56 & 323 & & 18 & 72 + & 29 & 117 & & 38 & 134 + & 104 & 1090 & & 40 & 189 + & 89 & 615 & & 51 & 280 + & 54 & 242 & & 74 & 512 + & 52 & 209 & & 73 & 326 + & & & & 59 & 353 + & 32 & 192 & & & + & 71 & 434 & & 31 & 184 + & 33 & 160 & & 81 & 537 + & 57 & 215 & & 46 & 261 + & 64 & 313 & & & + & 79 & 543 & & 32 & 147 +	with the goal of producing a reliable set of model atoms and singly - ionized ions for use in building nlte model atmospheres , we have combined measured energy levels , critically - compiled line transition probabilities , and resonance - averaged calculations of photoionization cross - sections . a majority of the elements from li to ca are considered , covering most of the important species in late - type atmospheres . these include elements which contribute free electrons and/or continuous opacity in the ultraviolet ( e.g. , mg , and si ) , as well as trace elements whose abundance determinations rely on ultraviolet lines ( e.g. , b from b i lines ) . the new data complement and , for the species in common , supersede a previous collection of model atoms originally designed for use in studies of early - type stars .
large amplitude millisecond oscillations have now been observed during thermonuclear x - ray bursts from six low mass x - ray binary ( lmxb ) systems with the rossi x - ray timing explorer ( rxte ) ( see strohmayer , zhang & swank 1997 ; smith , morgan & bradt 1997 , zhang _ et al . _ 1996 ; swank _ et al . _ 1997 ; and strohmayer _ et al . the thermonuclear instability which triggers an x - ray burst burns in a few seconds the nuclear fuel which has been accumulated on the neutron star surface over several hours . this @xmath1 10@xmath2 difference between the accumulation and burning timescales means that it is extremely unlikely that the conditions required to trigger the instability will be achieved simultaneously over the entire stellar surface . this realization , first emphasized by joss ( 1978 ) , led to the study of lateral propagation of the burning instability over the neutron star surface ( see fryxell & woosley 1982 , nozakura , ikeuchi & fujimoto 1984 , and bildsten 1995 ) . the subsecond risetimes of thermonuclear x - ray bursts suggests that convection plays an important role in the physics of the burning front propagation , especially in the low accretion rate regime which leads to large ignition columns ( see bildsten ( 1998 ) for a review of thermonuclear burning on neutron stars ) . bildsten ( 1995 ) has shown that pure helium burning on neutron star surfaces is in general inhomogeneous , displaying a range of behavior which depends on the local accretion rate . low accretion rates lead to convectively combustible accretion columns and standard type i bursts , while high accretion rates lead to slower , nonconvective propagation which may be manifested in hour long flares . these studies emphasize that the physics of thermonuclear burning is necessarily a multi - dimensional problem and that _ localized _ burning is to be expected , especially at the onset of bursts . there is now good evidence that the oscillations seen during the rising phase of bursts from 4u 1728 - 34 are produced by spin modulation of such a localized thermonuclear hotspot on the surface of the neutron star , and that the observed oscillation frequency is a direct measure of the neutron star spin frequency ( see strohmayer , zhang & swank 1997 ) . these observations provide the most compelling evidence to date that neutron stars in lmxb are rotating with near millisecond periods . in this letter we present new burst data from the lmxb 4u 1636 - 54 which provides further evidence in support of the spin modulation hypothesis for the millisecond burst oscillations . we present data from a thermonuclear burst from 4u 1636 - 54 which reveals a strong transient oscillation during the burst rise at 1.723 ms with an initial amplitude of @xmath0 . we also discuss the implications of these findings for the spin modulation interpretation and how they can be used to place constraints on the mass and radius of neutron stars . finally , we discuss some implications of our observations for the current theories of kilohertz quasiperiodic oscillations ( qpo ) in lmxb . zhang _ et al . _ ( 1996 ) reported the discovery of 581 hz oscillations during several thermonuclear x - ray bursts from 4u 1636 - 54 . here we focus on a single burst detected from this source with rxte at 22:39:22 utc on december 28 , 1996 . this burst is unique in that it reveals strong oscillations at 581 hz _ prior _ to the onset of photospheric radius expansion . oscillations are not detected during the photospheric expansion phase , but reappear with much weaker amplitude after photospheric touchdown in a manner very similar to that reported for bursts from 4u1728 - 34 and ks 1731 - 26 ( strohmayer _ et al . _ 1997 ; and smith , morgan , & bradt 1997 ) . for this burst we obtained event mode data with 125 @xmath3s ( 1/8192 s ) time resolution and 64 spectral channels across the 2 - 100 kev pca bandpass . this data mode employs a pair of data buffers which are alternately read out after a selectable accumulation interval . each buffer can store @xmath4 16,000 events . in the case that the source countrate is high enough to fill the buffer before the end of an accumulation interval gaps in the data can occur . for the event data shown here we did have short gaps when the burst flux was high , but telemetry constraints preclude selecting a shorter accumulation interval . for these bursts we also had 125 @xmath3s , 2 - 90 kev binnned timing data which does not suffer from gaps . at the time of the burst in question the persistent countrate from 4u1636 - 54 in the total pca bandpass was about 2100 cts / sec . 4u1636 - 54 is classified as an atoll source and it was in the so called banana branch during these observations ( see hasinger & van der klis 1989 ) . we will present more detailed analysis of the accretion driven flux and properties of all the x - ray bursts observed in a subsequent paper . figure 1 shows the 2 - 90 kev pca lightcurve ( top ) and the hardness ratio ( 6 - 18 kev)/(2 - 6 kev ) at 1/16 s resolution for this burst . the drop in the hardness ratio at about 1.5 s is a clear indication of photospheric radius expansion . this burst had a peak flux in the 2 - 60 kev band of @xmath5 ergs @xmath6 s@xmath7 and a fluence in the same band of @xmath8 ergs @xmath6 . at the system s probable distance of 6.5 kpc ( see lawrence _ et al . _ 1983 ) these correspond to @xmath9 ergs s@xmath7 and @xmath10 ergs . we used the 2 - 90 kev , 125 @xmath3s binned data to compute a dynamic power spectrum by calculating power spectra from 2 s of data every 1/8 s through the burst . since the data overlap in this analysis the individual power spectra are not independent , however , this procedure provides a straightforward way to investigate the time evolution of the oscillations . the resulting dynamic power spectrum is presented in figure 2 , which shows contours of constant power spectral amplitude along with the burst profile superposed . this analysis reveals the presence of strong oscillations at about 580 hz on the rising edge of the burst as well as oscillations after photospheric touchdown at a frequency about 1 hz higher . as we discuss below , the oscillations during the rise are transient , lasting about 0.25 s , this introduces a characteristic width to the peak in the power spectrum which results in the broader distribution in frequency of the power spectral contours near the burst rise . we emphasize that the contours centered near 580.5 hz at about 2 s are not indicative of a real drop in the oscillation frequency at the burst onset . however , the change in frequency from burst rise to post - peak is real and is not caused by finite sampling time effects or the transient nature of the oscillations . this trend for the oscillation frequency to increase with time during some bursts has now been reported in all sources with millisecond burst oscillations except ks1731 - 26 , from which only a single burst with oscillations has so far been detected ( see strohmayer _ _ 1997 ; smith , morgan , & bradt 1997 ) . to further characterize the oscillations seen during the rising phase we performed an epoch folding period search on the rising interval . figure 3 shows the lightcurve of the burst , focusing on the rising portion . the interval used for this analysis is denoted by the dashed lines in figure 3 . the epoch folding reveals a strong oscillation at 1.723 ms . we folded the data on this period to determine the pulse profile and amplitude . the resulting pulse profile is shown in figure 4a . we fit a model with a constant countrate plus a sinusoid , @xmath11 , to the pulse profile and find an amplitude @xmath12 with an acceptable fit . since there is no evidence that the persistent flux component is oscillating at the frequency observed during the burst , to determine the pulsed amplitude we first subtracted from the folded lightcurve the average persistent countrate calculated from a 20 s interval prior to the burst . we note that there are no detectable oscillations during the interval when the drop in hardness ratio at about 1.5 s ( see figure 1 ) indicates the onset of photospheric radius expansion . this is consistent with the idea that the expansion of the photosphere destroys the coherent oscillations , probably both by scattering away the amplitude as well as changing the frequency quickly via the expansion of the surface layers . after photospheric touchdown ( see figure 1 at about 3.5 s ) the oscillations are again detected , but with slightly higher frequency and a much lower amplitude of about 5.7 % in the 2 - 60 kev bandpass . the presence of large amplitude oscillations very near the onset of this burst is similar to the behavior of bursts from 4u 1728 - 34 reported by strohmayer , zhang & swank ( 1997 ) . we found even higher oscillation amplitudes by folding a 62 ms interval close to the onset of the burst . the folded interval is denoted by the solid vertical lines in figure 3 . the background subtracted folded lightcurve obtained from this 62 ms interval is shown in figure 4b . we again fit the sinusoidal model to this lightcurve and found an amplitude of @xmath0 for this interval . this is easily the largest amplitude of the 581 hz oscillation seen during this burst . such a large amplitude almost the entire thermal burst flux is modulated only about 0.1 s after the onset of the burst is strong evidence that we are seeing a rotationally modulated , thermonuclear hotspot . strohmayer , zhang & swank ( 1997 ) reported spectral evidence for an increasing x - ray emitting area during burst rise using bursts from 4u 1728 - 34 observed with rxte . we performed similar spectral analysis in order to determine whether the same process is plausible for this burst as an explanation for the strong oscillations seen during the rise . using the 125 @xmath3s , 64 channel data we accumulated spectra through the burst on either 0.0625 , 0.125 , 0.25 or 0.5 s intervals depending on the countrate . we also accumulated a 20 s interval just prior to the burst for use as a background estimate . we fit the background subtracted spectra to a blackbody model to determine temperatures and fluxes as a function of time through the burst . we created a plot of @xmath13 versus @xmath14 , where @xmath14 is the bolometric flux and @xmath15 is the blackbody temperature in kev . for a true blackbody the quantity @xmath16 , where @xmath17 is the x - ray emitting area and @xmath18 is the distance to the source . we found results very similar to those reported by strohmayer , zhang & swank ( 1997 ) for bursts from 4u 1728 - 34 , that is , @xmath19 increases during the rise . this is strong evidence for an expanding x - ray emission region and thus supports the spin modulation hypothesis in this burst from 4u 1636 - 54 . we emphasize that this result suggests an increase in the x - ray emitting area _ prior _ to the onset of photospheric radius expansion . the onset of photospheric radius expansion is reflected in the drop in hardness ratio at about 1.5 s in figure 1 . it also corresponds to an increase in the emitting surface area , but it is caused by _ expansion of the photosphere due to radiation forces and not the _ lateral _ propagation of the burning instability . the blackbody spectral analysis indicates that the inferred emitting areas were approximately the same both immediately before and after the photospheric radius expansion . this result combined with the large difference in oscillation amplitudes before and after photospheric expansion suggests that the thermonuclear flash covered the entire surface of the neutron star . if the millisecond oscillations seen during bursts are in fact due to spin modulation , then detailed study and modelling of the oscillation amplitudes , pulse profiles and spectral variability with pulse phase during x - ray bursts can provide a wealth of information on the mass and radius of the neutron star . for example , the maximum modulation amplitude that can be obtained from a hotspot of a given angular size on a rotating neutron star is set by the strength of general relativistic light bending . for the case that rotation is not rapid enough to substantially distort the exterior spacetime from the schwarzchild spacetime , and this is the case even for spin periods of a few milliseconds ( lamb & miller 1995 ) , then the maximum amplitude depends only on the compactness of the neutron star , that is , the ratio of stellar mass to radius @xmath20 . stars which are more compact produce lower amplitudes due to flattening of the pulse by light bending ( see pechenick , ftaclas & cohen 1983 ; strohmayer 1992 ; and miller & lamb 1997 ) . since the intrinsic rotational modulation amplitude can only be decreased by other effects such as photon scattering ( see miller , lamb & psaltis 1997 ; brainerd & lamb 1987 ; and kylafis & phinney 1989 ) or the viewing geometry of the spot , the maximum observed oscillation amplitude represents a lower limit to the intrinsic amplitude . thus an observed amplitude can be used to place an upper limit on the compactness of the neutron star , that is , if the star were more compact than some limit it would not be able to produce a modulation amplitude as large as that observed . in principle , stellar rotation will also play a role in the observed properties of spin modulation pulsations . for example , assuming the oscillation frequency of 581 hz represents the spin frequency of the neutron star in 4u 1636 - 54 , then for a 10 km radius neutron star the spin velocity is @xmath21 at the rotational equator . the motion of the hotspot produces a doppler shift of magnitude @xmath22 , thus the observed spectrum is a function of pulse phase ( see chen & shaham 1989 ) . measurement of a pulse phase dependent doppler shift in the x - ray spectrum would provide additional evidence supporting the spin modulation model and would also provide a means of constraining the neutron star radius . the rotationally induced velocity also produces an aberration which results in asymmetric pulses , thus the pulse shapes also contain information on the spin velocity and therefore the stellar radius ( chen & shaham 1989 ) . the component of the spin velocity along the line of site is proportional to @xmath23 , where @xmath24 is the lattitude of the hotspot measured with respect to the rotational equator . the modulation amplitude also depends on the lattitude of the hotspot , as spots near the rotational poles produce smaller amplitudes than those at the equator . thus we expect a correlation between the observed oscillation amplitude and the size of any pulse phase dependent doppler shift . dectection of such a correlation in a sample of bursts would definitively confirm the rotational modulation model in our opinion . we will present calculations of mass - radius constraints for 4u 1636 - 54 and 4u 1728 - 34 including the effects of light bending , rotation and angle dependent emission , based on the observed properties of burst oscillations as well as spectroscopy of eddington limited bursts in a subsequent paper . several models for the kilohertz qpo seen in 13 lmxb systems ( see van der klis 1997 for a recent review ) invoke some sort of beat - frequency interpretation for the twin khz peaks seen in many of the sources ( see miller , lamb & psaltis 1997 ; strohmayer _ et al . so far only in 4u 1728 - 34 does the frequency difference between the twin khz peaks match the frequency observed during x - ray bursts ( see strohmayer _ in two other sources ( ks 1731 - 26 and 4u 1636 - 54 ) the separation of the twin khz peaks is closer to 1/2 the frequency of oscillations observed during bursts ( see wijnands & van der klis 1997 ; and zhang _ et al . for example , wijnands _ ( 1997 ) report a frequency difference for the twin khz qpo in 4u1636 - 54 of @xmath25 hz . the effort to reconcile these observations with a beat - frequency interpretation has led to speculation that the oscillation frequency observed during bursts may sometimes be twice the spin frequency of the star , although in 4u1636 - 54 the difference frequency appears to be a bit less than 1/2 the burst oscillation frequency , and in some z sources the frequency difference is not constant ( van der klis _ if this scenario is correct it implies the existence of two antipodal spots on the neutron star surface during x - ray bursts . in addition to the daunting requirement of initiating the thermonuclear flash nearly simultaneously on opposite sides of the neutron star , the observation of large oscillation amplitudes shortly after burst onset in 4u1636 - 54 places severe constraints on the two hotspot scenario . to see this one can ask the following question . what maximum amplitude can be produced by a star with antipodal hotspots ? for two antipodal spots light bending strongly constrains the amplitudes that can be achieved ( see pechenick , ftaclas & cohen 1983 ) . calculations using the schwarzchild spacetime and isotropic emission from the stellar surface indicate that even a neutron star with an implausibly small compactness of @xmath26 , recall that rotational modulation amplitude increases with decreasing compactness , can only achieve a maximum amplitude of about 30 % , whereas we measured an amplitude of 75% from the burst described here . we note that with @xmath26 a 1.4 @xmath27 neutron star would have a radius of 21 km . this is far stiffer than any neutron star equation of state that we are familiar with . rapid rotation could in principle modify this result , but at the modest inferred spin period of 290.5 hz for the two spots to produce the observed 581 hz frequency one would still require implausibly large neutron star radii to make a significant rotational correction to the amplitude . another process which can increase the amplitude is beaming of radiation at the stellar surface . for example , to the extent that electron scattering is the dominant opacity process in neutron star atmospheres then one should expect a specific intensity distribution which approximates that from a grey atmosphere . such a distribution is proportional to @xmath28 , where @xmath29 is the angle from the normal to the stellar surface ( see mihalas 1978 ) , so this will modestly increase the rotational modulation amplitude . in addition , it is likely that the emergent spectrum will also be a weak function of @xmath29 ( miller & lamb 1997 ) . with more detailed modelling of the emission from hotspots of finite size it will be possible to place strong constraints on the antipodal hotspot interpretation for the burst oscillations . finally , if further analysis such as pulse phase spectroscopy continues to support the interpretation of the burst oscillation frequency as the neutron star spin frequency in 4u1636 - 54 , then khz qpo frequency separations near 1/2 the spin frequency would suggest that the beat frequency interpretation may be untenable . we thank lars bildsten , andrew cumming , keith jahoda , and john wang for helpful discussions . we thank the anonymous referee for helping us improve the manuscript . t. s. acknowledges the high energy astrophysics program under usra in the laboratory for high energy astrophysics ( lhea ) at nasa s goddard space flight center . bildsten , l. 1997 , to appear in proc . nato asi `` the many faces of neutron stars '' , lipari , italy , 1996 bildsten , l. 1995 , apj , 438 , 852 brainerd , j. j. , & lamb , f. k. 1987 , apj , 317 , l33 chen , k. & shaham , j. 1989 , apj , 339 , 279 fryxell , b. a. , & woosley , s. e. 1982 , apj , 261 , 332 hasinger , g. & van der klis , m. 1989 , a & a , 225 , 79 joss , p. c. 1978 , apj , 225 , l123 kylafis , n. & phinney , e. s. 1989 , in timing neutron stars , eds . h. ogelman and e. p. j. van den heuvel , ( dordrecht : kluwer ) , p. 731 lawrence , a. _ et al . _ 1983 , apj , 271 , 793 mihalis , d. 1978 , in stellar atmospheres , ( w. h. freeman ) miller , m. c. & lamb , f. k. 1997 , apj submitted . miller , m. c. , lamb , f. k. , & psaltis , d. 1997 , apj , to appear nozakura , t. , ikeuchi , s. & fujimoto , m. y. 1984 , apj , 286 , 221 pechenick , k. r. , ftaclas , c. , & cohen , j. m. 1983 , apj , 275 , 846 smith , d. , morgan , e. h. & bradt , h. v. 1997 , apj , in press . strohmayer , t. e. , zhang , w. & swank , j. h. 1997 , apj , 487 , l77 strohmayer , t. e. 1992 , apj , 388 , 138 strohmayer , t. e. , zhang , w. , swank , j. h. , smale , a. p. , titarchuk , l. , day , c. & lee , u. 1996 , apj , 469 , l9 strohmayer , t. e. , jahoda , k. , giles , a. b. & lee , u. 1997 , apj , 486 , 355 swank , j. h. _ et al . _ 1997 , in preparation van der klis , m. , wijnands , r. , chen , w. & horne , k. 1997 apj , 481 , l97 _ et al . _ 1997 , van der klis , m. 1997 , to appear in proc . nato asi `` the many faces of neutron stars '' , lipari , italy , 1996 . wijnands , r. _ et al . _ 1997 , 479 , l141 wijnands , r. a. d. , & van der klis , m. 1997 , apj , 482 , l65 zhang , w. , lapidus , i. , swank , j. h. , white , n. e. & titarchuk , l. 1996 , iauc 6541 figure 1 : 2 - 90 kev lightcurve and hardness ratio plot for the burst . the hardness ratio is defined as the ratio of counts in the 6 - 18 kev range to that in the 2 - 6 kev range . the drop in hardness ratio at about 1.5 s indicates the beginning of the radius expansion phase . the dashed vertical lines denote the time interval used in the epoch folding analysis . figure 2 : representation of the dynamic power spectrum for the burst shown in figure 1 . the contours are levels of constant power spectral amplitude . the power spectra were calculated from 2 s intervals with the center of each succesive interval shifted by 1/8 s. the countrate profile of burst 1 is also shown . the burst profile is aligned to the center of each data interval used to calculate the power spectra . figure 3 : lightcurve of the burst focusing on the rise . the interval denoted by dashed lines is that used for the folding analysis shown in figure 4a , while the solid vertical lines denote the 62.5 ms interval used for the folding analysis displayed in figure 4b . figure 4 : ( a ) folded , background subtracted lightcurve using the 0.25 second interval denoted with dashed vertcal lines in figure 3 . the best fitting sinusoidal model is also shown . the average amplitude during this interval was @xmath30 . ( b ) the same results are shown for the 62.5 ms interval shown in figure 3 . for this interval the amplitude was @xmath31 .	we present a study of 581 hz oscillations observed during a thermonuclear x - ray burst from the low mass x - ray binary ( lmxb ) 4u 1636 - 54 with the rossi x - ray timing explorer ( rxte ) . this is the first x - ray burst to exhibit both millisecond oscillations during the rising phase as well as photospheric radius expansion . we measure an oscillation amplitude within 0.1 s of the onset of this burst of @xmath0 , that is , almost the entire thermal burst flux is modulated near onset . the spectral evolution during the rising phase of this burst suggests that the x - ray emitting area on the neutron star was increasing , similar to the behavior of bursts from 4u 1728 - 34 with 363 hz oscillations reported recently . we argue that the combination of large pulsed amplitudes near burst onset and the spectral evidence for localized emission during the rise strongly supports rotational modulation as the mechanism for the oscillations . we discuss how theoretical interpretation of spin modulation amplitudes , pulse profiles and pulse phase spectroscopy can provide constraints on the masses and radii of neutron stars . we also discuss the implications of these findings for the beat frequency models of khz x - ray variability in lmxb .
star cluster pairs are common objects in the magellanic clouds and it is important to understand their formation and evolution processes . a list of 30 cluster pairs in the smc was first presented by hatzidimitriou & bhatia ( 1990 ) . bica & schmitt ( 1995 , hereafter bs95 ) revised previous data on smc extended objects ( star clusters , associations and emission nebulae ) and identified new ones using sky survey eso / serc r and j films . they presented a list of 40 pairs and 2 triple star clusters . pietrzyski & udalski ( 1999a ) reported 23 pairs and 4 triplets derived from pietrzyski et al.s ( 1998 ) star cluster catalogue in the ogle survey area . bica & dutra ( 2000 ) updated bs95 s catalogue considering the entries in pietrzyski et al s catalogue . bica & dutra ( 2000 ) indicated 75 pairs and multiplets comprising 176 individual objects . in recent years there has been growing evidence of interacting star clusters in the magellanic clouds , especially in the lmc . bhatia & hatzidimitriou ( 1988 ) concluded that more than 50% of lmc pairs must be physical systems . bhatia & mcgillivray ( 1988 ) found evidence that ngc2214 is a merging binary star cluster , based on the presence of a flattened core and an extended halo . indeed sagar et al . ( 1991 ) detected two turnoff points revealing the presence of two interacting clusters . bica et al . ( 1992 ) studied cluster pairs and multiplets in the lmc bar by means of integrated colours and found systems which resulted coeval and some with age differences . vallenari et al ( 1998 ) confirmed such scenarios by means of colour - magnitude diagrams . they also presented isophotal contours indicating physical interaction . several other studies have found binarity evidence in lmc cluster pairs ( e.g. kontizas et al . 1993 , dieball & grebel 2000a , 2000b ) . bhatia et al . ( 1991 ) presented a photographic atlas of binary star cluster candidates in the lmc . for the smc no morphological atlas is available and isophotal maps are required to test possible physical interactions . indeed comparisons of isophotes with isopleth maps of n - body simulations proved to be a useful tool ( rodrigues et al . 1994 , de oliveira et al . 1998 , hereafter odb98 ) , since the simulations produce features such as bridges , common envelopes and extensions . the observational importance of tidal tails as interaction signatures was also indicated by leon et al . ( 1999 ) . in this work we provide isophotal maps for smc pairs and multiplets to study the following properties of these candidate physical systems : ( i ) angular distribution ; ( ii ) projected centre - to - centre separation of members ; ( iii ) isophotal structures using the digitized sky survey@xmath6 ( hereafter dss ) ; ( iv ) ages derived by means of isochrone fitting , when possible . we discuss candidate physical systems and infer a scenario for their formation and evolution . in sect . 2 we gather the objects providing coordinates , sizes , centre - to - centre separations and other details for the smc pairs and multiplets . in sect . 3 we provide the isophotal contour atlas together with classifications of interaction features whenever present . a preliminary version of the present isophotal smc atlas together with one for the lmc was given in de oliveira ( 1996 ) . in sect . 4 we derive cluster ages by means of colour - magnitude diagrams extracted from the ogle - ii bvi photometric survey ( udalski et al . 1998 ) . in sect . 5 we discuss the properties of the systems and the possible scenarios for their origin and evolution . finally , concluding remarks are given in sect . table [ tab : catalog ] shows data for the 176 smc objects which form 75 star cluster pairs and multiplets ( bica & dutra 2000 ) , considering that the pair bs63/b67 in the latter study is possibly a triplet with ngc294 ( odb98 ) . by columns : ( 1 ) object cross - identification in the different catalogues : n ( henize 1956 ) , k ( kron 1956 ) , l ( lindsay 1958 ) , h ( hodge 1960 ) , sl ( shapley & lindsay 1963 ) , b ( brck 1976 ) , dem ( davies et al . 1976 ) , eso ( lauberts 1982 ) , h86- ( hodge 1986 ) , ma ( meyssonier & azzopardi 1993 ) , bs ( bica & schmitt 1995 ) , ogle ( pietrzyski et al . 1998 ) . note that some are embedded objects , named after the corresponding hii region ; ( 2 ) and ( 3 ) right ascension and declination for the epoch j2000 , respectively ; ( 4 ) object type following bs95 : c for star cluster , nc for small hii regions with embedded star clusters , cn for clusters which show some traces of emission , a for associations , ca and/or ac for star clusters of low density and objects with intermediate properties , an for associations which show some traces of emission and na for hii regions with embedded associations ; ( 5 ) and ( 6 ) major and minor sizes , respectively ; ( 7 ) position angle of major axis ( @xmath7=n , @xmath8=e ) ; ( 8) centre - to - centre angular separation ( 1 arcsec = 0.28 pc assuming an absolute distance modulus ( m - m ) = 18.9 for the smc , westerlund 1990 ) . for triplets and multiplets we measured the separation between the two main members , as indicated in the corresponding object line in the table ; ( 9 ) remarks : @xmath9 and @xmath10 indicate groups with six , five and four members , triplets and pairs , respectively . a running number identifies each candidate system . abbreviations ` br * ' indicates that a bright star is present ; ` att ' means attached to . a hierarchical indication is given for objects embedded in or superimposed on larger ones : ` in ' suggests a possible physical connection while ` sup ' suggests a projection . there are 56 pairs , 15 triplets , 2 quadruplets , 1 quintuplet and 1 sextuplet . the quintuplet is located in the star forming region ngc395 with dimensions 25 pc x 18 pc which in turn is embedded in the hii complex smc - dem126 with 66 pc x 35 pc according to angular sizes in bs95 . the sextuplet is mostly contained in the ob association h - a35 ( hodge 1985 ) with dimensions 57 pc x 47 pc . one quadruplet is located in h - a60 with 44 pc x 24 pc . note that dimensions ( columns 5 and 6 ) and position angle ( column 7 ) are from bica & dutra ( 2000 ) and were measured on plates . the dimensions are in general larger than the isophotal sizes in the subsequent analyses . l r r r r r r r l name & ra(2000 ) & dec(2000 ) & t & dmax & dmin & pa & separ . & comments & h : m : s & @xmath11 : : & & & & @xmath11 & & ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) bs3 & 0:30:01 & -73:20:01 & ac & 33 & 33 & - & 31 & mt-1 h86 - 2 & 0:30:04 & -73:20:35 & ac & 39 & 36 & 80 & 31 & mt-1 bs4 & 0:30:07 & -73:20:58 & c & 33 & 27 & 70 & & mt-1 bs7 & 0:32:43 & -73:37:59 & c & 33 & 33 & - & 59 & mp-1 bs8 & 0:32:56 & -73:38:33 & ac & 36 & 36 & - & & mp-1 h86 - 22 & 0:34:53 & -73:02:08 & c & 18 & 18 & - & 42 & mp-2 bs9 & 0:35:02 & -73:02:14 & ac & 45 & 39 & 100 & & mp-2 hw9 & 0:36:25 & -73:00:05 & c & 45 & 45 & - & 52 & mp-3 hw10 & 0:36:31 & -72:59:13 & c & 57 & 57 & - & & mp-3 b9 & 0:37:13 & -72:57:53 & c & 27 & 27 & - & 68 & mp-4 in h - a1 h86 - 41 & 0:37:29 & -72:57:48 & c & 33 & 24 & 50 & & mp-4 h86 - 38 & 0:37:24 & -73:01:50 & a & 72 & 72 & - & 46 & mp-5 bs10 & 0:37:34 & -73:01:30 & c & 39 & 39 & - & & mp-5 bs14,smc_ogle165 & 0:39:12 & -73:14:46 & c & 33 & 33 & - & 44 & mp-6 smc_ogle5 & 0:39:22 & -73:15:28 & ca & 51 & 45 & 90 & & mp-6 hw12a & 0:39:26 & -73:22:59 & c & 33 & 27 & 10 & 36 & mp-7 h86 - 54 & 0:39:35 & -73:22:58 & c & 33 & 33 & - & & mp-7 b15 & 0:39:42 & -72:58:39 & ca & 36 & 27 & 100 & 30 & mp-8 h86 - 57 & 0:39:47 & -72:58:56 & ca & 30 & 30 & - & & mp-8 bs13 & 0:40:08 & -72:45:30 & ca & 54 & 45 & 40 & 48 & mp-9 bs248 & 0:40:15 & -72:46:02 & ac & 39 & 30 & 70 & & mp-9 ngc220,k18,l22,eso29sc3 & 0:40:31 & -73:24:10 & c & 72 & 72 & - & 88 & mt-2,in h - a3 & smc_ogle8 ngc222,k19,l24,eso29sc4 & 0:40:44 & -73:23:00 & c & 72 & 72 & - & 88 & mt-2,in h - a3 & smc_ogle9 b23,smc_ogle170 & 0:40:55 & -73:24:07 & c & 36 & 30 & 40 & & mt-2,in h - a3 b19 & 0:40:44 & -73:03:42 & c & 27 & 21 & 170 & 32 & mt-3 b20 & 0:40:49 & -73:04:09 & c & 18 & 18 & - & 32 & mt-3 h86 - 62,smc_ogle10 & 0:40:48 & -73:05:17 & c & 36 & 36 & - & & mt-3 ngc231,k20,l25,eso29sc5 & 0:41:06 & -73:21:07 & c & 108 & 108 & - & 36 & mp-10,in h - a3 & smc_ogle11 bs15 & 0:41:21 & -73:20:31 & a & 138 & 138 & - & & mp-10,in h - a3 b21,smc_ogle171 & 0:41:15 & -72:49:55 & c & 21 & 21 & - & 34 & mp-11 b22 & 0:41:21 & -72:49:32 & c & 21 & 21 & - & & mp-11 b29 & 0:42:11 & -73:43:51 & ca & 30 & 21 & 100 & 46 & mp-12 hw16,smc_ogle13 & 0:42:22 & -73:44:03 & cn & 36 & 36 & - & & mp-12,in smc - dem7 ngc241,k22w , l29w , & 0:43:33 & -73:26:25 & c & 57 & 57 & - & 23 & mp-13 & eso29sc6w , smc_ogle17 ngc242,k22e , l29e , & 0:43:38 & -73:26:37 & c & 45 & 45 & - & & mp-13 & eso29sc6e , bh1,smc_ogle18 b31,smc_ogle19,smc_ogle175 & 0:43:38 & -72:57:31 & c & 30 & 24 & 150 & & mt-4 bs20,smc_ogle20 & 0:43:38 & -72:58:48 & c & 27 & 27 & - & 30 & mt-4 h86 - 70,smc_ogle21 & 0:43:44 & -72:58:36 & c & 39 & 27 & 50 & 30 & mt-4 bs27,smc_ogle177 & 0:44:55 & -73:10:27 & c & 24 & 21 & 80 & 19 & mp-14,in h86 - 72 smc - n10,l61 - 60,smc - dem11 , & 0:44:56 & -73:10:11 & nc & 21 & 21 & - & & mp-14,in h86 - 72 & ma85 ngc248n , smc - n13b , l61 - 67n & 0:45:24 & -73:22:34 & na & 42 & 36 & 110 & 21 & mp-15 & smc - dem16n , eso29en8n , ma101,smc_ogle26n ngc248s , smc - n13a , l61 - 67s & 0:45:26 & -73:23:04 & nc & 42 & 33 & 150 & & mp-15 & smc - dem16s , eso29en8s , ma103,smc_ogle26s b39,smc_ogle27 & 0:45:26 & -73:28:53 & c & 33 & 33 & - & 15 & mp-16 bs30 & 0:45:30 & -73:29:06 & c & 24 & 24 & - & & mp-16 b36 & 0:45:44 & -72:50:35 & c & 42 & 36 & 140 & 10 & mp-17 smc_ogle31 & 0:45:51 & -72:50:25 & ca & 33 & 27 & 90 & & mp-17 , not b36 b38 & 0:45:54 & -72:36:08 & c & 18 & 18 & - & 38 & mp-18 h86 - 79 & 0:45:58 & -72:35:36 & c & 27 & 27 & - & & mp-18 h86 - 76,smc_ogle182 & 0:46:02 & -73:23:44 & c & 27 & 27 & - & & mt-5,in smc - dem21 h86 - 78n , smc_ogle33n & 0:46:12 & -73:23:27 & cn & 27 & 27 & - & 16 & mt-5,in smc - n16 h86 - 78s , smc_ogle33s & 0:46:12 & -73:23:39 & cn & 27 & 24 & 60 & 16 & mt-5,in smc - n16 l31,smc_ogle36 & 0:46:35 & -72:44:32 & c & 66 & 51 & - & & mt-6 h86 - 83,smc_ogle35 & 0:46:34 & -72:46:26 & c & 42 & 42 & - & 30 & mt-6 h86 - 84,smc_ogle185 & 0:46:34 & -72:45:56 & c & 24 & 24 & - & 30 & mt-6 h86 - 86,smc_ogle40 & 0:47:01 & -73:23:35 & c & 48 & 39 & 110 & 71 & mp-19,in h - a9 h86 - 87,smc_ogle187 & 0:47:06 & -73:22:17 & c & 48 & 42 & 90 & & mp-19,in h - a9 h86 - 95 & 0:47:37 & -73:00:51 & ca & 21 & 12 & 100 & 30 & mp-20 h86 - 96 & 0:47:44 & -73:00:46 & c & 18 & 18 & - & & mp-20 bs35,smc_ogle42 & 0:47:50 & -73:28:42 & c & 42 & 42 & - & 55 & mp-21 k25,l35,smc_ogle45 & 0:48:01 & -73:29:10 & c & 72 & 72 & - & & mp-21 ma205 & 0:48:07 & -73:14:49 & nc & 15 & 15 & - & 37 & mt-7 smc - n25,l61 - 106,smc - dem38 , & 0:48:09 & -73:14:19 & na & 51 & 51 & - & & mt-7 & ma208,smc_ogle189 smc - n26,l61 - 107,ma206 & 0:48:08 & -73:14:53 & nc & 27 & 27 & - & 37 & mt-7 h86 - 99,smc_ogle190 & 0:48:13 & -72:47:35 & ca & 39 & 39 & - & 27 & mp-22 h86 - 100,smc_ogle191 & 0:48:20 & -72:47:42 & ca & 45 & 45 & - & & mp-22 b50 & 0:49:02 & -73:21:44 & c & 33 & 33 & - & 66 & mt-8 bs41,smc_ogle194 & 0:49:06 & -73:21:10 & c & 33 & 33 & - & & mt-8 l39,smc_ogle54 & 0:49:18 & -73:22:20 & c & 42 & 33 & 170 & 66 & mt-8,in bs43 smc - n33,l61 - 138,ma297 & 0:49:29 & -73:26:33 & nc & 18 & 15 & 80 & 12 & mp-23,in smc - dem44 ma301 & 0:49:30 & -73:26:23 & nc & 21 & 18 & 80 & & mp-23,in smc - dem44 smc_ogle56 & 0:49:36 & -72:50:13 & ca & 48 & 36 & 100 & 70 & mt-9,in smc - dem46e h86 - 110 & 0:49:44 & -72:51:14 & ca & 45 & 33 & 160 & 70 & mt-9 h86 - 109,smc_ogle58 & 0:49:45 & -72:51:58 & c & 27 & 27 & - & & mt-9 ma317 & 0:49:42 & -73:10:37 & nc & 18 & 15 & 20 & 20 & mp-24 smc - n34,l61 - 142,smc - dem50 , & 0:49:46 & -73:10:25 & nc & 39 & 27 & 120 & & mp-24 & ma322 b53,smc_ogle197 & 0:50:04 & -73:23:04 & c & 57 & 57 & - & 80 & mp-25 b55,smc_ogle60 & 0:50:22 & -73:23:16 & c & 42 & 36 & 110 & & mp-25 smc_ogle199 & 0:50:15 & -73:03:15 & ca & 15 & 15 & - & 35 & mp-26 , in smc - dem51 bs45,smc_ogle59 & 0:50:16 & -73:02:00 & ca & 60 & 54 & 70 & & mp-26,in smc - dem51 h86 - 106w & 0:50:31 & -73:20:11 & c & 30 & 24 & 90 & 23 & mp-27,in smc - dem52 h86 - 106e & 0:50:37 & -73:20:11 & c & 33 & 27 & 80 & & mp-27,in smc - dem52 h86 - 115,smc_ogle63 & 0:50:37 & -73:03:28 & ac & 96 & 72 & 40 & 77 & mp-28,in smc - dem51 smc_ogle65 & 0:50:55 & -73:03:27 & c & 39 & 39 & - & & mp-28 bs46,smc_ogle200 & 0:50:39 & -72:58:44 & c & 30 & 27 & 60 & 47 & mp-29,in l40 h86 - 116,smc_ogle64 & 0:50:40 & -72:57:55 & c & 30 & 30 & - & & mp-29,in l40 bs48,smc_ogle201 & 0:50:42 & -73:23:49 & ac & 51 & 33 & 80 & 46 & mp-30,in smc - dem53 h86 - 108,ma401 & 0:50:53 & -73:24:22 & na & 60 & 60 & - & & mp-30,in smc - dem53 b59,l61 - 183,ma488 , & 0:51:44 & -72:50:25 & cn & 48 & 36 & 80 & 26 & mp-31 & smc_ogle73 [ tab : catalog ] l r r r r r r r l name & ra(2000 ) & dec(2000 ) & t & dmax & dmin & pa & separ . & comments & h : m : s & @xmath11 : : & & & & @xmath11 & & ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) smc - n46,l61 - 184,smc - dem62 , & 0:51:47 & -72:50:47 & nc & 39 & 39 & - & & mp-31 & ma498 h86 - 120 & 0:51:46 & -73:28:01 & c & 21 & 21 & - & 43 & mt-10,in smc - dem70s bs53 & 0:51:49 & -73:28:38 & a & 39 & 33 & 150 & & mt-10,in smc - dem70s h86 - 122 & 0:51:58 & -73:27:41 & c & 27 & 27 & - & 43 & mt-10,in smc - dem70s bs56,smc_ogle77 & 0:52:13 & -73:00:12 & c & 42 & 33 & 90 & 56 & mp-32 h86 - 130,smc_ogle78 & 0:52:17 & -73:01:04 & c & 45 & 36 & 0 & & mp-32 b64,smc_ogle210 & 0:52:30 & -73:02:59 & c & 42 & 42 & - & 56 & mp-33,in h - a29 bs57,smc_ogle211 & 0:52:32 & -73:02:10 & c & 39 & 27 & 60 & & mp-33,in h - a29 h86 - 134w , smc_ogle212 & 0:52:45 & -72:59:24 & c & 30 & 30 & - & 19 & mt-11,in h - a30 b65,smc_ogle83 & 0:52:44 & -72:58:48 & c & 45 & 45 & - & & mt-11 h86 - 134e , smc_ogle213 & 0:52:48 & -72:59:22 & c & 30 & 27 & 0 & 19 & mt-11,in h - a30 bs61 & 0:52:43 & -73:01:45 & ca & 36 & 27 & 80 & 42 & mp-34,in h - a29 bs255 & 0:52:52 & -73:01:45 & c & 24 & 18 & 70 & & mp-34 bs63,smc_ogle84 & 0:52:47 & -73:24:25 & c & 30 & 24 & 150 & 19 & mt-12,in smc - dem73 b67,smc_ogle87 & 0:52:49 & -73:24:43 & c & 39 & 30 & 110 & & mt-12,in smc - dem73 ngc294,l47,eso29sc22 , & 0:53:06 & -73:22:49 & c & 102 & 102 & - & & mt-12 & smc_ogle90 h86 - 140,smc_ogle214 & 0:53:09 & -72:49:58 & c & 27 & 24 & 50 & 15&mp-35 h86 - 139 & 0:53:11 & -72:50:05 & c & 21 & 18 & 40 & & mp-35 bs67 & 0:53:32 & -73:21:03 & ac & 39 & 39 & - & 51 & mp-36 bs68,smc_ogle95 & 0:53:42 & -73:21:32 & ca & 54 & 45 & 130 & & mp-36 b72 & 0:53:26 & -72:40:57 & c & 72 & 72 & - & & m6,in h - a35 h86 - 143,smc_ogle93 & 0:53:31 & -72:40:04 & c & 48 & 48 & - & & m6,in h - a35 bs257 & 0:53:36 & -72:38:30 & ac & 48 & 39 & 0 & & m6 smc - n52a , l61 - 243 , & 0:53:40 & -72:39:35 & nc & 30 & 30 & - & 20&m6,in h - a35 & smc - dem77sw , ma696,smc_ogle94 smc - n52b , l61 - 244,b73 , & 0:53:42 & -72:39:15 & nc & 30 & 30 & - & 20 & m6,in h - a35 & smc - dem77ne , ma699,smc_ogle96 h86 - 148 & 0:53:55 & -72:40:08 & c & 30 & 30 & - & & m6,in h - a35 bs69,smc_ogle217 & 0:53:56 & -72:51:24 & ca & 36 & 24 & 45 & 75 & mp-37 bs72,smc_ogle97 & 0:54:11 & -72:51:54 & ca & 45 & 36 & 20 & & mp-37 b78 & 0:54:45 & -72:07:46 & c & 66 & 48 & 110 & 56&mp-38,in h - a37 l51,eso51sc7 & 0:54:54 & -72:06:46 & c & 60 & 45 & 170 & & mp-38,in h - a37 h86 - 159,smc_ogle102 & 0:55:12 & -72:41:00 & c & 30 & 24 & 130 & 45 & mp-39,in bs260 h86 - 160 & 0:55:21 & -72:40:10 & c & 24 & 24 & - & & mp-39,in bs260 bs81,smc_ogle223 & 0:56:26 & -72:29:45 & c & 36 & 33 & 0 & 43 & mp-40,in h - a40 h86 - 172,smc_ogle108 & 0:56:34 & -72:30:08 & c & 33 & 33 & - & & mp-40,in h - a40 h86 - 175,smc_ogle227 & 0:57:50 & -72:26:24 & c & 24 & 24 & - & 38 & mp-41 h86 - 179,smc_ogle112 & 0:57:57 & -72:26:42 & c & 24 & 24 & - & & mp-41 h86 - 177,smc_ogle226 & 0:57:50 & -72:30:29 & c & 45 & 45 & - & 46 & mp-42,in b - ob13 h86 - 176 & 0:57:53 & -72:29:48 & c & 36 & 30 & 90 & & mp-42 smc - n62,smc - dem93 & 0:57:56 & -72:39:26 & na & 72 & 72 & - & 75 & mt-13,in h - a42 smc - n63,l61 - 331,smc - dem94 , & 0:58:16 & -72:38:47 & na & 36 & 36 & - & & mt-13,in h - a42 & ma1065,smc_ogle113 smc - n64a , l61 - 335,smc - dem95 & 0:58:26 & -72:39:57 & nc & 48 & 39 & 70 & 75 & mt-13,in smc - n64 & h86 - 182,ma1071,smc_ogle114 bs269 & 0:58:19 & -72:13:10 & ca & 24 & 18 & 60 & 35 & mp-43 bs270 & 0:58:23 & -72:12:43 & ca & 33 & 30 & 130 & & mp-43 bs271 & 0:58:37 & -72:13:27 & nc & 39 & 30 & 30 & 35 & mp-44,in smc - dem98 bs272,smc_ogle229 & 0:58:38 & -72:14:04 & nc & 39 & 39 & - & & mp-44,in smc - dem98 bs93 & 0:59:36 & -71:44:13 & c & 24 & 21 & 10 & 17 & mt-14,in smc - dem105 b97 & 0:59:37 & -71:44:40 & c & 30 & 30 & - & & mt-14,in smc - dem105 bs273 & 0:59:40 & -71:44:34 & ac & 30 & 27 & 120 & 17 & mt-14,in smc - dem105 ic1611,k40,l61,eso29sc27 , & 0:59:48 & -72:20:02 & c & 90 & 90 & - & & m4 - 1 & smc_ogle118 h86 - 186,smc_ogle119 & 0:59:57 & -72:22:24 & c & 36 & 36 & - & 29 & m4 - 1,att smc - dem114 ic1612,k41,l62,eso29sc28 , & 1:00:01 & -72:22:08 & c & 72 & 48 & 20 & 29 & m4 - 1,att smc - dem114 & smc_ogle120 k42,l63,smc_ogle124 & 1:00:34 & -72:21:56 & c & 51 & 51 & - & & m4 - 1,att smc - dem114 b98sw & 1:00:21 & -73:52:51 & c & 36 & 27 & 40 & 33 & mp-45 b98ne & 1:00:28 & -73:52:32 & c & 33 & 27 & 130 & & mp-45 h86 - 189,smc_ogle123 & 1:00:33 & -72:14:23 & c & 24 & 24 & - & 68 & mp-46 h86 - 190,smc_ogle230 & 1:00:33 & -72:15:31 & c & 24 & 24 & - & & mp-46 h86 - 191,smc_ogle231 & 1:00:58 & -72:32:25 & c & 48 & 48 & - & 81 & mp-47,in ? smc - dem114 h86 - 194,smc_ogle232 & 1:01:14 & -72:33:03 & c & 51 & 51 & - & & mp-47,in ? smc - dem114 bs102 & 1:01:14 & -73:47:45 & c & 30 & 21 & 100 & 42 & mp-48 hw44 & 1:01:22 & -73:47:15 & c & 45 & 45 & - & & mp-48 b110 & 1:02:11 & -72:00:11 & c & 33 & 27 & 70 & 65 & mp-49 , brin , in h - a49 b112 & 1:02:23 & -72:00:11 & c & 51 & 45 & 60 & & mp-49 , brin , in h - a49 k45w , l69w & 1:02:45 & -73:44:19 & c & 33 & 27 & 40 & 47 & mp-50 k45e , l69w & 1:02:49 & -73:44:25 & c & 24 & 24 & - & & mp-50 ngc376,k49,l72,eso29sc29 , & 1:03:53 & -72:49:34 & c & 108 & 108 & - & 75 & mp-51 & smc_ogle139 bs114,smc_ogle235 & 1:03:59 & -72:48:18 & ac & 42 & 30 & 110 & & mp-51 smc_ogle138 & 1:03:53 & -72:06:11 & ca & 36 & 36 & - & 82 & mp-52,in h - a53 smc_ogle144,smc_ogle236 & 1:04:05 & -72:07:15 & ca & 36 & 36 & - & & mp-52 bs122 & 1:04:18 & -73:10:21 & c & 21 & 21 & - & 26 & mp-53 b119 & 1:04:19 & -73:09:54 & c & 36 & 36 & - & & mp-53 smc - n78a , l61 - 438,ma1512 & 1:05:04 & -71:59:01 & nc & 24 & 21 & 140 & 23 & m5,in ngc395 smc - n78b , l61 - 439 , & 1:05:04 & -71:59:25 & nc & 24 & 18 & 100 & & m5,in ngc395 & ma1508/1514,smc_ogle145 ma1520,smc_ogle147 & 1:05:08 & -71:59:45 & nc & 30 & 27 & 130 & 23 & m5,in ngc395 smc - n78d , smc - dem127 & 1:05:11 & -71:58:28 & na & 54 & 54 & - & & m5,att smc - n78 smc_ogle146 & 1:05:13 & -71 59 42 & na & 27 & 27 & - & & m5,in ngc395 bs130 & 1:05:56 & -72:04:11 & a & 57 & 36 & 140 & 30&mp-54,att smc - n78 bs132 & 1:06:01 & -72:03:37 & ca & 42 & 33 & 150 & & mp-54,att smc - n78 bs133 & 1:06:23 & -71:55:12 & a & 39 & 30 & 30 & & mt-15,in bs134 bs135 & 1:06:40 & -71:55:13 & ca & 39 & 39 & - & 47 & mt-15,in bs134 bs136 & 1:06:48 & -71:54:55 & ca & 42 & 42 & - & 47 & mt-15,in smc - dem132 b134 & 1:09:01 & -73:12:24 & ca & 48 & 33 & 80 & & m4 - 2,in h - a60 bs142 & 1:09:07 & -73:12:01 & c & 24 & 21 & 70 & 31 & m4 - 2,in h - a60 ic1644,smc - n81,l61 - 481 , & 1:09:12 & -73:11:42 & nc & 48 & 39 & 40 & 31 & m4 - 2,in h - a60 & eso29en35,ma1688/1687 b135 & 1:09:19 & -73:11:15 & c & 33 & 24 & 60 & & m4 - 2,in h - a60 ngc422,k62,l87,eso51sc22 & 1:09:25 & -71:46:00 & c & 60 & 60 & - & 64 & mp-55 ic1641,hw62,eso51sc21 & 1:09:39 & -71:46:07 & c & 45 & 39 & 40 & & mp-55 hw71nw & 1:15:30 & -72:22:36 & c & 18 & 18 & - & 17 & mp-56 hw71se & 1:15:33 & -72:22:50 & c & 39 & 33 & 170 & & mp-56 for isophotal analysis purposes we extracted digitized images of pairs and multiplets from the dss . the plates are from the serc southern sky survey and include iiia - j long ( 3600s ) , v band medium ( 1200s ) and v band short ( 300s ) exposures . the pds pixel values correspond to photographic density measures from the original plates , and are not calibrated . the images were processed with the iraf package at the instituto de fsica - ufrgs , applying a 2-d gaussian filter to smooth out individual stars , producing isodensity maps . we show the isophotal atlas of smc star cluster pairs and multiplets in figs . [ fig : atlas1 ] throughout [ fig : atlas7 ] , where morphological evidence of interactions can be searched for . in each panel we point members and provide designations . cluster pairs and multiplets with evidence of physical interaction are marked in column 6 of table [ tab : interact ] . their isophotal maps show features such as isophotal distortions , common envelope , isophotal twistings , etc . we classified these isophotal maps according to the following criteria : ( i ) isophotes of the members are detached ( * d * ) , but showing isophotal distortions ; ( ii ) isophotes of the members are connected by a `` bridge '' ( * b * ) ; ( iii ) the members are embedded in the same isophotal envelope ( * e * ) . these classification criteria are an important tool for a selection of interacting cluster pair candidates , since such isophotal morphologies are predominant in n - body model encounters ( rodrigues et al . 1994 , odb98 ) . the classifications above were made for objects not disturbed by gas emission . it is worth noting that both bridges and envelopes imply a common outer isophotal , however the former includes a dimension which is considerably smaller than the smaller cluster diameter . the pair b59/smc - n46 would be an envelope limiting case ( fig . [ fig : atlas4 ] ) . as examples , the pair h86 - 186/ic1612 ( fig . [ fig : atlas5 ] ) is embedded in the same isophotal envelope and it is classified as * e. the pair h86 - 140/h86 - 139 ( fig . [ fig : atlas4 ] ) shows a bridge linking its members and is classificated as b. in fig . [ fig : atlas6 ] the pair ngc376/bs114 shows isophotal distortions , however the outer isophotes of the components are not connected and it is thus classified as d*. we detected relevant isophotal features for about 25% of the sample distributed as follows : 6 candidate cluster systems with common envelope , 7 with bridge and 5 detached cases . high isophotal densities such as those observed for b78/l51 ( fig . [ fig : atlas5 ] ) are related to the high surface brightness which often occurs in blue clusters . it is important to note that non relevant isophotal features can appear in the maps caused by bright stars or background fluctuations . by means of previous inspections in the original images we verified the non relevant objects . for example , the compact object in the lower right corner of the b39/bs30 isophotal map ( fig . [ fig : atlas2 ] ) is a bright star , however the compact one to the right of the main cluster can be a star or a compact cluster , as seen in the original form . higher resolution images would be necessary to check the presence of a third cluster . l r r r r r l name & age & e(b - v ) & age@xmath12 & e(b - v)@xmath12 & morphologies & commments & myr & & myr & & & ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) bs3 & & & & & b & mt-1 outside ogleii h86 - 2 & & & & & b & mt-1 outside ogleii bs4 & & & & & & mt-1 outside ogleii h86 - 38 & 251@xmath1325 & 0.10&&&&mp-5 bs10 & @xmath14560 & 0.10&&&&mp-5 b15 & @xmath141000 & 0.07&&&&mp-8 h86 - 57 & @xmath141000 & 0.07&&&&mp-8 ngc220,smc_ogle8 & 65@xmath1313 & 0.07&100@xmath1323&0.07&d & mt-2 ngc222,smc_ogle9 & 70@xmath137 & 0.07&100@xmath1323&0.07&d & mt-2 b23,smc_ogle170 & @xmath15100 & 0.07&&&d & mt-2 ngc231,smc_ogle11 & 65@xmath138 & 0.12&79@xmath1318&0.08&&mp-10 bs15 & 70@xmath137 & 0.10&&&&mp-10 b29 & 150@xmath1370 & 0.10 & & & & mp-12 hw16,smc_ogle13 & @xmath1520 & 0.07&20@xmath1315&0.05&&mp-12 ngc241,smc_ogle17 & 55@xmath135 & 0.12&79@xmath1318&0.10&b&mp-13 ngc242,smc_ogle18 & 65@xmath1310 & 0.10&79@xmath1318&0.10&b&mp-13 b31,smc_ogle19,smc_ogle175 & 280@xmath1330 & 0.10&400@xmath1392&0.08&&mt-4 bs20,smc_ogle20 & 450@xmath1350 & 0.08&400@xmath1392&0.08&&mt-4 h86 - 70,smc_ogle21 & 450@xmath1350 & 0.07&&&&mt-4 bs27,smc_ogle177 & @xmath1525 & 0.18&79@xmath1337&0.08&&mp-14 smc - n10 & hii region & & & & & mp-14 ngc248n , smc_ogle26n & hii region & & & & & mp-15 ngc248s , smc_ogle26s & hii region & & & & & mp-15 b39,smc_ogle27 & 450@xmath1350 & 0.07&&&b&mp-16 bs30 & 450@xmath1350 & 0.07&&&b&mp-16 b36 & 315@xmath1350 & 0.03&&&&mp-17 smc_ogle31 & 450@xmath1350 & 0.03&&&&mp-17 h86 - 76,smc_ogle182 & 200@xmath1320 & 0.14&&&&mt-5 h86 - 78n , smc_ogle33n & @xmath1525 & 0.14&16@xmath139&0.15&&mt-5 h86 - 78s , smc_ogle33s & @xmath1530 & 0.14&16@xmath139&0.15&&mt-5 l31,smc_ogle36 & 250@xmath1350 & 0.10&&&&mt-6 h86 - 83,smc_ogle35 & 180@xmath1350 & 0.10&&&&mt-6 h86 - 84,smc_ogle185 & 250@xmath1350 & 0.10&&&&mt-6 h86 - 86,smc_ogle40 & 350@xmath1350 & 0.03&&&&mp-19 h86 - 87,smc_ogle187 & 140@xmath1320 & 0.07&158@xmath1336&0.04&&mp-19 bs35,smc_ogle42 & 400@xmath13100 & 0.03 & & & & mp-21 k25,smc_ogle45 & 250@xmath1350 & 0.07&250@xmath1358&0.07&&mp-21 ma205 & hii region & & & & b&mt-7 smc - n25,smc_ogle189 & hii region & & & & b&mt-7 smc - n26 & hii region & & & & & mt-7 h86 - 99,smc_ogle190 & 225@xmath1350 & 0.14&&&d&mp-22 h86 - 100,smc_ogle191 & 200@xmath1350 & 0.14 & & & d&mp-22 b50 & @xmath1530 & 0.10&&&&mt-8 bs41,smc_ogle194 & 70@xmath1330 & 0.10&79@xmath1318 & 0.07 & & mt-8 l39,smc_ogle54 & 80@xmath1320 & 0.10 & 100@xmath1323 & 0.10 & & mt-8 smc - n33 & hii region & & & & e&mp-23 ma301 & hii region & & & & e&mp-23 smc_ogle56 & 115@xmath1315 & 0.18&&&&mt-9 h86 - 110 & @xmath1520 & 0.18&&&&mt-9 h86 - 109,smc_ogle58 & 180@xmath1320 & 0.18&200@xmath1345&0.17&&mt-9 ma317 & hii region & & & & b&mp-24 smc - n34 & hii region & & & & b&mp-24 b53,smc_ogle197 & 200@xmath1350 & 0.07&250@xmath1355&0.08&&mp-25 b55,smc_ogle60 & 160@xmath1330 & 0.07&250@xmath1355&0.08&&mp-25 smc_ogle199 & & & & & & mp-26 bs45,smc_ogle59 & 65@xmath1330 & 0.10 & 63@xmath1314 & 0.10 & & mp-26 h86 - 106w & 200@xmath1350 & 0.10&&&e&mp-27 h86 - 106e & @xmath1530/200@xmath1350@xmath16 & 0.12/0.10&&&e&mp-27 h86 - 115,smc_ogle63 & & & & & & mp-28 smc_ogle65 & 200@xmath1320 & 0.10&&&&mp-28 bs46,smc_ogle200 & 80@xmath1310 & 0.07&100@xmath1323&0.06&b&mp-29 h86 - 116,smc_ogle64 & 125@xmath1315 & 0.10&126@xmath1329&0.10&b&mp-29 bs48,smc_ogle201 & 160@xmath1320 & 0.10&&&&mp-30 h86 - 108,ma401 & @xmath1530 & 0.12&&&&mp-30 b59,smc_ogle73 & 100@xmath1315 & 0.14&158@xmath1375&0.11&e&mp-31 smc - n46 & hii region & & & & e&mp-31 [ tab : interact ] l r r r r r l name & age & e(b - v ) & age@xmath12 & e(b - v)@xmath12 & morphologies & comments & myr & & myr & & & ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) bs56,smc_ogle77 & 140@xmath1320 & 0.07&79@xmath1338&0.08&d & mp-32 h86 - 130,smc_ogle78 & 65@xmath138 & 0.12&79@xmath1318&0.08&d & mp-32 b64,smc_ogle210 & @xmath1510 & 0.12&158@xmath1375&0.07&&mp-33 bs57,smc_ogle211 & @xmath1510 & 0.12&&&&mp-33 h86 - 134w , smc_ogle212 & & & & & & mt-11 b65,smc_ogle83 & 65@xmath1310 & 0.10&63@xmath1330&0.09&&mt-11 h86 - 134e , smc_ogle213 & & & & & & mt-11 bs61 & 250@xmath1350 & 0.03&&&&mp-34 bs255 & & & & & & mp-34 bs63,smc_ogle84 & 450@xmath1350 & 0.10&&&d&mt-12 b67,smc_ogle87 & 450@xmath1350 & 0.10&500@xmath13115&0.10&d&mt-12 ngc294,smc_ogle90 & 300@xmath1350 & 0.10&316@xmath1373&0.11&d & mt-12 h86 - 140,smc_ogle214 & 55@xmath1325 & 0.10&&&b&mp-35 h86 - 139 & @xmath1530 & 0.10&&&b&mp-35 bs67 & @xmath14300 & 0.07&&&&mp-36 bs68,smc_ogle95 & 500@xmath13100 & 0.07&&&&mp-36 b72 & 80@xmath1310 & 0.10 & & & & m6 h86 - 143,smc_ogle93 & 200@xmath1325 & 0.10&&&&m6 bs257 & 50@xmath1330@xmath16 & 0.10&&&&m6 smc - n52a , smc_ogle94 & hii region & & & & & m6 smc - n52b , smc_ogle96 & hii region & & & & & m6 h86 - 148 & 400@xmath13100 & 0.10&&&&m6 h86 - 159,smc_ogle102 & 500@xmath1350 & 0.10&&&&mp-39 h86 - 160 & & & & & & mp-39 bs81,smc_ogle223 & 80@xmath1320/250@xmath1350@xmath16 & 0.10&&&&mp-40 h86 - 172,smc_ogle108 & 280@xmath1330 & 0.07&&&&mp-40 h86 - 175,smc_ogle227 & 30@xmath1310 & 0.10&&&&mp-41 h86 - 179,smc_ogle112 & @xmath1530 & 0.10 & 32@xmath1323 & 0.11 & & mp-41 h86 - 177,smc_ogle226 & @xmath1530 & 0.10&&&d&mp-42 h86 - 176 & @xmath1530 & 0.10&&&d&mp-42 smc - n62 & hii region & & & & & mt-13 smc - n63,smc_ogle113 & hii region & & & & & mt-13 smc - n64a , smc_ogle114 & hii region & & & & & mt-13 bs269 & & & & & & mp-43 bs270 & @xmath1530 & 0.10&&&&mp-43 bs271 & @xmath1530 & 0.12&&&&mp-44 bs272,smc_ogle229 & @xmath1530 & 0.08&79@xmath1318&0.05&&mp-44 ic1611,smc_ogle118 & 100@xmath1320 & 0.07 & 160@xmath1337 & 0.08 & & m4 - 1 h86 - 186,smc_ogle119 & 180@xmath1320 & 0.07 & & & e&m4 - 1 ic1612,smc_ogle120 & 100@xmath1350 & 0.10 & 50@xmath1324 & 0.07 & e&m4 - 1 k42,smc_ogle124 & 20@xmath1310 & 0.10 & 40@xmath139 & 0.06 & & m4 - 1 h86 - 189,smc_ogle123 & 570@xmath1370 & 0.07 & & & & mp-46 h86 - 190,smc_ogle230 & 125@xmath1313 & 0.07 & 32@xmath1323&0.08&&mp-46 h86 - 191,smc_ogle231 & 200@xmath1325 & 0.07&&&&mp-47 h86 - 194,smc_ogle232 & 200@xmath1325 & 0.07&&&&mp-47 ngc376,smc_ogle139 & 20@xmath132 & 0.10&32@xmath137&0.07&e&mp-51 bs114,smc_ogle235 & 250@xmath1325 & 0.04&&&e&mp-51 smc_ogle138 & @xmath1530 & 0.04 & 25@xmath1320 & 0.04 & & mp-52 smc_ogle144,smc_ogle236 & @xmath1530 & 0.07 & 40@xmath1319 & 0.05 & & mp-52 smc - n78a & hii region & & & & & m5 smc - n78b , smc_ogle145 & hii region & & 79@xmath1337 & 0.07 & & m5 ma1520,smc_ogle147 & hii region & & 12@xmath139 & 0.06 & & m5 smc - n78d & hii region & & & & & m5 smc_ogle146 & hii region & & 20@xmath1315 & 0.06 & & m5 bs130 & 80@xmath1310 & 0.10&&&&mp-54 bs132 & @xmath1530/200@xmath1325@xmath17 & 0.10 & & & & mp-54 hw71nw & & & & & e&mp-56 outside ogleii hw71se & & & & & e&mp-56 outside ogleii notes to table [ tab : interact ] : 1- age depends on membership of bright stars ; 2- upper main sequence is possibly underpopulated . udalski et al . ( 1998 ) provided a bvi photometric survey of the smc central @xmath3 2.4 square degrees with results for about 2.2 million stars . the data were collected during the ogle - ii microlensing search project ( udalski et al . 1997 ) . pietrzyski & udalski ( 1999b ) studied colour - magnitude diagrams ( cmds ) extracted from the bvi database for 93 smc star clusters , taking into account neighbouring field extractions for comparisons . they estimated reddening values for each cluster using : ( i ) the mean i band magnitude of red clump stars in the cluster neighbourhood ; ( ii ) the assumed extinction - free magnitude of the smc s red clump stars , i = 18.34 mag ( udalski 1998 ) . they derived ages using bertelli et al.s ( 1994 ) isochrones , adopting as a rule for the smc a metallicity z = 0.004 , and an absolute distance modulus ( m - m ) = 18.65 . considering the ogle - ii angular coverage we conclude that out of the present 176 objects 133 are therein included . we obtained from the database v and i band cmds for these objects , using their coordinates and diameters to define a box extraction . guided by dss images of each system we selected representative field regions to extract stars and construct cmds for comparison purposes . ages and reddenings for the clusters were estimated by fitting the padova isochrones . we adopted bertelli et al.s ( 1994 ) isochrones rather than the new isochrones by girardi et al . ( 2000 ) , since the previous set includes younger ages . the isochrone age grid suitable for 4 , 5 , 6.3 , 7.9 , 10 , 12 , 16 , 20 , 25 , 32 , 40 , 50 , 63 , 79 , 100 , 126 , 160 , 199 , 251 , 316 , 400 , 500 , 630 , 790 and 1000 myr . the isochrone age range and resolution allow one to estimate errors taking into account the stellar statistics in the cluster and field cmds . we also assumed a smc metallicity of z = 0.004 , a foreground galactic reddening of e(b - v)@xmath18 = 0.03 in the smc direction and e(v - i)/e(b - v ) = 1.31 . table [ tab : interact ] shows the ages , in column 2 and reddening values ( foreground plus smc internal ) in column 3 determined in the present work via isochrone fitting . ages and reddening values derived by pietrzyski & udalski ( 1999b ) are in column 4 and 5 respectively , when available for comparisons . in our sample we could estimate cmd ages for 91 objects , 40 of them in common with pietrzyski & udalski ( 1999b ) . we conclude that despite the reddening method and distance modulus differences between the two studies , there is good overall agreement for the age determinations . we also included ages for 21 objects embedded in hii regions , classified as na and nc ( see column 4 of table [ tab : catalog ] , and also bs95 ) . we assumed for them an age of 3 myr . we overlap two age histograms in fig . [ fig : age ] . the dotted line histogram shows objects from table [ tab : interact ] with ages via cmds . there are three peaks , at about 80 , 220 and 450 myr . the first two peaks appear to be present in grebel et al.s ( 2000 ) sample for 200 smc clusters based on ogle - ii data , reported at t @xmath3 100 and 200 myr respectively . they suggested them as enhanced star formation epochs . considering pietrzyski & udalski s ( 1999b ) sample with cmd ages for 93 smc clusters , no peak is seen at t = 80 - 100 myr , having their histogram a maximum at the youngest bin . in their histogram there occurs a local maximum at t @xmath3 250 - 300 myr , but the statistics is low . the third peak in the present study has no counterpart in both previous analyses , and it is probably an artifact from the low isochrone age step resolution in that range . the solid line histogram in fig . [ fig : age ] shows additionally the embedded objects in hii regions for which we assumed an age of 3 myr , and the objects from table [ tab : interact ] with lower or upper limits by assuming them as the ages themselves . we can observe again a three peak distribution but the first peak is now shifted to the youngest bin similarly to pietrzyski & udalski s ( 1999b ) histogram . this suggests that only the 200 myr peak is relevant , being related to the smc / lmc last encounter ( gardiner et al . 1994 , grebel et al.s 2000 ) . the maximum seen in the youngest bin is possibly related to the cluster formation / destruction rates . since the present sample deals with pairs and multiplets a fast destruction rate might be caused by the internal dynamical evolution in each cluster complex , caused by merger and/or other effects . approximately 55% of cluster pairs and multiplets in table [ tab : interact ] present similar ages between their members indicating that they are coeval . this suggests that most of the pairs and multiplets had a common origin , possibly from the same molecular complex . note that about 60% of the pairs and multiplets embedded in ob associations ( h - a , hodge 1985 ) , as indicated in column 9 of table [ tab : catalog ] , have comparable ages between their members . this could be an explanation for the origin of cluster systems . considering triplet and multiplet members , we found that about 70% of them are younger than 100 myr . these results suggest a possible binary cluster formation scenario : clusters can be born in multiplet systems and coalesce by mergers and tidal disruptions forming binary clusters in a timescale of @xmath3 100 myr . this time is in agreement with dynamical times required for an interacting pair to merge into a single cluster ( odb98 , de oliveira et al . 2000 ) . the pairs with a bridge in the isophotal maps have comparable ages for their components ( table [ tab : interact ] ) . as examples , the pairs ngc241/ngc242 and b39/bs30 in fig . [ fig : atlas2 ] show a bridge linking their members which could be interpreted as a sign of interaction ( see the similarity with the n - body simulation model in fig . 11 of odb98 ) . a typical timescale for the bridge phenomenon is @xmath3 30 myr , as deduced from a series of n - body simulations related to bridge formation and evolution ( odb98 ) . another interesting isophotal feature is related to the cluster triplet ngc220/ngc222/b23 which shows distortions for the small cluster in a direction almost perpendicular to the line connecting itself to the large components ngc220 and ngc222 . this configuration and morphologies are compatible with a fast hyperbolic encounter with small impact parameter ( e.g. fig . 12 of odb98 ) . [ fig : distr_smc ] shows the angular distribution of pairs and multiplets together with smc hi column density isophotes from mathewson & ford ( 1984 ) . it can be seen that most of the objects are concentrated in the smc main body , close to the higher concentration of hi , so it is not unexpected that in general they result young ( sect . 4 ) . the objects appear to be distributed along an axis . such distribution is present in the overall smc cluster sample and there is growing evidence that it is related to a nearly edge - on disk containing the bulk of the young stellar population in the smc ( bica _ et al . _ 1999 , westerlund 1990 ) . a nearly edge - on disk in a low internal reddening galaxy like the smc would imply an increase of projection effects as compared to a simulation such as that carried out by bhatia & hatzidimitriou ( 1988 ) for the nearly face - on lmc disk where the physical pairs would be about 50% . consequently the fraction of physical pairs in the smc would be lower . the present approach including morphological evidence of interaction is an attempt to constrain this aspect . indeed the fraction with isophotal distortions is only 25% ( sect . projection effects can be responsible for the age spread in some multiplets . for example the sextuplet ( table [ tab : interact ] ) has component ages 80 myr ( b72 ) , 200 myr ( h86 - 143 ) , 50 myr ( bs257 ) , 3 myr ( smc - n52a and smc - n52b ) and 400 myr ( h86 - 148 ) . possibly only the 3 or 4 younger components could be related to ob - association h - a35 , the remaining objects would be captures or projection effects . this age spread is also present in ic1611 s quadruplet and in some triplets . on the other hand the quintuplet in the star forming complex ngc395 has all its members with the same age ( 3 myr ) thus forming a physical system . in fig . [ fig : diam_smc ] we show the distribution of the diameter ratio between members for all pairs in the sample . the diameter ratio is mostly in the range @xmath19 , with a peak at @xmath2 indicating that the majority of pair members have a comparable size . this effect was also observed in the lmc ( bhatia et al . 1991 ) . [ fig : separ_smc ] shows the distribution of the centre - to - centre angular separation between pair members . the separation range is @xmath3 @xmath20 arcsec ( @xmath33 - 22 pc ) with a pronounced peak at @xmath3 @xmath21 arcsec ( 13 pc ) . a similar peak was also observed by bhatia et al . ( 1991 ) and de oliveira ( 1996 ) who found a bimodal distribution for the lmc pairs with peaks at @xmath3 5 and 13 pc . the observed upper limit of the projected centre - to - centre linear separation @xmath3 23 pc ( 80 arcsec ) is comparable to bhatia & hatzidimitriou s ( 1988 ) separation criterion for pairs in the lmc ( 18.7 pc ) . the frequent separation value around 13 pc may reflect a preferred survival distance for the systems , combined to projection effects . we presented an isophotal atlas for 75 star cluster pairs and multiplets in the small magellanic cloud , comprising 176 objects . it was possible to derive ages from colour - magnitude diagrams using the ogle - ii photometric survey for 91 objects . in addition we included in the analysis ages for 21 embedded objects in hii regions . the age distribution has a maximum in the youngest bin with a profile related to the cluster formation / destruction rates , in particular cluster multiple systems can have a fast destruction rate caused by their internal dynamical evolution . there is a second peak around 220 myr which is probably related to the smc / lmc last encounter . we find that 55% of the pairs and multiplets in the sample are in general coeval , indicating that captures are a rare phenomenon . most of the cluster multiplets occur in ob stellar associations and/or hii region complexes which indicates a common origin and suggests that multiplets coalesce into pairs or single clusters in a short time scale ( @xmath3 100 myr ) . the majority of the cluster members have comparable sizes , with a diameter ratio ranging mostly between @xmath19 . the projected separation distribution between the members of a pair has a pronounced peak at @xmath3 13 pc . these observational results are important constraints to theoretical models of star cluster pair encounters and could be related with the formation process and subsequent dynamical evolution of cluster systems . the angular distribution of cluster pairs and multiplets shows that most of the objects are located in the smc main body . the overall smc cluster sample presents a similar distribution and there is evidence that it is related to a nearly edge - on disk in the smc . considering this , it is expected an increase of projection effects as compared to estimates for the lmc disk where physical pairs would be about 50% ( bhatia & hatzidimitriou 1988 ) . the atlas shows that around 25% of the isophote maps present relevant structures like bridges , common envelopes or detached distorted isophotes . n - body simulations have indicated that these structures arise from interactions between the members of the cluster systems . indeed cluster pairs as ngc241/ngc242 and b39/bs30 show in their isophotal maps bridges linking their members and have comparable ages . we conclude that multiplicity may have an important role in the early dynamical evolution of star clusters in general , and signatures of that may survive in the long term structure of large single clusters ( de oliveira et al . 2000 ) . grebel e.k . , zaritsky d. , harris j. , 2000 . in _ news views of the magellanic clouds . in : proceedings of the iau symp . 190 _ , eds . you - hua chu , nicholas b. suntzeef , james e. hesser , david a.bohlender ( eds ) . , pasp conf . series , p. 405 .	an isophotal atlas of 75 star cluster pairs and multiplets in the small magellanic cloud is presented , comprising 176 objects . they are concentrated in the smc main body . the isophotal contours were made from digitized sky survey@xmath0 images and showed relevant structural features possibly related to interactions in about 25% of the sample . previous n - body simulations indicate that such shapes could be due to tidal tails , bridges or common envelopes . the diameter ratio between the members of a pair is preferentially in the range @xmath1 , with a peak at @xmath2 . the projected separation is in the range @xmath3 @xmath4 pc with a pronounced peak at @xmath3 @xmath5 pc . for 91 objects it was possible to derive ages from colour - magnitude diagrams using the ogle - ii photometric survey . the cluster multiplets in general occur in ob stellar associations and/or hii region complexes . this indicates a common origin and suggests that multiplets coalesce into pairs or single clusters in a short time scale . pairs in the smc appear to be mostly coeval and consequently captures are a rare phenomenon . we find evidence that star cluster pairs and multiplets may have had an important role in the dynamical history of clusters presently seen as large single objects . the images in this study are based on photographic data obtained using the uk schmidt telescope , which was operated by the royal observatory edinburgh , with funding from the uk science and engineering research council , until 1988 june , and thereafter by the anglo - australian observatory . original plate material is copyright by the royal observatory edinburgh and the anglo - australian observatory . the plates were processed into the present compressed digital form with their permission . the digitized sky survey was produced at the space telescope science institute under us government grant nag w-2166 . ]
whiting1 ( @xmath13 , @xmath14 ) is a halo globular cluster at a heliocentric distance d@xmath945kpc , as first shown by @xcite . it was discovered by @xcite during a search for galaxies in the zone of avoidance , and originally it was classified as an open cluster by @xcite . later , @xcite confirmed its association with the milky way halo with deep @xmath15 photometry , and updated the heliocentric distance to [email protected] . they derived an age of 6.5gyr and metallicity @[email protected] from isochrone fitting and estimated a@xmath16=0.11 mag and ( m - m)@xmath17=17.34 . in latter study , measured caii indexes ( @xmath18caii ) of three stars in whiting 1 give metallicity [ fe / h ] ranging from @xmath21.1 to @xmath20.4 ( @xmath1 ranging from 0.002 to 0.008 ) . they pointed out that whiting1 is spatially close to the trailing stream of sgr dsph , and its kinematics is consistent with that of this galaxy . besides , it follows very nicely the sgr dsph age - metallicity relation @xcite . only a few globular clusters in the halo of the galaxy are known to have comparably young ages ( @xmath19 , pal1 rosenberg et al . 1998a ; pal12 rosenberg et al . 1998b ; segue3 ortolani et al . 2013 ; crater belokurov et al . 2014 ) , making whiting1 a member of a rare class of objects . up to now no nir photometry of whiting1 has been reported in the literature and in this study we aim to obtain the fundamental properties through the nir cmd using data from the eso science archive . also , a search for variable star population has never been performed . the next section describes our data . section 3 , 4 and 5 present our analysis and results of the nir data : cluster structural parameters and cluster fundamental parameters - age , metallicity and distance modulus . section 6 describes our search for variable stars in @xmath6 band and our conclusions are drawn in section 7 . whiting1 was observed in the nir on 2006 september 4 at the eso ntt ( new technology telescope ) with sofi ( son of isaac ; * ? ? ? * ) in large - field imaging mode . sofi is equipped with a 1024@xmath201024 detector , with a pixel scale of 0.292arcsecpix@xmath21 and fov of 4.98 x 4.98 . we had 15 images in @xmath22 ( each was the average of 4 frames of 15sec integration ) and 32 images in @xmath23 ( each was the average of 6 frames of 10sec integration ) . the seeing during the observations was 0.8 0.9 on average for both filters . the data reduction was carried out with iraf , and included flat fielding , sky / bias / dark subtraction , image alignment and combination . the sky frame was created by median combination of all images with the task _ imcombine _ , applying the _ avsigclip _ 3-@xmath24 rejection algorithm , and appropriate thresholds to excise the stars . a constant was added to the sky subtracted images during the final combination to set their median value to zero , accounting for the sky variations during the observations . next , we performed psf photometry ( stetson 1987 ) with daophot / iraf . only 99 stars were measured in both @xmath22 and @xmath23 . the 100% completeness limits were defined from the first drop in the luminosity function which give @xmath2519.5mag and @xmath2618.2mag ( magnitude bins varied from 0.1 to 0.3 ) , and the photometric errors for stars brighter than these limits are typically less than 0.1mag . the photometric calibration was based on 9 stars for @xmath22 , and 7 stars for @xmath23 , in common with the two micron all sky survey @xcite . @xmath27 @xmath28 where the lower case letters mark the instrumental magnitudes , and the upper case ones mark the standard magnitudes ( hereafter we will use the designation k for @xmath29 filter ) . optical images of whiting1 were collected with emmi ( eso multi - mode instrument ; dodorico 1990 ) and efosc2 ( eso faint object spectrograph and camera ; snodgrass et al . 2008 ) on the 3.58-m eso ntt ( new technology telescope ) at la silla , chile , typically to fill small gaps in other programs or during periods of poor weather conditions . binning by a factor of two was used on all occasions , making the pixels scales 0.33 and 0.24arcsecpix@xmath21 and the fov 5.6 x 9.9 and 4.1 x 4.1 , respectively for both instruments . five epochs were obtained in @xmath6 band , and the first was paired with @xmath30 band for quasi - simultaneous color information . the total integration for each epoch was split into three images , taken with some offsets to minimize the effects from the detector s cosmetics . a detailed log of the observations is presented in table [ table_obslog ] . .log of the optical monitoring observations . [ cols="^,^,^,^,^,^,^,^ " , ] the differential light curves ( fig.[light_curves ] ) of the candidate variables # 359 and # 834 were generated with respect to the reference stars # 672 and # 851 which are the nearest stable stars similar in magnitude and color . the maximum peak - to - peak variation of both candidates is @xmath9 0.2 mag over the monitoring period of 35 days with available data for 4 epochs . we plotted all 273 stars with three and more epochs in @xmath6 band on cmd @xmath6 vs. @xmath31 ( see fig.[cmd_optic ] ) . here , we used @xmath32 magnitudes from the first epoch . we selected the stars that lie within the determined cluster radius and one can see on the diagram that the cluster members mainly represent the lower part ms population . more massive stars have already evolved from the ms and the to - point is well visible at @xmath33 mag . some rgb stars are also present . for comparison , we plotted isochrones of parsec v1.1 @xcite , which are reddened with the proper extinction ( see fig . [ cmd_optic ] ) to match the observed ms , to - point , rgb and the clump . the used parameters are in consistence with the previously obtained ones . the two candidate variables ( empty circles in the latter figure ) lie in the middle and in the upper part of the ms but their luminosities are not high enough to suspect them neither as @xmath34 scuti nor as even dimmer @xmath35 doradus pulsating variables . according to the best fitting isochrones the masses of both stars are close to 1 solar mass and the detected amplitude variations about 2 tenths of magnitude within a time span of a month may be related to stellar eclipses by a companion star a likely possibility , taking into account their position on the cmd . another option is a rotational modulation by sun - like spots the variation study @xcite among a sample of sun - like stars reveals at least one object - hd 129333 with similar amplitude . however , due to the sparsity of the available data any further clarification of the nature of the variability of these stars is not possible . in an attempt to strengthen our knowledge of whiting1 we used nir photometry and determined the physical parameters of whiting1 . we fitted isochrones to @xmath7 vs. @xmath8 color - magnitude diagram and derived an age [email protected] , metallicity @[email protected] ( [ fe / h][email protected]@xmath00.1 ) and distance modulus @[email protected] . our results confirm that whiting 1 is a young and moderately metal - rich globular cluster . it is one of the youngest from the sgr dsph . in the nir cmd , we noticed a clustering of objects at @xmath36 , @xmath37 with radius [email protected]@xmath10 , presumably a background galaxy cluster . we estimated a redshift and z@xmath91 is based on a comparison with the colors of galev evolutionary synthesis models @xcite for e / sa galaxies . our multi - epoch @xmath6 band monitoring led up to the discovery of two candidate variable stars within the cluster radius . their differential light curves give maximum peak - to - peak variation of @xmath90.2 mag for both candidates over the monitoring period of 35 days and taking in to account their position on the cmd diagram the stars are likely to be eclipsing binaries . authors would like to thank venelin kozhuharov for his valuable help on fitting analysis and the anonymous referee for his / her constructive comments and suggestions that helped us to greatly improve to the paper . the data used in the publication were retrieved from the eso science archive . they were obtained under program ids : 59.a-9002(a ) , 59.a-9908(a ) , and 60.a-9013(a ) . this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .	whiting1 is a member of the fast - growing group of young globular clusters in the milky way halo . preliminary estimates of its fundamental parameters have been provided using optical photometry and low resolution spectroscopy . in an attempt to strengthen our knowledge of whiting1 , in this study we employ a complementary approach . isochrone fitting method was applied on the near - infrared color - magnitude diagram and yields an age [email protected] , metallicity @[email protected] ( [ fe / h][email protected]@xmath00.1 ) and distance modulus @[email protected] . our results confirm that whiting1 is a young and moderately metal - rich globular cluster . it is one of the youngest from the sgr dsph . we fitted an elson , fall and freeman ( eff ) profile to the near - infrared number counts , and measured cluster core radius @xmath4=9.1@[email protected]@xmath5 . two probable eclipsing variables in the cluster were found from multi - epoch @xmath6 band photometry . finally , an unknown galaxy cluster was identified on our @xmath7 vs. @xmath8 color - magnitude diagram . it has a redshift z@xmath91 , and it is located at about 1@xmath10 from the center of whiting1 at @xmath11 , @xmath12 , contaminating the cluster photometry . [ firstpage ] galaxy : globular clusters : general galaxy : globular clusters : individual : sgr dsph galaxy : globular clusters : individual : whiting1 galaxies : clusters : general galaxies : clusters : individual : cl j020256.6 - 031609
in a perfect , starless @xmath0cdm hierarchical universe filled with adiabatic gas and nfw ( navarro , frenk , & white 1996 ) dark halos , the bolometric x - ray bremsstrahlung luminosities of galaxy groups and clusters would scale in a self - similar fashion with gas temperature , @xmath1 ( kaiser 1986 ; 1991 ; evrard & henry 1991 ) . however , in our particular universe this relation is somewhat steeper ( @xmath2 ; e.g. arnaud & evrard 1999 ) and becomes very steep ( at least @xmath3 ) for groups having @xmath4kev ( helsdon & ponman 2000a ) . the observed properties of galaxy groups differ from those of massive clusters in several other respects . groups have a lower baryon fraction @xmath5 ( david 1997 ; renzini 1997 ) and a larger fraction of group baryons are stellar ( e.g. david et al . 1990 ; david & blumenthal 1992 ) . finally , the entropy factor @xmath6 evaluated at @xmath7 for groups exceeds the self - similar expectation , indicating that groups have received an additional entropy @xmath8 kev @xmath9 . if a comparable entropy increment were present in the hot gas in massive clusters , it would be difficult to detect because of the much larger entropy that the gas acquires passing through stronger accretion shocks that surround more massive clusters . this has been referred to as the `` entropy floor '' ( ponman et al . 1999 ; lloyd - davies et al . 2000 ) . these deviations from self - similarity has led to the hypothesis that gas in groups experienced some additional early heating before ( or when ) it flowed into the group halos . early `` pre - heating '' by 0.5 - 1.5 kev / particle could explain the aberrant behavior of groups in both the @xmath10 and @xmath11 plots . furthermore , since this level of heating would only be noticed in the shallower potentials of galaxy groups , most authors have adopted a much stronger assumption : that _ all _ baryonic gas , including gas currently in both groups and clusters , experienced some `` non - gravitational '' heating ( star formation , pop iii stars , agn , etc . ) prior to its entry into the dark halos . as a result of its dissipative nature , the gas entropy observed in galaxy groups today does not in itself indicate a unique heating history . entropy increases in shocks and is lost with radiative cooling . nevertheless , the level of heating required to heat all the gas at the same ( high ) redshift in `` pre - heating '' scenarios generally exceeds that produced by type ii supernovae following normal star formation : @xmath12 kev / particle . in this estimate we assume ten percent of the baryons form into stars ( fukugita , hogan & peebles 1998 ) with a typical salpeter imf , producing type ii supernovae that heat the remaining gas with 100 percent efficiency . nevertheless , many theoretical papers have appeared recently with estimates of the assumed universal pre - heating necessary to produce the `` entropy floor '' in groups and the related departure from @xmath13 self - similarity ( e.g. knight & ponman 1997 ; cavaliere , menci , & tozzi 1997 ; balogh , babul & patton 1999 ; cavaliere , giacconi & menci 2000 ; loewenstein 2000 ; tozzi & norman 2000 ; valageas & silk 1999 ; kravtsov & yepes 2000 ) . for a given energy release , the final entropy is larger if the energy is applied when the gas density is low . therefore , most of these authors have assumed that the gas was heated ( by some unspecified agency ) while in a low density intergalactic environment ( at redshift @xmath14 ) before it flowed into the galaxy group potentials . independent support for strong universal heating has come from estimates of the collective emission from gas - filled group halos at large redshifts which appear to exceed the observed soft x - ray background radiation ( pen 1999 ; wu , fabian & nulsen 1999b or wfn ) . these authors suggest that the intergalactic gas was heated to @xmath15 so that little of it flowed into group halos at early times . however , snii - driven galactic winds are expected to develop immediately in these small halos , perhaps reducing the contribution of groups to the unresolved x - ray background . in addition , warm - hot diffuse intergalactic gas at temperatures @xmath16k can result from ( gravitationally produced ) shocks in large - scale filaments ( cen & ostriker 1999 ; dav et al . 2000 ) . moreover , in the cosmological simulations of dav et al . most of the warm - hot gas that contributes to the unresolved soft x - ray background is at low densities and diffusely distributed , not concentrated in ( group ) halos as assumed by pen and wfn . consequently , dav et al . predict a soft x - ray flux @xmath17 times less than that of pen and wfn , sufficiently low to be consistent with the observed ( unresolved ) background . although the non - gravitational heating of intergalactic gas postulated by pen and wfn may occur , it is not supported by more detailed cosmological simulations . tozzi & norman ( 2000 ) study the consequences of universal intergalactic heating prior to collapse into dark halos in a flat @xmath0cdm cosmology . in their approximate hydrodynamic models with radiative cooling , all cooled baryonic gas is assumed to collect at the very center . the universal energy input ( at @xmath18 ) they require to match the observed `` entropy floor '' is in excess of normal snii expectations . in their preferred models , tozzi & norman assume that the intergalactic gas ( heated to @xmath19 for groups ) enters the group halos adiabatically , i.e. without shocking . as a consequence , the radial entropy profile for the tozzi - norman externally pre - heated galaxy groups is almost constant out to the virial radius @xmath20 , unlike those observed by lloyd - davies , ponman & cannon ( 1999 ) for which @xmath21 , and the gas temperature gradients of these tozzi - norman models are quite negative throughout , e.g. @xmath22 at @xmath23 , unlike the nearly constant temperature profiles observed . both of these key results of adiabatic inflow are in conflict with observed groups , at least within the small region that can be observed , @xmath24 . by contrast , an internal heating scenario is proposed by loewenstein ( 2000 ) who estimates the amount of heating required after the hot gas has reached hydrostatic equilibrium in group and cluster dark halos . from a series of approximate static hot gas models , loewenstein argues that most of the heating occurred during or after the assembly of the group or cluster gas , not throughout the intergalactic gas at an earlier time . we are in agreement with his interpretation , although it may be a minority opinion at the present time . perhaps the most convincing evidence that the origin and evolution of hot gas in galaxy groups can be understood with normal star formation and other standard astrophysical assumptions is the success of our own detailed calculations for the giant elliptical ngc 4472 ( brighenti & mathews 1999a ) , the dominant galaxy in a small virgo subcluster . this is a complete gas dynamical calculation beginning with a tophat perturbation in a flat cosmology . the gas evolves in a growing nfw dark halo , produces stars and snii heating , forms a central galaxy and the stars lose mass and produce type ia supernova . after 13 gyrs we were able to match the radial variation of density , temperature and iron abundance observed in ngc 4472 ( brighenti & mathews 1999b ) . the agreement is excellent everywhere except within @xmath25 kpc from the center where the gas density is too high ; we now think this is due to additional support by magnetic stresses there . our models also agree with the present day entropy variation and luminosity in ngc 4472 . in particular , our models for ngc 4472 produce the `` entropy floor '' in the @xmath26 plot and the observed deviation from the @xmath13 self - similar relation without universal pre - heating . in addition , they go much further in fitting the radial profiles of density , temperature and iron abundance ( buote 2000b ) observed in ngc 4472 . the pre - heating controversy devolves on a choice between _ internal _ supernova heating during galaxy group formation and _ external _ ( universal ) heating by agn or some other agency at an early time . given the apparent success of our calculations for ngc 4472 , it seems possible that the `` entropy floor '' and @xmath13 similarity breaking in groups can be explained with normal star formation . but the widely discussed pre - heating hypothesis can under some conditions also provide the necessary similarity - breaking . our objective in this paper is to perform a series of gas dynamical calculations that explore the consequences of various levels of external pre - heating that span the full range from galaxy groups to rich clusters . several sets of progressively more sophisticated calculations are considered , beginning with purely adiabatic flow . these externally heated cooling flows are compared with flows that only experience internal supernova - based heating . we shall find that the latter mode of heating is generally quite satisfactory in fitting the x - ray data , but the imf may need to be flatter than salpeter . in order to compare our gas dynamical calculations with internal supernova heating with discussions of external pre - heating in the current literature , we consider gas flowing into dark halos of three masses representing galaxy groups , poor clusters and rich clusters . the dark halo evolution is similar in all models . these structures all begin with a tophat perturbation in a @xmath0cdm universe ( @xmath27 , @xmath28 , @xmath29 ) which converts to a dark matter nfw profile ( navarro , frenk & white 1996 ) at very high redshifts . the dark halos are assumed to grow masswise from the inside out as described by bertschinger ( 1985 ) . the dark matter flow pattern consists of an outer , converging smooth collisionless flow which attaches to a stationary nfw core having a concentration appropriate to its current virial mass . the time - dependent intersection radius of the stationary halo with the converging flow is chosen to conserve dark matter mass . baryonic gas with mean density @xmath30 ( burles & tytler 1998a , b ) flows into the evolving dark matter potential . however , in order to explore the cosmic pre - heating hypothesis using the same approach as often employed in the current literature , we may choose to `` reset '' the density and temperature ( and therefore the entropy ) of the baryonic gas to be spatially uniform at a very early time @xmath31 gyr ( redshift @xmath32 ) when the stationary dark halo mass is only 0.1 - 0.2 of its current virial mass . at the reset , the gas velocity is assumed to retain its velocity in the cosmic flow outside the nfw halo at time @xmath33 , but within the halo the gas velocity is set to zero . this artificial reset of the gas parameters at this high redshift often has no long - lasting influence on the gas dynamics , however , if the gas temperature is less than the virial temperature of the dark matter structures at time @xmath33 , shocks may develop resulting in entropy fluctuations that persist until the present time @xmath34 gyrs . in the following discussion we consider a suite of increasingly more realistic physical models for the baryonic component . for dark halos of each mass , the baryonic gas is assumed to be pre - heated at time @xmath31 gyr by varying amounts . for the first set of models , we assume that the gas flow is perfectly adiabatic but with adiabatic ( non - radiating ) shocks , so most of the entropy increase occurs in the accretion shock transition . in the next series of models we include radiative cooling and allow the cooled gas to accumulate at the origin as in the models of tozzi & norman ( 2000 ) . next we repeat some of these models with a cooling recipe similar to that used by nulsen & fabian ( 1997 ) in which the cooled gas accumulates not at the origin , but in spatially extended structures , as if it had formed into a collisionless stellar system . this type of flow is similar to the mass dropout flows discussed by knight & ponman ( 1997 ) ( but who assumed bertschinger ( 1985 ) dark halos in a @xmath35 cosmology with a large baryon fraction and no heating ) . for each series of calculations we consider a range of virial masses and assumed entropies corresponding to different levels of external pre - heating at the reset time . finally , we repeat the same calculations again allowing for star formation with various amounts of energy released in type ii supernova ( snii ) feedback . in this series of models , which resembles our earlier calculation for ngc 4472 , the snii heating at early times is internal , a natural consequence of star formation . these models also include stellar mass loss and heating by type ia supernovae . the x - ray luminosity varies over a factor @xmath36 from galaxy groups to the richest clusters and the gas temperature spans a range of @xmath37 . because of the inhomogeneous observational data available over this vast range of parameters and the various bandpasses that have been used in these observations , we shall consider only those observations for which the bolometric x - ray luminosities are determined . all data is corrected to our assumed @xmath38 km s@xmath39 mpc@xmath39 . for galaxy group luminosities and temperatures we choose the recent observations of helsdon & ponman ( 2000a ) for loose groups . more recently , these authors have shown that there is no distinction in the x - ray properties of loose and compact groups ( helsdon & ponman 2000b ) . helsdon & ponman list rosat pspc bolometric luminosities for 24 groups . unfortunately , these groups are observed out to a radius which is typically only @xmath40 of the virial radius , so helsdon & ponman provide for each group an approximate correction factor to estimate the total bolometric luminosity within the virial radius ; we have employed this correction in their group data plotted with open squares in figure 1 . the @xmath13 variation is quite steep ( and uncertain ) for these groups , @xmath41 . for the entropy in galaxy groups and clusters we use the data of lloyd - davies , ponman & cannon ( 2000 ) who estimate the entropy factor @xmath42 at @xmath7 . for a sample of poor and rich clusters we use data for 24 clusters collected by arnaud & evrard ( 1999 ) from many disparate authors using different x - ray satellites . in selecting this sample arnaud & evrard avoided clusters with strong cooling flows . the arnaud - evrard bolometric luminosities have not been corrected to the virial radius , but this correction may not be large . for example , the mean correction found by helsdon & ponman for the groups is only a factor of @xmath43 . the arnaud - evrard data , shown with crosses in figure 1 , can be fit with @xmath44 , which is also steeper than adiabatic similarity @xmath1 . allen & fabian ( 1998 ) emphasize the important effect that a strong cooling flow has on the @xmath13 relation . about 70 - 90 percent of rich clusters have cooling flow cores . since the gas temperature has a steeper radial gradient ( and is generally lower ) in clusters with strong cooling cores , it is unclear which temperature is appropriate to enter in the @xmath13 diagram . if the asca x - ray spectra of allen & fabian are fit with isothermal models , ( their model a ) the cooling flow clusters lie at lower temperatures ( and higher luminosities ) than non - cooling flow clusters as shown in figure 1 . the non - cooling flow clusters have slope @xmath45 , very similar to the arnaud - evrard slope , and in fact appear to be fully consistent with the overlapping arnaud - evrard data . alternatively , if the x - ray spectra of cooling flow clusters are fit with a combination of single temperature thermal emission and xspec cooling flow models ( in which the emitted spectrum is an integrated sum over many gas temperatures ) as in model c of allen & fabian ( 1998 ) , then the background ( uncooled ) gas temperatures of cooling flow clusters increase and become very similar to those of the non - cooling flow clusters in the allen - fabian sample . this is a clear example of the substantial scatter that can occur in the @xmath13 diagram arising from the data reduction procedure when a wide range of temperatures contribute to the observed spectrum . in a related approach , markevitch ( 1998 ) showed that the @xmath13 correlation tightened considerably when the cooling flow cores are removed from both the luminosity and temperature data . when data for all groups and clusters are compared in figure 1 , it can be seen that the arnaud - evrard data have the same slope and ordinate as the non - cooling flow allen - fabian clusters . this is expected since the arnaud - evrard clusters were also chosen to have weak cooling flow cores . if arnaud & evrard had instead chosen clusters at random , they would have found a much larger number of normal cooling flow clusters scattered above their correlation in figure 1 , similar to the cooling flow clusters in the allen & fabian sample . the identification of the arnaud - evrard clusters with the lower envelope of the correlation in the @xmath13 plot also explains why the helsdon - ponman poor groups with the greatest temperatures may seem slightly overluminous relative to adjacent members of the arnaud - evrard sample having the lowest temperatures . this lack of consistency and apparent continuity between the various data sets demonstrates how systematic shifts in the @xmath13 plot may arise from different methods of sample selection and data reduction for the majority of cooling flows in which the intrinsic temperature necessarily varies over a wide range . since the @xmath13 data for groups are so important in setting limits on the amount of cosmic pre - heating , some possible systematic trends should be recognized . just as with rich clusters , groups with strong cooling flows tend to occupy the upper envelope of the group data in figure 1 . to demonstrate this we use the sample of 12 galaxy groups studied by buote ( 2000a ) in which the data was reduced with an xspec cooling flow model with a determination of the cooling rate @xmath46 . if the buote data is bisected with a line of slope @xmath2 or @xmath47 , the mean cooling flow rates of groups that lie above the line , @xmath48 @xmath49 yr@xmath39 , is much greater than the mean of the six groups below , @xmath50 @xmath49 yr@xmath39 . x - ray emission from the more luminous sample have spuriously lower temperatures due to strong cooling flows associated with the central galaxy . this is an important distinction for our models since we are interested in groups in which a strong cooling flow develops around a dominant , luminous elliptical galaxy located at the group center . for this reason we expect our successful models with concentrated cooling flows to lie near the upper envelope of the helsdon - ponman group data . additional scatter in the @xmath13 data for groups is introduced by the variable physical extents of the x - ray observations relative to the virial radius for each group . finally , the choice of cosmology has some influence on the observational group data . in extrapolating @xmath51 to the virial radius , helsdon & ponman ( 2000a ) used the scdm adiabatic results of navarro , frenk , & white ( 1995 ) to derive @xmath20 from the observed gas temperature . our adiabatic models and @xmath0cdm variables give @xmath52 mpc which is only about 10 percent smaller than that used by helsdon & ponman , probably within the observational and computational uncertainties . we consider three virial masses , @xmath53 , @xmath54 and @xmath55 @xmath49 which span a range from galaxy groups to moderately rich clusters . these dark halos form as a result of a tophat perturbation in a flat @xmath0cdm cosmology , @xmath27 @xmath28 and @xmath56 km s@xmath39 mpc@xmath39 , chosen so that the current time is exactly @xmath34 gyrs . the baryonic component @xmath30 is such that @xmath57 . when a tophat perturbation is imposed , the dark matter flows toward the perturbation as a pressure - free gas and accumulates in an extended , static , spherically symmetric structure . this structure is assumed to have the standard nfw ( navarro , frenk & white 1996 ) halo profile for which the virial mass is @xmath58 . here @xmath59 ( eke , navarro & frenk 1998 ) and @xmath60 gm @xmath61 is the critical density . the shape of the dark halo at any time is @xmath62 where @xmath63 , @xmath64 and @xmath65 is the concentration . cosmological simulations show that the concentration decreases slowly with @xmath66 and redshift ( bullock et al . we ignore the redshift variation , i.e. , @xmath67 . the virial radii for the three dark halos are @xmath68 , 1500 , 2640 kpc or @xmath69 and 3.42 respectively . the complete solution for the dynamical growth of a locally overdense region of collisionless dark matter in a flat einstein - de sitter universe was solved by bertschinger ( 1985 ) . he showed that the converging pressure - free flow accumulates in an essentially stationary core which grows masswise from the inside outward , as if newly arriving matter were added at the outer boundary of the core . since bertschinger made no allowance for the hierarchical , three dimensional nature of mass accumulation , the mass profile in his stationary core differs from the more appropriate nfw shape . we adapt the basic physics of bertschinger s collisionless solution for the evolution of a dark matter overdensity to our flat @xmath0cdm cosmology by computing the pressure - free collapse of the dark matter until it accumulates at the outer boundary of a stationary halo that is required to have the more correct nfw profile ( brighenti & mathews 1999b ) . the halo grows with time ( from the inside out ) at a cosmologically appropriate rate , conserving dark mass . however , at very high redshifts ( @xmath70 ) this simple recipe breaks down for large @xmath66 because the internal mass distribution of the nfw halo can be less than that of the converging @xmath0cdm flow at every radius . when this mismatch occurs at high redshift , we assume that the dark matter collects into an extended , mass - conserving power law halo @xmath71 within 40 kpc , which resembles the inner profile of a nfw halo . this transition in the dark matter halo is particularly necessary for the most massive halo we consider , nfw halos of smaller mass intersect the @xmath0cdm flow at very high redshifts ( @xmath72 ) . we find that the exact shape of the temporary , non - nfw dark halo for @xmath73 has a only a small influence on the integrated properties of the gas at time @xmath34 gyrs . while none of our conclusions are affected by assumed dark halo structure at @xmath73 , it is good to keep in mind that some of our computed results ( e.g. final @xmath51 , @xmath74 , etc . ) are uncertain at the @xmath75 percent level . finally , we note that the baryonic gas does not exactly follow the dark matter in the external @xmath0cdm flow since in general the gas velocity is slowed by radial pressure gradients . the equations for one dimensional eulerian gas dynamics that we use here are described in detail in brighenti & mathews ( 1999a ) . they are solved with an extensively modified 1d spherical version of the eulerian code zeus ( stone & norman 1992 ) . the pressure - free dark matter and the baryonic gas are computed as two separate fluids which interact gravitationally . the size of spatial zones increases logarithmically . we use `` outflow '' boundary conditions at the outermost spherical zone where the hubble flow receeds supersonically out of the grid , and the flow velocity vanishes at the origin . sink terms for radiative cooling and distributed mass dropout are included in the appropriate models . for those models involving gas flow within optical galaxies , the gas dynamic equations equations have additional source terms for stellar mass loss and heating by stars and type ia supernovae . the full set of equations are described in brighenti & mathews ( 1999a ) . the characteristic virial temperature of an nfw halo can be found directly from the condition for hydrostatic equilibrium , @xmath76 since by definition the mean density within the virial radius is always @xmath77 . the x - ray luminosity for pure bremsstrahlung emission , with emission coefficient @xmath78 , is @xmath79 where the characteristic gas density @xmath80 is assumed to scale with the dark halo density ( e.g. kaiser 1986 ; 1991 ) . therefore we expect @xmath1 provided the following assumptions hold : ( 1 ) the gas is in hydrostatic equilibrium bound to homologously identical halos , ( 2 ) the gas density profiles are homologous in the variable @xmath81 for all clusters , ( 3 ) the ratio of gas to dark mass is identical within all @xmath20 , and ( 4 ) x - ray line emission is ignored . although reasonable , these assumptions are not perfectly satisfied . for example , the gas variables within the accretion shock @xmath82 , which typically occurs very close to the virial radius , are not expected to be strictly homologous in @xmath81 and , furthermore , the halo concentration varies somewhat with @xmath66 . non - gravitational heating of the gas in excess of that received in the accretion shock can cause the mass fraction in gas to vary with virial mass . line emission becomes important at @xmath83 k , relevant for galaxy groups . at these low temperatures the radiative cooling coefficient @xmath84 where @xmath85 , which flattens the @xmath13 slope to at least @xmath86 . finally , the nfw halo for a given virial mass and cosmology represents a computational mean configuration about which there is considerable scatter ; for dark halos of the same @xmath66 , bullock et al . ( 2000 ) find a 1@xmath87 scatter of @xmath88 percent in the concentration parameter . we consider four levels of pre - heating for each of the three virial masses . this is done by arbitrarily raising the temperature to @xmath89 everywhere in the @xmath0cdm flow at time @xmath31 gyrs ( redshift @xmath90 ) when the cosmic baryonic density is @xmath91 gm @xmath61 . for comparison with other recent discussions of the pre - heating hypothesis , we also reset the temperature and density to be everywhere constant at time @xmath33 . by this time the nfw dark matter halos have masses about @xmath92 , @xmath93 and @xmath94 @xmath49 for groups , poor clusters and clusters respectively . gas that had already entered the dark halos at time @xmath33 is spread uniformly with the rest of the gas at the reset . the heat supplied to the gas at these early times places the universal gas on an adiabat which can be represented with an entropy factor @xmath95 as defined by tozzi & norman ( 2000 ) which is evaluated at time @xmath33 . the corresponding entropy factor defined by ponman , cannon & navarro ( 1999 ) is @xmath96 the four levels of heating we consider @xmath97 , @xmath98 , @xmath99 and @xmath100 k corresponding to 0.0013 , 0.65 , 1.3 , and 3.9 kev / particle are numerically characterized by 1 - 4 in our model nomenclature and correspond to entropies @xmath101 , 0.127 , 0.255 , and 0.764 respectively . the three virial masses are referred to as `` g '' , `` pc '' and `` c '' for group , poor cluster and cluster respectively . for example , model adg1 refers to an adiabatic model for gas flow in a galaxy group potential in which the temperature is set to @xmath97 k at @xmath33 . in all model calculations , as the gas evolves we never allow its temperature to drop below @xmath102 k , as would be expected in a photoionized intergalactic medium ( igm ) . previous discussions of pre - heating have not emphasized the possibly important role of compton cooling against the cosmic background radiation . after pre - heating , the electron gas is compton cooled but the proton temperature @xmath103 remains unchanged until equipartition is established by coloumb interactions . to explore this we assumed @xmath104 k for both electrons and protons at ( @xmath33 , @xmath105 ) = ( 0.5 gyrs , 9 ) and integrated equations for @xmath106 and @xmath107 in our assumed @xmath0cdm universe , allowing for compton cooling , equipartition and normal expansion cooling @xmath108 . while the electrons rapidly lose half of their thermal energy relative to @xmath109 by ( @xmath110 , @xmath111 ) = ( 0.64 gyrs , 7.7 ) , @xmath112 occurs at a much later time , ( @xmath110 , @xmath111 ) = ( 3.1 gyrs , 2.0 ) . by this latter time the proton temperature @xmath113 k is much less than halo virial temperatures of interest , and has negligible thermal energy as it enters the accretion shock . although our computed accretion flows are not greatly influenced by compton cooling , it should be noted that the epoch of ( universal ) pre - heating can not be much earlier than @xmath114 gyrs or the gas would rapidly cool and may also leave an observable perturbation on the cosmic microwave background . with this potential difficulty in mind , we shall assume @xmath31 gyrs in most of our calculations and consider a single temperature gas . in our `` adiabatic '' ad models the entropy only increases when the gas is shocked . by far most of the entropy increase occurs when the gas passes through the accretion shock as it enters the nfw halo , @xmath115 , and encounters nearly stationary gas that arrived earlier . the post - shock temperature is comparable to the halo virial temperature @xmath116 where @xmath117 is the mean molecular weight and @xmath118 from eke , navarro & frenk ( 1998 ) . the current virial temperature for the three dark halos we consider are @xmath119 , @xmath120 , and @xmath121 k. if the temperature of gas entering the dark halo @xmath122 is comparable to ( or significantly exceeds ) @xmath123 , then the accretion shock weakens and gas flows nearly adiabatically into the halo , preserving its entropy . the temperature of gas entering the accretion shock at the present time ( @xmath34 gyrs ) has cooled by adiabatic expansion to @xmath124 or @xmath102 k , whichever is larger . in any case , a shock must always propagate away from the origin ( where the flow velocity vanishes ) , even if the converging flow there is subsonic . finally , for models with the least amount of igm pre - heating ( ad1 ) , spurious shocks may develop just after time @xmath33 as gas with sub - virial temperatures flows into the dark halo . bolometric x - ray luminosities and emission - weighted mean temperatures for 12 adiabatic models ( 3 @xmath66 and 4 @xmath89 ) at time @xmath34 gyrs are shown with the observational data in figure 1a . the gas has been heated to @xmath89 and reset to spatial uniformity at @xmath31 gyrs . as the virial mass increases from groups to clusters ( filled circles @xmath125 squares @xmath125 triangles ) , the computed results follow the general trend of the observations , even for these simple adiabatic models . for each virial mass the final luminosity is progressively lower as the amount of assumed pre - heating increases , ad1 @xmath125 ad4 . this is expected since with increased pre - heating the entropy of the gas that enters the halo at early times may exceed the entropy that would have been achieved in the accretion shock . as the entropy at the center of the flow increases , the gas density and x - ray emissivity are reduced , corresponding to lower @xmath51 in figure 1a . with increasing virial mass , each cluster of four models ( 1 @xmath125 4 ) becomes more compact in figure 1a since @xmath126 is satisfied for the largest @xmath66 , even with the maximum amount of pre - heating . further details of the ad models are listed in table 1 . although the adiabatic series of models does not lose energy by radiation , to compute the luminosities plotted in figure 1a , it is necessary to assume that they do in fact radiate with emissivity @xmath127 . the cooling coefficient @xmath128 that we use ( taken from sutherland & dopita 1993 ) corresponds to abundance @xmath129 ( solar meteoritic ) for which @xmath130 reaches a minimum at @xmath131 kev because of increasing line emission at lower temperatures . finally , the results shown in figure 1a are sensitive to the time @xmath33 at which the reset is made and are very different if the reset is ignored altogether . for example if @xmath132 gyrs instead of 0.5 gyrs , the temperatures of the final adg models are relatively unchanged , but the luminosities are very much lower than those in figure 1a : @xmath133 for adg1 and -5.17 for adg4 . when the reset occurs at lower redshifts , more gas must be removed from the halos and the subsequent accretion in the @xmath0cdm universe is insufficient to re - establish the present - day x - ray luminosity . therefore explanations for both the amount of universal pre - heating and the epoch at which it is applied are required in external heating scenarios . if the reset is completely ignored , values of @xmath51 are increased enormously . for example , @xmath134 erg s@xmath39 for model adg1nr ( `` nr '' signifies no reset ) , as shown in table 1 . these no reset groups have x - ray luminosities comparable with the richest clusters observed . the reason for this is that dense , low entropy gas flowed into the group potential before time @xmath33 and , if not removed , must remain there in the absence of reset . the very dense central gas @xmath135 @xmath61 that entered the halo at very high redshifts accounts for the enormous @xmath51 of ad groups without reset . in more realistic models this gas would have radiated its energy away and formed into stars . ponman , cannon & navarro ( 1999 ) and lloyd - davies , ponman & cannon ( 2000 ) have emphasized a related non - selfsimilar deviation of galaxy group observations in the entropy - temperature plane . they compare the emission - weighted mean temperature @xmath74 to the entropy factor @xmath136 evaluated at @xmath137 , approximately the outer extent of current observations . the same idealized , self - similar group - cluster models that correspond to @xmath1 would appear as @xmath138 in figure 2a . however , it is seen that the entropy observed in groups generally exceeds the @xmath139 variation when extrapolated from observations of rich clusters . nevertheless , the deviation from similarity is less robust in the @xmath26 plane than in the @xmath13 diagram since @xmath140 refers to a single point in the observed or computed entropy distribution and , for the observations , also requires an estimate of the virial radius . the observational errors are substantial and can be found in the papers cited above . values for @xmath141 from the computed adiabatic models are plotted as open circles in figure 2a . entropies @xmath141 for models ad4 with the highest level of pre - heating , shown with the largest circles , depend weakly on @xmath66 since the entropy acquired from pre - heating dominates that from shocks . as the pre - heating at time @xmath33 is reduced , @xmath141 decreases with @xmath66 , approaching the expected @xmath139 . however , @xmath141 for the least - heated models , adpc1 and adc1 , exceeds the entropy for adpc and adc models that received more pre - heating . the source of this discrepancy can be traced to post - shock irregularities in the entropy profile @xmath142 created by weak shocks produced during the transient adjustment following the reset time @xmath33 . immediately after reset , cosmically inflowing gas outside the virial radius encounters stationary gas within . this gives rise to both outward and inward propagating shocks . the latter shock confronts a third outward moving shock produced as gas of uniform ( post - reset ) density falls toward the origin , reflecting as a shock . the local increase in entropy depends on the local strength of each shock that passes through . these artifacts of the reset assumption imprint long - lived entropy irregularities in the gas . the effect of these transient shocks is much reduced in models with more pre - heating since the sound speed is larger and shocks are greatly reduced in strength . the resulting fluctuations in @xmath142 are evident in figure 3 where we plot entropy profiles for groups and clusters ( adg & adc ) at the present time . for each @xmath142 profile there are several regions of interest : the relatively flat cosmic inflow region at the far right , the sudden vertical entropy increase at the accretion shock , a region of declining post - shock entropy , and an inner region where the entropy may decline further or be approximately constant . note also the decrease in the pre - shock entropy as gas approaches the shock when there is no pre - heating , ad1 . this is due to our lower limit on the inflowing gas temperature , @xmath143 k. since a low - temperature , adiabatically expanding universe would cool below this value , our lower limit on the temperature results in isothermal intergalactic flow in which the entropy factor increases as @xmath144 . for this isothermal pre - shock flow , the entropy decreases with rising gas density as the flow converges toward the shock in the ad1 flows . for a given dark halo , when the accretion shock is strong , the entropy just behind the shock is the same for all levels of pre - heating . the post - shock flow behaves approximately as @xmath21 , indicating that the accretion shock was weaker in the past when the dark halo was less massive . in model adg1 with no pre - heating this decline in the entropy continues to small radii ; the post - shock entropy was very low in the early universe . when the pre - heating is large , for some time after @xmath33 gas enters the dark halos without shocking , conserving @xmath145 ; this constant entropy gas is currently found closer to the center in the adg2 - adg4 solutions in figure 3 . the entire entropy profile for model adg4 is almost constant . however , converging subsonic flow must always produce an outward propagating shock since the flow velocity is assumed to vanish at the origin . the strength of this shock , which increases with virial mass @xmath66 , can produce central regions with @xmath146 , as in the adc1 - acc4 models in figure 3 . since these small inner regions are convectively unstable , the entropy would also become uniform in a more realistic three dimensional calculation . the choice of @xmath141 and emission averaged temperature @xmath74 in the @xmath26 diagram are determined by the realities of observational limitations , but they are probably not the best coordinates to test deviations from similarity . these inner flow regions near @xmath7 are also subject to computational difficulties from ancient shocks as apparent in the ad1 and ad1nr entropy profiles in figure 3 . even if the cosmic gas were perfectly adiabatic between shocks , we expect that the violent events that accompany the earliest mergers would disturb the entropy of the central regions near @xmath137 which in our models passed through the accretion shock at @xmath25 gyr ( redshift @xmath147 ) . the emission - weighted temperatures also heavily favor the same dense inner region sensitive to these early perturbations . aside from the entropy fluctuations apparent in the inner flow @xmath148 in figure 3 , our calculations are quite accurate elsewhere . this is evident in figure 4 where we plot entropies and temperatures that are mass - averaged values within the accretion shock . in these @xmath26 coordinates the unheated ad1 models lie exactly along the similarity variation @xmath139 . small deviations from this relation occur with increased pre - heating and smaller virial masses , just as one would expect . when the reset assumption is not made , as in the `` nr '' solutions listed in table 1 and adg1nr and adc1nr plotted in figure 3 , the entropy in the inner post - shock flow is very much lower . in these flows gas at @xmath102 k that entered the halo at high redshifts , when the cosmic flow was dense , had very low entropy . the high density of this gas is maintained during its evolution . in addition the emission - averaged temperatures of the `` nr '' solutions are much too low , biased by the central regions . this is still true even if all the gas is heated ( without reset ) to @xmath100 k at time @xmath31 gyr , as shown in the adg4nr , adpc4nr and adc4nr results in table 1 . some irregularities in the entropy profile in figure 3 for the adc1nr flow arise from weak shocks produced at very early times @xmath149 when the dark halo mass distribution changed to an nfw profile . clearly , adiabatic models without reset are unacceptable since their x - ray luminosities are excessive and they have not cooled or formed into stars and galaxies . the high luminosities of no reset flows are related to the strong high redshift emission from groups that would violate the soft x - ray background , as discussed by pen ( 1999 ) and wu , fabian & nulsen ( 1999b ) . but adiabatic solutions with the reset condition are also problematic because the time @xmath33 of universal heating must be carefully orchestrated for agreement with observed x - ray luminosities of galaxy groups . nevertheless , ad flows provide a useful reference for more detailed externally heated flows . moreover , the entropy profiles @xmath142 for the ad solutions in figure 3 share many of the same features with more realistic externally pre - heated models discussed below . this series of gas dynamic models , described with the prefix co , is similar to the adiabatic models except we now include radiative emission according to the sutherland - dopita cooling coefficient @xmath150 . when the gas cools at the center of the flow , we allow it to accumulate there , producing a point mass gravitational potential similar to a giant black hole . at time @xmath31 gyr we reset the gas temperature and density to be spatially uniform and apply the pre - heating . these assumptions are essentially identical to those made by tozzi & norman ( 2000 ) . the entropy decrease due to radiative losses has little effect on the temperature profile since for hydrostatic support the gas temperature must always be close to @xmath123 , i.e. , @xmath151 and @xmath152 is approximately proportional to @xmath153 . however , close to the central concentration of cooled gas , @xmath152 is nearly constant and the temperature increases toward the origin as @xmath154 . the locally higher temperatures in this dense central gas can influence the global emission - weighted temperature of the group or cluster . in some models with radiative cooling hot gas containing an appreciable amount of specific enthalpy and kinetic energy flows into the central numerical zone . since this gas must ultimately radiate this energy producing x - rays , the central region contributes an additional x - ray luminosity given by @xmath155 where the subscript 1 refers to quantities evaluated at the radius @xmath156 parsecs of the innermost numerical grid . this estimated correction to the total luminosity can be quite large , often exceeding @xmath51 from the rest of the cooling flow . since x - ray observations of cores of cooling flows do not reveal strong , high - temperature thermal point sources , the assumption that the gas cools only at the center of the flow is unrealistic . @xmath51 and the emission - weighted @xmath74 for the co ( radiative cooling ) models are illustrated in figure 1b . the result for cluster mass halos ( coc = filled triangles ) are similar to the adiabatic models with reset , as a result of the inefficiency of radiative cooling in these hot , low - density halos . in gas flows with the largest pre - heating cog3 , cog4 , copc3 , copc4 and coc4 none of the baryonic gas cools in @xmath157 , as shown in table 2 . these strongly pre - heated models fit the observations in figure 1b quite well even for low luminosity groups . however , the remaining cooling flow models with less pre - heating are currently accumulating uncooled gas in the origin . as shown in table 2 , for these flows the concentrated x - ray emission from the central zone @xmath158 is 3 - 13 times that from the rest of the flow within the virial radius . in addition , the colossal masses in the central baryonic singularity , @xmath159 @xmath49 , are similar to those of massive galaxies . clearly , none of these weakly pre - heated models are physically acceptable . another distinctive feature of the co series of calculations is the appearance of galactic drips at early times , @xmath25 gyr for clusters and poor clusters , and slightly later @xmath160 gyrs for groups . galactic drips are dense , narrow cooling waves that begin in the outer halo and proceed inward , crossing the cooling flow in @xmath161 gyr . the drips discussed by mathews ( 1997 ) actually lose mass en route by local cooling . but drips in the co models continue to the origin before cooling , so they are more massive and move faster than the drips discussed by mathews ( 1997 ) . the cooling evolution of spherically symmetric drips depends somewhat on the numerical resolution . spherical drip waves are an artifact of spherical symmetry in which the inherent three - dimensional character of rayleigh - taylor instabilities is suppressed . although drip waves have a physically plausible origin and may well occur in cooling flows , the wave fronts are unlikely to be globally spherical and their multi - dimensional evolution is currently uncertain . drips occur only in co models with little or no pre - heating ( e.g. cog1 , coc2 , copg1 and coc1 ) and only one or at most two drip waves occur in each of these models . as drips arrive at the origin , the mass of cooled gas there ( @xmath162 in table 2 ) suddenly increases . after a drip wave has passed through gas at the center of the flow , its entropy increases because the drip waves become supersonic in @xmath163 . the post - drip gas density is also lower since the drip wave transfers mass to the center . these factors contribute somewhat to the low @xmath51 for the cog1 model in figure 1b . both density and temperature gradients in the inner flow ( @xmath163 ) are considerably flattened following the passage of drip waves . since drips typically begin at @xmath164 , and move inward ; they have little influence on the outer @xmath165 percent of the cooling flows . entropy factors @xmath166 ( evaluated at @xmath167 ) for the co models are plotted against emission - averaged temperatures for groups and clusters in figure 2b . as the amount of pre - heating is reduced , the results approach the expected @xmath139 variation , but the entropy in these weakly pre - heated flows is increased by the passage of drip waves and shocks produced by transient flows following reset . because of these complications , the entropy @xmath141 computed at small radii is uncertain and the agreement with observations in figure 2b is poor . the globally mass - averaged entropy and temperature satisfy the similarity condition @xmath139 very well ( figure 4b ) , but can not be easily compared with currently available observations . in figure 3 we show the computed entropy profiles @xmath142 at @xmath34 gyrs for group and cluster co models . in general these entropy profiles are in good agreement with the approximate profiles computed by tozzi & norman ( 2000 ) , but models that experienced drips ( e.g. cog1 ) have higher central entropies and gas temperatures in @xmath163 due to the influence of drip waves . for the most strongly pre - heated models the entropy is high everywhere . in less strongly pre - heated flows the entropy is seen to decrease slightly as the gas approaches the accretion shock ; this occurs because the pre - shock flow is isothermal at @xmath168 k , as explained earlier . the accretion shock ( at all times ) occurs very close to the virial radius : @xmath169 and @xmath170 for groups and clusters respectively . the entropy just behind the strong accretion shock increases with virial mass . as with the ad solutions , the positive entropy gradient in the immediate post - shock flow , @xmath171 with @xmath172 , is due to the increasing strength of the accretion shock with cosmic time ( tozzi & norman 2000 ) . in the inner flow region , @xmath148 the entropy variation @xmath142 depends on several competing processes . the central entropy is increased by drip - induced shocks ( cog1 ) or ( in cog3 and cog4 ) by outward - propagating shocks generated by the initial nearly adiabatic flow of uniform post - reset gas into the origin . in this latter case @xmath146 so the ( small mass of ) gas very close to the center of the flow would be convectively unstable . radiation losses from the dense central regions lower the central entropy of co models below the corresponding ad model , although the effect of this can be reversed by drips ( cog1 , coc1 , and coc2 ) . we have also computed representative co flows with no reset , conr models , in which the gas acquired by the dark halo before @xmath33 is not removed . in these exploratory calculations the temperature of co4nr models is increased by @xmath100 k at @xmath33 but co1nr models receive no additional heating . the unsatisfactory results of these models are listed in table 2 . all of these `` nr '' flows suffer from early intense radiative ( over)cooling , producing baryon mass singularities about ten times larger than those of the ad models . as a result , the x - ray luminosity is currently totally dominated by emission from the core , @xmath173 . the unrealistic nature of these conr models without reset strongly argues for star formation and supernova heating that drives gas out of small dark halos , locally raising its entropy . in summary , our solutions for the co models are in good agreement with those of tozzi & norman ( 2000 ) , particularly in the post - shock entropy profiles shown in figure 3 . but our gas dynamical calculations have illustrated some additional features : the presence of drip waves , the concentrated x - ray emission from the baryonic singularity , a density enhancement as the gas approaches the accretion shock ( tozzi , scharf & norman 2000 ) and shock heating at the origin . our co results depend critically on the assumption used by many authors ( e.g. kaiser 1991 ; cavaliere et al . 1997 ; balogh , babul & patton 1999 ; wu , fabian , & nulsen 1999a ; tozzi & norman 2000 ) that the baryonic density and temperature are nearly uniform at the moment of pre - heating @xmath33 before gas flows into the dark halo potentials . in fact , however , baryonic gas is already concentrated in the dark halos at time @xmath33 and the most natural means of removing it is with supernova - driven winds . the do calculations are identical to the co models except the cooling flows are assumed to be inhomogeneous , containing entropy ( or magnetic ) irregularities that allow the gas to cool ( `` dropout '' ) at large distances from the origin . these do models are an improvement over the co models in which baryonic mass singularities form at the origin having masses that are much greater than the masses of black holes observed in luminous elliptical galaxies . the co models also produce extremely bright central x - ray sources that are not observed . although the details of mass dropout are poorly understood , spatially distributed radiative cooling is required in elliptical galaxy cooling flows . for example , the masses of central black holes in elliptical galaxies are about 10 times smaller than the total amount of diffuse cooling flow gas that has cooled over a hubble time . in addition , the x - ray isophotes of cooling flow gas in rotating galaxies does not exhibit rotational flattening ; this can be understood if angular momentum is being lost by distributed radiative cooling ( brighenti & mathews 2000a , b ) the local cooling dropout rate in an inhomogeneous cooling flow depends on the amplitude distribution of the entropy or magnetic inhomogeneities , which can not be directly observed or derived _ ab initio_. for a simple heuristic representation of mass dropout it is customary to introduce a sink term in the equation of continuity of the form @xmath174 where @xmath175 is the time for gas to cool locally by radiative losses at constant pressure . with this term the elliptical galaxy x - ray surface brightness distributions more nearly resemble those observed and the dynamical mass to light ratios are not greatly disturbed ( e.g. sarazin & ashe 1989 ; brighenti & mathews 2000b ) . successful models require that the dimensionless dropout parameter @xmath176 is close to unity . we assume @xmath177 in the do models computed here and it applies only to hot gas @xmath178 k. the influence of galactic drips is lessened in the do models , but not entirely eliminated ; their amplitude and velocity are reduced as gas cools in the wave and is locally deposited , similar to the drips described by mathews ( 1997 ) . since we view the do models as a variant of the co models , we continue to reset the gas temperature and density at time @xmath33 . flow irregularities introduced by the reset tend to be lessened by mass dropout in the do models . finally , we note that when @xmath177 , about half of the bolometric x - ray luminosity is produced by the cooling regions ( brighenti & mathews 1998 ) . an important beneficial effect of the distributed dropout models is that they do not produce the unrealistic central baryonic mass concentrations that occur in the co models . instead , the cooled baryons are assumed to remain at approximately the same radius at which the cooling dropout occurs . this would be expected if the cooled gas forms into stars that spend most of their time orbiting near the radius where they formed . as gas cools according to the @xmath177 sink term over many gyrs , an extended region of stars forms from the cooling dropout ( nulsen & fabian 1995 , 1997 ) . the density structure of the dropout stellar population has a remarkable resemblance to a de vaucouleurs profile , particularly at early times , @xmath179 gyrs . for the do models discussed here the dropout term is applied to the baryonic gas within the current virial radius ; at earlier times only a very small amount of dropout occurs in the low - density pre - shock cosmological flow . the results of the spatially distributed dropout ( do ) models in the ( @xmath13)-plane are shown in figure 1c . the do models exhibit many of the same trends as the co models in figure 1b but the cooled baryons are less concentrated so the x - ray surface brightness distribution is also less centrally concentrated . for each of the three virial masses we consider , the total mass of cooled gas in the do models @xmath180 within the current virial radius is insensitive to the amount of pre - heating , as shown in table 3 . the filled triangles in figure 1c , corresponding to models with cluster mass halos , doc , are in excellent agreement with observations and are largely unaffected by the various levels of pre - heating . the poor cluster solutions lie above the arnaud - evrard observations , but recall that their sample favored clusters with weak cooling flows which are hotter and therefore appear systematically underluminous in the @xmath13 plot . also arnaud & evrard did not extrapolate the observed x - ray luminosities to the virial radius . for both groups and poor clusters the luminosity decreases as the amount of pre - heating increases . the emission - weighted temperatures also decrease slightly as do1 @xmath125 do4 since the more strongly pre - heated models produce less total mass dropout and the lowered gravity of this dropout mass requires lower gas temperatures for hydrostatic support . overall , the results in figure 1c for group , poor cluster and rich cluster do models are in good agreement with the observations , particularly if the pre - heating is strong . models dog1 - dog3 , with little or no pre - heating , clearly lie above the group observations of helsdon & ponman ( 2000a ) . the most consistent models , when compared to the observations , are dog4 , dopc4 and doc4 . while this clearly supports the argument for cosmic pre - heating that has been widely discussed , we note that the amount of pre - heating required can not be explained solely by supernova heating associated with normal star formation ( e.g. loewenstein 2000 ) . the timing is also wrong since most galactic stars are thought to have formed at redshifts much less than @xmath32 , corresponding to the epoch of pre - heating @xmath31 gyr . since the dog4 model agrees so well with typical heldson - ponman groups in figure 1c , it is interesting to estimate the amount of pre - heating that would be required to raise the entropy to the dog4 ( preshock ) level at various redshifts . the amount of heating required to reach the same entropy of the dog4 model is 3.9 , 0.92 , 0.59 , 0.33 and 0.15 kev / particle if the heating occurred at redshift @xmath90 , 4 , 3 , 2 and 1 respectively . if about 0.1 of all baryons formed into stars ( @xmath181 fukugita , hogan & peebles 1998 ) with a salpeter imf ( as discussed above ) , this would generate only about 0.19 kev / particle . the required level of cosmic entropy in model dog4 ( figure 1c ) would be ( just ) consistent with supernova heating and normal star formation if most stars formed at redshifts @xmath182 . such recent supernova heating is clearly inconsistent with our knowledge that most stars in group - dominant early type galaxies were formed well before redshift @xmath183 . therefore , if the pre - heating is cosmic and universal it must have a non - stellar ( agn ? ) origin or be produced by ( pop iii ? ) stars that have no surviving imf counterparts today . in addition , the pre - heating in our models occurred at @xmath31 gyr before very much gas entered the dark potential ( e.g. wu , fabian & nulsen 1999b ) ; if the heating by stars occurred at later times , much luminous gas would have entered the dark potentials , in possible conflict with the observed x - ray background . the behavior of do models at time @xmath34 gyrs in the @xmath11 plot is illustrated in figure 2c . for each virial mass the entropy decreases with decreasing amounts of pre - heating . however , the dog models that fit best in the @xmath11 plot , dog1 and dog2 , agree less well with observations in the @xmath10 plot , figure 1c ; the best compromise model may be dog3 . however , we note again that the @xmath11 plot ( with @xmath145 and @xmath74 evaluated at @xmath184 ) is a less accurate indicator of the amount of pre - heating than the more globally representative @xmath11 plot in figure 4c or the @xmath10 plot ( figure 1c ) . figure 3 shows the final entropy profiles in the group and cluster do models . many of the same features shown for the co models appear again , but there are some significant differences , particularly in the inner regions @xmath148 . most of these differences can be understood in terms of the gas velocity . in the co solutions , the gas accelerates within @xmath137 as it approaches the central mass concentration , so less energy is radiated in this region and the entropy profiles become flat . by contrast , in the do solutions the central gas velocity is slowed by mass dropout and a much larger fraction of its entropy is radiated away . dropout causes subsonic flows to move even slower ( sarazin & ashe 1989 ; brighenti & mathews 2000b ) . this explains the continued decrease in @xmath142 in many of the do models as gas flows from @xmath185 to the origin . for the dog models this region would be dominated by the central elliptical galaxy . note that the group flow without pre - heating , dog1 , contains a low - entropy drip wave at @xmath186 which is moving slowly inward , but it has no strong influence on the solution elsewhere . radiation losses are less important in dropout flows with the largest pre - heating , dog3 , dog4 and doc4 , and the entropy remains more nearly constant throughout the inner flow . cooled mass profiles for representative do group and cluster models are shown in figure 5 and the total dropout mass within the virial radius @xmath180 is listed in table 3 . it is remarkable that the dropout mass distributions for the groups ( dog1 & dog4 ) resemble the de vaucouleurs profile of ngc 2563 shown with a dotted line in figure 5 . the total dropout mass @xmath180 is also similar to total stellar masses of galaxy groups . the similarity with a de vaucouleurs profile is coincidental , since hierarchical merging can also do this and is more plausible physically . moreover , in a more realistic model the mass dropout profiles would be lowered by internal supernova heating ( feedback ) not included in the do models . the dropout profiles in figure 5 for groups are almost identical for all models within @xmath187 since this gas cooled before @xmath33 when all the models for given @xmath20 were identical . in the more important interval @xmath188 all dropout profiles exceed that of the profile just before pre - heating at @xmath33 ( dashed - dot line ) which for dog(@xmath33 ) terminates at @xmath189 , the location of the accretion shock at @xmath33 . in this range we see that more dropout mass is deposited in models with less pre - heating ( dog1 ) or with no uniform reset at @xmath33 ( see below ) . the dropout profiles for cluster flows behave in a qualitatively similar manner , but are 30 - 100 times denser than ngc 2563 in @xmath190 . the dropout density in this region can be reduced by @xmath191 ( without much influencing @xmath51 and @xmath74 ) if the initial tophat perturbation is more extended , less dense and more massive . nevertheless , the high density mass dropout cores in do cluster models indicate that our mass dropout assumptions are inappropriate , at least on cluster scales . finally , each of the mass dropout profiles has a small density peak at @xmath192 kpc ; this feature is an artifact of mass dropout that accompanied the first compression and heating of baryonic gas in the tophat perturbation . do models are also sensitive to the epoch of pre - heating for example , when @xmath33 is increased to 2 gyrs , the final luminosities are reduced by factors of 6 , 30 , 75 and 140 for models dog1 @xmath125 dog4 . we have also computed `` no reset '' dropout models in which the temperature is increased at time @xmath31 gyr without altering the gas density profiles . in models do1nr , do2nr , do3nr and do4nr the temperature was increased throughout the flow at by 0 , 0.5 @xmath193 , 1.0 @xmath193 and 3.0 @xmath193 k respectively . global parameters for donr models for groups and clusters are listed in table 3 . the locus of clusters in the ( @xmath13)-plot is not strongly influenced by ignoring the reset , but the final group emission - weighted temperatures are increased by about 40 percent ( table 3 ) . emission - weighted temperatures reflect the temperature of gas in the high density cores . if we had used mass - weighted temperatures , the temperatures for dognr models would have been nearly equal to dog temperatures . in general pre - heating is less effective in the no reset models since the heating is radiated away by the high gas densities after time @xmath33 . as a result , the entropy profiles @xmath142 for these models vary as @xmath194 in @xmath148 . in this ga series of gas - dynamical models we abandon the hypothesis of universal pre - heating and assume that all heating results from normal star formation inside the group or cluster . this is similar to the approach we have used to reproduce the x - ray emission profile observed in the giant elliptical ngc 4472 ( brighenti & mathews 1999a ) . the dark halos evolve just as in the previous models , but the intergalactic baryonic gas is assumed to remain at @xmath143 k until it arrives at the accretion shock . we assume that stars form at some early time , @xmath195 gyrs ( redshift @xmath196 ) , after enough baryons have entered the dark potential . before time @xmath197 we use the dropout model to approximate the distribution of cooled baryonic mass . at time @xmath197 , however , the baryons are rearranged into a density profile similar to that of a luminous elliptical galaxy . we use this same galaxy core at the centers of poor cluster and cluster calculations with an additional extended stellar component approximated with a king distribution ; the gravity produced by these extended cluster stars has almost no influence on the gas dynamics . also at time @xmath197 we release the appropriate snii energy within the shock radius @xmath198 in proportion to the local gas density . shortly thereafter , a snii - driven starburst wind occurs in galaxy group halos , rapidly expelling most of the hot gas from the vicinity of the central group . a strong shock moves upstream against the converging cosmic gas . the gas that participated in the starburst wind reverses velocity and later shocks back onto the group for the second time , with a further entropy increase . for the deeper dark matter halos in poor clusters and clusters , the starburst can not expel much gas beyond the current virial radius . within the half - light radius of the central galaxy most of the hot gas is provided by stellar mass loss . for the ga series of models we include additional terms in the gas dynamical equations to allow for stellar mass loss and heating by stars and type ia supernovae ; these modified equations are described in detail in brighenti & mathews ( 1999a ) . our ga models are based on the mass distribution in galaxy group ngc 2563 . the central elliptical galaxy has luminosity @xmath199 @xmath200 ( distance = 78 mpc ) and total stellar mass @xmath201 @xmath49 . in addition , zabludoff & mulchaey ( 1998 ) have identified about 45 galaxies in the group surrounding ngc 2563 . using the approximate morphological types of zabludoff & mulchaey and @xmath202 ratios from fukugita , hogan & peebles ( 1998 ) , we estimate that the total stellar mass in the ngc 2563 group is at least 1.5 times that of the central elliptical galaxy . in the models discussed here we assume that the stellar mass of the entire group is twice that of the central galaxy ngc 2563 . the contribution of outlying group members to the total stellar mass of groups is highly variable in galaxy groups and is often considerably larger than that for ngc 2563 . immediately after the time of star formation @xmath197 the ratio of stellar to total baryons within the accretion shock is @xmath203 . this ratio is based on the estimated stellar mass of the ngc 2563 group , but we use this same ratio in poor clusters and cluster flows . we note that the fraction of stellar baryons exceeds the cosmic average @xmath181 at zero redshift ( fukugita , hogan & peebles 1998 ) . however , additional gas accretes into the virial radius after time @xmath197 and by the present time @xmath204 decreases to 0.27 in the gag2 model and to 0.15 in models gapc2 and gac2 . the final ratio of stellar to gas mass inside @xmath20 is @xmath205 for gag2 and 0.23 for gapc2 and gac2 . in figure 6 we show the spatial variation of the relative mass in baryons and gas for models gag1 an gag4 . the relative mass of baryons @xmath206 also includes the dropout mass of cooled gas . note that the gas fraction @xmath207 is lower than the cosmic value even at the virial radius and that its value is sensitive to the total amount of snii energy released . except for the @xmath0cdm cosmology assumed here , the gas dynamical calculations , stellar mass loss rates and type ia supernova rates that we use are identical to those described in brighenti & mathews ( 1999a ) . we assume distributed mass dropout with parameter @xmath177 . the energy provided by type ii supernovae is estimated by assuming @xmath208 ergs is released from stars with initial masses @xmath209 and that @xmath210 type ii supernovae are produced per solar mass of stars formed . for a typical salpeter imf ( slope @xmath211 , @xmath212 , @xmath213 ) we find @xmath214 . the snii energy delivered to the hot gas in a galaxy group or cluster of total stellar mass @xmath215 is @xmath216 where @xmath217 is the efficiency that the supernova energy is delivered to the thermal energy of the hot gas . in cosmological simulations the efficiency of type ii supernova feedback is often chosen to be @xmath218 , to allow for radiative losses . our calculation is somewhat different since we explicitly allow for radiative losses in the thermal energy equation , although we also assume that the supernova blast waves interact directly with the hot gas , not with cold , dense clouds as might be expected in star - forming regions . we therefore consider several values of the composite parameter @xmath219 which is expected to be of order unity . for the ga models we consider four values : @xmath220 , 1 , 2 and 4 , respectively designated as ga1 , ga2 , ga3 and ga4 . since @xmath221 is the same in all ga models , the gas temperature following the deposition of snii energy is increased by @xmath222 k. the gag models for galaxy groups are most satisfactory since the stellar mass distribution and formation epoch @xmath197 are designed with groups in mind . it is particularly gratifying that all gag models fall very close to the observed position of ngc 2563 in the @xmath13 plot , shown with a large open square in figure 1d . since we used ngc 2563 to estimate the stellar mass distribution for our group models , the agreement in figure 1d may confirm the approximations we have made in the early evolution of groups . the gag2 and gag4 luminosities and temperatures are within about 10 and 25 percent of the values observed in ngc 2563 ( see table 4 for further details ) . all gag models also lie near the upper envelope of the helsdon - ponman group data , where strong cooling flows are dominated by a massive central elliptical galaxy as we have assumed . generally , as @xmath223 increases along the sequence gag1 @xmath125 gag4 the x - ray luminosities decrease . however , the non - monotonic behavior of model gag3 , @xmath224 , can be understood by competing influences on @xmath51 : larger heating ( @xmath223 ) decreases the hot gas density , which lowers @xmath51 , but as @xmath223 increases less gas cools and the larger surviving hot gas mass increases @xmath51 . values of @xmath223 as large as 4 used in model gag4 may not be unreasonable . mathews ( 1989 ) estimated the stellar mass to light ratio @xmath225 of luminous elliptical galaxies using single burst stellar populations with power law imfs having many slopes and mass limits . he showed that the salpeter slope @xmath211 is inconsistent with @xmath226 which is typically observed in luminous elliptical galaxies , while shallower imf slopes , @xmath227 , give satisfactory @xmath225 for all reasonable upper and lower mass limits . a power law imf characterized with slope @xmath228 , and mass limits @xmath212 and @xmath213 , delivers much more snii energy than the salpeter imf , @xmath229 , which may justify our largest values of @xmath223 . for the gag models we have also explored varying the time @xmath197 when the snii energy is released and the radial distribution where the snii energy is deposited . for the standard model , gag2 , we varied @xmath230 from 2 to 1.5 or 1 gyr and found that the computed points in the @xmath13 plane at time @xmath34 gyrs were nearly unchanged . in addition , we repeated the gag2 calculation but applied all of the snii energy only to ( denser ) gas within half the accretion shock radius at time @xmath231 gyrs . the final luminosity increased slightly since the more central heated gas was denser and lost a larger fraction of its energy ( and entropy ) to radiation . we conclude that the positions of the group gag models in the @xmath13 plot ( figure 1d ) are not sensitive to the time or spatial dependence of the snii energy release . for completeness we have also computed gapc and gac models for poor clusters and clusters . these results shown in figure 1d and table 4 appear to have temperatures about 50 percent lower than those of the pre - heated models or the corresponding observations . while our ga assumptions are reasonable for group evolution , they are less appropriate for clusters . for example , we assume that clusters grow from a single seed galaxy group in the core , ignoring the complex merging events by which groups combine to form large clusters . the group models gag fall nicely among the observations in the @xmath26 plane shown in figure 2d regardless of the heating parameter @xmath223 . this insensitivity of @xmath232 to the initial snii energy @xmath223 is a result of radiative losses in this inner region . strong radiative cooling at the galactic center regulates the central density there to @xmath233 @xmath61 and the temperature profile adjusts to maintain hydrostatic equilibrium in the common potential of all ga models . therefore @xmath166 evolves to a similar value for all @xmath223 . the results for poor clusters and clusters are less satisfactory . some of the small perturbations of the entropy at @xmath234 discussed earlier are also present in the ga models , but tend to be smoothed due to radiative losses and mass dropout . as before , when the entropy and temperature are mass - averaged within @xmath20 the behavior is much more reasonable , as shown in figure 4d . final entropy profiles @xmath142 are shown for gag and gac models in figure 3 . the observed entropy profile for ngc 2563 is shown with the gag models . the agreement is quite good although our models are somewhat denser overall with slightly lower @xmath142 . as @xmath223 increases , the gag models have progressively more extended regions of high entropy that project beyond the current virial radius . these entropy features are unlike any of the other models and , if they could be observed , would be a clear signature of the internal heating scenario . the outer gag flows contain two outward - facing shocks . the outer shock is the starburst blast wave that began at time @xmath197 ; the inner shock which occurs close to @xmath20 is the accretion shock . variations in snii heating have little effect on the entropy profile or accretion shock radius in the gac solutions because of their much greater mass . a positive entropy gradient @xmath235 extends to very small radii in both gag and gac solutions in figure 3d . this is due to mass ejection of low entropy gas from galactic stars and radiative cooling in high density gas within the deep central stellar potential . in the foregoing we have described the heating and dynamical evolution of hot gas in galaxy groups and clusters in a flat @xmath0cdm universe with a special emphasis on the gas entropy . the motivation for our interest in this problem can be traced to kaiser ( 1986 ; 1991 ) who first noted that the x - ray luminosities of groups and clusters vary with gas temperature more steeply than that expected from bremsstrahlung emission in purely adiabatic flow with adiabatic shocks , @xmath1 . the observed steeper variation @xmath2 or @xmath47 can be understood if cosmic gas is heated by some non - gravitational process . the density and x - ray luminosity of hydrostatically supported gas in dark halos are both lower when the gas is heated . if hot gas in both groups and clusters experienced the same level of heating , the luminosity of groups should be disproportionally lower since the heating in the adiabatic accretion shock is much greater in clusters , typically masking the amount of non - adiabatic heating required to account for the steep @xmath13 variation . if gas and dark matter scale homologously within the virial radius and if all the gas that passes through the accretion shock remains within @xmath20 , the gas density should be the same at any fraction of @xmath20 regardless of the mass of the cluster . in this case the entropy factor @xmath166 is proportional to @xmath74 . lloyd - davies et al . ( 2000 ) have used the x - ray data to estimate the entropy factor @xmath136 at @xmath7 in group and cluster gas . they find that when the @xmath138 relation is normalized to massive clusters , the values of @xmath141 for groups are too large to fit the same relation . either some gas in groups flowed out because it was internally heated or an insufficient amount of gas entered the group at early times because the gas was pre - heated before the accretion occurred . this leads directly to the question we have addressed here : was the diffuse gas in galaxy groups and clusters pre - heated by some external process at high redshifts or was the heating internal , due to supernovae that accompanied normal star formation ? some recent studies of this question have invoked a universal cosmic pre - heating from sources unrelated to star formation . in view of its importance for galaxy formation , we have developed an ensemble of gas dynamical models with and without pre - heating . to investigate the consequences of universal pre - heating we study the accumulation of gas in group and cluster dark halos with three progressively more realistic assumptions : adiabatic ad flows with adiabatic shocks , co flows with radiative cooling ( tozzi & norman 2000 ) , and finally do flows with mass dropout and radiative cooling ( knight & ponman 1997 ; nulsen & fabian 1995 , 1997 ) . in order to compare our results with those of previous studies , we also assume that the gas temperature and density are reset to uniform values at the moment when the pre - heating occurred , i.e. excess gas inside the dark halo is removed . for most of our models the reset is assumed to occur at time @xmath31 gyr or redshift @xmath32 . in spite of its popularity in the current literature , there is some uncertainty about the reset and pre - heating procedure . while very few stars have formed at these high redshifts and the only collapsed objects correspond to very small stellar systems , the tophat perturbations that we use to generate present day group and cluster halos have collected baryonic gas which is removed at time @xmath33 by the reset assumption . we have also done many calculations with no reset . the dark halos evolve in a normal way , initiated by a tophat perturbation and cooling toward an nfw profile in a flat @xmath0cdm universe . the final halo is insensitive to the tophat perturbation , i.e. the same halo can be produced by a variety of different tophat density amplitudes with appropriately chosen tophat radii . in general we find that data in the @xmath13 plot ( figure 1 ) can be fit with ad , co or do flows with reset provided the pre - heating is sufficiently intense , 1.3 to 3.9 kev / particle ( @xmath236 ) . however , these strongly pre - heated flows typically have entropies @xmath141 for groups and poor clusters that exceed observed values ( figure 2 ) . values of the entropy and temperature that are mass - weighted within the virial radius obey the @xmath139 similarity relation except for the most strongly pre - heated group gas ( figure 4 ) , but these can not be compared with current observations . however , a variety of computational and physical problems have appeared in our models with cosmic pre - heating . for example , weakly heated ad models are not only less successful in the @xmath13 plot , but also show irregularities due to transient shocks following the reset that persist as irregularities in the entropy profiles @xmath142 in the final ad1 models ( figure 3 ) . @xmath141 may be increased by these spurious shocks . with more computational effort these irregularities could probably be removed . but it is clear that adiabatic ( ad ) models can not be correct in spite of their considerable success in fitting the @xmath13 data ( with large pre - heating ) and the entropy profiles ( with little pre - heating ) : adiabatic flows can not form stars nor can they radiate x - rays . when radiative losses are included , as in the co flows , we find that some pre - heated models agree with data the @xmath13 plot and have entropy profiles very similar to those of tozzi & norman ( 2000 ) who make similar assumptions . however , group models with less pre - heating and all cluster models except coc4 have enormously massive and luminous x - ray singularities produced by emission from uncooled gas flowing into the center of the cluster . because of their luminous cores , these models have absurdly high x - ray luminosities and do not correspond to any object observed . another somewhat technical , but physically real , aspect of co flows is their tendency to produce non - linear thermally unstable drip waves that propagate into the core of the flow sometimes at high velocity . the drip waves are rayleigh - taylor unstable so they can not remain perfectly coherent , unlike their representations in our one - dimensional models , and they will tend to be limited by mass dropout , not included in the co models . finally , our results differ from those of tozzi & norman ( 2000 ) in that the accretion shock is very close to the virial radius for all levels of pre - heating considered . to avoid these difficulties encountered in the co models , we performed a final do series of pre - heated models in which gas is allowed to cool to very low temperatures everywhere within the virial radius . this is a standard assumption used in models of cooling flows . as before , strongly pre - heated do models that fit the @xmath13 data have entropies @xmath141 that exceed observed values . while the ultra - luminous x - ray cores of the co flows are no longer present in the do models , the accumulated mass of cooled ( dropout ) mass can be large , particularly in massive clusters where it exceeds the central mass densities of luminous elliptical galaxies by 10 - 100 . we have also performed calculations of all types of externally pre - heated flows with no uniform reset of the gas density and temperature at time @xmath33 . in these models the cosmic gas temperature is heated by various amounts throughout the flow at @xmath33 , but the density profile is not altered . to reduce the final luminosity @xmath51 by the same amount as in the reset models , the no reset models require much higher levels of pre - heating to compensate for the energy radiated away from denser cores . the conr models are totally unacceptable because of the intense concentration of x - ray emission from the core where all the gas has cooled . donr models are more similar to their do counterparts since most of the heating is radiated away by dense gas in the core . although dognr flows are somewhat hotter than the corresponding dog flows , they nevertheless have lower central entropies @xmath141 . for a variety of reasons , none of our pre - heated models are particularly attractive , with or without reset . in addition to gas dynamical concerns , the agency that heats the gas is unclear . since the cosmic gas density is higher in the early universe , more energy is required to reach the same adiabat than at lower redshifts . in the flat @xmath0cdm universe global pre - heating by type ii supernovae during star formation is unable to heat all the gas sufficiently at redshifts @xmath237 to achieve agreement with galaxy group observations in the @xmath13 plane . there may be enough energy in agn to heat the gas at high redshifts , but it is unclear if this energy can be widely distributed in the intergalactic gas and not just concentrated near the agn . in comparing pre - heated models with observations , we place more emphasis on the @xmath13 plane than the @xmath238 plane . the @xmath141 data are of lower quality because they typically refer to the outermost detectable x - ray emission from groups where the background corrections may be troublesome . in addition the virial radius must be estimated to derive @xmath141 from the observations . some of our models are also less certain in the inner parts of the flow @xmath148 where the entropy can be increased by early shock waves or rapidly moving drip waves . in our final series of ga models the external pre - heating assumption is replaced with internal heating by type ii supernovae associated with star formation at time @xmath239 gyrs and redshift @xmath196 ( see also loewenstein 2000 ; bryan 2000 ) . this internal heating is distributed within the accretion shock ( @xmath240 ) in proportion to the local gas density . immediately following @xmath197 a strong starburst wind occurs in the ga group solutions . high entropy gas produced by the starburst shock currently extends beyond the virial radius . accompanying the snii energy release is the formation of a luminous elliptical galaxy at the center of the group or cluster flow . as a guide , we have chosen the stellar mass of the ngc 2563 group to calibrate the amount of snii energy delivered to the hot gas in groups . this same snii heating per unit baryonic mass is applied to ga models for clusters . we have explored the effect of varying the time @xmath197 of star formation and altering the distribution of the snii energy within @xmath20 , but the final results are surprisingly insensitive to these parameters . the ga models resemble do models without reset since both include cooling mass dropout and have dense baryonic cores ; the main difference is that the ga models are heated internally and do models are heated universally . while we have not attempted to produce a detailed model for the ngc 2563 group , we find that the x - ray luminosity and temperature observed in ngc 2563 are almost exactly matched by our ga group model at time @xmath34 gyrs when heated with snii energy expected from a salpeter imf . the data in the @xmath238 plane is also nicely fit by the group ga models . finally , we find that our standard ga group models agree well with the entropy distribution @xmath142 observed in ngc 2563 . excellent agreement with observations in both the @xmath13 and @xmath26 plots can be achieved with a range of 8 in the snii energy released per unit stellar mass , corresponding to different imfs . however , for the best results we require that at least @xmath12 of the energy released by snii from early star formation directly heats the hot gas . the local environment of the early snii explosions may differ from star - forming regions in our galaxy in that the supernova blast waves propagate into the hot gas , not into cold , dense clouds which can radiate away much of the supernova energy . of course the hot gas heating efficiency could be @xmath241 if the imf were more top - heavy than expected . we also present ga models for clusters , although our physical model of monolithic growth ( without detailed hierarchical merging ) is less appropriate for cluster masses . mushotzky and scharf ( 1997 ) have shown that the evolution of data in the @xmath13 plane is very weak out to redshifts @xmath242 . our ga group models also exhibit this same slow evolution . for example , the mean bolometric x - ray luminosity and gas temperature for the four ga models we consider differ at @xmath183 by only 45 and -7 percent respectively from the same averages at zero redshift , indicating that the temperature is stable and that the luminosity may be declining slightly . in view of the generally good agreement with observations , we conclude that galaxy group models with internal snii heating fit the data better and more plausibly than any of the externally pre - heated models we have considered . the production of snii energy and the transfer of this energy to the hot gas may be a concern for the internal heating models we prefer , but there has been no satisfactory explanation of the sources of the much greater energy associated with universal , external pre - heating at high redshifts . we wish to thank referee paolo tozzi for helpful criticism and advice . studies of the evolution of hot gas in elliptical galaxies at uc santa cruz are supported by nasa grant nag 5 - 3060 and nsf grant ast-9802994 for which we are very grateful . allen , s. w. & fabian , a. c. 1998 , mnras , 297 , 57 arnaud , m. & evrard , a. e. 1999 , mnras , 305 , 631 balogh , m. l. , babul , a. & patton , d. r. 1999 , mnras , 307 , 463 bertschinger , e. 1985 , apjs , 58 , 39 brighenti , f. & mathews , w.g . 2000a , apj , 539 , 675 brighenti , f. & mathews , w.g . 2000b , apj , 535 , 650 brighenti , f. & mathews , w.g . 1999a , apj , 512 , 65 brighenti , f. & mathews , w.g . 1999b , apj , 515 , 542 brighenti , f. & mathews , w.g . 1998 , apj , 495 , 239 bryan , g.l . 2000 , apjl , accepted ( astro - ph/0009286 ) bullock , j. s. et al . 2000 , mnras , accepted ( astro - ph/9908159 ) buote , d. a. 2000a , mnras 311 , 176 buote , d. a. 2000b , apj , accepted ( astro - ph/0001329 ) burles , s. & tytler , d. 1998a , apj 507 , 732 burles , s. & tytler , d. 1998b , apj 499 , 699 cavaliere , a. , giacconi , r. & menci , n. 2000 , apjl , accepted ( astro - ph/9912374 ) cavaliere , a. , menci , n. & tozzi , p. 1997 apj , 484 , l21 cen , r. & ostriker , j. p. 1999 , apj , 514 , 1 dav , r. et al . 2000 , apj , submitted ( astro - ph/0007217 ) david , l. p. 1997 , apj 484 , l11 david , l. p. , arnaud , k. a. , forman , w. , & jones , c. 1990 , apj 356 , 32 david , l. p. & blumenthal , g. r. 1992 , apj 389 , 510 eke , p. a. , navarro , j. f. & frenk , c. s. 1998 , apj , 503 , 569 evrard , a. e. & henry , j. p. 1991 , apj , 383 , 95 fukugita , m. , hogan , c. j. & peebles , p. j. e. 1998 , apj , 503 , 518 helsdon , s. f. & ponman , t. j. 2000a , mnras , accepted ( astro - ph/0002051 ) helsdon , s. f. & ponman , t. j. 2000b , mnras , accepted ( astro - ph/0008194 ) kaiser , n. 1991 , apj , 383 , 104 kaiser , n. 1986 , mnras 222 , 323 knight , p. a. & ponman , t. j. 1997 , mnras , 289 , 955 kravtsov , a. v. & yepes , g. 2000 , mnras , submitted ( astro - ph/0004333 ) lloyd - davies , e. j. , ponman , t. j. & cannon , d. b. 2000 , mnras , 315 , 689 loewenstein , m. 2000 , apj , 532 , 17 markevitch , m. 1998 , apj , 504 , 27 mathews , w. g. 1997 , aj , 113 , 755 mathews , w. g. 1989 , aj , 97 , 42 mushotzky , r. f. & scharf , c. a. 1997 , apj 482 , 13 navarro , j. f. , frenk , c. s. & white , s. d. m. 1995 , mnras , 275 , 720 navarro , j. f. , frenk , c. s. & white , s. d. m. 1996 , apj , 462 , 563 nulsen , p. e. j. & fabian , a. c. 1995 , mnras , 277 , 561 nulsen , p. e. j. & fabian , a. c. 1997 , mnras 291 , 425 pen , u - l . 1999 , apj , 510 , l1 ponman , t. j. , cannon , d. b. & navarro , j. f. 1999 , nature , 397 , 135 renzini , a. 1997 , apj , 488 , 35 sarazin , c.l . , ashe , g.a . 1989 , apj , 345 , 22 stone , j. m. & norman , m. l. 1992 , apjs , 80 , 573 sutherland , r.s . , dopita , m.a . 1993 , apjs , 88 , 253 tozzi , p. & norman , c. 2000 , apj , accepted ( astro - ph/0003289 ) tozzi , p. , scharf , c. & norman , c. 2000 , apj , 542 , 106 trinchieri , g. , fabbiano , g. & kim , d .- w . 1997 , a&a 318 , 361 valageas , p. & silk , j. 1999 , a&a 350 , 725 wu , k. k. s. , fabian , a. c. & nulsen , p. e. j. 1998 , mnras , 301 , l20 wu , k. k. s. , fabian , a. c. & nulsen , p. e. j. 1999a , ( astro - ph/9907112 ) wu , k. k. s. , fabian , a. c. & nulsen , p. e. j. 1999b , ( astro - ph/9910122 ) zabludoff , a. i. & mulchaey , j. s. 1998 , apj 496 , 39 lcccc adg1 & 0 .814 & 43.55 & 2.02 & 0.0918 adg2 & 0 .906 & 43.12 & 2.15 & 0.0900 adg3 & 0 .853 & 42.86 & 2.43 & 0.0856 adg4 & 0 .843 & 41.94 & 2.91 & 0.0420 adpc1 & 1.89 & 44.19 & 2.55 & 0.0931 adpc2 & 2.44 & 43.95 & 2.42 & 0.0916 adpc3 & 2.46 & 43.83 & 2.51 & 0.0899 adpc4 & 2.36 & 43.41 & 2.91 & 0.0811 adc1 & 6.13 & 44.99 & 2.87 & 0.0907 adc2 & 6.72 & 44.91 & 2.68 & 0.0902 adc3 & 6.96 & 44.85 & 2.82 & 0.0897 adc4 & 7.20 & 44.68 & 2.97 & 0.0877 adg1nr & 0 .392 & 46.14 & 1.83 & 0.121 adg4nr & 0 .425 & 44.76 & 1.85 & 0.068 adpc1nr & 0.760 & 46.41 & 1.63 & 0.121 adpc4nr & 0.908 & 45.47 & 2.22 & 0.100 adc1nr & 1.714 & 46.78 & 2.44 & 0.117 adc4nr & 2.399 & 46.14 & 2.81 & 0.114 lclccc cog1 & 1.10 & 43.67(44.55 ) & 1.76 & 0.083(0.095 ) & 11.77 cog2 & 1.32 & 43.82(44.62 ) & 1.81 & 0.080(0.096 ) & 11.59 cog3 & 0.886 & 43.11 & 2.21 & 0.092(0.092 ) & cog4 & 0.831 & 42.00 & 2.87 & 0.045(0.045 ) & copc1 & 2.45 & 44.21(44.80 ) & 2.15 & 0.09(0.094 ) & 11.94 copc2 & 2.39 & 44.24(44.90 ) & 2.28 & 0.091(0.094 ) & 11.80 copc3 & 2.32 & 44.10 & 2.34 & 0.092(0.092 ) & copc4 & 2.38 & 43.47 & 2.85 & 0.082(0.082 ) & coc1 & 7.12 & 44.87(45.40 ) & 2.78 & 0.088(0.091 ) & 12.55 coc2 & 7.14 & 44.92(46.04 ) & 2.89 & 0.088(0.091 ) & 12.53 coc3 & 6.15 & 45.11(45.90 ) & 2.91 & 0.089(0.090 ) & 12.16 coc4 & 7.12 & 44.77 & 2.90 & 0.088(0.088 ) & cog1nr & 2.58 & 43.00(45.35 ) & 2.40 & 0.065(0.126 ) & 12.48 cog4nr & 1.28 & 42.28(44.10 ) & 2.53 & 0.047(0.079 ) & 12.16 coc1nr & 14.93 & 44.91(47.69 ) & 2.94 & 0.085(0.118 ) & 13.60 coc4nr & 8.54 & 44.87(46.93 ) & 2.97 & 0.092(0.114 ) & 13.41 lccccc dog1 & 0.933 & 43.61 & 1.99 & 0.072(0.114 ) & 12.31 dog2 & 0.941 & 43.61 & 2.05 & 0.072(0.115 ) & 12.35 dog3 & 0.795 & 43.27 & 2.32 & 0.081(0.111 ) & 12.17 dog4 & 0.743 & 42.34 & 2.89 & 0.045(0.067 ) & 11.98 dopc1 & 2.27 & 44.35 & 2.25 & 0.084(0.106 ) & 12.68 dopc2 & 2.55 & 44.36 & 2.43 & 0.077(0.107 ) & 12.81 dopc3 & 2.02 & 44.21 & 2.47 & 0.084(0.105 ) & 12.64 dopc4 & 1.89 & 43.76 & 2.87 & 0.080(0.095 ) & 12.48 doc1 & 6.17 & 45.20 & 2.85 & 0.084(0.100 ) & 13.27 doc2 & 9.05 & 45.36 & 2.96 & 0.076(0.102 ) & 13.47 doc3 & 6.59 & 45.31 & 2.96 & 0.084(0.100 ) & 13.29 doc4 & 5.13 & 45.03 & 2.97 & 0.085(0.098 ) & 13.17 dog1nr & 1.24 & 43.59 & 1.98 & 0.066(0.123 ) & 12.45 dog2nr & 1.54 & 43.89 & 1.95 & 0.069(0.129 ) & 12.48 dog3nr & 1.29 & 43.49 & 2.24 & 0.083(0.125 ) & 12.33 dog4nr & 1.43 & 43.31 & 2.25 & 0.050(0.079 ) & 12.12 doc1nr & 6.74 & 45.47 & 2.86 & 0.088(0.117 ) & 13.55 doc2nr & 6.39 & 45.45 & 2.84 & 0.080(0.119 ) & 13.67 doc3nr & 4.77 & 45.70 & 2.79 & 0.088(0.118 ) & 13.55 doc4nr & 5.57 & 45.57 & 2.76 & 0.095(0.115 ) & 13.37 lccccc gag1 & 0.871 & 43.23 & 2.24 & 0.066(0.102 ) & 11.97 gag2 & 0.894 & 42.88 & 2.26 & 0.040(0.071 ) & 11.81 gag3 & 0.804 & 43.13 & 2.04 & 0.027(0.052 ) & 11.58 gag4 & 0.807 & 42.62 & 2.23 & 0.017(0.039 ) & 11.22 gapc1 & 1.866 & 44.26 & 2.34 & 0.087(0.123 ) & 12.86 gapc2 & 1.835 & 44.15 & 2.48 & 0.088(0.123 ) & 12.84 gapc3 & 1.774 & 43.91 & 2.62 & 0.083(0.116 ) & 12.81 gapc4 & 1.435 & 43.89 & 2.67 & 0.054(0.083 ) & 12.74 gac1 & 3.644 & 45.31 & 2.98 & 0.086(0.118 ) & 13.57 gac2 & 3.812 & 45.30 & 2.98 & 0.087(0.118 ) & 13.56 gac3 & 4.091 & 45.25 & 2.99 & 0.089(0.118 ) & 13.54 gac4 & 4.281 & 45.08 & 2.52 & 0.085(0.113 ) & 13.51	the entropy in hot , x - ray emitting gas in galaxy groups and clusters is a measure of past heating events , except for the entropy lost by radiation from denser regions . observations of galaxy groups indicate higher entropies than can be achieved in the accretion shock experienced by gas when it fell into the dark halos . these observations generally refer to the dense , most luminous inner regions where the gas that first entered the halo may still reside . it has been proposed that this non - gravitational entropy excess results from some heating process in the early universe which is external to the group and cluster halos and that it occurred before most of the gas had entered the dark halos . this universal heating of cosmic gas could be due to agn , population iii stars , or some as yet unidentified source . alternatively , the heating of the hot gas in groups may be produced internally by type ii supernovae when the galactic stars in these systems formed . we investigate here the consequences of various amounts of external , high redshift heating with a suite of gas dynamical calculations . we consider the influence of radiation losses and distributed mass dropout on the x - ray luminosity and emission - weighted temperature of the hot gas as well as its central entropy . in general we find that externally heated flows are unsatisfactory ; when the heating is high enough to bring the x - ray luminosities into agreement with observations , the gas entropy is too high . we compare these solutions with flows that are internally heated by type ii supernova ; this type of heating depends on the imf and the efficiency that the supernova energy is conveyed to the hot gas . these internally heated flows give much better agreement with x - ray observations of galaxy groups and are insensitive to the levels of supernova heating that we consider as well as to the epoch and spatial distribution of the supernova heating process . however , to fit x - ray observations a large fraction of the energy produced by high redshift type ii supernovae must heat the hot gas if the number of supernovae is based on a salpeter initial mass function . alternatively , only about 20 percent of the type ii supernova energy would be required to heat the gas if the initial mass function has a flatter slope than salpeter , as suggested by stellar mass to light ratios . .2 in
physical structures of an accretion disk are determined by various physical factors such as gravity of the central star , turbulence , magnetic fields . understanding the time evolution of dust particles in circumstellar disc - like structures around protostars and young stellar objects is crucial for explaining the formation of planetary systems . due to various complex physical processes that are playing significant roles in dust and gas discs , it is difficult to present a complete and self - consistent model for such systems . because of the gas drag force , the motion of the dust particles is different from that of gas ( e.g. , weidenschilling 1977 ) . the gas drag force takes angular momentum from particles , resulting in their inward migration ( whipple 1972 ) . garaud , barri@xmath0re - fouchet & lin ( 2004 ) found that small particles move smoothly to the midplane , while large bodies oscillate about the midplane with decreasing amplitude . takeuchi & lin ( 2002 ) ( here after tl ) studied grains motion in optically thick disc , using the assumption that radiation pressure on the dust is negligible . they showed that as the gas accretion proceeds , small dust particles are left behind in the disc , and the dust to gas ratio increases . in a subsequent paper , takeuchi & lin ( 2003 ) investigated the outflow of dust particles on the surface layers of optically thick discs . klahr & lin ( 2001 ) studied the motion of grains in optically thin disc , where the dust experiences both gas drag and radiation pressure . in a similar analysis , takeuchi & artymowicz ( 2001 ) suggested that ring - like features which have been observed in some circumstellar debris discs ( e.g. schneider et al . 1999 ; marsh et al . 2005 ) , arise naturally in discs passing through the transitional phase from gas - dominated to dust - dominated discs . the dust particles migrate radially until they are in co - rotation with the gas . to our knowledge , most of the previous studies are assuming that the motion of particles in the gas and dust accretion discs exclusively caused by gravitation , gas drag force and radiation pressure of the central protostar . however , collisional dust grain electrification in the early solar nebula has long been discussed to be a cause for lightning , which could explain the melting of meteoritic chondrule precursors . poppe , blum @xmath1 henning ( 2000 ) studied collisional grain charging experimentally . they showed that collisional grain charging is stronger than previously discussed with respect to preplanetary grains and should be considered concerning the preplanetary dust aggregation and the formation of lightning in the solar nebula . it also has influences on other physical conditions in gas and dust discs , because charged particles can interact with the magnetic field . although the strength and direction of magnetic fields in accretion discs are rather unknown , the observation of winds and outflows is explained by magnetic acceleration mechanisms that require the existence of a magnetic field with strength of @xmath2 t at 1 au solar distance ( k@xmath3nigl @xmath1 ruden 1993 ) . the possible effects of magnetic fields on the structure of accretion discs have been studied both numerically or analytically using simplified models ( e.g. lovelace , romanova & newman 1994 ; bisnovatyi - kogan & lovelace 2000 ; igumenshchev , narayan & abramowicz 2003 ) . these studies , generally show that magnetic fields may have significant effects on the structure of the discs . so , in gas and dust discs , one may expect such type of behaviors due to the magnetic fields . however , the ionization structure of protoplanetary discs which determines the coupling between the magnetic field and the grains is likely to be inhomogeneous and complex . the ionization fraction at different locations , and even different times , are not easy to calculate ( e.g. ilgner & nelson 2006 ) . in fact , the idea of a magnetically dead - zone was first proposed by gammie ( 1996 ) . he argued that there is a poorly ionized region where the growth of magnetorotational instability against ohmic dissipation can not be sustained . the dead zone typically stretching out to around 10 au to 15 au ( e.g. reyes - ruiz 2001 ; matsumura & pudritz 2006 ) . wardle ( 2007 ) studied the ionization equilibrium and magnetic diffusivity as a function of height from the disc midplane at radii 1 and 5 au . he showed that the dead zone may not extend over the entire vertical dimension of the disc , because not only the ionization fraction but the gas density are strong functions of height above the midplane and so the magnetic coupling is likely to increase away from the midplane . moreover , the ionization fraction itself is a strong function of the presence ( and size ) of the dust particles mixed with the gas and recent calculations have shown that once the grains have settled , the ionisation level is sufficient to couple the magnetic field to the gas at 1 au , even at the midplane ( wardle 2007 ) . but the disc becomes magnetically active beyond the dead zone ( e.g. reyes - ruiz 2001 ) . in this study , we are applying a simplified and phenomenological model for describing the gas component of a gas and dust disc for regions beyond the dead zone . we are assuming only the dust particles which are influenced by an imposed magnetic field . the properties of any external magnetic fields threading a dust and gas disc are not well - known . however , it is very unlikely that the dipolar field of the central protostar is important at several au from the protostar where the planets are believed to form . more probably such an external field comes from the the molecular cloud core out of which the gas and dust disc formed . these considerations show that magnetic forces on micron - sized particles in gas and dust disks will be worth further investigation . in fact , the structure of magnetic fields within protostellar discs may be studied via polarimetry provided that grains are aligned in respect to magnetic field within the discs . in this study , we assume that the charged particles are collisionless and indestructible spheres in a gaseous disc and , to keep the problem tractable , we also neglect the feedback of the particles drag on the motion of the gas . we mainly follow the approach of tl in analyzing the dynamics of grains , but in an imposed magnetic field . we show that a vertical imposed magnetic field affects the radial and the vertical motions of charged dust particles . the basic equations of the model are described in section 2 . in sections 3 we solve for individual particles velocities , which are charged and small with sizes smaller than the mean free path of the gas molecules . in the final section , we summarize our results . we study the motion of a dust particle with mass @xmath4 and charge @xmath5 in a magnetized gas disc . one can simply show that the collisional timescale is longer than the stopping time by at least an order of magnitude ( youdin & shu 2002 ) . so , we can ignore the effect of collisions . for a charged dust particle , newton s second law can be written as @xmath6 where @xmath7 is the radius vector of the particle . also , @xmath8 , @xmath9 , @xmath10 and @xmath11 are the gravitational force of the central star , the gas drag , the magnetic force and the electric force , respectively . we neglect radiation pressure and poynting - robertson drag for simplicity . it is more convenient to write the equations of motion in cylindrical coordinates @xmath12 . as for the magnetic field geometry , we consider the toroidal and the vertical components of the magnetic field . if a charged grain is placed in a rotating frame , it will invariably feel an electric field . the keplerian velocity is denoted by @xmath13 , where @xmath14 is the mass of the central star . we can imagine one of two approaches : ( a ) either state that the equation is written in a frame where electric field is zero , i.e. @xmath15 in which case lorentz force term will become @xmath16 ; or ( b ) explicitly include an electric field term . we follow the first approach . in highly conducting plasma the electric field in the frame co - moving with the neutrals vanishes , so that the current density remains finite . but if we include the evolution of the magnetic field via the induction equation , this is not a good approximation in the presence of the charged dust grains . because charged dust particles can easily decouple from the field lines . in order to avoid this difficulty , the magnetic field is treated as an imposed constraint in this study . now we can write the radial and the azimuthal equations of motion , @xmath17 @xmath18,\label{eq : main1}\ ] ] @xmath19 where @xmath20 is the nondimensional stopping time . also , @xmath21 and @xmath22 are the @xmath23 and @xmath24 components of the velocity of the dust particle , i.e. @xmath25 and @xmath26 . note that @xmath27 is the rotational velocity of the gas component . in order to write the @xmath28component of the equation of motion , we assume the particles reach terminal velocity , at which the gravity , the gas drag and the lorentz force balance with each other , @xmath29 where @xmath30 . the nondimensional stopping time @xmath20 can the be written as ( tl ) , @xmath31 where @xmath32 represents the @xmath33 times the mean thermal velocity @xmath34 of the gas . also , @xmath35 g @xmath36 is the grain bulk density and @xmath37 is the gas density . in writing the above relation a spherical shape with radius @xmath38 is assumed for the grain . we now specify our ansatz for the magnetic field . the vertical component of the field is assumed to take a form @xmath39 where @xmath40 au @xmath41 cm and @xmath42 t is the strength of the field at this radius . the normalization for the distance occurs relative to 1 au . we will also consider more general values of @xmath43 in order to explore the input parameter space . such a power - law relation for the vertical component of the magnetic field has been studied theoretically in self - similar models ( e.g. , shadmehri & khajenabi 2006 ; wu & lou 2006 ) . for example , a distributed @xmath44 field with @xmath45 has been suggested by blandford & payne ( 1982 ) . the paraboloidal field model of blandford ( 1976 ) corresponds to @xmath46 , and the models discussed by ostriker ( 1997 ) corresponds to the range @xmath47 . also , the toroidal component of the magnetic field is ( lovelace , romanova & newman 1994 ) @xmath48 where @xmath49 and @xmath50 is constant of order unity ( @xmath51 ) . the other physical variables of the gas component are @xmath52 @xmath53 @xmath54 where the power - law indices @xmath55 and @xmath5 are generally negative . also , the radial and the rotational velocities of the gas are @xmath56 @xmath57 @xmath58 where @xmath59 is coefficient for the turbulent viscosity ( shakura & sunyaev 1973 ) , and @xmath60 is written as @xmath61 for the overall disc properties , it is often sufficient to work with a disc model in which dust and gas are well mixed and coupled . in our analysis , we adopt the values correspond to a minimum - mass solar nebula : @xmath62 , @xmath63 g @xmath64 , @xmath65 au , @xmath66 , @xmath67 and @xmath68 . having these values the gas mass inside @xmath69 au is @xmath70 @xmath71 . we can obtain approximate analytical solutions for the equations of the motion by applying some simplifying assumptions . from equations ( [ eq : main3 ] ) and ( [ eq : btheta ] ) , we obtain @xmath72 z,\label{eq : vz1}\ ] ] where @xmath73 is , in fact , the _ larmor frequency _ which is defined for describing the motion of a collection of moving charged particles in a region of constant magnetic field . we assume that rotational profiles of the gas and the dust components are close to keplerian , i.e. , @xmath74 . thus , @xmath75 . now , we can rewrite equation ( [ eq : main2 ] ) as @xmath76.\ ] ] in equation ( [ eq : main1 ] ) , we can neglect the left hand side of the equation if @xmath77 . thus , equation ( [ eq : main1 ] ) can be written as @xmath78 @xmath79 = 0.\ ] ] substituting from equation ( [ eq : vz1 ] ) for @xmath80 into the above equation , an algebraic equation for @xmath81 is obtained and so , @xmath82 where @xmath83 @xmath84 we can study profile of the radial velocity of dust particles using equation ( [ eq : vr ] ) . if we substitute this relation into equation ( [ eq : vz1 ] ) , an analytical solution is obtained for the vertical velocity of the dust particles . actually , for neutral dust particles , both parameters @xmath85 and @xmath86 become one and our analytical equations ( [ eq : vz1 ] ) and ( [ eq : vr ] ) reduce to the solutions of tl . the effect of the magnetic field on charged particles appears through larmor frequency @xmath87 and the parameter @xmath50 . however , in our proposed geometry of the magnetic field , we have an order of unity value for @xmath50 . so , dependence of the solutions on this parameter is weak and in our subsequent analysis we assume @xmath88 . but the larmor frequency @xmath87 not only depends on the magnetic strength @xmath89 , but mass and the electrical charge of a dust particle are important . we consider micron - sized grains where the nondimensional stopping time is much smaller than unity through most of the disc . such small particles are well coupled to the gas component . poppe , blum & henning ( 2000 ) demonstrated experimentally that a micron - sized grains may acquire 100 elementary charge . however , in cold protoplanetary discs , grains are either neutral or have 1 to 2 electronic charges . it is very unlikely planetary disc model where grains can carry 100 charges near the mid - plane . it may be possible in the hot hii regions where temperature is very high , but in our analysis we consider 1 electrical charge however , this situation changes significantly as we consider higher vertical locations . for example , for @xmath90 au , the mean grain charge is @xmath91 above @xmath92 to @xmath93 ( wardle 2007 ) . at larger radii , this transition may occur even closer to the midplane , given the drop in fluid density away from the central object . since our solutions reach to @xmath94 , it is plausible that at those locations the typical grain carries more than a single charge . we are interested in regions of the disc where magnetically are active . as we discussed such regions must be beyond dead region which may extend up to @xmath95 au . although we consider the radial distance @xmath96 au in our figures as starting radius of active region , we may simply consider larger radial distances in our analytical solutions . figure [ fig:1 ] shows radial velocity @xmath81 versus vertical distance @xmath97 from the midplane of the disc for different dust sizes ( top plot ) and various exponent @xmath43 of the magnetic field ( bottom plot ) . in both plots the electrical charge of a dust particle is @xmath98 , where @xmath99 is magnitude of the charge of the electron . in the top plot of figure [ fig:1 ] , we fix the exponent @xmath100 . also , solid curves correspond to the magnetic solutions and the dashed curves represent nonmagnetic solutions . for a particle with radius @xmath101 @xmath102 m , the difference between magnetic and the nonmagnetic solutions is negligible . also , as the radius increases , the effect of the magnetic field becomes weaker so that for a particle with radius @xmath96 @xmath102 m , magnetic and nonmagnetic solutions are indistinguishable . but for particles with radii smaller than @xmath101 @xmath102 m , the radial profile of the nonmagnetic solution is represented by a dashed curve which is actually radial profile of the gas component as well . for such small particles , magnetic forces significantly modify profile of the radial velocity . in fact , the radial velocity of charged particles increases and as a particle becomes smaller , the radial velocity increases too . note to the minus sign of the radial velocity . in other words , the radial migration of charged dust particles with radii smaller than critical @xmath103 @xmath102 m significantly increases due to the magnetic force . bottom plot of figure [ fig:1 ] shows radial velocity of a particle with radius @xmath104 @xmath102 m for various exponent @xmath43 . the radial profile highly depends on the radial profile of the magnetic field . as the exponent @xmath43 decreases , the radial migration of a charged grain increases . it implies that the critical radius @xmath105 depends on the exponent @xmath43 as well . while for @xmath100 the critical radius @xmath105 is around 1 @xmath102 m , as the exponent @xmath43 decreases , the critical radius increases which means particles with larger radii are affected by the magnetic force . figure [ fig:2 ] shows the effect of the electrical charge polarity on the radial velocity of the dust grains . solid curves correspond to a positively charged grain with one elementary charge and the dashed curves are for a negatively charged grain . in this plot , the magnetic exponent is @xmath100 . although for grains with radii larger than @xmath105 the effect of magnetic field is negligible , for grains smaller than this critical size the direction of the radial velocity depends on the polarity of grain s charge . outward mass flux is carried primarily by the positively charged grains with radii smaller than the critical radius @xmath105 . note that this result is based on the assumed magnetic field configuration . we considered a negative @xmath50 according to lovelace , romanova & newman ( 1994 ) . top plot of figure [ fig:3 ] shows the vertical velocity @xmath106 of dust particles vs. @xmath107 for various sizes of grains and @xmath100 and @xmath98 . since the radial velocity significantly increases for grains with @xmath108 due to the magnetic force , we expect larger vertical velocity comparing to the nonmagnetic solutions according to equation ( [ eq : vz1 ] ) . although both the radial and the vertical velocities increase due to the magnetic force for grains with radii smaller than @xmath105 , sedimentation of the particles can be determined based on the ratio of the time - scales of the sedimentation and the radial migration . the timescales of the sedimentation and the radial migration are defined as @xmath109 and @xmath110 , respectively . bottom plot of figure [ fig:3 ] shows the ratio @xmath111 vs. distance from the midplane of the disc . solid curves correspond to the magnetic solutions and the dashed curves are for the nonmagnetic solutions . each curve is labeled by the radius of the grain . as the size of grain decreases , the ratio of timescales increases . when the effect of magnetic force becomes significant , the ratio of the timescales is larger than the nonmagnetic case . for example , grains with radius @xmath112 @xmath102 m may sediment before any large migration in the radial direction in the nonmagnetic case . but magnetic field force significantly enhances the ratio of the timescales to values larger than unity which implies large radial migration before sedimentation . note that grains with radii 0.1 , 1 and 10 @xmath102 m have positive radial velocities near to the midplane ( see figure [ fig:1 ] ) . however , as the distance from the midplane increases , the radial velocity tends to zero at @xmath113 . beyond this location the radial velocity is negative . on the other hand , the radial migration timescale is proportional to inverse of the radial velocity . the dip at the curves occurs at the location where the radial velocity tends to zero ( i.e. @xmath113 ) . but for grains with radius @xmath112 @xmath102 m , the radial velocity is negative for all the distances from the midplane . so , there is not a dip in its corresponding curve of the ratio of the timescales . we presented a model for the dynamics of charged grains in a magnetized , protoplanetary disc . properties of the gas component correspond to a minimum - mass solar nebula and the toroidal and the poloidal components of the magnetic field are assumed to be power - law . the vertical and the radial velocities of the charged dust particles depend on the geometry of the magnetic field , charge and mass of the grains . we assumed that each particle is charged one elementary charge . our model illustrates the importance of charge of dust grains in their motions beyond the dead zone . observations of protoplanetary discs , at least in the optical and infrared , mostly probe the upper disc layer . for charged grains with radii smaller than @xmath105 , effect of the magnetic force on their motion is stronger in the upper layers comparing to the midplane region . since magnetic force prevents sedimentation of such grains to the midplane of the disc , one may expect grains with radii larger than a critical radius in midplane region , unless smaller grains are neutral . in a disc resembling to the minimum - mass solar nebula , and assuming that all grains are charged , our model predicts that grains in midplane of the disc are mostly larger than @xmath103 @xmath102 m . dust sedimentation is known to affect the infrared spectra and the images of discs . in particular the far - infrared emission may be reduced by strong sedimentation . although we did not investigate spectra of the disc , it is interesting to study possible modifications to the spectra due to the charge of the grains and the magnetic force . however , in a more realistic model not only the coupling between the field lines and the charged grains should be considered but also the significant effects of magnetic fields on the gas component . takeuchi & lin ( 2003 ) studied the outflow of dust particles in the upper layers of optically disc . they showed that grains in surface layers are moving outward due to the stellar radiation . the outward mass flux is carried primarily by particles of size @xmath104 @xmath102 m according to their calculations . although we neglected the stellar radiation in our model , we expect strong outflows of sub - micron sized particles , if they are charged positively . in model of takeuchi & lin ( 2003 ) , radiation pressure is the main physical factor in making surface outflows of grains . but in our model , it is the magnetic force which drives positively charged grains smaller than @xmath105 to move outward . the actual optical depth along the radial direction in the midplane of the disk is huge , and thus , the radiation pressure from the central star onto the dust grains should be negligible . but neutral grains smaller than the critical radius in the midplane region can move outward once they charged positively . when they reach to the surface layers , not the magnetic force but the radiation pressure contribute to their outward motion . it will be interesting to study surface outflows under combined action of gas drag , the radiation pressure , and the magnetic force in order to understand their relative importance in making surface outflows . \(i ) motion of particles with radii smaller than @xmath105 are affected by the magnetic field , but larger particles move independent of their charge and the magnetic force . for input parameters corresponding to a minimum - mass solar nebula , the critical radius is around 1 @xmath102 m and so , sub - micron sized particles are affected by the field lines . i am grateful to the anonymous referee for suggestions and comments to improve the paper . this research was funded under the programme for research in third level institutions ( prtli ) administered by the irish higher education authority under the national development plan and with partial support from the european regional development fund .	we study the effect of an imposed magnetic field on the motion of charged dust particles in magnetically active regions of a protoplanetary disc . assuming a power law structure for the vertical and the toroidal components of the magnetic field for the regions beyond magnetically dead region of the disc , the radial and the vertical velocities of the charged particles , in the asymptotic case of small particles , are calculated analytically . while grains with radii smaller than a critical radius significantly are affected by the magnetic force , motion of the particles with larger radii is independent of the magnetic field . the critical radius depends on the magnetic geometry and the charge of the grains . assuming that a grain particle has one elementary charge and the physical properties of the disc correspond to a minimum - mass solar nebula , we show that only micron - sized grains are affected by the magnetic force . also , charge polarity determines direction of the radial velocity . for such small particles , both the radial and the vertical velocities increase due to the magnetic force .
the standard model of electroweak interactions ( sm ) so far has met all experimental challenges and is now tested at the @xmath5 level @xcite . however , there is little direct experimental information on the mechanism which generates the masses of the weak gauge bosons . in the sm , spontaneous symmetry breaking is responsible for mass generation . the existence of a higgs boson is a direct consequence of this mechanism . at present the negative result of direct searches performed at lep2 imposes a lower bound of @xmath6 gev @xcite on the higgs boson mass . indirect information on the mass of the higgs boson can be extracted from the @xmath7 dependence of radiative corrections to the @xmath1 boson mass , @xmath8 , and the effective weak mixing angle , @xmath9 . assuming the sm to be valid , a global fit to all available electroweak precision data yields a 95% confidence level upper limit on @xmath7 of 188 gev @xcite . in order to extract more accurate information on @xmath7 from electroweak data , it is very important to measure @xmath8 more precisely . currently , the @xmath1 boson mass is known to @xmath10 mev @xcite from direct measurements . further improvement in the @xmath1 mass uncertainty is expected from this years lep ii data taking , and run ii of the tevatron @xcite which is scheduled to begin in march 2001 . the ultimate precision expected for @xmath8 from the combined lep2 experiments is approximately 35 mev @xcite . at the tevatron , integrated luminosities of order 2 fb@xmath11 are foreseen for run ii , and one expects to measure the @xmath1 mass with a precision of approximately 40 mev @xcite per experiment and decay channel . preliminary studies indicate that measuring @xmath8 at the lhc with an with a precision of 25 mev @xcite per experiment and decay channel should be possible , although very challenging . in order to measure the @xmath1 boson mass with high precision in a hadron collider environment , it is necessary to fully understand and control higher order qcd and electroweak ( ew ) corrections to @xmath1 production . the determination of the @xmath1 mass in a hadron collider environment requires a simultaneous precision measurement of the @xmath2 boson mass , @xmath12 , and width , @xmath13 . when compared to the value measured at lep , the two quantities help to accurately determine the energy scale and resolution of the electromagnetic calorimeter , and to constrain the muon momentum resolution @xcite . it is therefore also necessary to understand the higher order ew corrections to @xmath2 boson production in hadronic collisions . accurate predictions for @xmath1 and @xmath2 production which include the complete @xmath3 corrections are also needed for a measurement of the @xmath1 cross section , the @xmath14 cross section ratio , the determination of the @xmath1 width , and for extracting @xmath9 from the forward backward asymmetry in the @xmath2 peak region . in addition , precise calculations of the drell - yan cross section are needed in searches for new physics beyond the sm , such as hidden extra dimensions or additional @xmath2 bosons . in a previous calculation of the ew radiative corrections to @xmath1 and @xmath2 production , only the final state photonic corrections were correctly included @xcite . the sum of the soft and virtual parts was estimated from the inclusive @xmath15 @xmath16 and @xmath17 ( @xmath18 ) width and the hard photon bremsstrahlung contribution . initial state , interference , and weak contributions to the @xmath3 corrections were ignored altogether . the unknown part of the @xmath19 ew corrections in ref . @xcite , combined with effects of multiple photon emission , have been estimated to contribute a systematic uncertainty of @xmath20 mev to the measurement of the @xmath1 mass in run i @xcite . clearly , in order to achieve the accuracies envisioned for run ii and the lhc , improved theoretical calculations are required . recently , new and more accurate calculations of the @xmath0 ew corrections to @xmath1 @xcite and @xmath2 boson production in hadronic collisions @xcite became available . in this talk i present an overview of these calculations . they include most of the contributions which were previously ignored . in section [ sec : two ] i briefly describe the calculation of the @xmath3 ew corrections to @xmath2 boson and high mass drell - yan production . in section [ sec : three ] i summarize the results of ref . @xcite . in section [ sec : four ] i present a brief summary and outlook . for @xmath22 , the pure qed corrections form a separately gauge invariant set of diagrams . the first step towards a full calculation of the @xmath3 corrections to @xmath2 boson production thus consists of performing a calculation of the pure qed corrections . the diagrams contributing to the @xmath3 qed corrections can be separated into gauge invariant subsets corresponding to initial and final state corrections . to perform the calculation , a monte carlo method for next - to - leading - order ( nlo ) calculations similar to that described in ref . @xcite was used . with the monte carlo method , it is easy to calculate a variety of observables simultaneously and to simulate detector response . the collinear singularities associated with final state photon radiation are regulated by the mass of the leptons . the associated mass singular logarithms of the form @xmath23 , where @xmath24 is the squared parton center of mass energy and @xmath25 is the charged lepton mass , are included in our calculation , but the very small terms of @xmath26 are neglected . the collinear singularities associated with initial state photon radiation can be removed by universal collinear counter terms generated by `` renormalizing '' the parton distribution functions ( pdf s ) @xcite , in complete analogy to gluon emission in qcd . in addition to the collinear counterterms , finite terms can be absorbed into the pdf s , introducing a qed factorization scheme dependence . we have carried out our calculation in the qed @xmath27 and qed @xmath28 scheme . in order to treat the @xmath3 initial state qed corrections to @xmath2 boson production in hadronic collisions in a consistent way , qed corrections should be incorporated in the global fitting of the pdf s using the same factorization scheme which has been employed to calculate the cross section . current fits to the pdf s do not include qed corrections , which introduces a small uncertainty into the calculation . in fig . [ fig : one]a we display the ratio of the @xmath29 and the born cross section as a function of the @xmath30 invariant mass in @xmath31 at tevatron energies . in the region @xmath32 gev , the cross section ratio is seen to vary rapidly . [ cols="^,^ " , ] the iba incorporates the running electromagnetic coupling constant , the @xmath2 propagator expressed in terms of the fermi constant , @xmath33 , and the @xmath2 boson mass and width measured at lep , and the vector and axial vector couplings expressed in terms of the effective leptonic weak mixing angle . the ratio shown in fig . [ fig : three]a directly displays the effect of the weak box diagrams and the energy dependence of the weak coupling form factors . while the additional weak contributions change the differential cross section by 0.6% at most , they do modify the shape of the @xmath2 resonance curve . figure [ fig : three]b compares the effect of the @xmath29 qed corrections and the full @xmath29 electroweak corrections on the di - muon invariant mass distribution at the lhc for @xmath34 values between 200 gev and 2 tev . due to the presence of logarithms of the form @xmath35 , the weak corrections become significantly larger than the qed corrections at large values of @xmath34 , and , eventually , may have to be resummed @xcite . for @xmath36 tev , the full @xmath37 electroweak corrections are found to reduce the differential cross section by more than 20% . the calculation of the @xmath3 corrections to @xmath1 boson production @xcite employs the same monte carlo method which was used in the @xmath2 case . the collinear singularities originating from initial state photon radiation are again removed by counter terms generated by renormalizing the pdf s . calculating the ew radiative corrections to @xmath1 boson production , the problem arises how an unstable charged gauge boson can be treated consistently in the framework of perturbation theory . this problem has been studied in ref . @xcite with particular emphasis on finding a gauge invariant decomposition of the ew @xmath3 corrections into a qed - like and a modified weak part . in @xmath1 production , the feynman diagrams which involve a virtual photon do not represent a gauge invariant subset . in ref . @xcite it was demonstrated how gauge invariant contributions that contain the infrared ( ir ) singular terms can be extracted from the virtual photonic corrections . these contributions can be combined with the also ir - singular real photon corrections in the soft photon region to form ir - finite gauge invariant qed - like contributions corresponding to initial state , final state and interference corrections . the ir finite remainder of the virtual photonic corrections and the pure weak one - loop corrections can be combined to separately gauge invariant modified weak contributions to the @xmath1 boson production and decay processes . since hadron collider detectors can not directly detect the neutrinos produced in the leptonic @xmath1 boson decays , @xmath38 , and can not measure the longitudinal component of the recoil momentum , there is insufficient information to reconstruct the invariant mass of the @xmath1 boson . instead , the transverse mass ( @xmath39 ) distribution of the @xmath40 system , or the @xmath41 distribution of the charged lepton , are used to extract @xmath8 . the @xmath39 distribution for electron and muon final states at the born level and including @xmath3 corrections at the tevatron is shown in fig . [ fig : four]a . the various individual contributions to the ew @xmath3 corrections of the @xmath39 distribution in the electron case are shown in fig . [ fig : four]b . = 2.9 in to model the detector acceptance , the following @xmath41 and @xmath42 cuts were imposed in fig . [ fig : four ] : @xmath43 @xmath44 these cuts are similar to the acceptance cuts used by the dcollaboration in their @xmath1 mass analyses in run i. as before , uncertainties in the energy and momentum measurements of the charged leptons in the detector are simulated in the calculation by gaussian smearing of the particle four - momentum vector . the initial state qed - like contribution uniformly increases the cross section by about 1% . it is largely canceled by the modified weak initial state contribution . the interference contribution is very small . it decreases the cross section by about @xmath45 for transverse masses below @xmath8 , and by up to @xmath46 for @xmath47 . the final state qed - like contribution significantly changes the shape of the transverse mass distribution and reaches its maximum effect in the region of the jacobian peak , @xmath48 . as for the initial state , the modified weak final state contribution reduces the cross section by about @xmath49 , and has no effect on the shape of the transverse mass distribution . since the final state qed - like contribution is proportional to @xmath50 , its size for muons is considerably smaller than for electrons . the initial state corrections and the interference contribution are very similar for electron and muon final states . in fig . [ fig : four ] , we have not taken into account the recombination of electrons and photons if their opening angle is small . as in @xmath2 boson production , when recombination is included , the mass singular logarithmic terms are eliminated . this significantly reduces the size of the ew corrections . figure [ fig : five ] demonstrates the effect for the transverse momentum distribution of the electron in @xmath51 at the lhc . = 2.5 in the solid histogram shows the cross section ratio taking only the transverse momentum and pseudorapidity cuts of eqs . ( [ eq : lepcut ] ) and ( [ eq : ptmisscut ] ) into account . the dashed histogram displays the result obtained when in addition the four momentum vectors are smeared according to the atlas specifications @xcite , and electron and photon momenta are combined if @xmath52 @xcite . qualitatively similar results are obtained for tevatron energies and cdf and d lepton identification criteria . as we have seen , final state bremsstrahlung has a non - negligible effect on the shape of the @xmath39 distribution in the jacobian peak region . as in the @xmath2 boson case , final state photon radiation shifts the @xmath1 boson mass extracted from data to a lower value . in the approximate treatment of the electroweak corrections used so far by the tevatron experiments , only final state qed corrections are taken into account ; initial state , interference , and weak correction terms are ignored . furthermore , the effect of the final state soft and virtual photonic corrections is estimated from the inclusive @xmath15 @xmath16 width and the hard photon bremsstrahlung contribution @xcite . when detector effects are included , the approximate calculation leads to a shift of about @xmath53 mev in the electron case , and approximately @xmath54 mev in the muon case @xcite . since only one of the @xmath1 decay products radiates photons , the shift in @xmath8 is about a factor 2 smaller than the shift in @xmath12 caused by photon radiation . initial state and interference contributions do not change the shape of the @xmath39 distribution significantly ( see fig . [ fig : four]b ) and therefore have little effect on the extracted mass . however , correctly incorporating the final state virtual and soft photonic corrections results in a non - negligible modification of the shape of the transverse mass distribution for @xmath47 . for @xmath1 production at the tevatron this is demonstrated in fig . [ fig : six ] , which shows the ratio of the @xmath39 distribution obtained with the qed - like final state correction part of our calculation to the one obtained using the approximation of ref . @xcite . = 3.in the difference in the line shape of the @xmath39 distribution between the @xmath29 calculation of ref . @xcite and the approximation used so far occurs in a region which is important for both the determination of the @xmath1 mass , and the direct measurement of the @xmath1 width . the precision which can be achieved in a measurement of @xmath8 using the transverse mass distribution strongly depends on how steeply the @xmath39 distribution falls in the region @xmath48 . any change in the theoretical prediction of the line shape thus directly influences the @xmath1 mass measurement . from a maximum likelihood analysis the shift in the measured @xmath1 mass due to the correct treatment of the final state virtual and soft photonic corrections is found to be @xmath55 . for the precision expected in run ii and for the lhc such a shift can not be ignored . two photon radiation has only a modest effect on the shape of the @xmath39 distribution @xcite . detailed simulations will be necessary to determine its effect on the @xmath1 mass . the calculation presented in ref . @xcite was carried out in the pole approximation , ie . the form factors associated with the modified weak corrections were evaluated at @xmath56 . this approximation is valid in the vicinity of the @xmath1 pole . away from the resonance region it will be important to go beyond the pole approximation . calculations which do so are in progress @xcite . preliminary results @xcite indicate that the radiative corrections for @xmath57 gev in a full calculation are up to a factor 2 larger than those calculated in the pole approximation . this may be important for a measurement of the @xmath1 width from the high transverse mass tail . accurate theoretical predictions for @xmath1 and @xmath2 boson production are essential for many important electroweak precision measurements in future hadron collider experiments , in particular the measurement of the @xmath1 mass and width . in addition , comparison of the @xmath2 boson mass and width with the values obtained at lep will help to calibrate detectors . all these measurements require a detailed understanding of the ew radiative corrections . i have described the current status of calculations of the @xmath3 ew corrections to @xmath1 and @xmath2 boson production in hadronic collisions . these calculations will be complete by the time run ii of the tevatron is expected to start . much more work is required to determine the effect of multiple photon radiation on the weak boson masses extracted from hadron collider experiments . this work has been supported by nsf grant phy-9970703 . d. abbaneo _ et al . _ ( the lep electroweak working group ) , cern - ep-2000 - 016 ( january 2000 ) . p. bock _ et al . _ ( the lep working group on higgs boson searches ) , cern - ep-2000 - 055 ( april 2000 ) . a. straessner , talk given at the xxxvth rencontres de moriond , `` electroweak interactions and unified theories '' , 11 18 march 2000 , les arcs , france . h. aihara _ et al . _ , in `` future electroweak physics at the fermilab tevatron : report of the tev_2000 study group '' , edited by d. amidei and r. brock , fermilab - pub-96/082 , 1996 , p. 63 . charlton , hep - ex/9912019 , to be published in the proceedings of the `` 19th international symposium on lepton and photon interactions at high - energies '' , stanford , california , 9 14 august 1999 . s. haywood _ et al . _ , hep - ph/0003275 , proceedings of the `` 1999 cern workshop on standard model physics ( and more ) at the lhc '' , cern yellow report cern-2000 - 004 , eds . g. altarelli and m. mangano . f. abe _ et al . _ ( cdf collaboration ) , phys . * 75 * , 11 ( 1995 ) and phys . * d52 * , 4784 ( 1995 ) ; t. affolder _ et al . _ ( cdf collaboration ) , hep - ex/0007044 ( july 2000 ) ; s. abachi _ et al . _ ( d collaboration ) , phys . * 77 * , 3309 ( 1996 ) ; b. abbott _ et al . _ ( d collaboration ) , phys . rev . * d58 * , 012002 ( 1998 ) , phys . rev . * d58 * , 092003 ( 1998 ) and phys . * 80 * , 3008 ( 1998 ) . [ ber ] f.a . berends and r. kleiss , z. phys . * c27 * , 365 ( 1985 ) . u. baur , s. keller and d. wackeroth , phys . rev . * d59 * , 013002 ( 1999 ) . [ saku ] u. baur , s. keller and w.k . sakumoto , phys . rev . * d57 * , 199 ( 1998 ) . u. baur , o. brein , w. hollik , c. schappacher and d. wackeroth , in preparation . h. baer , j. ohnemus , and j.f . owens , phys . rev . * d40 * , 2844 ( 1989 ) ; w. giele and e.w.n . glover , phys . * d46 * , 198 ( 1992 ) . j. kripfganz and h. perlt , z. phys . * c41 * , 319 ( 1988 ) ; h. spiesberger , phys . rev . * d52 * , 4936 ( 1995 ) ; a. de rujula _ et al . b154 * , 394 ( 1979 ) . u. baur and t. stelzer , phys . rev . * d61 * , 073007 ( 2000 ) . p. ciafaloni and d. comelli , phys . lett . * b446 * , 278 ( 1999 ) . j. h. khn and a. a. penin , hep - ph/9906545 ; j.h . khn , a.a . penin and v.a . smirnov , hep - ph/9912503 and hep - ph/0005301 ; v. s. fadin , l. n. lipatov , a. d. martin and m. melles , phys . rev . * d61 * , 094002 ( 2000 ) ; m. melles , hep - ph/0004056 and hep - ph/0006077 ; w. beenakker and a. werthenbach , hep - ph/0005316 . w. hollik and d. wackeroth , phys . rev . * d55 * , 6788 ( 1997 ) . atlas collaboration , `` atlas detector and physics performance '' , cern / lhcc/99 - 15 ( 1999 ) . u. baur and d. wackeroth , in preparation ; s. dittmaier and m. krmer , in preparation .	we discuss the @xmath0 electroweak radiative corrections to @xmath1 and @xmath2 boson production and their impact on the measurement of the @xmath1 mass at hadron colliders . the results of recent improved calculations are presented . we also briefly discuss the @xmath3 corrections to drell - yan production in the high invariant mass region . 1000 1000 @xmath4 \end{array } $ ]
recently there has been considerable interest in understanding the dynamics of systems of active interacting biological agents ( active systems , for short ) , such as flocking birds , schooling fish , swarming bacteria , etc @xcite . properties of these strongly self - organizing dissipative systems are of fundamental interest to nonequilibrium statistical dynamics @xcite and to potential technological applications @xcite . common swimming ( motile ) bacteria , such as _ bacillus subtilis _ , _ escherichia coli _ , and many others , are rod - shape microorganisms ( length about 5 @xmath0 , diameter of the order 1 @xmath0 ) , propelled by the input of mechanical energy at the smallest scales by the rotation of helical flagella attached to the cell wall . suspensions of swimming bacteria ( active suspensions ) are a convenient representative of self - organizing biological systems . at relatively high filling fractions of bacteria , they interact mostly through hydrodynamic entrainment induced by their swimming with respect to ambient fluid @xcite , which are instrumental in the establishment of long - range order . at the same time , they are amenable to accurate experimental studies , control , and manipulation @xcite . in the dense regime , the dynamics of active suspensions is dominated by the multiple body interactions and long - range self - organized coherent structures , such as recurring whorls and jets with the spatial scale exceeding the size of individual bacterium by an order of magnitude . this , however , makes the analysis rather challenging . in the dilute regime , the hydrodynamic interactions between bacteria ( or passive particles ) are often ignored as the interparticle distance exceeds the range of the flows resulting from the particle motion . therefore , in this case it is possible to isolate the effect of the particle - fluid interactions on the effective properties of active suspensions . this constitutes a step towards the full understanding and utilization of the novel properties of active suspensions . the first step was taken in @xcite by einstein , who derived an explicit formula for the effective viscosity of a dilute suspension of passive spheres . this formula shows an increase of the viscosity over that of the ambient fluid alone . the correction to the effective viscosity is to the first order in the volume fraction of the inclusions . later , batchelor and greene obtained the second order asymptotic formula , which takes into account pairwise interactions between particles @xcite . jeffrey calculated the effective viscosity of a suspension of ellipsoidal particles to first order in @xcite . unlike the spherical case , here the value of the viscosity is affected by the distribution of orientations of the inclusions , which is assumed to be uniform . in this work , einstein s classical dilute limit result is extended to the case of _ self - propelled _ disk inclusions in a 2d stokesean fluid . the choice of two - dimensional hydrodynamics is motivated by the tractable nature of calculations involving green s functions in 2d and also by the quasi - two - dimensional thin film geometry of the experiment @xcite . in particular , the correction to the effective viscosity is explicitly calculated as a function of the orientations of the bacteria s flagella , the intensity of the their force of self - propulsion , and the volume fraction of the suspension . in the case of an absence of self propulsion and , alternatively , the case of a uniform distribution of orientations , the result recovers the 2d version of einstein s result ( see , e.g.@xcite ) . the model of the bacterium and the corresponding hydrodynamics are described in section [ sec - model ] . following batchelor @xcite , the effective viscosity of a dilute suspension is defined in section [ sec - model ] , in terms of a suitable background flow and the disturbance flow produced by the inclusion of a bacterium in the background flow . in the case of self - propulsion the disturbance flow is due both to the passive response of the background flow to the inclusion of a particle as well as the response of the stationary fluid to the particle s self - propulsion . because of the linearity of stokesean hydrodynamics , the two components of the disturbance flow can be computed independently using the green s function for the disk . the disturbance flow @xmath1 due to the locomotion of a single bacterium is calculated in section [ sec - flow ] . this flow translates the bacterium . it is also shown that this disturbance flow decays at infinity . this result is emphasized since it is a point of well - known deficiency of 2d hydrodynamics . indeed , while the translation of a ball in 3d produces no flow at infinity , in the well - known stokes paradox , the 3d flow due to an infinite cylinder ( described by the corresponding force monopole ) translating transversely to its axis generates a non - zero flow at infinity ( see , e.g. , @xcite ) . thus prevents the decoupling of rods , which is needed for the dilute limit , where it is assumed that the flow due to a suspension can be approximated by the sum of solutions due to a single inclusion however , a self - propelled particle is constrained by the viscous drug force opposing the propulsion force , thus producing a force dipole which decays at infinity , unlike the force monopole in the case of the moving disk . for the other disturbance flow , @xmath2 , due to the bacterium s passive response to a background flow , care must be taken to ensure that no flow is produced at infinity . thus , in section [ sec - ev ] such a flow is selected to ensure the disturbance decays at infinity . once it has been done , the effective viscosity is calculated as a function of the flagella s orientations relative to the background flow . the time - dependent nature of the effective viscosity , due to active alignment of the inclusions to the flow , is also discussed . conclusions are discussed in section [ sec - conclusion ] . finally , the details of calculations are discussed in the appendices . a 2d dilute suspension of bacteria is modeled as a collection of discs of radius @xmath3 , each of which has an associated point force , representing the flagellum , placed at a distance @xmath4 from the disk , as shown in figure [ orientation ] . the point force is directed radially outward from the center of the bacterium and has some orientation angle @xmath5 , measured from the @xmath6-axis . the bacteria are distributed throughout an ambient fluid of viscosity @xmath7 which takes up the entire domain @xmath8 . a single bacterium in an infinite fluid moves with respect to the fluid with constant velocity driven by the point force @xmath9 . the suspension is assumed to contain sufficiently many bacteria so as to produce appreciable changes in the properties of the equivalent homogenized fluid . at the same time , the size of the bacteria and the propulsion force are assumed to be sufficiently small , so that the disturbance flow produced by bacteria is negligible at inter - bacterial distances and hence can be ignored at the locations of other bacteria . therefore , in the following calculations , it is sufficient to consider a single bacterium with orientation @xmath5 in an unbounded volume of fluid . this is analogous to einstein s assumptions for passive suspensions . this unbounded volume is an idealization of the microscopic volume element surrounding a single one of the particles located within a macroscopic fluid element . because of the decay assumption , this idealization is permissible for the solution of the stokes equation within that region . the viscous dissipation in the macroscopic fluid element is then the sum of the dissipations in all of the subordinate microscopic volumes , containing a collection of particles having some prescribed distribution of orientations . analogous to the method of batchelor in @xcite , the effective viscosity @xmath10 of a suspension in an ambient fluid of viscosity @xmath7 is defined as the viscosity of an equivalent fluid with no inclusions that produces the same energy dissipation in each macroscopic volume element . for a suspension of passive particles , the existence of an equivalent newtonian fluid with the scalar effective viscosity @xmath10 is a classical result of homogenization theory ( see e.g. , @xcite and references therein ) . the analysis for active particles is more subtle and has not been carried out in a rigorous mathematical context . the assumption is made that the effective viscosity can be defined in the same way for active particles . nevertheless , the technique is presented somewhat cryptically in the literature . to clarify the main points , the conceptual side of the calculation is presented here in some detail ; the technical details can be found in [ app - def_ev ] ( also see @xcite for the derivation in the case of passive particles ) . . ] since the effective viscosity is defined in terms of energy dissipation , a background flow ( which describes the flow throughout the homogenized fluid and on the boundary of the suspension ) is chosen so that it has a non - zero rate of energy dissipation ( i.e. , requiring a non - zero strain rate ) . choosing @xmath8 to represent the microscopic volume with coordinates @xmath11 the velocity components of the background flow are @xmath12 , where @xmath13 is constant and symmetric strain rate tensor but otherwise left to be specified later . adding a single self - propelled disk inclusion @xmath14 somewhere in the plane produces the disturbance flow @xmath15 . provided that @xmath16 is selected properly , @xmath15 vanishes at infinity , decoupling the bacteria , as required of the dilute limit . the disturbance flow can be computed explicitly and it is done below . the total flow of the suspension is @xmath17 and in principle it should be possible now to compare the dissipation rates of @xmath16 and @xmath18 by integrating over @xmath8 and @xmath19 respectively . as shown in @xcite , the integration over the unbounded volume @xmath8 leads to technical difficulties . this motivates the more efficient method of comparing the dissipation rates within a bounded domain @xmath20 which contains sufficiently many bacteria . figure [ fig - omega ] shows such a domain encompassing two bacteria . inside @xmath20 are the regions representing the bodies of the bacteria , @xmath21 and , for each , an @xmath22-disk @xmath23 around the corresponding point force . the disturbance flow in the domain @xmath24 is required to vanish at the outer boundary @xmath25 . [ fig - omega ] for which the modified dissipation rate is computed ; includes the complement of two @xmath3-disks ( bacteria ) and @xmath22-disks ( neighborhoods of point forces).,title="fig : " ] this models a finite container with the background flow applied at the boundary and simplifies the computation of the dissipation rate integrals by replacing a volume integral with an integral over the boundaries of the bacteria . calculating the disturbance flows @xmath26 , however , involves solving a boundary value problem in the two - dimensional domain @xmath24 . to avoid this , the total dissipation rate in this domain can be equivalently calculated as the integral of the work of forces at the boundary @xmath27 . by requiring that @xmath28 vanish at @xmath29 , the flow of the homogenized fluid and total flow of the suspension agree on the boundary @xmath30 , @xmath31 , but the corresponding values of the stress tensors @xmath32 and @xmath33 are not necessarily the same . now , the density of work at the boundary is @xmath34 and @xmath35 for the homogenized and total suspension flows respectively . the effective viscosity is defined by integrating these quantities over @xmath30 and setting them equal : @xmath36 using the stokes equation @xmath37 and repeatedly applying the divergence theorem ( see appendix [ app - def_ev ] ) , the integrals of these densities over @xmath27 can be reduced to integrals over the boundary of the bacterial domain only : @xmath38 . these integrals are independent of @xmath27 and upon passing to the limit @xmath39 they are expressed in terms of the disturbance flow @xmath15 vanishing at infinity . the effective viscosity is then determined from the expression resulting from summing over all inclusions : @xmath40 applying the dilute assumption , the solution to ( [ stokesfull ] ) is replaced by a superposition of solutions for the flow due to a single bacterium , derived in the next section . thus , for aligned bacteria , the above integral can be replaced with one over the boundary of one bacterium in @xmath8 , @xmath41 where @xmath42 is the volume fraction occupied by bacteria . let @xmath43 be the exterior of the unit disk in @xmath8 and @xmath44 be the dirac delta evaluated at @xmath45 , where @xmath46 is the location of the point force . in the low reynolds number limit , the flow due to the bacterium obeys the stokes equation @xmath47 where @xmath48 is the force strength and @xmath49 . for simplicity of calculations , set @xmath50 and @xmath51 . note that here , @xmath52 is set to @xmath53 on @xmath30 . this is the reference frame in which the bacterium is moving for the purpose of finding the total flow of the suspension , the reference frame in which the fluid is at rest at infinity must be used . by writing @xmath16 as the curl of a stream function @xmath54 and taking the divergence of ( [ stokes ] ) , this reduces to the inhomogeneous biharmonic equation @xmath55 the green s function @xmath56 for the domain @xmath43 with these boundary conditions is known and yields the solution @xmath57 explicitly , in polar coordinates , @xmath58 . \nonumber \ ] ] it is easy to see that @xmath59 as @xmath60 . subtracting this flow ( moving to the reference frame where the water is at rest at @xmath61 ) and rescaling to a disk of radius @xmath3 , is shown in figure [ velocity ] and is given asymptotically by @xmath62 now , by taking the divergence of ( [ stokes ] ) , one obtains @xmath63 the solution to ( [ pressurepde ] ) is found using complex function theory ( see appendix [ app - pressure ] ) , and is the real part of the complex function @xmath64 in order to calculate the effective viscosity , it remains to perform the integration in ( [ evintegral ] ) . the effective viscosity is independent of the background flow , so one is free to choose the flow which simplifies the necessary calculations . in ( [ evintegral ] ) , @xmath15 and @xmath65 are the solutions to @xmath66 subject to the boundary conditions @xmath67 note that , as mentioned in section [ sec - flow ] , one is not free to choose the value of the constant that @xmath16 takes on @xmath30 . the formula for the effective viscosity does not depend on the choice of background flow ( i.e. , the choice of the strain rate tensor @xmath13 in ( [ bgflow - bcs ] ) ) . however , the choice of this flow is not completely arbitrary . first , since the effective viscosity is defined in terms of energy dissipation , a non - zero rate of strain is required so that there is energy being dissipated due to viscosity . second , the background flow is chosen so that it does not rotate bacteria , which allows for the computation of the effective viscosity for a fixed ( time - independent ) distribution of orientations of bacteria ( see figure [ orientation ] ) . third , upon placing a disk ( i.e. , bacterium ) in the flow , it is necessary that the disturbance flow @xmath68 as @xmath60 . if this were not the case , the dilute assumption that this flow is negligible at the locations of other bacteria would be violated . in 2d , this is only possible for a background flow that satisfies the conditions for existence of such a decaying flow derived in @xcite . the background flow is chosen to be the simplest flow that matches these criteria the sum of two perpendicular shearing flows , @xmath69 where @xmath70 is the amplitude of the strain rate . the disturbance flow due to the presence of a bacterium will be @xmath71 with the pressure given by @xmath72 where , using notation , @xmath73 as a polar vector and @xmath74 , rotated to allow for an arbitrary orientation of a bacterium in the background flow . for a suspension of bacteria that are not fully aligned , the effective viscosity can be written in terms of the function @xmath75 which gives the probability density of orientations of the bacteria @xmath76 , which are assumed to be independent identically distributed random variables . in this case , ( [ evintegral ] ) becomes @xmath77 where @xmath42 is the volume fraction occupied by bacteria . performing the integration over @xmath78 , one obtains the following formula for the effective viscosity : @xmath79 , \nonumber\end{aligned}\ ] ] where @xmath42 is the two - dimensional volume fraction . note that for isotropically oriented bacteria ( @xmath80 ) , the angular term goes to zero and the effective viscosity reduces to the result in @xcite for the effective viscosity of a two - dimensional dilute suspension of disks , @xmath81 nevertheless , in experiments a reduction in viscosity is anticipated @xcite . in reality , bacteria are not spherical . an interaction with the background flow thus produces a torque on the bacteria which will tend to align them in particular preferred directions ( see e.g. @xcite ) . for a fully aligned suspension of bacteria , one obtains , from ( [ evformula ] ) , @xmath82.\ ] ] for the background flow ( [ bgflow ] ) , the corresponding preferred directions are @xmath83 . thus , after aligning , @xmath84.\ ] ] this formula can be used to explain the experimental data on the reduction of viscosity in bacterial suspensions observed in @xcite . here a rather puzzling result has been obtained : the effective viscosity of active bacterial suspension _ may decrease _ with the increase in volume fraction of particles . moreover , the viscosity may even become formally negative ( when the flagella add more energy than is being dissipated ) if the volume fraction @xmath42 and the magnitude of the propulsion force @xmath85 exceed critical values . manifestations of negative viscosity have been observed in experiments @xcite via large - scale instability and the formation of non - decaying whorls and jets of collective locomotion ( see also simulations @xcite for dense suspensions ) . the reduction of effective viscosity can be interpreted as a result of transformation by swimming bacteria of chemical energy of the surrounding nutrient medium into mechanical energy of fluid motion , and thus replacing energy loss due to viscous dissipation . it is also notable that the coefficient @xmath86 , which characterizes the strength of the background flow , shows up in the formula for effective viscosity ( unlike in a newtonian fluid ) . thus , the additional term due to self - propulsion is inherently non - newtonian . since @xmath86 appears in the denominator , there is a blow up in the effective viscosity as @xmath87 . nevertheless , this is to be expected , as the viscosity is , roughly , the ratio of stress of the entire flow to rate of strain of the background flow in the case of a bacterial suspension , as the rate of strain of the background flow goes to zero , the stress does not , since the bacteria will continue swimming . we also comment that the limit of vanishing background flow @xmath88 is not trivial . as mentioned above , the reduction of viscosity occurs only when the bacteria become aligned in a certain direction determined by the principal axis of the strain - rate tensor . however , when @xmath87 , the bacteria no longer have a tendency to align ( also , a preexisting aligned state would be destroyed by fluctuations , e.g. due to random tumbling of bacteria ) and hence the third term in eq . ( [ evformula ] ) does not blow up . thus , in the limit of @xmath88 , the steady state is in fact the isotropically oriented suspensions @xmath89 , which recovers the classical result for disks eq . ( [ eq : disks ] ) . moreover , since the effective viscosity for small strain rates is determined by the interplay between the response of the bacterium , through its orientation , to the shear strain and random fluctuations , one needs to solve self - consistently the equation for the orientation distribution function @xmath90 ) in the presence of non - zero strain rate . this isotropic orientation can also explain the results of recent simulations @xcite in which no reduction of the effective viscosity due to swimming bacteria ( modeled by spherically symmetric particles ) was detected : the reduction of viscosity requires particle reorientation by shear flow . however , the orientation of spherical particles is not affected by shear flow . in contrast , the effect is expected for elongated particles ( ellipsoids or cylinders with large aspect ratios ( of the order of 1:5 for swimming bacteria ) . obviously , more dedicated controlled experiments with bacterial suspensions and further generalizations of the obtained result to more general flow geometries , such as three - dimensional films and slabs , are keenly needed . the puzzling phenomenon of reduced viscosity in bacteria - laden fluids may find rather unexpected technological applications in bio - medical research and chemical technology , such as microscopic bacterial mixers and chemical reactors @xcite . * acknowledgments * the work of l. berlyand was partially supported by nsf grant dms-0708324 . aranson and d. karpeev were supported by us doe , grant doe grant de - ac02 - 06ch11357 . following @xcite , the rate of work being done at the boundary @xmath49 is given by @xmath91 where @xmath92 is the stress tensor , @xmath93 is the velocity of the fluid , @xmath94 is the unit outward normal for the surface @xmath30 , @xmath95 is the pressure of the fluid , and @xmath96 is the rate of strain tensor . henceforth , primed quantities will denote the disturbance values due to the presence of a bacterial suspension . the new stress tensor can be expressed as @xmath97 . additionally , on @xmath30 , @xmath98 . thus , the effective viscosity of the suspension @xmath99 is defined by setting @xmath100 noting that the terms involving @xmath95 are identical on both sides and employing the divergence theorem yields @xmath101 where @xmath102 is the area of the surface @xmath30 . the right hand side of ( [ defev ] ) is the additional rate of dissipation due to the suspension . employing the divergence theorem yet again , this integral is transformed into an integral over the surfaces of the particles , producing @xmath103 where @xmath104 is the volume occupied by a single bacterium , @xmath105 is a ball of radius @xmath106 around its corresponding point force , and the summation is taken over all bacteria inside @xmath43 . now , the fluid in @xmath43 obeys the inhomogeneous stokes equation @xmath107 where the subscript @xmath108 has been added to all quantities that can vary among the bacteria . in particular , @xmath109 is its velocity of the @xmath108th bacterium , @xmath110 is the location of its point force , and @xmath111 indicates the strength and orientation of each point force . thus , @xmath112 in @xmath113 . additionally , @xmath114 since the surface integral over @xmath30 vanishes . applying these observations to ( [ defev ] ) produces @xmath115 . recall that the velocity field @xmath16 solves the stokes equation , @xmath116 for simplicity , it is assumed that @xmath117 and @xmath118 . since @xmath119 , the velocity can be expressed as the ( 2d ) curl of a scalar stream function @xmath120 . this curl , which operates on scalar functions and produces a vector function , is defined by @xmath121 the scalar curl @xmath122 , which operates on vectors , is defined as @xmath123 so that @xmath124 . substituting @xmath125 into ( [ stokes ] ) and taking the scalar curl of both sides yields the inhomogeneous biharmonic equation @xmath126 with boundary conditions @xmath127 the green s function for the domain @xmath43 with these boundary conditions is derived in the exact same fashion as that for the unit disc , and has a simple form as a function of the complex variables @xmath128 and @xmath129 , taken from @xcite : @xmath130 this yields the solution formula @xmath131 and hence @xmath57 this derivation follows that in @xcite , with the only difference being that , in this case , the domain is the outside of the unit disk @xmath132 . nevertheless , the mathematical details are identical . the fundamental solution of the biharmonic equation is @xmath133 and the general solution of the biharmonic equation is @xmath134 where @xmath54 and @xmath135 are arbitrary analytic functions . thus , the problem is to find a function of the form ( [ generalsolution ] ) that cancels ( [ fundamentalsolution ] ) and its normal derivative on @xmath136 . to facilitate this , ( [ generalsolution ] ) can be equivalently written as @xmath137\phi(z)+\psi(z)\right\}.\ ] ] let @xmath138 . then , the condition @xmath139 on @xmath136 is equivalent to @xmath140 since @xmath141 there . since the equation inside the braces is analytic in the unit disk , it must be @xmath142 . additionally , the condition that @xmath143 on @xmath136 implies @xmath144 there , and so @xmath145 substituting @xmath135 yields @xmath146 and hence , using @xmath141 on @xmath136 once more , @xmath147 thus , @xmath148 by taking the divergence of ( [ stokes ] ) , one obtains @xmath149 thus @xmath150 is the real part of some function @xmath151 , which is holomorphic in @xmath152 , such that @xmath153 on @xmath30 . in fact , viewing @xmath154 and @xmath155 as functions of @xmath6 and @xmath156 , @xmath157 d\zeta , \nonumber\end{aligned}\ ] ] where @xmath158 , @xmath159 , and @xmath160 is any path from the origin to @xmath161 . it is clear that @xmath162 satisfies the compatibility condition and is harmonic , except where the integrand has singularities , so it remains to check that @xmath163 . performing the integration in ( [ pressureintegral ] ) gives @xmath164 taking the real part of ( [ analyticpressureappendix ] ) and expanding it about @xmath165 yields @xmath166 thus , indeed , @xmath167 . in terms of the stream function of the background flow , @xmath168 , the stream function for the disturbance flow @xmath169 , from @xcite , is given by @xmath170 . \nonumber \ ] ] this yields the velocity @xmath171 the corresponding pressure @xmath172 is , once more , the real part of @xmath173 d\zeta . \nonumber\end{aligned}\ ] ] performing this integration gives @xmath174 widest - label x .- l . wu and a. libchaber , lett . * 84 * , 3017 ( 2000 ) . kim and k.s . breuer , phys . fluids , * 16 * , 78 ( 2004 ) n.h . mendelson _ et al . _ , j. bacteriol . * 181 * , 600 ( 1999 ) . c. dombrowski _ et al . lett . * 93 * , 0980103 ( 2004 ) . riedel , k. kruse , and j. howard , science * 309 * , 300 ( 2005 ) . becco _ et al . _ , physica ( amsterdam ) * a367 * , 487 ( 2006 ) ; j. buhl _ et al . _ , science * 312 * , 1402 ( 2006 ) . t. feder , physics today , p. 28 , oct 2007 j. toner and y. tu , phys . lett * 75 * , 4326 ( 1995 ) . g. grgoire and h. chat , phys . lett . * 92 * , 025702 ( 2004 ) . t. vicsek , a. czirk , e. ben - jacob , i. cohen , and o. shochet , phys . lett * 75 * , 1226 ( 1995 ) ; a. czirk , h.e . stenley , and t. vicsek , j. phys . * a 30 * , 1375 ( 1997 ) . d. grossamn , i. s. aranson , and e. ben - jacob , new j. phys . * 10 * 023036 ( 2008 ) r.a . simha and s. ramaswamy , phys . lett . * 89 * , 058101 ( 2002 ) . kim and k.s . breuer , jour . fluids engin . * 129 * , 319 ( 2007 ) . a. sokolov , i. s. aranson , j. o. kessler , r. e. goldstein , phys.rev.lett . * 98 * , 158102 ( 2007 ) a. einstein _ investigations on the theory of the brownian movement _ , dover publications , new york , 1956 g.k . batchelor and j.t . green , j. fluid mech . * 56 * , 401 - 427 ( 1972 ) . jeffery , r. soc . london ser . a * 102 * , 161 - 79 ( 1922 ) . m. belzons , r. blanc , j.l . bouillot , and c. camoin , c.r . paris serie ii * 292 * , 939 - 44 ( 1981 ) . batchelor , _ an introduction to fluid dynamics _ , cambridge university press , cambridge , 1967 . t. levy , e. sanchez - palencia , j. non - newt . fluid mech . * 13 * ( 1983 ) 63 - 78 . landau and e.m . lifshitz , _ fluid mechanics _ , elsevier , oxford , 1987 . a. avudainayagam and b. jothiram , q. j. mechanics appl . math . * 41 * , 383 - 93 ( 1988 ) . a. sokolov , i.s . aranson , in preparation . pedley and j.o . kessler , annu . fluid mech . * 24 * , 313 - 58 ( 1992 ) . r. garabedian , _ partial differential equations _ , john wiley & sons , new york , 1964 . hernandez - ortiz , ch . g. stoltz , and m. d. graham , phys . * 95 * , 204501 ( 2005 ) . t. ishikawa and t. j. pedley , phys . rev . lett . * 100 * , 088103 ( 2008 ) t. ishikawa and t. j. pedley , j. fluid mech . * 588 * , 399 ( 2007 )	suspensions of self - propelled particles are studied in the framework of two - dimensional ( 2d ) stokesean hydrodynamics . a formula is obtained for the effective viscosity of such suspensions in the limit of small concentrations . this formula includes the two terms that are found in the 2d version of einstein s classical result for passive suspensions . to this , the main result of the paper is added , an additional term due to self - propulsion which depends on the physical and geometric properties of the active suspension . this term explains the experimental observation of a decrease in effective viscosity in active suspensions .
the resolution of the majority of the extragalactic submillimetre ( submm ) background into individual sources ( blain et al . 1999 ) has made it possible , in principle , to study the redshift distribution of these sources and hence investigate the history of obscured star formation in the universe . such investigations are important to compare the star formation which occurs in obscured sources with that which takes place apparently unobscured , e.g. , in the lyman break galaxies identified by steidel , pettini & hamilton ( 1995 ) . although the similar strengths of the extragalactic optical and submm backgrounds suggest that obscured and unobscured star formation occur in roughly equal quantities ( hauser et al.1998 ) , the more detailed relationship between the populations of galaxies within which these processes occur is unclear . are they coeval populations , or are each of them the dominant mode of star formation at a particular cosmic epoch ? spectroscopy of submm - selected sources has proven difficult and time - consuming , and has led to the emergence of an industry to estimate redshifts using template seds which cover the wavelength range from the far - infrared to the radio . since the initial work by carilli & yun ( 1999 , 2000 ) , the technique has been adapted by dunne , clements & eales ( 2000 ) , rengarajan & takeuchi ( 2001 ) , yun & carilli ( 2002 ) , and hughes et al . often these estimates disagree strongly , however ( see , e.g. , section 5 of this paper ) , due to differences in the assumed spectral energy distribution templates , and this only increases the need for spectroscopic measurements . there are two main issues which make it difficult to obtain spectroscopy of these objects . the first of these arises because the spatial resolution of the submillimetre common user bolometer array ( scuba ) is rather poor ( @xmath1 fwhm ) , leading to a number of optical / infrared counterparts within the source error circle . the case of hdf 850.1 , the brightest submm source in the hubble deep field , serves as a good example ( hughes et al . 1998 ; downes et al . 1999 ; dunlop et al . 2004 ) . to obtain an accurate position requires sensitive observations with an interferometer at millimetre or centimetre wavelengths in order to achieve the arcsecond resolution necessary to make an unambiguous identification . the second difficulty arises after an accurate position has been determined . in many cases , the correct identification is extremely faint and may require deep imaging observations to be identified . ivison et al.(2002 ; hereafter i02 ) report the optical / ir continuum characteristics of a relatively bright submm sample , having refined most of their positions via deep radio imaging . a large fraction was shown to be sufficiently bright for spectroscopy on large telescopes , as later demonstrated using the lris - b uv / optical spectrograph on the keck telescope ( chapman et al.2003 ) . the tendency for the optical / ir counterparts to be diffuse and/or multiple ( `` composite red / blue galaxies '' ; i02 ) means that the fraction of light transmitted by a spectroscopic slit is low compared to , say , a lyman - break galaxy of similar magnitude . continuum magnitudes are not a reliable indicator of likely success , however , if targets display strong emission lines , as many have been shown to do , including several with @xmath2 ( we use vega magnitudes throughout ) . nevertheless , the ability to detect continuum emission with a reasonable signal - to - noise ratio is essential if emission lines are weak or lie outside the available wavelength coverage , and conclusions about the underlying stellar population can only be drawn if there is sufficient signal - to - noise ratio in the continuum to detect breaks and/or stellar absorption feautures . given the optical faintness and frequent red colours of these sources , sensitive near - infrared spectroscopy may prove to be a fruitful avenue for sources where optical spectroscopy has proven unsuccessful . in this paper , we present the results from deep near - ir spectroscopy of a sample of submm galaxies , made with the extremely sensitive ohs instrument on subaru telescope . the format is as follows . in section 2 , we describe the selection of our sample and our observations , and in section 3 , we detail the careful data reduction procedures required to extract maximum sensitivity from these faint objects . section 4 presents our results , while in section 5 we describe the spectra of individual objects in detail . we discuss our results in section 6 , and speculate on future prospects for obtaining complete spectroscopic redshifts for submillimetre samples in section 7 . finally , we summarize our results and conclusions in section 8 . throughout this paper , we adopt @xmath3 , @xmath4 , and @xmath5 . our sample of targets was taken from the scuba 8-mjy survey of the lockman hole east and elais n2 fields ( scott et al . 2002 ; fox et al . 2002 ) and the canada uk deep submillimetre survey ( cudss ) 14-hour field ( eales et al . 2000 ; webb et al . we selected only targets which had been securely identified through follow - up radio observations which were able to provide sufficient astrometric accuracy to isolate a single object radio peaks makes the identification reasonably secure . ] . furthermore , we selected targets for which the identification had very red optical / ir colours ( see i02 ) , so as to exclude galaxies for which optical spectroscopy might be expected to be more successful . the sample is therefore by no means complete , and was constructed to test the limits of the most sensitive near - infrared spectrograph currently available on any telescope and assess the likely success of future ir spectroscopy . galaxies were observed with the oh - airglow suppressor ( ohs ; iwamuro et al.2001 ) and the cooled infrared spectrograph and camera for ohs ( cisco ; motohara et al . 2002 ) on subaru telescope during the period ut 2002 may 2024 . ohs is a spectroscopic filter which eliminates 224 strong oh airglow lines in the @xmath6 and @xmath7 bands from a @xmath8 region of sky and outputs an undispersed beam which is then fed to cisco to enable images or spectra to be taken . the broad - band sky background in this suppressed region ( the ` dark lane ' ) is reduced by a factor of @xmath9 , enabling the acquisition of extremely faint targets . conditions were clear for all the observations presented here , and the optical seeing was measured to be 0.50.7arcsec from the auto - guider camera . each target was observed in an identical manner . first , a nearby offset star was acquired and centred in the oh - suppressed ` dark lane ' , using cisco in imaging mode with an @xmath7-band filter . the telescope was then moved according to the offset between this star and the target , as measured from existing near - ir images . the location of the target on the array was then determined by nodding the telescope along the slit and taking images at each position , the number and length of which depended on the expected magnitude of the target . the images were combined to remove the sky background and allow the target to be seen . if the target was not within the dark lane , an appropriate telescope offset was applied and the process repeated until an accurate centroid could be determined . a small offset was then applied using the autoguider camera to move the target into the centre of the dark lane . the spectrograph slit was then closed to 1arcsec and aligned with the dark lane , and the _ jh _ grism inserted into the light path . eight exposures were then taken in two quads using the standard ` abba ' technique ( e.g. eales & rawlings 1993 ) , with exposure times of 900 or 1000seconds per position . the offset between positions was usually 10arcsec , although we chose to align the slit to include any nearby objects and a smaller offset was used if the the standard value would cause confusion between the beams . in some cases , a small offset along the slit was applied between the two quads . when these exposures had finished , the telescope was slewed to a nearby atmospheric ratioing standard which was acquired without opening the slit by making small telescope movements perpendicular to the slit position angle until the count rate from the star was sufficiently high . the stars used were hip 52256 , hip 71078 , and hip 80419 ( perryman et al . 1997 ) for the lockman hole , cudss 14-h , and elais n2 fields , respectively . eight exposures ( of 510seconds each ) were taken in a single quad with two frames per position . observations were also made of argon and neon calibration lamps and the planetary nebula ngc 7027 , to provide wavelength calibration and to measure the spectral resolution of the instrument . these observations were made with a 0.5-arcsec slit to more allow a more accurate determination of the locations of the emission lines . no shifts were found in the location of any spectral features , or in the locations of the unsuppressed night - sky lines , during the course of the run . an observing log is presented in table [ tab : obslog ] . .observing log . the dates refer to the spectroscopic observations . imaging observations were made immediately prior to the spectroscopy , except for le 850.1 , which was imaged on ut 2002 may 21 ( see section 5.1).[tab : obslog ] [ cols="<,^ , > , > , > " , ] the low level of signal coming from the extremely faint targets ( we stress that our targets are producing detector signals of about 1 electron per pixel per _ minute _ , which is less than the detector dark current ) makes it imperative to remove all detector artefacts . to do this , we first subtracted one image from each pair from the other to remove the contribution from the dark current . next , any differences in bias level between the four quadrants of the detector were removed by determining the signal in unexposed regions of each quadrant and subtracting this constant value from all the pixels in that quadrant . variance images for each combined pair were constructed based on the raw pixel values in the original images and the detector read noise . residual sky lines were removed by fitting a low - order function to the columns of the images , excluding any regions contaminated by objects . the two pairs in each quad were then combined , by a straight average if the pixel values in each pair were consistent at the @xmath10 level , or by the value closest to zero if they were not . next , the image was shifted by an amount equal to the offset between the positive and negative beams and this shifted image was subtracted from the original , with further rejection of deviant pixels . the variance images were combined in a statistically - appropriate manner . then the two quads were combined and spectra of all objects extracted in 1-arcsec ( 9-pixel ) apertures . the pixel - by - pixel uncertainty on this spectrum was determined from the variance image , although this is likely to underestimate the true noise level in regions of strong sky emission where the sky level may not be accurately determined . spectra of the atmospheric ratioing standards were obtained by simply averaging all the spectral images taken in the ` a ' position and subtracting the average of the spectral images taken in the ` b ' position . this image was shifted to align the two beams and subtracted from the original version of itself . a spectrum was extracted in a 1-arcsec aperture . a quadratic wavelength solution was determined from the arc and planetary nebula spectra and applied to all the spectra . the r.m.s . deviation of the individual emission lines from the fit was @xmath11 ( approximately 1 pixel ) . model spectra of the standard stars were constructed assuming they could be described by blackbodies with effective temperature and @xmath7-band magnitudes determined from the @xmath12-band magnitudes and spectral types listed in the hipparcos catalogue ( tokunaga 2000 ) . spectra of the targets in flux units were then determined by dividing each target spectrum by the appropriate atmospheric standard spectrum and multiplying by the model blackbody . since the standard stars were chosen to be close in the sky to the targets , their airmasses at the time of observation are similar to those of the target at the end of its observation , rather than an average target airmass . this may result in imperfect removal of atmospheric absorption features . the absolute flux calibration of the targets was determined from the acuqisition imaging . images at each of the two positions were combined and then subtracted from each other . this image was then shifted to align the two beams and subtracted from the original version of itself . photometry was performed in a 1-arcsec square aperture and therefore the flux scale of the final spectra represents the amount of light in the extraction aperture . the uncertainty in the photometry was estimated by fitting separate gaussians to sigma - clipped , logarithmically - scaled histogram of the pixel values both inside and outside the dark lane . the instrumental zeropoint of the photometry was tied to observations of the ukirt faint standards fs 27 and fs 146 ( hawarden et al . 2001 ) , taken at the end of the last night . any systematic uncertainties in the absolute calibration introduced by only taking such images at one point during the run are likely to be dwarfed by the 1020per cent random uncertainties in the target photometry . in cases where more than one object was in the spectrograph slit , the flux scale was tied to the brightest object . inevitably , we expect some spurious features from ` warm ' pixels which are not rejected by our rejection algorithm , since this algorithm needs to be conservative to avoid throwing out a large amount of good data ( in addition , the rejection limits need to be loose enough to not flag a change in the signal which is the result of changes in the seeing and/or atmospheric transparency ) . the reduced two - dimensional spectra are therefore used to determine the veracity of any putative emission lines , by determining whether they arise from only one of the two beams . the fwhm of the arc lamp and planetary nebula lines was measured to be 3540 , indicating that the object spectra will have a spectral resolution of @xmath13 . this is consistent with directly - made measurements of the resolution obtained with a 1-arcsec slit ( e.g. willott et al . 2003 ) , and means that any emission lines will be unresolved . the low resolution of the spectrum allows us to reject as spurious some features which might at first glance appear real ( especially in the smoothed spectra presented in fig . [ fig : spectra ] ) . this is important because the large number of pixels means that we have a @xmath14per cent chance of finding a @xmath15 ` emission line ' simply by chance . this probability is reduced by a factor of about 2 by excluding those ` lines ' where the flux is dominated by a small number of errant pixels . on the other hand , it is almost inevitable that each spectrum will contain at least one emission feature which is formally significant at the @xmath16 level , and we must treat such marginal ` detections ' with extreme caution . we can therefore only consider redshifts to be plausible if they are determined from a single , highly significant line , or from two or more less significant detections whose relative wavelengths correspond to a plausible pair of features . to detect such features , we use a 51-pixel running median to estimate the continuum level , and then search for groups of 9 pixels ( i.e. @xmath17 , or approximately the resolution fwhm ) whose total deviation from this continuum has a signal - to - noise ratio larger than a certain threshold , which was set at 2.2 after some experimentation . once features had been detected in this manner , the 2d spectrum was examined to ensure that they were not the result of poor sky subtraction or low - level cosmic rays which had not been removed by the rejection algorithm . not all detections are listed therefore in table [ tab : lines ] , where we summarize the results of our analysis , although we only exclude those features which are clearly spurious . the detections are described in detail on an object - by - object basis in the next section . it is important to note that the uncertainties quoted for the line fluxes are derived solely from the detector read noise and photon shot noise , and do not take into consideration the effects of improper sky subtractor or the placement of the continuum . incorrectly determining the continuum by as little as @xmath18 ( see fig . [ fig : spectra ] ) will cause the measured line flux to be in error by @xmath19 , which is typically 2040per cent of the flux . we therefore warn against any interpretation of line ratios , which are often substantially different from those expected . even though in some cases the median appears by eye to be wrong ( and will always be slightly biased by the presence of the detected line ) , any ` by eye ' estimate is likely to be just as wrong , so we choose to fix the continuum at the level derived by the running median , which at least has the advantage of being non - subjective . the line centres are determined from a simple centroiding of those pixels above the continuum level , and typically have an accuracy of @xmath2020 which , when combined with the residuals from the wavelength solution fit , imply uncertainties of @xmath21 in each individual redshift measurement . given the marginal nature of our line detections , we investigate the probability of obtaining false redshifts from our data . we first produce 100 artificial spectra from each of the ten noise spectra ( seven objects , plus three companions ) , by multiplying the noise spectrum by a spectrum of random numbers drawn from a normal distribution . we apply our line detection algorithm to each of these spectra , again rejecting those detections which are not consistent with unresolved lines . using the @xmath22 detection threshold , we find that most spectra produce one or two spurious features , but there is a pronounced tail , and 3per cent of our spectra produce five features . while this may seem like cause for alarm , the fact that we require more than half of the detected features to correspond to plausible lines ( and be absorption or emission features as appropriate ) makes it extremely unlikely that we will derive a redshift from five randomly - located lines . we investigate two distinct ways in which we can derive an erroneous redshift . first , the redshift can be produced from entirely spurious lines . this is not possible unless there are at least two lines , and in this case the redshifts derived from their identifications must agree to within 0.015 ( for the case of three spurious lines , we require that the redshifts from any pair agree , while for the cases of four or five spurious lines , we require that consistent redshifts are derived from a triplet ) . these probabilities can be calculated directly from our 1000 spectra , and amount to 10per cent . second , an erroneous redshift can be produced from a single real line , plus one or more spurious lines . to calculate this probability requires _ a priori _ knowledge of the probability of there being a genuine feature at a given wavelength . if we assume that the genuine emission line has an equal probability of appearing at any point in the spectrum , then the probability of an erroneous redshift is 20per cent , multiplied by the _ a priori _ probability that there is a genuine line in the @xmath6 or @xmath7 bands . the most likely cause of an erroneous redshift arises from cases where three lines are detected , and a pair of them correspond to plausible features . lcrlc & @xmath23 & + object & ( @xmath24 ) & ( @xmath25 ) & identification & @xmath26 + le 850.1 & 1.1544 & @xmath27 & + & 1.2440 & @xmath28 & + & 1.5998 & @xmath29 & + + le 850.3 & 1.1627 & @xmath30 & [ o ii ] @xmath313727 & 2.120 + & 1.5993 & @xmath32 & spurious + + cudss 14.1 & 1.2994 & @xmath33 & & + & 1.6675 & @xmath34 & + & 1.7227 & @xmath35 & + & 1.7543 & @xmath36 & + + cudss 14.3 & + + cudss 14.9 & 1.5216 & @xmath34 & ca k ? & 2.869 + & 1.5374 & @xmath37 & ca h ? & 2.874 + + n2 850.2 & 1.2889 & @xmath38 & [ o ii ] @xmath313727 & 2.458 + & 1.5637 & @xmath39 & spurious + & 1.5751 & @xmath40 & spurious + & 1.6777 & @xmath41 & h@xmath42 & 2.451 + & 1.7091 & @xmath43 & [ o iii ] @xmath314959 & 2.446 + & 1.7304 & @xmath44 & [ o iii ] @xmath315007 & 2.456 + + n2 850.12 & 1.5722 & @xmath45 & spurious + & 1.6678 & @xmath46 & h@xmath42 & 2.431 + & 1.7125 & @xmath47 & [ o iii ] @xmath315007 & 2.420 + + cudss 14.3 & + + n2 850.2 & 1.1700 & @xmath48 & + + n2 850.12 & + in this section , we describe our redshift determinations based on the data in table [ tab : lines ] . we list all plausible emission lines , even if they do not provide a definitive redshift measurement , so that our data may be of use if an independent redshift determination supports the reality of these lines . we deliberately refrain from making subjective comments on the relative believability of any lines in a given spectrum . due to the extremely faint and diffuse nature of this source , it took a substantial amount of imaging on the night of ut 2002 may 21 to make a firm imaging detection . although by the time we had done so , the hour angle of the source was no longer favourable , we confirmed the offsets from a nearby star as being 11.6arcsec west and 5.2arcsec south . we returned to le 850.1 two nights later , and used our previous imaging to hasten the acquisition . since the distance between the target and offset star is larger than the ohs field of view , we first located the bright star near the end of the field of view , and made a small ( 2arcsec ) movement of the telescope by altering the location of the centroiding box on the autoguider camera . the success of this technique depends on the accuracy to which the plate scales of cisco and the ag camera are known , and we estimate the error to be no more than 0.2arcsec . since we were unable to obtain reliable photometry from our acquisition image , we applied the same flux scaling factor for le 850.1 as was applied for le 850.3 ; from the flux in the ` clean ' part of our @xmath7-band spectrum ( 1.601.70@xmath49 m ) , we estimate @xmath50 in our extraction aperture . three emission lines are suggested by our line - finding algorithm , although all are extremely marginal . no pair of these lines corresponds to plausible features , and so we must conclude that at least two , and possible all three , of the lines are spurious . however , given that it is three times more likely for there to be only two spurious lines , rather than three , we consider it probably that one of the lines is real . the redshift estimate of @xmath51 ( model ` le2 ' of aretxaga et al . 2003 ; hereafter a03 ) suggests that either of the bluest two lines could be [ o ii ] ( at @xmath52 or @xmath53 ) , but we are unable to assign a redshift from our data . this source is the brightest of our targets in the @xmath7-band ( @xmath54 in our extraction aperture ) and has relatively good signal - to - noise in the clean parts of the spectrum . our detection algorithm produces two emission lines , both at marginal ( @xmath55 ) significance . the lines do not correspond to any plausible pair of transitions , but we note the presence of a clear continuum break at @xmath56 m . we confirm the reality of this break by subtracting the spectrum of blank sky at a distance of 1.5arcsec along the slit ( extracted in an identical 1-arcsec aperture ) . if we identify this break as the balmer jump , then the bluer of the two emission lines is [ o ii ] at @xmath57 , consistent with the radio submm estimate of i02 ) . [ fig : le850p3_fit ] shows the near - ir spectrum of le 850.3 ( shifted to the rest frame ) , overlaid with a synthetic starburst model ( from the pegase code of fioc & rocca - volmerange 1997 ) . this model has an decaying star - formation rate with an @xmath58-folding time of 100myr and is seen at an age of 250myr . it has been reddened by @xmath59 , according to pei s ( 1992 ) parametrization of the milky way extinction law . although this is not a fit , it provides a reasonable description of the observed continuum . the presence of a strong balmer jump requires that the rest - frame 4000- light be dominated by @xmath60yr - old stars , although this may be because younger stars are more heavily obscured ( rather than because the star - formation rate is decaying with time , as in our simple model ) . the reasonable fit that this model provides to the spectrum gives us confidence in assigning a redshift of @xmath61 to this source , even though it is rather inconsistent with the estimate of @xmath62 of a03 . this source ( also known as cudss 14a ) was the subject of intense study by gear et al . high - resolution _ hubble space telescope _ and ground - based ( canada france hawaii telescope ) adaptive optics images reveal it to be a red , compact galaxy . deep near - ir _ jhk _ spectroscopy by gear et al . failed to detect any emission lines to a level of 1@xmath63 . our new spectroscopy pushes this limit almost an order of magnitude fainter , but fails to find any significant ( @xmath64 ) lines . we detect four features ( three emission lines , and one absorption line ) of marginal significance , however . the measured wavelengths of the bluest and reddest emission lines are consistent with being h@xmath42 and h@xmath65 , respectively , at @xmath66 . in this interpretation , the third emission line and the absorption line are artefacts . alternatively , these two lines could be [ o ii ] at @xmath67 and [ o iii ] @xmath315007 at @xmath68 ; although this redshift discrepancy is large , it is not implausible given the low signal - to - noise ratios of the detections . with these identifications , the third emission line could be @xmath314959 at @xmath69 , which is barely consistent with the redshift from the other two lines , while the absorption feature is , again , spurious . the lower redshift is consistent with both the photometric redshift estimate of @xmath70 and the radio submm redshift estimate of @xmath71 ( clements et al.2004 ; hereafter c04 ) . we note , however , that a03 suggest a much higher redshift of @xmath72 , which is just consistent ( at the 90% level ) with both the above sets of line identifications . if this redshift is correct , then one of the @xmath7-band emission lines could be [ o ii ] at @xmath73 , while the remaining three features would be artefacts ( possibly the 1.2994-@xmath49 m line could be mg ii @xmath312798 at @xmath74 if the 1.7227-@xmath49 line is [ o ii ] at @xmath75 , but mg ii is not expected unless there is an agn present ) . ultimately , we decide against assigning a spectroscopic redshift to this object . apart from a spurious peak due to poor subtraction of the unsuppressed oh lines in the range 1.501.52@xmath49 m , there are no significant features in the spectrum of this faint ( @xmath76 ) object . a brighter , slightly bluer object is located approximately 2arcsec west and slightly north of cudss 14.3 ( fig . 4 of webb et al . 2003 ) , whose spectrum is similarly featureless ( fig . [ fig : comp_spectra ] ) . we are unable to obtain a redshift for either object . we note that both the photometric and radio submm redshift estimates of c04 are @xmath77 , at which redshift no strong emission lines would lie within our spectral coverage ( h@xmath65 would fall between the @xmath6 and @xmath7 spectroscopic windows ) . a03 obtain a much higher redshift estimate ( @xmath78 ) , which would place the [ o ii ] line somewhere between the 1.27-@xmath49 m o@xmath79 night sky emission and the long - wavelength end of our spectral coverage , and it could therefore easily fall between the @xmath6 and @xmath7 windows , or near the blue end of the @xmath7-band where the poor sky subtraction affects our sensitivity . the spectrum of cudss 14.9 ( @xmath80 ) bears a remarkable similarity to that of le 850.3 , with an apparent break in the continuum around @xmath81 m . however , the spectrum is noisier and the break is weaker if we make an _ ad hoc _ definition od the break strength as @xmath82 , we measure @xmath83 for cudss 14.9 , compared to @xmath84 for le 850.3 ( or @xmath85 after removal of the [ o ii ] line ) . two lines are detected by our algorithm ; absorption features at 1.5216 and 1.5374@xmath49 m with observed equivalent widths of @xmath86 and @xmath87 , respectively . the wavelengths of these lines are consistent with the ca h+k lines at @xmath88 . we are hesitant about assigning this redshift , however , since there is no clear continuum break associated with the absorption lines ( although there is insufficient coverage to the blue of them to make an accurate measurement of the continuum level ) and the redder of the two features is affected by a night sky emission line . furthermore , this redshift is inconsistent with c04 s photometric and radio submm estimates of @xmath89 and @xmath90 , respectively , although it agrees with a03 s estimate of @xmath91 . if our spectroscopic measurement is correct , however , none of the strong rest - frame optical emission lines associated with star formation ( h@xmath65 , h@xmath42 , [ o ii ] ) would lie within a spectroscopic window . the identification for this source is extremely faint ( @xmath92 in our spectroscopic aperture ) and the continuum is only seen in our 2d spectrum after substantial smoothing . the location of the extraction apertures were determined from the acquistion image . five plausible emission features ( one of which is detected at @xmath64 ) and one absorption feature are detected by our algorithm , although the absorption feature is close to one of the emission lines and may result from imperfect subtraction of the 1.58-@xmath49 m sky emission . the other four emission lines , however , are consistent with a redshift @xmath93 . this redshift is consistent with tha lack of formally - detected continuum in the @xmath6-band , since the 4000- break will lie between the @xmath6 and @xmath7 atmospheric windows , and also with the redshift estimates of i02 and a03 . it also agrees well with the optical redshift of 2.454 ( chapman et al . , in preparation ) . our spectroscopic slit was aligned to pass through a galaxy 2arcsec south southeast of elais n2 850.2 ( fig . 2 of i02 ) . this source shows faint continuum in the clean part of the @xmath6-band ( @xmath94 m ; the continuum is more clearly seen in the smoothed 2d spectrum than in fig . [ fig : comp_spectra ] ) but this does not rule out its being at the same redshift as the submm source , since it is brighter ( @xmath95 in our aperture ) . there is a single marginally - detected emission line at 1.1700@xmath49 m ; if real , this would mean that the ` companion ' is not physically associated with the submm source , since it does not correspond to any known emission line at the redshift of the submm source . this identification for this source is also very faint ( @xmath96 ) and the signal - to - noise ratio of our spectrum is low . however , we find three emission lines , of which we identify two as h@xmath42 and [ o iii ] @xmath315007 at @xmath97 ( the absence of detectable [ o iii ] @xmath314959 is simply due to its low flux ) . this redshift is also consistent with the lack of continuum at the red end of the @xmath6-band , since it places the 4000- break at @xmath98 m ( we note , however , that continuum is detected at the @xmath99 level in the ` clean ' part of the @xmath6-band ( 1.201.25@xmath49 m ) . at this redshift , the [ o ii ] @xmath313727 emission line would lie at @xmath100 m , where our sensitivity is low ( @xmath101 ) due to the o@xmath79 night - sky emission ; its non - detection is therefore not at odds with a ratio [ o ii]/h@xmath102 ( e.g. kennicutt 1998 ) , especially if there is some reddening . we note that this redshift is unexpectedly low , given the high ratio of submm to radio flux , though this inconsistency can be understood if the handful of 3@xmath103 radio peaks noted by i02 are the relics of a diffuse radio source . since this redshift is derived from two of three detected lines , it should be treated with caution , although it gains some support from the fact that there is positive flux at the expected position of [ o ii ] ( formally , the value is @xmath104 ) , and from the relatively high significance of the line at 1.7125@xmath49 m ( @xmath105 ) . our spectroscopic slit was aligned to pass through a second object located approximately 7arcsec west of n2 850.12 ( on the edge of the postage stamp shown in fig . 2 of i02 ) . this object is brighter ( @xmath106 ) and bluer and our algorithm failed to detect any emission lines in its spectrum ( see fig . [ fig : comp_spectra ] ) . we can not say whether it is physically associated with the submm galaxy . given the extreme star - formation rates implied by the submm luminosities of these objects and our extensive wavelength coverage , the lack of strong emission lines is perhaps a surprise . a star - formation rate of @xmath107 ( typical of these objects ; ivison et al . 1998 ; gear et al . 2000 ) corresponds to an h@xmath65 luminosity @xmath108w ( kennicutt 1998 ) , and we assume h@xmath65/h@xmath109 ( case b ; e.g. osterbrock 1989 ) and h@xmath65/[o ii@xmath110 = 0.9 $ ] ( kennicutt 1992 ) . these lines should all be well above our detection threshold ( see fig . [ fig : lineflux ] ) . in fact , in the absence of any extinction , our 2-hour integrations should be sensitive to star - formation rates as low as @xmath111 out to @xmath112 , although there is a gap in our coverage between @xmath113 when [ o ii ] lies in the gap between the @xmath6 and @xmath7 atmospheric windows . this gap could be significant since it lies close to the peak in the redshift distribution , @xmath114 , presented by chapman et al . ( 2003 ) , based on optical spectroscopy of bluer radio - detected submm galaxies . we have redshifts for only three of our seven targets . taken at face value , the measured line fluxes of le 850.3 , n2 850.2 , and n2 850.12 correspond to star - formation rates of @xmath20 , @xmath115 , and @xmath116 , respectively , well below the estimates from the radio submm data ( i02 ) . such large discrepancies are not unusual ; for example , in the extremely red object hr 10 ( hu & ridgway 1994 ) , the raw ( i.e. uncorrected for extinction ) h@xmath65-derived star - formation rate is a factor of @xmath117 lower than that derived from the submm flux ( cimatti et al . 1998 ; dey et al . the two rates can be reconciled by appealing to substantial quantities of dust which severely attenuate the observed line emisison . in all three cases , the emission - line - derived star - formation rate can be rectified with the submm radio - derived sfr if the emission lines are extinguished by @xmath1184mag ( compared to the @xmath119mag needed in hr 10/h@xmath120 , which they claim is `` at the extreme limit of observed values for local star - forming galaxies ( e.g. , kennicutt 1992 ) . '' however , fig . 12 of kennicutt ( 1992 ) indicates that this ratio is quite normal , since the data plotted are uncorrected for reddening ( a correction factor of @xmath121 ) or [ n ii ] contamination ( @xmath122 ) . ] ) . slit losses are unlikely to account for more than a factor of about 2 in our line fluxes since our slit width corresponds to a linear size of 8kpc , which is large for a starburst region ( smail et al . furthermore , the radio emission is not observed to be extended on scales much larger than this . arguing against a foreground screen hypothesis is the detection of ly@xmath65 emission in submillimetre galaxies ( chapman et al . the measured fluxes correspond to star formation rates of @xmath123 , broadly in line with our measurements , while a foreground screen with @xmath124 would produce at least four orders of magnitude of extinction at ly@xmath65 and completely suppress the line . the consistency between the uv and optical star formation diagnostic lines therefore suggests that we are seeing a small fraction ( @xmath125 per cent ) of the star formation in a relatively unobscured way , while the vast majority is completely extinguished at optical / uv wavelengths . some obscuration is still required since a @xmath126 sfr at @xmath127 produces an observed @xmath128 ( leitherer , robert , & heckman 1995 ) , which is brighter than our targets ( i02 ; c04 ) . this picture is quite reasonable , since massive stars are formed in the local universe inside the cores of dense molecular clouds , and the @xmath129myr it takes for these stars to disperse their parent clouds is longer than the lifetimes of stars with @xmath13020m@xmath131 ( see discussion in jimenez et al . 2000 ) . jimenez et al . suggest that the observed flux of far - ultraviolet photons represents only about one - sixth of the total star formation , and the observed line emission should also underestimate the true star formation rate by a similar factor if the massive stars are heavily obscured . larger discrepancies are possible if the starburst is younger than 15myr since even the longer - lived stars are embedded in their parent clouds . an alternative hypothesis which warrants investigation is that the star formation rates derived from the submm observations are substantially overestimated . the most likely reason for this would be erroneously high assumed dust temperatures . for a dust emissivity index @xmath42 , the bolometric luminosity per unit dust mass is proportional to @xmath132 , while the dust mass for a given monochromatic submm luminosity is inversely proportional to the dust temperature . therefore , for a source with a known redshift and measured 850-@xmath49 m flux , the inferred bolometric dust luminosity ( and hence star formation rate ) is proportional to @xmath133 ( e.g. blain et al . 2002 ) . this suggests that very large errors could be introduced by an incorrect estimate of the dust temperature . however , the expression is only accurate at low redshift , where @xmath134 ( @xmath135 being the frequency at which the observed 850-@xmath49 m radiation is emitted ) , and the possible error is more modest at @xmath136 , where a factor of 2 uncertainty in dust temperature corresponds to a factor of 10 uncertainty in luminosity ( e.g. hughes , dunlop & rawlings 1997 ; eales et al . 2000 ) . i02 adopt @xmath137k , which is near the hot end of the distribution found for ultraluminous infrared galaxies ( dunne et al . farrah et al . 2002 ) . in principle , therefore , the star formation rates of @xmath138 could be an order of magnitude too high if the temperature of the dust is @xmath139 . the ratio of 850-@xmath49 m to 450-@xmath49 m flux is a relatively sensitive function of dust temperature for @xmath140 , as the shorter wavelength point approaches the peak of the emission . eales et al . ( 2000 ) claim an average upper limit of @xmath141 for the cudss sources , which appears to rule out @xmath142 dust if the median source redshift is 2 . however , fox et al . ( 2002 ) only constrain this ratio to be @xmath143 for sources from the 8-mjy survey , which is low enough to allow hotter dust . in principle , the radio flux should provide an alternative estimate of the sfr if the redshift is known , since the tight radio fir correlation provides a way to derive the total fir luminosity without having to assume the dust temperature . systematically overestimated sfrs should result in high values of @xmath144 ( i.e. steeper spectral indices ) whereas there appears to be a strong tendency for the observations to produce shallower spectra ( e.g. c04 ) . the cause of this is unclear , and it seems that this issue will only be unequivocally resolved with shorter wavelength detections which can constrain the shape of the sed near the peak of the thermal emission . for now , we believe we are justified in claiming that it is highly improbable that the submm - derived sfrs are reconcilable with our spectroscopically - derived estimates unless most of the star formation is heavily obscured . we are unable to determine whether this obscured fraction is 90per cent or 99per cent of the total star formation . near - infrared spectroscopy of submm galaxies has been rather limited to date . ivison et al . ( 2000 ) presented @xmath145-band spectra of two sources behind the lensing cluster abell 1835 , smm j14011 + 0252 and smm j14010 + 0253 , which both showed h@xmath65 emission but lacked the strong [ n ii ] emission characteristic of an agn . using near - infrared spectrographs on larger telescopes , frayer et al.(2003 ) presented a spectrum of smm j04431 + 0210 , while smail et al.(2003a , b ) observed smm j17142 + 5016 ( in the protocluster around lbds 53w002 ) and elais n2 850.4 , respectively . all three of these galaxies showed strong emission lines indicative of the presence of an agn , which smail et al . suggest is a common feature of these objects . all the emission lines in these objects are brighter than our upper limits , by an order of magnitude in the cases of smm j17142 + 5016 and n2 850.4 . the spectra of our targets are dramatically different from those of the ( few ) objects in the literature . however , this fact may be readily explicable by the fact that our targets were selected to be optically faint , and are therefore perhaps biased towards more heavily extinguished objects . on the other hand , previously observed objects already had known redshifts from optical spectroscopy , and must therefore have had bright ly@xmath65 emission , and/or sufficiently bright optical continua to allow the detection of absorption lines . either case implies that the objects with near - ir spectra in the literature are lightly extinguished . in support of this idea , eisenhauer et al . ( 2003 ) recently obtained a high - quality near - infrared spectrum of the @xmath146 submm source smm j14011 + 0252 . they found that the rest - frame optical continuum of this object is dominated by stars with ages of a few hundred myr , similar to our estimate for le 850.3 ( the only object in which we detect an age - sensitive stellar feature ) although , unlike le 850.3 , the continuum of smm j14011 + 0252 appears to be unreddened , which presumably explains its unusually bright rest - frame uv spectrum ( ivison et al . such an age is perhaps larger than expected , since a constant star formation rate of @xmath138 can form a massive galaxy in only @xmath147 . the most serious difficulty which arises when attempting to understand the nature of submm galaxies as a class comes from a lack of clarity in the criteria with which submm galaxies are selected for spectroscopy by different groups . this difficulty is almost certainly compounded by the likelihood that spectra are only published where firm redshifts are obtained ( i.e. , objects with luminous emission lines , or those with bright enough continua to allow the detection of absorption features ) . this leads to the presence of strong and uncertain selection effects in the literature samples , and makes generalized statements about the properties of the population as a whole unreliable . nevertheless , almost 100 spectroscopic redshifts for submm galaxies have been obtained , and there is little reason to believe that the redshift distribution obtained from these objects is significantly different from that of the class as a whole ( chapman et al . 2003 ; but see also dannerbauer et al . it is therefore clearly time for the study of these important objects to progress from simple redshift determination and basic classification ( agn , starbust , composite ) to more detailed analyses . the difficulties posed by the presence of large quantities of dust and the existence of multiple components point to the need for spectroscopy over a long wavelength baseline , covering rest - frame wavelengths from ly@xmath65 to h@xmath65 . the most interesting new instrument for this purpose is fmos , the fibre multi - object spectrograph ( maihara et al . 2000 ; kimura et al.2003 ) which is scheduled to be commissioned on subaru telescope in late 2005 . this instrument will provide oh - suppressed spectroscopy with increased sensitivity , spectral resolution , and wavelength coverage ( 0.91.8@xmath148 m ) compared to ohs , using 400 fibres across a 30-arcminute field of view ( this corresponds to the source density at a flux level of @xmath149 ; e.g. , borys et al.2003 ) . the ability to take many spectra simultaneously enables much longer exposures than we were able to undertake here and , coupled with the increased sensitivity of the new instrument , it should be relatively straightforward to probe a further order of magnitude below our current detection limits . our emission line sensitivities will therefore correspond to star formation rates of @xmath150 , and it would be remarkable if a significant number of targets failed to display emission lines . furthermore , the increased sensitivity to the continuum emission will allow more accurate modelling of the galaxies stellar populations . as a demonstration of the likely data quality we can expect from fmos , we simulated the spectrum of n2 850.2 using the fmos spectral simulator developed by naoyuki tamura ( fig [ fig : fmos ] ) . our input spectrum assumes fluxes for the h@xmath42 , [ o ii ] , and [ o iii ] @xmath315007 lines given in table [ tab : lines ] , superimposed on a continuum of a 3-gyr - old elliptical , reddened by @xmath151 . all emission lines are assumed to have @xmath152 , and an exposure time of 7hours was adopted , representing the maximum likely on - source integration time possible in a single night s observing . a major caveat to this lies in the infrared `` spectroscopic desert '' at @xmath113 . here , [ o ii ] , the balmer jump , and the 4000- break are redshifted between the @xmath6 and @xmath7 windows ; h@xmath42 and [ o iii ] are redshifted between @xmath7 and @xmath145 ; and h@xmath65 is redshifted beyond the @xmath145 band . although this is a relatively small redshift range , it is close to the peak of the redshift distribution and may include more than 10 per cent of sources from a flux - limited sample ( chapman et al . 2003 ) . without the reddening - insensitive age measurements provided by the balmer jump and/or 4000- break , and with the low spectral resolution of fmos , detailed stellar population analysis of these targets is unlikely to prove fruitful . for now , further infrared spectroscopy is still worthwhile as a precursor to the above observations . many submm sources display optical emission lines and , regardless of whether these are powered by a starburst or an agn , there will be associated nebular lines redshifted into the near - infrared . the relative strengths of these lines ( in objects whose redshifts are suitable for near - infrared spectroscopy ) will test our idea that they come from a relatively unobscured region comprising only a small fraction of the total starburst . we have presented the results of deep near - ir spectroscopy of seven galaxies selected from the scuba 8-mjy and cudss submm surveys . we have obtained spectroscopic redshifts for three of these objects , two of which ( le 850.3 and elais n2 850.2 ) agree well with the prior estimates from radio submm data , while the other ( elais n2 850.12 ) is unexpectedly low . these redshifts are based on marginal ( @xmath153 ) detections of emission lines , although in all cases they come from more than one spectral feature , and in one instance ( n2 850.2 ) there are four emission lines , and the derived redshift agrees with the ( as yet unpublished ) ly@xmath65 and co redshifts . in all three cases , the star - formation rates derived from the emission - line luminosities are much lower than those derived from the radio submm data . by comparing the bulk properties of our sample with those observed at optical wavelengths , we suggest that we are seeing a small fraction of the total starburst through low extinction . further infrared spectroscopy of sources with secure optical redshifts are required to test this hypothesis . this paper is based on data collected at subaru telescope , which is operated by the national astronomical observatory of japan . the authors are grateful to the staff of subaru telescope for their help with the observations , and to the anonymous referee for his / her comments . cs and jsd thank the particle physics and astronomy research council for funding in the form of an advanced fellowship and senior research fellowship , respectively .	we present the results of deep near - infrared spectroscopy of seven submillimetre - selected galaxies from the scuba 8-mjy and cudss surveys . these galaxies were selected because they are too faint to be accessible to optical spectrographs on large telescopes . we obtain a spectroscopic redshift for one object , and likely redshifts for two more , based on a combination of marginal emission line detections and the shape of the continuum . all three redshifts broadly agree with estimates from their radio / submm spectral energy distributions . from the emission line strengths of these objects , we infer star formation rates of 1025m@xmath0 , while the lack of detections in the other objects imply even lower rates . by comparing our results with those of other authors , we conclude it is likely that the vast majority ( more than 90per cent ) of the star formation in these objects is completely extinguished at rest - frame optical wavelengths , and the emission lines originate in a relatively unobscured region . finally , we look at future prospects for making spectroscopic redshift determinations of submm galaxies . [ firstpage ] galaxies : distances and redshifts galaxies : evolution galaxies : formation galaxies : starburst cosmology : observations infrared : galaxies
it has long been recognized that the sun is not located precisely in the mid - plane of the galactic disk defined by @xmath6 but is displaced a few parsecs to the north of galactic plane ( gp ) ( see blitz & teuben 1996 for a review ) and understanding the exact value of @xmath0 is vital not only for the galactic structure models but also in describing the asymmetry in the density distribution of different kind of stars in the north and south galactic regions ( cohen 1995 , mndez & van altena 1998 , chen et al . several independent studies in the past have been carried out to estimate @xmath0 using different kind of astronomical objects , for example , gum , kerr & westerhout ( 1960 ) concluded that @xmath7 pc from the neutral hydrogen layer , kraft & schmidt ( 1963 ) and fernie ( 1968 ) used cepheid variables to estimate @xmath8 pc while stothers & frogel ( 1974 ) determined @xmath9 pc from the b0-b5 stars within 200 pc from the sun , all pointing to a broad range of @xmath0 . more recently various different methods have been employed to estimate @xmath0 e.g. cepheid variables ( caldwell & coulson 1987 ) , optical star count technique ( yamagata & yoshii 1992 , humphreys & larsan 1995 , chen et al . 2001 ) , wolf - rayet stars ( conti & vecca 1990 ) , ir survey ( cohen 1995 , binney , gerhard & spergel 1997 , hammersley et al . 1995 ) along with different simulations ( reed 1997 , mndez & van altena 1998 ) and models ( chen et al . 1999 , elias , cabrera - cao & alfaro 2006 , hereafter eca06 ) . most of these studies constrained @xmath0 in the range of 15 to 30 pc in the north direction of the gp . in recent years , the spatial distribution of open clusters ( ocs ) have been extensively used to evaluate @xmath0 since continued compilation of new clusters has brought together more extensive and accurate data than ever . using the ocs as a diagnostic tool to determine @xmath0 , janes & adler ( 1982 ) found @xmath0 = 75 pc for 114 clusters of age smaller than @xmath10 yr while lynga ( 1982 ) determined @xmath11 pc with 78 young clusters up to 1000 pc . pandey & mahra ( 1987 ) reported @xmath0 as 10 pc from the photometric data of ocs within @xmath12 and pandey , bhatt & mahra ( 1988 ) using a subsample of yocs within 1500 pc obtained @xmath13 pc . most recently , @xmath0 have been determined in three independent studies based on the analysis of ocs . considering about 600 ocs within @xmath14 of gp , we derived @xmath15 pc through the analysis of interstellar extinction in the direction of the ocs ( joshi 2005 , hereafter jos05 ) . bonatto et al . ( 2006 ) reported @xmath0 as 14.8 @xmath16 2.4 pc using 645 ocs with age less than 200 myrs while piskunov et al . ( 2006 , hereafter pkss06 ) estimated a value of 22 @xmath16 4 pc using a sample of 650 ocs which is complete up to about 850 pc from the sun . on the other hand using a few thousand ob stars within @xmath17 of the gp and 4 kpc from the sun , reed ( 1997 ) approximately estimated the value as 10 - 12 pc while maz - apellniz ( 2001 ) determined this values as @xmath18 pc using a sample of about 3400 o - b5 stars obtained from the hipparcos catalogue . the large range of @xmath0 derived from these different methods could be possibly caused by the selection of data of varying age , heliocentric distance @xmath19 , spectral type , etc . along with the method of the determination . the aim of the present paper is therefore to study the variation in @xmath0 following different methods by constraining different upper limits in @xmath20 and @xmath19 using a large sample of ocs and ob stars . the paper is organized as follows . first we detail the data used in this study in sect . 2 . in sect . 3 , we examine the distribution of @xmath20 with the age of clusters while sect . 4 deals their distribution with the different @xmath20 cut - off and @xmath19 cut - off in order to determine @xmath0 . the exponential decay of @xmath20 distribution of the ocs and ob stars and their variation over the galactic longitude are discussed in sects . 5 and 6 respectively . our results are summarized in sect . we use two catalogues in this study . the oc catalogue is complied by dias et al . ( 2002)wilton/. ] which includes information available in the catalogues of the lynga ( 1987 ) as well as webda with the recent information on proper motion , age , distance from the sun , etc . the latest catalogue ( version 2.7 ) that was updated in october 2006 gives physical parameters of 1759 ocs . of these , 1013 ocs have distance information for which it is possible to determine @xmath20 which is equivalent to @xmath21 where @xmath22 is the galactic latitude . out of the 1013 ocs , age information is available for 874 ocs with ages ranging from 1 myr to about 10 gyrs , although the majority of them are young clusters . though the clusters are observed up to a distance of about 15 kpc , it should be born in mind that the cluster sample is not complete owing to large distance and/or low contrast of many potential cluster candidates ( bonatto et al . 2006 ) and may be smaller by an order of magnitude since a good fraction of clusters are difficult to observe at shorter wavelengths due to large extinction near the gp ( lada & lada 2003 , chen , chen & shu 2004 , pkss06 ) . when we plot cumulative distribution of the clusters in our sample as a function of @xmath19 in fig . 1 , we notice that the present cluster sample may not be complete beyond a distance of about 1.7 kpc . a comprehensive discussion on the completeness of ocs has recently been given by bonatto et al . ( 2006 ) which along with pkss06 puts the total number of galactic ocs in the order of @xmath23 . the other sample used in the present study is that of the ob stars taken from the catalogue of reed ( 2006 ) which contains a total of 3457 spectroscopic observations for the 2397 nearby ob starsreed / obfiles.doc ] . the distance of ob stars are derived through their spectroscopic parallaxes . it is worth to note that the individual distance of ob stars may not be accurate ( reed 1997 ) , nevertheless , a statistical study with significant number of ob stars can still be useful for the determination of @xmath0 . although , several studies on the determination of @xmath0 using ob stars have already been carried out on the basis of hipparcos catalogue ( maz - apellniz 2001 , eca06 and references therein ) , however , it is noticed by some authors that the hipparcos catalogue gives a reliable distance estimation within a distance of only 200 - 400 pc from the sun ( cf . torra , fernndez & figueras 2000 ) . this is exactly the region where ob stars in the gould belt ( hereafter gb ) lie and this can cause an anomaly in the determination of @xmath0 if the stars belonging to the gb are not be separated from the data sample . further abt ( 2004 ) also noticed that classification of the stars in the hipparcos catalogue is uncertain by about + /-1.2 subclass in the spectral classifications and about 10% in the luminosity classifications . in the present study we therefore preferred reed s catalogue of ob stars over the hipparcos catalogue despite lesser in numbers but are reported up to a distance of about 1200 pc from the sun and @xmath24 mag . the ob stars which have two different distances in the catalogue are assigned the mean distance provided they do not differ by more than 100 pc , otherwise we remove them from our analysis . if there are more than two distances available for any ob star , we use the median distance . in this way , we considered a sample of 2367 ob stars in this study . it is a well known fact that ocs are born and distributed throughout the galactic disk . young clusters are normally seen in the thin disk while old clusters are found mainly in the thick disk of the galaxy which van den bergh ( 2006 ) termed as a ` _ cluster thick disk _ ' . in order to study the @xmath20 distribution of clusters with their age , we assemble the clusters according to their @xmath25 in 0.2 bins dex in width and estimate a mean value of @xmath20 for each bin . a distribution of mean @xmath20 vs @xmath25 is plotted in fig . 2 which clearly demonstrates that the distribution of clusters perpendicular to the gp has a strong correlation with their ages . while clusters with @xmath1 ( @xmath26 300 myrs ) have almost a constant width of @xmath20 distribution in both the directions of the gp , clusters older than this have mean @xmath27 pc which is continuously increases with the age . this indicates that the thickness of the galactic disk has not changed substantially on the time scale of about 300 myrs and most of the ocs , in general , formed somewhere inside @xmath16 100 pc of the gp . a similar study carried out by lynga ( 1982 ) using a smaller sample of 338 ocs found that clusters younger than one gyr formed within @xmath26 150 pc of the galactic disk . it is quite apparent from the figure that the clusters with @xmath28 are found not only far away from the gp but are also highly scattered in their distribution . however , this is not unexpected since it is a well known fact that clusters close to gp gets destroyed with the time in a timescale of a few hundred million years due to tidal interactions with the galactic disk and the bulge , encounters with the passing giant molecular clouds or mass loss due to stellar evolution . the few remaining survivors reach to outer parts of the galactic disk ( cf . friel ( 1995 ) , bergond , leon & guibert ( 2001 ) ) . if we just consider the clusters with @xmath1 , which we describe as yocs in our following analysis , we find that the 226 clusters ( @xmath26 38% ) lie above gp while 363 clusters ( @xmath26 62% ) lie below gp . the asymmetry in cluster density above and below the gp is a clear indication of inhomogeneous distribution of clusters around gp . this asymmetry can be interpreted as due to the location of the sun above the gp , displacement of the local dust layer from the gp or asymmetry in the distribution of young star formation near the sun with respect to the gp or a combination of all these effects as pointed out by the van den bergh ( 2006 ) . however , it is generally believed that it is the solar offset which plays a major role in this asymmetry . when we estimate weighted mean displacement along the gp for the clusters lying within @xmath1 , we find a value of @xmath29 pc above the gp and @xmath30 pc below the gp . if we consider a plane defined by the yocs at @xmath31 , then @xmath31 can be expressed as , @xmath32 where @xmath33 and @xmath34 are the mean @xmath20 for the yocs above and below the gp respectively ; @xmath35 and @xmath36 are number of yocs in their respective regions . this gives us a value of @xmath37 pc . if the observed asymmetry in the @xmath20 distribution of yocs is indeed caused by the solar offset from the gp then the negative mean displacement of @xmath20 perpendicular to gp can be taken as @xmath0 ( towards north direction ) which is about 25.4 pc . however , it is a well known fact that a large fraction of the young populations with ages under 60 myrs in the immediate solar neighbourhood belong to the gb ( gould 1874 , stothers & frogel 1974 , lindblad 1974 ) . it is widely believed that this belt is associated with a large structure of the interstellar matter including reflection nebulae , dark clouds , hi gas , etc . and is tilted by about 18 deg with respect to the gp and is stretches out to a distance of about 600 pc distance from the sun ( taylor , dickman & scoville 1987 , franco et al . 1988 , pppel 1997 ) . in our sample of 589 clusters , we found 38 such clusters which confined in the region of 600 pc from the sun and have age below 60 myrs . out of the 38 clusters , 26 ( @xmath38 ) follow a specific pattern in the @xmath39 plane as shown by the dark points in the shaded region of fig . 3(a ) which is slightly tilted with respect to the gp and resembles the gb . the association of these clusters with the gb seems to be confirmed by the fact that 23 out of 26 yocs are clumped in the longitude range of about 180 - 300 degrees as shown in fig . this contains the most significant structures accounting for the expansion of the gb ( torra , fernndez & figueras 2000 ) . a mean and median age of these 26 yocs are 24.4 and 21.2 myrs respectively . although no detailed study has been carried out on the fraction of the clusters actually belonging to the gb , however , on the basis of 37 clusters in the @xmath40 which lie within a distance of 500 pc from the sun , pkss06 found that about 55% of the clusters could be members of the gb . on the basis of ob stars in the hipparcos catalogue , torra et al . ( 2000 ) estimated that roughly 60 - 65% of the stars younger than 60 myr in the solar neighbourhood belong to the gb . although it is difficult to decide unambiguously which clusters belong to the gb , we believe that most of these 26 yocs could be associated with the gb instead of the local galactic disk ( hereafter lgd ) . hence to reduce any systematic effect on the determination of @xmath0 due to contamination of the clusters belong to the gb , we excluded all these 26 clusters from our subsequent analysis except when otherwise stated . when we re - derived the value of @xmath0 from the remaining 563 clusters , we find it to be @xmath41 pc north of the galactic plane . a further discussion on the @xmath0 and its dependence on various physical parameters shall be carried out below . various studies indicate that the plane of symmetry defined by the ocs is inclined with respect to the gp ( lynga 1982 , pandey , bhatt & mahra 1988 , jos05 ) . if this is the case , then @xmath0 shall be dependent on the distance of ocs from the sun and inclination angle between the two planes . therefore , a simple determination of @xmath0 considering all the ocs could be misleading . to examine to what extent @xmath0 depends on the distance , we study the distribution of clusters and their mean displacement from the gp as a function of the heliocentric distance ( @xmath42 ) taking advantage of the ocs observed up to a large distance . since yocs are primarily confined closer to the gp as discussed in the previous section , it seems worthwhile to investigate @xmath0 using only yocs despite the fact that the yocs are generally embedded in dust and gas clouds and many are not observed up to a large distance . although we found that some young clusters are reported as far as 9 kpc from the sun but only less than 5% yocs are observed beyond 4 kpc , most of them in the anti - center direction of the galaxy which we do not include in our analysis . following all the above cuts , we retain only 537 yocs observed up to 4 kpc from the sun as a working sample for the present study . their distribution normal to the gp as a function of galactic longitude is plotted in fig . 4(b ) shows the logarithmic distribution of the yocs as a function of @xmath43 . here we derive the number density in bins of 20 pc and error bars shown in the y - axis is the poisson error . following an exponential - decay profile , we estimate a scale height for the yocs as @xmath44 pc which is represented by a continuous straight line in the figure . however , a careful look in the figure suggests that the @xmath45 could be better described by the yocs lying within @xmath46 pc and a least square fit in this region gives a value of @xmath47 pc . it is however interesting to see if the scale height shows any shift in its value when considering a possible displacement of the cluster plane from the gp . in order to analyse any effect of the displacement on @xmath45 , we shift the cluster plane by 10 , 15 , 20 and 25 pc from the gp and recalculate @xmath45 using yocs within @xmath48 pc . our results are given in table 1 . it is seen that these values of @xmath45 are quite consistent and we conclude that the solar offset has no bearing in the determination of scale height . using a sample of 72 ocs younger than 800 myrs , janes & phelps ( 1994 ) reported a scale height of @xmath49 pc . recently bonatto et al . ( 2006 ) derived a scale height of @xmath50 pc using a sample of clusters younger than 200 myrs , however , they have also found a larger @xmath45 when considering ocs older than 200 myrs . pkss06 obtained a scale height of @xmath51 pc using the ocs within 850 pc from the sun . our value of @xmath47 pc obtained with the yocs within 4 kpc from the sun and @xmath48 pc is thus consistent with these determinations . cc shift & @xmath45 + 0 & @xmath4 + + 10 & @xmath52 + + 15 & @xmath53 + + 20 & @xmath54 + + 25 & @xmath55 + + an important issue that needs to be addressed in the determination of @xmath0 is the possible contamination by the outliers which are the objects lying quite far away from the gp that can seriously affect the @xmath0 estimation . hence it is worthwhile at this point to investigate @xmath0 using a subsample of yocs in different @xmath20 zone excluding the clusters far away from the gp without significantly reducing the number of clusters . if the observed asymmetry in the cluster distribution is really caused by an offset of the sun from the gp , then a single value of @xmath20 should result from the analysis . in order to study @xmath0 distribution using yocs , we select three different zones normal to the @xmath56 plane considering the clusters within @xmath57 pc , @xmath58 pc and @xmath59 pc . here , we have not made smaller zones than @xmath60 pc keeping in mind the fact that accounting lesser number of yocs could have resulted in a larger statistical error while zone larger than @xmath61 pc can cause significant fluctuations due to few but random clusters observed far away from the gp . to determine @xmath0 , we keep on moving the mid - plane towards the southwards direction in bins of 0.1 pc to estimate the mean @xmath20 till we get the mean value close to zero i.e. a plane defined by the yocs around which the mean @xmath20 is zero within the given zone that is in fact equivalent to @xmath0 . this approach of a running shift of @xmath20 in order to determine @xmath0 is preferred over the simple mean to remove any biases owing to the displacement of the cluster plane itself towards the southwards direction . hence it gives a more realistic value of the @xmath0 . we estimate @xmath0 with different cut - off limits in @xmath42 using an increment of 0.3 kpc in each step and for all the three zones . the variation in @xmath0 with @xmath42 for all the zones is illustrated in fig . 5 . the figure gives a broad idea of the variation in @xmath0 which increases with the increasing distance as well as zone size , however , it has to be noted that the range of variation is very small and varies between @xmath26 13 to 21 pc throughout the regions . here , it is necessary to look into the increasing trend in @xmath0 whether it is internal variation or due to our observational limitations . we note that 21 out of 25 yocs observed beyond 1 kpc in the region @xmath62 pc are observed in the direction of @xmath63 . moreover , most of these young clusters are observed below gp and majority of them are located in the direction of @xmath64 . this could be due to low interstellar extinction in the galactic anti - center direction which is least around the longitude range @xmath65 ( neckel & kare 1980 , arenou , grenon & gmez 1992 , chen et al . based on the study of extinction towards open clusters from the same catalogue of dias et al . ( 2002 ) , we found the direction of minimum extinction towards @xmath66 below the gp ( jos05 ) . hence a lower extinction allows us to have a higher observed cluster density in the surrounding area of the @xmath66 as well as observable up to farther distance which reflected in our larger value of @xmath0 with the increase of the distance . therefore , we conclude that the larger @xmath0 values obtained with the bigger zone or greater distance is not due to internal variation in @xmath0 but due to our observational constraint . in general , we found a value of @xmath67 pc for the @xmath0 . since yocs are on an average more luminous than the older clusters and also possess a large number of ob stars hence lends us an opportunity to compare the results with the independent study using massive ob stars which are also a younger class of objects and confined very close to the gp . in the present analysis , we use 2367 ob stars which are strongly concentrated towards the gp as those of the yocs . however , a natural problem in the determination of @xmath0 is to separate the ob stars belonging to the gb with the lgd . the issue has already been dealt with a great detail by several authors ( taylor , dickman & scoville 1987 , comeron , torra & gomez 1994 , cabrera - cao , elias & alfaro 1999 , torra , fernndez & figueras 2000 ) . a recent model proposed by the eca06 based on the three dimensional classification scheme allows us to determine the probability of a star belonging to the gb plane or lgd . a detailed discussion of the method can be found in the eca06 and we do not repeat it here . though it is not possible to unambiguously classify the membership of the stars among two populations but to statistically isolate the gb members from our sample , we used the results derived for the gb plane by the eca06 through the exponential probability density function for the o - b6 stars selected from the hipparcos catalogue while we used an initial guess value of 60 pc and -20 pc for the scale height and @xmath0 respectively for the gp . since typical maximum radius of the gb stars is not greater than about 600 pc ( westin 1985 , comeron , torra & gomez 1994 , torra , fernndez & figueras 2000 ) , we search ob stars belonging to gb up to this distance only . following the eca06 method , we found that 315 stars out of 2367 ob stars of our data sample belong to the gb . further , 22 stars do not seem to be associated with either of the planes . in this way , we isolate 2030 ob stars belonging to the lgd which are used in our following analysis . a @xmath68 distribution of the ob stars is shown in fig . 6(a ) ( in the cartesian galactic coordinate system , positive @xmath69 represents the axes pointing to the galactic center and positive @xmath70 to the north galactic pole ) and their distribution in the gp as a function of galactic longitude is displayed in fig . a clear separation of the gb plane from the gp can be seen in the figure which follows a sinusoidal variation along the galactic longitude and reaches its lower latitude at @xmath71 . a number density in the logarithmic scale of the ob stars belonging to lgd is shown in fig 6(c ) as a function of @xmath43 where stars are counted in the bins of 20 pc . we derive a scale height of @xmath72 pc from the least square fit that is drawn by a continuous straight line in the same figure . maz - apellniz ( 2001 ) using a gaussian disk model determined a value of @xmath73 pc which is well in agreement with our result . however , reed ( 2000 ) derived a broad range of @xmath74 pc using o - b2 stars while eca06 estimates smaller value of @xmath75 pc using o - b6 stars which are more in agreement with the @xmath76 pc derived with the self - gravitating isothermal disk model of maz - apellniz ( 2001 ) . it is seen in fig . 6(b ) that the ob stars are sparsely populated around the gp in comparison of the yocs and a significant fraction of them are below @xmath77 pc . in order to study the @xmath0 distribution with @xmath42 , we here make four different zones normal to the @xmath56 plane considering the ob stars within @xmath57 pc , @xmath58 pc , @xmath78 and @xmath79 pc . the @xmath0 is estimated by the same procedure as followed for the yocs . a variation in the @xmath0 with @xmath42 is illustrated in fig . 7 where we have made a bin size of 50 pc . it is seen that @xmath0 derived in this way for the ob stars show a continuous decay with the @xmath42 as well as size of the zone which seems to be due to the preferential distribution of the ob stars below the gp . when we draw the spatial distribution of ob stars in the x - y coordinate system in fig . 8 , we notice that most of the ob stars are not distributed randomly but concentrated in the loose group of the ob associations . this difference in density distribution of ob stars could be primarily related with the star forming regions . the number of ob stars below the gp are always found to be greater than the ob stars above the gp in all the distance bin of 100 pc . however , in the immediate solar neighbourhood within 500 pc distance , ob stars below the gp are as much as twice than those above the gp . this is clearly a reason behind a large value of @xmath0 in the smaller @xmath42 value which systematically decreases as more and more distant ob stars are included . a mean value of @xmath80 pc was obtained by reed ( 2006 ) using the same catalogue of 2397 ob stars , albeit without removing the gb members . in fact this is also noticeable in the present study ( see big dot in fig . however , we can not give a fixed value of @xmath0 from the present analysis of the ob stars as it depends strongly on the @xmath42 as well as selection of the @xmath20 cut - off . it is normally assumed that the cluster density distribution perpendicular to the gp could be well described in the form of a decaying exponential away from the gp , as given by , @xmath81 , \ ] ] where @xmath0 and @xmath45 are the solar offset and scale height respectively . we determine @xmath0 by fitting the above function . for example in fig . 9(a ) , we have drawn @xmath20 distribution in 30 pc bin considering all the 537 yocs which lie within @xmath59 pc and @xmath82 kpc . since we have already derived the scale height for the yocs as 56.9 pc in our earlier section hence kept it fixed in the present fit . a least square exponential is fitted for all the distance limits . here we do not divide the data sample in different zones of @xmath20 as we have done in the previous section since only the central region of @xmath16 150 pc has significant effect on the determination of solar offset in the exponential decay method as can be seen in fig . our results are shown in fig . 9(b ) where we have displayed @xmath0 derived for the yocs as a function of @xmath42 . we can see a consistent value of about 13 pc for @xmath0 except when only yocs closer to 1 kpc from the sun are considered . this may be due to undersampling of the data in that region . our estimate is close to the bonatto et al . ( 2006 ) who reported a value of @xmath83 pc following the same approach , however , clearly lower in comparison of @xmath0 determined in the previous section . here , it is worth to point out that following the same approach pkss06 found a significantly large value of @xmath0 ( @xmath84 pc ) when considering only those clusters within @xmath85 . however , the value of @xmath0 substantially comes down to @xmath86 pc for the clusters in the age range of @xmath87 in their study . if we confine our sample to @xmath85 only , we find that @xmath0 increases marginally up to @xmath88 pc which is not quite different than our earlier estimate but still considerably lower than the pkss06 and we suspect that their values are overestimated by a significant factor . a similar study for the @xmath20 distribution of ob stars is also carried out and our results are shown in fig . 9(c ) , as an example , considering all the data sample . the resultant variation of @xmath0 for the different @xmath42 are shown in fig . it is clearly visible that @xmath0 varies in the range of 6 to 12 pc which is substantially lower in comparison of the values obtained in the previous method for the same data set . reed ( 1997 , 2000 ) also reported a similar lower value of @xmath26 6 to 13 pc for the @xmath0 using exponential model . a significant feature we notice here is that the @xmath20 distribution to the left and right of the peak do not seem symmetric particularly in the bottom half of the region where exponential fit in the @xmath89 region is higher than their observed value while reverse is the case for the @xmath90 region . therefore , a single exponential profile fit to the distribution of the ob stars for the whole range results in a large @xmath91 since points are well fitted only over a short distance interval around the mid - plane . this may actually shift @xmath0 towards the lower value which results in an underestimation for the @xmath0 determination . we believe that a single value of @xmath0 determined through exponential decay method is underestimated and needs further investigation . a distribution of clusters in the galactic longitude also depends upon the age ( dias & lpine , 2005 ) and it is a well known fact that the vertical displacement of the clusters from the gp is correlated with the age of the clusters . hence , one alternative way to ascertain the mean displacement of sun from the gp is to study the distribution of yocs and ob stars projected on the gp as a function of the galactic longitude where it is noticeable that the distribution follows an approximately sinusoidal variation . we estimated @xmath0 in this way in our earlier study ( jos05 ) although analysis there was based on the differential distribution of interstellar extinction in the direction of ocs . to study the variation of @xmath20 as a function of galactic longitude , we assemble yocs in @xmath92 intervals of the galactic longitude and mean @xmath20 is determined for each interval . here we again divide the yocs in three different zones as discussed in sect . 4 and the results are illustrated in fig . 10 where points are drawn by the filled circles . considering the scattering and error bars in mind , we do not see any systematic trend in the @xmath20 variation and a constant value of @xmath93 pc ( in negative direction ) are found for @xmath94 , @xmath95 and @xmath96 pc respectively . however , when we consider all the yocs including possible gb members as drawn by open circles in the same figure , we found a weak sinusoidal variation as plotted in fig . 10 by the continuous lines and has a striking resemblance with @xmath20 distribution at maximum galactic absorption versus longitude diagram ( fig . 8 of jos05 ) . we fit a function , @xmath97 to the @xmath98 distribution with @xmath0 estimated from the least square fits in all the three zones and resultant values are given at the top of each panel in fig . it is clearly visible that the @xmath0 estimated in this way varies between 17 to 20 pc and it is not too different for the case when gb members are excluded . the largest shift in the mean @xmath20 below the gp occurs at about @xmath99 which is the region associated with the gb ( see fig . 6(b ) ) as can be seen by the maximum shift between filled and open circular points in fig . 10 . in fig . 11 , we plot a similar variation for the ob stars in four different zones as selected in sect . 4 and it is noticeable that the sinusoidal variation is more promising for the ob stars . the values of @xmath0 ranges from 8.4 to 18.0 and like in all our previous methods , it shows a significant variation among different @xmath42 for the ob stars . it is interesting to note that mean @xmath20 shows a lower value in the vicinity of @xmath100 region in both the yocs and ob stars . pandey , bhatt & mahra ( 1988 ) argued that since the maximum absorption occurs in the direction of @xmath101 as well as reddening plane is at the maximum distance from the gp in the same direction of the galactic longitude , it may cause a lower detection of the objects . we also found a similar result in jos05 . in his diagram of the distribution of ocs as a function of longitude , van den bergh ( 2006 ) also noticed that the most minimum number of ocs among various dips lies in the region of @xmath101 where there is an active star forming region , sagitta . however , the lack of visible ocs are compensated by the large number of embedded clusters detected from the 2mass data ( bica , dutra & soares 2003 ) . we therefore attribute an apparent dip in @xmath0 around the region @xmath101 to the observational selection effects associated due to star forming molecular clouds which may result in the non - detection of many potential yocs towards far - off directions normal to the gp . the spatial distribution of the young stars and star clusters have been widely used to probe the galactic structure due to their enormous luminosity and preferential location near the gp and displacement of the sun above gp is one issue that has been addressed before by many authors . in the present paper we considered a sample of 1013 ocs and 2397 ob stars which are available in the web archive . their @xmath20 distribution around the gp along with the asymmetry in their displacement normal to the gp allowed us to statistically examine the value of @xmath0 . the cut - off limit of 300 myrs in the age for yocs has been chosen on the basis of their distribution in the @xmath102 plane . we have made an attempt to separate out the ocs and ob stars belonging to the gb from the lgd . in our study , we have attempted three different approaches to estimate @xmath0 using 537 yocs lying within 4 kpc from the sun . we have studied @xmath0 variation with the maximum heliocentric distance and found that @xmath0 shows a systematic increase when plotted as a function of @xmath42 , however , we noticed that it is more related to observational limitations due to galactic absorption rather that a real variation . after analysing these yocs , we conclude that @xmath67 pc is the best estimate for the @xmath0 . a similar value has been obtained when we determined @xmath0 through the @xmath20 distribution of yocs as a function of galactic longitude , however , a smaller value of about 13 pc is resulted through exponential decay method . considering the yocs within @xmath48 pc , we determined that the clusters are distributed on the gp with a scale height of @xmath47 pc and noticed that the @xmath0 has no bearing in the estimation of @xmath45 . a scale height of @xmath72 pc has also been obtained for the ob stars belonging to the lgd . a comparative study for the determination of @xmath0 has been made using the 2030 ob stars lying within a distance of 1200 pc from the sun and belonging to the lgd . it is seen that the @xmath0 obtained through ob stars shows a substantial variation from about 8 to 28 pc and strongly dependent on the @xmath42 as well as @xmath20 cut - off limit . it is further noted that @xmath0 estimated through exponential decay method for the ob stars gives a small value in comparison of the yocs and ranges from 6 - 12 pc . therefore , a clear cut value of @xmath0 based on the ob stars can not be given from the present study , however , we do expect that a detailed study of ob associations in the solar neighbourhood by the future gaia mission may provide improved quality and quantity of data to precisely determine @xmath0 in order to understand the galactic structure . this paper presents our attempt to study the variation in @xmath0 due to selection of the data and method of determination using a uniform sample of yocs and ob stars as a tool . it is quite clear from our study that the differences in approach and choice of the data sample account for most of the disagreements among @xmath0 values . this publication makes use of the catalog given by w. s. dias for the ocs and by b. c. reed for the ob stars . author is thankful to the anonymous referee for his / her comments and suggestions leading to the significantly improvement of this paper . the critical remarks by john eldridge are gratefully acknowledged . 99 abt h. a. , 2004 , apjs , 155 , 175 arenou f. , grenon m. , gmez a. , 1992 , a&a , 258 , 104 bergond g. , leon s. , guibert j. , 2001 , a&a , 377 , 462 bica e. , dutra c. m. , soares j. , barbuy b. , 2003 , a&a , 404 , 223 binney j. , gerhard o. , spergel d. , 1997 , mnras , 288 , 365 blitz l. , teuben p. , 1997 , the observatory , 117 , 109 bonatto c. , kerber l. o. , bica e. , santiago b. x. , 2006 , a&a , 446 , 121 cabrera - cao j , elias f. , alfaro e.j . , 2000 , apss , 272 , 95 caldwell j. a. r. , coulson i. m. , 1987 , aj , 93 , 1090 chen b. , vergely j. l. , valette b. , carraro g , 1998 , a&a , 336 , 137 chen b. , figueras f. , torra j. , jordi c. , luri x. , galad - enrquez d. , 1999 , a&a , 352 , 459 chen b. , stoughton c. , smith , j. a. , et al . , 2001 , apj , 553 , 184 chen w. p. , chen c. w. , shu c. g. , 2004 , aj , 128 , 2306 cohen m. , 1995 , apj , 444 , 874 comeron f. , torra j. , gomez a. e. , 1994 , a&a , 286 , 789 conti p. s. , vacca w. d. , 1990 , aj , 100 , 431 dias w. s. , alessi b. s. , moitinho a. , lpine j. r. d. , 2002 , a&a , 389 , 871 dias w. s. , lpine j. r. d. , 2005 , apj , 629 , 825 elias f. , cabrera - cao j. , alfaro , e. j. , 2006 , aj , 131 , 2700 ( eca06 ) fernie j. d. , 1968 , aj , 73 , 995 franco j. , tenerio - tagle g. , bodenheimer p. , rozyczka m. , mirabel i. f. , 1988 , apj , 333 , 826 friel e. d. , 1995 , araa , 33 , 381 gould b. a. , 1874 , proc . am . ass . adv . sci , 115 gum c. s. , kerr f. j. , westerhout g. , 1960 , mnras , 121 , 132 humphreys r. m. , larsen j. a. , 1995 , aj , 110 , 2183 hammersley p. l. , graon f. , mahoney t. , calbet x. , 1995 , mnras , 273 , 206 janes k. , adler d. , 1982 , apjs , 49 , 425 janes k. , phelps r. l. , 1994 , aj , 108 , 1773 joshi y. c. , 2005 , mnras , 362 , 1259 ( jos05 ) kraft r. p. , schmidt m. , 1963 , apj , 137 , 249 lada c. j. , lada e. a. , 2003 , araa , 41 , 57 lindblad p. o. , 1974 , in : stars and milky way system , ed . l. n. marvridis , springer - verlag lynga g. , 1982 , a&a , 109 , 213 lynga g. , 1987 , computer based catalogue of open cluster data , 5th ed . ( strasbourg cds ) maz - apellniz j. , 2001 , aj , 121 , 2737 mndez r. a. , van altena , w. f. , 1998 , a&a , 330 , 910 neckel t. , klare g. , 1980 , aas , 42 , 251 pandey a. k. , mahra h. s. , 1987 , mnras , 226 , 635 pandey a. k , bhatt , b. c. , mahra , h. s. , 1988 , a&a , 189 , 66 piskunov a. e , kharchenko n. v. , rser s. , schilbach e. , scholz r. -d . , 2006 , a&a , 445 , 545 ( pkss06 ) pppel w. g. l. , 1997 , fundamental of cosmic physics , 18 , 1 reed b. c. , 1997 , pasp , 109 , 1145 reed b. c. , 2000 , aj , 120 , 314 reed b. c. , 2006 , jrac , 100 , 146 taylor d. k. , dickman r. l. , scoville n. z. , 1987 , apj , 315 , 104 stothers r. , frogel j. , 1974 , aj , 79 , 456 torra j. , fernndez d. , figueras f. , 2000 , a&a , 359 , 82 torra j. , fernndez d. , figueras f. , comeon f. , 2000 , apss , 272 , 109 van den bergh , s. , 2006 , aj , 131 , 1559 westin t. n. g. , 1985 , a&as , 60 , 99 yamagata t. , yoshii y , 1992 , aj , 103 , 117	we have carried out a comparative statistical study for the displacement of the sun from the galactic plane ( @xmath0 ) following three different methods . the study has been done using a sample of 537 young open clusters ( yocs ) with @xmath1 lying within a heliocentric distance of 4 kpc and 2030 ob stars observed up to a distance of 1200 pc , all of them have distance information . we decompose the gould belt s member in a statistical sense before investigating the variation in the @xmath0 estimation with different upper cut - off limits in the heliocentric distance and distance perpendicular to the galactic plane . we found @xmath0 varies in a range of @xmath2 pc from the analysis of yocs and @xmath3 pc from the ob stars . a significant scatter in the @xmath0 obtained due to different cut - off values is noticed for the ob stars although no such deviation is seen for the yocs . we also determined scale heights of @xmath4 and @xmath5 pc for the distribution of yocs and ob stars respectively . [ firstpage ] galaxy : structure , open clusters , ob stars , gould belt method : statistical astronomical data bases
recently there have been exciting developments in astrophysics and cosmology , associated with observations of ultra high energy cosmic rays ( uhecr ) , of energies larger than @xmath0 ev @xcite , as well as high - redshift ( @xmath1 ) supernovae observations @xcite claiming a current - era acceleration of the universe . the latter observations , when combined with data from measurements of the cosmic microwave background radiation @xcite , which provide evidence for a spatially flat universe , with a @xmath2 , seem to indicate that up to @xmath3 % of the total energy density of the universe is attributed to a _ dark energy _ component , which could be a positive cosmological constant ( de sitter type universe ) @xmath4 or something else . on the other hand , the presence of uhecr events appears puzzling from the point of view of lorentz invariance @xcite . the latter symmetry would impose an energy threshold ( greitsen - zatsepin - kuzmin ( gzk ) cut - off ) @xcite above which certain reactions can occur which would prevent such uhecr events to reach the observation point , if one makes the physically plausible assumption that such events come from extragalactic sources @xcite . these data present serious puzzles and challenges for the quantum field theory as we know it . first , if there is a non - zero cosmological _ constant _ @xmath4 , ( de sitter phase in the universe ) , this would eventually mean that the ( constant ) vacuum energy will dominate over matter in the universe , given the @xmath5 relaxation of the latter with the expanding - universe scale factor @xmath6 . when the vacuum energy is dominant the universe will enter again a de sitter inflationary phase , with exponential growth , @xmath7 ( @xmath8 is the gravitational ( newton s ) constant ) . with such scale factors , there will be cosmic horizons , since a light ray in a robertson - walker type universe will take an infinite time to traverse a _ finite distance _ @xmath9 the presence of horizons imply the impossibility of defining pure asymptotic states , which are essential for the proper definition of a heisenberg scattering @xmath10-matrix , and thus a conventional quantum field theory in such gravitational backgrounds . this is a serious challenge also for string theory @xcite , which , by its very nature , is supposed to be a theory of s - matrix . second , a relaxation of the gzk cutoff -in order to explain the uhecr events- would , in turn , imply a relaxation of lorentz invariance @xcite , which is another drastic modification of field theory . along these lines it has also been suggested @xcite that deviation from lorentz symmetry , in the sense of modified dispersion relations for matter probes , can also explain the presence of tev cosmic photon events , which again should have been prevented from reaching the observation point within the standard kinematics stemming from lorentz invariance requirement during the scattering of energetic photons with the infrared background radiation . a question i would like to address in this talk is whether there is a unified framework in which the above puzzles can be tackled , with the simultaneous ability to make concrete , experimentally falsifiable , predictions . in addition , i will attempt to take one more challenge , and tackle , within the above framework , another important issue , that of the _ hierarchy _ between the currently claimed @xcite `` observed value '' of the `` vacuum energy '' , @xmath11 in planck @xmath12 units , and the _ supersymmetry breaking scale _ , of a few tev , in supersymmetric theories . for the purposes of my talk i will concentrate in the string theory framework , and present some toy models of ( non - critical ) strings @xcite where the above issues can be tackled , as i will argue , in a mathematically consistent way . the structure of the talk is the following : in section 2 i present a very brief overview of possible violations of lorentz invariance , and the associated consequences , from various theoretical viewpoints . in section 3 i discuss the issue of lorentz invariance in string theory . i start from critical string theory , where lorentz invariance is valid , and then proceed to explain carefully how non - critical string models may lead to its violation , in the sense of modified dispersion relations for matter probes ( refractive indices _ etc _ ) . in section 4 i describe briefly two concrete models with the above properties : ( i ) one model for space time foam in liouville strings , involving the interaction of string matter with stringy space - time defects , playing the rle of space time `` foam '' , and ( ii ) a non - critical stringy model of cosmology , where the non - criticality , which is here viewed as departure from equilibrium , is provided by an initial catastrophic event , playing the rle of big bang . the model consists of two colliding branes . in the context of this model i will discuss relaxation scenaria for the cosmological vacuum energy , inflation and graceful exit from it , and how an s - matrix can be ultimately salvaged by exiting from the current - era accelerating phase . the model involves two time - like variables , one of which is the liouville mode , required for consistency of the non - critical string . i adopt the point of view that lorentz symmetry is a good symmetry of any isolated quantum field theory model on flat space time , and that its possible violations come from either placing the system in a ` medium ' or heat bath ( finite temperature ) , or coupling it to ( certain models of ) quantum gravity , which also behave like a stochastic medium @xcite . a theory of quantum gravity must describe physics at planck scales , @xmath13 m. at such small length scales the structure of space time might be very different from what we have experienced so far , _ e.g. _ the space time might be discrete , non commutative _ etc_. therefore lorentz symmetry might not be valid at such small scales , or at least its form may be different from the familiar one characterising the relatively low energy scales of the current experiments ( for instance , the lorentz symmetry might be non - linearly realized at planckian scales ) . there are instructive examples from condensed matter physics , that probably help understand physically a possible breakdown of lorentz symmetry at higher energies . for instance , antiferromagnetic spin systems at low energies are known to be described effectively by _ relativistic _ continuum field theories , for instance the large spin @xmath14 limit of planar antiferromagnetic lattice spin systems yields a relativistic @xmath15 @xmath16-model with action @xmath17 , @xmath18 are magnons ( spin degrees of freedom ) , and the non - dynamical gauge field @xmath19 takes proper account of the correct number of physical degrees of freedom in the system . another instructive example is that of systems with fermionic quasiparticle excitations which have _ nodes _ in their fermi surface ( or energy gap ) ( c.f . figure [ fig : nodes ] ) . the ( average ) radius of the fermi surface provides the energy reference scale above which all energies of quasiparticle excitations are measured . low energies in this setting are therefore viewed those near the fermi surface . near the node , where the gap function vanishes @xmath20 , the dispersion relation of the quasiparticle excitation reads : @xmath21 , with @xmath22 the fermi velocity of the node , playing here the rle of a limiting velocity ( ` speed of light ' ) for the relativistic problem at hand . linearizing the fermionic quasiparticle spectrum about the node one may therefore obtain a relativistic low - energy field theory in the continuum limit consisting of dirac fermions : @xmath23 , @xmath24 . such a condensed - matter inspired approach has been adopted in the literature in an attempt to discuss the origin of lorentz symmetry in quantum field theory @xcite . in the physics of fundamental interactions deviations from lorentz symmetry may become manifest in modifications of the dispersion relations of matter probes . for massless probes such modifications imply non - trivial refractive indices due to the modification in the group velocity of the probe . for instance , in theories with non - trivial vacuum polarization at finite temperature , such as quantum electrodynamics _ etc . _ , one obtains non - trivial refractive indices already at one loop @xcite , in the sense of a modified dispersion @xmath25 with @xmath26 the temperature , and the function @xmath27 being determined from the vacuum polarization graph . it is this sort of modification that may be induced also by quantum gravity effects . indeed , one of the most fascinating ideas about quantum gravity is the suggestion made by john wheeler , and subsequently adopted by stephen hawking , that space time at planckian scales might acquire a foamy structure ( c.f . figure [ fig : foam ] ) , which may thus behave like a stochastic ` medium ' , with non - trivial macroscopic consequences for the propagation of matter probes in such backgrounds @xcite . this results in modified dispersion relation and other non - trivial `` optical '' properties . such modified dispersion relations are known to characterise non - critical ( liouville ) string theory models of space - time foam @xcite , where the departure from criticality ( conformality of the associated @xmath16-model background ) is viewed here as a consequence of the interaction of string matter with the ( quantum gravity ) foam : @xmath28 where the function @xmath29 is suppressed by higher powers of the quantum - gravity scale ( planck ( or string ) length @xmath12 ( @xmath30 ) ) . subsequently , similar phenomena have been argued @xcite to characterise the so - called loop gravity approach @xcite . the latter is a background independent theory of gravity , in which the space time is characterised by a ` polymer - like ' discrete structure . as a result of this property , the basic state of the theory , the `` weavy state '' , @xmath31 is such that , although macroscopically the ( emergent ) space time looks minkowski , nevertheless there is non - trivial structure at scales smaller than the characteristic scale @xmath32 where quantum gravity effects set in : @xmath33 , where @xmath34 is the energy of the probe . such effects lead to non - trivial dispersion relations , similar in nature to those predicted in non - critical strings @xcite . there is an important difference , though , which distinguishes experimentally strings from these theories : in loop gravity superluminal propagation of matter probes is allowed , implying birefringence effects . such superluminal signals are forbidden in the liouville ( non - critical ) string approach to quantum gravity , for specifically stringy reasons @xcite . there is an important difference of the modification of the dispersion relation induced by quantum gravity , as compared to conventional vacuum polarization effects @xcite , in that the former increases with the energy of the probe , in contrast to the conventional field theoretic effects which decrease with energy . this allows a possible disentanglement of such effects in experiments . one of the most sensitive probes of such non - trivial optical properties is the observation of light from gamma ray bursters ( grb ) @xcite or gamma ray flairs @xcite . as mentioned in the introduction , such modified dispersion relations , have been used subsequently to explain the transparency of the universe in ultra high energy cosmic rays @xcite or tev photon events @xcite . for a recent and comprehensive review of astrophysical tests of such quantum - gravity - induced modified dispersion relations see @xcite . finally , modified dispersion relations for matter probes have been recently @xcite argued to characterise flat space theories in certain models in which the planck `` length '' is considered a real length . as such it should be transformed under usual lorentz boosts . the requirement that the planck length be , along with the speed of light , also _ observer independent _ leads quite naturally to modified `` lorentz transformations '' , and also to modified dispersion relations for particles . this approach is termed `` doubly special relativity '' ( dsr ) @xcite . the dsr approach should be distinguished from the other approaches , mentioned so far , where the planck length scale is associated to a ` coupling constant ' of the theory , and thus is observer independent by definition . this is clearly the case of einstein s general relativity , where the planck scale is related to the universal gravitational coupling constant ( newton s constant ) , and in string theory ( critical and non - critical ) , where the planck length is related to the characteristic string scale of the theory , which is also independent of inertial observers . there are two approaches so far in the problem of formulating dsr theories : in the first @xcite one postulates a modified dispersion @xmath35 , where @xmath36 is the inertial - observer independent length , proportional to the conventional planck length @xmath37 . the approach is as yet at a phenomenological level , and hence one can choose appropriate functions @xmath38 . let me illustrate the main features with one such choice : @xmath39 . the generators of the deformed lorentz boosts ( assumed , for definiteness and simplicity , to be along @xmath18 direction ) , that leave @xmath40 observer independent , read @xcite : @xmath41 as mentioned previously an important consequence of the formalism is that by construction there are modified dispersion relations : @xmath42 where @xmath43 is the rest mass of a matter point - like probe , defined by the appropriate quadratic casimir operator . note that @xmath43 is different from the _ rest energy _ @xmath44 , the latter defined as the energy at zero momentum @xmath45 : @xmath46 in this particular choice of @xmath38 note that the rest mass differs from the rest energy only by terms of order @xmath47 and higher : @xmath48 . it was remarked in @xcite that the modified dispersion implies a group speed of photon @xmath49 which diverges as @xmath50 . however i note here that the meaning of @xmath50 is not clear , especially in a theory with gravity , where in principle one expects the whole structure of space time to be changed drastically at energies close to planck scale , which thus acts as an effective ultraviolet cutoff . in the second approach @xcite , one combines lorentz boosts with dilatations in order to achieve the observer - independence of the planck length ( we assume again the boost along @xmath18-direction for comparison with the previous approach ) : @xmath51 + e\frac{\partial}{\partial p_z } - { \tilde \ell}_p p_z \left(e\frac{\partial}{\partial e } + p_j \frac{\partial}{\partial p_j}\right ) \label{modboost2}\ ] ] where the first term inside square brackets denotes the boost @xmath52 , and the rest the dilatation @xmath53 . the combined dilation and boost algebra closes with the three - rotations : @xmath54=\epsilon^{ijk}k_k~ , [ k^i , k^j]=\epsilon^{ijk}j_k~.\ ] ] there is a modified dispersion relation in this approach @xcite : @xmath55 notice that in this second approach , in contrast to that of @xcite , all energies are necessarily smaller than the ` gravity scale ' , @xmath56 , given that at such scales there appears to be a ` metric ' collapse in the dispersion relation . this approach is thus closer to the conventional understanding of the planck scale as an ultraviolet cutoff for any low - energy theory . another important feature is that the speed of light or , in general , of massless particles is @xmath57 , as in special relativity of einstein . again , the rest mass @xmath58 is different from the rest energy @xmath44 , but here the difference already occurs already at linear order in @xmath40 : @xmath59.it should be noted that there are experimentally testable differences between the dsr models of @xcite and @xcite , e.g. in connection with kinematical conditions for particle production in collisions @xcite . it remains to be seen what consequences these modified flat space transformations would have in a general relativistic framework . for instance , from ( [ moddisp2 ] ) one would be tempted to identify such effects with those in a specific curved space - time background , which as i mentioned above , seems to collapse at planck energies . are , then , such dsr models , when placed in a general relativistic context , selecting special metric backgrounds as , for instance , is the case of the modified dispersion relations obtained in non - critical string inspired models @xcite ? these and other questions , related to a reconciliation of dsr ( viewed as theories with a minimum length ) with quantum mechanics , should be pursued in future research . at this point i end this brief overview of possible theoretical models predicting violations or modifications of the lorentz symmetry . in what follows i will concentrate on specific models of such violations stemming from ( non - critical ( liouville ) ) string theory @xcite . let me start from `` old '' string theory @xcite , living in a critical dimension of target space time . from a first quantization viewpoint the motion of strings is described by a @xmath16-model , a two - dimensional world - sheet field theory , in background target - space fields . there is a basic symmetry , called conformal symmetry , which allows a consistent path integral formulation of the @xmath16-model , and restricts the background fields to their target - space classical equations of motion , thereby providing an important link between consistent world - sheet quantum geometry and target space dynamics . it is this symmetry that implies lorentz invariance in critical number of dimensions for flat space times , and the standard dispersion relations of stringy excitations . let us briefly see how . consider a bosonic @xmath16-model in flat space time , perturbed by certain tachyon excitations ( although tachyons are viewed as instabilities of bosonic strings , and are absent in superstrings , nevertheless we consider them here for simplicity and because they capture all the essential features we wish to discuss , which can be immediately generalized to the superstring case ) : @xmath60 where @xmath61 is a @xmath62-dimensional target space momentum , @xmath63 denotes the world sheet , @xmath64 are world - sheet derivatives , and @xmath65 is the regge slope @xcite , related to the square of the characteristic length scale of the string @xmath66 , which is observer independent from a target - space viewpoint . in old string theory this scale was identified with the four dimensional planck mass scale , although in the modern membrane ( d - brane ) extension the two scales may not be the same @xcite , the exponential vertex operator @xmath67 , which described the tachyonic excitation of the bosonic string spectrum , with polarization `` tensor '' @xmath68 , is not in general a marginal operator in a world - sheet renormalization group sense , and hence under the two - dimensional quantum field theory corrections will break the conformal invariance in the sense that its conformal dimension @xmath69 is different from one : @xmath70 insisting on conformal invariance , i.e. marginality of the operator , one obtains the standard dispersion relation ( _ on - shell condition _ ) for the tachyonic excitation compatible with lorentz invariance in target space : @xmath71 with the tachyon rest mass @xmath72 ( the negative sign indicates the abovementioned tachyonic instability of the bosonic string , but this does not affect the arguments on dispersion relation because the same procedure applies to all other string excitations ) . the number of target - space dimensions @xmath62 is also restricted by the same requirement of conformal world - sheet invariance if one performs the path integral computation of the trace of the world - sheet stress tensor @xmath73 in curved world sheets . even for a free string , then , one finds that @xmath74 , where @xmath75 denotes a two - dimensional path integral average , @xmath76 is the world - sheet curvature , and the -26 appeared due to the fact that world - sheet reprametrization invariance acts from a two - dimensional viewpoint as a gauge symmetry which needs fixing , thereby implying fadeev - popov ghosts . their contribution to the conformal anomaly is then given by -26@xmath76 . to ensure conformal invariance one must have @xmath77 which select the number of space time dimensions for the bosonic string to 26 ( for the superstring this number reduces to 10 ) . critical string theory has proven a very successful and elegant formalism so far to understand many questions related to the structure of space time at planckian distances , to count microstates in singular space times such as black holes _ etc_. however , there are some situations which critical strings appear inadequate to describe . these include formation of black holes and other dynamical space - time boundaries , stochastic space - time foam backgrounds ( if one adopts the point of view that the latter exist ) , recoil effects of stringy solitons , and in general situations which involve the change of a background over which the string propagates . critical strings , and conventional conformal field theories , describe _ fixed background _ situations . since , however , strings are supposed to be a theory of space time itself , they should be able to provide in principle an answer as to how the space time was created , and for this reason they should exhibit background independence in some sense . this is an advantage of the loop gravity approach @xcite which start from abstract fundamental units ( `` spin networks '' ) and then proceeds to construct a dynamical space time out of them . one may argue that such background independence issues or background changes in string theory can only be tackled in the context of string field theory , which is not developed as yet to a satisfactory level of precision or computational power . one , however , may be less ambitious and attempt to tackle such issues perturbatively at first instance , within a first quantization @xmath16-model approach . the key to this is probably abandoning _ conformal invariance _ which fixes the background , and use instead _ non - critical _ ( liouville ) strings @xcite , which is a mathematically consistent way to deal with @xmath16-models away from their conformal ( fixed ) points on the space of string theories ( c.f . figure [ fig : flow ] ) . i will argue below that such a procedure describes consistently background changes in string theory , and implies , under certain circumstances , modified dispersion relations stemming from spontaneous breaking of lorentz symmetry . to understand better non - critical strings we should first mention a few things about theory space . this is considered as the space of all possible background target - space - time string configurations , which is therefore an infinite dimensional manifold . the manifold is endowed with a metric , which is nothing other than the zamolodchikov metric of a conformal field theory @xcite given by the appropriate two point function of the vertex operators describing the non - conformal deformations : @xmath78 , where the vertex operators are deforming the @xmath16-model as : @xmath79 here @xmath80 is a fixed - point @xmath16-model background action , and @xmath81 are the non - conformal backgrounds / couplings corresponding to the vertex operators @xmath82 . in stringy models of interest to us here @xmath83 i.e. is a set of target - space fields . the non - criticality of the deformation is expressed by the non - triviality of the renormalization - group ( rg ) @xmath84-function of @xmath81 on the world - sheet : @xmath85 , where @xmath22 is a world - sheet scale provided by the world - sheet area @xcite . perturbation theory requires that one lies close to fixed point , which implies that one should work with @xmath86 expandable in power series in the couplings @xmath81 . quadratic order is sufficient for our purposes here , and to this order the @xmath84-function read : @xmath87 , where no sum is implied in the first term , @xmath88 is the anomalous dimension , and @xmath89 are the operator product expansion ( ope ) coefficients . an important property of the _ off - shell _ @xmath86 functions of stringy @xmath16-models is their gradient flow form @xcite i.e. the fact that they can be derived as @xmath81-space gradients of a flow function @xmath90 $ ] . this function is a renormalization - group invariant function and coincides with the running central charge of the theory between fixed points . one has the following relations for its flow along a renormalization - group trajectory , the celebrated zamolodchikov @xmath91-theorem @xcite : @xmath92}{\partial t } = -\frac{1}{12}\beta^i { \cal g}_{ij}\beta^j~ , \qquad \beta^i = { \cal g}^{ij}\frac{\partial c[g , t]}{\partial g^j } \label{ctheorem}\ ] ] in unitary @xmath16-model the metric @xmath93 is positive definite and hence the flow is such that the @xmath90 $ ] acts like a thermodynamic @xmath94 function , decreasing _ monotonically _ along the direction of the flow . at fixed points @xmath95}{\partial g^j}=0 $ ] and taking into account the positive - definiteness of the zamolodchikov metric one then obtains that the conformal invariance conditions are equivalent to target - space equations of motion for the flow function @xmath90 $ ] . the latter , thus , plays the rle of a low - energy effective action for the string theory at hand , which notably includes gravitational interactions . at fixed points the flow function becomes the central charge characterising the two - dimensional conformal field theory . in critical string theories with target space - time interpretation the flow function becomes 26 for bosonic strings ( or 10 in superstrings ) . for non - unitary @xmath16-models , such as stringy @xmath16-models with time - like @xmath96 fields , and time - like dilaton couplings , the flow may not be monotonic all the way , but one may still expect @xcite an overall decrease of the running central charge from one fixed point to the other this is due to the information `` loss '' beyond the ultraviolet cutoff of the underlying world - sheet quantum field theory . in fact the central charge counts degrees of freedom of the system , and hence there must be an inherent irreversibility if there is a cutoff scale . in non - critical strings with time - like fields , the running central charge does make oscillations before settling to a fixed point @xcite . this will become important later on ( section 4 ) when we discuss cosmology in this context . let me now argue how non - critical strings become consistent world - sheet ( two - dimensional ) field theories upon liouville dressing @xcite . the liouville mode @xmath97 seizes to decouple in a non - critical string which is characterized by a ( rg running ) central charge deficit @xmath98 - c^\right)$ ] caused by departure from criticality due to a given deformation ( for strings with a target - space interpretation @xmath99 ( 9 for superstrings ) ) . a detailed analysis @xcite shows that the liouville dressed theory is described by a @xmath16-model action : @xmath100 where @xmath101 are the liouville dressed couplings and the gravitational anomalous dimension @xmath102 is chosen in such a way so that the conformal dimension of the dressed deformation operator @xmath103 becomes one ( marginal ) . note that this marginality is exact , including interactions ( @xmath89 _ etc _ in the @xmath84-function ) and this constitutes the main rle of the liouville dressing , _ the restoration of conformal symmetry _ , in such a way that the non - critical , rg flowing @xmath16-model , becomes conformal at the expense of the introduction of the liouville mode @xmath97 in the path integral . from the quadratic equation that the gravitational anomalous dimensions satisfy ( [ dressingliouv ] ) it becomes evident that there are in general two solutions : @xmath104 usually in liouville theory ( central charge @xmath105 ) one keeps only the @xmath106 solution , given that @xmath107 corresponds to states that `` do not exist '' in the sense of not having a well - defined ( bounded , normalizable ) behaviour in the semiclassical limit where the central charge @xmath108 . however in supercritical string theory ( central charge @xmath109 ) this is not the case and one may , and in some circumstances does , keep both dressings . we shall encounter a situation like this in section 4 , when we discuss specific examples of liouville strings . upon redefining the liouville mode @xmath110 one may arrive at _ liouville - generalized conformal invariance conditions _ @xcite : @xmath111 - 25)\beta^i(\lambda^i(\phi ) ) \label{liouvcond}\ ] ] where the overdot denotes differentiation with respect to the normalized liouville mode @xmath112 ( [ normal ] ) . from ( [ dressingliouv ] ) it is obvious that the liouville mode plays the rle of an extra target - space dimension , which is _ time - like _ if the string is supercritical @xcite @xmath113 and _ spacelike _ is the string is subcritical . it is a general property of liouville strings that during their flows they remain either subcritical or supercritical @xcite . in what follows we shall concentrate on deformations that induce supercriticality of the string , in which case the liouville mode may be identified with the target time @xcite . in fact , what one identifies with the target time is the world - sheet zero mode of the normalized field @xmath114 in what follows i shall describe the most important physical consequences of such an identification . from ( [ liouvcond ] ) it becomes clear that the restoration of conformal symmetry leads to dynamical equations for the background fields of the non - critical @xmath16-model that have a `` frictional '' form , the damping being provided by the ( square root of the ) central charge deficit @xmath115 . although the sign of @xmath115 is fixed , however the sign of its square root @xmath116 is not determined in liouville strings , and as we shall see in section 4 , one may encounter situations @xcite in which @xmath117 evolves in such a way so as to change sign ( in such a case the string passes through a metastable critical point in theory space ) . from ( [ dressingliouv ] ) it also becomes clear that after the identification ( [ identify ] ) there will be a dilaton coupling in target space ( defined as the background coupling of the world - sheet curvature term in the @xmath16 model action ) @xmath118 which will contain linear dependences on target time for @xmath116 constant , but in general will be a complicated function of time . string theories in such time - like supercritical backgrounds can be formulated as consistent conformal field theories only for some special cases of constant @xmath116 ( linear dilatons ) @xcite but the corresponding s - matrix aspects are problematic . one can define factorisable and modular invariant amplitudes ( @xmath10-matrix ) in such a case only for certain _ discrete _ constant values of @xmath116 , as required for consistency of the mapping of the corresponding conformal theory to a coulomb gas representation with appropriate screening charges at infinity @xcite . nevertheless it is possible to consider more general situations , with continuously varying @xmath116 , provided that one abandons the concept of a factorisable scattering amplitude @xcite . to understand this latter point let us consider a generic correlation function among @xmath119 vertex operators @xmath82 of a liouville string : @xmath120 where @xmath112 is the normalized liouville mode ( [ normal ] ) , and @xmath115 denotes the central charge deficit , quantifying the departure of the non - conformal theory from criticality @xcite . a detailed analysis @xcite shows that , upon performing the world - sheet zero - mode @xmath121 integration of the liouville mode @xmath112 in ( [ liouvcor ] ) , one obtains that the dominant contributions to the path integral are proportional to @xmath122 where the prime in @xmath123 indicates the absence of world - sheet zero mode @xmath124 of the liouville field @xmath112 , with @xmath125 the world - sheet area ( in units of the ultraviolet cutoff of the @xmath16-model ) , @xmath126 and @xmath102 are the liouville anomalous dimensions ( [ dressingliouv ] ) , and @xmath127 is the anomalous dimension of the so - called identity operator @xmath128 . in order to compute the integral for arbitrary @xmath129 ( and hence @xmath116 ) it is necessary to make an analytic continuation to positive integer @xmath130 which immediately calls for regularization of the prefactors @xmath131 . this can be achieved @xcite by representing the integral over the liouville zero mode by a steepest - descent contour of @xmath121 as indicated in fig . [ fig : contour ] . the interpretation of the liouville zero mode as the target time @xcite implies a direct analogy of this contour with _ closed time like paths _ in _ non - equilibrium _ field theories @xcite . it can be seen @xcite that the world - sheet ultraviolet limit of the contour @xmath132 is ill defined and a regularization is needed ( dashed line in the contour of fig . [ fig : contour ] ) . this can be inferred by considering infinitesimal weyl shifts of the world - sheet metric ( which is integrated over ) in the correlator ( [ liouvcor ] ) . such a divergent behaviour implies that the generic liouville correlator ( [ liouvcor ] ) is not an s - matrix amplitude , as conventionally assumed in critical strings , but rather a $ matrix element connecting asymptotic density matrices @xcite @xmath133 $ @xmath134 with $ @xmath135 as a result of the world - sheet ultraviolet divergences . from a target space time view point small world - sheet areas may be seen as an infrared limit of a target space theory , where the strings shrink to their point - like limits . thus , in this picture , the infrared divergences in target space are held responsible for the failure of factorisability of the liouville string $ matrix . it is this feature that , together with the _ irreversibility _ of the world - sheet renormalization group flow function , expressed by the c - theorem ( [ ctheorem ] ) , makes manifest the _ non equilibrium nature _ of non - critical string theory . there is thus a direct analogy of liouville strings with open systems in which there is an entropy change ( here the entropy is associated with the flow function @xmath90 $ ] @xcite ) . further support of the non - equilibrium nature comes from the modification of the evolution equation of the density matrix of string matter moving in non - critical string backgrounds . renormalization - group considerations on the @xmath16-model theory , and in particular renormalization - group invariance of physical target space quantities , imply the following rg flow equation for the density matrix of low - energy string matter @xmath136 @xcite : @xmath137 + : \beta^i { \cal g}_{ij}[g^j , \rho]:\ ] ] where the quantization is achieved in this framework by summing up world - sheet topologies . in fact it can be shown that canonical quantization in the case of liouville strings is possible , given that the generalized conformal invariance conditions ( [ liouvcond ] ) satisfy the helmholtz conditions for being derived from an off - shell lagrangian in @xmath81 space @xcite . note the similarity of ( [ modifiedevol ] ) with the evolution of open dynamical systems . thus we see that the non - criticality of the string , expressed by the non - vanishing of the world - sheet rg @xmath84-functions , leads to the existence of `` environment '' for the subsystem of the low - energy string matter . in fact an important comment is in order here : there are quantum ordering ambiguities of the last term due to the fact that @xmath138 are complicated power series in the operators @xmath81 which do not commute with @xmath136 . one might have thought that these ambiguities may be resolved by postulating an ordering that respects formally the lindblad form of the interaction of the string matter with its non - critical string environment . however , things may not be that simple . it must be noted that there is a hidden non - linearity in the evolution equation ( [ modifiedevol ] ) due to the string wavefunction dependence of the zamolodchikov metric @xmath139 . when one writes the perturbative expansion @xmath140 this dependence is hidden in the coefficients @xmath141 where @xmath142 denote free liouville string correlators . such non - linearities may imply significant deviations from standard linear lindblad environmental entanglement @xcite . the situation should be examined case by case unfortunately , as it seems that general statements on this issue can not be made at this stage . to study the induced modifications on the dispersion relations of string excitations in the case of non critical deformations we concentrate on the generic liouville dressing ( [ dressingliouv ] ) . upon the identification of the liouville mode with time ( [ identify ] ) we may rewrite this relation as : @xmath143 where @xmath144 depends on the ope coefficients @xmath89 which encode the non - critical interactions of the string probe with the `` environment '' . in string theory the ope , which are related to string target - space scattering amplitudes , are power series in @xmath145 for closed strings , and @xmath146 for open strings . let us concentrate in the case of closed strings for definiteness , and also for physical concreteness , given that closed strings include gravity . we restrict ourselves to _ massless modes _ for our purposes below . to order @xmath147 ( i.e. close to a fixed point ) , where we restrict ourselves for our purposes in this article , the ope are assumed of order @xmath145 . on the other hand , for vertex operators proportional to @xmath148 ( in fourier space ) , describing the string excitation spectrum in flat space times , the anomalous dimensions @xmath149 . this can be found from ( [ dressings ] ) by noting that the central charge deficit @xmath115 is determined from the c - flow theorem ( [ ctheorem ] ) @xmath150 , since @xmath151 for reasons stated previously . thus , for massless closed string excitations one has the estimate : @xmath152 the complete liouville - dressed vertex operator , therefore , describing the massless string excitation reads : @xmath153 the coefficient of the liouville mode ( time ) is defined as the energy of the probe , and hence from ( [ fullvertex ] ) one obtains a modified dispersion relation for massless string excitations @xcite : @xmath154 where the coefficient @xmath155 and its sign is to be determined in specific models of non - critical strings , and the string scale @xmath156 may be taken to be the planck scale @xmath157 of quantum gravity ( but , we repeat again , in modern versions of string theory where our world is viewed as a string membrane @xcite @xmath66 is an arbitrary free parameter , which is in general different from the four - dimensional planck scale @xmath157 ) . from the modified dispersion one obtains a refractive index for the massless probe in the sense of its group velocity not being equal to one : @xmath158 which implies violation of lorentz invariance . the latter should not come as a surprise given the non - equilibrium stochastic nature of the non - critical string ( c.f . ( [ modifiedevol ] ) ) which resembles an open system in which string matter propagates in a ` stochastic ' dissipative way ( c.f . ( [ liouvcond ] ) ) . however , in our approach @xcite the lorentz violation is _ spontaneous _ in the sense that the full string theory may be critical , and the non - criticality is only a result of restricting oneself in an effective low - energy theory ( `` ground state '' not respecting the symmetry ) . in this last sense the presence of the liouville mode is a _ collective _ description of `` environmental '' effects associated with degrees of freedom essentially unobserved by a local observer @xcite . the precise nature of such degrees of freedom can be unveiled in realistic concrete examples of non - critical strings with physical significance . we examine two such examples in the next section . we describe below a toy model for stringy space - time foam @xcite , consisting of a bulk space time `` punctured '' with d(irichlet)-0-brane point - like defects . the latter are solitonic states in string theory @xcite , and as such they are not simply structureless point - like objects , but have substructure in the sense of an infinity of internal degrees of freedom corresponding to the massive string states . once a closed string , representing light string matter in the model , is scattered off the defects , the latter recoil and distort the surrounding space time . the distortion is expressed through the dynamical formation of ( unstable ) horizons , the interior of which is found to have a different refractive index from the exterior flat minkowski space time . the situation is schematically depicted in figure [ fig : dfoam ] . the mathematical formalism is based on a conformal field theory treatment of recoil in d - brane theory @xcite which allows for changes of @xmath16-model backgrounds . this can be achieved by the so - called logarithmic conformal field theories ( lcft ) , which lie on the border line between conventional conformal field theories and general two - dimensional field theories , and can still be classified by conformal data . below we review briefly the main results . a d - particle , recoiling with velocity @xmath159 , and initially positioned at @xmath160 , can be viewed as a d - particle with open string excitations attached to it ( c.f . middle figure [ fig : dfoam ] ) , which from a @xmath16-model point of view are described by adding to the free string @xmath16-model the following boundary ( world - sheet disk ) deformations @xmath161 where @xmath162 is a normal world - sheet boundary derivative , and we assumed that the d - particle was initially _ at rest_. these operators form a lcft pair of operators @xcite . in the above formula @xmath163 for @xmath164 is a regularized heaviside operator , with @xmath165 the ordinary heaviside function . the coordinate @xmath96 obey neumann boundary conditions on the world - sheet boundary @xmath166 , whilst @xmath167 , @xmath168 spatial index , obey dirichlet boundary conditions , and express the fact that the d - particle defect can trap strings with their ends attached on it ( c.f . figure [ fig : dfoam ] ) . in the limit @xmath169 which we consider here the first term in ( [ recoil ] ) is subleading and from now on we ignore it . we next extend the discussion to motion of string matter through a gas of moving d - particles @xcite , as is likely to be the case for a laboratory on earth , e.g. , if the d - particle foam is comoving with the cosmic microwave background ( cmb ) frame . assuming this to be moving with three - velocity @xmath170 relative to the observer , the recoil deformation takes the following form to leading order in @xmath169 : @xmath171 where the suffix @xmath172 denotes quantities in the boosted frame . the recoil velocity @xmath173 depends in general on @xmath172 , and is determined by momentum conservation during the scattering process , as discussed in @xcite . the main novelty in the @xmath174 case is that now there are two @xmath16-model operators in ( [ changerec ] ) . the deformations ( [ recoil ] ) and ( [ changerec ] ) are relevant world - sheet deformations in a two - dimensional renormalization - group sense , with anomalous scaling dimensions @xmath175 @xcite . their presence drives the stringy @xmath16-model non - critical , and requires dressing with the liouville mode @xmath112 . as mentioned previously , in section 3 , in liouville strings there are two screening operators @xmath176 , where the @xmath177 are the liouville - string anomalous dimensions given by : @xmath178 and the central charge deficit @xmath116 was computed in @xcite with the help of the c - theorem ( [ ctheorem ] ) , and found to be of higher order than @xmath179 . hence @xmath180 . the @xmath107 screening operator is sometimes neglected because it corresponds to states that do not exist in liouville theory . however , this is not the case in string theory , where one should keep both screenings as above this is essential for recovering the correct vanishing - recoil limit in the case of infinite d - particle mass . the liouville - dressed boosted deformation then reads : @xmath181 using stokes theorem , and ignoring terms that vanish using the world - sheet equations of motion , one arrives easily at the following bulk world - sheet operator : @xmath182 where the @xmath183 denote terms subleading as @xmath169 as we shall explain below . notice the cancellation of the terms proportional to @xmath172 due to the opposite screening dressings . recalling that the regularised heaviside operator @xmath184 , where @xmath185 is the standard heaviside function , we observe that one can identify the _ boosted time coordinate _ @xmath186 with the liouville mode @xmath112 : @xmath187 @xcite . at long times after the scattering , the liouville - dressed theory leads to target - space metric deformations of the following form to order @xmath179 : @xmath188 as explained in detail in @xcite , at the long times after the scattering event when the @xmath16-model formalism is valid , one has @xmath189 : @xmath190 and the world - sheet zero mode of @xmath112 , @xmath121 , are not independent variables , as @xmath190 is linked with the world - sheet renormalization - group scale . the lorentz breaking effects are proportional to @xmath173 , which is determined by momentum conservation and hence is independent of @xmath172 in the non - relativistic limit . in the case of photon scattering off d - particles , one would have @xmath191 . _ our form of lorentz violation exhibits galilean invariance _ for small @xmath172 . this is an important feature which differentiates our liouville foam model from generic lorentz - violating models of space time foam considered in the literature @xcite . for our purposes , the main effect of our d - particle model for space - time foam is the modification of the dispersion relation for an energetic particle , which causes it to propagate more slowly than a less energetic particle . the formation of horizons , mentioned earlier , and depicted in figure [ fig : dfoam ] , is obtained upon considering quantum fluctuation effects , which express the probability of finding a particular defect configuration as a fluctuation around a mean value . formally , one can evaluate quantum fluctuations in the recoil process by making an appropriate resummation over the world - sheet genera , which lead to stochastic probability distribution functions in the @xmath16-model path integral @xcite . performing appropriate coordinate transformations , which from a @xmath16-model view point amount to redefining path integral variables , and averaging over quantum fluctuations one can map the metric ( [ metricdressed ] ) in the limit @xmath192 into one which has an expanding horizon surrounding the defect at : @xmath193 , with @xmath194 , @xmath195 is the kinetic recoil energy of the non - relativistic d - particle , @xmath196 is the string coupling , assumed weak for the validity of the @xmath16-model perturbative expansion , and @xmath197 is the d - brane mass scale with @xmath30 the string scale . often one assumes @xmath30 close to the four dimensional planck mass of @xmath198 gev , although in general @xmath199 are parameters which can be constrained by data @xcite . inside the horizon , the space - time metric has the form : @xmath200 and one can match this space - time with a flat external minkowski space - time , as explained in detail in @xcite . the internal ` bubble ' metric configuration ( [ metricrecoil ] ) violates the positive energy conditions and hence is _ unstable_. thus a ` breathing bubble ' picture emerges from such a world - sheet path integration of the liouville zero mode @xcite , in which the average lifetime of the bubble is planckian , during which the bubble expands to its maximum size , which is of order the planck length , and then recontracts . we have identified three possible experimental implications of the formation of such quantum space - time bubbles . _ first _ , low - energy particles may get trapped inside the bubble , and their absorption within the space - time foam provides an explicit mechanism for the apparent loss of information accessible to external observers . _ secondly _ , in the case of electrically - charged particles , their non - uniform ( spiral ) motion inside the bubble causes the emission of radiation , which may escape in the form of photons . since the bubble interior has a non - trivial refractive index @xcite , these particles exhibit transition radiation , i.e. , the emission of photons accompanying an electrically - charged particle when it crosses an interface separating two media with different refractive indices . a fraction of this radiation escapes from the bubble , yielding photons that accompany the charged particle . _ thirdly _ , propagation inside the bubble causes the particle to travel more slowly , as seen by an external observer . this effect is equivalent to the retardation induced by the modification ( [ metricdressed ] ) of the space - time metric , as the d - brane defect recoils when struck by the propagating particle . the refractive index in this d - brane foam can be computed as @xcite @xmath201 where @xmath34 is the kinetic energy transfer of the matter probe during the collision with the d - particle defect . the index ( [ branerefr ] ) is always _ subluminal _ , thereby differentiating the model from certain models of loop gravity where superluminal propagation also occurs @xcite . the subluminal nature is due to specific stringy properties of the model , in particular it stems from the fact that the effective action describing the recoil excitations is of born - infeld form of non - linear `` electrodynamics '' @xcite , and hence there is a limiting ( light ) velocity for signal propagation . for astrophysical and other tests of these string - inspired dispersion relation see @xcite . we should mention , though , that the non - critical string model of space time foam of @xcite , despite its minimal planck - scale suppression of the effects , seems to avoid severe constraints , especially from atomic physics experiments @xcite , which seem to exclude the loop - gravity minimal suppression models . thus , for this model of foam , grb arrival - time tests @xcite seem , as yet , to provide the most sensitive probes . as mentioned in the introduction , there has been recent exciting evidence from astrophysical observations on high redshift type i supernovae @xcite that our universe is currently accelerating . the evidence has been seconded by cosmic microwave background data @xcite pointing towards total flatness of our universe @xmath202 , a result which , when combined with the estimate for the total matter contribution ( including dark matter ) @xmath203 , implies that there is a dark energy component in our universe which is 70 % of the total energy density . this is in agreement with the supernovae data implying acceleration of the universe , and implies that the universe is currently in a phase where this dark energy component begins to take over . these data are in agreement with naive cosmological constant in the universe ( de sitter ) , but this issue present an enormous theoretical challenge , due to the cosmic horizon problem ( [ cosmichorizon ] ) and its incompatibility with the definition of a scattering matrix @xcite , and hence string theory . as we have seen in section 3 , liouville non - equilibrium strings , are not characterised by a proper s - matrix , nevertheless they could be quantized consistently on the world - sheet . it is therefore legitimate to argue @xcite that , if our universe has a cosmological constant , then the only way to quantize it within string theory is to postulate some sort of liouville string cosmology . one , however , may actually attempt to go one step further and try to find cosmological non - critical string models which eventually asymptote ( in time ) to equilibrium ( critical string ) cosmologies . in such models , which resemble _ quintessence models _ @xcite in conventional cosmology , there is an initial cause for departure from criticality ( conformality of the respective @xmath16-model , i.e. the presence of a central - charge deficit @xmath204 ) , which imply that the universe undergoes a non - equilibrium , and then a relaxation phase , which lasts until today . eventually the central charge deficit vanishes , and the string theory reaches its critical equilibrium situation . this allows an appropriate definition of asymptotic states and an @xmath10-matrix , since in such models one eventually exits from the de sitter accelerating phase . in what follows i will give a concrete example of such a non - equilibrium cosmology . the model involves two colliding branes worlds , one of which is assumed to be our world . we shall be very brief in our exposition . for details we refer the reader to @xcite where the model is described in some detail . i will argue below that the model produces inflation and a relaxing to zero cosmological `` constant '' hierarchically small as compared to the supersymmetry breaking ( tev ) scale . supersymmetry breaking is induced by compactification of the brane worlds on magnetized tori . the crucial ingredient is the non - criticality ( non conformality ) of string theory on the observable brane world induced at the collision , which is thus viewed as a cause for departure from equilibrium in this system . the hierarchical smallness of the present - era vacuum energy , as compared to the susy breaking scale , is attributed to relaxation phenomena . the model consists of two five - branes of type iib string theory , embedded in a ten dimensional bulk space time . two of the longitudinal brane dimensions are assumed compactified on a small torus , of radius @xmath205 . in one of the branes , from now on called _ hidden _ , the torus is _ magnetized _ with a constant magnetic field of intensity h. this amounts to an effective four - dimensional vacuum energy contribution in that brane of order : @xmath206 . notice that such compactifications provide alternative ways of breaking supersymmetry @xcite , which we shall make use of in the current article . in scenaria with two branes embedded in higher - dimensional bulk space times it is natural to assume ( from the point of view of solutions to bulk field equations ) that the two branes have _ opposite _ tensions . we assume that before the collision the visible brane ( our world ) has positive tension @xmath207 . the presence of opposite tension branes implies that the system is not stable , but this is o.k . from a cosmological view point . for our purposes we assume that the two branes are originally on _ collision course _ in the bulk , with a relative velocity @xmath208 for the validity of the @xmath16-model perturbation theory , and to allow for the model to have predictive power . the collision takes place at a given time moment . this constitutes an event , which in our scenario is identified with the _ initial cosmological singularity _ ( big bang ) on the observable world . we note that similar scenaria exist in the so - called ekpyrotic model for the universe @xcite . it must be stressed , though , that the similarity pertains only to the brane - collision event . in our approach the physics is entirely different from the ekpyrotic scenario . the collision is viewed as an event resulting in non - criticality ( departure from conformal invariance ) of the underlying string theory , and hence in non - vanishing @xmath84 functions at a @xmath16-model level . on the contrary , in the scenario of @xcite the underlying four - dimensional effective theory ( obtained after integration of the bulk extra dimensions @xcite ) is assumed always critical , satisfying classical equations of motion , and hence vanishing @xmath16-model @xmath84 functions . this latter property leads only to contracting and _ not expanding _ four - dimensional universes according to the work of @xcite , which constitutes one of the main criticisms of the ekpyrotic universe . on the other hand , in our non - critical description of the collision we do not assume classical solutions of the equations of motion , neither specific potentials associated with bulk branes , as in @xcite . the physics of our colliding worlds model can be summarized as follows : during the collision one assumes electric current transfer from the hidden to the visible brane , which results in the appearance of a magnetic field on the visible brane . we also assume that the entire effect is happening very slowly and amounts to a slow flow of energy and current density from the hidden to the visible brane ( our world ) . in turn , this results in a positive energy component of order @xmath209 in the vacuum energy of the visible brane world . at the moment of the collision the conformal invariance of the @xmath16-model describing ( stringy ) excitations on the observable brane world is spoiled , thereby implying the need for liouville dressing @xcite . this procedure restores conformal invariance at the cost of introducing an extra target space coordinate ( the liouville mode @xmath112 ) , which in our model has time - like signature . hence , initially , one faces a two - times situation . we argued @xcite , though , that our observable ( cosmological ) time @xmath96 parametrizes a certain curve , @xmath210 , on the two - times plane @xmath211 , and hence one is left with one physical time . the appearance of the magnetic field on the visible brane , on the dimensions @xmath212 , is described ( for times long after the collision ) within a @xmath16-model superstring formalism by the boundary deformation : @xmath213 , where @xmath214 , and @xmath215 denotes tangential @xmath16-model derivative on the world - sheet boundary . this @xmath16-model deformation describes open - string excitations attached to the brane world ( c.f . fig . [ fig : collbr ] ) . the presence of the quantity @xmath216 reflects the _ adiabatic _ switching on of the magnetic field after the collision . it should be remarked that in our approach the quantity @xmath217 is viewed as a world - sheet renormalization - group scale parameter @xcite . in addition to the magnetic field deformation , the @xmath16-model contains also boundary deformations describing the ` recoil ' of the visible world due to the collision : @xmath218 , where @xmath219 , @xmath220 denotes normal @xmath16-model derivative on the world - sheet boundary , and @xmath221 is the recoil velocity of the visible brane world . the presence of the exponential factors @xmath222 in _ both _ the magnetic field and recoil deformations implies a small but negative world - sheet anomalous dimension @xmath223 , and hence the relevance of both operators from a renormalization - group point of view . the induced central - charge deficit @xmath115 , which quantifies the departure from the conformal point of the pertinent @xmath16-model , can be computed by virtue of the zamolodchikov s c - theorem @xcite : @xmath224 , with @xmath225 the world - sheet infrared ( ultraviolet ) cutoff scale . as discussed in detail in @xcite , the correct scaling behaviour of the operators necessitates the identification @xmath226 which we assume from now on . the quantity @xmath227 is the equilibrium vacuum energy density , which we take to be zero @xmath228 due to an assumed cancellation between anti - de - sitter bulk and brane vacuum energies after the collision . we shall come back to this point later on . the non - conformal deformed @xmath16-model can become conformal by liouville dressing @xcite , as discussed in section 3 : @xmath229 , where @xmath230 are the liouville anomalous dimensions @xcite and @xmath112 is the normalized liouville mode , whose zero - mode is related to the renormalization - group scale @xmath231 as : @xmath232 . from the work of @xcite it becomes clear that the coupling constant @xmath94 is associated with supersymmetry - breaking mass splittings . this has to do with the different way fermions and bosons couple to an external magnetic field . the mass splittings squared of an open string are in our case @xcite : @xmath233 the so - obtained mass splittings are constant upon the requirement that the flow of time @xmath96 and of liouville mode @xmath112 are correlated in such a way that @xmath234 or at most slowly varying . notice that deviations from the condition ( [ liouvtime ] ) would result in very large negative - mass squares , which are clearly unstable configurations . hence , the identification ( [ liouvtime ] ) seems to provide a resolution of this problem . to ensure the phenomenologically reasonable order of magnitude of a tev scale , one must assume very small @xcite @xmath235 in planck units . note also that parametrizing the condition ( [ liouvtime ] ) as @xmath236 , @xmath237 , and taking into account that , for convergence of @xmath16-model path integration , it is formally necessary to work with euclidean signature @xmath96 @xcite , the induced metric on the hypersurface ( [ liouvtime ] ) in the extended space time acquires a minkowskian - signature robertson - walker form : @xmath238 . where @xmath239 denote spatial parts . in @xcite , where we refer the interested reader , we have given some arguments on a a _ dynamical stability _ of the condition ( [ liouvtime ] ) in the context of liouville strings . physically , one may interpret this result as implying that a time - varying magnetic field induced by the collision implies back reaction of strings onto the space time in such a way that the mass splittings of the string excitation spectrum , as a result of the field , are actually stabilized . we now notice that in our case the dilaton field is @xmath240 , that is , one faces a situation with an asymptotically constant dilaton . this is a welcome fact , because otherwise , the space - time would not be asymptotically flat , and one could face trouble in appropriately defining masses . in the case of a constant dilaton the vacuum energy is determined by the central - charge charge deficit @xmath115 , which in our case is : @xmath241 where @xmath242 is the world - sheet zero mode of the liouville field to be identified with the target time on the _ hypersurface _ ( [ liouvtime ] ) of the extended space - time resulting after liouville dressing . we next remark that the restoration of the conformal invariance by the liouville mode results in the following equations ( c.f . ( [ liouvcond ] ) ) for the @xmath16-model background fields / couplings @xmath81 near a fixed - point of the world - sheet renormalization group ( large - times cosmology ) we restrict ourselves here @xcite : @xmath243 where the prime denotes derivative with respect to the liouville zero mode @xmath121 , and the sign on the right - hand - side is appropriate for supercritical strings we are dealing with here . these equations replace the equilibrium equations @xmath244 of critical string theory , and should be used in our colliding brane scenario to determine the evolution of the scale factor of the four - dimensional robertson - walker universe . a preliminary analysis has been performed in @xcite , where we refer the reader for details . below we only describe briefly the main results . in our non - critical string scenario , one does indeed obtain an expanding universe , in contrast to standard ekpyrotic scenaria @xcite , based on critical strings and specific solutions to classical equations of motion . one of the most important features of the existence of a non - equilibrium phase of string theory due to the collision is the possibility for an _ inflationary phase_. although the physics near the collision is strongly coupled , and the @xmath16-model perturbation theory is not reliable , nevertheless one can give compelling physical arguments favoring the existence of an early phase of the brane world where the four - dimensional universe scale factor undergoes exponential growth ( inflation ) . this can be understood as follows : in our model we encounter two type - ii string theory branes colliding , and then bouncing back . from a stringy point of view the collision and bounce will be described by a phase where open strings stretch between the two branes worlds ( which can be thought of as lying a few string scales apart during the collision , c.f . [ fig : collbr ] ) . during that early phase the excitation energy of the brane worlds can be easily computed by the same methods as those used to study scattering of type ii d - branes in @xcite . in type ii strings the exchange of pairs of open strings is described by annulus world - sheet diagrams , which in turn results in the appearance of `` spin structure factors '' in the scattering amplitude . the latter are expressed in terms of appropriate sums over jacobi @xmath245 functions . due to special properties of these functions , the spin structures start of at quartic order in @xmath221 @xcite . the resulting excitation energy is therefore of order @xmath246 and may be thought off as an initial value of the central charge deficit of the non - critical string theory describing the physics of our brane world after the collision . the deficit @xmath115 is thus cut off at a finite value in the ( world - sheet ) infrared scale ( early target times , c.f . [ fig : collbr ] ) . one may plausibly assume that the central charge deficit remains constant for some time , which is the era of _ inflation_. it can be shown @xcite that for ( finite ) constant @xmath247 the liouville equations imply a scale factor exponentially growing with the liouville zero mode @xmath248 ( assuming a negative @xmath249 ) . upon the condition ( [ liouvtime ] ) , then , one obtains an early inflationary phase after the collision , in contrast to the critical - string based arguments of @xcite . the duration of the inflationary phase is @xmath250 , which yields the conventional values of inflationary models of order @xmath251 for @xmath252 . exit from the inflationary phase is possible , and the de sitter phase is assumed to be succeeded by a phase , to be identified with the _ nucleosynthesis era _ , in which the central charge deficit passes through a metastable critical phase in which @xmath116 vanishes , and then changes sign ( cf . [ fig : collbr ] ) . this can happen in liouville strings with time - like fields @xmath96 , as we mentioned above , where the central charge deficit can oscillate before it reaches its equilibrium value @xcite . during the metastable critical phase the central charge deficit is zero and hence there is no liouville dressing . this implies that during that epoch any time dependence of fields on our brane world will be given by the ordinary critical string time dependence . in brane cosmologies the friedman equation on the three brane , where matter lives , assumes the form @xcite ( for spatially flat branes ) : @xmath253 where @xmath254 is the gravitational constant in the bulk geometry , @xmath255 is an integration constant ( a mass of a bulk schwarzschild black hole in the brane cosmology of @xcite , which for simplicity , and without any important physical consequences , we assume zero for our purposes ) , @xmath256 is the bulk anti de sitter cosmological constant , and @xmath257 is the scale factor on the brane world . the density @xmath136 denotes the total energy density on the brane , including matter contributions as well as cosmological constant ( brane tension ) contributions . shifting @xmath258 to make such tension contributions explicit , @xmath259 , where @xmath260 is the matter density , one observes that for late times , the friedman equation on the brane assumes the standard form , linear in @xmath261 , _ provided _ one cancels the total vacuum energy as perceived by a brane ( four - dimensional ) observer : @xmath262 . in our non - critical string case @xmath263 may be assumed positive , and one may identify @xmath264 ( the equilibrium value of the string central charge deficit ) with @xmath265 which vanishes . during the nucleosynthesis era , therefore , the rate of expansion of the non - critical string universe is that predicted by standard cosmology , which is a welcome feature given the phenomenological success of the nucleosynthesis model in explaining the abundance of light elements in the universe . at the end of the nucleosynthesis one faces a non - critical situation , with a central charge deficit significantly smaller than that of the inflationary era . the above considerations on the two times , liouville @xmath112 and coordinate time @xmath96 , arise again . the generalized conformal invariance conditions ( [ liouvcond2 ] ) apply in this case , and hence the scale factor depends now on both @xmath96 and @xmath112 , but the evolution is confined for energetic reasons on the hypersurface ( [ liouvtime ] ) , as explained above . we may assume that at the end of the nucleosynthesis @xmath244 for the scale factor , and hence one obtains from ( [ liouvcond2 ] ) that a natural solution is @xmath266 since after the end of nucleosynthesis ( present era ) the universe is in a matter dominated phase , one may assume , from the above considerations , that @xmath267 with @xmath268 the robertson - walker observable time to be connected with the liouville time @xmath112 on the hypersurface ( [ liouvtime ] ) . thus the present era scale factor scales with time as : @xmath269 . this suffices to yield a current acceleration of the universe , thereby indicating that the dark energy ( in the form of gravitational recoil - liouville contributions ) takes over the expansion . unfortunately , such a scaling must not remain for ever , if one wants to recover an s - matrix equilibrium theory asymptotically ( although it must be noted that this is not formally necessary in this framework , since liouville strings do not admit s - matrix description in general , as we have discussed above ) . remarkably a solution to this problem can be provided in our model by noting the possibility of decompactification of the extra toroidal dimensions on the five branes , as a result of the collision . if one assumes a very slow decompactification ( much slower than any other rate in the problem ) then eventually the radius of the torus @xmath270 , while keeping the magnetic field energy @xmath271 constant . the masses of the matter particles on the brane will therefore eventually go to zero , and hence one will enter a radiation era in which @xmath272 on the five brane . from ( [ branefried ] ) this will imply an asymptotic scaling @xmath273 , which , upon liouville dressing , will yield a scale factor @xmath274 , as @xmath275 , thereby resolving the cosmic horizon problem ( [ cosmichorizon ] ) , and thus saving the s - matrix in this model . this concludes our discussion on this toy model . it remains to be seen whether such conjectures / speculations can be accommodated in realistic string models of brane cosmologies . i wish to thank prof . h. klapdor - kleingrothaus and the organizers of the conference _ beyond the desert 2002 _ ( oulu , finland , june 2002 ) for the invitation . this work is supported by the european union ( contract hprn - ct-2000 - 00152 ) . 999 for a recent review see : p. biermann and g. sigl , lect . notes phys . 576 * ( 2001 ) 1 [ arxiv : astro - ph/0202425 ] , and references therein s. perlmutter _ et al . _ [ supernova cosmology project collaboration ] : nature * 391 * , 51 ( 1998 ) ; a. g. riess _ et al . _ [ supernova search team collaboration ] : astron . j. * 116 * , 1009 ( 1998 ) ; p. m. garnavich _ et al . _ , astrophys . j. * 509 * , 74 ( 1998 ) . for a subset of references see : s. r. coleman and s. l. glashow , phys . d * 59 * ( 1999 ) 116008 [ arxiv : hep - ph/9812418 ] . s. r. coleman and s. l. glashow , arxiv : hep - ph/9808446 . s. r. coleman and s. l. glashow , phys . b * 405 * ( 1997 ) 249 [ arxiv : hep - ph/9703240 ] ; o. bertolami , gen . * 34 * ( 2002 ) 707 [ arxiv : astro - ph/0012462 ] ; nucl . * 88 * ( 2000 ) 49 [ arxiv : gr - qc/0001097 ] . o. bertolami and c. s. carvalho , phys . d * 61 * ( 2000 ) 103002 [ arxiv : gr - qc/9912117 ] . t. kifune , astrophys . j. * 518 * ( 1999 ) l21 [ arxiv : astro - ph/9904164 ] ; g. amelino - camelia and t. piran , phys . rev . d * 64 * ( 2001 ) 036005 [ arxiv : astro - ph/0008107 ] . phys . rev . d * 64 * ( 2001 ) 036005 [ arxiv : astro - ph/0008107 ] ; y. j. ng , d. s. lee , m. c. oh and h. van dam , phys . b * 507 * ( 2001 ) 236 [ arxiv : hep - ph/0010152 ] . j. r. ellis , n. e. mavromatos and d. v. nanopoulos , phys . d * 63 * ( 2001 ) 124025 [ arxiv : hep - th/0012216 ] . t. banks , w. fischler : hep - th/0102077 ; s. hellerman , n. kaloper and l. susskind , jhep * 0106 * ( 2001 ) 003 [ arxiv : hep - th/0104180 ] ; w. fischler , a. kashani - poor , r. mcnees and s. paban , jhep * 0107 * ( 2001 ) 003 [ arxiv : hep - th/0104181 ] ; e. witten , hep - th/0106109 ; p. o. mazur , e. mottola : phys . d * 64 * , 104022 ( 2001 ) , and references therein . r. j. protheroe and h. meyer , phys . b * 493 * ( 2000 ) 1 [ arxiv : astro - ph/0005349 ] . see , however , k. okumura _ et al . _ , astro - ph/0209487 which questions the experimental findings ( thanks to a. sakharov for pointing this out ) . j. ellis , n. e. mavromatos , d. v. nanopoulos : phys . b * 293 * ( 1992 ) 37 [ arxiv : hep - th/9207103 ] ; mod . a * 10 * ( 1995 ) 1685 [ arxiv : hep - th/9503162 ] ; for reviews see : erice subnuclear series ( world sci . , singapore ) * 31 * 1 , ( 1993 ) ; [ hep - th/9304133 ] ; j. chaos , solitons and fractals * 10 * , 345 ( eds . c. castro amd m.s . el naschie , elsevier science , pergamon 1999 ) [ hep - th/9805120 ] , and references therein ; g. amelino - camelia , j. r. ellis , n. e. mavromatos and d. v. nanopoulos , int . j. mod . a * 12 * ( 1997 ) 607 [ arxiv : hep - th/9605211 ] ; j. r. ellis , n. e. mavromatos and d. v. nanopoulos , phys . d * 62 * ( 2000 ) 084019 [ arxiv : gr - qc/0006004 ] . phys . d * 62 * ( 2000 ) 084019 [ arxiv : gr - qc/0006004 ] ; e. gravanis and n. e. mavromatos , arxiv : hep - th/0103122 . l. gonzalez - mestres , arxiv : physics/0003080 ; l. j. garay , phys . rev . d * 58 * ( 1998 ) 124015 [ arxiv : gr - qc/9806047 ] . s. chadha and h. b. nielsen , nucl . b * 217 * ( 1983 ) 125 . h. b. nielsen and i. picek , nucl . b * 211 * ( 1983 ) 269 [ addendum - ibid . b * 242 * ( 1984 ) 542 ] ; j. l. chkareuli , c. d. froggatt and h. b. nielsen , phys . * 87 * ( 2001 ) 091601 [ arxiv : hep - ph/0106036 ] . j. i. latorre , p. pascual and r. tarrach , nucl . b * 437 * ( 1995 ) 60 [ arxiv : hep - th/9408016 ] . g. amelino - camelia , j. r. ellis , n. e. mavromatos , d. v. nanopoulos and s. sarkar , nature * 393 * ( 1998 ) 763 [ arxiv : astro - ph/9712103 ] ; j. r. ellis , k. farakos , n. e. mavromatos , v. a. mitsou and d. v. nanopoulos , astrophys . j. * 535 * ( 2000 ) 139 [ arxiv : astro - ph/9907340 ] ; j. ellis , n. e. mavromatos , d. v. nanopoulos and a. s. sakharov , arxiv : astro - ph/0210124 . r. gambini and j. pullin , phys . d * 59 * ( 1999 ) 124021 [ arxiv : gr - qc/9809038 ] ; arxiv : gr - qc/0110054 . a. ashtekar and j. lewandowski , class . * 14 * ( 1997 ) a55 [ arxiv : gr - qc/9602046 ] ; a. ashtekar , c. rovelli and l. smolin , phys . * 69 * ( 1992 ) 237 [ arxiv : hep - th/9203079 ] ; c. rovelli and l. smolin , phys . lett . * 61 * ( 1988 ) 1155 ; nucl . b * 331 * ( 1990 ) 80 ; for a comprehensive review ( and more references ) see : a. ashtekar , arxiv : gr - qc/0112038 . g. amelino - camelia , int . j. mod . d * 11 * ( 2002 ) 35 [ arxiv : gr - qc/0012051 ] ; nature * 418 * ( 2002 ) 34 [ arxiv : gr - qc/0207049 ] ; g. amelino - camelia , d. benedetti and f. dandrea , arxiv : hep - th/0201245 ; j. kowalski - glikman , phys . a * 286 * ( 2001 ) 391 [ arxiv : hep - th/0102098 ] ; arxiv : hep - th/0209264 . f. david , modern physics letters * a3 * ( 1988 ) 1651 ; j. distler , h. kawai nucl b321 * ( 1989 ) 509 ; see also n.e . mavromatos and j.l . miramontes , mod . . lett . * a4 * ( 1989 ) 1847 ( 1989 ) ; e. d hoker and p.s . kurzepa , mod . a5 * ( 1990 ) 1411 ; e. d hoker , mod . . lett . * a6 * ( 1991 ) 745 . g. a. diamandis , b. c. georgalas , n. e. mavromatos , e. papantonopoulos and i. pappa , int . j. mod . a * 17 * ( 2002 ) 2241 [ arxiv : hep - th/0107124 ] . e. gravanis and n. e. mavromatos , arxiv : hep - th/0205298 , physics lett . * b * in press , and work in progress . the ctp formalism may be attributed to : j. schwinger , j. math . * 2 * , 407 ( 1961 ) ; for reviews see : e. calzetta and b.l . . rev . * d37 * , 2878 ( 1988 ) ; e. calzetta , s. habib and b.l . hu , phys . rev . * d37 * , 2901 ( 1988 ) ; h. umezawa , _ advanced field theory : micro , macro and thermal concepts _ ( american inst . of physics , n.y . 1993 ) . a non - linear difusive non - critical stringy system is examined in : n. e. mavromatos and r. j. szabo , int . j. mod . phys . a * 16 * ( 2001 ) 209 [ arxiv : hep - th/9909129 ] ; see , however , how world - sheet supersymmetry may eliminate such non - linearities : n. e. mavromatos and r. j. szabo , jhep * 0110 * ( 2001 ) 027 [ arxiv : hep - th/0106259 ] . i. i. kogan , n. e. mavromatos and j. f. wheater , phys . b * 387 * ( 1996 ) 483 [ arxiv : hep - th/9606102 ] ; n. e. mavromatos and r. j. szabo , phys . d * 59 * ( 1999 ) 104018 [ arxiv : hep - th/9808124 ] . j. ellis , e. gravanis , n. e. mavromatos and d. v. nanopoulos , arxiv : gr - qc/0209108 . d. sudarsky , l. urrutia and h. vucetich , arxiv : gr - qc/0204027 ; j. alfaro , h. a. morales - tecotl and l. f. urrutia , phys . lett . * 84 * , 2318 ( 2000 ) [ arxiv : gr - qc/9909079 ] ; j. alfaro and g. palma , phys . d * 65 * , 103516 ( 2002 ) [ arxiv : hep - th/0111176 ] . j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , phys . d * 64 * , 123522 ( 2001 ) . for a recent and concise review of the criticisms of the model of @xcite see : a. linde , arxiv : hep - th/0205259 , and references therein . c. bachas , phys . b * 374 * , 37 ( 1996 ) [ arxiv : hep - th/9511043 ] . for a recent review see : d. langlois , arxiv : hep - th/0209261 and references therein ; see also : p. binetruy , c. deffayet and d. langlois , nucl . b * 565 * ( 2000 ) 269 [ arxiv : hep - th/9905012 ] ; p. binetruy , c. deffayet , u. ellwanger and d. langlois , phys . b * 477 * ( 2000 ) 285 [ arxiv : hep - th/9910219 ] .	i review some aspects of non - critical strings in connection with lorentz - invariance violating approaches to quantum gravity . i also argue how non - critical strings may provide a unifying framework where string cosmology and quantum gravity may be tackled together .
while attending the august 2010 `` new trends in noncommutative algebra '' conference in seattle , the author of the present paper was told about the question whether the @xmath0 functor from af c-algebras to dimension groups _ splits _ , that is , has a ( categorical ) right inverse . we describe here three reasons why this is not the case . the first and the third reason are more conveniently expressed _ via _ the _ nonstable @xmath0-theory _ ( often also called `` nonstable k - theory '' ) of a ring @xmath1 , mainly described by a commutative monoid , usually denoted by @xmath2 . if @xmath1 is unital , then @xmath2 is the monoid of all isomorphism types of finitely generated projective right @xmath1-modules and @xmath14 is the grothendieck group of @xmath2 . corollary [ c : firstnonlift ] , implies that the more general question , whether the functor @xmath8 has a right inverse from simplicial monoids ( cf . section [ s : basicmon ] ) to _ exchange rings _ ( cf . warfield @xcite , and ara @xcite for the non - unital case ) , has a negative answer . actually , corollary [ c : firstnonlift ] is stated for an even more general class of rings that we call _ weakly v - semiprimitive rings_. it also suggests that the original question might get more interesting if one required the morphisms between dimension groups be _ one - to - one_. _ our second and third reason will address the modified @xmath15-splitting question_. proposition [ p : nonsplit ] , involves the author s construction , introduced in wehrung @xcite , of a dimension group with order - unit , of cardinality @xmath16 , which is not isomorphic to @xmath2 for any ( von neumann ) regular ring @xmath1 . if @xmath0 had a right inverse from simplicial groups ( cf . section [ s : basicmon ] ) to af c-algebras , then it would also have a right inverse from simplicial groups to regular rings . however , due to the abovementioned counterexample , this is impossible . there are some complications , due to the requirement that the maps between simplicial groups be one - to - one , that we solve by using tools from lattice theory and universal algebra . theorem [ t : nolift ] , is the combinatorial core of our second reason . it involves a commutative diagram @xmath7 of simplicial monoids , indexed by a cube ( i.e. , the powerset lattice @xmath6 of a three - element set ) , that has no lifting , with respect to the functor @xmath8 , by _ semiprimitive exchange rings_. in particular , it can be lifted neither by c-algebras of real rank @xmath5 nor by regular rings . in particular , the negative result of proposition [ p : nonsplit ] extends both to regular rings ( this could have been deduced directly from @xcite ) and to c-algebras of real rank @xmath5 ( this is new ) . besides those three reasons , we also provide two positive representation results of the abovementioned diagram @xmath7 : while theorem [ t : nolift ] implies that @xmath7 can not be lifted , with respect to the functor @xmath8 , by any commutative diagram of semiprimitive exchange rings , theorem [ t : lulift ] implies that a certain `` collapsed '' image , denoted by @xmath17 , of the diagram @xmath7 , can be lifted with _ exchange rings _ , while theorem [ t : luliftc * ] implies that @xmath17 can be lifted with _ c-algebras _ ( of real rank and stable rank both equal to @xmath9 ) . in order to be able to state that @xmath17 is really a `` diagram '' , we need the extended notion of diagram given in definition [ d : diagram ] . by `` unfolding '' @xmath17 back to @xmath7 , we obtain ( proposition [ p : noliftlu ] ) that @xmath7 can also be lifted with exchange rings on the one hand , and with c-algebras on the other hand . there is no `` best of two worlds '' , because a c-algebra is an exchange ring iff it has real rank @xmath5 , and @xmath7 can not be lifted by c-algebras of real rank @xmath5 . the reason why we need a diagram indexed by a finite lattice lies in a quite technical result of categorical algebra by gillibert and the author @xcite called the _ condensate lifting lemma _ , cll in short . cll makes it possible to turn _ diagram counterexamples _ to _ object counterexamples _ , with possible jumps of cardinality . it works only on diagrams indexed by so - called _ almost join - semilattices _ , in particular for finite lattices . by applying various instances of cll to the negative lifting versus positive lifting results of the cube @xmath7 , we obtain ( cf . theorems [ t : cxakrr0 ] and [ t : ccxakrr0 ] ) that _ there exists a unital exchange ring @xmath1 , with stable rank @xmath9 and index of nilpotence @xmath11 , such that @xmath2 is the positive cone of a dimension group and @xmath2 is not isomorphic to @xmath12 for any ring @xmath13 which is either a c-algebra of real rank @xmath5 or a regular ring_. due to the cardinality - jumping properties of the `` condensate '' construction underlying cll , the exchange ring that we construct has @xmath10 elements , while the c-algebra that we construct is @xmath10-separable ( it has also real rank @xmath9 ) . both the ring and the c-algebra are stably finite , have stable rank @xmath9 and index of nilpotence @xmath11 , while the nonstable @xmath0-theory is the positive cone of a dimension group with order - unit of index @xmath11 . although the present paper is written in the language of ring theory , a substantial part of our results and proofs draw their basic inspiration from related results in universal algebra and lattice theory . for example , * corollary [ c : firstnonlift ] ( the `` first reason '' ) originates in tma and wehrung ( * ? ? ? * section 8) , which is a non - lifting result by universal algebras with respect to the congruence lattice functor . * proposition [ p : nonsplit ] ( the `` second reason '' ) originates in wehrung @xcite , which solves , in the negative , representation problems with respect to various functors , including @xmath0 , by using the lattice structure of the set of all principal right ideals in a regular ring . * the cube @xmath7 ( cf . figure [ fig : diagrd ] ) , which is the key ingredient of theorem [ t : nolift ] ( the `` third reason '' ) , originates in a cube of boolean semilattices introduced in tma and wehrung ( * ? ? ? * section 3 ) . the author is indebted to pere ara , francesc perera , enrique pardo , george elliott , chris phillips , and ken goodearl for stimulating interaction and helpful comments at various stages of this paper s preparation . special thanks are due to ken goodearl for his improvement of the author s original argument , which led to theorem [ t : firstnonlift ] , and to chris phillips for most inspiring conversations about c-algebras and noncommutative rings at the seattle conference in august 2010 . all our rings will be associative , but not necessarily unital . we denote by @xmath19 , or @xmath9 if @xmath1 is understood , the unit of a ring @xmath1 . although we shall use standard commutative diagrams such as the one , denoted by @xmath7 , introduced in figure [ fig : diagrd ] , we shall also use sometimes diagrams of a more general kind , a formal definition of which follows . [ d : diagram ] a _ quiver _ is a quadruple @xmath20 where @xmath21 and @xmath22 ( the `` vertices '' and the `` edges '' , respectively ) are sets and @xmath23 ( the `` source '' and `` target '' map , respectively ) . an element @xmath24 is thus viewed as an `` arrow '' from @xmath25 to @xmath26 . a _ diagram _ in a category @xmath27 , _ indexed by @xmath28 _ , is a functor from @xmath28 to @xmath27 . formally , it is given by a pair @xmath29 , where @xmath30 ( resp . , @xmath31 ) is a map from @xmath21 ( resp . , @xmath22 ) to the object class ( resp . , morphism class ) of @xmath27 , such that @xmath32 is a morphism from @xmath33 to @xmath34 for any @xmath24 . there is no composition of arrows in @xmath28 , thus nothing to worry about at that level . the usual definitions of a _ natural transformation _ and a _ natural equivalence _ carry over to diagrams without modification . for diagrams @xmath35 , we denote by @xmath36 the natural equivalence of @xmath37 and @xmath38 . for categories @xmath39 and @xmath40 , we say that a diagram @xmath41 _ lifts _ a diagram @xmath42 with respect to a functor @xmath43 if @xmath44 ( cf . figure [ fig : liftdiag ] ) . in all our lifting problems , we will need to specify additional composition relations among the arrows in the range of @xmath45 . ( the first such instance in this paper is theorem [ t : firstnonlift ] : the additional requirement is @xmath46 . ) @xmath47^{\operatorname{\operatorname{\gamma}}}\ar[rrd]_{\psi } & & { \mathbf{a}}\ar[d]^{\phi}\\ & & { \mathbf{b}}}\ ] ] in many of our applications ( but not all of them ) , @xmath39 will be a category of rings , @xmath40 will be a category of commutative monoids , and @xmath37 will be the nonstable @xmath0-theory functor @xmath8 . we shall now give a brief section by section overview of the paper . in section [ s : basicmon ] * , we recall the basic monoid - theoretical concepts used in the paper . * in section [ s : rv(r ) ] * , we recall some basic facts about the functor @xmath8 , from rings to commutative monoids , describing nonstable @xmath0-theory , for either rings or c-algebras . we also recall basic facts about _ semiprimitive _ rings . in section [ s : basicexch ] * , we recall some basic facts about _ exchange rings _ , including the non - unital case . * in section [ s : firstdiagr ] * , we introduce a notion of orthogonality of elements in a commutative monoid , then the notion of a _ weakly v - semiprimitive ring _ ( _ v - semiprimitive rings _ will be introduced in section [ s : vsemi ] ) . we prove that every exchange ring is weakly v - semiprimitive . we also introduce a diagram ( in the sense of definition [ d : diagram ] ) @xmath48 of simplicial monoids with order - unit that has no lifting , with respect to the functor @xmath8 , by weakly v - semiprimitive rings ( theorem [ t : firstnonlift ] ) . it follows ( corollary [ c : firstnonlift ] ) that the functor @xmath8 has no right inverse from simplicial monoids with order - unit , with normalized monoid _ homomorphisms _ , to weakly v - semiprimitive rings . * in section [ s : nosplit ] * , we prove ( cf . proposition [ p : nonsplit ] ) that the functor @xmath8 has no right inverse from simplicial monoids with unit , and their _ embeddings _ , to af c-algebras . the proof uses universal algebra , lattice theory , and a counterexample of cardinality @xmath16 constructed in wehrung @xcite . as proposition [ p : nonsplit ] will be later superseded by stronger results , we give only an outline of the proof . nevertheless we chose to keep it there , because a large part of the material in further sections originates from the proof of that result . in section [ s : measring ] * we introduce a tool that will prove essential for our work , the notion of a _ measured ring_. by using some results of section [ s : rv(r ) ] and [ s : basicexch ] , we prove ( cf . lemma [ l : reflsubcat ] ) that the category of _ measured _ exchange rings is a reflective subcategory of the one of _ premeasured _ exchange rings . we also prove ( cf . lemma [ l : reflsubrr0 ] ) an analogous result for c-algebras of real rank @xmath5 . in section [ s : vsemi ] * , we introduce a common generalization of semiprimitive exchange rings , principal ideal domains , and polynomial rings over fields , that we call _ v - semiprimitive rings_. we also observe there that not every exchange ring , and not every c-algebra , is v - semiprimitive . the c-algebra @xmath49 constructed in example [ ex : semiprnotv ] will play a crucial role in further sections . * in section [ s : unlift ] * , we introduce the main combinatorial object of the paper , namely the diagram @xmath7 . it is a @xmath6-indexed commutative diagram of simplicial monoids , and it originates in the proof of proposition [ p : nonsplit ] . theorem [ t : nolift ] implies that @xmath7 has no lifting , with respect to the functor @xmath8 , by any commutative diagram _ of v - semiprimitive _ rings . this is the central non - representability result of the paper : all the other negative lifting results of the paper , with the exception of theorem [ t : firstnonlift ] , follow from that one . the proof of theorem [ t : nolift ] is direct : it does not involve any universal algebra , lattice theory , or transfinite cardinals . we also introduce an auxiliary diagram , denoted there by @xmath50 , of simplicial monoids and monoid _ embeddings _ , such that any lifting of @xmath50 , with respect to the functor @xmath51 introduced in definition [ d : twofunct ] , gives rise to a lifting of @xmath7 . this implies , in particular , that the functor @xmath8 has no right inverse from simplicial monoids with order - unit , and their _ embeddings _ , to c-algebras of real rank @xmath5 ( resp . , regular rings ) , see corollary [ c : nolift3 ] . in section [ s : collapse ] * , we take advantage of the many symmetries underlying the diagram @xmath7 to collapse it to another diagram ( now in the sense of definition [ d : diagram ] ) , denoted by @xmath17 , much simpler looking than @xmath7 , such as every lifting of @xmath17 gives rise to a lifting of @xmath7 ( this is shown in the argument of the proof of proposition [ p : noliftlu ] ) . we also prove , by a direct construction , that @xmath17 ( thus also @xmath7 ) can be lifted , with respect to the functor @xmath8 , by _ exchange rings_. in a sense , everything in that section boils down to finding a matrix @xmath52 ( given in ) satisfying the constraints given in and . * in section [ s : calglift ] , we construct a lifting of @xmath17 by _ c-algebras_. these c-algebras have both real rank and stable rank equal to @xmath9 , while they have index of nilpotence @xmath11 . by using theorem [ t : nolift ] , it can be seen that these values are optimal . * in section [ s : al3 ] * , we take advantage of the conflict between theorem [ t : nolift ] ( a non - lifting result of @xmath7 ) and proposition [ p : cdliftedexch ] ( a lifting result of @xmath7 by exchange rings ) to construct , for every field @xmath53 , a unital exchange @xmath53-algebra @xmath54 such that @xmath55 , although being the positive cone of a dimension group , is never isomorphic to @xmath12 for either a c-algebra of real rank @xmath5 or a regular ring @xmath13 . furthermore , @xmath54 has at most @xmath56 elements , and it has index of nilpotence @xmath11 . in order to achieve this , we are using a fair amount of heavy machinery established earlier in the book gillibert and wehrung @xcite . that work introduces tools of categorical algebra and infinite combinatorics , called _ larders _ and _ lifters _ , that are designed to turn _ diagram counterexamples _ to _ object counterexamples_. these tools are effective mainly on lattice - indexed diagrams , which is the reason why we need to keep the cube @xmath7 instead of the better looking @xmath17 . the results of @xcite being now used as a toolbox , it should be possible to read the proofs of section [ s : al3 ] without needing to ingest large amounts of larder technology . in particular , the ring - theoretical and c-algebraic arguments used there are elementary . * in section [ s : al3sep ] * , we show how to extend to c-algebras what we did for exchange rings in section [ s : al3 ] . in particular , we construct a unital c-algebra @xmath22 , of real rank and stable rank both equal to @xmath9 , such that @xmath57 , although being the positive cone of a dimension group , is never isomorphic to @xmath12 for either a c-algebra of real rank @xmath5 or a regular ring @xmath13 . furthermore , @xmath22 is @xmath10-separable , and it has index of nilpotence @xmath11 . in section [ s : pbs ] * , we list some open problems . we represent on figure [ fig : ringclass ] the implications between the main attributes of rings used in the paper . @xmath58 & \text{v - semiprimitive}\ar@{=>}[u ] & \text{semiprimitive}\\ & \text{semiprimitive exchange}\ar@{=>}[lu]\ar@{=>}[ru]\ar@{=>}[u ] & \text{c-algebra}\ar@{=>}[u]\\ & \text{regular}\ar@{=>}[u ] & \text{c-algebra of real rank } 0\ar@{=>}[lu]\ar@{=>}[u ] & } \ ] ] we refer to goodearl @xcite for partially ordered abelian groups , interpolation groups , dimension groups . we set @xmath59 and @xmath60 . we shall write all our commutative monoids additively . a submonoid @xmath61 of a commutative monoid @xmath62 is an _ o - ideal _ of @xmath62 if @xmath63 implies that @xmath64 and @xmath65 , for all @xmath66 . we say that @xmath62 is _ conical _ if @xmath67 is an o - ideal . every commutative monoid @xmath62 can be endowed with its _ algebraic preordering _ @xmath68 , defined by @xmath69 if there exists @xmath70 such that @xmath71 . we say that an element @xmath72 is an _ order - unit of @xmath62 _ if for each @xmath73 there exists @xmath74 such that @xmath75 . for @xmath76 , an element @xmath77 has _ index at most @xmath78 _ if @xmath79 implies that @xmath80 , for each @xmath73 . if @xmath61 is an o - ideal of a commutative monoid @xmath62 , the binary relation @xmath81 on @xmath62 defined by @xmath82 is a monoid congruence of @xmath62 , and @xmath83 is a conical commutative monoid . the _ grothendieck group _ of @xmath62 is the initial object in the category of all monoid homomorphisms from @xmath62 to a group . it consists of an abelian group @xmath84 with a monoid homomorphism @xmath85 , and we shall always endow it with the unique compatible preordering with positive cone @xmath86 ( i.e. , @xmath69 iff @xmath87 ) . a commutative monoid @xmath62 has the _ refinement property _ , or is a _ refinement monoid _ ( cf . dobbertin @xcite ) , if for all @xmath88 such that @xmath89 , there are @xmath90 ( for @xmath91 ) such that @xmath92 and @xmath93 for each @xmath94 . a very special class of refinement monoids consists of the positive cones of _ dimension groups_. a partially ordered abelian group @xmath84 is a dimension group if it is directed ( i.e. , @xmath95 ) , unperforated ( i.e. , @xmath96 implies @xmath97 for each @xmath98 and each @xmath99 ) and the positive cone @xmath100 satisfies refinement . simplicial group _ is a group of the form @xmath101 , for a natural number @xmath78 , ordered componentwise . every simplicial group is a dimension group . conversely , every dimension group is a direct limit of simplicial groups ( this follows from effros , handelman , and shen @xcite , but the form of this result established earlier in grillet @xcite is easily seen to be equivalent ) . we shall also call a _ simplicial monoid _ the positive cone of a simplicial group ( i.e. , @xmath102 for some @xmath76 ) . [ d : pointedmon ] for a subcategory @xmath27 of the category @xmath103 of conical commutative monoids with monoid homomorphisms , we shall denote by @xmath104 the category of _ monoids with order - unit _ in @xmath27 : the objects of @xmath104 are the pairs @xmath105 , where @xmath62 is an object of @xmath27 and @xmath106 is an order - unit of @xmath62 , and a morphism from @xmath105 to @xmath107 is a monoid homomorphism @xmath108 such that @xmath109 ( i.e. , @xmath110 is _ normalized _ ) . [ d : catrings1 ] not all of our rings will be unital , and consequently not all our ring homomorphisms will preserve the unit . for a subcategory @xmath111 of the category @xmath112 of rings and ring homomorphisms , we shall denote by @xmath113 the subcategory of @xmath111 consisting of all unital rings in @xmath111 , with unital ring homomorphisms , possibly with any additional structure present in @xmath111 . for example , we shall denote by @xmath114 the category of all c-algebras of real rank @xmath5 with c-homomorphisms , and by @xmath115 the category of all unital c-algebras of real rank @xmath5 with unital c-homomorphisms . a useful subcategory of @xmath114 is the full subcategory @xmath116 of all the _ approximately finite _ , or _ _ , c-algebras . by definition , a c-algebra is af if it is a ( possibly uncountable ) c-direct limit of finite - dimensional c-algebras ( not all our af algebras will be separable ) . for basic concepts about ( stable or nonstable ) k - theory of rings and c-algebras , we refer to goodearl ( ? ? ? * section 4 ) for the unital case , ara ( * ? ? ? * section 3 ) for the general case , blackadar ( * ? ? ? * chapter 3 ) for the case of c-algebras . for basic notions about c-algebras we refer to murphy @xcite . for a ring @xmath1 ( associative but not necessarily unital ) , we shall often identify the ring @xmath117 of all @xmath118 matrices over @xmath1 with its image , _ via _ the embedding @xmath119 , in @xmath120 . hence the elements of the ( non - unital ) ring @xmath121 can be identified with the countably infinite matrices with entries from @xmath1 and only finitely many nonzero entries . we define @xmath122 as the set of all idempotent elements of @xmath117 , for each @xmath123 . for a ring homomorphism @xmath124 , we denote by @xmath125 the map from @xmath122 to @xmath126 defined by the rule @xmath127 we shall sometimes write @xmath128 instead of @xmath129 , for @xmath130 . idempotents @xmath131 are _ orthogonal _ if @xmath132 . the ( murray - von neumann , algebraic ) _ equivalence _ of idempotents @xmath133 is defined by @xmath134 iff there are @xmath135 such that @xmath136 and @xmath137 . replacing @xmath138 by @xmath139 and @xmath140 by @xmath141 , we see that we can assume that @xmath142 and @xmath143 . we denote by @xmath144_r$ ] , or @xmath144 $ ] if @xmath1 is understood , the @xmath145-equivalence class of @xmath106 . for all @xmath131 there exists @xmath146 such that @xmath147 ; it follows that equivalence classes can be added , _ via _ the formula @xmath148 } [ a]+[b]:=[a+b]\text { if } ab = ba=0\,,\text { for all } a , b\in{\mathrm{i}}_\infty(r)\,.\ ] ] the set @xmath149}\mid{a\in{\mathrm{i}}_\infty(r)}\right\}}}$ ] , endowed with the addition defined in , is a conical commutative monoid . if @xmath1 is unital , then @xmath150 $ ] is an order - unit of @xmath2 and @xmath2 is isomorphic to the monoid of all isomorphism types of finitely generated projective right ( resp . , left ) @xmath1-modules , with addition defined by @xmath151+[y]=[x\oplus y]$ ] ( where @xmath151 $ ] now denotes the isomorphism class of @xmath152 ) . we shall also set @xmath153)$ ] . the assignment @xmath154 can be extended to a _ functor _ from @xmath112 to @xmath103 : for a homomorphism @xmath124 of rings ( resp . , of unital rings ) , there is a unique monoid homomorphism @xmath155 such that @xmath156_r)=[{\mathrm{i}}_\infty(f)(a)]_s$ ] for each @xmath157 . the functor @xmath8 preserves all direct limits ( i.e. , in categorical language , _ directed colimits _ ) and finite products : for instance , @xmath158 ( for direct limits ) and @xmath159 . all these facts extend to the unital case , giving a functor @xmath160 , with @xmath161)$ ] for every unital ring @xmath1 . if @xmath1 is unital , @xmath14 is defined as the ( preordered ) grothendieck group of @xmath2 . for a two - sided ideal @xmath61 of a ring @xmath1 and the inclusion map @xmath162 , the map @xmath163 embeds @xmath164 into @xmath2 . we shall thus often identify @xmath164 with its image , under @xmath163 , in @xmath2 . it is an _ o - ideal _ of @xmath2 ( cf . section [ s : basicmon ] ) . [ l : vmeasdec ] let @xmath1 be a ring , let @xmath165 , and let @xmath166 . if @xmath167=\alpha+\beta$ ] , then there are orthogonal idempotents @xmath131 such that @xmath168 , @xmath144=\alpha$ ] , and @xmath169=\beta$ ] . observe , in particular , that if @xmath170 , then @xmath171 . there are orthogonal idempotents @xmath172 such that @xmath173=\alpha$ ] and @xmath174=\beta$ ] . as @xmath175=\alpha+\beta=[c]$ ] , there are @xmath135 such that @xmath176 while @xmath177 . the matrices @xmath178 and @xmath179 are as required . the following result is observed , for unital exchange rings @xmath1 and @xmath180 , in the course of the proof of pardo ( * ? ? ? * teorema 4.1.7 ) . it is contained , in full generality , in ara and goodearl ( * ? ? ? * proposition 10.10 ) . [ l : genv(i ) ] let @xmath1 be a ring , let @xmath181 , and denote by @xmath61 the two - sided ideal of @xmath1 generated by the entries of all the elements of @xmath22 . then @xmath164 is the o - ideal of @xmath57 generated by @xmath182_r}\mid{x\in e}\right\}}}$ ] . a ring @xmath1 has _ index of nilpotence at most @xmath78 _ if @xmath183 implies that @xmath184 for all @xmath185 . it is proved in yu ( * ? ? ? * corollary 4 ) that every unital exchange ring with finite index of nilpotence has stable rank @xmath9 . the following easy result gives some more information . [ l : basicvcak ] let @xmath78 be a positive integer and let @xmath1 be a unital ring with index of nilpotence at most @xmath78 . then @xmath150 $ ] has index at most @xmath78 in @xmath2 . furthermore , if @xmath2 satisfies the refinement property , then it is the positive cone of a dimension group . let @xmath186 such that @xmath187 $ ] . by lemma [ l : vmeasdec ] , there are orthogonal idempotents @xmath188 , , @xmath189 , @xmath106 of @xmath1 such that @xmath190 while @xmath191=\xi$ ] for each @xmath192 . for each @xmath193 , let @xmath194 be an isomorphism of right @xmath1-modules , and denote by @xmath195 the unique endomorphism of @xmath196 extending all the @xmath197 such that @xmath198 . setting @xmath199 , we obtain that @xmath200 , thus @xmath183 , and thus , by assumption , @xmath184 , so @xmath201 , and so , as @xmath202 , we get @xmath203 , therefore @xmath204 . this shows that @xmath205 $ ] has index at most @xmath78 in @xmath2 . suppose now that @xmath2 is a refinement monoid . as @xmath205 $ ] is an order - unit in that monoid , it follows that every element of @xmath2 has finite index ( cf . wehrung ( * ? ? ? * corollary 3.12 ) ) , and so @xmath2 is the positive cone of a dimension group ( cf . wehrung ( * ? ? ? * proposition 3.13 ) ) . in case @xmath1 is a c-algebra , each @xmath117 , for @xmath74 , is also a c-algebra , and it is often convenient to replace _ algebraic equivalence of idempotents _ by _ * -equivalence of projections _ ( a _ projection _ is a self - adjoint idempotent ) , namely , @xmath206 this can be done , because every idempotent in a c-algebra is equivalent to a projection ( and even generates the same right ideal , cf . kaplansky ( ? ? ? * theorem 26 , p. 34 ) ) and two projections are equivalent iff they are * -equivalent ( cf . kaplansky ( * ? ? ? * theorem 27 , p. 35 ) ) . [ l : ibarc * ] let @xmath207 be a left ideal in a c-algebra @xmath208 . then every projection of @xmath209 is equivalent to some projection of @xmath207 . let @xmath210 be a projection of @xmath209 . there exists a self - adjoint @xmath211 such that @xmath212 . by rrdam , larsen , and laustsen ( ? ? * lemma 2.2.3 ) , the spectrum of @xmath106 is contained in @xmath213\cup[2/3,4/3]$ ] . let @xmath214 the unique continuous function such that @xmath215 for each @xmath216 , @xmath217 for each @xmath218 , and @xmath110 is affine on @xmath219 $ ] . observe that @xmath220 for @xmath221 $ ] and @xmath222 for @xmath223 $ ] , thus , using continuous function calculus , @xmath224 is a projection . it belongs to @xmath207 as @xmath211 , and @xmath225 . now @xmath226 , so @xmath227 . it is known ( cf . jacobson ( * ? ? ? * theorem 1 ) , herstein ( * ? ? ? * theorem 1.2.3 ) ) that the jacobson radical @xmath228 of a ring ( not necessarily unital ) @xmath1 is the largest two - sided ideal @xmath229 of @xmath1 such that @xmath230 . this concept is left - right symmetric . a ring @xmath1 is _ semiprimitive _ ( _ semi - simple _ in jacobson ( * ? ? ? * section 2 ) or herstein @xcite , but we are using the seemingly more common terminology ) if @xmath231 . not every exchange ring is semiprimitive ( consider the upper triangular matrices over a field , cf . example [ ex : trmat ] ) . the following is contained in jacobson ( * ? ? ? * theorem 8) . [ p : regsemipr ] every regular ring is semiprimitive . the following is contained in dixmier ( * ? ? ? * thorme 2.9.7 ) . [ p : csemipr ] every c-algebra is semiprimitive . _ exchange rings _ have been first defined in the unital case in warfield @xcite , then in the general case by ara @xcite . a ring @xmath1 is an exchange ring if for all @xmath185 , there are an idempotent @xmath232 and @xmath233 such that @xmath234 . this condition is left - right symmetric . not every exchange ring is a two - sided ideal in a unital exchange ring , see ( * ? ? ? * example 4 , page 412 ) . a ring @xmath1 is ( von neumann ) _ regular _ if for all @xmath185 , there exists @xmath235 such that @xmath236 . every regular ring is an exchange ring , and the converse fails ( cf . example [ ex : trmat ] ) . the following result is established in the unital case in ara , goodearl , omeara , and pardo ( * ? ? ? * theorem 7.2 ) , and then extended to the non - unital case in ara ( * ? ? ? * theorem 3.8 ) . [ p : rr0exch ] a c-algebra has real rank @xmath5 iff it is an exchange ring . the following is contained in proposition 1.3 and theorem 1.4 in ara @xcite . [ l : ereerch ] the following statements hold , for any exchange ring @xmath1 . 1 . @xmath237 is an exchange ring , for any idempotent @xmath232 . @xmath117 is an exchange ring , for any positive integer @xmath78 . it is well - known that lemma [ l : ereerch ] extends to c-algebras of real rank @xmath5 ( cf . corollary 2.8 and theorem 2.10 in brown and pedersen @xcite ) . the following result is proved in ( * ? ? ? * proposition 1.5 ) . [ p : v(exch ) ] the monoid @xmath2 has refinement , for every exchange ring @xmath1 . the first part of the following result is proved in ara ( * ? ? ? * theorem 2.2 ) . the second part is proved , in the unital case , in ara , goodearl , omeara , and pardo ( * ? ? ? * proposition 1.4 ) . the proof can be trivially extended to the non - unital case , by using lemma [ l : ereerch ] . [ p : v(r / i ) ] let @xmath61 be a two - sided ideal in a ring @xmath1 . then @xmath1 is an exchange ring iff @xmath61 and @xmath238 are both exchange rings and idempotents can be lifted modulo @xmath61 . furthermore , if @xmath1 is an exchange ring , then the canonical homomorphism @xmath239 is surjective , and it induces an isomorphism @xmath240 . in the statement of proposition [ p : v(r / i ) ] , @xmath164 is identified with its image in @xmath2 , which is an o - ideal of @xmath2 . the quotient @xmath241 is defined in and the comment following . from now on we shall denote by @xmath242 ( resp . , @xmath243 ) the category of all regular rings ( resp . , exchange rings ) with ring homomorphisms . [ d : perp ] elements @xmath244 and @xmath245 in a conical commutative monoid @xmath62 are _ orthogonal _ , in notation @xmath246 , if the following statement holds : @xmath247 equivalently , the o - ideals generated by @xmath244 and @xmath245 , respectively , meet in @xmath67 . [ l : v(a)capv(b ) ] let @xmath1 be an exchange ring , let @xmath131 with @xmath144_r\perp[b]_r$ ] , and denote by @xmath208 and @xmath13 the two - sided ideals of @xmath1 generated by the entries of @xmath106 and @xmath248 , respectively . then @xmath249 is contained in @xmath228 . set @xmath250 . it follows from lemma [ l : genv(i ) ] that for each @xmath251 , there exists @xmath74 such that @xmath252\leq n[a]$ ] and @xmath252\leq n[b]$ ] . as @xmath144\perp[b]$ ] , it follows that @xmath253 . hence @xmath254 . as @xmath61 is an exchange ring without nonzero idempotents , it must be contained in @xmath228 . [ d : wvsem ] a ring @xmath1 is _ weakly v - semiprimitive _ if for all @xmath131 , the relation @xmath144_r\perp\nobreak[b]_r$ ] implies that @xmath255 . as an immediate consequence of lemma [ l : v(a)capv(b ) ] , we record the following . [ p : exch2wvsem ] every exchange ring is weakly v - semiprimitive . the converse of proposition [ p : exch2wvsem ] fails : not every weakly v - semiprimitive ring is an exchange ring ( e.g. , the ring @xmath256 of all integers ) . consider the diagram @xmath48 represented in the left hand side of figure [ fig : oneloop ] : by definition , @xmath257 and @xmath258 , for all @xmath259 . @xmath260+(-5,7)}@+{[0,0]+(-15,0)}@+{{[0,0]+(-5,-7)}}}}_{{\boldsymbol{s}}}\ar[rr]^{{\boldsymbol{h } } } & & { \mathbb{z}}^+ & & & & r{\ar@'{@+{[0,0]+(-5,7)}@+{[0,0]+(-15,0)}@+{{[0,0]+(-5,-7)}}}}_{s}\ar[rr]^h & & s } \ ] ] the diagram @xmath48 and theorem [ t : firstnonlift ] were both suggested to the author by ken goodearl , thus improving the author s original formulation of corollary [ c : firstnonlift ] ( which was stated for functors @xmath261 ) . [ t : firstnonlift ] the diagram @xmath48 has no lifting , with respect to the functor @xmath8 , by any diagram labeled as on the right hand side of figure in such a way that @xmath1 is weakly v - semiprimitive and @xmath46 . let @xmath262 be a natural equivalence , with component isomorphisms @xmath263 and @xmath264 . we obtain a commutative diagram as on figure [ fig : unfoldingloop ] . @xmath265^{\operatorname{v}(s)}\ar[d]_{\eta_r } & & \operatorname{v}(r)\ar[rr]^{\operatorname{v}(h)}\ar[d]_{\eta_r } & & \operatorname{v}(s)\ar[d]_{\eta_s}\\ ( { \mathbb{z}}^+)^2\ar[rr]^{{\boldsymbol{s } } } & & ( { \mathbb{z}}^+)^2\ar[rr]^{{\boldsymbol{h } } } & & { \mathbb{z}}^+ } \ ] ] there are @xmath131 such that @xmath266_r)=(1,0)$ ] and @xmath267_r)=(0,1)$ ] . as @xmath268 is an isomorphism , @xmath144_r\perp[b]_r$ ] . chasing around the diagram of figure [ fig : unfoldingloop ] , we get @xmath269_r)=(0,1)={\boldsymbol{s}}(1,0)=({\boldsymbol{s}}\circ\eta_r)([a]_r)= ( \eta_r\circ\operatorname{v}(s))([a]_r)=\eta_r([s(a)]_r)\,,\ ] ] thus @xmath169_r=[s(a)]_r$ ] , and thus @xmath144_r\perp[s(a)]_r$ ] , and so , as @xmath1 is weakly v - semiprimitive , @xmath270 . hence there exists @xmath271 such that @xmath272 . by applying @xmath273 to each side of this equation and observing that @xmath274 while @xmath46 , we obtain that @xmath275 , thus , multiplying this equation on the left by @xmath276 , we get @xmath277 . therefore , @xmath278_r)=(\eta_s\circ\operatorname{v}(h))([a]_r)= \eta_s([h(a)]_s)=0\quad\text{in } { \mathbb{z}}^+\,,\ ] ] a contradiction . [ c : firstnonlift ] denote by @xmath279 the category of all simplicial monoids with order - unit and normalized monoid homomorphisms , and by @xmath280 the category of all weakly v - semiprimitive rings and ring homomorphisms . then there is no functor @xmath281 such that @xmath4 . we shall see in proposition [ p : vecclift ] that while the diagram @xmath48 has no lifting , with respect to the functor @xmath8 , by weakly v - semiprimitive rings with @xmath46 , it has one by unital c-algebras . as the following result will be superseded by corollary [ c : nolift3 ] , we shall only outline its proof . [ p : nonsplit ] denote by @xmath282 the category of all simplicial groups with order - unit with normalized positive group embeddings . then there is no functor@xmath283 such that @xmath284 . we shall first invoke some ideas from universal algebra and lattice theory ( cf . mckenzie , mcnulty , and taylor @xcite for background ) . a _ bounded lattice _ is a lattice with smallest element and largest element ( usually denoted by @xmath5 and @xmath9 , respectively ) . a _ variety of bounded lattices _ is the class of all bounded lattices that satisfy a given set of identities , written in the language @xmath285 . for a variety @xmath286 of bounded lattices and a set @xmath152 , we denote by @xmath287 the _ free object _ on @xmath152 within the variety @xmath286 . we shall denote by @xmath288 the lattice of length two with three atoms ( cf . figure [ fig : m3 ] ) and by @xmath289 the variety that it generates . ] as @xmath288 is finite , the variety @xmath289 is _ locally finite _ , that is , @xmath290 is finite whenever @xmath152 is finite ( this is a classical argument of universal algebra , see , for example , the proof of mckenzie , mcnulty , and taylor ( ? ? ? * lemma 4.98 ) ) . furthermore , @xmath291 with canonical transition maps and limiting maps , for any set @xmath152 . we proceed by invoking a lattice - theoretical tool . the _ dimension monoid _ @xmath292 of a lattice @xmath207 , defined in wehrung ( * ? ? ? * definition 1.1 ) , is always a conical commutative monoid ; furthermore , the assignment @xmath293 extends to a _ functor_. as in @xcite , we define @xmath294 as the ( preordered ) grothendieck group of @xmath292 . now fix any set @xmath152 with at least @xmath16 elements . by ( * corollary 10.30 ) , there is no regular ring @xmath1 such that @xmath295 . furthermore , as the @xmath296 functor preserves direct limits ( cf . * proposition 1.4 ) ) , it follows from that @xmath297 and thus @xmath298 for each finite @xmath299 , the lattice @xmath300 is finite modular , thus its dimension monoid @xmath301 is a finitely generated simplicial monoid ( cf . * proposition 5.5 ) ) ; in particular , it embeds into its grothendieck group @xmath302 . furthermore , @xmath300 is a retract of @xmath290 ( send every element of @xmath303 to @xmath5 ) , thus @xmath304 is a retract of @xmath302 . finally , all the transition morphisms in are trivially seen to preserve the canonical order - units ( namely , using the notation of @xcite , @xmath305 ) . therefore , the direct limit in is , in fact , a _ direct union of dimension groups with order - unit_. hence , by assumption on @xmath3 , we can define an af c-algebra as a c-direct limit , @xmath306 let @xmath1 be a dense locally matricial algebra ( over @xmath307 ) in @xmath208 . then @xmath308 . furthermore , as the functor @xmath0 preserves c-direct limits and @xmath284 , it follows from that @xmath309 so @xmath310 , and so @xmath311 . however , as @xmath1 is von neumann regular , this is a contradiction . the proof of proposition [ p : nonsplit ] may feel quite outlandish to a number of readers : while its statement is , essentially , combinatorial , its proof involves the counterexample of cardinality @xmath16 introduced in @xcite . extracting the gist of that proof will lead us to the proof of theorem [ t : nolift ] , which is , to a large extent , a finitary result , involving neither transfinite cardinals , nor universal algebra , nor lattice theory . it will also extend proposition [ p : nonsplit ] to c-algebras of real rank @xmath5 . [ d : meas ] let @xmath62 be a _ conical _ commutative monoid and let @xmath1 be a ring . @xmath62-valued premeasure on @xmath1 _ is a map @xmath312 such that * @xmath313 ; * @xmath314 , for all orthogonal idempotents @xmath131 ; * @xmath134 implies that @xmath315 , for all @xmath131 . ( _ recall that @xmath145 stands for murray - von neumann equivalence , see section _ . ) we say that the premeasure @xmath316 is a _ measure _ if it satisfies the condition * @xmath317 iff @xmath253 , for each @xmath318 . we say that the premeasure @xmath316 is a _ v - premeasure _ if it satisfies the following _ v - condition _ : * for all @xmath165 and all @xmath319 such that @xmath320 , there are orthogonal idempotent @xmath131 such that @xmath168 , @xmath321 , and @xmath322 . we say that @xmath316 is a _ v - measure _ if it is both a measure and a v - premeasure . we shall use the notation @xmath323 ( the _ codomain _ of @xmath316 ) . a fundamental example of a v - measure is the following . [ ex : canvmeas ] let @xmath1 be a ring . then the assignment @xmath324 , @xmath325_r$ ] is a measure on @xmath1 . due to lemma [ l : vmeasdec ] , @xmath326 is actually a v - measure . we call it the _ canonical v - measure _ on @xmath1 . [ d : vhom ] let @xmath62 and @xmath327 be conical commutative monoids . a homomorphism @xmath108 is a _ pre - v - homomorphism _ if whenever @xmath328 and @xmath329 with @xmath330 , there are @xmath331 such that @xmath332 , @xmath333 , and @xmath334 . if , in addition , @xmath335 , then we say that @xmath110 is a _ v - homomorphism _ dobbertin ( * ? ? ? * definition 1.2 ) ) . the following result relates the definition of a measure given above , the canonical v - measure , and the notion of v - homomorphism . it expresses the universality of @xmath326 as a premeasure on @xmath1 , and its proof is a straightforward exercise . [ p : factmeas ] let @xmath1 be a ring , let @xmath62 be a conical commutative monoid , and let @xmath316 be an @xmath62-valued premeasure on @xmath1 . then there exists a unique monoid homomorphism @xmath336 such that @xmath337 . furthermore , 1 . @xmath316 is a measure iff @xmath338 ; 2 . @xmath316 is a v - premeasure iff @xmath339 is a pre - v - homomorphism . [ d : meascat ] for a category @xmath111 of rings and ring homomorphisms , we shall denote by @xmath340 the category where the objects ( _ premeasured rings _ ) are the pairs @xmath341 , where @xmath316 is a premeasure on the ring @xmath342 ( with values in some conical commutative monoid ) , and , for objects @xmath341 and @xmath343 , a _ morphism _ from @xmath341 to @xmath343 is a pair @xmath344 , where @xmath124 is a ring homomorphism , @xmath345 is a monoid homomorphism , and @xmath346 ( cf . figure [ fig : meashom ] ) . @xmath347_{f } & & s & & { \mathrm{i}}_\infty(r)\ar[rr]_{{\mathrm{i}}_\infty(f)}\ar[d]_{\mu } & & { \mathrm{i}}_\infty(s)\ar[d]^{\nu}\\ m_{\mu}\ar[rr]^{\tilde{f } } & & m_{\nu } & & m_{\mu}\ar[rr]^{\tilde{f } } & & m_{\nu } } \ ] ] the composition of morphisms is defined componentwise ( i.e. , @xmath348 ) . we define similarly @xmath349 , @xmath350 , @xmath351 , using measures , v - premeasures , and v - measures , respectively , instead of premeasures . next , we introduce two functors : one _ to _ @xmath352 , and one _ from _ @xmath352 . [ d : twofunct ] the _ canonical functor _ @xmath353 is defined by @xmath354 ( cf . example [ ex : canvmeas ] ) and @xmath355 , for any rings @xmath1 and @xmath356 and any ring homomorphism @xmath124 . the _ projection functor _ @xmath357 is defined by @xmath358 and @xmath359 , for any premeasured rings @xmath341 and @xmath343 and any morphism @xmath360 . [ rk : picirctau ] it is trivial that @xmath361 . in particular , if a diagram can be lifted by _ rings _ with respect to the functor @xmath8 , then it can be lifted by _ premeasured rings _ with respect to the functor @xmath51 . from now on , we shall often omit the second coordinate @xmath362 in a morphism @xmath344 from @xmath352 , thus simply writing @xmath363 and specifying what is @xmath362 in case needed . [ l : reflsubcat ] the category @xmath364 is a reflective subcategory of @xmath365 , with reflector @xmath366 satisfying @xmath367 . furthermore , 1 . @xmath368 sends surjective premeasures to surjective measures ; 2 . @xmath368 sends v - premeasures to v - measures ; 3 . @xmath368 sends regular rings to regular rings . let @xmath316 be a premeasure , with values in a conical commutative monoid @xmath62 , on an exchange ring @xmath1 , and denote by @xmath369 the canonical monoid homomorphism . denote by @xmath61 the two - sided ideal of @xmath1 generated by the entries of all @xmath370 such that @xmath371 . by lemma [ l : genv(i ) ] together with the conicality of @xmath62 , @xmath317 for each @xmath251 ; so @xmath372 . therefore , denoting by @xmath373 the canonical projection , there exists a unique monoid homomorphism @xmath374 such that @xmath375 . then @xmath376 is an exchange ring , while , composing @xmath377 with the canonical isomorphism @xmath378 ( cf . proposition [ p : v(r / i ) ] ) and using the fact that idempotents can be lifted modulo @xmath61 , we obtain a ( necessarily unique ) monoid homomorphism @xmath379 such that @xmath380_{r / i})=\mu(e)\,,\quad \text{for each } e\in{\mathrm{i}}_\infty(r)\,.\ ] ] the map @xmath381 defined by the rule @xmath382_{r^})$ ] for each @xmath383 is an @xmath62-valued premeasure on @xmath384 , and @xmath385 for each @xmath318 such that @xmath386 , that is , @xmath317 , it follows from the definition of @xmath61 that @xmath387 . therefore , _ @xmath381 is an @xmath62-valued measure on @xmath384_. @xmath388 is the @xmath364-reflection of @xmath341 , with @xmath364-reflection morphism the canonical projection @xmath389 with @xmath390 . we must prove that for every measured exchange ring @xmath343 and every morphism @xmath391 , there exists a unique morphism@xmath392 such that @xmath393 . for each @xmath64 , @xmath394 , thus , as @xmath395 is a measure , @xmath396 . it follows that there exists a unique ring homomorphism @xmath397 such that @xmath393 . setting @xmath398 , we obtain that @xmath195 is as desired . denote by @xmath399 the @xmath364-reflection functor . as @xmath316 and @xmath381 have the same codomain and @xmath398 in the proof of the claim above , @xmath400 and the surjectivity of @xmath316 implies the one of @xmath381 . if @xmath316 is a v - premeasure , then , as idempotents can be lifted modulo @xmath61 , so is @xmath381 ( cf . ) . finally , if @xmath1 is a regular ring , then @xmath401 is also regular . the following analogue of lemma [ l : reflsubcat ] for c-algebras of real rank @xmath5 is valid . [ l : reflsubrr0 ] the category @xmath402 is a reflective subcategory of @xmath403 , with reflector @xmath404 satisfying @xmath405 . furthermore , 1 . @xmath406 sends surjective premeasures to surjective measures ; 2 . @xmath406 sends v - premeasures to v - measures . using proposition [ p : rr0exch ] , the proof is almost identical to the one of lemma [ l : reflsubcat ] . the main difference is the need to replace the ideal @xmath61 of @xmath1 by its topological closure @xmath407 ( so @xmath408 is still a c-algebra , necessarily of real rank @xmath5 , see brown and pedersen ( ? ? ? * theorem 3.14 ) ) . we need to prove that @xmath371 for each projection @xmath138 of @xmath407 . by lemma [ l : ibarc * ] , @xmath138 is equivalent to a projection @xmath140 of @xmath61 . by the proof of lemma [ l : reflsubcat ] , @xmath409 ; thus @xmath371 . now the reflection morphism @xmath404 is defined by @xmath410 , where this time @xmath411 while @xmath412 is the canonical projection and @xmath413 . the rest of the proof runs as the one of lemma [ l : reflsubcat ] . the following definition introduces a strengthening of the weakly v - semiprimitive rings introduced in definition [ d : wvsem ] . [ d : vsemi ] a ring @xmath1 is _ v - semiprimitive _ if for all @xmath131 , @xmath144_r\perp[b]_r$ ] in @xmath2 implies that @xmath414 . if @xmath2 is totally ordered with respect to its algebraic preordering , then @xmath1 is v - semiprimitive . in particular , this holds if @xmath415 . this isomorphism occurs in case @xmath1 is a principal ideal domain . due to the quillen - suslin theorem , it also holds in case @xmath1 is a polynomial ring over a field . another important class of v - semiprimitive rings , which also explains the terminology , is provided by the following result . [ p : seexvsem ] every semiprimitive exchange ring is v - semiprimitive . we emphasize here that the rings in question are not necessarily unital . let @xmath1 be a semiprimitive exchange ring and let @xmath131 such that @xmath144\perp\nobreak[b]$ ] in @xmath2 . denoting by @xmath208 and @xmath13 the two - sided ideals of @xmath1 generated by the entries of @xmath106 and @xmath248 , respectively , it follows from lemma [ l : v(a)capv(b ) ] that @xmath249 is contained in @xmath228 , thus , by assumption , @xmath416 , so @xmath414 . the following two examples show that none of the assumptions of semiprimitivity and exchange property can be dispensed with in the statement of proposition [ p : seexvsem ] . [ ex : trmat ] let @xmath53 be a field ( any unital exchange ring would do ) . then the ring @xmath1 of all upper triangular @xmath417 matrices over @xmath53 is an exchange ring , but it is not v - semiprimitive . to show the latter statement , set @xmath418 and @xmath419 . observe that @xmath420 , and once the two monoids are identified , @xmath144=(1,0)$ ] while @xmath169=(0,1)$ ] . in particular , @xmath144\perp[b]$ ] , although @xmath421 is nonzero . [ ex : semiprnotv ] we denote by @xmath422 the c-algebra of all @xmath208-valued continuous functions on a topological space @xmath152 , for any c-algebra @xmath208 . we denote by @xmath423 $ ] the unit interval of @xmath424 and we set @xmath425,{\mathrm{m}}_2({\mathbb{c}}))}\mid{x(0)\text { is a diagonal matrix}}\right\}}}\,.\ ] ] this type of construction appears in su ( * ? ? ? * section 1.2 ) . the ring @xmath49 is a key ingredient in the main construction in section [ s : calglift ] . then @xmath49 is a semiprimitive unital ring ( because it is a unital c-algebra , cf . proposition [ p : csemipr ] ) . however , @xmath49 is not v - semiprimitive . indeed , denoting by @xmath426{\hookrightarrow}{\mathbb{c}}$ ] the inclusion map and setting @xmath427 then @xmath106 and @xmath248 are both idempotent and @xmath428 is nonzero . however , we prove in section [ s : calglift ] that @xmath429 , _ via _ an isomorphism that is easily seen to send @xmath144 $ ] to @xmath430 and @xmath248 to @xmath431 . in particular , @xmath144\perp[b]$ ] in @xmath432 . therefore , @xmath49 is not v - semiprimitive . [ l : bracklift ] let @xmath1 be a v - semiprimitive ring , let @xmath62 be a conical commutative monoid , and let @xmath312 be a measure . let @xmath433 with @xmath434 and @xmath435 while @xmath436 and @xmath437 in @xmath62 . then @xmath438 . denote by @xmath336 the unique v - homomorphism such that @xmath439)$ ] for each @xmath370 ( cf . proposition [ p : factmeas ] ) . we claim that @xmath144\perp[e - b]$ ] in @xmath2 . indeed , otherwise there would exist a nonzero @xmath370 such that @xmath440\leq[a]$ ] and @xmath440\leq[e - b]$ ] . by applying @xmath339 to those inequalities , we obtain @xmath441 and @xmath442 , a contradiction as @xmath443 and @xmath436 , and thus proving our claim . as @xmath1 is v - semiprimitive , it follows that @xmath444 , thus , as @xmath434 , we get that @xmath445 . by symmetry , @xmath446 . the main combinatorial object underlying the present paper is a commutative diagram , which we shall denote by @xmath7 , of simplicial monoids and monoid homomorphisms . this diagram is indexed by the _ cube _ , that is , the partially ordered set of all subsets of a three - element set . its vertices are the commutative monoids @xmath447 and @xmath448 ( @xmath449 ) . the monoid homomorphisms labeling its edges are the identity map , together with the maps @xmath450 , @xmath451 , and @xmath452 defined by @xmath453 , @xmath454 , and @xmath455 for all @xmath456 . ( the maps @xmath457 and @xmath458 were already introduced in section [ s : firstdiagr ] . ) we shall later consider the diagram , denoted by @xmath459 , obtained by associating order - units to the vertices of the diagram @xmath7 : @xmath9 is associated to @xmath460 , @xmath461 is associated to both @xmath462 and @xmath463 for each @xmath449 , and @xmath11 is associated to @xmath464 . observe that @xmath459 is a commutative diagram of _ pointed monoids _ , which means here that @xmath465 and @xmath466 . the diagrams @xmath7 and @xmath459 are represented on the left- and right - hand side of figure [ fig : diagrd ] , respectively . @xmath467^{{\boldsymbol{h } } } & { \boldsymbol{b}}_1\ar[u]_{{\boldsymbol{h } } } & { \boldsymbol{b}}_0\ar[lu]_{{\boldsymbol{h } } } & ( { \boldsymbol{b}}_2,(1,1))\ar[ru]^{{\boldsymbol{h } } } & ( { \boldsymbol{b}}_1,(1,1))\ar[u]_{{\boldsymbol{h } } } & ( { \boldsymbol{b}}_0,(1,1))\ar[lu]_{{\boldsymbol{h}}}\\ { \boldsymbol{a}}_0\ar@{=}[u]\ar@{=}[ur ] & { \boldsymbol{a}}_1\ar[ul]_(.7){{\boldsymbol{s}}}\ar@{=}[ur ] & { \boldsymbol{a}}_2\ar@{=}[ul]\ar@{=}[u ] & ( { \boldsymbol{a}}_0,(1,1))\ar@{=}[u]\ar@{=}[ur ] & ( { \boldsymbol{a}}_1,(1,1))\ar[ul]_(.65){{\boldsymbol{s}}}\ar@{=}[ur ] & ( { \boldsymbol{a}}_2,(1,1))\ar@{=}[ul]\ar@{=}[u]\\ & { \boldsymbol{a}}\ar[lu]^{{\boldsymbol{e}}}\ar[u]_{{\boldsymbol{e}}}\ar[ru]_{{\boldsymbol{e } } } & & & ( { \boldsymbol{a}},1)\ar[lu]^{{\boldsymbol{e}}}\ar[u]_{{\boldsymbol{e}}}\ar[ru]_{{\boldsymbol{e } } } & } \ ] ] the following theorem will involve the functor @xmath357 introduced in definition [ d : twofunct ] , together with v - semiprimitivity ( cf . definition [ d : vsemi ] ) . [ t : nolift ] there exists no lifting of @xmath7 , with respect to the functor @xmath51 , by any commutative diagram of premeasured rings satisfying the following conditions : 1 . the lifts of @xmath468 , @xmath469 , @xmath470 are surjective v - premeasures ; 2 . the lifts of @xmath471 , @xmath472 , @xmath473 are surjective v - measures ; 3 . the underlying rings of the lifts of @xmath471 , @xmath472 , @xmath473 are v - semiprimitive . suppose that we are given a lifting @xmath474 , with respect to the functor @xmath51 , of the diagram @xmath7 , labeled as on figure [ fig : ringlift ] ( _ we remind the reader that @xmath362 is now dropped from the notation @xmath344 for morphisms in @xmath352 _ ) . we further assume that the conditions ( i)(iii ) above hold . @xmath475^{h_2 } & ( b_1,\beta_1)\ar[u]\|-{h_1 } & ( b_0,\beta_0)\ar[lu]_{h_0}\\ & & \\ ( a_0,\alpha_0)\ar[uu]^{f_2}\ar[uur]^(.75){f_1 } & ( a_1,\alpha_1)\ar[uul]\|-(.72){{\vrule height 10pt depth 4pt width 0pt}g_2}\ar[uur]\|-(.72){{\vrule height 10pt depth 4pt width 0pt}f_0 } & ( a_2,\alpha_2)\ar[uul]_(.75){g_1}\ar[uu]_(.7){g_0}\\ & ( a,\alpha)\ar[lu]^{e_0}\ar[u]_{e_1}\ar[ru]_{e_2 } & } \ ] ] by composing the premeasures in the diagram @xmath474 with the arrows of the natural equivalence given by the relation @xmath476 , we may assume that @xmath477 . this means that the following equalities hold for each @xmath449 : @xmath478 , @xmath479 , @xmath480 , @xmath481 , @xmath482 , and @xmath483 . there exists @xmath484 such that @xmath485 . set @xmath486 , for each @xmath449 . then @xmath487 . likewise , `` propagating '' those new idempotents up the diagram @xmath474 , we obtain idempotent matrices @xmath488 , for @xmath449 , and @xmath489 such that @xmath490 and cyclically , @xmath491 , @xmath492 , and @xmath493 . define new matrices @xmath494 , for @xmath449 , as follows . as @xmath495 in @xmath462 and @xmath496 is a v - premeasure , there exists @xmath494 such that @xmath497 , @xmath498 , and @xmath499 . set @xmath500 then @xmath501 and @xmath502 both belong to @xmath503 and @xmath504 , for each @xmath449 . further calculations yield easily @xmath505 for example , @xmath506 while @xmath507 by evaluating the equation @xmath508 at @xmath509 , we obtain @xmath510 . by evaluating @xmath511 at @xmath512 , we obtain @xmath513 . by evaluating @xmath514 at @xmath515 , we obtain @xmath516 . now , as each @xmath517 is v - semiprimitive and by , it follows from lemma [ l : bracklift ] that @xmath518 for each @xmath449 . therefore , @xmath519 a contradiction as @xmath520 and @xmath521 are both idempotent with @xmath522 and @xmath523 . we observed in remark [ rk : picirctau ] that any lifting of a diagram with respect to the functor @xmath8 yields a lifting of the same diagram with respect to the functor @xmath51 . as the canonical measure on a ring is a v - measure ( cf . example [ ex : canvmeas ] ) , we obtain the following immediate consequence of theorem [ t : nolift ] . [ c : nolift1 ] there exists no lifting of @xmath7 , with respect to the functor @xmath8 , by any commutative diagram of rings and ring homomorphisms in which the lifts of @xmath471 , @xmath472 and @xmath473 are all v - semiprimitive . [ c : noreglift ] there exists no lifting of @xmath7 , with respect to the functor @xmath51 , by any commutative diagram of v - premeasured regular rings . suppose otherwise , and let @xmath474 be a commutative diagram of v - premeasured regular rings such that @xmath476 . by lemma [ l : reflsubcat ] , the diagram @xmath524 is a commutative diagram of _ v - measured _ regular rings and @xmath525 . but every regular ring is a semiprimitive exchange ring ( cf . proposition [ p : regsemipr ] ) , thus it is v - semiprimitive ( cf . proposition [ p : seexvsem ] ) . this contradicts theorem [ t : nolift ] . a similar proof , using lemma [ l : reflsubrr0 ] instead of lemma [ l : reflsubcat ] , yields the following . [ c : norr0lift ] there exists no lifting of @xmath7 , with respect to the functor @xmath51 , by any commutative diagram of v - premeasured c-algebras of real rank @xmath5 . by using remark [ rk : picirctau ] , we obtain the following negative lifting results with respect to the functor @xmath8 . [ c : novlift ] there is no lifting , with respect to the functor @xmath8 , of the diagram @xmath7 by any diagram of regular rings . nonetheless , we shall see that @xmath7 can be lifted , with respect to the functor @xmath8 , by a diagram of exchange rings ( cf . proposition [ p : cdliftedexch ] ) and by a diagram of c-algebras ( cf . proposition [ p : cdliftedc * ] ) . in order to be able to extend proposition [ p : nonsplit ] to c-algebras of real rank @xmath5 , we need to find an analogue of the diagram @xmath7 in which the arrows are all embeddings . fortunately , this can be done . we denote by @xmath526 and @xmath527 , @xmath528 , @xmath529 the maps defined by @xmath530 for all @xmath531 . further , we denote by @xmath50 the commutative diagram of simplicial monoids and normalized monoid order - embeddings represented in figure [ fig : diagre ] . @xmath532^{{\boldsymbol{a } } } & { \vrule height 10pt depth 4pt width 0pt}{\boldsymbol{b}}'_1:=({\mathbb{z}}^+)^4 \ar@{_(->}[u]_{{\boldsymbol{b } } } & { \vrule height 10pt depth 4pt width 0pt}{\boldsymbol{b}}'_0:=({\mathbb{z}}^+)^4\ar@{^(->}[lu]_{{\boldsymbol{c}}}\\ { \boldsymbol{a}}_0:=({\mathbb{z}}^+)^2\ar@{_(->}[u]^{{\boldsymbol{f}}}\ar@{^(->}[ur]_(.75){{\boldsymbol{f } } } & { \boldsymbol{a}}_1:=({\mathbb{z}}^+)^2\ar@{_(->}[ul]_(.6){{\boldsymbol{g}}}\ar@{^(->}[ur]_(.7){{\boldsymbol{f } } } & { \boldsymbol{a}}_2:=({\mathbb{z}}^+)^2\ar@{_(->}[ul]^(.7){{\boldsymbol{g}}}\ar@{^(->}[u]_{{\boldsymbol{g}}}\\ & { \boldsymbol{a}}:={\mathbb{z}}^+\ar@{_(->}[lu]^{{\boldsymbol{e}}}\ar@{_(->}[u]_{{\boldsymbol{e}}}\ar@{^(->}[ru]_{{\boldsymbol{e } } } & } \ ] ] [ l : frometod ] there exists a natural transformation @xmath533 in which all the arrows are surjective pre - v - homomorphisms . we describe the components of the natural transformation . the morphism @xmath534 is defined by the rule@xmath535 . * the morphism @xmath536 is defined by the rule @xmath537 . * the morphism @xmath538 is defined by the rule @xmath539 . * the morphism @xmath540 is defined by the rule @xmath541 . * the morphism @xmath542 is the identity on @xmath543 for each @xmath544 , and @xmath545 is the identity on @xmath546 . each of these arrows is the canonical projection from @xmath547 onto @xmath62 , for suitable ( simplicial ) monoids @xmath62 and @xmath327 ; hence it is a surjective pre - v - homomorphism . verifying that @xmath548 is a natural transformation ( which , in reality , amounts to verifying the commutativity of only five squares ) is trivial . denote by @xmath549 the category of all simplicial monoids with order - unit , with normalized monoid order - embeddings ( the latter means that the maps @xmath110 satisfy @xmath550 implies @xmath69 , where @xmath68 denotes the algebraic preordering , which is here an ordering , on the monoid ) . actually , all arrows of the diagram @xmath50 are not only order - embeddings but _ coretractions _ , that is , they all have a left inverse , so corollaries [ c : nolift2 ] and [ c : nolift3 ] extend to the category of all simplicial monoids endowed with those maps as well . [ c : nolift2 ] there is no functor @xmath551 such that @xmath552 . we give the proof for @xmath553 . it suffices to prove that the diagram @xmath50 has no lifting , with respect to the functor @xmath51 , by a diagram of v - premeasured c-algebras of real rank @xmath5 . if @xmath554 were such a lifting , then , composing all the v - premeasures in @xmath554 with the corresponding pre - v - homomorphisms given by lemma [ l : frometod ] , we would obtain a lifting of the diagram @xmath7 , with respect to the functor @xmath51 , by a diagram in @xmath553 , a contradiction by corollary [ c : norr0lift ] . the proof for regular rings is similar , using this time corollary [ c : noreglift ] . by using remark [ rk : picirctau ] , we obtain the following extension of proposition [ p : nonsplit ] . [ c : nolift3 ] there is no functor @xmath555 such that @xmath556 . we shall see in the subsequent sections that the v - semiprimitivity assumption can not be dispensed with in the statement of theorem [ t : nolift ] . we do not know whether corollary [ c : nolift2 ] extends to all c-algebras , or even to all rings , see problems [ pb : ringsfunct ] and [ pb : calgvlift ] . five arrows out of six in the middle part of the diagram @xmath7 ( cf . figure [ fig : diagrd ] ) are identities . this suggests that the structure of @xmath7 is highly collapsible . in the present section we shall take advantage of this observation and define another diagram , denoted by @xmath17 , obtained by identifying as many `` symmetric '' parts of @xmath7 as possible while keeping its main unliftability property ( theorem [ t : nolift ] ) . denote by @xmath557 the diagram obtained by adding the natural order - units to the vertices of @xmath17 . the diagrams @xmath17 and @xmath557 are represented in figure [ fig : diagrl ] . the maps @xmath558 , @xmath458 , and @xmath457 are already arrows of @xmath7 . @xmath559+(5,-7)}@+{[0,0]+(15,0)}@+{{[0,0]+(5,7)}}}}_{{\mathrm{id } } } & & ( { \mathbb{z}}^+,2){\ar@'{@+{[0,0]+(5,-7)}@+{[0,0]+(25,0)}@+{{[0,0]+(5,7)}}}}_{{\mathrm{id } } } & & \\ { \mathbb{z}}^+\times{\mathbb{z}}^+\ar[u]_{{\boldsymbol{h } } } & & ( { \mathbb{z}}^+\times{\mathbb{z}}^+,(1,1))\ar[u]_{{\boldsymbol{h } } } & & { \boldsymbol{h}}(x , y):=x+y\\ { \mathbb{z}}^+\times{\mathbb{z}}^+{\ar@'{@+{[0,0]+(5,-7)}@+{[0,0]+(15,0)}@+{{[0,0]+(5,7)}}}}_{{\boldsymbol{s } } } \ar@/^.5pc/[u]^{{\mathrm{id}}}\ar@/_.5pc/[u]_{{\mathrm{id } } } & & ( { \mathbb{z}}^+\times{\mathbb{z}}^+,(1,1)){\ar@'{@+{[0,0]+(5,-10)}@+{[0,0]+(30,0)}@+{{[0,0]+(5,10)}}}}_{{\boldsymbol{s } } } \ar@/^.5pc/[u]^{{\mathrm{id}}}\ar@/_.5pc/[u]_{{\mathrm{id } } } & & { \boldsymbol{s}}(x , y):=(y , x)\\ { \mathbb{z}}^+\ar[u]_{{\boldsymbol{e } } } & & ( { \mathbb{z}}^+,1)\ar[u]_{{\boldsymbol{e } } } & & { \boldsymbol{e}}(x):=(x , x ) } \ ] ] [ p : noliftlu ] there is no lifting @xmath474 , with respect to the functor @xmath8 , of the diagram @xmath17 , such that , labeling @xmath474 as on figure , @xmath560 and @xmath561 is v - semiprimitive . in particular , @xmath561 can not be a semiprimitive exchange ring , and thus it can be neither a regular ring nor a c-algebra of real rank @xmath5 . @xmath562+(5,-7)}@+{[0,0]+(15,0)}@+{{[0,0]+(5,7)}}}}_{v } & & \operatorname{v}(v)={\mathrm{id}}_{{\mathbb{z}}^+}\\ c\ar[u]_{h } & & \operatorname{v}(h)={\boldsymbol{h}}\\ b{\ar@'{@+{[0,0]+(5,-7)}@+{[0,0]+(15,0)}@+{{[0,0]+(5,7)}}}}_{s } \ar@/^.5pc/[u]^{f}\ar@/_.5pc/[u]_{g } & & \operatorname{v}(f)=\operatorname{v}(g)={\mathrm{id}}_{{\mathbb{z}}^+\times{\mathbb{z}}^+}\,,\ \operatorname{v}(s)={\boldsymbol{s}}\\ a\ar[u]_{e } & & \operatorname{v}(e)={\boldsymbol{e}}}\ ] ] equations such as @xmath563 , @xmath564 , and so on , are , strictly speaking , stated only up to the natural equivalence between @xmath565 and @xmath557 . suppose that @xmath474 satisfies the given conditions . then we can `` unfold '' @xmath474 to the commutative diagram represented in figure [ fig : cubiclift ] , which lifts @xmath7 with respect to the functor @xmath8 ; a contradiction by corollary [ c : nolift1 ] . @xmath566^{h } & c\ar[u]\|-{{\vrule height 10pt depth 4pt width 0pt}vh } & c\ar[lu]_{vh}\\ b\ar[u]^{g}\ar[ur]_(.68){\ ! g } & b\ar[ul]_(.72){\ ! fs}\ar[ur]^(.72){f\!\ ! } & b\ar[ul]^(.68){g\!}\ar[u]_{g}\\ & a\ar[lu]^{e}\ar[u]_{e}\ar[ru]_{e } & } \ ] ] [ t : lulift ] let @xmath53 be a field . then the diagram @xmath557 has a lifting , with respect to the functor @xmath567 , by unital exchange algebras over @xmath53 , in such a way that the equations are satisfied and @xmath568 and @xmath521 are both involutive automorphisms . denote by @xmath569 ( resp . , @xmath570 $ ] ) the field of rational functions ( resp . the ring of polynomials ) over @xmath53 in the indeterminate @xmath571 . we denote by @xmath572 the unique automorphism of @xmath569 such that @xmath573 . observe that @xmath574 . the subalgebra @xmath575\text { and } q(0)\neq0}\right\}}}\ ] ] is a local subring of @xmath569 , with maximal ideal @xmath576 . in particular , @xmath577 is an exchange ring . the matrices @xmath578 and @xmath579 are both idempotent elements in the @xmath53-algebra @xmath580 as @xmath581 are exchange rings , it follows from nicholson ( * ? ? ? * corollary 2.6 ) that @xmath561 is an exchange subring of @xmath582 . we denote by @xmath521 the involutive automorphism of @xmath583 defined by the rule @xmath584 the set @xmath585 is a two - sided ideal of @xmath561 . furthermore , the relation @xmath586 holds for each @xmath587 ; in particular , @xmath588 is invertible in @xmath561 . hence @xmath229 is contained in the jacobson radical of @xmath561 . set @xmath589 , @xmath590 , and @xmath591 . then the map @xmath592 , @xmath593 is an involutive automorphism of @xmath13 with @xmath594 . furthermore , @xmath595 with simplicial basis @xmath596_b,[b_1]_b\right\}}$ ] . we can define a surjective homomorphism @xmath597 of @xmath53-algebras by the rule @xmath598 as the jacobson radical of @xmath599 is zero , it follows that @xmath229 contains , and thus is equal to , the jacobson radical of @xmath561 . from @xmath600 and proposition [ p : v(r / i ) ] it follows that @xmath601 _ via _ @xmath252_c\mapsto[e+{\mathrm{m}}_\infty(j)]_{c / j}$ ] , thus @xmath602 with simplicial basis @xmath603_c,[c_1]_c\right\}}$ ] . now the matrix @xmath604 is idempotent , and it belongs to @xmath561 . as @xmath605 , that is , @xmath606 , we get the first key property of @xmath52 : @xmath607_c=[c_0]_c\quad\text{and}\quad [ 1-c]_c=[1-c_0]_c\,.\ ] ] furthermore , an elementary calculation yields the second key property of @xmath52 : @xmath608 therefore , defining @xmath273 as the inclusion map from @xmath561 into @xmath583 and defining embeddings of unital @xmath53-algebras @xmath609 and @xmath610 by the rules @xmath611 we obtain the equations @xmath612 and ( using ) @xmath613 . finally , set @xmath614 and @xmath615 , @xmath616 . obviously , @xmath617 and @xmath618 . we obtain a diagram @xmath474 as on figure [ fig : liftl ] . verifying that @xmath474 lifts @xmath557 , with respect to the abovementioned choices of simplicial bases , amounts to verifying the following : @xmath619_a)&=[b_0]_b+[b_1]_b&&\text{(obvious);}\\ \operatorname{v}(f)([b_j]_b)&=[c_j]_c\quad\text{for each } j<2&&\text{(obvious);}\\ \operatorname{v}(g)([b_j]_b)&=[c_j]_c\quad\text{for each } j<2 & & \text{(this follows from~\eqref{eq : cequivc0})}\\ \operatorname{v}(h)([c_j]_c)&=[c_0]_d=[c_1]_d\quad\text{for each } j<2 & & \text{(obvious);}\\ \operatorname{v}(v)([1_d]_d)&=[1_d]_d&&\text{(obvious ) . } \end{aligned}\ ] ] this concludes the proof . the `` unfolding '' technique used in the proof of proposition [ p : noliftlu ] yields then immediately the following consequence of theorem [ t : lulift ] . [ p : cdliftedexch ] the diagram @xmath459 can be lifted , with respect to the functor @xmath567 , by a commutative diagram of unital exchange rings . for any topological space @xmath152 and any positive integer @xmath78 , we define the c-algebra @xmath620 . in case @xmath152 is given a distinguished point @xmath5 , we set @xmath621 we shall often identify @xmath622 with @xmath623 and @xmath307 with the subalgebra of @xmath624 consisting of all constant functions . this way , the c-algebra @xmath625 is a unital subalgebra of @xmath626 . convention in all the notations @xmath622 , @xmath627 , @xmath626 , @xmath628 , @xmath629 , defined above , we shall drop the subscript @xmath152 and the superscript @xmath630 in case @xmath631 $ ] and @xmath632 . we introduced the algebra @xmath633}$ ] in example . [ p : rrsrac ] the algebras @xmath634 , @xmath49 , and @xmath635 have both real rank and stable rank equal to @xmath9 . it follows from brown and pedersen ( * ? ? ? * proposition 1.1 ) that @xmath636,{\mathbb{c}})$ ] has real rank @xmath9 , while , by rieffel ( * ? ? ? * proposition 1.7 ) , @xmath634 has stable rank @xmath9 . by rieffel ( * ? ? ? * theorem 3.3 ) , it follows that @xmath635 has also stable rank @xmath9 . by brown and pedersen ( * proposition 1.2 ) , it follows that @xmath635 has real rank @xmath9 as well . as @xmath634 has stable rank @xmath9 , the argument of the second part of the proof of ( * theorem 3.3 ) shows easily that @xmath49 has stable rank @xmath9 . by ( * proposition 1.2 ) , it follows that the real rank of @xmath49 is either @xmath5 or @xmath9 . as @xmath634 is isomorphic to a hereditary subalgebra of @xmath49 , it follows from ( * ? ? ? * corollary 2.8 ) that @xmath49 has real rank @xmath9 . for each idempotent @xmath637 , the trace @xmath638 is a nonnegative integer for each @xmath639 . as the map @xmath640 is continuous on @xmath152 , it follows that if @xmath152 is connected ( _ which will always be the case throughout this section _ ) , then the value of @xmath638 , for @xmath639 , is constant . we shall denote this value by @xmath641 and call it the trace of @xmath210 . observe that the trace of @xmath210 is equal to the rank of @xmath642 , for each @xmath639 . the following result is contained in examples 1.4.2 and 5.1.3(c ) of blackadar @xcite . [ l : vc[01 ] ] let @xmath643 with @xmath644 . then @xmath645,{\mathbb{c}}))\cong{\mathbb{z}}^+$ ] . furthermore , for any positive integer @xmath78 , two idempotent elements of @xmath646,{\mathbb{c}}))$ ] are equivalent iff they have the same trace . [ l : eequivrhonxe ] let @xmath647}$ ] be a projection . then @xmath210 and @xmath648 are unitarily equivalent in @xmath649}$ ] . set @xmath650 . the common value @xmath651 is an integer with @xmath652 . as @xmath210 is continuous at @xmath5 , there is @xmath653 $ ] such that @xmath654 for each @xmath655 $ ] . thus we obtain the inequality ( between elements of @xmath656}$ ] ) @xmath657}-{\overline{e}}{\mathbin{\restriction}}_{[0,{\varepsilon}]}}\vert}\leq 1/2\,.\ ] ] as @xmath658}$ ] and @xmath659}$ ] are both projections of @xmath656}$ ] , they are unitarily equivalent in @xmath656}$ ] , that is , there exists a unitary @xmath660}$ ] such that @xmath661}=u{\overline{e}}{\mathbin{\restriction}}_{[0,{\varepsilon}]}u^\,.\ ] ] extend @xmath662 to a unitary element of @xmath649}$ ] by setting @xmath663 for each @xmath664 $ ] . by replacing @xmath210 by @xmath665 , we may thus assume the following : @xmath666\,.\ ] ] as @xmath667}$ ] and @xmath668}$ ] are projections of @xmath669,{\mathbb{c}}))$ ] with the same rank @xmath670 , it follows from lemma [ l : vc[01 ] ] that they are unitarily equivalent in @xmath669,{\mathbb{c}}))$ ] , so there exists a unitary @xmath671,{\mathbb{c}}))$ ] such that @xmath672}=v{\overline{e}}{\mathbin{\restriction}}_{[{\varepsilon},1]}v^\,.\ ] ] extend @xmath521 to a unitary element of @xmath649}$ ] by setting @xmath673 for each @xmath655 $ ] . by putting and together , we obtain easily that @xmath674 . as in section [ s : collapse ] , we set @xmath578 and @xmath579 . by using the canonical isomorphism @xmath675}$ ] together with lemma [ l : eequivrhonxe ] , we obtain the following description of the nonstable @xmath0-theory of @xmath49 . [ c : v(b1[01 ] ] denote by @xmath676 , @xmath677 the canonical retraction . then @xmath678 is an isomorphism , and @xmath603_c,[c_1]_c\right\}}$ ] is a simplicial basis of @xmath432 . in particular , idempotent matrices @xmath679 are equivalent in @xmath680 iff @xmath681 and @xmath682 are equivalent in @xmath683 . observe that @xmath684 . [ t : luliftc * ] the diagram @xmath557 has a lifting , with respect to the functor @xmath567 , by unital c-subalgebras of @xmath685,{\mathbb{c}}))$ ] , in such a way that the equations are satisfied and @xmath568 and @xmath521 are both involutive automorphisms . the argument is similar to the one of the proof of theorem [ t : lulift ] . the role played by @xmath569 in that proof is taken up here by @xmath686)$ ] . we denote by @xmath572 the automorphism of @xmath634 defined by @xmath687\,.\ ] ] we set @xmath688 , @xmath689 , and we denote by @xmath521 the involutive automorphism of @xmath635 defined by the rule @xmath690 set @xmath691 , @xmath590 , and @xmath591 . then the map @xmath592 , @xmath593 is an involutive c-automorphism of @xmath13 , which switches @xmath692 and @xmath693 . furthermore , @xmath595 with simplicial basis @xmath596_b,[b_1]_b\right\}}$ ] . denote by @xmath426{\hookrightarrow}{\mathbb{c}}$ ] the inclusion map . the matrix @xmath694 is a projection of @xmath49 . as @xmath605 , that is , @xmath695 , it follows from corollary [ c : v(b1[01 ] ] that @xmath696_c=[c_0]_c\quad\text{and}\quad [ 1-c]_c=[1-c_0]_c\,.\ ] ] furthermore , an elementary calculation yields the second key property of @xmath52 , namely @xmath697 therefore , defining @xmath273 as the inclusion map from @xmath49 into @xmath635 and defining c-embeddings @xmath698 and @xmath699 by the rules @xmath611 we obtain the equations @xmath612 and ( using ) @xmath613 . finally , let @xmath700 , @xmath616 . obviously , @xmath617 and @xmath618 . we obtain a diagram @xmath474 as on figure [ fig : liftl ] , with @xmath307 in place of @xmath208 . verifying that @xmath474 lifts @xmath557 , with respect to the abovementioned choices of simplicial bases , is done in a similar fashion as at the end of the proof of theorem [ t : lulift ] . the `` unfolding '' technique used in the proof of proposition [ p : noliftlu ] yields then immediately the following consequence of theorem [ t : luliftc ] . [ p : cdliftedc * ] the diagram @xmath459 can be lifted , with respect to the functor @xmath567 , by a commutative diagram of unital c-algebras and unital c-homomorphisms . the algebra @xmath49 can also be applied to find a lifting of the diagram @xmath48 represented in figure [ fig : oneloop ] . [ p : vecclift ] the diagram @xmath48 can be lifted , with respect to the functor @xmath8 , by a diagram of unital c-algebras and unital c-homomorphisms such that , labeling the lifting as on figure , @xmath46 . we shall define unital c-algebras @xmath1 and @xmath356 with unital c-homomorphisms @xmath701 and @xmath702 such that @xmath46 and , denoting by @xmath474 the diagram represented in the right hand side of figure [ fig : oneloop ] , @xmath703 . set @xmath704 and @xmath705 , with the already introduced @xmath706 and @xmath707 . we set @xmath708\,.\ ] ] observe that @xmath709 is a unitary element of @xmath635 and @xmath710 . furthermore , set @xmath711 , for each @xmath712 . then @xmath713 and , similarly , @xmath714 . it follows that @xmath568 is an automorphism of @xmath49 , and , by lemma [ l : eequivrhonxe ] , @xmath715 and @xmath716 in @xmath49 . hence , @xmath717 switches @xmath718_{\mathbb{d}}$ ] and @xmath719_{\mathbb{d}}$ ] . observe that by corollary [ c : v(b1[01 ] ] , @xmath432 is simplicial with simplicial basis @xmath603_c,[c_1]_c\right\}}$ ] . now let @xmath720 , @xmath721 . then @xmath722 for each @xmath94 . furthermore , from @xmath723 it follows that for each @xmath712 , @xmath724 so @xmath46 , and so @xmath568 and @xmath273 are as desired . we do not know whether it is possible to ensure simultaneously @xmath46 and @xmath725 in a lifting of @xmath48 , with respect to the functor @xmath8 , by unital c-algebras . our first non - lifting result , proposition [ p : nonsplit ] , started with a counterexample of cardinality @xmath16 . we applied the combinatorial core of that result in theorem [ t : nolift ] . now the cycle swings back on itself : we shall get further negative representation results , algebraically stronger than those that we used in order to get proposition [ p : nonsplit ] , now in cardinality @xmath10 ( the present form of theorem [ t : nolift ] would not yield @xmath16 ) . throughout this section we fix a field @xmath53 , and we denote by @xmath577 ( cf . ) and @xmath726 ( cf . ) the @xmath53-algebras introduced in the proof of theorem [ t : lulift ] . we also denote by @xmath727 the closure , under finite direct products and direct limits ( of unital @xmath53-algebras ) , of the set @xmath728 as each member of @xmath729 is a unital exchange @xmath53-algebra with index of nilpotence at most @xmath11 , and as these properties are preserved under finite products and direct limits , every member @xmath1 of @xmath727 is a unital exchange @xmath53-algebra with index of nilpotence at most @xmath11 . as , by proposition [ p : v(exch ) ] , @xmath2 is a ( conical ) refinement monoid , it follows from lemma [ l : basicvcak ] that @xmath2 is the positive cone of a dimension group in which @xmath150 $ ] has index at most two . in particular , by yu ( ? ? ? * theorem 9 ) , @xmath1 has stable rank @xmath9 . now recall that @xmath103 denotes the category of all conical commutative monoids with monoid homomorphisms ( cf . definition [ d : pointedmon ] ) . we shall also define the subcategory @xmath730 with the same objects but where the morphisms are the pre - v - homomorphisms ( cf . definition [ d : vhom ] ) , and the full subcategory @xmath731 of all _ countable _ ( i.e. , at most countable ) conical commutative monoids . denote by @xmath732 the functor @xmath8 ( i.e. , @xmath154 , @xmath733 ) . we introduce in definition 3.8.1 and definition 3.8.2 of gillibert and wehrung @xcite attributes of certain collections of categories and functors , called _ left larderhood _ and _ right larderhood_. we shall use the notation used in these two definitions , yet trying to spell out in detail the verifications to be performed , and we shall start with the ( traditionally easier ) left part . [ l : cakleftlard ] the quadruple @xmath734 is a left larder . we check one after another the items defining left larderhood . * ( clos@xmath735 ) and ( prod@xmath735 ) : by definition , @xmath727 is closed under direct limits and finite products . * ( cont@xmath736 ) : it is well - known that the functor @xmath8 preserves direct limits . * ( proj@xmath737 ) : we must check that @xmath163 is a double arrow ( i.e. , a pre - v - homomorphism ) , for each morphism @xmath124 in @xmath727 which is a direct limit ( in the category of all arrows of @xmath727 ) @xmath738 of projections @xmath739 , @xmath740 , for each @xmath741 ( @xmath229 is a directed partially ordered set ) . as @xmath742 , with @xmath743 the canonical projection @xmath744 , @xmath743 is obviously a pre - v - homomorphism . as the class of pre - v - homomorphisms is easily seen to be closed under direct limits ( within the category of all arrows of @xmath103 ) and the functor @xmath8 preserves direct limits , it follows that @xmath163 is a pre - v - homomorphism . in order to fill the `` right larder '' part , we need , for c-algebras of real rank @xmath5 , @xmath114 , the category of all c-algebras of real rank @xmath5 with c-homomorphisms ( cf . definition [ d : catrings1 ] ) . * @xmath745 , the category of all _ separable _ members of @xmath114 . * @xmath746 denotes the functor @xmath8 ( i.e. , @xmath747 , @xmath733 ) from @xmath114 to @xmath103 . for the side of regular rings , we need * @xmath242 , the category of all regular rings with ring homomorphisms . * @xmath748 , the category of all _ countable _ regular rings . * @xmath749 denotes the functor @xmath8 ( i.e. , @xmath747 , @xmath733 ) from @xmath242 to @xmath103 . back to c-algebras of real rank @xmath5 , we prove [ l : cbrightlard ] the @xmath750-uple @xmath751 is a right @xmath752-larder . the part ( pres@xmath753 ) involves the notion of a _ weakly @xmath752-presented structure _ introduced in gillibert and wehrung ( ? ? ? * definition 1.3.2 ) . it is , basically , easy : every @xmath754 is separable , thus ( cf . the note on page 28 in blackadar @xcite ) @xmath12 is countable , thus ( cf . gillibert and wehrung ( * ? ? ? * proposition 4.2.3 ) ) weakly @xmath752-presented . the part ( ls@xmath755 ) , for a given @xmath756 , is less obvious . we are given a countable conical refinement monoid @xmath62 and a pre - v - homomorphism @xmath757 , together with a sequence @xmath758 of unital c-homomorphisms ( the @xmath759 can be assumed to be monomorphisms , but this will play no role here ) . we are trying to find a separable , unital c-subalgebra @xmath561 of @xmath13 , of real rank @xmath5 , containing @xmath760 , such that , if @xmath761 denotes the inclusion map , then @xmath762 is a surjective pre - v - homomorphism . for each @xmath76 , denote by @xmath763 the continuous function defined by @xmath764 if @xmath765 , @xmath766 if @xmath767 , and @xmath768 is affine on the interval @xmath769 $ ] . we say that a countable subset @xmath152 of @xmath13 is _ ffc - closed _ ( `` ffc '' stands for `` flat function calculus '' ) if @xmath770 for each self - adjoint element @xmath771 and each @xmath76 . furthermore , denote by @xmath772 the set of all countable , ffc - closed , * -closed unital @xmath773$]-subalgebras of @xmath13 . a classical argument about approximating projections on a dense set ( see , for example , blackadar ( * * proposition 4.5.1 ) ) shows that every member @xmath152 of @xmath772 satisfies the following : @xmath774 for all @xmath775 and all @xmath319 with @xmath776_b)=\alpha+\beta$ ] , there are orthogonal idempotents @xmath777 such that @xmath168 while @xmath778_b)=\alpha$ ] and @xmath779_b)=\beta$ ] . as @xmath780 is a pre - v - homomorphism , there are @xmath781 such that @xmath167_b=\alpha'+\beta'$ ] while @xmath782 and @xmath783 . by lemma [ l : vmeasdec ] , there are orthogonal idempotents @xmath777 such that @xmath168 while @xmath144_b=\alpha'$ ] and @xmath169_b=\beta'$ ] . hence @xmath106 and @xmath248 are as required . for each @xmath76 , pick a countable dense subset @xmath784 . as @xmath780 is surjective and @xmath62 is countable , there exists @xmath785 containing @xmath786 such that @xmath787_b)}\mid{e\in{\mathrm{i}}_\infty(y_0)}\right\}}}$ ] . suppose that @xmath788 has been constructed for each @xmath789 . as @xmath13 has real rank @xmath5 and @xmath790 is countable , there exists @xmath791 containing @xmath790 such that every self - adjoint element of @xmath773\times y_{2k}$ ] , viewed as a subset of the unitization of @xmath13 , lies within @xmath792 of some invertible self - adjoint element of the unitization of @xmath793 . as @xmath793 and @xmath62 are both countable and by the claim above , there is @xmath794 containing @xmath793 such that for all @xmath795 and @xmath319 with @xmath776_b)=\alpha+\beta$ ] , there are orthogonal idempotents @xmath796 such that @xmath168 while @xmath778_b)=\alpha$ ] and @xmath779_b)=\beta$ ] . by construction , the union @xmath797 belongs to @xmath772 and the closure @xmath798 has real rank @xmath5 . furthermore , it follows from that every projection of @xmath799 is equivalent to some projection in @xmath800 ; hence , by the construction of @xmath801 from @xmath793 , it follows that @xmath802 is a pre - v - homomorphism . furthermore , from @xmath803 it follows that @xmath804 . therefore , @xmath805 is as required . consider again the set @xmath729 of exchange algebras defined in . as @xmath806 is contained in @xmath731 , we obtain the following consequence of lemmas [ l : cakleftlard ] and [ l : cbrightlard ] together with the ( obvious ) ( * ? ? ? * proposition 3.8.3 ) . [ l : akrr0larder ] the @xmath807-uple @xmath808 is an @xmath752-larder . the analogue of lemma [ l : cbrightlard ] for regular rings is the following . [ l : cbrightlardreg ] the @xmath750-uple @xmath809 is a right @xmath752-larder . the proof of lemma [ l : cbrightlardreg ] is similar , and actually slightly easier ( because there is no topology involved ) than the one of lemma [ l : cbrightlard ] , thus we omit it . lemma [ l : akrr0larder ] is modified in a similar fashion . [ l : akreglarder ] the @xmath807-uple @xmath810 is an @xmath752-larder . now everything is ready for the proof of the following result . [ t : cxakrr0 ] for any field @xmath53 , there exists a unital exchange @xmath53-algebra @xmath54 that satisfies the following properties : 1 . @xmath54 has index of nilpotence @xmath11 . 2 . for every c-algebra of real rank @xmath5 @xmath13 , there is no surjective pre - v - homomorphism from @xmath12 onto @xmath55 . in particular , there is no c-algebra of real rank @xmath5 @xmath13 such that @xmath811 . furthermore , @xmath54 can be constructed as the direct limit of a system of @xmath10 finite products of members of @xmath729 . in particular , if @xmath53 is countable , then @xmath812 . we apply ( * ? ? ? * lemma 3.4.2 ) , called there cll , to the following data : * @xmath813 ; * @xmath814 is the powerset lattice of the three - element set @xmath815 ; * @xmath554 is the commutative diagram of unital exchange @xmath53-algebras , given in proposition [ p : cdliftedexch ] , such that @xmath816 . observe that all objects of @xmath554 belong to @xmath729 ( cf . ) ; * @xmath817 is one of the @xmath752-larders given by either lemma [ l : akrr0larder ] or lemma [ l : akreglarder ] . the structure @xmath818 involved in the statement of cll , which will turn out to be the desired counterexample @xmath54 , is the same for both larders ( it depends only of @xmath727 , @xmath554 , @xmath814 , and @xmath152 ) . in order for the assumptions underlying cll to be fulfilled , we need @xmath814 to admit an @xmath752-lifter ( cf . * definition 3.2.1 ) ) @xmath819 such that @xmath820 . as @xmath814 is a finite lattice , it is a so - called _ almost join - semilattice _ ( cf . * definition 2.1.2 ) ) , thus , according to ( * ? ? ? * corollary 3.5.8 ) , it is sufficient to prove that the relation @xmath821 ( cf . * definition 3.1 ) , also ( * ? ? ? * definition 3.5.1 ) ) holds . according to the definition of the kuratowski index @xmath822 given in ( * definition 4.1 ) , it suffices to prove that @xmath823 . as @xmath814 has exactly three join - irreducible elements , this is a trivial consequence of ( * ? ? ? * proposition 4.2 ) . the statement of cll involves a `` @xmath814-scaled boolean algebra '' , denoted there by @xmath824 ( where @xmath819 is the abovementioned @xmath752-lifter ) , and the `` condensate '' @xmath818 . a @xmath814-scaled boolean algebra is a boolean algebra endowed with a collection , indexed by @xmath814 , of ideals , subjected to certain constraints . for our present purposes , neither the exact definition of a @xmath814-scaled boolean algebra ( cf . * definition 2.2.3 ) ) , nor the exact descriptions of the constructions of @xmath824 ( cf . * lemma 2.6.5 ) ) and @xmath818 ( cf . * section 3.1 ) ) , will matter . we set @xmath825 . the statement that @xmath54 is a direct limit of finite products of members of @xmath729 , together with the cardinality bound on @xmath54 , are immediate consequences of the following facts : * @xmath824 is a @xmath814-scaled boolean algebra with at most @xmath10 elements . * thus @xmath826 , for a directed poset @xmath61 of cardinality @xmath10 and `` compact '' ( i.e. , here , finitely presented ) @xmath814-scaled boolean algebras @xmath463 ( cf . * proposition 2.4.6 ) ) . * @xmath827 . this follows from ( * ? ? ? * proposition 3.1.4 ) . * each @xmath828 is a finite product of members of @xmath554 . this follows from ( * ? ? ? * definition 3.1.1 ) . suppose that there exists a surjective pre - v - homomorphism @xmath829 , for a c-algebra @xmath13 with real rank @xmath5 . as all the assumptions underlying cll are satisfied , there are a @xmath814-indexed diagram @xmath830 of c-algebras of real rank @xmath5 and a natural transformation @xmath831 . as @xmath832 ( cf . proposition [ p : cdliftedexch ] ) , we get a natural transformation , that we shall denote again by @xmath548 , now @xmath833 . the double arrow notation for the natural transformation @xmath548 means here that each component @xmath834 ( for @xmath835 ) of @xmath548 is a surjective pre - v - homomorphism . it follows that the rule @xmath836_{b_p})$ ] , for @xmath837 , defines a @xmath838-valued surjective v - measure on @xmath839 . we thus obtain a diagram @xmath840 of surjective v - premeasures on ( unital ) c-algebras of real rank @xmath5 such that @xmath841 . but this contradicts corollary [ c : norr0lift ] . this completes the proof that @xmath54 is the desired counterexample for c-algebras of real rank @xmath5 . we also need to prove that there exists no surjective pre - v - homomorphism @xmath829 , for any regular ring @xmath13 . the proof is similar to the one for c-algebras of real rank @xmath5 . we need to use lemmas [ l : cbrightlardreg ] and [ l : akreglarder ] instead of lemmas [ l : cbrightlard ] and [ l : akrr0larder ] , and corollary [ c : novlift ] instead of corollary [ c : norr0lift ] . for an infinite cardinal @xmath842 , we say that a c-algebra is _ @xmath842-separable _ if it has a dense subset of cardinality at most @xmath842 . in particular , @xmath843-separable means separable . we claim that what we did in section [ s : al3 ] , using the diagram of exchange rings constructed in section [ s : collapse ] , can be done for c-algebras , using the diagram of c-algebras constructed in section [ s : calglift ] . consider again the c-algebra @xmath49 used in example [ ex : semiprnotv ] and section [ s : calglift ] , and set @xmath844 the relevant analogue of theorem [ t : cxakrr0 ] is the following . [ t : ccxakrr0 ] there exists an @xmath10-separable unital c-algebra @xmath22 that satisfies the following properties . 1 . @xmath22 has index of nilpotence @xmath11 , and real rank and stable rank both equal to @xmath9 . 2 . @xmath845)$ ] is the positive cone of a dimension group with order - unit of index two . 3 . for every c-algebra of real rank @xmath5 @xmath13 , there exists no surjective pre - v - homomorphism from @xmath12 onto @xmath57 . in particular , there is no c-algebra of real rank @xmath5 @xmath13 such that @xmath846 . furthermore , @xmath22 can be constructed as the c-direct limit of a system of @xmath10 finite products of members of @xmath847 . while stating that the nonstable @xmath0-theory of c-algebras is contained neither in the one of regular rings nor in the one of c-algebras of real rank @xmath5 is no big deal ( for @xmath848 may not have refinement , even for a unital c-algebra @xmath208 , cf . blackadar ( ? ? ? * example 5.1.3(e ) ) ) , theorem [ t : ccxakrr0 ] states this fact even for those c-algebras whose nonstable @xmath0-theory is a refinement monoid , which is far more difficult . the proof of theorem [ t : ccxakrr0 ] is very similar to the one of theorem [ t : cxakrr0 ] , now getting the @xmath814-indexed diagram @xmath554 from proposition [ p : cdliftedc ] instead of proposition [ p : cdliftedexch ] . the right larder part requires no change . the left larder part needs to be modified in the obvious way : the category @xmath727 needs to be replaced by the closure @xmath39 of @xmath847 under finite products and c-direct limits . as the functor @xmath8 preserves finite direct products and c-direct limits , @xmath849)$ ] is a refinement monoid with order - unit of index at most two , for each @xmath850 . as in the proof of lemma [ l : cakleftlard ] , the quadruple @xmath851 is a left larder . as every member of @xmath847 has real rank either @xmath5 or @xmath9 while it has stable rank @xmath9 ( cf . proposition [ p : rrsrac ] ) , this is also the case for every direct limit of finite products of members of @xmath847 , in particular for the condensate @xmath852 . the remaining changes that need to be applied to the proof of theorem [ t : cxakrr0 ] are trivial . [ pb : c0vsreg ] does any of the classes @xmath853 and@xmath854 contain the other ? and on positive cones of dimension groups ? [ pb : c0vsregdiagr ] are there finite , lattice - indexed commutative diagrams of simplicial monoids that can be lifted , with respect to the functor @xmath8 , by diagrams in one of the classes @xmath242 and @xmath114 but not the other ? the results and methods of elliott @xcite and goodearl and handelman @xcite imply that the nonstable k - theories of regular rings and of c-algebras of real rank @xmath5 ( and also of af c-algebras ) agree on dimension groups of cardinality at most @xmath752 . our next question is about the remaining cardinality gap @xmath16 . as to the non - representability by regular rings , it looks plausible that a positive solution to problem [ pb : al2 ] would follow from the methods of wehrung @xcite . in that paper , the fact that the principal right ideals in a regular ring form a lattice is crucial . as the analogous result for c-algebras of real rank @xmath5 does not hold , it sounds unlikely that the methods of @xcite are ready to help finding a solution to problem [ pb : al2 ] for those algebras . on the other hand , it is plausible that one may be able to use special functors , similar to those involved in gillibert and wehrung ( ? ? ? * chapter 5 ) ( which is pure lattice theory ! ) , for which a suitable analogue of theorem [ t : nolift ] would remain valid . every conical commutative monoid with order - unit is isomorphic to @xmath855 for some unital hereditary ring @xmath1 : this is proved in theorems 6.2 and 6.4 of bergman @xcite for the finitely generated case , and in bergman and dicks @xcite for the general case . the general , non - unital case is proved in ara and goodearl ( * ? ? ? * proposition 4.4 ) . in light of this result , the following problem is natural . [ pb : exchringslift ] does there exist a functor @xmath3 , from the category of simplicial monoids with normalized monoid _ embeddings _ to the category of exchange rings and ring homomorphisms , such that @xmath4 ? [ pb : calgsimplift ] does there exist a functor @xmath3 , from the category of simplicial monoids with normalized monoid _ embeddings _ to the category of c-algebras and c-homomorphisms , such that @xmath4 ? p. gillibert and f. wehrung , _ from objects to diagrams for ranges of functors _ , lecture notes in mathematics , vol . 2029. springer - verlag , heidelberg - dordrecht - london - new york . isbn : 978 - 3 - 642 - 21773 - 9 r.n . mckenzie , g.f . mcnulty , and w.f . taylor , `` algebras , lattices , varieties . volume i. '' the wadsworth & brooks / cole mathematics series . monterey , california : wadsworth & brooks / cole advanced books & software , 1987 . xii+361 p. isbn : 0 - 534 - 07651 - 3 e. pardo , `` monoides de refinament i anells dintercanvi '' , ph . d. thesis ( in catalan ) , universitat autnoma de barcelona , april 1995 . available online at ` https://sites.google.com/a/gm.uca.es/enrique-pardo-s-home-page/phd-thesis ` . m. rrdam , f. larsen , and n. laustsen , `` an introduction to @xmath53-theory for @xmath860-algebras '' . london mathematical society student texts 49*. cambridge university press , cambridge , 2000 . xii+242 p. isbn : 0 - 521 - 78334 - 8 ; 0 - 521 - 78944 - 3	the assignment ( _ nonstable @xmath0-theory _ ) , that to a ring @xmath1 associates the monoid @xmath2 of murray - von neumann equivalence classes of idempotent infinite matrices with only finitely nonzero entries over @xmath1 , extends naturally to a functor . we prove the following lifting properties of that functor : 1 . there is no functor @xmath3 , from simplicial monoids with order - unit with normalized positive _ homomorphisms _ to exchange rings , such that @xmath4 . 2 . there is no functor @xmath3 , from simplicial monoids with order - unit with normalized positive _ embeddings _ to c-algebras of real rank @xmath5 ( resp . , von neumann regular rings ) , such that @xmath4 . 3 . there is a @xmath6-indexed commutative diagram @xmath7 of simplicial monoids that can be lifted , with respect to the functor @xmath8 , by exchange rings and by c-algebras of real rank @xmath9 , but not by semiprimitive exchange rings , thus neither by regular rings nor by c-algebras of real rank @xmath5 . by using categorical tools ( _ larders _ , _ lifters _ , _ cll _ ) from a recent book from the author with p. gillibert , we deduce that there exists a unital exchange ring of cardinality @xmath10 ( resp . , an @xmath10-separable unital c-algebra of real rank @xmath9 ) @xmath1 , with stable rank @xmath9 and index of nilpotence @xmath11 , such that @xmath2 is the positive cone of a dimension group but it is not isomorphic to @xmath12 for any ring @xmath13 which is either a c*-algebra of real rank @xmath5 or a regular ring .
the study of oscillations in the atmosphere of sunspots dates back to the detection of umbral flashes in the k line ( beckers & tallant 1969 ) , followed by the detection in h@xmath1 of penumbral waves ( zirin & stein 1972 , giovanelli 1972 ) and umbral oscillations ( giovanelli 1972 ) . the nature and mode identification of the 3 min oscillations in the umbral chromosphere have been debated . based on linear theory two cavities have been discussed , i.e. a cavity for fast modes in the low photosphere and sub - photosphere and a cavity for resonant slow modes in the umbral chromosphere as well as a unified model with contributions from both cavities ( e.g. thomas & weiss 1992 ) . however , the 3 minute umbral oscillations in the chromosphere show a non - linear character , suggesting the presence of upward propagating shock waves ( lites 1992 , bard & carlsson 1997 ) . knowledge of the oscillations in the transition region originates from the study of eight sunspots in the c iv @xmath01548 line by gurman et al . ( 1982 ) , see also henze et al . interestingly , the observed waves with periods of 129 - 173 s , show no signs of shocks and appear to be upward propagating acoustic waves . recently , rendtel et al . ( 1998 ) reported transition region intensity oscillations in the 2 min range in sunspot noaa 7986 . we focus on transition region sunspot oscillations in noaa 8156 , observed simultaneously in o v @xmath0629 and n v @xmath01238 , @xmath01242 . the observations were obtained on 1998 february 18 between 16:00 ut and 21:07 ut with the solar ultraviolet measurements of emitted radiation ( sumer ; wilhelm et al . , 1995 ) instrument as part of a joint observing program on the _ solar and heliospheric observatory ( soho)_. the pointing of the sumer slit is kept constant to 525@xmath2 w and 332@xmath2 s off the solar disk center , i.e. at a heliocentric distance of 40@xmath3 . the solar rotation moves the image of the sunspot , with umbral size 15@xmath2 and penumbral size 50@xmath2 , over the narrow slit , 0.3@xmath4 120@xmath2 . from each spectrum 90@xmath4 2.2 spectral windows , centered on o v @xmath0629 and n v @xmath01238 , @xmath01242 , are recorded with detector b. with an exposure time of 15 s it takes 20 min to observe 80 consecutive spectra , called one raster , while the solar rotation moves the image 2.2@xmath2 . figure 1 ( bottom ) shows the positions of the 12 rasters and the location of the sunspot , between 300@xmath2 and 350@xmath2 s off disk center . between each raster , exposures with the rear slit camera give the slit position in relation to the sunspot and 100 s exposures of 90@xmath4 40 spectra give the wavelength scale , from the wavelength positions of chromospheric lines . the data reductions include corrections for the fixed pattern noise , defects of the detector and corrections for the geometrical distortions , see wilhelm et al . a least squares fit of a single gaussian to each observed line profile gives the peak line intensity , @xmath5 , the relative line - of - sight velocity , @xmath6 , and the line width , @xmath7 . we estimate the accuracy in the relative line - of - sight velocity determinations to be 1 km s@xmath8 for o v @xmath0629 and 1.5 km s@xmath8 ( 2 km s@xmath8 ) for n v @xmath01238 ( @xmath01242 ) . [ fig1 ] from the sumer observations alone it is difficult to separate temporal from spatial changes in oscillatory power . simultaneous observations with the coronal diagnostic spectrometer ( cds ; harrison et al . 1995 ) in ten emission lines , including the o v @xmath0629 line , with a cadence of 25 min , show only small temporal variations in the overall intensity and line - of - sight velocity patterns during the time span of the observations . for a description of sunspot velocity field observations with cds , see brynildsen et al . the cds results encourage us to use the sumer observations to investigate the spatial distribution of the power in the transition region oscillations . by deriving the observed characteristics within each raster for 2@xmath2 sections along the slit the power distribution is studied within an area of 33.5@xmath2 @xmath9 90@xmath2 with a spatial resolution of 2.2@xmath2 @xmath10 2@xmath2 . as an example figure 1 shows the observed o v @xmath0629 umbral oscillations in relative line - of - sight velocity , @xmath11 , and relative peak line intensity , @xmath12 . the hatched area in the umbra marks the location of this oscillation . slow variations in the background fields have been removed from the average values , @xmath13 and @xmath14 . this corresponds to applying a low frequency filter in the fourier domain . note that the observed oscillations are linear in character without clear signs of shocks . the amplitudes of the oscillations are listed in table 1 . figure 2 shows the corresponding power spectra for the oscillations in line - of - sight velocity and peak line intensity . the power spectra are nearly identical with one maximum within the frequency interval 5.0 - 6.7 mhz , corresponding to periods between 150 and 200 s , and maximum power close to 170 s. we also find a 5% variation in the line width . the corresponding power spectrum is similar to those shown in figure 2 , but is more noisy . the results are scrutinized by tests where artificial noise is added to the observed signal and new values of the line parameters are derived . the tests show that the results for the intensity and the line - of - sight velocity oscillations are confirmed , but the results are less certain for the line width . figure 1 shows near temporal coincidence of maxima in peak line intensity and maxima in blueshift . from the corresponding power spectra we deduce a phase difference @xmath15 - @xmath16 = 173@xmath3 between the line - of - sight velocity and the peak line intensity for the o v line . the corresponding phase difference for the n v line , @xmath17 - @xmath18 = 172@xmath3 . we give less weight to the observed phase difference of 160@xmath3 in the @xmath01242 line since this line is weaker than the other lines . in figure 5 , below , we present this phase difference for the o v @xmath0629 line for the entire sunspot region . the determination of the phase difference in velocity or intensity between the n v @xmath01238 and o v @xmath0629 lines is difficult since it involves careful alignment along the slit of lines observed in different locations on the detector . for the umbra we find that the wave arrives on average 10 s @xmath19 5 s earlier in n v @xmath01238 line than in o v @xmath0629 line . if the n v line originates at a lower height than the o v line , as seems reasonable considering the temperatures of formation for the n v ( 1.5 @xmath10 10@xmath20k ) and o v ( 2.1 @xmath10 10@xmath20k ) lines , this suggests an upward propagating wave . consider next the observed variation along the slit in raster 7 , i.e. in the 2.2@xmath21 wide raster that includes the central part of the umbra . figure 3 gives the o v @xmath0629 variations in peak line intensity , @xmath12 , and relative line - of - sight velocity , @xmath22 , as a function of time and position along the slit . the visibility and shape of the oscillation wave front is enhanced by subtracting the observed signal at time @xmath23 from the signal observed at @xmath23 - 90 s , a time difference close to half of the dominant period . figure 3 shows that waves with periods close to 170 s are observed mainly within the umbra and do not extend into the penumbra to any great extent . the wave front in the central part of the umbra leads the wave front at the rim of the umbra by nearly a full wave period . clearly the observed oscillation affects the entire umbral transition region , but the generation or transmission of the wave depend on the position within the umbra . to investigate the suggestion by gurman et al . ( 1982 ) that the oscillations are caused by upward propagating acoustic waves , we make use of the property that the relative perturbation in number density , @xmath24 , varies in phase with the line - of - sight velocity , @xmath25 , ( e.g. hansteen & maltby 1992 ) . regarding @xmath25 as positive when directed away from the observer we may write , for an upward propagating acoustic wave with a sinusoidal variation . here @xmath27 is the sound speed , @xmath28 is the aspect angle and the index @xmath29 denotes the amplitude . for n v and o v , formed in a gas with a mean molecular weight of 0.60 at temperatures close to 1.5 @xmath10 10@xmath20 k and 2.1 @xmath10 10@xmath20 k , we estimate the sound speeds to be 58 km s@xmath8 and 69 km s@xmath8 . we will use equation ( 1 ) to calculate line profiles that may be compared with observations . since the lines may be regarded as optically thin , the intensity is determined by an integration along the line - of - sight of the product of the line profile , the electron number density , @xmath30 , the n v ( o v ) ion density , @xmath31 , and a temperature dependent function , @xmath32 ( e.g. mariska 1992 ) . the influence of the wave on the function @xmath32 is small ( e.g. gurman et al . assuming ionization balance , such that @xmath33 , and neglecting possible variations in the thickness of the transition region , we may write the line intensity as @xmath34 . this implies that we may use the observed intensity amplitudes of 9% and 11% , see table 1 , to determine the corresponding density amplitudes and equation ( 1 ) to calculate the line profile for different values of ( @xmath35 ) . figure 4 shows the calculated line profiles for o v @xmath0629 and n v @xmath01238 , with @xmath36 = 0.055 and @xmath37 = 0.045 , and line broadening owing to the instrument and non - thermal velocities of 27 km s@xmath8 and 24 km s@xmath8 , respectively . for ( @xmath35 ) = @xmath38 both the peak line intensity and the line shift towards shorter wavelengths reach their maxima , whereas for ( @xmath35 ) = @xmath39 a maximum in line shift towards longer wavelengths occurs while the peak line intensity reaches a minimum . the observed values at each wavelength are averages of six intensity values from the center of the umbra at the corresponding phases . the averages are determined from the time series in raster 7 , where six minima and maxima are observed . the average is constructed from a 2@xmath2 wide strip centered at position -326@xmath2 , see figure 3 . the agreement between observed and calculated profiles in figure 4 is remarkable and supports the interpretation of the oscillations as caused by upward propagating acoustic waves . to obtain a deeper understanding of this result , the wave propagation through the transition region should be thoroughly discussed , see zugzda , staude , & locans ( 1984 ) for a discussion of this topic . figure 5 gives the spatial distributions of power in line - of - sight velocity and peak line intensity oscillations between 5.0 - 6.7 mhz for the o v @xmath0629 line . the phase difference @xmath40 throughout most of the umbra , i.e. the maximum intensity is nearly in phase with the maximum velocity directed towards the observer . the oscillations are restricted to the umbra and interestingly , the spatial distributions of power in velocity and intensity differ . whereas the power in velocity shows one maximum , the power in intensity also show maxima closer to the umbral rim , not far from the sunspot plumes observed with cds ( figure 5 , right ) , the plumes have peak line intensity @xmath41 . for a recent discussion of sunspot plumes , see brynildsen et al . possibly the direction of wave propagation changes with position in the umbra . in summary , significant oscillations in line - of - sight velocity and peak line intensity are observed , with periods close to 170 s. the observed transition region oscillations are concentrated to the umbra and the entire umbral transition region oscillates . we study the observed line profiles and find support for the suggestion that the waves are upward propagating acoustic waves . discussions of the relation between the shock waves observed in the chromosphere and the present observations are of interest , but outside the scope of this letter . we are indebted to the international sumer team and thankful to the mdi team for permission to use their data . sumer is supported by dlr , cnes , nasa and the esa prodex programme ( swiss contribution ) . soho is a project of international cooperation between esa and nasa . bard , s. & carlsson , m. 1997 in fifth soho workshop : the corona and solar wind near minimum activity , ed . a. wilson , esa sp-404 , 189 beckers , j. m. & tallant , p. e. 1969 , sol . phys . , 7 , 351 brynildsen , n. , maltby , p. , brekke , p. , fredvik , t. , haugan , s. v. h. , kjeldseth - moe , o. , & wikstl , . 1998 , apj , 502 , l85 giovanelli , r. g. 1972 , sol . 27 , 71 gurman , j. b. , leibacher , j. w. , shine , r. a. , woodgate , b. e. , & henze , w. 1982 , apj , 253 , 939 hansteen , v. & maltby , p. 1992 , comments on astrophys . , 16 , 137 harrison , r. a. et al . 1995 , sol . phys . , 162 , 233 henze , w. , tandberg - hanssen , e. , reichmann , e. j. , & athay , r. g. 1984 , sol . 91 , 33 lites , b. w. 1992 in sunspots : theory and observations , ed . j. h. thomas & n. o. weiss ( dordrecht , kluwer ) , 261 mariska , j. t. 1992 , the solar transition region ( cambridge , cambridge univ . press ) rendtel , j. , staude , j. , innes , d. , wilhelm , k. , & gurman , j. b. 1998 in a crossroads for european solar and heliospheric physics , ed . r. a. harris , esa sp-417 , 277 scherrer , p. h. et al . 1995 , sol . phys . , 162 , 129 thomas , j. h. , & weiss , n. o. 1992 in sunspots : theory and observations , ed . j. h. thomas & n. o. weiss ( dordrecht , kluwer ) , 3 wilhelm , k. et al . 1995 , sol . phys . , 162 , 189 zirin , h. & stein , a. 1972 , apj , 178 , l85 zugzda , y. d. , staude , j. , & locans , v. 1984 , sol . phys . , 91 , 219	based on observations obtained with the _ solar and heliospheric observatory - soho _ joint observing program for velocity fields in sunspot regions , we have detected 3 min transition region umbral oscillations in noaa 8156 . simultaneous recordings of o v @xmath0629 and n v @xmath01238 , @xmath01242 with the sumer instrument give the spatial distribution of power in the 3 min oscillations , both in intensity and line - of - sight velocity . comparing loci with the same phase we find that the entire umbral transition region oscillates . the observed maxima in peak line intensity are nearly in phase with the maxima in velocity directed towards the observer . we discuss the suggestion that the waves are upward propagating acoustic waves .
activity - related noise caused by rotation , active features , and irregularities of stellar surfaces ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) provide currently the greatest challenges in improving the precision of radial velocity surveys to such an extent that detections of earth - like planets become possible . despite being largely induced by rotation , the nature of stellar noise related to activity gives rise to aperiodic and/or quasiperiodic variations that can mimic periodic doppler signals of keplerian origin and thus prevent the detections of low - amplitude planetary signals by littering the posterior densities ( or periodograms ) with spurious signals that can be mistaken for genuine signals of planetary origin . in order to increase the sensitivity of radial velocity surveys , it is necessary to develop statistical models ( e.g. for correlated noise : * ? ? ? * ; * ? ? ? * ) accounting for these irregular variations . furthermore , it should only be accepted that a weak planetary signal is detected if it is necessary in describing the radial velocity variations and that these variations can not be explained by stochastic variability that might be connected to measures of stellar activity and/or intrinsic correlations in the data ( e.g. * ? ? ? * ; * ? ? ? corot-7 is a moderately active g9 ( @xmath1 ; [ fe / h ] @xmath2 dex ; @xmath3 m@xmath4 ; @xmath5 ) dwarf star @xcite that has a planetary system orbiting it , including one transiting planet corresponding to a valuable benchmark planet for estimating the compositions and structures of such super - earths @xcite . however , determination of the mass of this planet based on high - precision radial velocities has proved difficult ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) due to the difficulties in detecting the signal in the precense of at least equally significant variations related to stellar activity and rotation @xcite and because modelling approaches based on different assumptions and approaches pre - whitening or high - pass filtering corresponding to the removal of periodicities estimated to correspond to activity - related phenomena @xcite or modelling starspot - induced radial velocity variations @xcite have provided a wide range of estimates that are consistent at best but sometimes very diverse ( see the estimates listed in * ? ? ? in this work we obtain the harps spectra of corot-7 from the european southern observatory archive , process them with the _ template - enhanced radial velocity re - analysis application _ ( harps - terra ; * ? ? ? * ) , and obtain velocity products together with activity indicators . we aim to filter out the activity - related variations from the corot-7 radial velocity data with our noise model . spurious quasiperiodic signals related to the stellar rotation of roughly 23 d @xcite corresponding to the `` dominant frequency in the power spectrum '' according to @xcite , have been affecting the obtained solutions to the radial velocities . our goal is the detection of the signal of the transiting planet without any prior knowledge of its orbital period and , more generally , to be able to distinguish between activity - related variability and genuine doppler signals of transiting planets . stellar activity can contribute considerably to the variations in precision radial velocity data ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? it is thus necessary to account for these variations with the statistical model in order to detect low - amplitude signals of planetary origin . our key point is that activity variations are not strictly periodic and comprise quasi- and/or aperiodic variability that is unlikely to be modelled accurately by using keplerian periodicities or sinusoids . this is the case even when such variations appear as periodicities in the commonly used lomb - scargle periodograms @xcite , although such an interpretation might be tempting in practice ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) in the context of so - called `` pre - whitening '' that aims at removing periodogram powers related to activity and/or rotation by subtracting the corresponding sinusoids from the data . since such variations can not be expected to be strictly sinusoidal , this procedure gives rise to harmonics of the subtracted signals that need to be subtracted as well in order to decrease the contribution of such features to the data . such an approach can not be considered very satisfactory in practice because such a subtraction of sinusoids that are not orthonormal in the case of unevenly sampled time series causes biases to the genuine signals in the data ( e.g. * ? ? ? * ) , but also , because it requires prior knowledge of the nature of the signals seen in the periodograms , e.g. information from transit observations , such that activity - related spurious signals are subtracted but planetary signals are left intact @xcite . we note that more sophisticated periodogram - based methods have been proposed as well , including the generalised log - likelihood periodograms ( e.g. * ? ? ? * ; * ? ? ? * and references therein ) that evaluate the significances of signals based on likelihood - ratio tests and can be applied to arbitrary noise models , and the minimum mean squared error technique for determining the optimal number of signals in a data set @xcite that also provides means for testing whether the signals are stationary in time . instead , we adopt the approach of @xcite and assume that a large fraction of activity - induced variability can in fact be filtered out from the velocities by accounting for intrinsic correlations in the data . in practice , this means that the radial velocity noise is not white but has a significant red - noise component even for stars that are known to be rather inactive @xcite . however , such models for correlated noise might not be sufficient in explaining the variations in radial velocities of moderately active stars such as corot-7 . well - known measures of stellar activity , obtained from the same spectra as the velocities themselves , can be useful ( e.g. * ? ? ? * ) to estimate how much the stellar activity contributes to the radial velocity variability . in particular , we consider using the line bisector span ( bis ) that respond to spots and other surface inhomogeneities co - rotating on the stellar surface @xcite , line full - width at the half - maximum ( fwhm ) , and the @xmath6-index based on the caii h&k lines as reasonable measures of activity ( e.g. * ? ? ? the above considerations suggest , as a first - order approximation , a model for the radial velocities that can be written ( see also * ? ? ? * ) as @xmath7 where the measurement @xmath8 made at time @xmath9 is modelled by using the reference velocity @xmath10 ; a linear trend quantified by using the parameter @xmath11 that is can be present in the radial velocities of any given target due to the possible existence of a long - period stellar and/or substellar companion ; a white - noise component @xmath12 that we assume to have a gaussian density with a zero mean and a variance of @xmath13 , where @xmath14 represents the estimated instrument noise and @xmath15 is a free parameter of the model ; a correlated noise component that we describe according to a @xmath16th order moving average ( ma ) model with an exponential smoothing in the time - scale of @xmath17 @xcite ; and linear correlations with the activity indices @xmath18 quantified by using the parameters @xmath19 . although it is possible that the linear correlations between velocities and activity indices described above are not sufficient as the relationship between those two can not be expected to be linear ( or in fact to follow any simple functional description ) , we adopt such a description as a first - order approximation . while we do not expect this model to be optimal , if velocity variations are indeed impacted by activity this will be better than modelling the velocity variations without such a component . moreover , the ma(@xmath16 ) term might not be an optimal description of the correlated noise either . for instance , @xcite and @xcite describe the noise such that they set all @xmath20 equal and take the summation over all @xmath21 , @xmath22 . we do not consider this a realistical description . first , such a model corresponds to a violation of causality as the measurements made after the @xmath23th one contribute to its noise , although this might be justifiable as simply a statistical description without such an interpretation . second , the assumption that the parameter @xmath20 is the same for all @xmath21 might be too restrictive . furthermore , using such models requires inversions of @xmath24 matrices , where @xmath25 is the number of measurements , which becomes computationally heavy even for moderately high @xmath25 . we choose an ma(1 ) model that does not increase the computational requirements and appears to be a reasonably good noise model in practice @xcite and because increasing the number of ma components did not improve the model . moreover , as we measure activity - related information based on three activity indicators , namely bis , fwhm , and @xmath6-indices , we set @xmath26 and @xmath27 in eq . ( [ eq : model ] ) . the model described in eq . [ eq : model ] implies a likelihood model for the measurements that can be calculated for the @xmath23th measurement based on the obtained values @xmath28 , where @xmath29 . we used the priors as discussed in @xcite such that the eccentricity prior was defined as @xmath30 and the jitter prior as @xmath31 , where @xmath32 2 ms@xmath33 . see @xcite and @xcite for discussion and justification . we analyse the radial velocity data sets by applying the delayed - rejection adaptive - metropolis ( dram ) algorithm @xcite that is a generalisation of the adaptive - metropolis algorithm @xcite based on the metropolis - hastings algorithm @xcite . the basics of the dram algorithm are as follows . we first draw a proposal ( @xmath34 ) from a proposal density @xmath35 . this density is a function of the current state of the chain @xmath36 ( typically centered at it ) . the proposal is then accepted with the probability @xmath37 where @xmath38 is the posterior density given the data @xmath39 . if the proposal density is a symmetric one , as is the case when applying a multivariate gaussian density , the acceptance probability depends only on the posterior ratio of the two vectors because @xmath40 . assuming that the new vector @xmath34 is rejected , another vector @xmath41 can then be proposed by using another proposal density @xmath42 @xcite . in addition to the current state @xmath36 and the newly proposed vector @xmath41 , this new proposal density can depend on the rejected vector @xmath34 as well . the new proposal is then accepted with probability @xmath43}{[1 - \alpha_{1}(\theta_{0 } , \theta_{1 } ) ] } \bigg\ } .\end{aligned}\ ] ] this process can be repeated for an arbitrary number of rejected proposals in the above manner . the equation for the general case , i.e. @xmath44th acceptance probability , can be found in @xcite . analyses of radial velocity time series in a search for periodic signals are typically complicated because the period parameter ( its logarithm in our analyses ) has a highly multimodal likelihood function and thus posterior density . therefore , it is necessary to perform the dram samplings in such a way that the chains visit all the modes in the posterior regularly and that the highest maxima are identified robustly in the samplings . for this reason , we choose the proposal densities such that @xmath45 is a multivariate gaussian density with a covariance matrix @xmath46 . the second proposal is then a multivariate gaussian as well with a covariance of @xmath47 that is obtained by multiplying the row and the column corresponding to the period parameter of the signal by a factor of @xmath48 such that the width of the proposal density is decreased in the dimension of the period to enable the chains to find the narrow probability maxima in the period space . we allowed three delayed - rejection steps before finally rejecting a proposed vector . such a practice has been proposed by @xcite albeit in a different context . when searching for periodic signals in the data , we used tempered samplings such that a posterior raised to the power of @xmath49 was used instead of the original posterior . this ensured , by decreasing the relative heights of the modes in the posterior , that the corresponding markov chains visited all the relevant areas in the period space . this is important because the posterior density is typically highly multimodal in such examples of analyses of radial velocity data ( e.g. * ? ? ? * ; * ? ? ? dram samplings were not used to obtain point and uncertainty estimates for the model parameters but only for ensuring that there were no additional significant maxima in the period space . for parameter estimation purposes , we set @xmath50 to draw samples from the original posterior density with the adaptive - metropolis algorithm . the corot-7 doppler data have been obtained over two rather distinct observing runs between 2454527 - 2454885 jd and between 2455939 - 2455965 jd . we call these simply the first ( @xmath51 after removing few measurements corresponding to fwhm outliers ) and second observing run ( @xmath52 ) , respectively . these velocities are tabulated in the appendix . because there is a gap of almost three years between these two observing runs , we believe that it can not be expected that the noise properties , including all possible correlations as modelled according to eq . ( [ eq : model ] ) , are equivalent in these runs due to differences in stellar activity , e.g. the evolution of spot patterns , and the designs of the observations in terms of e.g. cadence and exposure time . therefore , we assume that the noise parameters of these two observing runs are independent of one another . even if they are not strictly independent , we believe that treating them as free parameters for each run , as one would for data from different instruments , enables us to determine whether there are indeed differences that should be accounted for and to obtain a better statistical model than when making the much stronger assumption that the noise properties were not evolving as a function of time , which is unlikely to be the case ( e.g. * ? ? ? due to this choice , our model contains eight independent and two common parameters for the two observing runs , which yields a baseline model with 18 parameters when there are no keplerian signals in the model . we started the analyses of the harps velocity data by performing tempered samplings of a one - keplerian model . these samplings enabled us to identify a signal at 3.7 d as a unique probability maximum in the period space together with local maxima corresponding to its double and a daily alias at a period of 1.4 d ( fig . [ fig : dram_searches ] , right hand panel ) . similar samplings , albeit with parameter @xmath53 set to values closer to unity , of a two - keplerian model yielded a maximum at a period of 9.0 d for the second signal ( fig . [ fig : dram_searches ] middle panel ) together with local maxima at 0.86 d and 22 d. after this , additional samplings revealed a unique third signal in the data at a period of 0.86 d ( fig . [ fig : dram_searches ] , right hand panel ) without considerable local maxima . this sequence of signal detections was consistent with the signal detection criteria of @xcite as the log - bayesian evidences were increased at these steps by 43.0 , 13.2 , and 19.0 . these log - bayesian evidences correspond to model probabilities that increase by factors of @xmath54 , @xmath55 , and @xmath56 , respectively , when assuming equal prior probabilities for each model . the phase - folded signals are shown in fig . [ fig : signals ] . we did not detect any additional signals in the data . the observed rotation period of the star of 23 d that has been detected in the radial velocities in earlier studies @xcite did not correspond to the global maximum for any of the models used in our analyses . although it can be seen as a local maximum in the dram search for the second most significant signal ( fig . [ fig : dram_searches ] , middle panel ) , it did not correspond to a significant solution even when we started additional samplings in its vicinity . furthermore , we did not observe significant signals at that period even after accounting for the three candidate planets orbiting the star . this indicates that the statistical model ( eq . [ eq : model ] ) accounts for the aperiodic and quasiperiodic variations corresponding to the stellar activity sufficiently well and makes it unnecessary to model the rotation period ( and/or its harmonics ) as additional signals in the data @xcite . we interpret this result as a suggestion that the noise model is an adequate description of the activity - related variations and , more generally , that it might be possible to remove such variability from radial velocity data of at most moderately active stars . this could therefore help in the detection of low amplitude radial velocity variations corresponding to low - mass planets around such stars . we did not find significant correlations between the radial velocities and bis values or the @xmath6-indices . however , the fwhm appears to be connected to the velocity variations as we obtain @xmath57 0.094 [ 0.016 , 0.198 ] for the first observing run and @xmath57 0.065 [ -0.013 , 0.192 ] for the second when using the maximum a posteriori ( map ) estimates and the 99% credibility intervals . the former correlation is significantly present in the data considering that the 99% interval does not overlap with zero but the latter is significant in this sense with respect to only a 95% credibility interval . the excess noise , as measured by obtaining estimates for the parameters @xmath58 , is likely different between the two observing runs . we obtained estimates of 3.89 [ 3.07 , 4.97 ] ms@xmath33 and 1.88 [ 1.42 , 3.21 ] ms@xmath33 for the first and second observing runs , respectively . similarly , the first observing run contained significant intrinsic correlations because we obtained @xmath59 0.88 [ 0.68 , 1 ] , whereas the same parameter was consistent with zero for the second observing run . the time - scale ( @xmath17 ) of these intrinsic correlations was found to be roughly 6 d ( see also * ? ? ? * ; * ? ? ? not accounting for these differences might weight the first observing run too much and the second one too little and lead to biased results .	variations related to stellar activity and correlated noise can prevent the detections of low - amplitude signals in radial velocity data if not accounted for . this can be seen as the greatest obstacle in detecting earth - like planets orbiting nearby stars with doppler spectroscopy regardless of developments in instrumentation and rapidly accumulating amounts of data . we use a statistical model that is not sensitive to aperiodic and/or quasiperiodic variability of stellar origin . we demonstrate the performance of our model by re - analysing the radial velocities of the moderately active star corot-7 ( @xmath0 ) with a transiting planet with a transiting planet whose doppler signal has proven rather difficult to detect . we find that the signal of the transiting planet can be robustly detected together with signals of two other planet candidates . our results suggest that rotation periods of moderately active stars can be filtered out of the radial velocity noise , which enables the detections of low - mass planets orbiting such stars . [ firstpage ] methods : statistical , numerical techniques : radial velocities stars : individual : corot-7
a kind of rare - earth compounds called kondo insulator or semiconductor undergoes a crossover from a metallic state into an insulating ground one with decreasing temperature , where a narrow gap or a pseudogap is formed at the fermi level ( @xmath2 ) in a coherent heavy fermion band.@xcite these energy gap openings are generally considered to be derived from c - f hybridization between a wide conduction band and a 4f band that is renormalized as a heavy fermion band at @xmath2 at low temperature . the smallest gap kondo insulators , cerhsb and cenisn with orthorhombic structure , are exotic among them : they appear to develop an anisotropic energy gap , while other systems such as ce@xmath3bi@xmath4pt@xmath3 and ybb@xmath5 with cubic structure have a well - defined size of energy gap . the anisotropic energy gap has been suggested by nmr in cerhsb and cenisn for the first time , where the power law of nuclear spin - lattice relaxation rate @xmath6 in its temperature dependence is explained by a narrow v - shaped pseudogap at @xmath2 in the density of states ( dos).@xcite these results , together with transport,@xcite specific heat,@xcite and tunneling spectroscopy measurements,@xcite indicate the formation of a new type of semimetal with an anisotropic gap for cerhsb and cenisn . such an anisotropic gap is proposed to originate from an anisotropic hybridization of conduction band with a certain ground state of 4f electron in the crystalline field @xcite and , therefore , it may be closely related to its crystal lattice . on the other hand , it has been reported that cerhas shows a gap opening over the entire fermi surface by various measurements,@xcite although cerhas is isostructural to cerhsb and cenisn at room temperature . recent investigation in cerhas indicate that successive structural modulations at low temperature are closely related to its gap property.@xcite thus , the detailed mechanism of the anisotropic gap formation has received new attention in connection with a crystal lattice . recently , another isostructural compound ceirsb with orthorhombic @xmath7-tinisi type structure was synthesized.@xcite the lattice parameters for ceirsb are @xmath8 7.351 , @xmath9 4.5744 , and @xmath10 7.935 . on going from cerhsb to ceirsb , the @xmath11 and @xmath12 parameters decrease by 0.9% but the @xmath13 parameter increases by a similar amount , thus the unit cell volume is decreased by 0.8% . this anisotropic shrink in the lattice may strengthen c - f hybridization . actually , the large negative curie - weiss temperature of @xmath14 k and the gradual decrease in resistivity with decreasing temperature suggest that ceirsb is a valence fluctuating compound with stronger c - f hybridization than that for cerhsb.@xcite the value of @xmath15 below 10 k is less than that of cerhsb , which proves that the dos at @xmath2 is strongly reduced . it is notable that the resistivity of ceirsb increases below 8 k , which is reminiscent of the similar behavior of resistivity in cerhsb . although this upturn in the resistivity may be associated with an energy gap opening , thermopower does not show an enhanced value at low temperature , which disagrees with a distinct characteristic of narrow gap semiconductors.@xcite thus , it has been still controversial whether the ground state of ceirsb is a gapped one or not . in order to make this point clear from a microscopic viewpoint , we have performed sb - nmr / nqr measurements on ceirsb and its lanthanum analog lairsb . detailed preparation method of the polycrystalline samples was reported in the literature.@xcite x - ray powder diffraction ( xrd ) and electron - probe microanalysis ( epma ) detected impurity phases of cesb and ceir@xmath16 with several percent.@xcite the amounts of impurity phases of cesb and ceir@xmath16 were estimated to be 5% and 3% , respectively , from the combined analysis of xrd , epma and specific - heat anomaly due to the antiferromagnetic order of cesb at 17 k.@xcite nmr / nqr method is , however , a very local probe and therefore yields information on the majority phase , ceirsb . in this meaning , physical quantities obtained by nmr / nqr method have its inherent advantage over macroscopic ones that are affected by the magnetic impurity cesb , for example.@xcite we grounded these polycrystals into grains with smaller size than 100 @xmath17 m for nmr / nqr measurement . the measurements were made at temperatures between 1.6 k and 300 k by employing a phase - coherent pulsed nmr / nqr spectrometer . nmr ( nqr ) spectrum was measured by tracing the integrated spin - echo signal as the function of an external magnetic field ( the frequency of pulsed rf field in zero external magnetic field ) . we obtained nuclear spin - lattice relaxation rate , @xmath6 by fitting the longitudinal nuclear magnetization recovery after saturating pulse to theoretical functions given in refs . [ 24 ] and [ 25 ] . the high quality of the fittings with single component of @xmath6 guarantees that obtained @xmath6 s will make clear intrinsic nature of ceirsb , unaffected by any interruption from the impurity phases.@xcite sb - nqr spectra of ( a ) ceirsb and ( b ) lairsb at 4.2 k. ] figure 1 displays the @xmath18sb - nqr spectra of ( a ) ceirsb and ( b ) lairsb for two sb isotopes at 4.2 k.@xmath19sb ( @xmath20sb ) has a nuclear spin @xmath21 ( 7/2 ) and therefore exhibits two ( three ) nqr transitions . in ceirsb , the resonance lines for @xmath19sb ( @xmath20sb ) are found at 13.68 and 26.95 mhz ( 8.50 , 16.30 and 24.56 mhz ) , corresponding to @xmath22 mhz ( @xmath23 mhz ) with @xmath24 = 0.107 . here , the nuclear quadrupole frequency @xmath25 and the asymmetry parameter @xmath24 are defined as @xmath26 and @xmath27 , respectively , with the nuclear quadrupolar moment @xmath28 and the electric field gradient at the position of the nucleus @xmath29 ( @xmath30 ) . since no change in the spectral shape across 17 k , at which cesb exhibits the magnetic order,@xcite has been observed , we consider these signals originate from ceirsb . any other signal from impurity phase is not observed in the nqr spectra , presumably because these signal intensities are below the sensitivity of our spectrometer . the line width for the transition @xmath31 of @xmath20sb in ceirsb is 89 khz , which is about four times larger than 23 khz for the corresponding transition observed in cerhsb,@xcite but smaller than 200 - 300 khz observed for cerhas.@xcite as for lairsb , the resonance lines for @xmath19sb ( @xmath20sb ) are found at 12.29 and 24.43 mhz ( 7.59 , 14.79 and 22.27 mhz ) , corresponding to @xmath32 mhz ( @xmath33 mhz ) with @xmath24 = 0.078 . figure 2 displays the temperature dependence of @xmath0 for ceirsb ( closed circles ) with that for cerhsb ( crosses [ 4 ] ) in zero field . here , these @xmath6 s were measured at the resonance line arising from the @xmath34 transitions of @xmath19sb for both compounds . the value of @xmath35 2.9 at 4.2 k for ceirsb , is comparable to @xmath36 = 3.4 expected when magnetic interaction is dominantly responsible for the nuclear spin relaxation . here , @xmath37 and @xmath38 are the relaxation rate and the nuclear gyromagnetic ratio of sb isotopes with its mass number @xmath39 . this fact proves that the leading contribution to the nuclear spin relaxation arises from magnetic interactions . for ceirsb ( closed circles ) , lairsb ( closed squares ) and cerhsb ( crosses [ 4 ] ) in zero field . the open squares and circles represent @xmath0 for ceirsb in magnetic fields of 1 and 105.5 koe , respectively . the solid line indicates the calculation by assuming an effective dos with v - shaped gap structure at @xmath2 as drawn in the bottom left . ] the overall temperature dependence of @xmath0 for ceirsb is qualitatively similar to that for cerhsb . @xmath0 for ceirsb has a maximum around @xmath40 = 300 k and decreases significantly as @xmath1 below this temperature , followed by a weak upturn below 4 k. this decrease in @xmath0 with decreasing temperature indicates that ceirsb has an energy gap at @xmath2 in the dos at low temperature . the similarity in @xmath6 is demonstrated in fig . 3 , where the normalized relaxation rate @xmath41 is plotted against the normalized temperature @xmath42 for ceirsb and cerhsb . here , @xmath40 and @xmath43 are the temperature at which @xmath0 has a maximum and @xmath6 at @xmath40 for each compound , respectively . the temperature dependences of @xmath6 for both compounds are scaled with @xmath40 except those in low temperature region . this deviation from the scaling at low temperature is explained by the difference in the amount of residual dos at @xmath2 as discussed later . it should be noted that this scaling between @xmath6 and @xmath40 is good although the difference in @xmath40 s for ceirsb and cerhsb is as large as 10 times . this result indicates that these two compounds have a similar gap structure and that the magnitudes of the energy gap are scaled with @xmath40 . the weak upturn in @xmath0 below 4 k for ceirsb is probably due to the relaxation process by spin fluctuations of paramagnetic ce spins through direct mutual spin flipping between electron and nuclear spins , which has been also observed in cerhsb.@xcite in order to suppress this extrinsic relaxation channel , we have measured @xmath0 in a small magnetic field of @xmath44 = 1 koe , because the flipping may be prevented by a magnetic field which leads to a different zeeman splitting between electron and nuclear spin levels . as a matter of fact , the upturn is suppressed and @xmath0 stays constant down to 1.6 k in 1 koe as indicated by open squares in fig . 2 . similar suppression has been also observed in cerhsb @xcite by applying @xmath45 koe as shown in fig . 2 . in small magnetic fields , where the extrinsic relaxation contribution is suppressed , @xmath0 for ceirsb is independent of temperature below 20 k. this result indicates the existence of residual dos at @xmath2 even though a v - shaped gap is made . normalized by the value at @xmath40 as a function of the normalized temperature @xmath42 for ceirsb ( closed circles ) and cerhsb ( crosses [ 4 ] ) in zero field . the open squares and circles represent the data for ceirsb at @xmath46 1 and 105.5 koe , respectively . the solid curve gives the reproduction by the best fitted v - shaped gap . ] since the temperature dependences of @xmath0 for ceirsb and cerhsb are quite similar , the v - shaped gap model used in cerhsb @xcite is applicable to the gapped state in ceirsb . namely , we assume an effective dos , @xmath47 having a v - shaped gap structure with a residual dos at @xmath2 as shown in the bottom left of fig . 2 . the temperature dependence of @xmath0 in ceirsb could be well fitted to the relation @xmath48 where @xmath49 is the fermi distribution function , with the band width @xmath50 = 1800 k , the gap @xmath51 = 350 k and the fraction of residual dos against a value without gap at @xmath2 for the lorentzian band @xmath52 = 0.193 . the result of calculation , indicated by solid lines in figs . 2 and 3 , is in good agreement with the experimental data when the above - mentioned extrinsic relaxation channel is suppressed in a small magnetic field . these parameters for ceirsb as well as those for cerhsb @xcite and cenisn @xcite are listed in table i. the values of @xmath50 and @xmath51 for ceirsb are much larger than those for cerhsb and cenisn , respectively . the larger value of @xmath50 for ceirsb is consistent with the weaker temperature dependence of the magnetic susceptibility , suggesting a stronger c - f hybridization.@xcite despite the values of @xmath50 and @xmath51 for ceirsb by one order of magnitude larger than those for others , it is notable that @xmath51 is scaled with @xmath50 among these compounds , which is expected when the c - f hybridization is essential for the v - shaped gap formation . in this meaning , it is considered that these compounds are classified into the same group revealing a v - shaped gap due to c - f hybridization and that the much larger band width for ceirsb brings about the much larger magnitude of the energy gap . the larger value of @xmath52 for ceirsb than those for cerhsb and cenisn may originate from a possible off - stoichiometry in composition or effective carrier doping effect,@xcite which may be associated with the larger nqr line width of 89 khz for ceirsb than that of 23 khz for cerhsb . lairsb , a metallic and non - magnetic analog of ceirsb , may offer information of conduction electron states without 4f electrons . as indicated by closed squares in fig . 2 , @xmath0 of lairsb is almost independent of temperature , characteristics of a normal metallic state , where a dos near @xmath2 is finite and almost flat with a temperature - independent band structure . in such a metallic state , we may consider @xmath53 as approximately obtained from eq . ( [ eq:1 ] ) . the value of @xmath0 below 10 k in ceirsb , 0.134 s@xmath54k@xmath54 , is by one order of magnitude smaller than that of lairsb , 2.32 s@xmath54k@xmath54 , although a lanthanum has no 4f electrons . therefore , if we assume a comparable hyperfine coupling constant for ceirsb and lairsb , the smaller value of @xmath0 in ceirsb is considered to be related to the reduction in @xmath55 of ceirsb at low temperature due to the v - shaped energy gap . the ratio of the residual dos for ceirsb against that for lairsb is estimated to be @xmath56 , about 25% larger than @xmath57 obtained in the previous paragraph . a possible contribution of 4f electrons to the conduction band in ceirsb due to c - f hybridization may be responsible for this difference . .list of quasi - particle band width @xmath50 , energy gap @xmath51 and the fraction of residual dos against a value without gap at @xmath2 for the lorentzian band @xmath52 for ceirsb , cerhsb @xcite and cenisn.@xcite [ cols="<,<,<,<",options="header " , ] to gain further insight into the gap property of ceirsb , we have performed @xmath19sb - nmr measurements in high magnetic field . the field - swept @xmath19sb - nmr spectrum measured at fixed frequency of 107.3 mhz at 10 k is shown in the inset of fig . 4 . it indicates a distribution of resonance field at sb site characterized by six peaks at 9.34 , 9.84 , 10.39 , 10.58 , 11.00 and 11.69 t in the polycrystalline sample , so - called powder pattern for nuclei with @xmath21 . here , a second order perturbation of nuclear quadrupole interaction added to zeeman interaction yields a splitting of the central line into two peaks . neither broadening nor splitting in the resonance lines has been observed down to 5 k ( not shown ) , indicating that ceirsb is non magnetic at the ground state and that these resonance lines come from ceirsb . small peaks at 9.69 t and 11.10 t may come from cesb . sb knight shift for ceirsb . inset shows the field - swept @xmath19sb - nmr spectrum at 10 k measured with the fixed rf frequency of 107.3 mhz . ] we show the temperature dependences of @xmath0 and @xmath41 measured at the central resonance lines by open circles in figs . 2 and 3 , respectively . apparently , these relaxation rates in high magnetic field are almost identical to those in zero and low field , indicating that the gap structure is not affected by the applied magnetic field at least up to 10 t. it is a remarkable contrast to the case for cenisn , in which the pseudogap is known to be progressively suppressed with increasing the applied magnetic field higher than 2 t , while no dramatic change was found below 2 t.@xcite these facts have been interpreted as a result of the presence of a flat part at the bottom of the pseudogap and the respective zeeman shift of valence up - spin and conduction down - spin bands which induces the overlapping of these bands at high magnetic field.@xcite the robust gap unaffected by the external magnetic field in ceirsb is consistent with the much larger sizes of energy gap and residual dos than those for cenisn . in the main panel of fig . 4 , we show the temperature dependence of the knight shift , @xmath58 , which is obtained from the two peaks of the @xmath59 transition with taking non - zero value of @xmath24 = 0.107 into account . it is notable that @xmath58 reflects the intrinsic magnetic property of ceirsb , while @xmath60 is largely affected by the impurity phase cesb giving rise to the curie - weiss - like upturn below 250 k.@xcite in general , @xmath58 is dominated by two contributions , a temperature dependent spin part @xmath61 and a temperature independent van vleck one @xmath62 , as @xmath63 . the temperature dependent portion of @xmath58 arises from @xmath61 part , which is proportional to a corresponding spin susceptibility @xmath64 via a dominant transferred hyperfine coupling . therefore , the very weak temperature dependence in @xmath58 above 200 k suggests that this compound is in the regime of intermediate valence systems . the decrease in @xmath58 below 200 k is considered to be associated with the decrease in the dos at @xmath2 in ceirsb . at the lowest temperature , there remains large @xmath58 of 0.7 % , which may be attributed to the large residual dos and the van vleck contribution . by subtracting @xmath65 arising from the residual dos , @xmath66 , at low temperature , the van vleck contribution is estimated to be 0.65 % . such a large van vleck contribution has been also observed in other kondo insulators and semiconductors @xcite and explained in a strongly correlated regime , where the van vleck susceptibility is enhanced by the same renormalization factor as for the effective mass.@xcite although the 4f electronic state of ceirsb is not expected to be highly renormalized with the conduction band as in a heavy fermion material , some renormalization effects might enhance @xmath62 and yield the large value of @xmath58 at low temperature . the valence fluctuating compound ceirsb and its la analog lairsb have been investigated by sb - nmr / nqr measurements . the @xmath0 for ceirsb has a maximum around 300 k and decreases significantly as @xmath1 , followed by a @xmath67 = const . relation at low temperature . this temperature dependence of @xmath0 indicates a v - shaped energy gap with a residual dos at @xmath2 . the size of energy gap for ceirsb is estimated to be about 350 k , which is by one order of magnitude larger than the respective 28 k and 14 k for cerhsb and cenisn . despite the very large difference in the size of energy gap , the temperature dependence of @xmath6 has revealed the scaling between the band width and the magnitude of energy gap among these compounds . this fact indicates that these compounds are classified into the same group exhibiting a v - shaped gap due to c - f hybridization . the temperature dependence of the knight shift for ceirsb measured in a high magnetic field is consistent with the formation of the pseudogap at @xmath2 . t. takabatake , f. iga , t. yoshino , y. echizen , k. katoh , k. kobayashi , m. higa , n. shimizu , y. bando , g. nakamoto , h. fujii , k. izawa , t. suzuki , t. fujita , m. sera , m. hiroi , k. maezawa , s. mock , h.v . lhneysen , a.brckl , k. neumaier and k. andres , j. magn . , * 177 - 181 * , 277 ( 1998 ) . z. fisk , j.l . sarrao , j.d . thompson , d. mandrus , m.f . hundley , a. miglori , b. bucher , z. schlesinger , g. aeppli , e. bucher , j.f . ditusa , c.s . oglesby , h - r . ott , p.c . canfield and s.e . brown , physica b * 206 - 207 * , 798 ( 1995 ) . m. kyogaku , y. kitaoka , h. nakamura , k. asayama , t. takabatake , f. teshima and h. fujii , j. phys . * 59 * , 1728 ( 1990 ) . k. nakamura , y. kitaoka , k. asayama , t. takabatake , h. tanaka and h. fujii , j. phys . jpn . * 63 * , 433 ( 1994 ) . t. takabatake , h. tanaka , y. bando , h. fujii , s. nishigori , t. suzuki , t. fujita and g. kido , phys . b , * 50 * , 623 ( 1994 ) . s. nishigori , h. goshima , t. suzuki , t. fujita , g. nakamoto , h. tanaka , t. takabatake and h. fujii , j. phys . jpn . , * 65 * , 2614 ( 1996 ) . t. ekino , t. takabatake , h. tanaka and h. fujii , phys . lett . , * 75 * , 4262 ( 1995 ) . h. ikeda and k. miyake : j. phys . * 65 * , 1769 ( 1996 ) . j. moreno and p. coleman , phys . * 84 * , 342 ( 2000 ) . s. yoshii , m. kasaya , h. takahashi and n. mori , physica b , * 223 - 224 * , 421 ( 1996 ) . k. shimada , k. kobayashi , t. narimura , p. baltzer , h. namatame , m. taniguchi , t. suemitsu , t. sasakawa and t. takabatake , phys . rev . b , * 66 * , 155202 ( 2002 ) . m. matsumura , t. sasakawa , t. takabatake , s. tsuji , h. tou and m. sera , j. phys . , * 72 * , 1030 ( 2003 ) . d.t . adroja , j .- park , k.a . mcewen , k. shigetoh , t. sasakawa , t. takabatake and j .- y . so , physica b , * 378 - 380 * , 788 ( 2006 ) . t. sasakawa , t. suemitsu , t. takabatake , y. bando , k. umeo , m.h . jung , m. sera , t. suzuki , t. fujita , m. nakazima , k. iwasa , m. kohgi , ch . paul , st . berger and e. bauer , phys . b , * 66 * , 041103(r ) ( 2002 ) . t. sasakawa , k. mine , k. shigetoh and t. takabatake , j. phys . jpn . , * 74 * , 3329 ( 2005 ) . m. matsunami , h. okamura , t. nanba , t. suemitsu , t. yoshino , t. takabatake , y. ishikawa and h. harima , j. phys . * 71 * , suppl . 291 ( 2002 ) . n. ogita , m. udagawa , t. sasakawa , t. suemitsu and t. takabatake , physica b , * 328 * , 151 ( 2003 ) . t. sasakawa , h. miyaoka , k. umeo , s. aoyagi , k. kato , f. iga and t. takabatake , j. phys . jpn . * 73 * , 262 ( 2004 ) . hasse , t. schmidt , c.g . richter , h. block and w. jeitschko , j. solid state chem . , * 168 * , 18 ( 2002 ) . t. sasakawa , k. shigetoh , d. hirata , k. umeo and t. takabatake , physica b , * 359 - 361 * , 111 ( 2005 ) . m.f . hundley , p.c . canfield , j.d . thompson and z. fisk , phys . b , * 50 * , 18142 ( 1994 ) . f. iga , t. suemitsu , s. hiura , k. takagaki , k. umeo , m. sera and t. takabatake , j. magn . , * 226 - 230 * , 137 ( 2001 ) . f. hulliger , m. landolt , h.r . ott and r. schmelczer , j. low . phys . , * 20 * , 269 ( 1975 ) . d.e . maclaughlin , j.d . williamson and j. butterworth , phys . rev . , * 4 * , 60 ( 1971 ) . a. narath , phys . , * 162 * , 320 ( 1967 ) . y. kawasaki , t. tanaka , m. izumi , y. kishimoto , t. ohno , h. tou , y. inaoka , m. sera , k. shigetoh and t. takabatake , j. phys . solids , to be published . k. nakamura , y. kitaoka , k. asayama , t. takabatake , g. nakamoto , h. tanaka and h. fujii , phys . b , * 53 * , 6385 ( 1996 ) . k. izawa , t. suzuki , m. kitamura , t. fujita , t. takabatake , g. nakamoto , h. fujii and k. maezawa , j. phys . * 65 * , 3119 ( 1996 ) . k. nakamura , y. kitaoka , k. asayama , t. takabatake , g. nakamoto and h. fujii , phys . b , * 54 * , 6062 ( 1996 ) . h. kontani and k. yamada , j. phys . * 65 * , 172 ( 1996 ) .	sb - nmr / nqr study has revealed a formation of a pseudogap at the fermi level in the density of states in a valence fluctuating compound ceirsb . the nuclear spin - lattice relaxation rate divided by temperature , @xmath0 has a maximum around 300 k and decreases significantly as @xmath1 , followed by a @xmath0 = const . relation at low temperature . this temperature dependence of @xmath0 is well reproduced by assuming a v - shaped energy gap with a residual density of states at the fermi level . the size of energy gap for ceirsb is estimated to be about 350 k , which is by one order of magnitude larger than those for the isostructural kondo semiconductors cerhsb and cenisn . despite the large difference in the size of energy gap , ceirsb , cerhsb and cenisn are indicated to be classified into the same group revealing a v - shaped gap due to c - f hybridization . the temperature dependence of the knight shift measured in a high magnetic field agrees with the formation of this pseudogap .
when one analyzes data that arrive sequentially over time , it is important to detect changes in the underlying model which can then be adjusted accordingly . such problems arise in many engineering ( signal processing , speech recognition , communication systems ) , econometric and biomedical applications and can be found in an extensive literature widely scattered in these fields . inference on time - varying parameters in stochastic systems is therefore of fundamental interest in sequential analysis . consider an @xmath0-valued time series @xmath1 , @xmath2 , @xmath3 , @xmath4 , such that at time moment @xmath5 the first observation @xmath6 and subsequently at each time moment @xmath7 a new observation @xmath8 arrives according to the model @xmath9 , where @xmath10 . suppose we are interested in certain characteristics of the conditional distribution of @xmath8 given the past @xmath11 : @xmath12 . here @xmath13 is an operator mapping conditional distributions @xmath14 into measurable @xmath15-valued functions @xmath16 , @xmath17 , @xmath15 is a compact subset of @xmath18 . the goal is to estimate ( or to track ) @xmath19 at time instant @xmath20 , based on the data @xmath21 ( and prior information ) available by that time moment . the traditional parametric formulation is the most simple particular case of the above setting : the observations are independent and the parameter @xmath22 is a constant vector . the simplest nonparametric formulation deals again with independent observations and time - varying parameter @xmath23 , @xmath20 ( cf . . modeling observations by a markov chain with a time varying parameter of the transition law would add a next level of complexity ( cf . for the autoregressive model in @xcite ) . the proposed time series formulation admits an arbitrary dependence structure between the observations . another important and peculiar feature of our approach is that the multidimensional parameter @xmath24 , @xmath7 , besides being time - varying , is also allowed to depend on the past of the time series . it is thus a predictable process with respect to the natural filtration : @xmath25 . an example of such characteristics is the conditional expectation @xmath26 $ ] . this time series formulation , with time - varying parameter of interest which is also allowed to depend on the past , represents the most general sequential framework , independent observations and markov chains are typical examples of models that fit into this framework . since the data arrives in a successive manner , conventional methods based on samples of fixed size are not easy to use . a more appropriate approach is based on sequential methods , stochastic recursive algorithms , which allow fast updating of parameter or state estimates at each instant as new data arrive and therefore can be used to produce an `` online '' inference , that is , during the operation of the system . stochastic recursive algorithms , also known as stochastic approximation , take many forms and have numerous applications in the biomedical , socio - economic and engineering sciences , which highlights the interdisciplinary nature of the subject . there is a vast literature on stochastic approximation beginning with the seminal papers of robbins and monro @xcite and of kiefer and wolfowitz @xcite . there is a big variety of techniques in the area of stochastic approximation which have been developed and inspired by the applications from other fields . we mention here the books @xcite . a classical topic in adaptive control concerns the problem of tracking drifting parameters of a linear regression model , or somewhat equivalently , tracking the best linear fit when the parameters change slowly . this problem also occurs in communication theory for adaptive equalizers and noise cancellation , etc . , where the signal , noise , and channel properties change with time . successful stochastic approximation schemes for tracking in the time - varying case were given by @xcite ( see further references therein ) . coming back to our time series model , the problem of tracking a time - varying parameter that is a functional of conditional distribution of the current observation given the past , is clearly unfeasible , especially in such general formulation , without some conditions on the model . in general , some knowledge about the structure of underlying time seres and some control over the variability of the parameter of interest over time are needed . interestingly , in this seemingly very general time series framework , we actually do not require the ( full ) knowledge of the observational model . instead , all we do need is to be able to compute a so called _ gain vector _ at each time moment @xmath7 , which is a certain ( vector ) function of the previous estimate of the parameter @xmath19 , new observation @xmath8 and prehistory @xmath27 . the essential property of such gain vector is that it , roughly speaking , `` pushes '' in the right direction of the current value of true parameter to track . although the assumption about the existence of that gain vector seems to be rather strong , we demonstrate on a number of interesting examples when such an assumption indeed holds . basically , in case of markov chain observations , if the form of transition density is known as function of the underlying parameter and it satisfies certain regularity assumptions , then the gain vector can always be constructed , for example , as a score function corresponding to the conditional maximum likelihood method . under appropriate regularity conditions ( the existence of the conditional fisher information and @xmath28-differentiability of the conditional log likelihood ) , such a score function has always the property of gain vector at least locally . a gain function , together with a step sequence and new observations from the model , can be used to adjust the current approximation of the drifting parameter , resulting in a tracking algorithm . to ease the verification of our assumptions on the gain function , we formulate them in two equivalent forms . under some assumptions on the gain vectors , we establish a uniform non - asymptotic bound the @xmath29 error of the resulting tracking algorithm , in terms of the variation of the drifting parameter . under the extra assumption that the gain function is bounded , we can strengthen this result to a uniform bound on the @xmath30 error ( and then an almost sure bound ) . these error bounds constitute our main result and they also guide us in the choice of the step size for the algorithm . some extensions are also presented where we allow for approximation terms and approximate gains . based on our main result , we specify the appropriate choice for the step sequence in three different variational setups for the drifting parameter . we treat first the simple case of a constant parameter . although we are mainly concerned with tracking time - varying parameters , our algorithm is still of interest in the constant parameter case since it should result in an algorithm which is both recursive and robust . we also consider a setup where the parameter is stabilizing . this covers both the case where the parameter is converging and where we sample the signal with increasing frequency . the third variational setup covers the important case of tracking smooth signals . this setup is somewhat different in that we make observations with a certain frequency from an underlying continuous time process which is indexed by a parameter changing like a lipschitz function . our result can then either be interpreted as a uniform , non - asymptotic result for each fixed sampling frequency or as an asymptotic statement in the observation frequency . examples are also given for different possible gain functions . these fall into two categories : general , score based gain functions for tracking multidimensional parameters in regular models and specialized gains for tracking more specific quantities . the latter include gains to track level sets or maxima of drifting functions ( extending the classical robbins - monro and kiefer - wolfowitz algorithms ) and gains to track drifting conditional quantiles . we also propose modifications for a given gain function ( rescaling , truncation , projection ) which can be used to design gains tailored specifically to verify our assumptions . we illustrate our method by treating some concrete applications of the proposed algorithm , in particular , we elaborate on the problem of tracking drifting parameters in autoregressive models . results on tracking algorithms for these models already exist in the literature ( cf.@xcite ) and we can derive similar results by choosing an appropriate gain function . using our approach , obtaining error bounds on the resulting tracking algorithm reduces to verifying our assumptions for the chosen gain function which considerably simplifies the derivation of results . this paper is structured as follows . in section [ sec : preliminaries ] we summarize the notation that will be used thought the paper , as well as our model and two equivalent formulations for our assumptions . section [ sec : main_result ] contains our main result and respective proof as well as some straightforward extensions of the main result . the construction and modification of gain functions for different models and different parameters of interest is explained in section [ sec : gains ] . section [ sec : variational_setups ] contains three examples of variational setups for the time - varying parameter for which we specify the tracking error implied by our main result . we collect in section [ sec : examples ] some examples of applications . section [ sec : proofs ] contains the proofs for our lemmas . first we introduce some notation that we are going to use throughout the paper . all vectors are always column vectors . we use bold letters to represent matrices and families of vectors , uppercase letters for families of random vectors and matrices ; denote @xmath31 for a set of vectors @xmath32 . for @xmath33 , @xmath34 ( of course , @xmath35 ) , @xmath36 . we use interchangeably @xmath37 for @xmath38 . for vectors @xmath39 , denote by @xmath40 and @xmath41 the usual euclidean norm and the inner product in @xmath18 , respectively , and by @xmath42 the @xmath43 norm ( with @xmath44 ) on vectors in @xmath18 . for an event @xmath45 , we denote by @xmath46 the indicator of the event @xmath45 . for a symmetric @xmath47-matrix @xmath48 , let @xmath49 and @xmath50 be the smallest and the largest eigenvalues of @xmath48 respectively . denote by @xmath51 denote the zero matrix and by @xmath52 the identity matrix , whose dimensions will be determined by the context . besides , we adopt the convention that @xmath53 and @xmath54 for matrices @xmath55 and @xmath56 . when applied to matrices , @xmath57 represents the matrix norm induced by the @xmath43 vector norm : @xmath58 . assume that by time @xmath59 , time series data @xmath60 ( which we may not be fully observable ) occur according to the following model : @xmath61 here vector @xmath8 takes value in some set @xmath62 with @xmath4 , i.e. , @xmath63 , @xmath20 . by @xmath64 we denote the conditional distribution of @xmath8 given @xmath27 and @xmath65 for @xmath20 , with @xmath66 . thus , @xmath67 takes values in @xmath68 . clearly , the distribution of @xmath69 , @xmath59 , is given by @xmath70 where @xmath71 should be understood as @xmath72 . for the sake of consistent notation , @xmath73 ( also @xmath74 etc . ) means void variables from now on . introduce an increasing sequence of @xmath75-algebras @xmath76 ( i.e. , @xmath77 if @xmath78 ) such that @xmath79 , @xmath20 , where @xmath80 denote the @xmath75-algebra generated by @xmath67 and @xmath81 is a @xmath75-algebra such that @xmath82 . unless otherwise specified , we assume that @xmath83 . as is discussed in remark [ rem : filtration ] below , the case @xmath84 is not conceptually new because it can be reduced to the situation of the time series @xmath85 with @xmath86 for some @xmath87-measurable functions @xmath88 s such that @xmath89 , @xmath90 . now we describe the statistical model which is a family probability measures for the observed data . for each @xmath20 , let @xmath91 be a given family of conditional probability measures of @xmath8 given @xmath92 . then the underlying statistical model is determined by imposing @xmath93 , @xmath20 . thus , at time @xmath94 , the underlying ( growing ) statistical model is @xmath95 . for some compact subset @xmath15 of @xmath96 , denote by @xmath97 the borel @xmath75-algebra on @xmath15 and by @xmath98 the set of @xmath15-valued @xmath99-measurable functions on @xmath68 , @xmath20 . consider a sequence of operators @xmath100 , @xmath20 , so that for a @xmath93 @xmath101 with @xmath102 . we will often abbreviate @xmath103 , remembering that this is a measurable function of @xmath104 . for @xmath5 , since @xmath73 is void , @xmath105 means a constant . our goal is to design an online algorithm for tracking the drifting _ parameter of interest _ @xmath19 at time moment @xmath90 on the basis of the data @xmath67 observed by that time moment . the time - varying parameter @xmath25 , @xmath20 , is thus allowed to depend on the past of the time series , i.e. , it is a predictable process with respect to the filtration @xmath106 . recall further that @xmath107 is assumed to take values in some compact subset @xmath15 of @xmath18 , to be precise , @xmath108 for all @xmath90 . denote @xmath109 at time @xmath110 , given @xmath67 , the model @xmath111 contains all the relevant information about the next observation . actually , we do not consider the model to be ( completely ) known . instead , we assume that our prior knowledge about the model is formalized as follows : for each @xmath20 , we have certain @xmath112-measurable functions @xmath113 at our disposal ( which we call _ gain functions _ or _ gain vectors _ or just _ gains _ ) , @xmath114 . we use these gain functions to construct a recursive algorithm for tracking the sequence @xmath115 from the observations ( [ model ] ) : @xmath116 for some positive sequence @xmath117 and some ( arbitrary ) initial value @xmath118 , measurable with respect to @xmath81 . since @xmath119 is @xmath120-measurable , then @xmath121 is predictable with respect to the filtration @xmath106 . notice that @xmath122 can be a random vector if @xmath81 is not the trivial @xmath75-algebra . of course , it is not to be expected that the tracking algorithm ( [ eq : algorithm_main ] ) performs well for arbitrary gains . intuition suggests that the gain @xmath113 should `` push '' @xmath123 in the direction of @xmath107 . the following conditions formalize this requirement . * for all @xmath20 , the quantity , which we call _ ( conditional ) average gain _ , @xmath124\ ] ] is well defined ( recall that @xmath125 and @xmath123 is defined by ( [ eq : algorithm_main ] ) ) and there exist a @xmath120-measurable symmetric positive definite matrix @xmath126 ( its entries are @xmath120-measurable functions ) such that , almost surely @xmath127 @xmath128,\quad \lambda_1 \le\lambda_{(d)}(\bm{m}_k ) \le \lambda_2,\ ] ] for some fixed constants @xmath129 . * there exists a constant @xmath130 such that @xmath131 as one can see from ( [ eq : lambda ] ) , a @xmath75-algebra @xmath132 , such that @xmath133 , is also needed . without loss of generality , assume that @xmath134 , the trivial @xmath75-algebra on @xmath135 . we will often use shorthand notation : @xmath136 and @xmath137 . condition ( a1 ) means , in a way , that the gain @xmath138 shifts any @xmath139 , on average , towards the `` true '' value @xmath140 . this elucidates the idea of algorithm ( [ eq : algorithm_main ] ) . suppose at time instant @xmath141 an observer had a reasonable estimator @xmath142 of the `` old '' value of the parameter of interest @xmath143 , and a new data vector @xmath8 arrives . then the available data is @xmath144 and the observer can construct an estimator @xmath145 of the new value of the parameter @xmath25 by calculating the gain @xmath146 and using a rescaled ( by a step size @xmath147 ) version of it to update the `` old '' estimator @xmath148 towards @xmath25 . the upper bound in relation ( [ eq : lambda ] ) means that the gain is of a bounded magnitude , and the lower bound has the meaning of the so called _ persistence of excitation _ as it is termed in control theory literature . assumption ( a2 ) is trivially satisfied if the gain vectors @xmath138 are almost surely uniformly bounded . this is not so difficult to arrange , for example , by dividing the gain vector by a multiple of its length or by truncating . in doing so , we make the resulting gain vector bounded , whereas retaining its direction . we discuss these approaches in more detail in section [ sec : gains ] . assumption ( a2 ) is also satisfied if @xmath149 indeed , for a random vector @xmath150 with a finite second moment of its norm and any @xmath75-algebra @xmath151 , @xmath152\le { { \mathbb{e}}}\\|x\\|^2 $ ] . combining this with ( [ g_bounded ] ) yields @xmath153 , @xmath90 . on the other hand , from ( [ c_theta ] ) , ( a1 ) and ( a2 ) it follows that @xmath154 this relation and lemma [ lemma_bound ] below ( thus the conditions of lemma [ lemma_bound ] must hold ) will in turn imply the uniform bound ( [ g_bounded ] ) . generally , there is no universal way to find gain vectors which satisfy conditions ( a1 ) and ( a2 ) . in many practical situations , the model @xmath155 is typically specified and it is an art to find gain vectors which satisfy ( a1 ) and ( a2 ) ; we discuss this issue in more detail in section [ sec : gains ] . the assumptions above look somewhat unnatural and cumbersome because they are assumed to hold for all @xmath156 , whereas functions involved in the conditions depend in general on @xmath27 whose dimension increases unlimitedly as @xmath110 increases . however , the assumptions become reasonable in the important case of markov chain observations @xmath157 of order , say , @xmath158 . in this case , for any @xmath141 we can use vector of bounded dimension @xmath159 instead of @xmath27 ( of growing dimension ) in all the quantities from conditions ( a1 ) and ( a2 ) . independent observations is a next simplification , also important in many practical applications . in this case there is no past involved in the function @xmath107 , @xmath160 , it will only be a function of time . [ rem : filtration ] suppose that conditions ( a1 ) and ( a2 ) hold for the filtration @xmath161 and for some measurable gain functions @xmath162 , @xmath20 , but the parameter sequence @xmath163 is predictable with respect to a coarser filtration @xmath76 , i.e. , @xmath164 , @xmath20 , where @xmath165 for some measurable @xmath88 s . for example , the vector @xmath8 consist of two subvectors @xmath166 and @xmath167 ( i.e. , @xmath168 ) and @xmath169 with @xmath170 ( think of @xmath171 as unobservable part of @xmath67 and @xmath172 as observable ) . then , by the tower property of the conditional expectation , conditions ( a1 ) and ( a2 ) hold for the filtration @xmath76 as well if we take the new gain function @xmath173 $ ] , @xmath20 . in fact , this means that the case of a coarser filtration @xmath76 , with respect to which the parameter sequence @xmath163 is predictable , can be reduced to the above described setup in terms of `` new time series '' @xmath172 , @xmath90 , and the filtration @xmath76 generated by this new time series ( or other way around ) . thus , if the parameter sequence @xmath163 is known to be predictable with respect to a coarser filtration @xmath76 , we could have considered this coarser filtration and the corresponding time series @xmath174 from the very beginning , and impose conditions ( a1 ) and ( a2 ) in terms of @xmath174 and @xmath76 . in fact , the coarser the filtration @xmath76 , the weaker the conditions . on the other hand , if the parameter sequence @xmath163 is known to be predictable with respect to a finer filtration @xmath76 ( i.e. , @xmath175 , @xmath20 ) , then the key relation ( [ eq : d ] ) will most likely not hold since the expression on the left hand side is @xmath80-measurable and the expression on the right hand side is @xmath176-measurable . however , in such situation ( [ eq : d ] ) may still hold with some ( small ) error @xmath177 , we address this issue in remark [ rem : close_parameter ] below . intuitively , the information of what we observe should match ( or , at least , not be less than ) the information of what we want to track . [ rem : parametrized ] consider one particular case of our general setting . at time @xmath94 , we observe @xmath60 such that @xmath178 the model in this case is @xmath179 and the operator @xmath180 . this is a convenient formulation when the time series model is parametrized by a time - varying parameter which we would like to recover by using an online tracking algorithm . also in this case we can actually allow the parameter @xmath107 to depend on the past of the time series , i.e. , @xmath25 so that the sequence @xmath181 is predictable with respect to the filtration @xmath76 . condition ( a1 ) can be reformulated as condition ( 1 ) below , which gives some intuition about the role of the average gain @xmath182 defined in ( a1 ) and which may , in certain situations , be easier to verify . * for @xmath160 , the average gain @xmath183 defined in ( a1 ) satisfies , almost surely , the following conditions : there exist random variables @xmath184 , @xmath185 and constants @xmath186 , @xmath187 such that @xmath188 and @xmath189 with @xmath190 $ ] and @xmath191 . in view of the lemma below , if ( a1 ) holds , then ( 1 ) will also hold ( and vice versa ) ; the values of the constants @xmath192 and @xmath193 appearing in the assumptions are different , though . the proof of this lemma is deferred to section [ sec : proofs ] . [ lemma1 ] let @xmath194 . if there exists a symmetric positive definite matrix @xmath48 such that @xmath195 and @xmath196 for some @xmath197 , then @xmath198 and @xmath199 for some @xmath200 ( depending only on @xmath201 ) such that @xmath202 and @xmath203 . conversely , if @xmath198 and @xmath199 for some @xmath200 such that @xmath202 and @xmath203 , then there exists a symmetric positive definite matrix @xmath48 such that @xmath195 and @xmath196 for some constants @xmath197 depending only on @xmath204 and @xmath205 . we start with a lemma which we will need in the proof of the main result . heuristically , since the gain vector @xmath146 moves , on average , @xmath123 towards @xmath107 and the sequence @xmath206 is bounded ( since @xmath15 is compact ) , the resulting estimating sequence @xmath123 should also be well - behaved . the following lemma states that the second moment of @xmath123 is uniformly bounded in @xmath20 for sufficiently small @xmath207 s . [ lemma_bound ] let assumptions ( a1 ) and ( a2 ) hold . then for sufficiently small @xmath207 there exists a constant @xmath208 such that @xmath209 the proof of this lemma is given in section [ sec : proofs ] . in fact , it is enough to assume that @xmath207 is sufficiently small for all @xmath210 for some fixed @xmath211 . this is the case if @xmath212 as @xmath213 , which is typically assumed . this lemma will be used in the proof of the main theorem below . from now on we assume that the sequence @xmath207 is such that lemma [ lemma_bound ] holds . the following theorem is our main result , it provides a non - asymptotic upper bound on the quality of the tracking algorithm ( [ eq : algorithm_main ] ) in terms of of the algorithm step sequence @xmath214 and oscillation of the process to track @xmath215 between arbitrary time moments @xmath216 , @xmath217 . [ theo : bound ] let assumptions ( a1 ) and ( a2 ) hold , the tracking sequence @xmath123 be defined by ( [ eq : algorithm_main ] ) and @xmath218 , @xmath90 . then for any @xmath219 and sequence @xmath220 ( satisfying the conditions of lemma [ lemma_bound ] ) such that @xmath221 and @xmath222 for all @xmath223 , the following relation holds : @xmath224^{1/2 } \!\!+ c_3\max_{k_0\le i\le k } { { \mathbb{e}}}\\|\theta_{i+1}-\theta_{k_0}\\|,\ ] ] where @xmath225 , @xmath226 , @xmath227 , constants @xmath228 are from assumptions ( a1 ) and ( a2 ) , @xmath229 is defined by ( [ c_theta ] ) and @xmath208 is from lemma [ lemma_bound ] . [ rem : norms ] by using ( [ eq : bounde ] ) , one can derive a bound for @xmath230 , with @xmath44 . indeed , as @xmath231 for any @xmath232 and @xmath233 , we obtain that @xmath234 for @xmath235 and @xmath236 for @xmath237 . for the sake of brevity , denote @xmath238 , @xmath239 , @xmath240 and @xmath241 , @xmath20 . we have @xmath242= { { \mathbb{e}}}[(g_k - g_k)\|\mathcal{f}_{k-1}]=g_k - g_k=0 , \quad k \in \mathbb{n}_0.\ ] ] it follows that @xmath243 , is a ( vector ) martingale difference sequence with respect to the filtration @xmath106 . rewrite the algorithm equation ( [ eq : algorithm_main ] ) as @xmath244 in view of ( a1 ) , the decomposition @xmath245 holds almost surely , with an @xmath120-measurable symmetric positive definite matrix @xmath246 , so that @xmath247 by iterating the above relation , we obtain that for any @xmath248 @xmath249 \delta_{k_0 } + \sum_{i = k_0}^k \big[\prod_{j = i+1}^k(\bm{i}- \gamma_j \bm{m}_j)\big ] ( \delta\theta_i+\gamma_i d_i).\end{aligned}\ ] ] denote @xmath250 , @xmath251 and @xmath252 . applying the abel transformation ( lemma [ lemma : abel ] ) to the second term of the right hand side of ( [ eq : recursed ] ) yields @xmath253 ( \delta\theta_i+\gamma_i d_i ) = h_k - \sum_{i = k_0}^{k-1}\ ! \gamma_{i+1}\bm{m}_{i+1 } \big[\!\prod_{j = i+2}^k(\bm{i}-\gamma_j \bm{m}_j)\big]h_i.\ ] ] in particular , note that if we take @xmath254 , @xmath255 and @xmath256 for @xmath257 , @xmath258 and @xmath259 for @xmath260 , we derive that ( since @xmath261 for @xmath262 ) @xmath263 which we will use later . using ( [ abel_transform ] ) , we can rewrite our expansion of @xmath264 in ( [ eq : recursed ] ) as follows : @xmath265 \delta_{k_0 } + h_k - \sum_{i = k_0}^{k-1}\gamma_{i+1}\bm{m}_{i+1 } \big[\prod_{j = i+2}^k(\bm{i}-\gamma_j\bm{m}_j)\big]h_i.\end{aligned}\ ] ] the previous display , the minkowski inequality and the sub - multiplicative property of the operator norm ( @xmath266 ) imply that @xmath267 in view of ( a1 ) and the condition @xmath222 for @xmath268 , @xmath269 , @xmath268 , almost surely . hence @xmath270 , @xmath268 , almost surely . this , lemma [ lemma : eig ] and the fact ( see ( [ eq : lambda ] ) from ( a1 ) ) that @xmath271 $ ] , @xmath268 , almost surely , imply that @xmath272\notag\\ & = { { \mathbb{e}}}\big [ { { \mathbb{e}}}\big [ \big(1-\gamma_k \lambda_{(1)}(\bm{m}_k)\big)\big\|\mathcal{f}_{k-2}\big ] \prod_{i = k_0}^{k-1 } \big(1- \gamma_i \lambda_{(1)}(\bm{m}_i)\big ) \big ] \notag\\ & \label{eq : m_i } \le ( 1-\gamma_k \lambda_1 ) { { \mathbb{e}}}\prod_{i = k_0}^{k-1}\big(i-\gamma_i \lambda_{(1)}(\bm{m}_i)\big ) \le\ldots \le \prod_{i = k_0}^k ( 1- \gamma_i \lambda_1 ) .\end{aligned}\ ] ] by ( [ c_theta ] ) and lemma [ lemma_bound ] , we have @xmath273 . using this fact , the cauchy - schwartz inequality , ( [ eq : m_i ] ) and the elementary inequality @xmath274 , @xmath275 , leads to @xmath276 & \le \big[{{\mathbb{e}}}\\|\delta_{k_0}\\|^2\ , { { \mathbb{e}}}\prod_{i = k_0}^k \\|\bm{i}- \gamma_i \bm{m}_i\\|^2\big]^{1/2 } \notag\\ & \label{eq : products } \le c_1\prod_{i = k_0}^k ( 1- \gamma_i \lambda_1 ) ^{1/2 } \le c_1\exp\big\{-\frac{\lambda_1}{2}\sum_{i = k_0}^k\gamma_i\big\}.\end{aligned}\ ] ] let @xmath277 denote the @xmath278-th coordinate of the vector @xmath279 . clearly , for each @xmath280 , @xmath281 is a martingale difference with respect to the filtration @xmath106 . using ( [ eq : g2 ] ) from ( a2 ) and the fact that martingale difference terms are uncorrelated , we derive that for all @xmath282 @xmath283 denote for brevity @xmath284 , so that @xmath285 , @xmath282 . the obtained relation for @xmath286 , together with the minkowski and hlder inequalities , imply that for all @xmath282 @xmath287 notice that @xmath288 from ( [ eq : after_triangle_lp ] ) is @xmath289-measurable for all @xmath290 . therefore , by ( [ eq : lambda ] ) from ( a1 ) , ( [ eq : eh_i ] ) and lemma [ lemma : eig ] , @xmath291 = { { \mathbb{e}}}{{\mathbb{e}}}\big[\\|\bm{m}_{i+1}\\|\\|h_i\\| \prod_{j = i+2}^k \\|\bm{i}- \gamma_j \bm{m}_j\\| \big\| \mathcal{f}_{k-2 } \big]\notag\\ & = { { \mathbb{e}}}\big [ { { \mathbb{e}}}\big[1-\gamma_k \lambda_{(1)}(\bm{m}_k ) \big\| \mathcal{f}_{k-2 } \big ] \\|\bm{m}_{i+1}\\|\\|h_i\\| \prod_{j = i+2}^{k-1 } \big(1- \gamma_j \lambda_{(1)}(\bm{m}_j)\big ) \big ] \notag\\ & \le(1-\gamma_k\lambda_1 ) { { \mathbb{e}}}\big[\\|\bm{m}_{i+1}\\|\\|h_i\\| \prod_{j = i+2}^{k-1 } \big(1- \gamma_j \lambda_{(1)}(\bm{m}_j)\big)\big ] \le \ldots \notag\\ & \le { { \mathbb{e}}}\big(\\|\bm{m}_{i+1}\\| \\|h_i\\|\big ) \prod_{j = i+2}^k(1- \gamma_j \lambda_1 ) \le \lambda_2 { { \mathbb{e}}}\\|h_i\\| \prod_{j = i+2}^k(1- \gamma_j \lambda_1 ) \notag\\ & \label{eq : c_i } \le \lambda_2 \big[c_g^{1/2 } \gamma_{k_0,k } + { { \mathbb{e}}}\\|b_i\\| \big ] \prod_{j = i+2}^k(1- \gamma_j \lambda_1).\end{aligned}\ ] ] now we take the expectation of relation ( [ eq : after_triangle_lp ] ) and use relations ( [ eq : products ] ) , ( [ eq : eh_i ] ) , ( [ eq : c_i ] ) and ( [ eq : bound_on_coefs ] ) to derive that @xmath292 is bounded from above by @xmath293 + { { \mathbb{e}}}\\|h_k\\| + \sum_{i = k_0}^{k-1 } \gamma_{i+1 } { { \mathbb{e}}}\big [ \\|\bm{m}_{i+1}\\| \\|h_i\\| \prod_{j = i+2}^k \\|\bm{i}-\gamma_j\bm{m}_j\\| \big]\\ & \le c_1\exp\big\{-\frac{\lambda_1}{2}\sum_{i = k_0}^k\gamma_i\big\ } + { { \mathbb{e}}}\\|h_k\\| + \lambda_2 \sum_{i = k_0}^{k-1 } \gamma_{i+1 } { { \mathbb{e}}}\\|h_i\\| \prod_{j = i+2}^k ( 1- \gamma_j \lambda_1 ) \\ & \le c_1\exp\big\{-\frac{\lambda_1}{2}\sum_{i = k_0}^k\gamma_i\big\}+ { { \mathbb{e}}}\\|h_i\\| \big[1+\sum_{i = k_0}^{k-1}\lambda_2 \gamma_{i+1 } \prod_{j = i+2}^k \big(1-\gamma_j \lambda_1\big)\big ] \\ & \le c_1\exp\big\{-\frac{\lambda_1}{2}\sum_{i = k_0}^k\gamma_i\big\}+ \big[c_g^{1/2 } \gamma_{k_0,k}+\max_{k_0\le i\le k } { { \mathbb{e}}}\\|\theta_{i+1}-\theta_{k_0}\\| \big ] \big(1+\frac{\lambda_2}{\lambda_1}\big),\end{aligned}\ ] ] where we also used in the last bound that @xmath294 , @xmath295 , is a telescopic sum . this completes the proof of the the theorem . at this stage , it may not be clear how the non - asymptotic bound from theorem [ theo : bound ] can be utilized . the obtained result is not useful unless we assume some sort of damping of the oscillations of the parameter process @xmath181 . looking ahead , in section [ sec : variational_setups ] we impose certain settings for damping of the parameter process oscillations ( either `` stabilizing '' in time or increasing the observation frequency ) and derive results in various asymptotic regimes by using our main theorem [ theo : bound ] . if we assume a slightly stronger version of ( [ eq : lambda ] ) in ( a1 ) , @xmath296 almost surely , then a slightly better version of bound ( [ eq : products ] ) holds : @xmath297 \le \bar{c}_1 \prod_{i = k_0 + 1}^k ( 1- \gamma_i \lambda_1 ) \le \bar{c}_1\exp\big\{-\lambda_1\sum_{i = k_0}^k\gamma_i\big\},\ ] ] since @xmath298 , by ( [ c_theta ] ) and lemma [ lemma_bound ] . we can derive a bound alternative to ( [ eq : products ] ) , which leads to slightly better constants in the first term of the right hand side of ( [ eq : bounde ] ) . indeed , @xmath299 is @xmath120 -measurable , and , instead of ( [ eq : products ] ) , we derive @xmath300 = { { \mathbb{e}}}{{\mathbb{e}}}\big[\\|\delta_{k_0}\\| \prod_{i = k_0}^k \big(1- \gamma_i \lambda_{(1)}(\bm{m}_i)\big)\big\|\mathcal{f}_{k-2}\big ] \\ & \le{{\mathbb{e}}}\big [ \\|\delta_{k_0}\\| \prod_{i = k_0}^{k-1}\big(1-\gamma_i \lambda_{(1)}(\bm{m}_i)\big)\big ] ( 1-\gamma_k \lambda_1 ) \le\ldots \\ & \le { { \mathbb{e}}}\big[\\|\delta_{k_0}\\|\big(1-\gamma_{k_0 } \lambda_{(1)}(\bm{m}_{k_0})\big ) \big ] \prod_{i = k_0 + 1}^k ( 1- \gamma_i \lambda_1 ) \le { { \mathbb{e}}}\big [ \\|\delta_{k_0}\\|\big ] \prod_{i = k_0 + 1}^k ( 1- \gamma_i \lambda_1 ) \\ & \le ( \bar{c}_\theta^{1/2}+c_\theta^{1/2 } ) \prod_{i = k_0 + 1}^k ( 1- \gamma_i \lambda_1 ) .\end{aligned}\ ] ] if we assume that @xmath301 for all @xmath302 and @xmath303 , then we can prove lemma [ lemma_bound ] in another way : first take the expectation of the second power of the relation ( [ eq : after_triangle_lp ] ) with @xmath304 to establish that @xmath305 is uniformly bounded in @xmath20 , and then @xmath306 , by ( [ c_theta ] ) . [ rem_version_theorem ] one can try to establish a version of theorem [ theo : bound ] where , instead of ( [ eq : lambda ] ) , one assumes @xmath307\big ) \le \lambda_{(d ) } \big(\mathbb{e } [ \bm{m}_k \|\mathcal{f}_{k-2}]\big ) \le \lambda_2 \quad \mbox{almost surely}.\ ] ] the point is that there may be situations with certain gain functions when ( [ eq : lambda ] ) does not hold but ( [ different_bound ] ) does ; see remark [ rem_mouilines ] below . the idea of the proof would be to first introduce @xmath308 $ ] , @xmath309 , and then , beginning with the relation ( [ eq : recursed ] ) , work with the representation @xmath310 instead of just @xmath311 , using the relation ( [ different_bound ] ) for @xmath312 and the fact that @xmath313 , is a ( matrix ) martingale difference sequence with respect to the filtration @xmath314 . we will not pursue this here . imposing somewhat stronger versions of conditions ( a1 ) and ( a2 ) enables us to derive a similar non - asymptotic bound for the expectation of @xmath315 for all @xmath44 . of course , the bigger @xmath158 , the bigger the constants involved in the bound . the next theorem is a strengthened version of the previous result . [ theo : bound2 ] suppose that the conditions of theorem [ theo : bound ] are fulfilled . if , in addition ( to assumption ( a1 ) ) , @xmath316 and @xmath317 ( instead of ( a2 ) ) almost surely for all @xmath318 , then for any @xmath44 @xmath319^{p/2 } + c'_3\max_{k_0\le i\le k}{{\mathbb{e}}}\\|\theta_{i+1}-\theta_{k_0}\\|_p^p,\end{aligned}\ ] ] where @xmath320 , @xmath321 , @xmath322 and @xmath323 is the constant from lemma [ lemma : eig ] . now we have stronger versions of assumptions ( a1 ) and ( a2 ) : @xmath324 hold almost surely . along the same lines as for ( [ eq : after_triangle_lp ] ) , by using lemma [ lemma : eig ] , ( [ eq : a1_stronger ] ) , ( [ eq : bound_on_coefs ] ) and the elementary inequality @xmath325 , we obtain that @xmath326\\ & \le k_p \\|\delta_{k_0}\\|_p \exp\big\{-\lambda_1\sum_{i = k_0}^k\gamma_i\big\ } + \big[1+\frac{k_p^2\lambda_2}{\lambda_1}\big ] \big(\max_{k_0\le i\le k } \\|a_i\\|_p+\max_{k_0\le i\le k } \\|b_i\\|_p\big)\end{aligned}\ ] ] almost surely , where constant @xmath323 is from lemma [ lemma : eig ] . take now the @xmath158-th power of both sides of the inequality and apply the hlder inequality @xmath327 for @xmath328 to get @xmath329 recall that the sequence @xmath330 is a martingale with respect to the filtration @xmath331 and that the coordinates of @xmath332 verify @xmath333 almost surely , @xmath334 , @xmath262 . applying the maximal burkholder inequlity for @xmath335 and the davis inequality for @xmath336 ( cf . @xcite ) yields @xmath337^{p/2 } \le d b_p 2^p \bar{g}^p \bigg[\sum_{j = k_0}^k \gamma_j^2 \bigg]^{p/2},\end{aligned}\ ] ] for some constant @xmath338 . one can take @xmath339 for @xmath335 , cf . the second inequality of the theorem now follows by taking expectations on both sides of the bound on @xmath315 above and by using the last inequality . one can derive a similar result for the @xmath340 , by simply taking the @xmath158-th power of the inequality ( [ eq : after_triangle_lp ] ) and then proceeding in the same way as in the proof of theorem [ theo : bound2 ] , with minor modifications in the argument for the martingale @xmath341 . once a bound on @xmath340 is established , one can use it for proving theorem [ theo : bound2 ] in another way . namely , since @xmath231 for any @xmath232 and @xmath233 , @xmath342 , with @xmath343 if @xmath235 and @xmath344 if @xmath237 . thus , a bound for @xmath345 will immediately follow from the obtained bound for @xmath340 . the bound will be of the same form as in theorem [ theo : bound2 ] , but with different constants @xmath346 . [ rem : close_parameter ] consider the following situation , which we will call case i. suppose we are not interested in tracking the , say , _ natural _ parameter @xmath107 of the model , but rather some other time - varying parameter @xmath347 , which is also assumed to be predicable with respect to the filtration @xmath76 . denote @xmath348 , @xmath20 . the difference @xmath349 , @xmath160 , can be seen as an approximation error . similar to ( [ eq : recursed ] ) , the following expansion can be derived for the quantity @xmath350 : @xmath351 \delta_{k_0}^ * + \sum_{i = k_0}^k \big[\prod_{j = i+1}^k(\bm{i}- \gamma_j \bm{m}_j)\big ] ( \delta\theta^_i + \gamma_i\bm{m}_i \varepsilon_i+\gamma_i d_i).\end{aligned}\ ] ] now consider case ii : we want to track the natural parameter @xmath107 but the average gain makes an error @xmath177 , i.e. , @xmath352 , @xmath90 . the error term @xmath177 may be random but must be measurable with respect to @xmath120 . again , similar to ( [ eq : recursed ] ) , we can derive @xmath353 \delta_{k_0 } + \sum_{i = k_0}^k \big[\prod_{j = i+1}^k(\bm{i}- \gamma_j \bm{m}_j)\big ] ( \delta\theta_i + \gamma_i\eta_i + \gamma_i d_i).\ ] ] now notice that case i can actually be reduced to case ii by putting in the last relation @xmath354 and @xmath355 ( where @xmath356 ) , @xmath309 . therefore , consider only case ii from now on . under the conditions of theorem [ theo : bound ] , in the same way as for ( [ eq : bounde ] ) , we can derive the following bound : @xmath357^{1/2}\\ & + c_3{{\mathbb{e}}}\max_{k_0\le i\le k } \\|\theta_{i+1}-\theta_{k_0}\\| + c_3 { { \mathbb{e}}}\sum_{i = k_0}^k \gamma_i\\|\eta_i\\| . \end{aligned}\ ] ] similarly , under the conditions of theorem [ theo : bound2 ] , @xmath358^{p/2}\\ & + c'_3 { { \mathbb{e}}}\big[\max_{k_0\le i\le k } \\|\theta_{i+1}-\theta_{k_0}\\|_p + \sum_{i = k_0}^k\gamma_i\\|\eta_i\\|_p\big]^p . \end{aligned}\ ] ] clearly , ( [ eq : bounde2 ] ) and ( [ eq : boundas2 ] ) generalize the bounds of theorems [ theo : bound ] and [ theo : bound2 ] , where we had @xmath359 , @xmath309 . in case i , we have @xmath354 and @xmath360 with @xmath349 , @xmath309 , in relations ( [ eq : bounde2 ] ) and ( [ eq : boundas2 ] ) . noting that @xmath361 for all @xmath44 and @xmath309 , we can rewrite bounds ( [ eq : bounde2 ] ) and ( [ eq : boundas2 ] ) in terms of @xmath362 instead of @xmath363 with appropriate adjustments of corresponding constants . any gain function for which conditions ( a1 ) and ( a2 ) hold may be used with our algorithm , and whether a particular gain function is suitable or not depends on the model under study and the quantity that we wish to track . for certain types of models and quantities to track , there are natural choices for the gain function . many different settings are investigated in the literature . in this section we consider the construction of appropriate gain functions to be used in the algorithm ( [ eq : algorithm_main ] ) in several traditional settings . in particular , we relate our general approach to well known classical procedures such as robbins - monro and kiefer - wolfowitz algorithms and outline possible extensions . the traditional ` signal+noise ' situation can be represented by the following observation model : @xmath364 where @xmath365 , @xmath163 is a predictable process ( @xmath25 ) we are interested in tracking , @xmath366 is a martingale difference noise , with respect to the filtration @xmath76 . we use the algorithm ( [ eq : algorithm_main ] ) for tracking @xmath107 , and in this case we can simply take the following gain function @xmath367 since @xmath368 = - ( \hat{\theta}_k - \theta_k ) , \quad k \in \mathbb{n}_0,\ ] ] i.e. , @xmath369 . clearly , condition ( a1 ) holds and condition ( a2 ) follows as well if we assume @xmath370 , @xmath20 . indeed , according to ( [ g_bounded ] ) , it is enough to show the boundedness of the second moment of @xmath113 : @xmath371 \le c , \quad k \in \mathbb{n}_0,\ ] ] by virtue of the hlder inequality , lemma [ lemma_bound ] and ( [ c_theta ] ) . the classical nonparametric regression model fits into this framework so that our results can be applied . for example , the simplest nonparametric regression model with an equidistant design on @xmath372 $ ] is as follows : @xmath373 , @xmath374 , with independent noises @xmath375 s , @xmath376 , @xmath377 ; we will return to this issue in subsection [ sec : variational_setups : lipschitz ] . [ gain_autoregression ] possibly , @xmath378 = { \varphi}(\theta_k)$ ] for some smooth function @xmath379 . in this case , one should consider @xmath380 , @xmath381 , so that @xmath382 , @xmath160 , which should be comparable to @xmath383 . autoregressive models , for example , fall into this category ( cf . section [ sec : examples : ard ] ) . let us turn to more dynamical situations where the observations themselves depend on our tracking sequence . in their seminal paper , @xcite studied the problem of finding the unique @xmath384-root @xmath385 of a monotone function @xmath386 , i.e. , the equation @xmath387 has a unique solution at @xmath388 . the function @xmath386 can be observed at any point @xmath389 but with noise @xmath390 : @xmath391 so that @xmath392 . a stochastic approximation algorithm of design points converging to @xmath385 is known as classical robbins - monro procedure . we now illustrate how this also fits into our general tracking algorithm scheme . in fact , the following model essentially extends the original setup of @xcite . suppose there is a time series @xmath393 ( with @xmath171 taking values in @xmath394 ) running at the background , which is not ( fully ) observable . instead , some other @xmath395-dimensional ( related ) time series @xmath396 is observed , which we introduce below . as usual , let @xmath397 , @xmath398 . further , for a sequence of functions @xmath399 , let a @xmath395-dimensional measurable function @xmath25 be the unique solution of the equation @xmath400 , where @xmath401 $ ] , for some measurable function @xmath402 , @xmath90 . ( here @xmath171 may contain @xmath27 . ) the goal is to track the sequence @xmath403 . at a time moment @xmath90 , we observe the noise corrupted value of @xmath404 at some design point @xmath123 ( which can be picked on the basis of the previous observations @xmath27 , i.e. , @xmath119 ) : @xmath405 where @xmath406 is a martingale difference noise sequence with respect to the filtration @xmath106 ( indeed , let simply @xmath407 ) . of course , we could assume a more general model @xmath408 , @xmath160 , but this would not have made any principal difference , since variable @xmath375 can be incorporated into the vector @xmath167 . let the design points @xmath409 in ( [ robbins_monro ] ) be determined by the algorithm ( [ eq : algorithm_main ] ) and we want this algorithm to track @xmath107 . theorem [ theo : bound ] is applicable if the gain @xmath113 in ( [ eq : algorithm_main ] ) satisfies ( a1 ) and ( a2 ) . as in the robbins - monro algorithm , the gain is taken to be @xmath410 then @xmath411 , @xmath160 , and ( a2 ) is fulfilled if , for example , @xmath412 and @xmath413 , @xmath160 . condition ( a1 ) , or equivalently ( 1 ) , is fulfilled if , for some @xmath414 , @xmath415 almost surely . in the last display , one should recognizes the usual regularity requirements for the function @xmath386 in the classical robbins - monro setting : @xmath254 , @xmath416 and @xmath417 ( so that @xmath418 is the non - random solution of the equation @xmath419 ) : @xmath420 . in multidimensional case , this can be seen as a generalized identifiability requirement for the sequence @xmath181 . for example , if @xmath421 is a differentiable mapping in @xmath422 for each @xmath423 , then a sufficient condition for ( a1 ) is positive definiteness of the jacobian matrix of @xmath421 ( with respect to @xmath424 ) , uniformly in @xmath423 and over the support of @xmath123 . one can possibly relax this to a vicinity of the root @xmath107 under other appropriate conditions which guarantee that @xmath123 eventually gets into a neighborhood of @xmath107 . a particular example is @xmath425 $ ] , where @xmath166 is a subvector of @xmath167 , independent of @xmath120 . another classical example is the algorithm of @xcite for successive estimating the maximum of a function @xmath386 which can be observed at any point , but gets corrupted with a martingale difference noise ( similarly , one can formulate the problem of tracking minima of a sequence of functions ) . the algorithm is based on a gradient - like method , the gradient of @xmath386 being approximated by using finite differences . there are many modifications of the procedure , including multivariate extensions , and they are all based on estimates of the gradient of @xmath386 . the following scheme essentially contains many such procedures considered in the literature and even extends them to a time - varying predictable maxima process @xmath181 . as in the previous subsection , suppose there is a time series @xmath393 , with @xmath171 taking values in @xmath394 , running in the background , which is not ( fully ) observable . instead , some other related time series @xmath396 is observed , which we introduce below . let @xmath397 , @xmath398 . suppose we are given a sequence of measurable functions @xmath426 , @xmath427 , @xmath20 , such that the function @xmath428 $ ] has a unique maximum @xmath25 on @xmath15 , i.e. , @xmath429 again , @xmath171 may contain @xmath27 . we want to track the sequence @xmath403 . for that we use the sequence of design points @xmath409 defined by the tracking algorithm ( [ eq : algorithm_main ] ) , with @xmath430-measurable gain functions @xmath113 to be specified later . at a time moment @xmath90 , we observe the so called `` noisy approximate gradients '' at those design points : @xmath431 where @xmath432 and @xmath413 for all @xmath160 , and @xmath406 is a martingale difference noise sequence with respect to the filtration @xmath76 . the @xmath395-dimensional approximate gradient @xmath433 is not necessarily the exact pathwise @xmath424-derivative of @xmath434 at @xmath123 ( i.e. , @xmath435 ) but such that @xmath436 = \bar{f}_k(\hat{\theta}_k,\bm{x}_{k-1})=-\bm{m}_k(\hat{\theta}_k-\theta_k ) + \eta_k , \quad k \in \mathbb{n}_0,\ ] ] almost surely , where a symmetric positive definite matrix @xmath126 satisfies conditions ( [ eq : lambda ] ) and @xmath437 is some predictable approximation error . of course , such a representation ( [ appr_gradient ] ) is always possible : simply take @xmath438 for some symmetric positive definite matrix @xmath246 satisfying ( [ eq : lambda ] ) ; useful ones are those for which the @xmath177 s are under control basically , the @xmath177 s should be small . as a choice for the gain , take now @xmath439 , so that @xmath440 = \bar{f}_k(\hat{\theta}_k,\bm{x}_{k-1})=-\bm{m}_k(\hat{\theta}_k-\theta_k ) + \eta_k$ ] , @xmath160 . clearly , ( a2 ) holds in view of moment conditions on the quantities in ( [ kiefer_wolfowitz ] ) , however ( a1 ) is not satisfied in general since there is an approximation ( possibly nonzero ) term @xmath177 involved . yet , we are in the position of remark [ rem : close_parameter ] and thus the bound ( [ eq : bounde2 ] ) for the tracking error holds in this case . this bound is however useful only if the approximation errors @xmath177 s get sufficiently small as @xmath110 gets bigger . the most desirable situation is when @xmath441 , @xmath160 . for each particular model of form ( [ kiefer_wolfowitz ] ) , one needs to determine conditions that should be imposed on the approximate gradients @xmath442 s in order to be able to claim a reasonable quality of the tracking algorithm by using our general result . conditions on approximate gradients @xmath442 s from ( [ kiefer_wolfowitz ] ) which provide control on the magnitude of the approximation errors @xmath177 s are comparable to the ones proposed in many papers . examples can be found in ; see further references therein . commonly , a finite difference form of the gradient estimate is used as noisy approximate gradient . below we outline two settings . first consider the following situation which is very close to the classical kiefer - wolfowitz setting : @xmath443 for some subvector @xmath166 of @xmath167 , independent of @xmath27 defined below and we wish to maximize the function @xmath444 . for simplicity , let @xmath445 and all @xmath166 s are identically distributed ( although the generalization to the time - varying case is straightforward ) so that @xmath446 is to be maximized : @xmath447 . let @xmath448 be a positive sequence , @xmath449 be the standard orthonormal basis vectors in @xmath18 , @xmath450 en @xmath451 have the same distribution as @xmath166 , @xmath452 . denote @xmath453 , @xmath454 , likewise for @xmath455 and @xmath456 . the observations are the noisy finite difference estimates of the gradient : @xmath457 here @xmath458 is a martingale difference noise sequence with respect to the filtration @xmath76 , @xmath123 denotes the @xmath110th estimate of the maximum point @xmath385 according to the algorithm ( [ eq : algorithm_main ] ) with the gain @xmath459 . then , under some regularity conditions , @xmath460 where the magnitude of @xmath177 is controlled by @xmath461 . usually @xmath462 as @xmath463 in an appropriate way . to ensure that @xmath464 ( possibly with a small approximation error ) for some positive definite matrix @xmath246 satisfying ( [ eq : lambda ] ) , concavity of @xmath465 is typically required , either global or over a compact set which is known to include the maximum location @xmath385 . for example , if function @xmath465 is sufficiently smooth and strongly concave , then by taylor s expansion @xmath466 , the hessian matrix of @xmath465 at some point @xmath347 between @xmath123 and @xmath385 , the relations ( [ eq : lambda ] ) are fulfilled and the approximation error @xmath177 is small if @xmath461 is small . another approach ( due to @xcite ) is based on random direction instead of the unit basis vectors . we use the same notations as in the previous setting with one simplification : assume now that there are no vectors @xmath166 s involved in the model so that @xmath467 . let @xmath468 denote a sequence of independent ( @xmath279 is also assumed to be independent of @xmath27 ) random unit vectors in @xmath18 . at time moment @xmath160 we observe @xmath469 where the tracking sequence @xmath123 is defined by the algorithm ( [ eq : algorithm_main ] ) with the gain function @xmath470 . notice that one step in the previous ( classical kiefer - wolfowitz ) observation scheme requires in essence @xmath471 observations in design points @xmath472 , @xmath452 , whereas only two measurements must be made in the case of the above random direction observation scheme . this property was the main motivation for the random direction method introduced by @xcite . then , under some regularity conditions , @xmath473 \\ & = { { \mathbb{e}}}\big [ d_kd_k^t \big]\nabla f(\hat{\theta}_k)+ \eta_k = -\bm{m}_k(\hat{\theta}_k-\theta_k ) + \eta_k,\end{aligned}\ ] ] where @xmath474h(f)(\theta_k^)$ ] and again the magnitude of @xmath177 is controlled by @xmath461 . the relations ( [ eq : lambda ] ) hold if , for example , we assume that the random directions were chosen in such a way that @xmath475 $ ] are positive definite matrices and the hessian @xmath476 is negative definite . a particular choice of function @xmath477 is @xmath478 , @xmath20 , where @xmath479 s is a sequence of observations with values on a measurable space @xmath480 and @xmath481 is a loss function . then @xmath482 is the prediction risk of the predictor given by @xmath424 . classical examples are least squares and logistic regression ( cf . ): @xmath483 or @xmath484 $ ] , where @xmath485 , @xmath486 and @xmath487 , or @xmath488 for logistic regression . consider one more example . suppose @xmath489 and we would like to track the conditional quantile of the distribution of our observed time series @xmath490 , i.e. , @xmath25 such that @xmath491 , where the levels @xmath492 are of our choice and @xmath493 is the conditional distribution function of @xmath8 given the past @xmath27 . assume that this conditional distribution posses a density @xmath494 . in this case it makes sense to use @xmath495 in the algorithm ( [ eq : algorithm_main ] ) for tracking @xmath107 , since @xmath496 for some @xmath497 between @xmath123 and @xmath107 . under some mild conditions theorem [ theo : bound ] is applicable . note also that the algorithm based on this gain function only requires knowledge of the values of the indicators @xmath498 which means that we may still track the required quantiles without explicitly observing @xmath8 . this problem is treated in detail for the case of independent observations in . for certain models it may not be obvious how gain functions can be constructed , especially when tracking multi - dimensional parameters . it is therefore important to have a general procedure that can be used to construct candidate gain functions that can either be used directly or , if needed , modified to verify ( a1 ) and ( a2 ) . in this subsection we assume that we are in the framework of remark [ rem : parametrized ] , i.e. , we are dealing with a parameterized model @xmath179 . assume further that for each @xmath156 , each distribution from the family of conditional distributions @xmath91 has a density with respect to some @xmath75-finite dominating measure and denote this conditional density by @xmath499 , @xmath500 . assume also that there is a common support @xmath0 for these densities , and that for any @xmath501 and @xmath502 , the partial derivatives @xmath503 , @xmath504 , exist and are finite , almost surely . as before , the `` true '' value of the time - varying parameter at time moment @xmath20 is denoted by @xmath25 . under these assumptions , the _ conditional _ gradient vector @xmath505 and the random matrices @xmath506 , @xmath160 , with entries @xmath507 , \quad i , j=1,\dots , d,\ ] ] can be defined , almost surely . a possible gain function for the algorithm ( [ eq : algorithm_main ] ) is simply the conditional score of the model , i.e. , the gradient vector @xmath508 if @xmath506 is almost surely non - singular in point @xmath123 , then one might also consider @xmath509 we now outline some heuristic arguments why these choices are reasonable . take @xmath510 . it is not uncommon for the kullback - leibler divergence @xmath511 to be a quadratic form in the distance between the parameters @xmath385 and @xmath424 , i.e. , equal to a multiple of @xmath512 for some ( eventually random ) positive semi - definite matrix @xmath48 . actually , this is also in general true under some regularity conditions , at least locally , in a vicinity of the `` true '' @xmath385 . for example , suppose that we can interchange integration and differentiation and that @xmath48 does not depend on @xmath424 , then @xmath513 the score in principle depends on the past of the time series @xmath27 and the previous argument might only be valid for a certain subset of values @xmath27 in @xmath514 . this dependence could prevent ( a1 ) from holding . in such cases , using a gain of the form ( [ eq : gain_canonical_2 ] ) might be a good alternative since the matrix @xmath515 acts as an appropriate scaling factor . the dependence of the gain function on the past of the time series is in fact one of the main issues one has to deal with when checking ( a1 ) and ( a2 ) . on one hand , to ensure that the gain function has , on average , the right direction , as required by ( [ eq : g ] ) , the gain will often need to depend on previous observations . this might , however , affect either the range or the variance of the gain . gain functions , such as ( [ eq : gain_canonical_1 ] ) and ( [ eq : gain_canonical_2 ] ) , can be modified , or rescaled , to ensure that the respective conditional expectation @xmath516 verifies the assumptions of theorem [ theo : bound ] . one can for example truncate certain entries or factors in both @xmath138 and @xmath506 to ensure that the resulting @xmath516 meets the required assumptions . another possibility is to rescale , or directly truncate , the length of a given gain vector and consider , for example , one of the following gains @xmath517,\\ { \bar g}_k(\vartheta,\bm{x}_k ) & = g_k(\vartheta,\bm{x}_k ) \frac{\min\{s_k(\bm{x}_{k-1 } ) , \kappa \ } } { s_k(\bm{x}_{k-1})},\end{aligned}\ ] ] for @xmath113 an arbitrary gain function , @xmath518 and some functions @xmath519 . note that @xmath520 , @xmath521 and @xmath522 all preserve the direction of @xmath113 and have norm bounded by respectively @xmath523 , @xmath524 and the norm of @xmath113 , almost surely . the gain @xmath522 is a rescaling of @xmath113 for situations when the corresponding conditional gain @xmath182 is of the form @xmath525 , where @xmath246 has eigenvalues as prescribed by ( a1 ) . consequently the conditional rescaled gain is @xmath526 so that the largest eigenvalue of the matrix @xmath527 is almost surely upper bounded . as to the lower bound , in certain situations it will be possible to show that @xmath528\ge c\lambda_1 $ ] almost surely , for some @xmath529 and sufficiently large @xmath524 , by using the fact that @xmath530\ge\lambda_1 $ ] almost surely . this would establish condition ( a1 ) for the rescaled gain @xmath522 . since @xmath531 for all @xmath532 , then @xmath533\notag\\ & \quad\quad\quad \label{eq : scaled_gain } = { { \mathbb{e}}}\big [ \big(\frac{\min\big ( s(\bm{x}_{k-1 } ) , \kappa \big)}{s(\bm{x}_{k-1})}\big)^2 { { \mathbb{e}}}\big[\\|g_k - g_k\\|^2\big\|\bm{x}_{k-1 } \big ] \big ] \le { { \mathbb{e}}}\\|g_k - g_k\\|^2.\end{aligned}\ ] ] thus , if @xmath113 verifies ( a2 ) , then so does @xmath522 . another possible modification one might consider is to truncate the iterates of the our algorithm ( [ eq : algorithm_main ] ) . this might be motivated by practical considerations in the case where the parameter being tracked has some physical meaning and is bounded for that reason . the algorithm should be restricted as well . we can then consider an algorithm of the form @xmath534 where @xmath535 acts as a projection on a convex compact set @xmath536 : @xmath535 is an identity on @xmath537 and maps any point from @xmath538 to the closest point in @xmath537 . we provide concrete examples of gain functions later in section [ sec : examples ] . in section [ sec : variational_setups ] , we present some examples of different types of parameter variation such that our algorithm is capable of adequately tracking the time - varying parameter . it is clear and in fact explicit in ( [ eq : bounde ] ) and ( [ eq : boundas ] ) that the changes in the parameter have a non - negligible contribution to the accuracy of our tracking algorithm . this is reasonable since , if the parameter changes arbitrarily in - between observations , we should not expect it to the trackable " . we should then specify how the parameter is allowed to vary and , based on that assumption , pick an appropriate sequence @xmath207 which minimizes the general bounds in ( [ eq : bounde ] ) or ( [ eq : boundas ] ) . in this section , we specify different settings for the variation of the parameter to be tracked . these settings refer only to how the parameter is assumed to change and are unrelated to the actual model in question ; examples of specific models can be found in section [ sec : examples ] . to avoid overloaded notations , we use letters @xmath205 and @xmath539 for constants whose values are not important to us and which can be different in different expressions . we assume in this section that @xmath540 , @xmath541 , almost surely , for some unknown @xmath542 so that @xmath543 ( zero vector ) for all @xmath541 , almost surely , and we are actually in a parametric setup . in this case the second terms in both ( [ eq : bounde ] ) and ( [ eq : boundas ] ) obviously vanish . take then @xmath544 and for @xmath545 , @xmath546 $ ] , where @xmath547 $ ] is the whole part of @xmath548 . let @xmath549 such that @xmath550 . for any @xmath551 there is a large enough @xmath552 such that , for all @xmath553 , @xmath554 = c_\gamma \log n_0 \log\big[\frac{n+1}{n_0}\big ] \ge c \log n. \ ] ] moreover , it is easy to see that , under the conditions of theorem [ theo : bound2 ] , @xmath555 . thus , in both ( [ eq : bounde ] ) and ( [ eq : boundas ] ) the first term can be upper bounded by @xmath556 for any @xmath551 by taking sufficiently large @xmath552 . next note that @xmath557^{p/2}\le c ( n^{-1/2 } \log n)^p$ ] , we conclude that , for a sufficiently large @xmath552 , we can rewrite ( [ eq : bounde ] ) and ( [ eq : boundas ] ) as respectively , @xmath558^p \le c , \quad p\ge1.\ ] ] if we let @xmath559 , this is almost ( up to a log factor ) parametric convergence rate , the @xmath560-factor in the rate can not be avoided and is in some sense a price for the recursiveness of the algorithm . if we are in the situation of theorem [ theo : bound2 ] , then by taking @xmath561 ( where @xmath562 is some small fixed number ) and by using markov s inequality and the second bound in the previous display , we derive that @xmath563 in view of the borel - cantelli lemma , it follows that @xmath564 as @xmath565 with probability 1 at a rate @xmath566 . the particular setup presented in this section , where the parameter is fixed , might seem out of place since we are mainly concerned with tracking time - changing parameters . we would like to point out that recursive algorithms in parametric situation can also be useful ; for example , the classical robbins - monro and kiefer - wolfowitz algorithms deal with the parametric case . recursive procedures often produce estimates in a fast , straightforward fashion . this is an advantage especially over offline " estimators obtained , say , as solutions to a certain system , which require iterative likelihood or least squares optimization or are obtained via other indirect methods , a situation which is common when dealing with markov models ( cf . section [ sec : examples : ard ] . ) suppose now that the parameter we want to track is stabilizing . this situation might arise if the expectation of the sequence of values that the parameter takes is converging to some limiting value . it could also be the case that the data is being sampled with increasing frequency from an underlying , continuous time process which depends on a parameter varying continuously ; in this case , the parameter varies less because it has less time to change . regardless , we assume that @xmath567 verifies @xmath568 for @xmath569 and some positive sequence @xmath570 . assume that @xmath571 for some @xmath572 and @xmath573 . consider first the case @xmath574 . in this case , the variation of the parameter vanishes so quickly that we are essentially in the setup of the previous section , i.e. , as if the parameter is constant . indeed , take @xmath575 and @xmath576 as in the previous section . the first and second terms in both ( [ eq : bounde ] ) and ( [ eq : boundas ] ) can be bounded in the same way as in the previous section . using the relations between norms from remark [ rem : norms ] , we upper bound the third term in ( [ eq : bounde ] ) by a multiple of @xmath577 using the hlder inequality , we upper bound the third term in ( [ eq : boundas ] ) by a multiple of @xmath578^p \le c n^{-(\beta-1)p } \le c n^{-p/2}.\end{aligned}\ ] ] clearly , in both ( [ eq : bounde ] ) and ( [ eq : boundas ] ) the third term is of a smaller order than the second term . thus , the relations ( [ constant_parameter ] ) remain valid for the case @xmath574 . consider now the case @xmath579 . let @xmath580 , @xmath581 . by using the elementary inequality @xmath582 for @xmath583 and @xmath584 , we obtain that for any @xmath551 there is a sufficiently large constant @xmath585 such that @xmath586\\ & \ge\frac{c_\gamma ( \log n_0)^{1/3}}{1 - 2\beta/3 } \big[n^{1 - 2\beta/3 } -n^{1 - 2\beta/3 } \big(1-n^{2\beta/3 - 1 } ( \log n)^{2/3}(1 - 2\beta/3)\big)\big]\\ & = c_\gamma ( \log n_0)^{1/3 } ( \log n)^{2/3 } \ge c \log n \ ] ] for sufficiently large @xmath587 , i.e. , @xmath588 . this yields the same upper bound for the first term in ( [ eq : bounde ] ) and ( [ eq : boundas ] ) as for the static parameter , namely , @xmath556 for any @xmath551 by taking sufficiently large @xmath552 . let us bound now the second term in ( [ eq : bounde ] ) and ( [ eq : boundas ] ) : @xmath589 for @xmath590 . for sufficiently large @xmath587 ( i.e. , @xmath591 ) the third terms in ( [ eq : bounde ] ) and ( [ eq : boundas ] ) are bounded similarly to ( [ eq : bound00 ] ) and ( [ eq : bound01 ] ) by , respectively , @xmath592 and @xmath593 finally we obtain that for @xmath594 and sufficiently large constant @xmath552 in the algorithm step @xmath580 , ( [ eq : bounde ] ) and ( [ eq : boundas ] ) can be rewritten as respectively @xmath595^p \le c,\ ] ] where @xmath596 is the burn - in period of the algorithm . if we choose @xmath597 and @xmath598 , @xmath599 , @xmath600 , @xmath601 in case @xmath594 , then we get the following bound of the convergence rate : for sufficiently large @xmath587 and sufficiently large constant @xmath552 @xmath602 thus , the choices @xmath603 , @xmath604 , @xmath605 are optimal in the sense of the minimum of the right - hand side of the above inequality . much in the same way as for ( [ eq : almost_sure ] ) , we can establish that for any @xmath562 , @xmath606 with probability 1 . finally , consider the case @xmath607 , i.e. , we assume the following weak requirement : @xmath608 , @xmath309 , for some uniform constant @xmath539 . take @xmath609 , @xmath610 for some @xmath611 , @xmath612 . then theorem [ theo : bound ] implies that @xmath613 we thus have that the algorithm will track down the parameter in the proximity of size @xmath614 , which we can try to minimize by choosing appropriate constants @xmath615 and @xmath616 . we consider now a different setup where we assume that the parameter is changing , on average , like a lipschitz function . in this setup we let the time series ( [ model ] ) be sampled from a continuous time process @xmath617 , @xmath618 $ ] , which we observe with frequency @xmath587 . this means that we deal with a triangular sequence of models , i.e. , for each @xmath619 we have a different model , namely , @xmath620 where the parameter @xmath621 verifies , for some @xmath44 , @xmath622 , @xmath623 , @xmath624 assume for example that @xmath625 , where @xmath626\}$ ] for some @xmath623 and @xmath187 , a space of vector valued lipschitz functions . let @xmath627 ( constant in @xmath110 ) for @xmath628 , and @xmath629 for @xmath630 . note that for @xmath631 as @xmath632 for any @xmath623 . we have @xmath633 so that once again the first term in ( [ eq : bounde ] ) and ( [ eq : boundas ] ) can be upper bounded by @xmath556 for any @xmath551 by taking sufficiently large @xmath552 . as to the second term , we evaluate @xmath634 from our assumption on the variation of the parameter , we have @xmath635^p.\end{aligned}\ ] ] combining these three bounds , we get that ( [ eq : bounde ] ) ( we also need the relations between norms from remark [ rem : norms ] ) and ( [ eq : boundas ] ) imply @xmath636^p.\end{aligned}\ ] ] if we consider step sizes of the form @xmath637 , the above proposed choices of @xmath638 and @xmath639 are optimal in the sense of tracking error minimum . note that the obtained convergence rate ( the asymptotic regime : the observation frequency @xmath559 ) coincides , up to a log factor , with the minimax rate of convergence in the problem of estimating nonparametric regression function over lipschitz functional class @xmath640 . in this section we present some examples of particular models to which our algorithm may be applied . we start with two toy examples and present thereafter some more involved examples . the toy examples illustrate the type of results that can be obtained from our main result and its extensions , how a gain function can be picked and modified , and how conditions ( a1 ) and ( a2 ) are checked . suppose we are monitoring @xmath619 independent poisson processes on @xmath372 $ ] with unknown intensity function @xmath641 . this is equivalent to observing @xmath642 , a poisson process with intensity @xmath643 , @xmath644 . we would like to track the intensity function @xmath641 which is assumed to be upper bounded by @xmath645 . assume that we observe the process with frequency @xmath587 , in that our observations are @xmath646 , so that for each @xmath619 we have a markov model @xmath647 where @xmath648 represents a poisson law with parameter @xmath649 . from now on , we will skip the dependence on @xmath587 for notational simplicity : write @xmath8 instead of @xmath650 , @xmath107 instead of @xmath651 etc . introduce the conditional , shifted poisson mass function given by @xmath652 the moving parameter is given by @xmath653 , @xmath654 , which is the average of function @xmath655 over the interval @xmath656 $ ] . assume that @xmath655 is continuous , then @xmath657 for @xmath587 large enough . consider now the gain function @xmath113 of the type ( [ eq : gain_canonical_2 ] ) for the algorithm ( [ eq : algorithm_main ] ) so that @xmath658 = - ( \hat{\theta}_k-\theta_k ) , \notag\end{aligned}\ ] ] it follows that @xmath659 since @xmath660}\lambda(t)\le l$ ] . we thus conclude that the gain function ( [ eq : gain_poisson ] ) satisfies both ( a1 ) and ( a2 ) . this gain function can now be used for the three setups outlined in section [ sec : variational_setups ] and we can attain the rates indicated there . for a constant intensity function @xmath661 , @xmath662 , the algorithm will simply estimate the parameter of the underlying homogeneous poisson process @xmath385 , because we matched the sampling frequency @xmath663 with the sample size @xmath587 . if we had sampled the process with frequency , say , @xmath664 , then @xmath665 and the algorithm would track @xmath666 and not @xmath385 . the tracking sequence would then have to be rescaled by a factor @xmath667 to obtain a tracking sequence for @xmath385 itself . in the setup where we assume that the parameter is stabilizing , take @xmath668 so that @xmath669 is the mean number of events per time unit @xmath656 $ ] . note that @xmath670 and the average number of events per time unit will stabilize in time if , for example , @xmath671 as @xmath672 . the algorithm will then track the mean number of events per time unit . we can also assume that the intensity function @xmath641 belongs to @xmath673 for some @xmath623 and @xmath187 . let @xmath674 , @xmath675 . it follows that @xmath676 the tracking sequence based on the gain ( [ eq : gain_poisson ] ) will then track the sequence @xmath677 , @xmath675 ( as well as @xmath107 ) with the asymptotics seen in section [ sec : variational_setups ] ( cf . remark [ rem : close_parameter ] ) . assume that we observe , with fixed frequency @xmath619 , a process @xmath678 , @xmath679 , taking values on @xmath680 , @xmath681 . the observations available up to time moment @xmath682 is a random vector @xmath683 , with @xmath684 . we again skip the dependence on @xmath587 , although all the quantities below do depend on @xmath587 . the increments @xmath685 are assumed to be conditionally gaussian in the sense that given the past of the process , each increment has a multivariate normal distribution : @xmath686 the dependence on the past in the model comes from the fact that both the mean and the covariance processes of the above conditional distributions are predictable , i.e. , @xmath25 and @xmath687 , @xmath20 , with respect to the filtration @xmath106 . if the covariance structure of the process is known , we can use the gain ( [ eq : gain_canonical_1 ] ) which verifies @xmath688 for this gain , we assume that almost surely @xmath689 for some positive @xmath690 . we then obtain that @xmath691 and assumptions ( a1 ) and ( a2 ) are thus met for the gain from ( [ eq : gain_gaussian1 ] ) . now suppose that the covariance matrix of the process is unknown or difficult to invert . then we can use the gain ( [ eq : gain_canonical_2 ] ) , so that @xmath692 clearly , assumptions ( a1 ) and ( a2 ) are again met for the gain from ( [ eq : gain_gaussian2 ] ) if @xmath693 for some @xmath203 , @xmath160 , almost surely . the results of section [ sec : variational_setups ] can be applied to the algorithm based on the gain functions presented above for all three considered asymptotic regimes : constant parameter process , stabilizing ( on average ) process and lipschitz on average . although designed for different frameworks , it is interesting to compare the above resulting tracking algorithm with the famous _ kalman filter_. for simplicity , consider the one dimensional situation . suppose we observe @xmath694 where the parameter of interest @xmath107 , evolves according to @xmath695 with @xmath696 . at each step , the initial state and the noises @xmath697 are assumed to be mutually independent . one can show ( by combining both prediction and update steps ) that the kalman filter in this case reduces to @xmath698 we also derive the exact expression for the mean squared error of the algorithm : @xmath699 coming back to our framework , suppose we have observations ( [ model_kalman ] ) with predictable process @xmath403 such that @xmath700 , @xmath156 ; cf . section [ sec : variational_setups : stabilizing ] . then the kalman filter ( [ kalman_algorithm ] ) coincides with our tracking algorithm with the gain ( [ eq : gain_gaussian1 ] ) and a particular choice of the step sequence @xmath207 given by ( [ kalman_step ] ) . one should keep in mind that the two frameworks are different , but it would still be interesting to compare the convergence rates for some particular settings for stabilizing the parameter @xmath107 . for example , one can consider @xmath701 , @xmath702 , as in section [ sec : variational_setups : stabilizing ] . the above kalman filter setting has more structure and we expect therefore that the rate in this case ( which is of order @xmath703 , with @xmath704 defined by ( [ kalman_step ] ) ) should be faster than the rate @xmath705 obtained in section [ sec : variational_setups : stabilizing ] for our general framework . we were however unable to solve the recursive rational difference equation ( [ kalman_step ] ) for @xmath701 . note that the trivial case @xmath706 leads to the situation of a constant parameter @xmath707 and the sample mean @xmath708 as an estimator for that parameter . consider the following arch(@xmath523 ) model with drifting parameter @xmath709 where @xmath710 almost surely , @xmath711 is predictable and @xmath712 is a martingale difference noise with respect to the filtration @xmath713 , @xmath714=\sigma_k^2 $ ] for some known @xmath715 , @xmath7 . without loss of generality assume @xmath716 . assume further that @xmath717 and @xmath718 \le \rho$ ] , @xmath156 , for some @xmath719 . consider the gain function @xmath720 for some truncating constant @xmath721 . since @xmath722 and @xmath723 = 1+\theta_k x_{k-1}^2 $ ] , @xmath724 = - \min(x_{k-1}^2 , t)(\vartheta-\theta_k).\ ] ] we have that @xmath725 almost surely . besides , @xmath726 & = { { \mathbb{e}}}\big[\min\{(1+\theta_{k-1}x_{k-2}^2)\epsilon_{k-1}^2 , t\ } \big\|\bm{x}_{k-2}\big ] \\ & \ge{{\mathbb{e}}}\big[\min(\epsilon_{k-1}^2 , t)\|\bm{x}_{k-2}\big].\end{aligned}\ ] ] using the hlder inequality and the facts that @xmath727 and @xmath728 , it is straightforward to check that @xmath729 & = { { \mathbb{e}}}\big [ t+\epsilon_{k-1}^2 - \|\epsilon_{k-1}^2-t\| \big\|\bm{x}_{k-2}\big ] \\ & \ge t+1- \big({{\mathbb{e}}}[(\epsilon_{k-1}^2-t)^2\|\bm{x}_{k-2}]\big)^{1/2 } \\ & \ge t+1 + 2 t - \big({{\mathbb{e}}}[\epsilon_{k-1}^4 \|\bm{x}_{k-2}]+t^2\big)^{1/2 } \\ & \ge1 + 2 t - \big({{\mathbb{e}}}[\epsilon_{k-1}^4 \|\bm{x}_{k-2}]\big)^{1/2 } \ge 1,\end{aligned}\ ] ] as long as @xmath730 /4 $ ] , @xmath156 . for example , we can take @xmath731 . we conclude that ( a1 ) holds for the gain ( [ eq : gain_arch_1 ] ) . to ensure ( a2 ) , we evaluate @xmath732 \\ & \le 3{{\mathbb{e}}}\big[\big(\frac{\min(x_{k-1}^2 , t)}{x_{k-1}^2}\big)^2 \big ( ( 2 + 2\theta_k^2x_{k-1}^4 ) \epsilon_k^4 + 1+\theta_k^2 x_{k-1}^4\big ) \big ] \\ & \le 9 + 3 { { \mathbb{e}}}\big[\big(\frac{\min(x_{k-1}^2 , t)}{x_{k-1}^2}\big)^2 ( 2c_\theta \rho + c_\theta ) x_{k-1}^4\big ] \\ & \le 9 + 3c_\theta t^2(2\rho+1 ) . \ ] ] in this section we use the notation @xmath733 for the vector of the @xmath395 consecutive observations ending with @xmath8 . consider an autoregressive model with @xmath395 time varying parameters : @xmath734 where @xmath735 is @xmath736-measurable , @xmath737 is a martingale difference noise with respect to the filtration @xmath713 such that @xmath738 , @xmath156 , starting random vector @xmath739 is given and such that @xmath740 , for some @xmath741 . for @xmath742 , associate with the ar(d ) model its polynomial @xmath743 it is well know that an ar(d ) model with autoregressive parameters @xmath424 is stationary if , and only if , the ( complex ) zeros of the polynomial @xmath744 are outside the unit circle . this motivates the definition of the parameter sets @xmath745 for some @xmath746 : @xmath747 cf . @xcite who also showed that the following embeddings hold : @xmath748 where @xmath749 is a uniform ball around zero in @xmath18 with radius @xmath750 . this gives some feeling about the size of the parameter set @xmath745 and implies in particular that the set @xmath745 is non - empty and bounded for all @xmath751 . the ar(d ) model can also be described by the following inhomogeneous difference equation @xmath752 where @xmath753 and , for any @xmath754 , @xmath755 is the square matrix of order @xmath395 @xmath756 this matrix is usually called the _ companion matrix _ to the autoregressive polynomial @xmath757 ; it is also sometimes called the _ state transition matrix_. one can show that the eigenvalues of @xmath755 are exactly the reciprocals of the zeros of @xmath744 . this means that the absolute values of the eigenvalues of @xmath755 for @xmath758 are all at most @xmath759 . this in turn implies that for any sequence of vectors @xmath760 , the pair of sequences @xmath761 forms a so called _ exponentially stable _ pair ( cf . @xcite ) . among other things , this gives us that so long as the @xmath158-th moments of both the initial @xmath739 and the noise terms @xmath375 are bounded , then the @xmath158-th moments of all @xmath8 , @xmath7 , will be bounded as well ( cf . proposition 10 of @xcite ) . in @xcite the model ( [ eq : ar_p_model ] ) is considered with nonrandom but time varying @xmath762 for some smooth function @xmath763 , @xmath764 $ ] . one has a triangular array of models and the studied asymptotics is in sampling frequency @xmath559 ; cf . section [ sec : variational_setups : lipschitz ] . the considered gain function is an appropriately rescaled version of the gain from remark [ gain_autoregression ] , namely , @xmath765 for an appropriately chosen @xmath766 depending on the observation frequency @xmath587 . one should mind the difference in indexing in our algorithm ( [ eq : algorithm_main ] ) and algorithm ( 3 ) from @xcite . this is not an issue since we can make the correspondence between the algorithms exact by treating @xmath767 as an estimate of @xmath107 rather than of @xmath768 , the error can be absorbed into the third term of the right hand side of ( [ eq : bounde ] ) . although assumption ( a2 ) is trivially satisfied , our general theorem [ theo : bound ] can not be applied for @xmath769 because assumption ( a1 ) does not hold . indeed , @xmath770 and the matrix @xmath246 is of the form @xmath771 for some @xmath772 and column vectors @xmath39 . but the matrix @xmath773 has @xmath774 zero eigenvalues and one eigenvalue @xmath775 , so that always @xmath776 and thus ( [ eq : lambda ] ) does not hold . [ rem_mouilines ] on the other hand , the authors of @xcite do manage to establish a convergence results for the gain function ( [ gain_moulines ] ) . it is instructive to understand where the difference in the two approaches is . careful inspection of the proofs in @xcite reveals that the analogue of the lower bound ( [ eq : lambda ] ) , the persistence of excitation condition , is established in lemma 17 ( p. 2627 of @xcite ) . the basic difference is that the quantity to bound from below in our case is the conditional expectation of the smallest eigenvalue of the matrix @xmath246 , whereas in @xcite ) it is the smallest eigenvalue of the conditional expectation of the matrix @xmath246 . for the gain function ( [ gain_moulines ] ) , the lower bound for the former is zero ( as is demonstrated above ) and it is positive for the latter ( cf . lemma 17 of @xcite ) . a way to fix this would be to establish a version of the general theorem , where ( [ different_bound ] ) is assumed instead of ( [ eq : lambda ] ) , see also remark [ rem_version_theorem ] . we do not consider this here . consider the case @xmath254 and gain ( [ gain_moulines ] ) for which theorem [ theo : bound ] can be applied . assume that @xmath777 is bounded , @xmath778 and @xmath779 , @xmath156 . we have @xmath780 where @xmath781 . besides , if @xmath782 , then @xmath783 , and if @xmath784 , then @xmath785 condition ( a1 ) is fulfilled . as to condition ( a2 ) , @xmath786 \le \sigma^2 \max_{u>0 } \big\{\frac{u}{(1+\mu u)^2}\big\ } = \frac{\sigma^2}{4\mu}.\end{aligned}\ ] ] both ( a1 ) and ( a2 ) are thus satisfied . interestingly , there is no issue of stability in this case : we do not have to assume that @xmath787 , @xmath90 , almost surely . just almost sure boundedness @xmath788 , @xmath90 , for a constant @xmath229 , is sufficient . consider now another gain function for the case @xmath254 . this time we assume that @xmath789=\mathbb{e } [ \xi_k^3\|\bm{x}_{k-1}]=0 $ ] , @xmath790=\sigma^2>0 $ ] , and @xmath791 = c \sigma^4 $ ] , @xmath156 , for some constant @xmath792 . the proposed gain and the corresponding average gain are as follows : @xmath793 with some @xmath794 . note that this is a rescaled gain function of type @xmath795 from section [ sec : gains ] . clearly , @xmath796 and , according to lemma [ lemma : truncated_conditional ] , @xmath797 = \mathbb{e}\big[\min\{x_{k-1}^2,t\}\|\bm{x}_{k-2}\big ] \ge \frac{(5-c)\sigma^2}{4},\ ] ] so that ( a1 ) holds . assumption ( a2 ) also holds since @xmath798 \le \max\{t^2,1\ } \sigma^2.\end{aligned}\ ] ] finally consider a version of general ar(d ) model . we will only outline the main steps , leaving out the details . assume that the noise terms @xmath375 in ( [ eq : ar_p_model ] ) form a gaussian white noise sequence with mean zero and variance @xmath799 and that the parameter process @xmath800 is constant within the batch of @xmath395 consecutive observations . for a @xmath801-dimensional vector @xmath802 , introduce the toeplitz matrix @xmath803 associated with that vector whose entries are @xmath804 , @xmath805 , so that this matrix has constant ( from left to right ) diagonals . thus , @xmath806 is the column vector formed by starting at the top right element of @xmath807 , going backwards along the top row of @xmath807 and then down the left column of @xmath807 . denote @xmath808 and introduce @xmath809 and @xmath810 , the toeplitz matrices created from the vectors @xmath811 and @xmath812 respectively . under the imposed assumptions , we can rewrite the model ( [ eq : ar_p_model ] ) as follows : @xmath813 the matrix @xmath814 is upper triangular with a diagonal consisting of ones , whence invertible . from this point on , we regard vector @xmath815 , @xmath156 , as an observation at time moment @xmath110 so that we can specify our observation model in terms of conditional distribution of @xmath816 given @xmath817 : @xmath818 where @xmath711 is a predictable process with respect to the filtration @xmath819 . notice that the observation process is of a markov structure . even if the normality of the noise is assumed in the model ( [ eq : ar_p_model ] ) , the models ( [ eq : ar_p_model ] ) and ( [ eq : ar_d_kernel ] ) still differ since in general the parameter process @xmath711 varies also within the batches of @xmath395 observations in the model ( [ eq : ar_p_model ] ) . however , this is not an issue . indeed , even though the parameter is allowed to vary within each batch of @xmath395 observation , we still can use the gain function ( which we derive below ) as if the parameter process is constant within the batches and establish an upper bound of type ( [ theo : bound ] ) for the quality of such a procedure . the error that is made by pretending that the parameter is constant within the batches can be absorbed into the third term of the right hand side of ( [ theo : bound ] ) . in this case we propose a gain of the type ( [ eq : gain_canonical_1 ] ) : @xmath820 where @xmath821 is the conditional density of ( [ eq : ar_d_kernel ] ) . thus , the tracking sequence is updated with batches of @xmath395 observations from the autoregressive process . below , to ease the notation , we will often write @xmath150 and @xmath822 instead of @xmath815 and @xmath823 , respectively . as explained in section [ sec : gains ] , the corresponding average gain @xmath824 can be found as minus the gradient of the kullback - leibler divergence between the two conditional distributions with two different parameters . this observation is particularly useful if we are able to write this kullback - leibler divergence as an appropriate quadratic form . the kullback - leibler divergence between two @xmath395-dimensional multivariate normal distributions @xmath825 and @xmath826 is given by @xmath827 let @xmath828 , i.e. , @xmath829 ( not to be confused with the vectors @xmath107 , @xmath20 ) and @xmath830 . according to ( [ eq : ar_d_kernel ] ) , @xmath831 and @xmath832 . now we compute @xmath833 . let @xmath834 be the toeplitz matrix associated with the vector @xmath835 where @xmath523 is in the @xmath836-th position . matrix @xmath837 has ones above the main diagonal and zeros elsewhere and it is sometimes called _ upper shift matrix_. for @xmath838 , the powers @xmath839 are the toeplitz matrices associated with the vectors @xmath840 where @xmath523 occupies the @xmath841-th position , @xmath842 , the zero matrix of order @xmath395 , and @xmath843 should be read as @xmath52 , the identity matrix of order @xmath395 . it follows that @xmath844 , so that @xmath845 and @xmath846 for all @xmath754 , the matrices @xmath847 have all eigenvalues equal to one ( so do their inverses ) , hence @xmath848 and we conclude that the logarithm in ( [ eq : kl_normals ] ) is zero . also , using basic properties properties of the trace and the representation for @xmath849 derived above , @xmath850- d = \operatorname{tr}\big[\big(\bm{a}^{-1}(\vartheta)\bm{a}^{-t}(\vartheta)\big)^{-1}\big(\bm{a}^{-1}(\theta)\bm{a}^{-t}(\theta)\big ) \big]- d\\ & = \operatorname{tr}\big[\bm{a}^t(\vartheta)\bm{a}(\vartheta)\bm{a}^{-1}(\theta)\bm{a}^{-t}(\theta)\big]- d = \operatorname{tr}\big[\big(\bm{a}(\vartheta)\bm{a}^{-1}(\theta)\big)^t\bm{a}(\vartheta)\bm{a}^{-1}(\theta)\big]-d\\ & = 2\sum_{i=1}^d \operatorname{tr}\big[\bm{s}^i\bm{a}^{-1}(\theta)\big](\theta_i-\vartheta_i)+ \sum_{i=1}^d\sum_{j=1}^d \operatorname{tr}\big[\bm{a}^{-t}(\theta)(\bm{s}^i)^t\bm{s}^j\bm{a}^{-1}(\theta ) \big ] ( \theta_i-\vartheta_i)(\theta_j-\vartheta_j).\end{aligned}\ ] ] since the inverse of an upper - triangular matrix is upper - triangular , @xmath851=0 $ ] , for all @xmath504 and @xmath852 . for any @xmath853-matrix @xmath48 , denote by @xmath854 the column vector containing the @xmath855 entries of @xmath48 in any ( fixed ) order . let @xmath856 , @xmath857 , and note that @xmath858 is always a zero vector . note that @xmath859 = v_i^t(\theta)v_j(\theta)$ ] , @xmath860 . we conclude that the previous display can be written as @xmath861- d = ( \vartheta-\theta)^t\big[v_1(\theta ) v_2(\theta ) \ldots v_d(\theta)\big]^t \big[v_1(\theta ) v_2(\theta ) \ldots v_d(\theta)\big ] ( \vartheta-\theta),\ ] ] where the matrices on the right are comprised by columns . consider now the quadratic form in the kullback - leibler divergence ( [ eq : kl_normals ] ) . for any @xmath862 , @xmath863 since @xmath864 , we have @xmath865 which , together with the representation for @xmath866 derived above , imply @xmath867 where @xmath868 , @xmath452 ; notice also that @xmath869 . then @xmath870(\vartheta-\theta).\ ] ] summarizing , we obtained that @xmath871 with @xmath872^t \big[v_1(\theta ) v_2(\theta ) \ldots v_d(\theta)\big ] \notag\\ \label{matrix_m } & \quad + \sigma^{-2 } \big[\bm{c}_1(\theta)y \ldots \bm{c}_d(\theta)y \big]^t \big[\bm{c}_1(\theta)y \ldots \bm{c}_d(\theta)y\big].\end{aligned}\ ] ] according to ( [ kl - gain ] ) , we can derive the expression for the average gain : @xmath873 where @xmath48 is given by ( [ matrix_m ] ) . note that the matrix @xmath48 does not depend on @xmath424 and is clearly positive semidefinite . we evaluate now its eigenvalues . in the representation ( [ matrix_m ] ) , the first matrix in the sum is positive semi - definite but has at least one zero eigenvalue . it is also clear that the entries of this matrix are polynomials of the coordinates of @xmath385 , so that , if @xmath22 for a bounded set @xmath15 , then the largest eigenvalue of this matrix is upper bounded , uniformly over @xmath15 , by some constant , say , @xmath874 . as to the second ( also positive semidefinite ) matrix in the sum of matrices from ( [ matrix_m ] ) , note that @xmath875 the entries of the matrices @xmath876 , @xmath504 , are polynomials in @xmath877 which are bounded uniformly over a bounded set @xmath15 . recall also that the trace of a matrix is equal to the sum of its eigenvalues . we conclude that @xmath878 uniformly in @xmath879 for any bounded @xmath880 . to derive a lower bound on the smallest eigenvalue of the matrix @xmath881 , note that this matrix can be rewritten in the form @xmath882 + \left [ \begin{array}{ccc\|c } c_{1,1}(\theta ) & \cdots & c_{1,d-1}(\theta ) & c_{1,d}(\theta)\\ \vdots & \ddots & \vdots & \vdots\\ c_{d-1,1}(\theta ) & \cdots & c_{d-1,d-1}(\theta ) & c_{d-1,d}(\theta)\\ \hline c_{d,1}(\theta ) & \cdots & c_{d , d-1}(\theta ) & 0 \end{array } \right]\ ] ] for @xmath883 and @xmath884 , where we swapped the @xmath885-th entries of the matrices in the sum from ( [ matrix_m ] ) . we used also that @xmath886 and @xmath887 . note that the top left matrices in the block matrices above are gram matrices and therefore positive semidefinite . the matrix @xmath888_{i , j=1,\dots , d-1}$ ] is the gram matrix associated with the vectors @xmath889 . since @xmath890 is a triangular matrix with @xmath523 s in its main diagonal , it follows that these vectors are linearly independent . hence the associated gramian is actually positive definite for each @xmath385 . the determinant of this gramian is a polynomial in the entries of the matrix which in turn are polynomials in @xmath877 . if @xmath385 lies in a compact set @xmath15 , the infimum of the determinant of this matrix over @xmath879 must be lower bounded by some positive constant , say , @xmath891 . using the same reasoning , we conclude that its determinant is upper bounded by some constant @xmath892 . a lower bound on the smallest eigenvalue can then be obtained by noting that for any positive definite matrix @xmath48 of order @xmath395 , @xmath893 we conclude that the smallest eigenvalue of the block matrix on the left is at least @xmath894 . the block matrix on the right is clearly positive semidefinite . we conclude that the smallest eigenvalue of the matrix @xmath895 by using weyl s monotonicity theorem , see for example @xcite . this means that @xmath896 for all @xmath897 . swapping back the @xmath885-th entries of the matrices in the sum from ( [ matrix_m ] ) , we see that there must exist an @xmath822 such that @xmath897 and for which the smallest eigenvalue of the second matrix in the sum from ( [ matrix_m ] ) is bounded from below by @xmath898 for some @xmath899 . by using renormalization arguments , we conclude that @xmath900 for some @xmath901 . condition ( a2 ) is not difficult to check . the gain ( [ eq : gain_ar_d ] ) can be written in the following form @xmath902 where @xmath903 represents the jacobian operator . to verify ( a2 ) , it suffices to check that the expectation of the norm of @xmath904 is bounded . we omit the details but it is clear from the expression derived above that the norm of the gain function squared is a polynomial of degree four in the coordinates of @xmath905 . we have already mentioned that if the initial values for the autoregressive process and the noise terms have uniformly bounded second moments , then this transfers to the each observation @xmath8 , provided that the sequence of parameters of the model , @xmath107 , lives in the parameter set @xmath745 for some @xmath759 . for the matrix @xmath48 defined by ( [ matrix_m ] ) , we established above that almost surely @xmath906 uniformly in @xmath879 for any bounded @xmath880 . we can get rid of the dependence on @xmath907 ( recall that @xmath908 ) by using the rescaled gain @xmath909 defined in section [ sec : gains ] : @xmath910 then @xmath911 , where @xmath912 and @xmath48 is defined by ( [ matrix_m ] ) . the argument in shows that ( a2 ) still holds for this rescaled gain . next , @xmath913 almost surely by construction . thus to establish ( a1 ) , we need to verify that @xmath914 \ge c$ ] . one can then proceed as in ( [ m_bound_1 ] ) ( and lemma [ lemma : truncated_conditional ] ) to show that for an appropriately large @xmath524 , @xmath915 > c$ ] , we omit this derivation . one could drop the requirement for the errors to be gaussian and still use the same gain @xmath904 . we expect the same results to hold , under appropriate moment assumptions . instead of using the kullback - leibler representation , one has to work with quantities @xmath904 and @xmath824 directly and assure the validity of ( a1 ) and ( a2 ) based on moment assumptions on the error terms ( and possibly the initial conditions ) as we did in the one dimensional case . first suppose that @xmath195 for some symmetric positive definite matrix @xmath48 such that @xmath916 . then @xmath917 and therefore @xmath918 and @xmath919 now we prove the converse assertion . suppose @xmath39 and @xmath920 for some @xmath921 such that @xmath202 and that @xmath922 . let @xmath923 be the linear space spanned by @xmath389 and @xmath924 . first consider the case @xmath925 , i.e. , @xmath926 for some @xmath927 . then @xmath928 so that @xmath929 . thus @xmath930 with symmetric and positive @xmath931 so that @xmath932 . let @xmath941 be chosen in such a way that @xmath942 ( which is always possible . ) now , we change the basis of @xmath936 as follows : @xmath943 we thus rotate the basis @xmath935 by the angle @xmath385 . in these new basis we have @xmath944 let @xmath952 be the orthonormal basis of @xmath953 , so that @xmath954 is an orthonormal basis of @xmath18 . take @xmath955 \quad \hbox { with } \quad \bm{d}=\left[\begin{array}{cc } \alpha_y/\alpha_x & 0\\ 0 & \beta_y/\beta_x \end{array}\right]\ ] ] where the @xmath51 s indicate null matrices of the appropriate dimensions . we then have @xmath956 in the basis @xmath957 and @xmath958 . we can finally obtain @xmath48 by using the orthogonal matrix @xmath959 to change the basis @xmath960 to the canonical basis of @xmath18 as @xmath961 . clearly , @xmath48 has the same eigenvalues as @xmath962 and is symmetric . for the sake of brevity , we use the notations @xmath25 , @xmath963 and @xmath240 , @xmath20 . recall that @xmath15 is compact and @xmath964 . by iterating ( [ eq : algorithm_main ] ) , it is easy to see that @xmath965 for each @xmath160 . first assume @xmath966 , for some @xmath967 to be chosen later . by ( [ bound_g_k ] ) , we obtain @xmath968 , which implies , in view of ( [ eq : algorithm_main ] ) and the fact that @xmath117 , @xmath969 next , consider the case @xmath970 , which , together with ( [ c_theta ] ) , implies that @xmath971 . recall that , in view of ( a1 ) , @xmath246 is a symmetric positive definite matrix such that @xmath972 almost surely . therefore we obtain that , almost surely , @xmath973 and , by the cauchy - schwarz inequality , @xmath974 by using the last two relation , ( [ c_theta ] ) , ( [ eq : d ] ) , ( [ eq : g2 ] ) and ( [ eq : algorithm_main ] ) , we evaluate @xmath975 : @xmath976 + \gamma^2_k { { \mathbb{e}}}\\|g_k\\|^2\\ \le & { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2 - 2 \gamma_k { { \mathbb{e}}}\big(\hat{\theta}_k^t \bm{m}_k ( \hat{\theta}_k - \theta_k)\big ) + \gamma_k^2 \bar{c}_g\\ \le & { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2 - 2 \gamma_k \big[\lambda_1 { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2 - { { \mathbb{e}}}\big(\hat{\theta}_k^t\bm{m}_k \theta_k ) \big]+\gamma_k^2 \bar{c}_g(kc_\theta)^{-1 } { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2\\ \le & { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2 - 2 \gamma_k \big[\lambda_1 { { \mathbb{e}}}\\|\hat{\theta}_k\\|^2 - \lambda_2 { { \mathbb{e}}}\big(\\|\hat{\theta}_k\\| \\|\theta_k\\|\big)\big ] + \gamma_k^2 c_3{{\mathbb{e}}}\\|\hat{\theta}_k\\|^2,\end{aligned}\ ] ] with @xmath977 . since @xmath978 , it follows that @xmath979 . using this , we proceed by bounding the previous display as follows : @xmath980 for sufficiently large @xmath981 and sufficiently small @xmath207 . thus , for sufficiently large @xmath981 and sufficiently small @xmath207 , @xmath982 with @xmath983 as defined above . [ lemma : eig ] let @xmath48 be a symmetric positive definite matrix of order @xmath395 , @xmath569 and a constant @xmath612 be such that @xmath984 . then @xmath985 and @xmath986 besides , @xmath987 for some constant @xmath988 . let @xmath989 s be the eigenvalues of @xmath48 , so that the matrix @xmath990 has eigenvalues @xmath991 , @xmath992 . since @xmath984 , then , for all @xmath992 , @xmath993 , implying @xmath994 , so that @xmath995 . the first two assertions follow . it remains to prove the last assertion . for @xmath996 , let @xmath997 and @xmath998 . according to theorem 5.6.18 from , @xmath999 recall that @xmath1000 and @xmath231 for any @xmath232 and @xmath233 . from the last relation it is easy to get the following bounds : @xmath1001 if @xmath235 , @xmath1002 if @xmath1003 ; @xmath1004 if @xmath235 , @xmath1005 if @xmath1003 . these bounds imply that @xmath1006 if @xmath235 and @xmath1007 if @xmath1003 . this completes the proof of the lemma . we prove this by induction in @xmath110 . for @xmath1013 we simply have @xmath1014 and the assertion holds true . assume that the equality holds for @xmath1015 and let us prove the result for @xmath1016 . we have @xmath1017 [ lemma : truncated_conditional ] consider an ar(1)-model with a measurable @xmath25 : @xmath1018 where @xmath789=\mathbb{e } [ \xi_k^3\|\bm{x}_{k-1}]=0 $ ] , @xmath790=\sigma^2>0 $ ] , and @xmath791=c \sigma^4 $ ] , @xmath156 , for some constant @xmath792 . then , for any @xmath1019 such that @xmath1020 , @xmath1021 \ge \frac{(5-c)\sigma^2}{4},\quad k \in \mathbb{n}.\ ] ] we compute @xmath1022 & = x_{k-1}^2 \theta_k^2 + 2x_{k-1}\theta_k \mathbb{e}[\xi_k\|\bm{x}_{k-1 } ] + \mathbb{e}[\xi_k^2\|\bm{x}_{k-1 } ] = x_{k-1}^2 \theta_k^2+\sigma^2,\\ \mathbb{e}[x_k^4 \|\bm{x}_{k-1 } ] & = x_{k-1}^4\theta_k^4 - 4x_{k-1}^3\theta_k^3 \mathbb{e}[\xi_k\|\bm{x}_{k-1 } ] + 6 x_{k-1}^2\theta_k^2 \mathbb{e}[\xi_k^2\|\bm{x}_{k-1 } ] \\ & \;\;\ ; - 4x_{k-1}\theta_k \mathbb{e}[\xi_k^3\|\bm{x}_{k-1}]+\mathbb{e}[\xi_k^4\|\bm{x}_{k-1 } ] = x_{k-1}^4\theta_k^4 + 6x_{k-1}^2\theta_k^2\sigma^2+c\,\sigma^4.\end{aligned}\ ] ] for @xmath1023 we have @xmath1024 . using this relation , conditional version of jensen s inequality and the last display , we derive : @xmath1025 & = \frac{1}{2}\mathbb{e}\big[x_k^2 + \rho\sigma^2-\|x_k^2-\rho\sigma^2\| \big\|\bm{x}_{k-1}\big]\\ & \ge \frac{1}{2 } \big[x_{k-1}^2\theta_k^2 + ( \rho+1)\sigma^2 -\big(\mathbb{e}\big[\big(x_k^2-\rho\sigma^2\big)^2\|\bm{x}_{k-1}\big]\big)^{1/2}\big],\end{aligned}\ ] ] for @xmath719 . we now have , by plugging in the expressions derived above and simplifying , @xmath1026 = \mathbb{e}\big[x_k^4\|\bm{x}_{k-1}\big ] - 2\rho\sigma^2\mathbb{e}\big[x_k^2\|\bm{x}_{k-1}\big ] + \rho^2\sigma^4\\ & = x_{k-1}^4\theta_k^4 + 2(3-\rho)x_{k-1}^2\theta_k^2\sigma^2 + ( c-2\rho+\rho^2)\sigma^4 = \big(x_{k-1}^2\theta_k^2+\frac{c+3}4\sigma^2\big)^2,\end{aligned}\ ] ] if we pick @xmath1027 . combining the previous two displays , we conclude that @xmath1028 \ge \frac{(5-c)\sigma^2}{4 } , \quad k \in \mathbb{n},\ ] ] and the statement of the lemma follows .	we propose an online algorithm for tracking a multidimensional time - varying parameter of a time series , which is also allowed to be a predictable process with respect to the underlying time series . the algorithm is driven by a gain function . under assumptions on the gain , we derive uniform non - asymptotic error bounds on the tracking algorithm in terms of chosen step size for the algorithm and the variation of the parameter of interest . we also outline how appropriate gain functions can be constructed . we give several examples of different variational setups for the parameter process where our result can be applied . the proposed approach covers many frameworks and models ( including the classical robbins - monro and kiefer - wolfowitz procedures ) where stochastic approximation algorithms comprise the main inference tool for the data analysis . we treat in some detail a couple of specific models . * keywords : * on - line tracking ; predictable drifting parameter ; recursive algorithm ; stochastic approximation procedure ; time series ; time - varying parameter .
ta@xmath0s@xmath0c has a unique layered structure , where a sandwiched structure of c - ta - s - vdw - s - ta - c is periodically stacked along the @xmath2 axis . @xcite a van der waals ( vdw ) gap is between adjacent s layers . there are two polytypes : 1t- ta@xmath0s@xmath0c ( space group p3m1 , @xmath14 , @xmath15 ) and 3r - ta@xmath0s@xmath0c ( space group r3 m , @xmath16 , @xmath17 ) . in ta - c - ta layer ( for simplicity ta - c layers are used hereafter ) , each [ ta@xmath18c ] octahedron shares six of its 12 edges with adjacent octahedra and the corner ta atom is shared by three octahedra . this edge linking of [ ta@xmath18c ] octahedra is the common structural feature of transition metal carbides . relatively weak vdw interactions between s layers give this compound a graphitic character . the structure of ta@xmath0s@xmath0c can be viewed as a structural sum of ta - c layers and tas@xmath0 , where the structural part corresponding to ta - c layers is represented by [ ta@xmath18c ] octahedra and the structural part corresponding to tas@xmath0 is identical to the atom disposition of either 1t - tas@xmath0 in the case of 3r - ta@xmath0s@xmath0c or a hypothetical 2h@xmath19-tas@xmath0 ( mos@xmath0-type ) in the case of 1t- ta@xmath0s@xmath0c . x - ray photoelectron core level spectra@xcite of ta@xmath0s@xmath0c show that the binding energy of carbon is close to that in a graphene sheet on ( 111 ) face of ta rather than carbon in tantalum carbides ( tac , ta@xmath0c ) . the binding energy of sulfur is close to that in 1t - tas@xmath0 . the structure of the ta - c layers in ta@xmath0s@xmath0c is different from that of the bulk tac ( cubic ) and ta@xmath0c ( hexagonal ) which are cubic compounds and do not show any layered structure . in this paper we have undertaken an extensive study on the magnetic properties of ta@xmath0s@xmath0c from squid ( superconducting interference device ) dc and ac magnetic susceptibility . we show that this compound undergoes successive superconducting phase transitions at @xmath9 ( @xmath20 k ) and @xmath8 ( @xmath21 k ) . the intermediate phase between @xmath9 and @xmath8 is an intra - grain superconductive state , while the low temperature phase below @xmath9 is an inter - grain superconductive state . the superconducting properties of ta@xmath0s@xmath0c are compared with those of the type - ii superconductor nb@xmath0s@xmath0c with a critical temperature @xmath22 k.@xcite in the presence of a magnetic field ( @xmath12 ) well above the upper critical field @xmath13 ( @xmath23 koe ) , the magnetic susceptibility is described by a sum of a curie - like behavior and a diamagnetic susceptibility . we show that the curie - like behavior is due to localized magnetic moments of conduction electrons associated with the anderson localization effect , which may occur in the 1t - tas@xmath0 type structure in ta@xmath0s@xmath0c . the pristine tac@xmath24 is a superconductor with a fairly high @xmath25 (= 9.7 k ) for @xmath26 , but for @xmath27 no superconductivity is found.@xcite the small change of the fermi surface due to the nonstoichiometry of the carbides considerably reduces the bulk phonon anomaly ( dip ) resulting in the reduction of @xmath25.@xcite in the pristine 1t - tas@xmath0 the charge density wave ( cdw ) becomes commensurate with an undistorted host lattice in a first order transition below 200 k.@xcite the electrical resistivity increases about 10-fold at the 200 k transition and below 2 k the resistivity diverges following the relation @xmath28 $ ] with @xmath29 , where @xmath30 is a constant resistivity and @xmath31 is a characteristic temperature.@xcite this is characteristic of the anderson localization of the conduction electrons due to a random potential . the susceptibility shows a curie - like behavior due to the localized magnetic moments of conduction electrons.@xcite powdered samples of ta@xmath0s@xmath0c were prepared by pablo wally . the detail of the synthesis and structure is described by wally and ueki.@xcite x - ray powder diffraction pattern shows that ta@xmath0s@xmath0c sample consists of a 3r phase as a majority phase and a 1 t phase as a minority phase.@xcite the sample characterization was also carried out by scanning tunneling microscopy@xcite and x - ray photoelectron spectroscopy.@xcite the measurements of dc and ac magnetic susceptibility were carried out using a squid magnetometer ( quantum design mpms xl-5 ) . a polycrystalline powdered sample ( mass 253.2 mg ) was used in the present work . before setting up a sample at 298 k , a remnant magnetic field was reduced to less than 3 moe using an ultra - low field capability option . for convenience , hereafter this remnant field is noted as the state @xmath32 . the detail of the measurements of dc and ac magnetic susceptibility is described in sec . [ result ] . @xmath35 dependence of ( a ) @xmath33 and ( b ) @xmath34 with and without @xmath12 for ta@xmath0s@xmath0c . @xmath36 oe . @xmath37 the inset shows the detail of @xmath33 at @xmath32 around @xmath9.,width=302 ] figure [ fig : one ] shows the @xmath35 dependence of the ac magnetic susceptibility [ ( a ) the dispersion @xmath33 and ( b ) the absorption @xmath34 ] , where the ac frequency @xmath39 (= 1 hz ) and the magnitude of the ac field ( @xmath36 oe ) are used . the @xmath35 dependence of @xmath33 and @xmath34 is strongly dependent on @xmath12 . the sign of @xmath33 is negative at least for @xmath40 k , while the sign of @xmath34 is positive . for @xmath32 ( see the inset of fig . [ fig : one](a ) ) , @xmath33 increases with increasing @xmath35 . it shows a kink at a critical temperature @xmath9 (= 3.61 k ) where the derivative d@xmath33/d@xmath35 undergoes a discontinuous jump . the system undergoes a superconducting transition at @xmath9 . the dispersion @xmath33 increases with further increasing @xmath35 and tends to zero around @xmath8 (= 9.0 k ) . similar kink - behavior is observed at higher @xmath12 , although the kink is hardly seen in fig . [ fig : one](a ) because of very small magnitude of @xmath33 at @xmath41 . the critical temperature @xmath41 decreases with increasing @xmath12 , forming the @xmath12-@xmath35 phase diagram ( see sec . [ resulte ] ) . the absorption @xmath34 at @xmath32 shows a drastic decrease around @xmath9 . it shows a tail above @xmath9 and is reduced to zero around 5 k. the temperature at which the tangential line @xmath34 vs @xmath35 with the steepest slope intersects the @xmath42 axis coincides with @xmath9 . ( a ) @xmath43-@xmath12 loop at 1.9 k. ( b ) @xmath12 dependence of @xmath44 at various @xmath35.,width=302 ] figure [ fig : two](a ) shows the hysteresis loop of the magnetization @xmath43 at @xmath45 k. after the sample was quenched from 298 to 1.9 k at @xmath32 , the measurement was carried out with varying @xmath12 from 0 to 1 koe at @xmath35 , from @xmath46 to -1 koe , and from @xmath47 to 1 koe . the @xmath43-@xmath12 curve at 1.9 k shows a large hysteresis and a remnant magnetization . structural imperfections or defect in the sample may play an role of flux pinning , resulting in a inhomogeneous type - ii superconductor . figure [ fig : two](b ) shows typical data of zero - field cooled ( zfc ) magmetization @xmath44 vs @xmath12 at various @xmath35 . before each measurement , the sample was kept at 20 k at @xmath32 for 20 minutes and then it was quenched from 20 k to @xmath35 ( @xmath48 k ) . the magnetization @xmath44 at @xmath35 was measured with increasing @xmath12 ( @xmath49 oe ) . the magnetization @xmath44 exhibits a single local minimum at a characteristic field for @xmath50 , shifting to the low-@xmath12 side with increasing @xmath35 . the lower critical field @xmath51 is defined not as the first minimum point of the @xmath44 vs @xmath12 , but as the first deviation point from the linear portion due to the penetration of magnetic flux into the sample : typically , @xmath52k@xmath53 oe and @xmath54k@xmath55 oe . the least - squares fit of the data of @xmath56 vs @xmath35 to a conventional relation@xcite @xmath57 $ ] yields @xmath58 oe . @xmath35 dependence of @xmath61 ( = @xmath62 ) at various @xmath12 around @xmath9.,width=302 ] the measurements of @xmath59 (= @xmath63 ) and @xmath60 (= @xmath64 ) were carried out as follows , where @xmath65 is the field - cooled ( fc ) magnetization . after annealing at 50 k for 1200 sec in the absence of @xmath12 , the sample was quenched from 50 to 1.9 k. the magnetic field @xmath12 was applied at 1.9 k and then @xmath59 was measured with increasing @xmath35 . the sample was again heated up and annealed at 50 k for 1200 sec in the presence of @xmath12 . then @xmath60 was measured with decreasing @xmath35 . figures [ fig : three](a ) and ( b ) show the @xmath35 dependence of @xmath59 , @xmath60 , and @xmath61 at @xmath46 oe , where @xmath66 . the scale of the susceptibility is enlarged in fig . [ fig : three](b ) . the susceptibility @xmath59 ( @xmath60 ) exhibits a kink at @xmath9 (= 3.61 k ) , where d@xmath59/d@xmath35 ( d@xmath60/d@xmath35 ) undergoes a drastic decrease . the deviation of @xmath59 from @xmath60 is clearly seen below @xmath9 (= 3.61 k ) , indicating that the extra magnetic flux is trapped during the fc process . between @xmath9 and @xmath8 (= 9.0 k ) , @xmath61 is still positive but nearly equal to zero . the sign of @xmath59 ( @xmath60 ) changes from negative to positive at @xmath8 . similar behavior is also observed at @xmath12 = 20 oe as shown in figs . [ fig : three](c ) and ( d ) . figure [ fig : four ] shows the @xmath35 dependence of @xmath61 at various @xmath12 . the value of @xmath61 at fixed @xmath35 decreases with increasing @xmath12 . the difference @xmath61 undergoes a drastic decrease at @xmath41 . the minimum value of @xmath59 at @xmath46 oe is @xmath67 emu / g at 1.9 k , while the minimum value of @xmath33 at @xmath32 is @xmath68 emu / g at 1.9 k. using the value of @xmath33 at 1.9 k ( @xmath69 emu / g ) and the calculated density @xmath70 g/@xmath71 for 3r - ta@xmath0s@xmath0c,@xcite the fraction of flux expulsion relative to complete diamagnetism ( @xmath72 emu/@xmath71 ) is estimated as 21 % , suggesting that the system consists of small grains . this is in contrast to 38 % of the diamagnetic volume fraction reported for nb@xmath0s@xmath0c by sakamaki et al.@xcite ( a ) and ( b ) @xmath35 dependence of @xmath59 at various @xmath12.,width=302 ] @xmath35 dependence of ( a ) @xmath61 and ( b ) d@xmath59/d@xmath35 at various @xmath12 for @xmath73.,width=302 ] figures [ fig : five](a ) and ( b ) show the @xmath35 dependence of @xmath59 at various @xmath12 , where the measurement was carried out between 1.9 and 11 k with increasing @xmath35 . there is a drastic increase in the diamagnetic contribution in @xmath59 associated with the superconducting transition at @xmath41 . nevertheless , a diamagnetic contribution in @xmath59 still remains above @xmath41 , increases with further increasing @xmath35 , and reduces to a zero at a upper critical temperature @xmath74 . for @xmath75 oe , for example , @xmath59 exhibits a kink at @xmath41 . the sign of @xmath59 changes from negative to positive around 9 k with increasing @xmath35 . at @xmath76 koe , @xmath59 is positive at least between 1.9 and 6 k , showing a broad peak at 2.65 k. at @xmath77 koe , @xmath59 decreases with increasing @xmath35 , showing a curie - like behavior . figure [ fig : six](a ) shows the @xmath35 dependence of @xmath61 at various @xmath12 . the magnitude of @xmath61 between @xmath41 and @xmath74 is very small compared to that below @xmath41 , but still show the irreversible effect of magnetization . the difference @xmath61 at fixed @xmath12 decreases with increasing @xmath35 and reduces to zero at @xmath74 . note that @xmath61 at fixed @xmath35 ( for example 5 k ) increases increasing @xmath12 , showing a maximum around @xmath78 oe , and decreases with further increasing @xmath12 . this feature is in contrast to the @xmath12 dependence of @xmath61 below @xmath41 , which decreases with increasing @xmath12 . figure [ fig : six](b ) shows the @xmath35 dependence of the derivative d@xmath59/d@xmath35 for 5 oe @xmath79 koe . clearly d@xmath59/d@xmath35 at low @xmath12 undergoes two step - like changes around @xmath41 and @xmath74 . the critical temperatures @xmath41 and @xmath74 decreases with increasing @xmath12 , forming the @xmath12-@xmath35 phase diagram ( see sec . [ resulte ] ) . here we present our peculiar results on the @xmath35 dependence of @xmath80 . the method of the measurement for @xmath80 is a little different from that for the conventional @xmath59 . first , the sample was annealed at 50 k for 1200 sec in the absence of @xmath12 and then it was quenched to 1.9 k. after the sample was kept at 1.9 k for 100 sec in the presence of fixed @xmath12 , it was quickly heated up to @xmath31 between @xmath9 and @xmath8 . then @xmath80 was measured with increasing @xmath35 from @xmath31 to 11 k. figures [ fig : seven](a)-(d ) show the @xmath35 dependence of @xmath80 , @xmath59 , and @xmath60 at various @xmath12 ( @xmath81 oe ) , where @xmath82 k. the @xmath35 dependence of @xmath80 at the same @xmath12 is independent of @xmath31 when @xmath31 is at least between 4 and 8 k. the data of @xmath80 vs @xmath35 at @xmath46 and 20 oe are shown in figs . [ fig : three](b ) and ( d ) , respectively . the susceptibility @xmath80 increases with increasing @xmath35 and reduces to zero at @xmath74 . the magnitudes of @xmath59 , @xmath80 , and @xmath60 at the same @xmath12 are strongly dependent on @xmath12 : @xmath83 for @xmath84 oe , @xmath85 for @xmath86 oe , and @xmath87 for @xmath88 and 150 oe . note that @xmath80 at @xmath46 oe ( see fig . [ fig : three](b ) ) , whose sign is positive , decreases with increasing @xmath35 and reduces to zero at @xmath8 . between @xmath9 and @xmath8 , the space of states is divided into at least three states , the zfc state , zfc@xmath89 state , and fc state . the system lies in the fc states under the cooling from 11 k , the zfc state under the slow heating from 1.9 k , and the zfc@xmath89 state under the rapid heating from 1.9 k. the value of @xmath12 (= 30 oe ) is a little higher than the lower critical field @xmath52k@xmath53 oe . the susceptibility of each state provides a measure for the corresponding induced magnetic flux density @xmath90 which is trapped in the superconducting grains : @xmath91 . the inequality @xmath92 indicates that the induced magnetic flux density ( or the number of fluxoids over the system ) in the zfc@xmath89 state is higher than that in the fc state for @xmath93 such a relatively high flux density in the zfc@xmath89 state may be due to a flux compression as a result of the rapid redistribution of the grain - pinned vortices which occurs during a change of @xmath35 from 1.9 k to @xmath31 . this effect exists only at low @xmath12 . in this sense , the present effect is similar to the paramagnetic meissner effect observed in @xmath60 of the pristine nb@xcite : @xmath60 at low @xmath12 becomes positive below the superconducting transition temperature . according to koshelev and larkin,@xcite the surface supercurrent inhomogeneously trap the magnetic flux in the sample interior , as a vortex . in sec . [ disa ] we discuss a possible distribution of vortices around the grains in the zfc and fc states below @xmath9 and between @xmath9 and @xmath8 . @xmath12-@xmath35 phase diagram , where @xmath41 is determined from the measurements of @xmath33 vs @xmath35 ( @xmath94 ) and @xmath34 vs @xmath35 ( @xmath95 ) , and @xmath74 is a temperature where @xmath59 becomes zero ( @xmath96 ) . solid lines are least - squares fitting curves for the data of @xmath41 and @xmath74 to the form @xmath97 $ ] . the fitting parameters are given in the text . @xmath98 k. @xmath99 k.,width=302 ] figure [ fig : eight ] shows the @xmath12-@xmath35 phase diagram , where @xmath41 is determined from the data of @xmath33 vs @xmath35 and @xmath34 vs @xmath35 , and @xmath74 is determined from the data of @xmath59 vs @xmath35 . these lines correspond to the lines @xmath100 and @xmath101 . the least squares - fit of the data of @xmath12 vs @xmath35 for the line @xmath100 to a conventional relation @xmath102 $ ] yields @xmath103 koe and @xmath104 k , where the data of @xmath33 vs @xmath35 and @xmath34 vs @xmath35 are used . the value of @xmath13 thus obtained is comparable to that estimated using an empirical relation given by werhammer et al.,@xcite @xmath105k@xmath106d@xmath107d@xmath108 . in fact , the value of @xmath13 is calculated as @xmath109 koe , where we use @xmath110 k and a slope ( d@xmath107d@xmath111 ( oe / k ) obtained from the linear relation in @xmath112 vs @xmath35 in the vicinity of @xmath113 and @xmath114 . the coherence length @xmath115 and the magnetic penetration depth @xmath116 are related to @xmath117 and @xmath13 through relations @xmath118 and @xmath119 , where @xmath120 ( @xmath121 gauss @xmath122 ) is the fluxoid.@xcite when the values of @xmath117 (= 27.7 oe ) and @xmath13 (= 7730 oe ) are used , the values of the ginzburg - landau parameter @xmath123 ( @xmath124 ) , @xmath116 and @xmath115 can be estimated as @xmath125 , @xmath126 and @xmath127 . our results of @xmath9 , @xmath117 , @xmath13 , @xmath123 , @xmath116 , and @xmath115 in ta@xmath0s@xmath0c thus obtained are compared to those in nb@xmath0s@xmath0c@xcite : @xmath22 k , @xmath128 oe and @xmath129 oe , @xmath130 , @xmath131 , and @xmath132 . both the pristine nb and ta are superconductive elements with the critical temperatures @xmath133nb@xmath134 k and @xmath133ta@xmath135 k,@xcite and have a body - centered cubic structure . we find that the ratio @xmath25(nb@xmath0s@xmath0c)/@xmath25(ta@xmath0s@xmath0c ) (= 2.11 ) is close to the ratio @xmath25(nb)/@xmath25(ta ) (= 2.06 ) . the values of @xmath25 and @xmath136 for ta@xmath0s@xmath0c are comparable to those of 2@xmath137-tas@xmath138se@xmath139 ( @xmath140 ) : @xmath141 k and @xmath142 koe for @xmath143 , @xmath144 k and @xmath145 koe for @xmath146 , @xmath141 k and @xmath147 koe at @xmath148.@xcite what kind of the superconductivity occurs at @xmath74 ? the least squares - fit of the data of @xmath12 vs @xmath35 for the line @xmath149 to a conventional relation @xmath150 $ ] yields @xmath151 koe and @xmath152 k , where the data of @xmath59 vs @xmath35 are used . the origin of the superconductivity at @xmath74 may be due to ta - c layers in ta@xmath0s@xmath0c . according to giorgi et al.,@xcite the critical temperature @xmath25 of the pristine tac@xmath24 increases with decreasing @xmath153 and is equal to 9.0 k at @xmath154 . fink et al.@xcite have reported that the values of @xmath155 and @xmath136 at @xmath156 k for tac are 220 oe and 4.6 koe . these values are in good agreement with the values of @xmath8 and @xmath157 in ta@xmath0s@xmath0c . the origin of the superconductivity at @xmath9 and @xmath8 will be discussed in sec . [ disa ] . @xmath35 dependence of @xmath60 at various @xmath12 near @xmath8.,width=302 ] in fig . [ fig : nine ] we show the @xmath35 dependence of @xmath59 and @xmath60 at various @xmath12 . the susceptibility @xmath60 ( also @xmath59 ) at @xmath158 oe show a sharp peak around 9.5 k and decreases with further increasing @xmath35 . the deviation of @xmath60 from @xmath59 is observed for @xmath159 k. the sign of @xmath60 changes from positive to negative around 16 k. the discontinuous jump of @xmath160 observed in the vicinity of @xmath161 is an artifact due to the squid measurement . in contrast , @xmath60 (= @xmath59 ) at @xmath77 koe decreases with increasing @xmath35 from 1.9 k and merge to the curves of @xmath60 at @xmath158 oe above 9.5 k. @xmath35 dependence of @xmath60 at @xmath77 koe . the inset shows the @xmath35 dependence of @xmath60 at @xmath77 , 20 , and 30 koe . the solid line in the inset denotes the least - squares fitting curve to eq.([eq : one ] ) . the fitting parameters are given in the text.,width=302 ] figure [ fig : ten ] shows the @xmath35 dependence of @xmath60 at @xmath77 koe for @xmath162 k. the susceptibility @xmath160 consists of a diamagnetic susceptibility and a curie - like susceptibility . the susceptibility @xmath60 slightly decreases with increasing @xmath35 above 20 k and reaches a negative constant ( diamagnetic susceptibility ) above 150 k : @xmath163 emu / g at 298 k. similar @xmath35 dependence of @xmath160 is observed in 1t - tas@xmath0 below 150 k.@xcite a step - like change @xmath160 near 200 k for 1t - tas@xmath0 is due to the cdw transition between a high - temperature incommensurate phase and a low - temperature commensurate phase . as seen in fig . [ fig : ten ] there is no such anomaly above 20 k in ta@xmath0s@xmath0c . the inset of fig . [ fig : ten ] shows the @xmath35 dependence of @xmath60 at @xmath77 , 20 , and 30 koe . the susceptibility is strongly dependent on @xmath12 below 10 k , indicating that @xmath65 is nonlinear in @xmath12 ( see also the data of @xmath44 vs @xmath12 in fig . [ fig : eleven ] ) . note that the discontinuous jump of @xmath60 observed in the vicinity of @xmath164 is an artifact due to the squid measurement . almost all the data between 10 and 16 k are removed from the figure . the least - squares fit of the data of @xmath60 vs @xmath35 at @xmath77 koe for @xmath165 k to @xmath166 yields the @xmath35-independent susceptibility @xmath167 emu / g , the curie - weiss constant @xmath168 emu k / g , and the curie - weiss temperature @xmath169 = -2.37 @xmath170 0.07 k. the magnitude of @xmath171 is a little larger than that of @xmath172 determined above . the susceptibility @xmath60 consists of a curie - like behavior at low @xmath35 and a diamagnetic contribution at high @xmath35 . we assume that the curie - like behavior is due to the localized electron spins having the effective magnetic moment @xmath173^{1/2}\mu_{b } = \sqrt{3 } \mu_{b}$ ] , where the land g - factor @xmath174 = 2 and the spin @xmath175 . this localized magnetic moment could be related to the anderson localization ( see sec . [ disb ] ) . through the repulsive coulomb interaction between the electrons , a singly occupied state exists in the vicinity of @xmath176 . then the number of spins per gram of ta@xmath0s@xmath0c ( @xmath177 ) can be estimated as @xmath178/g ( or @xmath179/@xmath71 ) from the curie - weiss constant . in summary , the superconductivity below @xmath9 , which is dominant at low @xmath12 , is weakened as @xmath12 increases and overcome by the localization effect at high @xmath12 well above @xmath13 . there may be another possibility that the curie - weiss behavior is due to magnetic impurities ( for example , fe@xmath180 ) , which may be contained in original ta ( typically 15 ppm fe in the pristine 1t - tas@xmath0).@xcite however , this possibility may be ruled out in the following way . if each fe@xmath180 ion has the effective magnetic moment ( @xmath181 ) , then the number of fe@xmath180 spins per gram of ta@xmath0s@xmath0c is estimated as @xmath182 /g . the magnitude of fe@xmath180 impurities is estimated as 1800 ppm , which is too large compared to @xmath183 ppm as major metallic impurities of ta@xmath0s@xmath0c . similar curie - weiss behavior is observed in the susceptibility of 1t - tas@xmath0 : @xmath184 emu / g , @xmath185 k , and @xmath186 emu k / g.@xcite these values of @xmath171 , @xmath169 , and @xmath187 are comparable to those derived in the present work for ta@xmath0s@xmath0c . the curie - like behavior in 1t - tas@xmath0 is not due to magnetic impurities , but due to the localized magnetic moments related to the anderson localization effect.@xcite the diamagnetic susceptibility is common to the 1t - polytypes showing cdw s . in 1t - tas@xmath0 , the effect of the ordering of the commensurate cdw below 200 k is clearly observed in a discontinuous jump in the electrical resistivity.@xcite in ta@xmath0s@xmath0c the @xmath35 dependence of the electrical resistivity @xmath188 has been reported by ziebarath et al.@xcite using a pressed- and sintered - sample . the resistivity increases with increasing @xmath35 for @xmath189 k ( @xmath190 m@xmath191 cm at 4.2 k and 2.2 m@xmath191 cm at 300 k ) , showing a metallic behavior . unlike the resistivity of 1t - tas@xmath0 , no discontinuous change in @xmath188 has been observed below 300 k. magnetization @xmath44 as a function of @xmath12 at low @xmath35.,width=302 ] difference @xmath192 ( @xmath193 ) as a function of @xmath194 where @xmath195 emu / g , @xmath196 ( @xmath12 and @xmath197 are in the units of oe and tesla , respectively ) , and the data of fig . [ fig : eleven ] for @xmath44 are used . the solid lines denote brillouin functions given by @xmath198 with @xmath199/g.,width=302 ] figure [ fig : eleven ] shows the @xmath12 dependence of @xmath44 at fixed @xmath35 below @xmath9 . the susceptibility @xmath44 is negative at low @xmath12 because of the meissner effect . the sign of @xmath44 changes to positive at a zero - crossing field @xmath200 . the value of @xmath201 ( @xmath202 koe at 1.9 k ) coincides with that evaluated from the relation @xmath203 $ ] . the magnetization @xmath44 increases with further increasing @xmath12 . it shows a broad peak , which shifts to the high-@xmath12 side with increasing @xmath35 : 22.5 koe at 1.9 k and 29 koe at 3.5 k. this peak arises from a competition between the curie - like behavior ( @xmath192 ) and the diamagnetic contribution ( @xmath204 ) . in fig . [ fig : twelve ] we show the plot of a magnetization @xmath192 as a function of @xmath194 , where @xmath205 with @xmath195 emu / g , and @xmath206 ( @xmath12 and @xmath197 are in the units of oe and tesla , respectively ) . the curve of @xmath192 vs @xmath194 , which are slightly different for different @xmath35 , increases with increasing @xmath194 . the function form of @xmath192 vs @xmath194 will be discussed in sec . we find that ta@xmath0s@xmath0c undergoes successive superconducting transitions at @xmath9 and @xmath8 . the diamagnetic volume fraction ( 21 % ) indicates that our system is formed of many small grains . the @xmath35 and @xmath12 dependence of @xmath59 and @xmath60 below @xmath9 is very different from that between @xmath9 and @xmath8 : @xmath207 for @xmath208 and @xmath209 ( @xmath207 ) for @xmath210 . the difference @xmath61 at a fixed @xmath35 below @xmath9 decreases with increasing @xmath12 . in contrast , @xmath61 at a fixed @xmath35 between @xmath9 and @xmath8 increases with increasing @xmath12 , showing a maximum around @xmath78 oe , and decreases with further increasing @xmath12 . we note that similar successive transitions have been reported in a ceramic superconductor yba@xmath0cu@xmath211o@xmath212 ( @xmath213 k and @xmath214 k),@xcite which consists of small grains . below @xmath8 , @xmath59 ( @xmath215 ) becomes negative . the difference @xmath61 appears below @xmath9 . the difference @xmath61 at a fixed @xmath35 below @xmath9 decreases with increasing @xmath12 . the successive phase transitions at @xmath9 and @xmath8 in yba@xmath0cu@xmath211o@xmath212 can be qualitatively explained by kawachi et al.@xcite by taking into account of the role of the superconducting grain below @xmath8 and clusters below @xmath9 . according to their model , our results of ta@xmath0s@xmath0c can be explained as follows . below @xmath8 the superconductivity occurs in each grain . in the presence of @xmath12 well above @xmath216 , the fluxoids are pinned by pinning centers such as defects and vacancies within grains . a very weak irreversible effect of magnetization suggests that the density of fluxoids pinned in each grain for the fc state is slightly larger than the zfc state . the distribution of fluxoids in the fc state is more uniform , while in the zfc state more fluxoids are concentrated in the grain - boundary , which does not contributes to @xmath59.@xcite below @xmath9 , the superconductivity occurs in clusters formed of grains coupled through weak inter - grain josephson couplings . in the presence of @xmath12 , the system is divided into relatively free regions ( the cluster - boundary region ) and strong - pinned regions ( within the clusters ) . in the fc state , the distribution of the fluxoids is more uniform inside the clusters . in the zfc state , more fluxoids are concentrated in the cluster - boundary regions which may have little contribution to @xmath59 . in the presence of @xmath12 ( @xmath217 ) applied to the system in the zfc state , the fluxoids will enter only the cluster - boundary regions around the clusters . the fluxoids in the inter - cluster region can have a path through the system without being caught by clusters . when @xmath218 , some free fluxoids are caught by the clusters and pinned strongly , which contributes to @xmath59 . because of such an increase in @xmath59 , @xmath11 decreases with increasing @xmath12 . the possible existence of mesoscopic grains in the ta - c layers of ta@xmath0s@xmath0c would be essential to the successive transitions having a hierarchical nature . the superconductive ordering proceeds in two steps from the intraplanar josephson couplings between grains in the same ta - c layers to the interplanar josephson interaction between grains in adjacent ta - c layers separated by tas@xmath0-type structure . in the intermediate phase between @xmath8 and @xmath9 , each grain in ta - c layers becomes a superconductor . through the intraplanar josephson coupling between grains , the region of the superconducting grains becomes larger as @xmath35 decreases below @xmath8 , forming a 2d superconducting phase . thermal fluctuations overcome the interplanar josephson coupling . just below t@xmath219 , the effective interplanar josephson coupling becomes strong enough to give rise to a 3d superconducting phase . the 2d superconducting systems are coupled to each other through a weak interplanar josephson coupling between adjacent ta - c layers . we note that the magnetic analogy to the successive phase transitions of such a hierarchical nature is seen in a stage-2 cocl@xmath0 graphite intercalation compound ( gic):@xcite @xmath8 (= 8.9 k ) and @xmath9 ( @xmath220 k ) . the existence of islands is essential to the successive phase transitions . the nearest neighbor spins inside islands are ferromagnetically coupled with intraplanar exchange interactions . on approaching @xmath8 from the high-@xmath35 side , spins come to order ferromagnetically inside islands . at @xmath8 these ferromagnetic islands continue to order over the same layer through interisland interactions ( mainly ferromagnetic ) , forming a 2d ferromagnetic long - range order . below @xmath9 a reentrant spin - glass - like phase order is established through effective antiferromagnetic interplanar interactions between spins in adjacent intercalate layers , where the antiferromagnetic phase and the spin glass phase coexist . plot of @xmath221 for @xmath45 k ( dash - dotted line ) and @xmath222 k ( dotted line ) , and the brillouin function given by @xmath223 ( solid line ) as a function of @xmath194 as the intrastate coulomb interaction @xmath224 ( in the units of k ) is changed as a parameter . k for 1t - tas@xmath0.@xcite @xmath226 is defined by eq.([eq : four ] ) . @xmath227 is the pauli paramagnetism given by @xmath228.,width=302 ] we have shown that in ta@xmath0s@xmath0c , the susceptibility @xmath60 at low @xmath35 and high @xmath12 obeys the curie law , which is a direct evidence of the appearance of local magnetic moments of unpaired electrons due to the anderson localization effect . a theory for the magnetic susceptibility in systems with both localization and electron - electron interactions is presented by kobayashi et al.,@xcite and kamimura and aoki.@xcite it is assumed that there are localized states with energies very close to the fermi level @xmath176 . singly occupied states are energetically favorable by virtue of intra - state coulomb interaction ( @xmath229 ) between spin - up and spin - down electrons in the same localized state . in the presence of @xmath12 , the magnetization contribution @xmath230 of the @xmath231-th localized state with energy @xmath232 , relative to @xmath176 , can be written as @xmath233/z \label{eq : two}\ ] ] where @xmath234 , @xmath235 is the bohr magneton , and @xmath236 is the partition function given by @xmath237 corresponding to an empty state , a state occupied by a spin - up electron with the energy @xmath232 - @xmath238 , a state occupied by a spin - down electron with the energy @xmath232 + @xmath238 , and a state occupied by spin - up and spin - down electrons with 2 @xmath232 + @xmath224 . then the total magnetization @xmath226 is obtained as @xmath239^{1/2 } } \ln { \frac{\cosh ( \beta \mu_{b } h)+\lbrack \cosh ^{2 } ( \beta \mu _ { b } h)-e^{-\beta u } ] ^{1/2}}{\cosh ( \beta \mu _ { b } h)-\lbrack \cosh ^{2 } ( \beta \mu _ { b } h)-e^{-\beta u } ] ^{1/2 } } } , \label{eq : four}\ ] ] where @xmath240 is the density of states at @xmath176 . for @xmath241 , @xmath43 is equal to the pauli paramagnetism @xmath242 . here we define @xmath243 as @xmath221 , where @xmath12 is in the unit of oe and @xmath224 is in the unit of k : @xmath225 k for the pristine 1 t tas@xmath0 . figure [ fig : thirteen ] shows the plot of @xmath243 as a function of @xmath194 for @xmath244 tesla and @xmath245 and @xmath246 k as @xmath224 is changed as a parameter , where @xmath197 is in the units of tesla . for comparison , the brillouin function given by @xmath247 is also plotted for each @xmath224 . although @xmath243 depends on both @xmath194 and @xmath248 for each @xmath224 , the curve of @xmath243 vs @xmath194 for each @xmath224 fits well with @xmath249 with the same @xmath224 . note that @xmath249 is a little larger than @xmath243 at the same @xmath194 for 0@xmath2503 . the magnetization @xmath243 at @xmath251 is equal to the saturation magnetization ( @xmath252 ) of @xmath249 . this implies that @xmath243 with @xmath224 coincides with the magnetization of free electron spins whose number is given by @xmath253 . here we discuss our result shown in fig . [ fig : twelve ] based on the above model . as shown fig . [ fig : twelve ] , all the data of @xmath254 vs @xmath194 do not fall on a single - valued function of @xmath194 . this is consistent with the expression given by eq.([eq : four ] ) . negative values of @xmath192 for @xmath255 is due to the meissner effect . in fig . [ fig : twelve ] a solid line denotes a brillouin function given by @xmath256 with @xmath257 emu / g or @xmath258/g , where @xmath259 . our data of @xmath192 vs @xmath194 greatly deviates downward from this brillouin function for @xmath260 . in fig . [ fig : twelve ] the magnetization @xmath192 reaches 0.012 emu / g at @xmath261 . since the saturation magnetization is @xmath262 , the value of @xmath177 can be estimated as @xmath263/g . we also note @xmath264/g ( @xmath265 emu / g ) in the limit of @xmath266 , which is evaluated from the curie - weiss constant ( see sec . [ resultf ] ) . alternative method to determine @xmath177 is as follows . we assume that @xmath267 . the magnetization @xmath44 at fixed @xmath268 has a local maximum at @xmath269 : @xmath270 @xmath271 . using the values of @xmath272 and @xmath268 determined from the inset of fig . [ fig : eleven ] and @xmath273 emu / g , @xmath177 is calculated as @xmath274/g . this value of @xmath177 is close to that for 1t - tas@xmath0 ( @xmath263/g ) which is calculated from the curie - weiss constant ( @xmath275 emu / g ) obtained by disalvo and waszczak.@xcite for comparison , in fig . [ fig : twelve ] we also show the brillouin function with these values of @xmath177 , as a function of @xmath194 . we do not find any reasonable value of @xmath177 , which leads to good agreement between the result and the brillouin function over the whole range of @xmath194 . similar behavior has been also observed in 1t - tas@xmath0 : the downward deviation of the observed magnetization from the brillouin function ( corresponding to the case of @xmath276/g in fig . [ fig : twelve ] ) occurs for @xmath277.@xcite one of the reason for such a difference between the theory and experiment is that that @xmath224 is assumed to be constant in the above model . the intra - state coulomb energy @xmath278 is considered to decrease with increasing @xmath232 . the total energy @xmath279 is needed for the state i to be occupied by the spin - up and spin - down electrons . if @xmath280 , the @xmath231-th state is occupied by a single electron which behaves as a free spin . if @xmath281 , the @xmath231-th state is occupied by paired electrons.@xcite the number of free spins is given by @xmath282 , where @xmath283 is the region where free spins can situate on the localized states . the replacement of @xmath224 by @xmath283 ( @xmath284 ) leads to a decrease in the saturation magnetization . thus the downward deviation of the magnetization from the brillouin function with @xmath224 in ta@xmath0s@xmath0c and 1t - tas@xmath0 can be qualitatively explained in terms of this replacement . two phenomena ( the superconductivity and the anderson localization effect ) are observed in ta@xmath0s@xmath0c . the structure of ta@xmath0s@xmath0c can be viewed as a sum of ta - c layers and tas@xmath0-type structure . the superconductivity is mainly due to the ta - c layers , while the anderson localization effect is due to tas@xmath0-type structure . ta@xmath0s@xmath0c undergoes successive superconducting transitions of a hierachical nature at @xmath104 k and @xmath6 k. the intermediate phase between @xmath8 and @xmath9 is a intra - grain superconductive state occurring in the ta - c layers in ta@xmath0s@xmath0c , while the low temperature phase below @xmath9 is a inter - grain superconductive state . the @xmath35 dependence of magnetic susceptibility at @xmath12 well above @xmath13 and the @xmath12 dependence of @xmath44 at low @xmath35 and high @xmath12 are described by a sum of a diamagnetic susceptibility and a curie - like behavior . the latter shows that the anderson localization effect occurs in the 1t - tas@xmath0-type structure in ta@xmath0s@xmath0c , leading to the localized magnetic moments due to unpaired electrons just below @xmath176 . the authors are grateful to pablo wally , technical university of vienna , austria ( now littlefuse , yokohama , japan ) for providing them with the samples . one of the authors ( j.w . ) acknowledges financial support from the ministry of cultural affairs , education and sport , japan , under the grant for young scientists no . 70314375 and from kansai invention center , kyoto , japan . r. brec , j. ritsma , g. ouvrard , and j. rouxel , inorganic chemistry * 16 * , 660 ( 1977 ) . h. boller and h. blaha , j. solid state chem . * 45 * , 119 ( 1982 ) . r. p. ziebarath , j. k. vassiliou , and f. j. disalvo , j. less - common metals * 156 * , 207 ( 1989 ) . h. boller and k. hiebl , j. alloys and compounds * 183 * , 438 ( 1992 ) . h. boller , int . j. of refractory metals and hard materials * 12 * , 195 ( 1993 - 94 ) . wally and m. ueki , j. solid state chem . * 138 * , 250 ( 1998 ) . wally and m. ueki , j. alloys and compounds * 268 * , 83 ( 1998 ) . wally and m. ueki , wear * 215 * , 98 ( 1998 ) . j. walter , h. shioyama , and s. hara , j. vac . b * 18 * , 1203 ( 2000 ) . j. walter , w. boonchuduang , and s. hara , j. alloys and compounds * 305 * , 259 ( 2000 ) . k. sakamaki , h. wada , h. , y. nuki , and m. kawai , solid state commun . * 118 * , 113 ( 2001 ) . a. l. giorgi , e. g. szklarz , e. k. storms , a. l. bowman , and b. t. matthias , phys . * 125 * , 837 ( 1962 ) . h. j. fink , a. c. thorsen , e. parker , v. f. zackay , and l. toth , phy . rev . * 138 * , a1170 ( 1965 ) . c. oshima , r. souda , m. aono , s. otani , and y. ishizawa , phys . * 56 * , 240 ( 1986 ) . see also references therein . a. wilson , f.j . disalvo , and s. mahajan , adv . * 24 * , 117 ( 1975 ) . f. j. disalvo and j. e. graebner , solid state commun . * 23 * , 825 ( 1977 ) . f. j. disalvo and j. v. waszczak , phys . b * 22 * , 4241 ( 1980 ) . r. inada , y. nuki , and s. tanuma , j. phys . * 50 * , 1217 ( 1981 ) . see for example , j. b. ketterson and s. n. song , superconductivity ( cambridge university press , cambridge , 1999 ) . p. kosti , b. veal , a.p . paulikas , u. welp , v.r . todt , c. gu , u. geiser , j.m . williams , k.d . carlson , and r.a . klemm , phys . b * 53 * , 791 ( 1996 ) . l. pust , l. e. wenger , and m. r. koblischka , phys . b * 58 * , 14191 ( 1998 ) . a. e. koshelev and a. i. larkin , phys . b * 52 * , 13559 ( 1995 ) . n. r. werthamer , e. helfand , and p. c. hohenberg , phys . rev . * 147 * , 1295 ( 1966 ) . r. c. morris and r. v. colmen , phys . b * 7 * , 991 ( 1973 ) . m. kawachi , m. hagiwara , k. koyama , and m. matsuura , j. phys . 63 * , 3405 ( 1994 ) . m. matsuura , m. kawachi , k. miyoshi , m. hagiwara , and k. koyama , j. phys . 64 * , 4540 ( 1995 ) . l. ji , m. s. rzchowski , and m. tinkham , phys . rev . b * 42 * , 4838 ( 1990 ) . l. ji , m. s. rzchowski , n. anand , and m. tinkham , phys . b * 47 * , 470 ( 1993 ) . m. suzuki , i. s. suzuki , and t .- y . huang , j. phys . condensed matter * 14 * , 5583 ( 2002 ) . s. kobayashi , y. fukagawa , s. ikehata , and w. sasaki , j. phys . . jpn . * 45 * , 1276 ( 1978 ) . h. kamimura and h. aoki , _ the physics of interacting electrons in disordered systems _ ( clarendon press , oxford 1989 ) . see references therein for the anderson localization effect . s. ikehata , t. ema , s. kobayashi , and w. sasaki , j. phys . * 50 * , 3655 ( 1981 ) .	a complex carbide ta@xmath0s@xmath0c consists of van der waals ( vdw)-bonded layers with a stacking sequence @xmath1 c - ta - s - vdw - s - ta - c- @xmath1 along the @xmath2 axis . the magnetic properties of this compound have been studied from dc and ac magnetic susceptibility . ta@xmath0s@xmath0c undergoes successive superconducting transitions of a hierachical nature at @xmath3 k [ @xmath4 oe and @xmath5 koe ] and @xmath6 k [ @xmath7 koe ] . the intermediate phase between @xmath8 and @xmath9 , where @xmath10 , is an intra - grain superconductive state occurring in the ta - c layers in ta@xmath0s@xmath0c . the low temperature phase below @xmath9 , where @xmath11 clearly appears , is an inter - grain superconductive state . the magnetic susceptibility at @xmath12 well above @xmath13 is described by a sum of a diamagnetic susceptibility and a curie - like behavior . the latter is due to the localized magnetic moments of conduction electrons associated with the anderson localization effect , occurring in the 1t - tas@xmath0 type structure in ta@xmath0s@xmath0c .
the hcn molecule plays a fundamental role in interstellar chemistry as it is the simplest molecular species containing both c and n atoms ( while the radical cn is the simplest interstellar species composed of c and n ) ; reactions involving hcn can produce key species intermediate to the formation of amino acids , such as aminoacetonitrile @xcite . hcn has been observed in the gas phase in various environments including molecular clouds @xcite , circumstellar envelopes @xcite , the comae of comets @xcite , the atmospheres of neptune @xcite and titan @xcite , and was recently discovered in the murchison meteorite @xcite . typical abundances with respect to h@xmath14 are 10@xmath15 10@xmath16 towards molecular clouds @xcite and massive protostars @xcite , rising to 10@xmath17 in the hot envelope preceding a hot core @xcite due to a combination of gas phase chemistry and the sublimation of ices from dust grain surfaces . as such , hcn has been shown to be a linear tracer of ongoing star formation and is used to probe dense gas in star forming regions @xcite . the gas phase ratios between isomers such as hnc and hcn indicate the extent of photochemical isomerisation that has occurred in astrophysical environments . the hnc / hcn ratio is highly dependent on temperature , and has been observed to be @xmath18 1 in prestellar cores @xcite , 0.6 0.9 in infrared dark clouds @xcite , and less than 0.3 in neptune s atmosphere @xcite . the gas phase ratio cn / hcn can change by orders of magnitude between the edge of a molecular cloud and the densest regions of the core , as it is sensitive to photodissociation channels @xcite . a greater knowledge of the grain surface chemistry network between cn and hcn is needed to understand cn / hcn ratios in dense cores @xcite . cn@xmath3 is the first diatomic anion to have been positively identified in an astrophysical environment , having been observed recently in the gas phase in the c - star envelope irc + 10216 @xcite . one environment where hcn is of particular interest is in comets and their comae . the cn moiety ( part ) is key in the production of amino acids @xcite , recently observed in comets for the first time by the _ stardust _ spacecraft @xcite . both hcn and cn have been observed extensively in the gas phase on lines of sight towards cometary comae ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? hcn is believed to be the major source of the cn radical @xcite , although other sources , such as refractory organic compounds , are necessary in order to produce the observed cn abundances @xcite . in the solid phase , hcn is unlikely to exist solely in its monomeric form , but also in the form of polymers and complex species which act as hcn reservoirs , such as hexamethyltetramine ( hmt , * ? ? ? * ; * ? ? ? * ) or poly(methylene - imine ) ( pmi , * ? ? ? solid phase hcn has yet to be positively identified in comets , but a tentative detection of solid phase hcn on triton was made recently using the akari telescope @xcite . experimentally , solid phase hcn has been produced under astrophysically relevant conditions upon irradiation of ch@xmath19:n@xmath14 mixtures with 5 kev electrons @xcite . hcn can react as an electrophile , as in the reaction of hcn with h atoms in the solid phase which produces methylamine @xcite , or as a nucleophile , as in its reaction with h@xmath14co to form hydroxyacetonitrile hoch@xmath14cn @xcite . the participation of hcn in a low temperature solid phase strecker synthesis ( a series of reactions to synthesise an amino acid ) yields aminoacetonitrile , nh@xmath14ch@xmath14cn , via thermal processing of an ice mixture of ch@xmath14nh , nh@xmath0 and hcn @xcite . in addition , the irradiation of hcn in ice mixtures and aqueous solutions with photons or charged particles has been seen to yield complex ions @xcite , amino acids @xcite , amides , carboxylic acids and bases @xcite . hcn is also a weak acid , therefore it could react with bases to produce salts . previously , acid - base reactions have been shown to yield salts under astrophysically relevant conditions . the reaction hnco + nh@xmath0 produces nh@xmath2ocn@xmath3 @xcite with an activation energy of 0.4 kjmol@xmath9 @xcite ; hnco + h@xmath14o produces h@xmath0o@xmath20ocn@xmath3 @xcite with an activation energy of 26 kjmol@xmath9 @xcite , which then further reacts to form the isomerisation product hocn with an activation energy of 36 kjmol@xmath9 @xcite ; hcooh + nh@xmath0 yields nh@xmath2hcoo@xmath3 @xcite . such species have relatively high sublimation temperatures ( greater than @xmath18 150 k ) compared with the simple molecules such as co@xmath14 , h@xmath14o and co which compose the bulk of the icy mantles on dust grains @xcite . as such , it is likely that salts will remain on the grain surface following heating processes that sublime the majority of the ice mantle , ultimately forming part of the refractory organic residue observed on comet grains @xcite . in this paper , we report the results of a study of the purely thermal reaction of the weak acid hcn with the base nh@xmath0 : @xmath21 nh@xmath0 has been observed in astrophysical ices at abundances of 2 15 % h@xmath14o @xcite , and is therefore a key basic , nucleophilic species available to react with acidic , electrophilic molecules in grain ice mantles . the 6.0 and 6.85 @xmath22 m bands are observed in ices towards many astrophysical environments , including young stellar objects ( ysos , * ? ? ? * ; * ? ? ? * ) and quiescent molecular clouds @xcite , but have yet to be fully characterised . nh@xmath2 is considered to be a likely candidate to account for some of the feature at 6.85 @xmath22 m @xcite . it has been suggested that most of the 6.0 @xmath22 m excess , and at least some of the 6.85 @xmath22 m band , is related to highly processed ices @xcite . however , given that the bands are observed in quiescent regions , some low - energy formation route to their carriers must be possible . laboratory studies have shown that nh@xmath2 can be produced via acid - base reactions in ices at 10 k with no irradiation ( e.g. * ? ? ? thus , reaction ( [ eqn_hcn_nh3 ] ) is important as it involves the abundant ice species nh@xmath0 , produces nh@xmath2 , and also involves hcn , the simplest molecule containing a cn bond , which could be a prerequisite for the formation of amino acids @xcite . in this work , the nh@xmath2cn@xmath3 product of the nh@xmath0 + hcn reaction is spectroscopically characterised using fourier transform infrared ( ftir ) spectroscopy . the kinetics of the nh@xmath0 + hcn reaction are investigated using isothermal experiments employing ftir spectroscopy . the temperature dependence of the reaction rate is determined to be k(t ) = 0.016@xmath5 s@xmath9@xmath23 . the band strength of cn@xmath3 is determined to be a@xmath7 = 1.8 @xmath24 1.5 @xmath25 10@xmath26 cmmolec@xmath9 . temperature programmed desorption ( tpd ) experiments using mass spectrometry allows calculation of the desorption rate as @xmath10 = 10@xmath11 molecules@xmath12s@xmath9@xmath27 for a zeroth - order multilayer desorption . nh@xmath2cn@xmath3 is therefore more refractory than h@xmath14o , and thus can participate in a warm , water - free chemistry . experiments were performed using the ring experimental set - up , as described elsewhere @xcite . briefly , ring consists of a gold - plated copper surface within a high vacuum chamber ( a few 10@xmath15 mbar ) . the temperature of the surface is controlled using a lakeshore model 336 temperature controller , a closed - cycle helium cryostat ( ars cryo , model de-204 sb , 4 k cryogenerator ) , and a heating resistor ; the temperature is measured using a dtgs 670 silicon diode , with a 0.1 k uncertainty . .experiments performed in the current work . @xmath28 [ cols="<,<,^,^,^,^,^,^,^,^,^ " , ] in order to determine the activation energy of the nh@xmath0 + hcn @xmath1 nh@xmath2cn@xmath3 reaction , isothermal kinetic experiments were performed , following @xcite . a series of isothermal experiments was carried out in which mixtures of hcn : nh@xmath0 with large excesses of nh@xmath0 ( the ratios were typically 1:15 ) were deposited at 15 k , heated as quickly as possible ( @xmath18 10 kmin@xmath9 ) to a specific temperature , and held at that fixed temperature for a number of hours , in order to study the reaction kinetics . the temperatures chosen were 60 , 70 , 80 , 90 , 100 , and 105 k. above 105 k , desorption of the reactant molecules becomes important , and below 60 k the reaction proceeds so slowly that the experiments become prohibitively long to carry out . the abundances of nh@xmath0 and nh@xmath2 in the ice mixtures were monitored via their characteristic absorption features in the ir spectra at 3375 and @xmath18 1440 cm@xmath9 , respectively . the time evolution of the species molar fractions for four experiments are plotted in figure [ figure_ratea ] . the other experimental data is omitted for clarity . it is evident from this figure that an increase in temperature increases the rate of the reaction , particularly in the first few hours of each experiment . as nh@xmath0 was in excess with respect to hcn , each hcn molecule was surrounded by a homogeneous nh@xmath0 environment . thus , we can express the rates of destruction of nh@xmath0 and formation of nh@xmath2 with first - order reaction kinetics , in terms of the molar fraction of the reactants : @xmath30 the first - order kinetics in hcn implies that the molar fraction of nh@xmath0 is constant . although not necessary in the first - order case , it is convenient to express the kinetic equations in terms of unitless molar fractions , which allows the calculation of a reaction rate constant @xmath31 in @xmath32 . resolving equation [ kinetics ] for the isothermal experimental curves gives a reaction rate constant at a fixed temperature . the fits of four of the experiments are overplotted on the experimental data in figure [ figure_ratea ] . the reaction rate constants , determined at the chosen temperatures by analysis of the nh@xmath2 band at @xmath18 1440 cm@xmath9 only , are displayed in table [ kt ] . the temperature dependence of the reaction rate constant is plotted in figure [ figure_rateb ] , in the form of @xmath33 as a function of the inverse of the temperature . assuming @xmath4 follows an arrhenius law , a least - squares straight line fit to the experimental data allows the calculation of the activation energy of the reaction hcn + nh@xmath0 . we derive an activation energy of 2.7 @xmath24 0.4 kjmol@xmath9 with a pre - exponential factor of 0.016@xmath5 s@xmath9 . the uncertainty on the calculated values includes the dispersion of the results measured at 80 k and 90 k , where two experiments were performed at each temperature . the value for the activation energy is consistent with values measured previously for different thermal reaction systems @xcite . similarly , as observed previously , the pre - exponential factor is low , which may be due to the solvent cage effect in the solid solutions , _ i.e. _ the need for the molecules to orientate within the solid into favourable steric configuration before reaction . therefore the reaction rate constant has a @xmath4 = 0.016@xmath5 s@xmath9@xmath23 temperature dependence within the small temperature interval ( 60 105 k ) where we have been able to measure it . the question of the extrapolation to lower temperatures is critical to interstellar ice chemistry , but difficult to address experimentally . as a function of the inverse of the temperature , and the fit against an arrhenius law ( dot - dashed line ) . , scaledwidth=50.0% ] a tpd spectrum of nh@xmath2cn@xmath3 ( m / z 27 and m / z 17 ) is shown in the lower panel of figure [ figure_nh3_hcn_tpd ] . it is evident in this tpd trace that some of the product ( or unreacted hcn and nh@xmath0 ) co - desorbs with the dominant species between @xmath18 120 140 k ( similarly to the pure hcn in the upper panel ) and that the nh@xmath2cn@xmath3 desorbs at higher temperature . it is not possible from the experiments presented here to determine at what temperature nh@xmath2cn@xmath3 first starts to desorb from the surface , as in none of our experiments was the nh@xmath2cn@xmath3 present on the surface with no hcn . thus in all tpd traces , the signals of the desorption of hcn and nh@xmath2cn@xmath3 overlap . the dominant mass fraction for nh@xmath2cn@xmath3 is m / z 27 , but as mentioned in [ identification ] above , other masses observed desorbing at this temperature include m / z 26 ( 17 % ) , 17 ( 20 % ) , 16 ( 15 % ) , and 15 ( 2 % ) . under current experimental conditions there is no evidence of the desorption of m / z 44 or 43 , and thus it is concluded that nh@xmath2cn@xmath3 desorbs from the surface entirely as hcn and nh@xmath0 , following the reverse reaction nh@xmath2 + cn@xmath3 @xmath1 nh@xmath0 + hcn , rather than as nh@xmath2cn@xmath3 . the zeroth - order shape of the m / z 27 nh@xmath2cn@xmath3 curve in figure [ figure_nh3_hcn_tpd ] is less marked since only a few monolayers of nh@xmath2cn@xmath3 are desorbing . using the two methods described in [ sec - pure - hcn - nh3 ] above ( fitting to the m / z 27 desorption peak only , and taking the average of values calculated from five desorption spectra ) the energy of desorption of nh@xmath2cn@xmath3 was calculated as @xmath34 = 38.0 @xmath24 1.4 kjmol@xmath9 ( method 1 ) and @xmath34 = 35.4 @xmath24 1.3 kjmol@xmath9 ( method 2 , with @xmath35 = 1.3 @xmath25 10@xmath36 @xmath24 6.0 @xmath25 10@xmath37 molecules@xmath12s@xmath9 ) . when considering the energies derived for all three molecular species , the values calculated using the fixed @xmath35 method are consistently 6 7 % higher than those calculated using the second method ( @xmath35 as a function of @xmath38 ) . in order to investigate the reaction of hcn and nh@xmath0 under conditions relevant to ices in the interstellar medium , h@xmath14o must be included in the mixtures . h@xmath14o is the most abundant solid phase interstellar molecule , with abundances of 10@xmath39 with respect to h@xmath14 . experiment ( v ) investigates the effect of dilution of hcn in a water ice at a ratio of 1:18 . it is clear from the band positions tabulated in table [ table_freq ] that the peak position of the cn stretching absorption in hcn is shifted redward ( with respect to that of pure hcn ) when diluted in a h@xmath14o ice ( to 2091 cm@xmath9 ) , although the shift is not as marked as that seen for the hcn / cn@xmath3 band in a hcn : nh@xmath0 mixture ( 2077 cm@xmath9 , experiment ( iii ) ) . as expected , no reaction is seen for a mixture of hcn : h@xmath14o , neither at 10 k nor during heating to 180 k , where both hcn and h@xmath14o have desorbed . experiment ( vi ) is a mixture of hcn : nh@xmath0:h@xmath14o with concentration ratio 1:3:30 . this experiment investigates the hcn + nh@xmath0 reaction when heavily diluted in h@xmath14o , as would be expected in interstellar ices . the ir spectrum of this mixture at 10 k is shown in figure [ figure_astro_mixtures ] , trace a. the addition of h@xmath14o shifts the hcn / cn@xmath3 and nh@xmath0 bands with respect to their positions in hcn : nh@xmath0 mixtures . the hcn / cn@xmath3 @xmath40 band is observed at 2090 cm@xmath9 , while the nh@xmath0 @xmath41 and @xmath40 bands are centred at 1101 and 3378 cm@xmath9 , respectively . for this mixture , the nh@xmath2 band at 1460 cm@xmath9 is not observed as a distinct peak immediately after deposition at 10 k , unlike in mixtures of hcn : nh@xmath0 only . there is perhaps a trace amount of nh@xmath2cn@xmath3 initially present in the mixture , but it is likely that the dilution of hcn and nh@xmath0 in a h@xmath14o matrix prevents their direct reaction in the gas phase during the deposition . during the heating of the ice mixture , a wide and shallow nh@xmath2 band , centered at @xmath18 1490 cm@xmath9 develops in the spectra . at temperatures of 160 170 k ( when almost all of the h@xmath14o has desorbed from the surface ) , this band shifts redwards to 1437 cm@xmath9 . the hcn / cn@xmath3 band at 2090 cm@xmath9 shifts slightly redwards ( 6 7 cm@xmath9 , to around 2086 cm@xmath9 ) during the heating of the ice mixture , due to the h@xmath14o environment , the conversion of hcn to cn@xmath3 , the crystallisation of cn@xmath3 , or a combination of these effects . upon desorption of the h@xmath14o , when nh@xmath2cn@xmath3 is the only species remaining on the surface , the peak of the band shifts to 2092 cm@xmath9 . produced during the temperature ramps in experiments ( iii ) and ( vi ) . the reaction between hcn and nh@xmath0 without h@xmath14o ( experiment ( iii ) , diamonds ) occurs more rapidly than the reaction in a h@xmath14o matrix ( experiment ( vi ) , triangles ) . values are normalised to the abundance at 100 k , before the desorption of nh@xmath0 and hcn from the surface begins.,scaledwidth=50.0% ] further experiments were performed on astrophysically relevant ice mixtures ( experiments ( vii ) and ( viii ) ) to investigate the effect of including co@xmath14 and co in the ice on the position of the cn@xmath3 absorption band . we wanted to investigate the possibility that the cn@xmath3 ion could contribute to the band at 2041 cm@xmath9 ( 4.90 @xmath22 m ) , assigned to ocs @xcite , or to the so - called `` xcn '' band at 2165 cm@xmath9 ( 4.62 @xmath22 m ) observed towards many astrophysical environments , including high and low mass ysos ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? it has been previously observed that the ice absorption features of ionic species can be very sensitive to their chemical environment @xcite and could shift by a few wavenumbers . the ice mixture containing co@xmath14 at 15 k is shown in figure [ figure_astro_mixtures ] ( left panel , trace b ) , while the ice mixture containing co at 10 k is shown in trace c of the same figure . the hcn / cn@xmath3 and nh@xmath0 absorption bands are shifted by the presence of these two species in the ice , as tabulated in table [ table_freq ] . figure [ figure_astro_mixtures ] ( right panel ) shows the spectral region containing the hcn / cn@xmath3 stretching absorption for all three astrophysically relevant ice mixtures studied . the traces are labelled as in figure [ figure_astro_mixtures ] ( trace a ) hcn : nh@xmath0:h@xmath14o , trace b ) hcn : nh@xmath0:h@xmath14o : co@xmath14 , trace c ) hcn : nh@xmath0:h@xmath14o : co ) . the cn absorption feature of hcn / cn@xmath3 is observed at 2100 cm@xmath9 in pure hcn , at 2091 cm@xmath9 when diluted into h@xmath14o , and at 2086 cm@xmath9 in hcn : h@xmath14o : co@xmath14 and hcn : h@xmath14o : co mixtures , is around 45 60 cm@xmath9 from the 2041 cm@xmath9 band and around 65 80 cm@xmath9 from the 2165 cm@xmath9 band . by comparison of the absorption features in figure [ figure_astro_mixtures ] , and the band positions in table [ table_freq ] , it is clear that the addition of h@xmath14o , co@xmath14 and co to ice mixtures containing hcn and nh@xmath0 is not sufficient to induce a significant shift in the position of the hcn or cn@xmath3 features . the presence of co and co@xmath14 in the mixtures shifts the peak to slightly lower frequency ( 2086 cm@xmath9 ) . in the spectra presented here , it is not possible to deconvolve the contributions of hcn and cn@xmath3 to the absorption feature centred at @xmath18 2090 cm@xmath9 . no absorption features are observed in the region of 2165 2175 cm@xmath9 . we conclude that cn@xmath3 can not contribute to either the 2041 cm@xmath9 band , assigned to ocs , or to the `` xcn '' band . if present in ices , it would likely be observed at or very close to 2092 cm@xmath9 . regarding the kinetics of the reaction between hcn and nh@xmath0 , it is evident that dilution in a h@xmath14o matrix slows the formation of nh@xmath2cn@xmath3 , as illustrated in figure [ fig_h2o_matrix ] . quantification of the effect of diffusion on the kinetics of a thermal reaction is extremely complicated , and will be the subject of a forthcoming article ( mispelaer et al . , in preparation ) . in this article , we have not attempted to further study the kinetics of the hcn + nh@xmath0 reaction in mixtures containing h@xmath14o . in low mass dark molecular clouds , the average abundance of gas phase hcn is around 1 1.5 times higher than that of hnco ( see references in * ? ? ? * ) , and thus if ocn@xmath3 is present in interstellar ices , it is reasonable to assume that hcn and cn@xmath3 could equally be present . gas phase hcn is present in molecular clouds at abundances ( with respect to h@xmath14 ) of @xmath18 10@xmath42 10@xmath16 , while solid phase h@xmath14o , the most abundant solid phase species , is present at abundances of @xmath18 10@xmath39 . we can thus assume that the abundance of hcn in the solid phase is approximately 0.1 % h@xmath14o . with values of 5.1 @xmath25 10@xmath43 cmmolecule@xmath9 @xcite and 1.8 @xmath25 10@xmath26 cmmolecule@xmath9 ( this work ) for the cn stretch of hcn and cn@xmath3 , respectively , the band strengths of these molecules are not high enough to compensate for their expected low interstellar abundances . thus , given the sensitivity of current infrared telescopes , these two molecules are unlikely to be detected in ice mantles . an hcn detection has been reported in titan , where atmospheric chemistry has resulted in a higher abundance of solid phase hcn @xcite . moreover , neither the hcn nor the cn@xmath3 band position in different ice environments matches the 4.90 @xmath22 m ( 2041 cm@xmath9 ) band currently assigned to ocs , nor can these species contribute to the 4.42 @xmath22 m `` xcn '' band . however , nh@xmath2cn@xmath3 may contribute to the 6.85 @xmath22 m band through its nh@xmath2 bending absorption band , as discussed above . the chemistry of hcn is of primary importance in interstellar ices , since hcn is the simplest molecular species with a cn bond and , as such , hcn is probably at the origin of some prebiotic molecules . an illustration of the chemical network centred upon hcn and nh@xmath2cn@xmath3 is shown in figure [ figure_hcn_network ] . hcn can be produced by the hydrogenation of the cn radical , and then hydrogenated into ch@xmath14nh and ch@xmath0nh@xmath14 @xcite . as the cn@xmath3 moiety of the nh@xmath2cn@xmath3 species is more nucleophilic than hcn , it could be the source of a rich ice chemistry . its recent detection in the gas phase highlights its potential importance , and validates studies into the formation routes of interstellar ions @xcite . nh@xmath2cn@xmath3 reacts with h@xmath14co to form hydroxyacetonitrile hoch@xmath14cn @xcite . nh@xmath2cn@xmath3 can also react with the hcn hydrogenation product ch@xmath14nh to give aminoacetonitrile nh@xmath14ch@xmath14cn , which can go on to form glycine according to the strecker synthesis . however , the competing reaction of nh@xmath2cn@xmath3 with ch@xmath14nh can also form the poly(methylene - imine ) cn-(ch@xmath14-nh)@xmath44-h ( cn - pmi , * ? ? ? the reaction product depends on the starting concentrations of nh@xmath2cn@xmath3 and ch@xmath14nh ; an excess of ch@xmath14nh promotes the polymerisation of ch@xmath14nh and the production of cn - pmi , while an excess of cn@xmath3 promotes the production of aminoacetonitrile ( and ultimately glycine ) . it is likely that cn@xmath3 will react with other electrophiles in interstellar ices . the reaction hcn + nh@xmath0 @xmath1 nh@xmath2cn@xmath3 is the preliminary step necessary to the reactions discussed above and , as such , it is important to quantify the kinetics of this rate limiting reaction . indeed , without h@xmath14o , hcn does not react with ch@xmath14o @xcite . it is worth emphasising that the reactions in figure [ figure_hcn_network ] are thermally activated , meaning that they do not need any non - thermal process in order to occur at low temperature in astrophysical ices . the competition and the resulting branching ratios for these thermal reactions are entirely determined by the temperature dependence of their rate constants . as the reaction nh@xmath0 + hcn is the first step in the network , its kinetics are limiting to the overall network . the measured barrier to this reaction ( 2.7 @xmath24 0.4 kjmol@xmath9 ) is a little high compared to the formation of the salt nh@xmath2ocn@xmath3 ( 0.4 kjmol@xmath9 , * ? ? ? * ) , but lower than the formation of more complex species such as aminomethanol ( 4.5 kjmol@xmath9 , * ? ? ? * ) . this simplified network is not complete ; in order to fully explain the kinetics surrounding nh@xmath2cn@xmath3 , it is necessary to include all reactions , in particular those originating from nh@xmath0 , such as the nh@xmath0 + co@xmath14 reaction , which can be either thermally activated @xcite or activated by low - energy electrons @xcite . in addition to constraining its reaction network , it is important to understand how much hcn can be stored in the nh@xmath2cn@xmath3 form , whether in molecular clouds , or in higher temperature environments such as comets . when the hcn hydrogenation product ch@xmath14nh reacts with hcooh , rather than with nh@xmath2cn@xmath3 , the complex molecules hexamethyltetramine ( hmt , c@xmath45h@xmath46n@xmath19 ) or a second form of poly(methylene - imine ) ( hcoo-(ch@xmath14-nh)@xmath44-h ) are formed . in particular , hmt is known to decompose under vuv irradiation , producing hcn molecules and cn radicals , and therefore is a possible extended source of the observed hcn and cn in comets @xcite . at higher temperature the reverse reaction nh@xmath2cn@xmath3 @xmath1 nh@xmath0 + hcn / hnc can occur , releasing nh@xmath0 and hcn ( and possibly hnc ) into the gas phase . this type of isomerisation has previously been shown for the hnco / hocn system in water ice @xcite . the formation of nh@xmath2cn@xmath3 allows hcn to be stored within a less volatile species and made available to a high temperature chemistry , when the species fragments and eventually desorbs . as a follow up to studying its formation kinetics , it would be interesting to study the nh@xmath2cn@xmath3 photodissociation under a vuv field . does this sort of irradiation release the simple molecules hcn and nh@xmath0 , as upon heating the salt , or rather yield a more complex chemistry ? moreover , the hcn / hnc ratio produced by the reverse reaction could be measured in the gas phase using gas phase spectroscopy . this type of experiment could aid interpretation of gas phase hcn / hnc ratios in interstellar environments . the inventory of the thermal reactions involving the most abundant solid state species must be continued , and the kinetics of the rate limiting reactions should be investigated , in order to achieve a complete network of interstellar grain reactions . the competition of this series of coupled reactions must be quantified as a function of temperature and of initial abundance ratios in order to understand interstellar ice chemistry . + hcn @xmath1 nh@xmath2cn@xmath3 thermal reaction . the final products are a ) hydroxyacetonitrile ( hoch@xmath14cn ) , b ) aminoacetonitrile ( nh@xmath14ch@xmath14cn ) , c ) poly(methylene - imine ) ( pmi , r-(ch@xmath14-nh)@xmath44-h , where r = hcoo or cn ) , and d ) hexamethylenetetramine ( hmt , c@xmath45h@xmath46n@xmath19).,scaledwidth=50.0% ] in this work , we have shown experimentally that the purely thermal reaction between hcn and nh@xmath0 produces nh@xmath2cn@xmath3 in interstellar ice analogues at low temperature . nh@xmath2cn@xmath3 has a characteristic cn@xmath3 absorption band at 2092 cm@xmath9 ( 4.78 @xmath22 m ) that can be slightly redshifted by the ice environment . however , the environmental effect is not sufficient that the cn@xmath3 band can match the 2041 cm@xmath9 ( 4.90 @xmath22 m ) band observed in the ir spectra of interstellar ices . in addition , at low temperatures , the contributions of hcn and cn@xmath3 to the observed cn stretching mode absorption are difficult to differentiate . we have estimated the cn@xmath3 ion band strength to be @xmath47 = 1.8 @xmath24 1.5 @xmath25 10@xmath26 cmmolec@xmath9 at 20 k and at 140 k ; when coupled with its expected low abundance in interstellar ices , it is unlikely that solid phase cn@xmath3 could be detected by current ir telescopes . we have measured the kinetics of the pure hcn and nh@xmath0 reaction with an excess of nh@xmath0 , and we have derived the temperature dependence for the reaction rate constant to be @xmath4 = 0.016@xmath5 s@xmath48 . when the reactants are diluted in water ice , the reaction is slowed down . we have also measured a desorption energy of @xmath49 = 38.0 @xmath24 1.4 kjmol@xmath9 , assuming a pre - exponential factor of 10@xmath11 molecules@xmath12s@xmath9 . this work has been funded by the french national program physique chimie du milieu interstellaire ( pcmi ) and the centre national detudes spatiales ( cnes ) . j.a.n . is a royal commission for the exhibition of 1851 research fellow . the authors would like to thank prof . h. cottin for the hcn synthesis protocol . agndez , m. , cernicharo , j. , gulin , m. , et al . 2010 , , 517 , l2 aikawa , y. , kamuro , d. , sakon , i. , et al . 2012 , , 538 , a57 bertin , m. , martin , i. , duvernay , f. , et al . 2009 , phys . , 11 , 1838 bernstein , m. p. , sandford , s. a. , allamandola , l. j. , chang , s. , & scharberg , m. a. 1995 , , 454 , 327 bernstein , m. p. , sandford , s. a. , & allamandola , l. j. 1997 , , 476 , 932 bernstein , m.p . , dworkin , j.p . , sandford , s.a . , cooper , g.w . , & allamandola , l. j. 2002 , nature , 416 , 401 bieging , j. h. , chapman , b. , & welch , w. j. 1984 , , 285 , 656 bockelee - morvan , d. , crovisier , j. , despois , d. , et al . 1987 , , 180 , 253 boger , g. i. , & sternberg , a. 2005 , , 632 , 302 boogert , a. c. a. , pontoppidan , k. m. , knez , c. , et al . 2008 , , 678 , 985 boogert , a. c. a. , huard , t. l. , cook , a. m. , et al . 2011 , , 729 , 92 boonman , a. m. s. , stark , r. , van der tak , f. f. s. , et al . 2001 , , 553 , l63 borget , f. , danger , g. , duvernay , f. , et al . 2012 , , 541 , a114 bossa , j. b. , theule , p. , duvernay , f. , borget , f. , & chiavassa , t. 2008 , , 492 , 719 bossa , j. b. , theule , p. , duvernay , f. , & chiavassa , t. 2009 , , 707 , 1524 bottinelli , s. , boogert , a. c. a. , bouwman , j. , et al . 2010 , , 718 , 1100 burgdorf , m. , cruikshank , d. p. , dalle ore , c. m. et al . 2010 , , 718 , l53 burke , d. j. , & brown , w. a. 2010 , pccp , 12 , 5947 chapillon , e. , guilloteau , s. , dutrey , a. , pitu , v. , & gulin , m. 2012 , , 537 , a60 clark , f. o. , buhl , d. , & snyder , l. e. 1974 , , 190 , 545 clutter , d. r. , & thompson , w. e. 1969 , j. chem . phys . , 51 , 153 colin - garca , m. , negrn - mendoza , a. , & ramos - bernal , s. 2009 , bioastronomy 2007 : molecules , microbes and extraterrestrial life , 420 , 175 collings , m. p. , dever , j. w. , fraser , h. j. , & mccoustra , m. r. s. 2003 , , 285 , 633 collings , m. p. , anderson , m. a. , chen , r. , et al . 2004 , , 354 , 1133 danger , g. , borget , f. , chomat , m. , et al . 2011 , , 535 , a47 danger , g. , duvernay , f. , theul , p. , borget , f. , & chiavassa , t. 2012 , , 756 , 11 dartois , e. 2005 , , 119 , 293 demyk , k. , dartois , e. , dhendecourt , l. , et al . 1998 , , 339 , 553 elsila , j. e. , glavin , d. p. , & dworkin , j. p. 2009 , meteoritics and planetary science , 44 , 1323 ferris , j. p. , & hagan , w. j. 1984 , tetrahedron , 40 , 1093 fray , n. , bnilan , y. , cottin , h. , gazeau , m .- c . , & crovisier , j. 2005 , , 53 , 1243 fuchs , g. w. , acharyya , k. , bisschop , s. e. , et al . 2006 , faraday discussions , 133 , 331 glvez , o. , mat , b. , herrero , v. j. , & escribano , r. 2010 , , 724 , 539 gerakines , p. a. , moore , m. h. , & hudson , r. l. 2004 , , 170 , 202 gibb , e. l. , whittet , d. c. b. , schutte , w. a. , et al . 2000 , , 536 , 347 gibb , e. l. , & whittet , d. c. b. 2002 , , 566 , l113 gibb , e. l. , whittet , d. c. b. , boogert , a. c. a. , & tielens , a. g. g. m. 2004 , , 151 , 35 glavin , d. p. , dworkin , j. p. , & sandford , s. a. 2008 , meteoritics and planetary science , 43 , 399 hasegawa , t. i. , herbst , e. , & leung , c. m. 1992 , , 82 , 167 dhendecourt , l. b. , & allamandola , l. j. 1986 , , 64 , 453 herbst , e. , terzieva , r. , & talbi , d. 2000 , , 311 , 869 hirota , t. , yamamoto , s. , mikami , h. , & ohishi , m. 1998 , , 503 , 717 huebner , w. f. , buhl , d. , & snyder , l. e. 1974 , , 23 , 580 ip , w .- h . , balsiger , h. , geiss , j. , et al . 1990 , annales geophysicae , 8 , 319 jamieson , c. s. , chang , a. h. h. , & kaiser , r. i. 2009 , advances in space research , 43 , 1446 keane , j. v. , tielens , a. g. g. m. , boogert , a. c. a. , schutte , w. a. , & whittet , d. c. b. 2001 , , 376 , 254 king , d. a. 1975 , surf . , 47 , 384 lacy , j. h. , faraji , h. , sandford , s. a. , & allamandola , l. j. 1998 , , 501 , l105 lahuis , f. , & van dishoeck , e. f. 2000 , , 355 , 699 mispelaer , f. , theule , p. , duvernay , f. , roubin , p. , & chiavassa , t. 2012 , , 540 , a40 moreno , r. , lellouch , e. , lara , l. m. , et al . 2011 , , 536 , l12 noble , j. a. , congiu , e. , dulieu , f. , & fraser , h. j. 2012a , , 421 , 768 noble , j. a. , theule , p. , mispelaer , f. , et al . 2012b , , 543 , a5 padovani , m. , walmsley , c. m. , tafalla , m. , hily - blant , p. , & pineau des forts , g. 2011 , , 534 , a77 paganini , l. , villanueva , g. l. , lara , l. m. , et al . 2010 , , 715 , 1258 pizzarello , s. 2012 , , 754 , l27 raunier , s. , chiavassa , t. , marinelli , f. , allouche , a. , & aycard , j .- 2003a , chem . lett . , 368 , 594 raunier , s. , chiavassa , t. , allouche , a. , marinelli , f. , & aycard , j .- p . 2003b , chem . , 288 , 197 raunier , s. , chiavassa , t. , marinelli , f. , & aycard , j .- p . 2004 , chem . , 302 , 259 rosenqvist , j. , lellouch , e. , romani , p. n. , paubert , g. , & encrenaz , t. 1992 , , 392 , l99 sandford , s. a. , & allamandola , l. j. 1993 , , 417 , 815 schreyer , k. , helmich , f. p. , van dishoeck , e. f. , & henning , t. 1997 , , 326 , 347 schutte , w. a. , boogert , a. c. a. , tielens , a. g. g. m. , et al . 1999 , , 343 , 966 schutte , w. a. , & khanna , r. k. 2003 , , 398 , 1049 snyder , l. e. , & buhl , d. 1971 , , 163 , l47 soifer , b. t. , puetter , r. c. , russell , r. w. et al . 1979 , , 232 , l53 theule , p. , borget , f. , mispelaer , f. et al . 2011a , , 534 , a64 theule , p. , duvernay , f. , ilmane , a. , coussan , s. , & chiavassa , t. 2011b , , 530 , a96 tokunaga , a. t. , beck , s. c. , geballe , t. r. , lacy , j. h. , & serabyn , g. 1981 , , 13 , 701 van broekhuizen , f. a. , keane , j. v. , & schutte , w. a. 2004 , , 415 , 425 van broekhuizen , f. a. , pontoppidan , k. m. , fraser , h. j. , & van dishoeck , e. f. 2005 , , 441 , 249 vasyunina , t. , linz , h. , henning , t. , et al . 2011 , , 527 , a88 vinogradoff , v. , duvernay , f. , danger , g. , theule , p. , & chiavassa , t. 2011 , , 530 , a128 wagner , e. l. , & hornig , d. f. 1950 , j. chem . phys . , 18 , 296 woodney , l. m. , ahearn , m. f. , schleicher , d. g. , et al . 2002 , , 157 , 193 woon , d. e. 2001 , , 149 , 277	hcn is a molecule central to interstellar chemistry , since it is the simplest molecule containing a carbon - nitrogen bond and its solid state chemistry is rich . the aim of this work was to study the nh@xmath0 + hcn @xmath1 nh@xmath2cn@xmath3 thermal reaction in interstellar ice analogues . laboratory experiments based on fourier transform infrared spectroscopy and mass spectrometry were performed to characterise the nh@xmath2cn@xmath3 reaction product and its formation kinetics . this reaction is purely thermal and can occur at low temperatures in interstellar ices without requiring non - thermal processing by photons , electrons or cosmic rays . the reaction rate constant has a temperature dependence of @xmath4 = 0.016@xmath5 s@xmath6 when nh@xmath0 is much more abundant than hcn . when both reactants are diluted in water ice , the reaction is slowed down . we have estimated the cn@xmath3 ion band strength to be a@xmath7 = @xmath8 cmmolec@xmath9 at both 20 k and 140 k. nh@xmath2cn@xmath3 exhibits zeroth - order multilayer desorption kinetics with a rate of @xmath10 = 10@xmath11 molecules@xmath12s@xmath13 . the nh@xmath0 + hcn @xmath1 nh@xmath2cn@xmath3 thermal reaction is of primary importance because _ ( i ) _ it decreases the amount of hcn available to be hydrogenated into ch@xmath14nh , _ ( ii ) _ the nh@xmath2 and cn@xmath3 ions react with species such as h@xmath14co , or ch@xmath14nh to form complex molecules , and _ ( iii ) _ nh@xmath2cn@xmath3 is a reservoir of nh@xmath0 and hcn , which can be made available to a high temperature chemistry . [ firstpage ] astrochemistry ism : molecules molecular processes molecular data methods : laboratory .
before the mid-1990s it was believed that young neutron stars are all fast - spinning objects with high surface magnetic field strengths of @xmath4 g , emitting radio pulses . however , recent discoveries of new populations of neutron stars , including central compact objects ( ccos ) , magnetars , and x - ray dim isolated neutron stars , have challenged this simple picture ( see reviews by * ? ? ? * ; * ? ? ? ccos are the most enigmatic class . they are found at the center of supernova remnants ( snrs ) and can not be as easily classified as other types of objects . there are nine confirmed ccos sharing the following properties : ( 1 ) they are located near the centers of young snrs ; ( 2 ) they show no radio or optical counterparts ; ( 3 ) they have no detectable pulsar wind nebulae ; and ( 4 ) they exhibit a thermal spectrum in the soft x - ray band with high luminosity @xmath5ergs@xmath2 ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * for reviews ) . since ccos are generally associated with very young snrs , their nature and evolution are highly relevant to the neutron star production rate and the physics underlying the branching ratios of core collapse @xcite . however , the active lifetime and evolution of ccos are poorly understood due to the small sample . it is also unclear if these neutron stars are intrinsically radio - quiet , or if they are radio pulsars beamed away from us ( see e.g. , * ? ? ? * ) . only three ccos have x - ray pulsations firmly detected . they show periods of @xmath60.4s and long - term timing revealed small period derivatives of @xmath7 @xcite . their spin parameters are plotted in the diagram in figure [ fig : ppdot ] . these suggest spin - down luminosities of @xmath8@xmath9ergs@xmath2 , where @xmath10 is the neutron star moment of inertia . these values are nearly two orders of magnitude smaller than the cco s x - ray luminosities , implying that the sources can not be entirely powered by rotation . the characteristic age @xmath11 of the three ccos is over @xmath12years , much older than their true ages of a few thousand years , estimated from their associated snrs . the inferred surface dipole field strengths @xmath13 g are of the order of @xmath14 g . these are much lower than those of young pulsars but still lie well above the radio pulsar `` death line '' ( e.g. , @xmath15gs@xmath16 ; * ? ? ? * ) , which is the ( uncertain ) theoretical limit for producing radio emission . indeed , radio pulsars have been detected in the cco range of @xmath17 and ( see figure [ fig : ppdot ] ) , hence with comparable dipole field strengths . this raises a fundamental question : are ccos and radio pulsars the same class of objects or do they belong to disjointed sets of neutron stars ? if ccos are ordinary radio pulsars born with weak magnetic fields ( see , e.g. , * ? ? * ; * ? ? ? * ) , their small @xmath18 values imply very slow spin evolution , such that they take a long time to reach the death line . the detection of nine young ccos implies that there should be over @xmath19 older ones in the galaxy @xcite . this is in contrast to the small number of known radio pulsars with low @xmath20 g as shown in the diagram in figure [ fig : ppdot ] , given that there is no observational bias against these pulsars in radio surveys ( see * ? ? ? * ; * ? ? ? the discrepancy could be reconciled if ccos are strong - field objects but appear to have weak surface fields at a young age . this is supported by modeling of ccos x - ray light curves , which suggest a strong crustal field @xcite . it was proposed that the @xmath21-fields of ccos could be buried by supernova fallback @xcite , and hence the radio emission is suppressed ( as in the case of accreting millisecond pulsars in quiescence ; see e.g. , @xcite ) . after accretion stops , the @xmath21-field is expected to diffuse back to the surface with a timescale depending sensitively on the amount of accreted mass and ranging from @xmath22years to over @xmath19years ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the radio emission would then presumably switch on @xcite . this picture predicts that young ccos should be radio - quiet and aged ones could become ordinary radio pulsars ( see * ? ? ? * ; * ? ? ? no radio emission has yet been observed from any cco ( see * ? ? ? * and references therein ) , but given the small number of known pulsating ccos and the possibility of small beaming fractions , the result is inconclusive and a more systematic study is needed . in addition to deeper radio observations , we can turn the search around to look for cco - like x - ray emission from selected weak - field radio pulsars . indeed , if ccos manifest as radio pulsars , this could be a more efficient way to detect them , especially after their natal snrs fade away in @xmath23years . any detection of x - ray emission from the radio - selected sample will confirm ccos as a subset of radio pulsars , providing direct evidence to rule out the scenario of ongoing accretion . there were previous attempts to identify ccos from radio pulsars positionally coincident with snrs @xcite and aged ccos with parent snrs that have dissipated ( so - called orphaned ccos ) from disrupted recycled pulsars @xcite . however , no new ccos have been found . to complete the study , here we present an x - ray census of all isolated weak - field radio pulsars near the galactic plane to search for cco - like emission . the sample selection and data analysis are described in section [ sect : data ] and we report the detection limits and discuss the implications in section [ sect : discuss ] . we select weak - field radio pulsars from the atnf catalog @xcite with similar spin parameters as those of the known ccos , according to the following criteria : 1 . weak surface dipole @xmath21-fields of @xmath24 g inferred from spin - down ; 2 . isolated and having periods @xmath25s to avoid recycled pulsars ; and 3 . located @xmath26pc from the galactic plane to exclude old objects . the pulsar height from the plane is calculated from the source s galactic latitude and estimated distance . neutron stars are born at an average distance of 50pc from the plane with a mean space velocity of @xmath27kms@xmath2@xcite . at this velocity , a pulsar travels only 36pc in @xmath28years , which is well below the cut even in the rare case that a pulsar moves exactly perpendicular to the plane . cccccccccc & 1.2 & 189.2 & 1.0 & 3.9 & @xmath29 & 0.06 & 0.24 & 38 & 0.37 + & 1.3 & 289.9 & 1.1 & 4.0 & @xmath29 & 0.25 & 9.0 & 4.5 & 4.8 + & 8.3 & 310.3 & @xmath300.2 & 55 & @xmath31 & 0.28 & 3.1 & 14 & 3.0 + & 1.2 & 315.4 & 3.2 & 4.3 & @xmath32 & 0.35 & 22 & 2.5 & 8.9 + & 7.5 & 341.6 & 0.5 & 67 & @xmath33 & 0.31 & 17 & 2.9 & 7.3 + & 4.6 & 342.3 & 0.4 & 34 & @xmath34 & 0.42 & 15 & 4.4 & 8.0 + & 7.1 & 344.1 & @xmath300.3 & 63 & @xmath35 & 0.23 & 11 & 3.2 & 5.2 + & 0.8 & 4.0 & @xmath304.7 & 2.5 & @xmath36 & 0.34 & 20 & 2.7 & 8.3 + & 2.4 & 2.4 & @xmath301.1 & 12 & @xmath37 & 0.26 & 14 & 3.0 & 6.1 + & 5.6 & 12.1 & 0.3 & 45 & @xmath38 & 0.15 & 52 & 0.47 & 9.1 + & 5.3 & 15.9 & @xmath300.1 & 42 & @xmath39 & 0.23 & 7.9 & 4.6 & 4.3 + & 0.7 & 65.3 & 0.8 & 0.79 & @xmath40 & 0.43 & 1.7 & 40 & 2.7 cccccccc j0609 + 2130 & & 5 & 6.3 & 51 & @xmath41 & 20 + j1107@xmath305907 & & 5 & 14 & 55 & @xmath42 & 20 + j1355@xmath306206 & & 3.4 & 9.1 & 134 & @xmath43 & + j1425@xmath305723 & & 5 & 8.9 & 53 & @xmath44 & 20 + j1650@xmath304341 & & 8 & 3.9 & 124 & @xmath45 & + j1653@xmath304315 & & 5 & 6.3 & 98 & @xmath46 & 0.4 + j1702@xmath304217 & & 8.6 & 6.4 & 126 & @xmath47 & + j1739@xmath303951 & & 5 & 6.3 & 46 & @xmath48 & 30 + j1755@xmath302725 & , & 4 & 12 & 79 & @xmath49 & 1 + j1810@xmath301820 & & 5 & 14 & 119 & @xmath50 & + j1819@xmath301510 & 0406450201 , 0505240101 & 71.5 & 1.4 & 117 & @xmath51 & 0.2 + b1952 + 29 & 12684 & 5 & 6.3 & 45 & @xmath52 & 30 our sample consists of 12 candidates . they are plotted in the diagram in figure [ fig : ppdot ] and their properties are listed in table [ tab : candid ] . while the pulsars have large characteristic ages of @xmath53years , we note that their true age could be much younger , particularly if they were born spinning slowly ( see * ? ? ? * ; * ? ? ? there is an archival _ chandra _ on - axis acis - s observation of psr j1355@xmath306206 with 3ks exposure , and psr j1755@xmath302725 was located far off - axis in two _ chandra_acis - i exposures with a total of 4ks . psr j1819@xmath301510 falls onto the edge of the field of view in two _ xmm - newton _ mos exposures . after filtering the background flaring periods and correcting for vignetting , we obtained an effective total exposure of 72ks for this source . the _ xmm - newton _ pn data were not used in this study , as the total exposure is only 21ks . for the remaining nine candidates , we obtained _ chandra _ acis - s observations with 5ks exposure each , except for psrs j1702@xmath304217 and j1650@xmath304341 , which have roughly 8ks each . the observation ids ( obsids ) and effective exposures are listed in table [ tab : results ] . note that observations of three sources : psrs j0609 + 2130 , j1355@xmath306206 , and b1952 + 29 were recently reported in an independent study @xcite . we carried out standard _ chandra _ data reduction with ciao v4.5 and caldb v4.5.8 . we reprocessed the data using the task ` chandra_repro ` and checked that there was no background flaring occuring during the _ chandra _ observations ; all data are therefore used in the analysis . the _ xmm - newton _ data reduction was performed with sas v13.5 . we used the task ` emchain ` to reprocess the mos data , then applied filtering using standard flags and rejected periods with high background . the analysis is restricted to events with pattern @xmath54 . finally , we generated x - ray images in the 0.57kev energy range for source detection . the x - ray images show no obvious x - ray sources at the radio pulsar positions . to establish the pulsar flux limits , we extracted events from 5 and 70 radius source and background regions , respectively , for the 10 on - axis _ chandra _ observations . the 5 radius was chosen to ensure that the region encircles over 95% of the flux . a larger source region of 10radius was used for psr j1755@xmath302725 , due to the large off - axis angle . for the _ xmm - newton _ observations of psr j1819@xmath301510 , we used a 60 radius source region and excluded a background source inside . we follow @xcite to compute the detection limits at a 2@xmath55 confidence level , using the task ` aprates ` in ciao , which is based on bayesian statistics . the results are reported in table [ tab : results ] . the count rate limits above can be converted to luminosity limits using a spectral model . following @xcite , we assume uniform blackbody emission from the entire star with an observed radius of @xmath56 km ( which corresponds to a neutron star radius of 11.6 km for a gravitational redshift of 1.25 ) . the pulsar dispersion measure ( dm ) distance is adopted and the x - ray absorption column density ( @xmath57 ) is inferred from dm using the empirical relation @xmath58 @xcite . we generated spectral response files for individual observations and determined the flux limits in the ` sherpa ` environment @xcite by comparing the expected count rates with the observed limits . the surface temperature and bolometric luminosity limits are listed in table [ tab : results ] . the detection limits of psrs j0609 + 2130 , j1355@xmath306206 , and b1952 + 29 are compatible with those reported in the previous study @xcite . as shown in the table , the luminosity limits are @xmath59ergs@xmath2 for nearby ( @xmath60kpc ) sources and a few times @xmath61ergs@xmath2 for the rest . in figure [ fig : cooling ] , these results are compared with the observed bolometric luminosities of known ccos . the latter are generally above @xmath61ergs@xmath2 , suggesting that our candidates are unlikely to be young ccos . as mentioned , the characteristic age is not a good indicator of the true age of a pulsar . we therefore attempt to constrain the age by comparing the luminosity limits above with standard neutron star cooling curves . we employed a cooling code based on @xcite with some modifications as described in @xcite . the temperature evolution of an isolated neutron star is determined by the relativistic equations of energy balance and heat flux ( see , e.g. , * ? ? ? two important factors affecting the evolution are the composition of the outer layer ( i.e. , the envelope ) and whether the stellar interior is superfluid and/or superconducting . for the former , the envelope serves as a heat blanket @xcite . if it is composed of light elements , which have higher thermal conductivity , it would be more heat transparent @xcite . however , the exact composition of the envelope is uncertain . for the second factor , superfluidity and superconductivity suppress heat capacities and some emission mechanisms , such as modified urca processes . on the other hand , emission due to the cooper pairing of nucleons would be enhanced when the temperature decreases just below the critical value ( see * ? ? ? * ; * ? ? ? * for reviews ) . here we consider a @xmath62 and 11.6 km radius ( corresponding to an observed radius @xmath56 km ) star built with the akmal - pandharipande - ravenhall ( apr ) equation of state ( eos ) @xcite and the standard ( also known as minimal ) cooling scenario , in which neutrino emission by fast direct urca processes does not take place ( see , e.g. , * ? ? ? * ; * ? ? ? we show in figure [ fig : cooling ] two shaded regions , one for a maximally thick @xmath63 hydrogen envelope and one for an iron envelope . the upper boundary of each region is determined by a cooling model with only superconducting protons ( gap model ccdk ) and the lower boundary accounts for superfluid neutrons as well ( singlet gap model sfb and triplet gap model ttoa ) , which induce more rapid cooling ( see * ? ? ? * for details ) . to illustrate the effect of different masses , we plot in the figure inset cooling curves of @xmath64 and @xmath62 stars for models with only superconducting protons . the model cooling curves are compared with the candidates luminosity limits in figure [ fig : cooling ] to constrain the source age . the result slightly depends on the neutron star mass in the model , but the main uncertainty comes from the whether or not neutrons in the core become superfluid . the inferred age lower limits are listed in table [ tab : results ] . the results are not very constraining . for nearby objects that have good luminosity limits , we deduce age limits of @xmath65years , but no useful limits are obtained for the rest , and deeper x - ray observations are needed for further investigation . unlike the cco candidates in our previous study @xcite , none of the sources here coincide with known snrs , and we do not find evidence of remnant emission from the x - ray data . this indeed gives a hint that the sources are likely older than @xmath23years , such that the snrs have already dissipated in the interstellar environment . the non - detection of x - ray emission from our sample of weak - field radio pulsars seems to support the idea that the two are not evolutionary connected classes of neutron stars , similar to what previous studies have suggested ( e.g. , @xcite , @xcite ) . it was proposed that the magnetic fields of ccos could be buried by accretion from supernova materials at birth @xcite , resulting in the suppression of radio emission . as the field diffuses back to the surface , the radio emission would subsequently turn on so that a cco becomes a radio pulsar . to investigate the effect of magnetic field diffusion on the neutron star spin evolution , we applied a model described in @xcite to solve the induction equation . we assumed that the @xmath21-field is confined in the crust and is initially buried at a high density of @xmath66g@xmath67 , which corresponds to a depth of @xmath68 m or an accreted mass of @xmath69 ( see * ? ? ? * ) . we consider different initial surface @xmath21-field strengths from @xmath70 g to @xmath71 g and the resulting spin evolution is overplotted in the diagram in figure [ fig : ppdot ] . we found that even at this large burial depth , the field emerges with a relatively short timescale of a few @xmath72years , resulting in a vertically upward track in the diagram . more importantly , a cco with an emerging surface field is expected to pass through the parameter space of our candidates ( i.e. , @xmath73@xmath74 g ) at a young age of the order of @xmath75yr , such that it should remain hot and luminous , with the parent snr probably still visible . if this model is true , our candidates are likely to just be old ordinary pulsars rather than orphaned ccos . moreover , we expect that when an aged cco turns on as a radio pulsar , it will have a stronger @xmath21-field than those of our candidates . for further study , identifying cco descendants would require x - ray observations of more radio pulsars to search for thermal emission ( see also * ? ? ? in particular , it would be useful to extend the survey to include sources with @xmath76 g . on the other hand , it is more difficult to detect an orphan cco if the radio emission never turns on ( see * ? ? ? this is best done with all - sky surveys in x - rays , such as the upcoming _ erosita _ mission @xcite . finally , pulsar braking index measurements from long - term timing could reveal a growing dipole field ( e.g. , * ? ? ? * ) , which could indicate @xmath21-field diffusion to the surface ( see * ? ? ? we present an x - ray study of all isolated radio pulsars within 100pc of the galactic plane , which has spin - down - inferred @xmath21-fields weaker than @xmath74 g , with the aim of searching for cco - like thermal x - ray emission . none of our 12 candidates were detected in the _ chandra _ and _ xmm - newton _ observations and we obtained bolometric luminosity limits of @xmath9@xmath77ergs@xmath2 . the limits were compared with a standard neutron star cooling model to constrain the source ages . together with the lack of known associated snrs , we conclude that all candidates are unlikely to be young ccos . this result supports the idea that young ccos and currently known weak - field radio pulsars are not connected through evolution . this could be a consequence of field burial by supernova materials . we model the pulsar spin evolution in this scenario and show that the surface @xmath21-field could reemerge rapidly with a timescale shorter than 100kyr . it could therefore be fruitful to include radio pulsars with slightly stronger fields ( @xmath78 g ) in future searches for cco descendants . we thank the referee for a careful reading and useful comments . this work was funded in part by the nasa _ grant go2 - 13079x awarded through columbia university and issued by the _ chandra x - ray observatory _ center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 03060 . a portion of the results presented was based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and nasa . the scientific results reported in this article are based in part on observations made by the _ chandra x - ray observatory_. this research has made use of the nasa astrophysics data system ( ads ) and software provided by the chandra x - ray center ( cxc ) in the application package ciao . wcgh acknowledges support from stfc in the uk . . receives support from an nserc discovery grant and accelerator supplement , the centre de recherche en astrophysique du quebec , an r. howard webster foundation fellowship from the canadian institute for advanced study , the canada research chairs program , and the lorne trottier chair in astrophysics and cosmology .	central compact objects ( ccos ) are a handful of young neutron stars found at the center of supernova remnants ( snrs ) . they show high thermal x - ray luminosities but no radio emission . spin - down rate measurements of the three ccos with x - ray pulsations indicate surface dipole fields much weaker than those of typical young pulsars . to investigate if ccos and known radio pulsars are objects at different evolutionary stages , we carried out a census of all weak - field ( @xmath0 g ) isolated radio pulsars in the galactic plane to search for cco - like x - ray emission . none of the 12 candidates are detected at x - ray energies , with luminosity limits of @xmath1ergs@xmath2 . we consider a scenario in which the weak surface fields of ccos are due to a rapid accretion of supernova materials and show that as the buried field diffuses back to the surface , a cco descendant is expected to leave the parameter space of our candidates at a young age of a few @xmath310kyr . hence , the candidates are likely to just be old ordinary pulsars in this case . we suggest that further searches for orphaned ccos , which are aged ccos with parent snrs that have dissipated , should include pulsars with stronger magnetic fields .
numerous lines of evidence support the hypothesis that massive elliptical galaxies form the majority of their stars quickly at early epochs , including : stellar population synthesis studies of nearby elliptical galaxies ( renzini 2006 ; collins et al . 2009 ) , a faster decline with cosmic time in the star formation rate per unit stellar mass with increasing galaxy mass ( moresco et al . 2010 ) , and the direct observation of ` red and dead ' ellipticals in clusters at high redshift ( kurk et al . 2009 ; doherty et al . 2009 ; andreon & huertas - company 2011 ) . plausible progenitors for these evolved galaxies at high redshift are even higher redshift submm galaxies ( smgs ) , corresponding to dusty , luminous starburst galaxies found in wide - field submm surveys ( blain et al . these hyper - luminous high - z galaxies may trace high over - densities , and are likely related to the formation of clusters and large ellipticals ( stevens et al . 2003 ; aravena et al . 2010 ; although cf . chapman et al . 2009 ; williams et al . 2011 ) . while the space density of smgs peaks around @xmath12 ( chapman et al . 2003 ) , it is becoming clear that there is a substantial high - redshift tail of the smg population , extending to @xmath13 ( capak et al . 2011 , riechers et al . 2010 ; wardlow et al . daddi et al . ( 2009a ) conclude , based on smg space densities and duty cycles , that there are likely enough smgs at @xmath14 to account for the known populations of old massive galaxies at @xmath15 to 3 . a key question for the smgs is : what drives the prolific star formation ? tacconi et al . ( 2006 ; 2008 ) argue , based on imaging of higher - order co emission from a sample of @xmath15 smgs , that smgs are predominantly nuclear starbursts , with median sizes @xmath16 ( @xmath17kpc ) , ` representing extreme , short - lived , maximum star forming events in highly dissipative mergers of gas rich galaxies . ' this conclusion is supported by vlbi imaging of the star forming regions in two smgs ( momjian et al . 2005 ; 2010 ) . however , recent imaging of the lower order co emission in a few smgs ( ivison et al . 2010 ; carilli et al . 2010 ; riechers et al . 2011 in press ) , suggests that the lower - excitation molecular gas reservoirs can be significantly more extended . daddi et al . ( 2009a ) have identified a unique region in goods - n , containing the three galaxies : gn20 , gn20.2a , and gn20.2b , at @xmath18 , within 20@xmath19 of each other ( projected physical separation = 140 kpc ) . gn20 is among the brightest submm galaxies known , with a 350 ghz flux density of 23mjy . the galaxies gn20.2a and b are separated by only a few arcseconds , and hence are not spatially distinct in typical submm bolometer images . the two galaxies have a summed flux density of @xmath20mjy . the implied ir luminosities ( 8 to 1000@xmath21 m ) are @xmath22 l@xmath4 , with star formation rates are @xmath23 m@xmath4 yr@xmath24 ( daddi et al . high order co emission ( 4 - 3 and/or 6 - 5 ) has been detected from all three galaxies , although only marginally in the case of gn20.2b ( daddi et al . 2009a ) . the gn20 field also contains numerous lyman break galaxies ( lbgs ) at similar redshift , and an overdensity of @xmath25 irac selected galaxies ( daddi et al . hence , this field offers an unprecedented opportunity to perform a definitive study of the gas distribution , kinematics , and physical conditions in a proto - cluster of galaxies within 1.6 gyr of the big bang . in this paper we present a survey with the expanded very large array ( evla ; perley et al . 2011 ) , of the co 2 - 1 emission from the gn20 proto - cluster . the low order transitions are critical for determining the total gas mass . we also search for co emission from other galaxies in the cosmic volume surveyed . the hst i - band image of the bright smg gn20 shows diffuse emission about 1.5@xmath26 in extent ( daddi et al . 2009a ) , although offset from the radio and submm emission by @xmath27 , implying high obscuration of the most active star forming regions in the galaxy . gn20 is detected at 1.4 ghz , and resolved on a scale of @xmath28 , with a total flux density at @xmath29jy ( morrison et al . 2010 ; casey et al . 2009 ) . high resolution imaging of the 850@xmath21 m emission shows resolved structure , with a north - south extension possibly as large at 1.5@xmath26 ( younger et al . 2008 ; iono et al . the 6.2@xmath21 m pah spectral feature has been detected in gn20 using spitzer ( riechers et al . in prep ) . the integrated co 1 - 0 and 2 - 1 emission from gn20 has been imaged at resolutions down to @xmath30 with the vla ( carilli et al . the molecular gas is extended on a scale of at least 1.5@xmath26 , and at high resolution forms a partial ring , or disk . unfortunately , the old vla correlator provided no velocity information . the excitation of the integrated co emission from gn20 is higher than the milky way , but lower than high redshift quasar hosts and the nuclear starburst regions of ir - luminous galaxies . the co 4 - 3 line strength in gn20 is more than a factor of two lower than expected for thermal excitation ( carilli et al . 2010 ) . gn20.2a shows a complex and extended morphology in the hst images ( daddi et al . 2009a ) , and it is a relatively strong radio source ( @xmath31jy ) , likely corresponding to a radio agn . the sed in the optical through near - ir is consistent with a star forming galaxy . gn20.2a has been detected in co 4 - 3 emission by daddi et al . ( 2009a ) . daddi et al . derive an ir luminosity ( 8 to 1000@xmath21 m ) of @xmath32 l@xmath4 . gn20.2b is faint and compact in the hst image . it is also detected in the radio continuum at @xmath33jy , consistent with star formation . co 4 - 3 emission is marginally detected with the plateau de bure interferometer , and appears to be broad ( 700 km s@xmath24 ; daddi et al . 2009a ) . daddi et al . ( 2009a ) find 15 b - band dropout galaxies in a 25@xmath19 radius centered on gn20 , an overdensity of a factor of 6 compared to the full goods - n area , which is significant at the 7@xmath10 level . a spike in the redshift distribution of galaxies at @xmath34 is observed in all of goods - n ( 13 spectroscopic redshifts in total at this redshift ) . lastly , the smg gn10 at @xmath35 is located 9@xmath36 from gn20 ( daddi et al . 2009b ; dannerbauer et al . therefore , it appears that the gn20 volume has a very significant overdensity , indicating a proto - cluster environment at @xmath37 . daddi et al . ( 2009a ) estimate a total mass for this structure of @xmath38 m@xmath4 . we observed the gn20 field with the expanded very large array in the d ( 1 km ) configuration in march and april , 2010 . observations were made of the co 2 - 1 line , using a total bandwidth of 246 mhz and in each of two polarizations , centered at 45.655 ghz . the velocity coverage is 1600 km s@xmath24 , including the co 2 - 1 lines in the three galaxies . the total observing time was 28 hours , but only about 22 to 24 antennas were available at a time due to on - going evla work . dynamic scheduling to ensure good weather is now standard at the evla . fast switching phase calibration was employed ( carilli & holdaway 1999 ) on timescales between two and three minutes using the vla calibrator j1302 + 5748 . data were edited to remove time ranges of poor phase stability . the source 3c286 was used for flux density calibration . standard evla data calibration and editing was performed using aips . after calibration and data editing , images were synthesized using the robust weighting scheme of the uv data with r=2 ( cornwell et al . 1999 ) . the final resolution for the images was @xmath39 . spectral line cubes were generated at 12mhz per channel ( 78 km s@xmath24 ) . the rms noise per channel at 12mhz resolution is 0.11 mjy beam@xmath24 . all images were corrected for the vla primary beam response , which has a fwhm @xmath40 at 45 ghz . the field pointing center was located @xmath41 west of gn20 . figure 1 shows the co 2 - 1 emission integrated over the full frequency range covered by these observations for the gn20 group . the crosses mark the 1.4 ghz continuum positions of the three galaxies gn20 , 20.2a , b ( morrison et al . 2010 ) , and the optical position of one lbg with a spectoscopic redshift in the redshift range of the co 2 - 1 observations ( section 5.2 ) . we have added the previously published d array observations ( carilli et al . 2010 ) to improve signal - to - noise . unfortunately , the previous data did not have the exact same velocity coverage , and hence the velocity - integrated flux densities are not accurate . hence , this image acts as a finding chart for the co emitting regions , but accurate line fluxes will be based on subsequent analysis of the new evla data itself . we detect co 2 - 1 emission from gn20 , gn20.2a , and gn20.2b . gn20 is extended on a scale @xmath5 ( see below ) . gn20.2a is not spatially resolved by these observations . gn20.2b may be extended , although the signal - to - noise of these data are not conclusive . figure 2 shows spectra of the three sources . for gn20 , we integrate over an area of @xmath9 . for the other two sources we only consider the spectra at the peak of the velocity integrated co image . results for gaussian fits to the lines are given in table 1 . the lines are broad for gn20 and gn20.2a ( @xmath42 km s@xmath24 ) , although not outside the range of the zero intesity line widths seen in some nearby ulirgs ( downes & solomon 1998 ) . however , we note athat , due to the bandwidth limitation for the early evla , the very lowest frequency ( highest velocity ) emission for gn20 may be truncated by these observations , although the higher order co transitions show that we expect little emission beyond our lowest spectral channel ( daddi et al . 2009a ) . gn20.2b shows a narrower profile at the peak position . however , the detection in the velocity integrated image over the full frequency range ( figure 1 ) , suggests that the emission may be extended both spatially and spectrally . a possibly broader line is also suggested for the higher order co emission , but again , the high order co detection is only marginal ( daddi et al 2009a ) . we have investigated this possibility with the current data , and conclude that more sensitive observations are required to better characterize the co emission from gn20.2b , and we simply present the spectrum at the peak position herein . figure 3 shows the velocity channel images for the gn20 co 2 - 1 emission at 78 km s@xmath24 channel@xmath24 and @xmath39 resolution . the emission clearly moves from the south to the north with increasing frequency . an interesting feature is seen in channel 6 , corresponding to a frequency of 45.598 ghz ( or @xmath43 ) . the emission in channel 6 appears to be substantially more extended than in other channels . inspecting our bandpass calibration , we find no channel - dependent calibration errors that would lead to such a difference in morphology for this single channel . in channel 6 the emission extends well to the north , with a total of extent of @xmath44 . admittedly , the signal - to - noise is not high , but this extended emission is suggestive of a gravitationally induced tidal feature . such tidal tails in nearby galaxies often occur over a narrow velocity range ( hibbard & mihos 1995 ) . figure 4a shows the co 2 - 1 emission from gn20 integrated over the velocity range shown in the channel maps of figure 3 . the emission appears extended , in particular to the southwest . a formal gaussian fit to the emission yields a total flux density of @xmath45 mjy , a peak surface brightness of @xmath46 jy beam@xmath24 , and a deconvolved source size of @xmath47 with major axis position angle pa = 69@xmath48 . figure 4b shows the iso - velocity contours ( first moment ) of the co 2 - 1 emission , ie . the weighted mean velocity derived after first blanking the channel images below 3@xmath10 per channel . there is a velocity gradient north - south , with a magnitude of @xmath49 km s@xmath24 . given the low spatial resolution of these images , this corresponds to a lower limit to the true projected rotational velocity . we derive the mass in molecular gas ( h@xmath1 ) from the observed co 2 - 1 luminosities using the standard relationships in solomon & vanden bout ( 2005 ) . we extrapolate from co 2 - 1 to 1 - 0 luminosity assuming constant brightness temperature , which is certainly correct to within 10% for these galaxies ( carilli et al . we adopt a co luminosity to h@xmath1 conversion factor of @xmath50 m@xmath4 k km s@xmath24 pc@xmath51 , appropriate for nearby nuclear starburst galaxies ( downes & solomon 1998 ) . the resulting values are listed in table 1 . admittedly , the uncertainty on @xmath52 is significant . a minimum gas mass can be derived assuming optically thin emission , and adopting a temperature and a co to h@xmath1 abundance ratio . ivison et al . ( 2010 ) show that for reasonable assumptions ( galactic co abundance in molecular clouds , and temperatures 40 k to 50 k ) , the lower limit to @xmath53 ( see also aalto et al . 1995 ) . the gas mass in gn20 is @xmath54 m@xmath4 . daddi et al . ( 2009a ) derive a stellar mass of @xmath7 m@xmath4 from ir through optical sed fitting . we can obtain a very rough estimate of the gravitational mass of gn20 from the co velocity field , although there are substantial uncertainties due to the low spatial resolution of these observations . for the rotational velocity , we adopt the extremes of the velocity channels with significant emission in figure 3 , or @xmath55 km s@xmath24 . for the radius , we adopt the value for the co ring seen in the high resolution 2 - 1 observations presented in carilli et al . ( 2010 ) , @xmath56 kpc . the inclination angle remains uncertain . the gravitational mass inside this radius is @xmath57^{-2}$ ] m@xmath4 . this naively derived dynamical mass is comparable to the sum of the stellar and gas masses . there is one lyman - break galaxy within our field with a spectroscopic redshift placing the co 2 - 1 line within our band : j123711.48 + 622155.8 at @xmath58 ( bd29079 ; daddi et al . 2009a ; shim et al . we find no co emission from this galaxy , to a @xmath59 limit of 0.17 mjy beam@xmath24 at 78 km s@xmath24 channel@xmath24 , after primary beam correction . convolving to 312 km s@xmath24 , we set a 3@xmath10 limit to the co 1 - 0 luminosity of @xmath60 , assuming constant brightness temperature from co 2 - 1 to 1 - 0 . the implied molecular gas mass of this galaxy is @xmath61 m@xmath4 . the stellar mass of this galaxy is @xmath62 m@xmath4 ( daddi et al . 2009a ) , hence the gas - to - stellar mass ratio is @xmath63 . daddi et al . ( 2009a ) estimate a star formation rate of 150 m@xmath4 year@xmath24 for bd29079 , or an ir luminosity of @xmath64 l@xmath4 . daddi et al . ( 2010 ) have considered the star formation law in high @xmath65 galaxies , ie . the relationship between ir luminosity and co luminosity , and find for galaxies of this luminosity at @xmath66 : @xmath67 . adopting this relation implies an expected co luminosity of @xmath68 l@xmath4 . hence , bd29079 is under - luminous in co emission relative to the standard star formation law , although within the broad ( factor three ) scatter at any given luminosity . there are 11 other lbgs within our field , and with photometric redshifts that are consistent with the sampled range , although the @xmath69 error bars are such that the galaxies could be outside the velocity range of the co observations . we have looked for co emission at all these positions , and find no significant detections to similar limits to those quoted above . we emphasize that there is a large uncertainty in the derived h@xmath1 mass for the lbgs , due to the unknown conversion factor , @xmath52 . it is possible that metal poor star forming galaxies have a much larger value of @xmath52 than ulirgs or even the milky way due to co dissociation by the more pervasive uv radiation field ( madden et al . 1997 ; papadopoulos & pelupessy 2010 ) . we have detected co 2 - 1 emission from three galaxies in the gn20 proto - cluster at @xmath70 using the evla . the molecular gas masses range from @xmath71 m@xmath4 to @xmath54 m@xmath4 . hence , we are observing a group of molecular gas rich galaxies undergoing extreme starburts within 1.6 gyr of the big bang . gn20 presents a particularly interesting case , given its very high luminosity , and large spatial extent . we find that the sum of the stellar and gas mass in gn20 is comparable to the dynamical mass . while the assumptions involved in all three mass calculations are uncertain , it appears that the baryons likely dominate the mass content within 5 kpc radius of gn20 . moreover , their is little room for a co luminosity to gas mass conversion factor substantially larger than 0.8 m@xmath4 k km s@xmath24 pc@xmath51 . high resolution observations are required to confirm the rotational aspect of the emission , and to obtain a more accurate derivation of the dynamical mass . the morphology and velocity field for gn20 suggest a rotating disk , with a possible tidal tail extending 10kpc to the north in a single 78 km s@xmath24 channel . we can speculate that gn20 is a gas rich disk galaxy that has been gravitationally torqued by its neighboring galaxies . this gravitational disturbance has not distrupted the disk , but has greatly enhanced star formation in the disk , as well as generating the extended , tidal gas distribution . lastly , we have searched for co emission from lbgs in the gn20 protocluster . we set an upper limit of @xmath72 m@xmath4 to the molecular gas mass in these galaxies , including one with an accurate spectroscopic redshift . our limit is below the expected value based on the standard star formation law at this redshift , but within the broad scatter . we are obtaining high resolution co imaging observations of the gn20 proto - cluster with the evla and the plateau de bure interferometer to perform a more detailed dynamical analysis , and study the spatial distribution of the co excitation and gas - to - dust ratio of this forming cluster of massive galaxies at @xmath70 . we thank the referee for useful comments . ed acknowledges the funding support of erc - stg - upgal-240039 , and anr-08-jcjc-0008 . dr acknowledges support from from nasa through hubble fellowship grant hst - hf-51235.01 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 . & velocity@xmath73 & lco & m(h@xmath1 ) + & & mjy & km s@xmath24 & jy km s@xmath24 & km s@xmath24 & k km s@xmath24 pc@xmath51 & m@xmath4 + gn20 & 4.0554 & [email protected] & 679@xmath7488 & [email protected] & -11@xmath7432 & @xmath75 & @xmath76 + gn20.2a & 4.0508 & [email protected] & 723@xmath74110 & [email protected] & -283@xmath7445 & @xmath77 & @xmath78 + gn20.2b & 4.0567 & [email protected] & 290@xmath7470 & [email protected] & 67@xmath7429 & @xmath79 & @xmath80 +	we present observations of the molecular gas in the gn20 proto - cluster of galaxies at @xmath0 using the expanded very large array ( evla ) . this group of galaxies is the ideal laboratory for studying the formation of massive galaxies via luminous , gas - rich starbursts within 1.6 gyr of the big bang . we detect three galaxies in the proto - cluster in co 2 - 1 emission , with gas masses ( h@xmath1 ) between @xmath2 and @xmath3 m@xmath4 . the emission from the brightest source , gn20 , is resolved with a size @xmath5 , and has a clear north - south velocity gradient , possibly indicating ordered rotation . the gas mass in gn20 is comparable to the stellar mass ( @xmath6 m@xmath4 and @xmath7 m@xmath4 , respectively ) , and the sum of gas plus stellar mass is comparable to the dynamical mass of the system ( @xmath8^{-2}$ ] m@xmath4 ) , within a 5kpc radius . there is also evidence for a tidal tail extending another @xmath9 north of the galaxy with a narrow velocity dispersion . gn20 may be a massive , gas rich disk that is gravitationally disturbed , but not completely disrupted . there is one lyman - break galaxy ( bd29079 ) in the gn20 proto - cluster with an optical spectroscopic redshift within our search volume , and we set a 3@xmath10 limit to the molecular gas mass of this galaxy of @xmath11 m@xmath4 .
sax j1808.43658 ( sax j1808 in the following ) was the first neutron star transient during whose outbursts coherent pulsations were detected ( wijnands & van der klis 1998 ) . this source , together with six other low mass transients ( see wijnands 2005 for a review ) , form a separate ( sub-)class within neutron star transients ( see campana et al . 1998a for a review ) . distinctive properties , besides coherent pulsations , are weaker outburst peak luminosities ( @xmath0 ) , very low mass companions ( mass functions @xmath1 ) , short orbital periods ( @xmath2 hr ) , very faint quiescent luminosities ( @xmath3 ) and absence of a soft x ray spectral component in quiescence . a further difference with respect to ` classical ' neutron star transients concerns the return to quiescence following an outburst . the best light curve to date of ` classical ' neutron star transients is represented by bepposax observations of aql x-1 ( campana et al . 1998b ) . below a luminosity level of @xmath4 aql x-1 turned off with an exponential decay with an @xmath5folding time of @xmath6 d. the source then remained quiescent for at least one month following the 1998 outburst ( campana et al . 1998b ) and for five months following the 2000 outburst ( rutledge et al . 2001 ; campana & stella 2003 ) . the return to quiescence for transient pulsating neutron stars has been monitored for sax j1808 ( wijnands et al . 2003 ; wijnands 2005 ) and for xte j1751305 ( markwardt et al . 2002 ) and swift j1756.92508 ( krimm et al . the behaviour of the latter two sources was somewhat similar to aql x-1 with a fast decay , even though in the case of xte j1751305 there was still activity 15 d after the transition to quiescence had occurred ( according to rxte / pca , corresponding to a 0.510 kev unabsorbed luminosity of @xmath7 for a source distance of 7 kpc ) . in case of sax j1808 the decay was completely different from that observed in aql x-1 . the most striking feature is the strong erratic variability by a factor of @xmath8 . this erratic behaviour was not concentrated in the first few weeks after turn off but lasted for months ( e.g. wjinands et al . the most likely explanation for this behaviour comes from the similarities with dwarf novae where a similar behaviour has been observed ( kato et al . 2004 ; patterson et al . 2002 ) and to a lesser extent in black hole transients ( kuulkers , howell & van paradijs 1996 ) . in the case of dwarf novae this has been interpreted as due to a combination of short orbital period and low mass ratio ( osaki & meyer 2002 ; truss et al . 2002 ; campana et al . 2008 , in preparation ) . here we report on a monitoring campaign carried out with swift , following the 2005 outburst . at variance with previous monitoring with rxte / pca we are able to follow the source down to very low levels , a factor of @xmath9 fainter than before , close to quiescence . in sect . 2 we discuss the data and their analysis . sect . 3 contains our discussion and conclusions . swift carried out 23 observations of sax j1808 between jun 17 , 2005 and oct 28 , 2005 . xrt collected a total of 4899 s data in window timing ( wt ) mode and 72922 s data in photon counting mode ( see table 1 ) . in wt mode a 1d image is obtained reading data compressing along the central 200 pixels in a single raw . pc mode produces standard 2d images ( for more details see hill et al . wt data were mainly collected during the early stages of the outburst when the source was brighter . pc data collected during this period are piled - up . [ cols="^,^,^,^,^,^ " , ] at lower luminosities the number of collected photons is very small and we have to stack together different observations . one important consideration comes from the observation that 8 out of 13 observations found the source with a count rate in the narrow interval @xmath10 c ks@xmath11 , whereas the other 5 all lie within @xmath12 c ks@xmath11 . the mean of the first group is @xmath13 c ks@xmath11 and of the second is @xmath14 c ks@xmath11 . this is rather peculiar and hints for the existence of two different states in the deep faint end tail of the outburst . due to the small number of photons these spectra can be easily fit with single power law models . for state a we derive a photon index of @xmath15 and a mean unabsorbed 0.310 kev luminosity of @xmath16 , for state b we have @xmath17 and a luminosity of @xmath18 . these values have been derived binning the data to 5 photons per energy bin and applying the churazov weighting in the fitting procedure ( see fig . this indicates that state b is not yet the true quiescent state , characterized by a single power law spectrum with index @xmath19 and a source luminosity of @xmath20 ( heinke et al . 2007 ) , i.e. a factor of @xmath21 smaller that state b luminosity . assuming this power law component as a stable component ( at least in photon index ) present in state a and b spectra , one can investigate if a black body component is present . fitting state a spectrum with a fixed power law photon index ( with @xmath22 ) and a free black body component , we derive @xmath23 kev and a radius of @xmath24 km . the improvement over the simple power law fit is at a level of @xmath25 by means of an f - test . for state b we are not able to constrain the black body component . however , if we assume the same black body temperature as in state a ( i.e. 0.2 kev ) we can derive an upper limit on its radius of @xmath26 km , indicating that if present and at the same temperature it must be smaller in size . sax j1808 is the prototype of the accreting millisecond x ray pulsar ( amp ) class . the source underwent several outbursts to date which have been monitored in great details thanks to rxte observations ( wijnands 2005 ; wijnands et al . these observations however were limited to the brightest part of the outburst , being the pca on rxte a collimated instrument ( and therefore heavily background limited ) . here we report on the first campaign aimed at studying the faintest tail of the outburst and return to quiescence . thanks to the fast repointing and flexible scheduling capabilities of the swift satellite , we monitored the return to quiescence of sax j1808 during the 2006 outburst . the bright phase of the outburst is similar to what has been observed in the past : following the peak and the smooth decay a flaring behaviour sets in with at least three rebrightening episodes . after this the source turns to quiescence even if same low level activity ( factor of @xmath27 ) is still present . the puzzling result of this observational campaign on sax j1808 is that out of 13 observations with luminosity below @xmath28 , 8 times the source was found with a luminosity around @xmath29 ( see fig . 1 ) , i.e. at a luminosity level @xmath30 times higher than the true quiescent level ( campana et al . 2002 ; heinke et al . 2007 ) . given the wild luminosity variations this is somewhat strange and calls for the presence of a ` metastable ' state . this luminosity level is difficult to interpret in case of standard disk accretion but finds a natural explanation if the accretion process is mediated by the magnetic field of the fast spinning neutron star . when the magnetospheric radius ( i.e. the radius at which the neutron star magnetic field starts controlling the accretion flow ) is larger than the corotation radius , the system is in the ( so - called ) propeller regime . in this case matter is halted from the rotating magnetosphere and does not accrete onto the neutron star surface ( see e.g. campana & stella 2000 and references therein ) . when the magnetospheric radius becomes larger than the light cylinder radius ( @xmath31 , where @xmath32 is the velocity of light and @xmath33 the spin period ) , the magnetic field can not corotate any more with the neutron star and a dipole losses will take place . the luminosity range spanned in the propeller regime is @xmath34 ( where @xmath35 is the spin period in units of 2.5 ms and @xmath36 the neutron star mass in units of @xmath37 ; see campana & stella 2000 ) . in order to have the ` metastable ' state to coincide with the lowest luminosity in the propeller regime ( i.e. just before the reactivation of the pulsar ) , we have to require a magnetic field of @xmath38 g , in line with previous estimates : @xmath39 g using disk - magnetosphere interaction models ( psaltis & chakrabarty 1999 ) , @xmath40 g using simple considerations on the position of the magnetospheric radius during quiescent periods ( di salvo & burderi 2003 ) ; @xmath41 g modelling period changes during outbursts with a magnetic torque model ( hartman et al . 2008 ) . the lowest luminosity level in the proposed interpretation is ascribed to the interaction of a turned - on pulsar with the interbinary and circumstellar environment . variability is expected at this stage due to the rapidly changing environment as observed , e.g. , in the millisecond radio pulsar psr j17405340 ( damico et al . 2001 ; ferrario et al . 2001 ) which gets eclipsed over a range of orbital phases for different orbital cycles .	the transient accreting millisecond pulsar sax j1808.43658 has shown several outbursts to date but the transition from outburst to quiescence has never been investigated in detail . thanks to the swift observing flexibility , we monitored for the first time the decay to quiescence during the 2005 outburst . at variance with other transients , wide luminosity variations are observed . in addition , close to quiescence , sax j1808.43658 seems to switch between two different states . we interpret them in terms of the accretion states accessible to a magnetized , fast rotating neutron star .
lovelock gravity is one of the natural generalization of einstein s general relativity , introduced by david lovelock @xcite , the action of which contains higher - order curvature terms . the lovelock action in @xmath1-dimensional spacetime reads @xmath2}\alpha_{(p)}{\mathcal l}_{(p)}+i _ { matter},\\ { \mathcal l}_{(p)}:=&\frac{1}{2^p}\delta^{\mu_1\cdots \mu_p\nu_1\cdots \nu_p}_{\rho_1\cdots \rho_p\sigma_1\cdots \sigma_p}r_{\mu_1\nu_1}^{\phantom{\mu_1}\phantom{\nu_1}\rho_1\sigma_1}\cdots r_{\mu_p\nu_p}^{\phantom{\mu_p}\phantom{\nu_p}\rho_p\sigma_p},\end{aligned}\ ] ] where @xmath3 is a constant related to @xmath4 via @xmath5 with @xmath6 , the coupling constant @xmath7 has dimension of @xmath8 , and @xmath9 is the euler density of a 2@xmath10-dimensional manifold . the symbol @xmath11 describes a totally antisymmetric product of kronecker deltas , normalized to take values @xmath12 and @xmath13 @xcite , defined by @xmath14},\end{aligned}\ ] ] where @xmath15 is related to the cosmological constant @xmath16 by @xmath17 . the lovelock action @xmath18 reduces to the einstein - hilbert action in four dimensions , and its second term is the gauss - bonnet invariant . lovelock theories are distinct , among the larger class of general higher - curvature theories , in having field equations involving not more than second derivatives of the metric . consequently , lovelock gravity theories are free from many of the pathologies that affect general higher derivative gravity theories . as higher - dimensional members of einstein s general relativity family , lovelock gravities allow us to explore several conceptual issues of gravity in depth in a broader setup . most interestingly , one can include features of black holes such as their existence and uniqueness theorems , their thermodynamics , the definitions of their mass and entropy , their horizon properties , etc . also , such a theory may be used in the context of the ads / cft correspondence to investigate the effects of including higher - curvature terms @xcite . it has , therefore , been explored to a large extent , also possibly for its appearance in strings theories at low energies @xcite . in this paper , we will be concerned with the black hole solutions of this lovelock theory , and we will discuss how higher - curvature corrections to black hole physics substantially change the qualitative features we know from our experience with black holes in general relativity . since their inception , steady attention has been devoted to black hole solutions , including their formation , stability , and thermodynamics . the spherically symmetric static black hole solution for second - order lovelock gravity ( the theory that is usually referred as the einstein - gauss - bonnet theory ) was first obtained by boulware and deser @xcite , and this kind of solution for third - order lovelock gravity was introduced in ref . exact black hole solutions of the former can be found in ref . @xcite and the latter in refs . the black hole solution in einstein - gauss - bonnet theory in a string cloud model was considered in refs . @xcite . the recent theoretical developments signal toward a scenario in which the fundamental building blocks of the universe are extended objects instead of point objects and have been considered quite seriously @xcite . the most natural and popular candidate is one - dimensional strings object . this resulted in the intense level of activity towards the study of the gravitational effects of matter in the form of clouds of both cosmic and fundamental strings @xcite . in addition , the intense level of activity in string theory has led to the idea that many of the classic vacuum schemes , such as the static schwarzschild black hole ( point mass ) , may have atmospheres composed of a fluid or field of strings . further , this two - fluid atmosphere can model an array of physical situations at diverse distance scales , which can depict the atmosphere around a black hole with a distance scale of multiples of schwarzschild radii . it could also describe a globular cluster with components of a dark matter at a scale tuned to the order of parsecs . the event horizon for the classical schwarzschild metric in the background of a cloud of strings has a modified radius @xmath19 with @xmath20 as a string cloud parameter @xcite , thereby enlarging the schwarzschild radius of the black hole by the factor @xmath21 , which may have several astrophysical consequences , e.g. , on a wormhole @xcite . further , glass and krisch @xcite have demonstrated that allowing the schwarzschild mass to be a function of radial position builds an atmosphere with a string fluid stress energy around a static , spherically symmetric , object . thus , the study of einstein s equations coupled with a string cloud , in both general relativity and modified theories , may be very important because relativistic strings at a classical level can be used to construct applicable models @xcite . intense activity of studying black hole solutions in modified theories of gravity including lovelock theories of gravity is due to the fact that , besides theoretical results , cosmological evidence , e.g. , dark matter and dark energy , the possibility of modification of the einstein gravity is suggested . many other authors have found exact black hole solutions with a string cloud background , for instance , in general relativity @xcite , in einstein - gauss - bonnet gravity @xcite , and also in lovelock gravity @xcite . it is the purpose of this paper to obtain an exact black hole solution in second- and third - order lovelock theories with a cloud of strings in the background . we shall present a class of black hole solutions endowed with a cloud of strings . in particular , we explicitly bring out how the effect of the background string cloud can alter black hole solutions and their properties as we know from our knowledge of black holes in general relativity . our attention will be given to the second - order einstein - gauss - bonnet case , which exhibits most of the relevant qualitative features . we obtain @xmath22-dimensional static spherically symmetric black hole solutions in a string cloud background with the three terms of lovelock gravity that are einstein or general relativity , gauss - bonnet and third - order lovelock terms . we analyze their thermodynamic properties and also perform a stability analysis the for einstein - gauss - bonnet case . lovelock gravity is the most general second - order gravity theory in higher - dimensional spacetimes , which is free of ghosts @xcite . the lovelock tensor is nonlinear in the riemann tensor and nontrivially differs from the einstein tensor only if @xmath23 . the third - order lovelock action with matter in @xmath23 dimensions reads @xcite@xmath24 + \mathcal{i}_{s}.\ ] ] with @xmath25 . the einstein term @xmath26 , the second - order lovelock ( gauss - bonnet ) term @xmath27 is @xmath28 and @xmath29 is the third - order lovelock lagrangian . here , @xmath30 , @xmath31 , and @xmath32 are the ricci tensors , riemann tensors , and ricci scalar , respectively . the variation of the action with respect to the metric @xmath33 gives the einstein - gauss - bonnet - lovelock equations,@xmath34 where @xmath35 is the einstein tensor , while @xmath36 and @xmath37 are given explicitly , respectively , by @xcite@xmath38 and @xmath39 and @xmath40 is the energy - momentum tensor of matter that we consider as a cloud of strings . note that for third - order lovelock gravity , the nontrivial third term requires that the dimension of spacetime should satisfy @xmath41 . let us consider a cloud of strings as matter . for completeness , we give a brief review of the theory of a cloud of strings ( see ref . @xcite for further details ) . the nambu - goto action of a string evolving in spacetime is given by @xmath42 where @xmath43 is a positive constant , @xmath44 and @xmath45 being timelike and spacelike parameters @xcite . the string world sheet @xmath46 is given by @xmath47 and @xmath48 = det @xmath49 . associated with the strings world sheet , we have the bivector of the form @xmath50where @xmath51 denotes the two - dimensional levi - civit@xmath52 tensor given by @xmath53 . within this setup , the lagrangian density becomes @xmath54^{1/2}.\ ] ] further , since @xmath55 , we obtain the energy - momentum tensor for one string as @xmath56 hence , the energy - momentum tensor for a cloud of strings is @xmath57where @xmath58 is the proper density of a string cloud and quantity @xmath59 is the gauge - invariant density . the strings is characterized by a surface - forming bivector @xmath60 and conditions to be a surface - forming are @xmath61 } = 0 , \nonumber \\ & & \nabla_{\mu } \sigma^{\mu [ \alpha } \sigma^{\beta \gamma ] } = 0,\end{aligned}\ ] ] where the square brackets in eq . ( [ eq : imp ] ) indicate antisymmetrization . the above equation , in conjunction with eq . ( [ eq : bivector ] ) , leads to the useful identity @xmath62 which will be used in subsequent calculations . further , the conservation of energy - momentum tensor @xmath63 implies that @xmath64 which upon multiplication by @xmath65 leads to @xmath66 . contracting the previous identity with @xmath67 and using eq . ( [ eq : usable ] ) , we obtain @xmath68 , and finally adapting to parametrization , we get @xmath69 here , we want to obtain @xmath22-dimensional static spherically symmetric solutions of eq . ( [ ee ] ) with a cloud of strings as source and investigate its properties . we assume that the metric has the form @xmath70 where @xmath71 is the metric of a @xmath72-dimensional constant curvature space @xmath73 or -1 . in this paper , we shall stick to @xmath74 . to find the metric function @xmath75 , we consider the components of eq . ( [ ee ] ) . using this metric _ ansatz _ , the einstein - gauss - bonnet - lovelock @xmath76 equation of motion ( [ ee ] ) reduces to @xmath77 f^{\prime}\left ( r\right ) + \nonumber\\ & & \left ( n-1\right ) r^{4}\left ( f\left ( r\right ) -1\right ) -\left ( n-3\right ) \tilde{\alpha}_{2}r^{2}\left ( f\left ( r\right ) -1\right ) ^{2}+\nonumber\\ & & \left ( n-5\right ) \tilde{\alpha}_{3}\left ( f\left ( r\right ) -1\right)^{3 } = \frac{2r^6 } { n}t^r_r,\end{aligned}\ ] ] in which a prime denotes a derivative with respect to @xmath78 @xmath79 @xmath80 @xmath81 , and @xmath82 @xmath83 . in general , eq . ( [ master ] ) has one real and two complex solutions . it may have three real solutions as well under some conditions . here , we consider only the real solution . here , the density @xmath58 and the bivector @xmath84 are the functions of @xmath85 only as we seek static spherically symmetric solutions . the only surviving component of the bivector @xmath46 is @xmath86 . thus , @xmath87 , and from eq . ( [ eq : div ] ) , we obtain @xmath88 , which implies @xmath89 for some real constant @xmath20 . in the rest frame associated with the observer , the energy density of the matter will be given by @xmath90 the energy - momentum of the source can be written as @xmath91 . \label{emt}\ ] ] the weak energy condition demands that for every timelike vector field -the matter density observed by the corresponding observer is always non - negative : @xmath92 the strong energy condition requires @xmath93 clearly , all energy conditions hold . in three and four dimensions , lovelock theory coincides with einstein theory @xcite , e.g. , for @xmath94 , we get@xmath95 which clearly is independent of @xmath96 and @xmath97 , and therefore it will be the einstein equation in four dimensions admitting the solution @xmath98 this solution was first obtained by letelier @xcite , and the metric represents the black hole spacetime associated with a spherical mass @xmath43 centered at the origin of the system of coordinates , surrounded by a spherical cloud of strings . the event horizon of the metric is placed at @xmath99 . in the limit @xmath100 , we recover the schwarzschild radius , and close to unity . the event horizon radius tends to infinity . on the other hand , a cloud of strings alone ( @xmath101 ) does not have a horizon ; it represents only a naked singularity at @xmath102 . besides , the metric ( [ metric ] ) with ( [ lfour dimensions ] ) can be understood as the metric associated with a global monopole . but in higher dimensions , the lovelock theories are actually different . in fact , for @xmath103 , einstein gravity can be thought of as a particular case of lovelock gravity since the einstein - hilbert term is one of several terms that constitute the lovelock action . hence , for @xmath103 and @xmath104 , we obtain @xmath105 which admits the solution @xmath106 we also observe that the four - dimensional solution ( [ lfour dimensions ] ) is recovered in the limit @xmath107 . the integration constant @xmath43 in eq . ( [ ndsol ] ) is related to the arnowitt - deser - misner ( adm ) mass @xmath108 via @xmath109 where @xmath110 is the volume of the @xmath72-dimensional unit sphere . to study the general structure of the solution given by ( [ eehd ] ) , we look for the essential singularity . it is seen that the kretschmann scalar ( @xmath111 ) for the metric ( [ metric ] ) reduces to @xmath112which on inserting ( [ eehd ] ) becomes @xmath113 the kretschmann scalar ( [ ks ] ) diverges as @xmath114 , indicating the scalar polynomial or essential singularity at @xmath115 . it is interesting to see that the metric ( [ metric ] ) is well behaved even if @xmath116 , as ( [ ks ] ) indicates . in the higher - dimensional case , a fact that deserves to be emphasized is that a cloud of strings alone , unlike in four dimensions , can have an event horizon located at @xmath117 . thus , we have extended the letelier @xcite solutions to higher - dimensional spacetime . now , we look for the existence of event horizons , and therefore for possible black hole solutions . the horizons , if they exist , are given by zeros of @xmath118 . the black hole horizon of the solution ( [ eehd ] ) is located at , e.g. , in the five - dimensional case , @xmath119 and at @xmath120 with @xmath121 for six - dimensional case . we note that the gravitational mass of a black hole is determined by @xmath122 , which , from equation ( [ ndsol ] ) , reads @xmath123equation ( [ mass ] ) , takes the form of the @xmath22-dimensional schwarzschild black hole when @xmath124 . in @xmath22-dimensional spacetime , we can see the roles of both second- and third - order lovelock parameters simultaneously . a simplifying , yet interesting case in the solution ( [ sol : ll ] ) can be obtained if @xmath125 the metric function @xmath75 reads simply@xmath126 with @xmath127 in figs . [ ellf(r ) ] and [ ellf(r)1 ] , we have plotted @xmath75 for the ell case in various dimensions . figure [ ellf(r ) ] shows the variation of @xmath75 taking fixed @xmath128 and varying @xmath129 ; fig . [ ellf(r)1 ] shows the variation of @xmath75 taking fixed @xmath129 and varying @xmath128 . 99 d. lovelock , j. math . ( n.y . ) * 12 * , 498 ( 1971 ) ; n. deruelle and l. farina - busto , phys . rev . d * 41 * , 3696 ( 1990 ) ; g. a. menamarugan , phys . rev . d * 46 * , 4320 ( 1992 ) ; * 46 * , 4340 ( 1992 ) ; d. g. boulware and s. deser , phys . rev . lett . * 55 * , 2656 ( 1985 ) ; b. zwiebach , phys . b * 156 * , 315 ( 1985 ) ; r. g. cai and n. ohta , phys . rev . d * 74 * , 064001 ( 2006 ) ; a. chakrabarti and d. h. tchrakian , phys d * 65 * , 024029 ( 2001 ) . d. kastor and r.b . mann , j. high energy phys . 04 , ( * 2006 * ) 048 . m. fukuma , s. matsuura , and t. sakai , prog . . phys . * 105 * , 1017 ( 2001 ) . d. j. gross and e. witten , nucl . b277 * , 1 ( 1986 ) ; d. j. gross and j. h. sloan , nucl . phys . * b291 * , 41 ( 1987 ) ; r. r. metsaev and a. a. tseytlin , phys . b * 191 * , 354 ( 1987 ) ; b. zwiebach , phys . b * 156 * , 315 ( 1985 ) ; r. r. metsaev and a. a. tseytlin , nucl . phys . * b293 * , 385 ( 1987 ) . d.g . boulware and s. deser , phys . * 55 * , 2656 ( 1985 ) ; j.t . wheeler , nucl . phys . * b268 * 737 ( 1986 ) ; r.c . myers and j.z . simon , phys . d * 38 * , 2434 ( 1988 ) . m. h. dehghani and m. shamirzaie , phys . d * 72 * , 124015 ( 2005 ) . y. m. cho and i. p. neupane , phys . d * 66 * , 024044 ( 2002 ) ; i. p. neupane , phys . d * 67 * , 061501(r ) ( 2003 ) ; * 69 * , 084011 ( 2004 ) ; r. g. cai , phys . d * 65 * , 084014 ( 2002 ) ; r. g. cai and q. guo , phys . d * 69 * , 104025 ( 2004 ) ; t. torii and h. maeda , phys . d * 71 * , 124002 ( 2005 ) ; m. h. dehghani , phys . rev . d * 69 * , 064024 ( 2004 ) ; m. h. dehghani and r. b. mann , phys . d * 72 * , 124006 ( 2005 ) ; m. h. dehghani and s. h. hendi , phys d * 73 * , 084021 ( 2006 ) ; m. h. dehghani , g. h. bordbar , and m. shamirzaie , phys . d * 74 * , 064023 ( 2006 ) ; m. h. dehghani and s. h. hendi , int . j. mod . d * 16 * , 1829 ( 2007 ) ; a. padilla , classical quantum gravity * 20 * , 3129 ( 2003 ) ; n. deruelle , j. katz , and s. ogushi , classical quantum gravity * 21 * , 1971 ( 2004 ) ; m. cvetic , s. nojiri and s. d. odintsov , nucl . b628 * , 295 ( 2002 ) . m. h. dehghani and r. b. mann , phys . d * 72 * , 124006 ( 2005 ) ; m. h. dehghani , n. alinejadi , and s. h. hendi , phys . rev . d * 77 * , 104025 ( 2008 ) ; s. h. hendi and m. h. dehghani , phys . b * 666 * , 116 ( 2008 ) . s. h. mazharimousavi , o. gurtug , and m. halilsoy , int . d * 18 * , 2061 ( 2009 ) . s. h. mazharimousavi and m. halilsoy , phys . b * 681 * , 190 ( 2009 ) . e. herscovich and m. g. richarte , phys . b * 689 * , 192 ( 2010 ) . j. l. synge , _ relativity : the general theory _ , ( north holland , amsterdam , 1966 ) , p. 175 . p. letelier , phys . d * 20 * , 1294 ( 1979 ) . s. h. mazharimousavi , o. gurtug , and m. halilsoy , classical quantum gravity * 27 * , 205022 ( 2010 ) . s. g. ghosh and s. d. maharaj , phys . d , * 89 * , 084027 ( 2014 ) . m. g. richarte and c. simeone , int . j. mod . d * 17 * , 1179 ( 2008 ) . e.n . glass and j.p . krisch , phys . d * 57 * , 5945 ( 1998 ) ; classical quantum gravity * 16 * , 1175 ( 1999 ) ; j. math . phys . * 40 * , 4056 ( 1999 ) . x. o. camanho and j. d. edelstein , classical quantum gravity * 30 * , 035009 ( 2013 ) . c. sahabandu , p. suranyi , c. vaz , and l. c. r. wijewardhana , phys . d * 73 * , 044009 ( 2006 ) . j. d. bekenstein , phys . d * 7 * , 2333 ( 1973 ) .	lovelock theory is a natural extension of the einstein theory of general relativity to higher dimensions in which the first and second orders correspond , respectively , to general relativity and einstein - gauss - bonnet gravity . we present exact black hole solutions of @xmath0-dimensional spacetime for first- , second- , and third - order lovelock gravities in a string cloud background . further , we compute the mass , temperature , and entropy of black hole solutions for the higher - dimensional general relativity and einstein - gauss - bonnet theories and also perform thermodynamic stability of black holes . it turns out that the presence of the gauss - bonnet term and/or background string cloud completely changes the black hole thermodynamics . interestingly , the entropy of a black hole is unaffected due to a background string cloud . we rediscover several known spherically symmetric black hole solutions in the appropriate limits .
according to the standard model ( sm ) , in @xmath0 collisions at the tevatron top quarks can be created in pairs via the strong force , or singly via the electroweak interaction . due to its higher cross - section and smaller associated backgrounds , the strong production ( @xmath1 ) represents the main channel used for performing detailed measurements of the properties of the top quark . in terms of its decay , a top quark is expected to disintegrate into a @xmath2 boson and a @xmath3-quark almost 100% of the time . thus , the top candidate events can be classified based on the subsequent decays of the @xmath2 daughter into : a ) `` dilepton '' events if both @xmath2 bosons decay leptonically , b ) `` lepton+jets '' events if one @xmath2 decays leptonically and the other one hadronically , and c ) `` all hadronic '' events if both @xmath2 s decay hadronically . the last category of top candidates ( all hadronic ) are typically dominated by multijet background contributions , and will not be used in the results described here . in the following sections , we will present the results from top charge measurement , as well as measurements related to the production and the decay of top quarks . while the top charge is expected to be @xmath4 , there are models which include an exotic @xmath5 top quark that couples to a right - handed @xmath3 quark @xcite and decays into a negative @xmath2 boson and a @xmath3 quark . both the cdf and d experiments have performed measurements to test if the top charge is @xmath6 or @xmath5 . the challenge is therefore to reconstruct the charges of the @xmath2 bosons and the @xmath3 quarks in the event , and subsequently pair the @xmath2 and the @xmath3 s to reconstruct the top decay chain . in the cdf analysis , dilepton and lepton+jets top candidate events are selected by first requiring evidence for the leptonic decay of a @xmath2 : an electron ( muon ) with large transverse energy @xmath7(@xmath8)@xmath920 gev and large transverse missing energy from the undetected neutrino : @xmath1020 gev . for the lepton+jets subsample , three jets with @xmath1120 gev and a fourth jet with @xmath1112 gev are required . to further reduce background contamination , at least two of these jets are required to be identified ( tagged ) as @xmath3-jets using displaced vertex information as measured by the cdf silicon vertex detector . in the case of the dilepton subsample , the presence of another high transverse momentum , opposite - sign lepton is demanded , along with the presence of at least two jets with @xmath1115 gev . of the two jets , one or both should have been tagged as @xmath3-jets by the silicon detector . a minimum threshold of 200 gev on the total transverse energy in the event ( scalar - sum including @xmath12 ) is also imposed . the charge of the @xmath3-jets is determined by performing a @xmath13-weighted is the projection of track s momentum along the jet axis . ] sum of the charges of the tracks found within a cone of radius @xmath140.4 around the jet axis . a negative ( positive ) weighted sum indicates a @xmath3 ( @xmath15 ) quark . for real @xmath3-jets present in cdf data this method have been shown to determine the correct charge roughly 60% of the time . finally , the pairing of @xmath2 and @xmath3 in dilepton events is performed by computing the invariant mass of the lepton and @xmath3-jet ( @xmath16 ) . this mass can be abnormally high for the wrong combinations ; the cdf pairing exploits this fact , first by requiring one combination over a certain threshold and then choosing the opposite @xmath17 pairing . this procedure was estimated to have a 95% success rate , with an efficiency of 39% . in lepton+jets events , a kinematic fitter @xmath18 using reconstructed top and @xmath2 mass constraints is calculated and the lowest @xmath18 combination is retained . the correct pairing is found 86% of the time , with an efficiency of 53% . the results of this analysis are presented in figure [ chgplot ] . these results favour the sm hypothesis over the @xmath5 top charge hypothesis . the latter hypothesis is rejected at 87% confidence level ( c.l . ) . we note that a top charge measurement has previously been performed by the d collaboration @xcite with similar conclusions . at the fermilab tevatron , the top pairs are expected to originate predominantly from a @xmath19 initial state : @xmath21% @xmath22 85% , with somewhat large ( @xmath23%))uncertainties . two cdf measurements @xcite attempt to measure the production cross section ratio . the first measurement takes advantage of the large correlation between the average number of gluons and the average number of low @xmath8 charged particles present in a given sample . the @xmath2 + 0jet sample and the dijet ( 80 - 100gev ) sample are then used to extract the distributions of the low-@xmath8 track multiplicity in the no - gluon `` case , and the ' ' gluon - rich " case , respectively ( figure [ combo ] ) . these two templates are used in a simple maximum likelihood fit to derive the fractions of the no - gluon and the gluon - rich events present in the cdf data . after background subtraction is performed , it is found that @xmath24(stat)@xmath25(syst ) . the other cdf measurement of the production mechanism relies on an artificial neural network ( ann ) designed to separate events originating from @xmath19 and @xmath20 initial state . the ann uses eight variables , of which two ( six ) variables characterize the production ( decay ) of the top pair . a maximum likelihood fit yields @xmath26 = -0.075 $ ] , or @xmath26 < 0.61 $ ] at 95% c.l . the two measurements are combined to obtain @xmath26 = 0.07^{+0.15}_{-0.07}$ ] ( stat+syst ) ( see figure [ combo ] ) . the d collaboration has recently published the first measurement of the forward - backward asymmetry in top production @xcite . while a small asymmetry of the order of 5% is expected to arise from qcd calculations @xcite , several new physics phenomena could drastically increase this asymmetry @xcite . the asymmetry @xmath27 is defined as the count ratio @xmath28 , where an event is counted as forward ( backward ) if the rapidity difference @xmath29 is positive ( negative ) . this measurement is performed in the reconstructed @xmath1 rest frame . the results are shown in table [ dtab ] . while the values of the asymmetry are somewhat higher than their expected values ( a few percent , depending on jet channel ) they also have sizable uncertainties , and are therefore in statistical agreement with these expectations . .d : number of selected events and fit results in the data . [ cols="<,^,^,^",options="header " , ] [ dtab ] a similar search has recently been performed by the cdf collaboration @xcite . the @xmath27 was measured in both the reconstructed rest frame of the @xmath1 system and in the laboratory frame , and corrected to the intrinsic parton - level value to allow an easy comparison between the experimental @xmath27 result and the theoretical predictions . the results frame asymmetries is : @xmath30 . ] @xmath31(stat)@xmath32(syst ) and @xmath33(stat)@xmath32(syst ) are higher than , but consistent with the theoretical sm calculation . the d and cdf collaborations have performed searches for narrow - width , high - mass particles x decaying to top quark pairs . such particles are predicted in many extensions of the sm , such as extended gauge theories @xcite , kaluza - klein excited states of gluons or @xmath34 bosons @xcite , and topcolor @xcite to name a few . the d search uses the lepton+jets subsample . at least three high-@xmath7 jets are required to be present in every event ; at least one of these jets should be identified as a @xmath3-jet using a neural networks tagging algorithm @xcite . the four momenta of the jets , lepton , and neutrino - component of the neutrino momentum is obtained from the @xmath2 mass constraint @xmath35 . this condition yields a quadratic equation in @xmath36 ; if there are two real solutions , the smaller @xmath37 one is chosen . if there are no real solutions , then @xmath38 is set to zero . ] are used to reconstruct the invariant mass of the @xmath1 system . the left plot of figure [ d_restt ] shows the distribution of the reconstructed top pair mass @xmath39 in the lepton+@xmath404 jets channel for data ( points ) and sm expectation ( colored histograms ) . a 650 gev top - color - assisted signal @xcite is shown by the white contribution stacked on top of the sm expectation ( open histogram ) . the data are found to agree well with the sm expectation . limits are set on @xmath41 for different @xmath42 values , as shown in the right plot of figure [ d_restt ] . one can see that a leptophobic @xmath43 with a width @xmath44%@xmath45 is excluded ( 95% c.l . ) up to a mass of 760 gev . a similar search at cdf observes no significant evidence for new particles decaying to @xmath1 . using the lepton+@xmath46 jets sample , the cdf collaboration has also measured the differential top pair production cross - section in bins of @xmath39 . the result is shown in figure [ alis ] where a good agreement between cdf data ( points ) and expected distribution ( histogram ) is apparent . to quantify this agreement , a p - value " , defined as the fraction of sm simulated experiments which look as discrepant as the data or worse , is calculated ; the result of 0.45 confirms no excess over sm prediction is observed . other resonance searches in the top sector at the tevatron include @xmath47 searches and scalar top searches . the cdf collaboration has performed a search for a hypothetical new quark , @xmath47 , whose decay features the same final state particles as the sm top . the @xmath47 particle could belong to a fourth quark generation in which the mass of the @xmath48 partner is large enough to satify : @xmath49 ; in this case , @xmath47 would decay predominantly into @xmath50 ( @xmath51 ) . the variables used to discriminate the @xmath47 signal from the associated background ( which includes sm @xmath1 production ) are the @xmath47 reconstructed mass @xmath52 and the total transverse energy in the event @xmath53 . a simultaneous analysis of these variables excludes the presence of a @xmath47 signal up to a mass of 284 gev ( at 95% c.l . ) . searches for scalar top quarks ( stop ) have been performed by both the cdf and d collaborations . because of the large top mass , the mixing between the stop eigenstates can be substantial , and the lighter stop @xmath54 could be lighter than the top quark @xcite . the d search focuses on this scenario in which @xmath55 pairs are produced and subsequently decay as @xmath56 . the final state therefore resembles that of @xmath1 production , with some additional missing energy from the neutralinos . multivariate likelihood discriminants are built for six models , defined by @xmath57 and @xmath58 . in all cases , no significant evidence of a signal is found , and upper limits are set on the cross section of these models , ranging from 7 - 12 times the theoretical calculations ( at 95% c.l . ) . the cdf analysis @xcite draws similar conclusions . the v@xmath59a structure of the weak current @xmath60 requires the following helicity fractions @xmath61 , @xmath62 , and @xmath63 for the longitudinal , left - handed , and right - handed @xmath2 contributions , respectively . an @xmath64 value significantly different from zero would indicate the presence of physics beyond the sm . the d collaboration has recently published @xcite results from the measurement of the fractions of @xmath2 helicity contributions using the @xmath65 * variable . * is the angle between the top quark and the lepton from @xmath2 decay , measured in the @xmath2 boson rest frame . ] both the dilepton and the lepton+jets subsamples are used for this measurement . the @xmath66 and @xmath64 fractions are fitted simultaneously , while @xmath67 is constrained to @xmath68 . the result of the fit is shown in figure [ heli ] : @xmath69(stat)@xmath70(syst ) and @xmath71(stat)@xmath72(syst ) the cdf collaboration has employed a three - prong approach for performing this measurement . the first two methods rely on the cosine of the decay angle @xmath73 * of the charged lepton in the @xmath2 rest frame measured with respect to the direction of motion of the @xmath2 boson in the top - quark rest - frame . fitting for both @xmath66 and @xmath64 fractions simultaneously and combining the two analyses , cdf obtains @xmath74(stat+syst ) and @xmath75(stat+syst ) . the third method relies on the matrix element technique previously developed in run i @xcite . in this measurement the @xmath64 and @xmath67 fractions are fixed to zero and 1-@xmath66 , respectively . the method yields @xmath76(stat)@xmath77(syst ) . the results from the three cdf methods are summarized in figure [ heli ] . to summarize , the d and cdf results are compatible with each other and with the sm expectations . more data will be needed for a conclusive determination of the three polarization fractions . in the sm , the flavor changing neutral current ( fcnc ) @xmath78 is predicted to occur at a tiny rate @xmath79 ; however , several scenarios involving physics beyond the sm could enhance this branching ratio up to values of the order @xmath80 . thus , any sign of this rare decay at the tevatron would be an indication of new physics . a cdf search for @xmath78 decays is based on a template fit to a @xmath81 variable using kinematic constraints present in the top fcnc events ; one of these constraints is that the reconstructed mass of the @xmath82 system approach the top mass , within a small window . the subsample used for this study is the @xmath834 jets , where the @xmath34 decayed to a pair of oppositely - charged electrons or muons . the analysis of the kinematic @xmath81 reveals no significant evidence for a signal beyond the sm . a feldman - cousins approach yields the tightest constraint to date in this top decay channel : @xmath84% at 95% c.l . charged higgs bosons are predicted in many extensions of the sm , such as supersymmetry or grand unified symmetries if these bosons have massed below that of the top quark , the process @xmath85 would lead to a decrease in the sm branching ratio br(@xmath86 ) assumed to be very close to unity within the sm . the d collaboration has considered a purely leptophobic model in which br(@xmath87=100% , and the mass of the charged higgs is close to the @xmath2 boson mass . such a model would decrease the measured @xmath1 cross - section in the lepton+jets channel , while leaving the dilepton @xmath1 cross section unchanged . using the ratio between the top pair cross section measured in the dilepton and the lepton+jets subsamples , one can then derive upper bounds on br(@xmath88 . at 95% c.l . this upper limit is br(@xmath890.35 . the cdf collaboration has also search for charged higgs bosons from top decays . as opposed to the d approach , the cdf analysis seeks to exploit the difference between the dijet mass spectra from the regular @xmath2 decays @xmath90 and the charged higgs decays @xmath91 . invariant dijet mass templates are built for higgs masses ranging between 90 and 150 gev . no significant deviation from the sm is observed , and limits are set for the br(@xmath92 as function of the charged higgs mass . the 95% c.l . limits on this branching ratio vary between 0.32 ( @xmath93 gev ) and 0.08 ( @xmath94 gev ) . we presented selected results from the studies of the properties of the top quarks at the tevatron . both the d and the cdf collaborations have very mature and diversified top physics programs . the large amounts of data accumulated by the two experiments allow an unprecedented precision for most sm and beyond the sm measurements in the top sector . for updates of these results and numerous other top - quark - related measurements , we invite the reader to consult the d and cdf top group webpages @xcite . 9 d. chang , w. chang , e. ma , phys . d 59 , 091503 ( 1999 ) . v. m. abazov @xmath95 @xmath96 ( the d collaboration ) , phys . 98 , 041801 ( 2007 ) . t. aaltonen @xmath95 @xmath96 ( the cdf collaboration ) , arxiv:0807.4262v1 ( 2008 ) . v. m. abazov @xmath95 @xmath96 ( the d collaboration ) , phys . 100 , 142002 ( 2008 ) . j. h. kuhn and g. rodrigo , phys . lett . * 81 * , 49 ( 1998 ) ; j. h. kuhn and g. rodrigo , phys . d * 59 * , 054017 ( 1999 ) ; m. t. bowen , s. ellis , and d. rainwater , phys . d 73 , 014008 ( 2006 ) ; s. dittmaier , p. uwer , and s. weinzierl , phys . 98 , 262002 ( 2007 ) ; l. g. almeida , g. sterman and w. vogelsang , phys . d * 78 * , 014008 ( 2008 ) . o. antunano , j. h. kuhn and g. rodrigo , phys . d * 77 * , 014003 ( 2008 ) ; j. rosner , phys . lett . b 387 , 113 ( 1996 ) ; p. frampton and s. glashow , phys . lett . b 190 , 157 ( 1987 ) ; l. sehgal and m. wanninger , phys . b 200 , 211 ( 1988 ) . t. aaltonen @xmath95 @xmath96 ( the cdf collaboration ) , arxiv:0806.2472v2 ( 2008 ) . a. leike , phys . 317 , 143 ( 1999 ) . b. lillie , l. randall , and l .- t . wang , jhep 09 , 074 ( 2007 ) . t. g. rizzo , phys . d 61 , 055005 ( 2000 ) . c. t. hill and s. parke , phys d 49 , 4454 ( 1994 ) . t. scanlon , fermilab - thesis-2006 - 43 . c. balazs , m. carena , and c. wagner , phys 015007 ( 2004 ) . w. beenakker @xmath95 @xmath96 , nucl . b , 515 , 3 ( 1998 ) . r. erbacher , a. ivanov , w. johnson , public cdf note 9343 available at : + ` http://www-cdf.fnal.gov/physics/new/top/2008/tprop/stop/stopdilcdfnote.pdf ` v. m. abazov @xmath95 @xmath96 ( the d collaboration ) , phys . 100 , 062004 ( 2008 ) . v. m. abazov @xmath95 @xmath96 ( the d collaboration ) , phys . b617 , 1 ( 2005 ) ; m. f. canelli , fermilab - thesis-2003 - 22 . d top quark public webpage : ` http://www-d0.fnal.gov/run2physics/top/index.html ` ; cdf top quark public webpage : ` http://www-cdf.fnal.gov/physics/new/top/top.html ` .	in this document we present several recent ( july 2008 ) results from studies of the top quark properties at the cdf and d experiments at the tevatron . measurements of several top quark properties , as well as tests of new physics in the top quark production and decay sectors are presented . in the latter case , no significant evidence for physics beyond the standard model is uncovered , and the tightest constraints to date are placed on most of the new physics scenarios investigated .
electron - impact excitation rates are a fundamental component of collisional radiative modelling . in particular , the excitation rates for o - like ne ( ne@xmath3 ) , the focus of this paper , are applicable in diagnosing the spectral lines not only from astrophysical observations but also from fusion plasma studies . it has been shown that transitions within the first five levels of an atom with the @xmath2 ground configuration can provide a valuable diagnostic of the electron temperature @xcite and the electron density @xcite . recently , an updated electron - impact excitation calculation of munoz _ et al . _ @xcite for the @xmath5 ground state complex of o - like ar ( ar@xmath3 ) found that the plasma diagnostics were particularly sensitive to near threshold resonances , and therefore , ne@xmath3 may offer a similar diagnostic potential , but without the high sensitivity to single near threshold resonance as found in the ar@xmath3 case . neon plays an important role in a variety of astrophysical phenomena , being the most abundant rare gas in the universe after helium . forbidden ne iii lines have been observed in a variety of astrophysical plasmas . the lines at 36 @xmath6 m and 3869 , due to forbidden @xmath7 and @xmath8 transitions within the ground configuration , have been observed in h ii regions by a number of authors @xcite . the ne iii 3869 line is also present in ne - rich filaments in the supernova cas a remnant @xcite and can be used to investigate the neon abundance and its spatial distribution throughout the remnant . in terms of laboratory measurements , the identification of ne iii lines were reviewed by persson _ et al . _ @xcite and the low lying terms are depicted in fig 1 . the present r - matrix calculation now includes shells of sufficiently high principal quantum numbers for a direct comparison with significantly more of their listed lines . more recently , james and co - workers @xcite measured electron - impact excitation cross sections for transitions in the far ultraviolet wavelength range between 1200 and 2700 using an electron - ion impact collision chamber . et al . _ @xcite have measured a pair of lines within the ground - state complex . the spin forbidden line @xmath9s@xmath10@xmath11@xmath12p@xmath13 and the quadrupole @xmath9s@xmath10@xmath11@xmath9d@xmath14 line offer diagnostic potential . finally , fig 2 illustrates the transitions of interest in the planetary nebulae ngc 3918 @xcite for ne iii . the wavelength of the astrophysically observed transitions are labelled in units of angstroms . an overview of the various theoretical models investigating electron impact excitation of ne@xmath3 is summarized in the introduction of mclaughlin and bell @xcite . this includes the work of seaton @xcite in the 1950s , and continues through a progression of the perturbative calculations ( e.g. , coulomb - born @xcite and distorted wave @xcite in 1960s-1970s ) yet only for transitions within the @xmath15 ground state complex . these were followed by early r - matrix calculations of pradhan @xcite and butler and mendoza @xcite using a program of saraph @xcite to transform ls - coupled k - matrices into level - level cross sections using the jajom code . this work included six states in the close coupling expansion and clearly illustrated the sensitivity of excitation rates when adopting either theoretically calculated or experimentally observed energy level values . between 1999 and 2001 , mclaughlin and co - workers in a couple of papers @xcite extended this model to 28 ls terms or 56 levels arising from the following @xmath16(@xmath17=0@xmath182 ) , @xmath19 configurations , with the extra @xmath20 , @xmath21 orbitals used only as an improvement in the description of the target s states . however , as evident in fig . 5 of ref.@xcite there is a breakdown in their calculation for the 28 terms model , manifesting as broad unphysical 2 ev resonances in transitions from the @xmath12p@xmath10 ground state term . unsurprisingly , this leads to corresponding unphysical increase in the effective collision strength ( see fig . 6 in ref.@xcite ) at approximately @xmath22 kelvin . we point out that a less elaborate six - level model does not exhibit the same problem as indicated in ref.@xcite . therefore , it would not be unreasonable , given the increase in computational power over the last decade to pursue a full breit - pauli r - matrix calculation for the 56 levels under consideration that would resolve and correct this problem . however , our present studies revealed that the 6 configurations listed above , actually give rise to a possible 49 terms or 95 levels . this imbalance between the configuration - interaction of the target and those levels actually used in the close coupling expansion will give rise to secondary , but a less pronounced set of pseudo - resonances above 3 rydbergs . therefore , to correct and also expand the scope of ne@xmath3 r - matrix calculations currently in the literature we have carried out an intermediate coupling frame transformation ( icft ) @xcite calculation for 554 levels . as illustrated in ref.@xcite , the use of the icft method gives cross sections that are extremely close to a breit - pauli ( bp ) calculation if both employing the same target configurations . the remainder of the paper is organized as follows . in section 2 we outline our small breit - pauli r - matrix and a large scale level icft r - matrix calculation used in the current study . section 3 , gives results for selected collision strengths and effective collision strengths and compares the bp and icft calculations . a small collisional radiative calculation is then performed in section 4 , to predict observed astronomical line ratios . finally , in the summary , we assess the impact of this now complete data set on the predictive modelling of ne@xmath3 line ratios . in the smaller r - matrix breit - pauli model , the atomic orbitals employed were generated from the atomic structure code civ3 @xcite ; in which the orbitals were energy - optimized on the lowest 28 terms ; details can be found in the earlier work of mclaughlin and co - workers @xcite . a subset of 28 ls terms constructed from the following configuration list ; @xmath15 , @xmath23 , @xmath24 and @xmath25 give rise to 56 fine - structure levels which were used in the close - coupling expansion . extra diffuse 4@xmath26 and 4@xmath27 pseudo - orbitals were used only to improve the target states , but also ensured a larger r - matrix box of 15.2 a.u . a total of forty basis orbitals were used to represent the continuum basis , easily spanning the energy range from threshold to 20 rydbergs . in the larger icft calculation , the target radial wavefunctions were generated using gasp ( graphical autostructure package ) @xcite , which is a java front - end for the atomic - structure code autostructure @xcite . the orbitals were generated within an thomas - fermi - dirac - amaldi potential , using slightly modified lambda scaling parameters as given by landi and bhatia @xcite . specifically we used @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 and @xmath37 . however , we are able to include all 24 configurations : @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath19 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 and @xmath60 resulting in 554 levels ; which were subsequently included in close - coupling expansion of the scattering calculation . the energies of the low lying @xmath39 terms were much improved over the earlier reported civ3 values @xcite , with only the first @xmath9d@xmath14 level at a 9.6@xmath61 difference being the greatest outlier relative to nist energy level values @xcite . a representative sample of the full 554 level calculation , is illustrated in table 1 , where it is seen that the first 33 level energies agree to within an average of @xmath62 with the nist values . ccccccc level no . & configuration & term@xmath63 & energy & energy & @xmath64 & th . order + & & & ( experiment ) & ( theory ) & & + 1 & 2s@xmath652p@xmath66 & @xmath12p@xmath14 & 0.00000 & 0.00000 & 0.00 & 1 + 2 & 2s@xmath652p@xmath66 & @xmath12p@xmath13 & 0.00585 & 0.00569 & 2.74 & 2 + 3 & 2s@xmath652p@xmath66 & @xmath12p@xmath10 & 0.00839 & 0.00837 & 0.24 & 3 + 4 & 2s@xmath652p@xmath66 & @xmath9d@xmath14 & 0.23548 & 0.25804 & 9.58 & 4 + 5 & 2s@xmath652p@xmath66 & @xmath9s@xmath10 & 0.50806 & 0.49160 & 3.24 & 5 + 6 & 2s@xmath92p@xmath67 & @xmath12p@xmath14 & 1.86163 & 1.91958 & 3.11 & 6 + 7 & 2s@xmath92p@xmath67 & @xmath12p@xmath13 & 1.86694 & 1.92512 & 3.12 & 7 + 8 & 2s@xmath92p@xmath67 & @xmath12p@xmath10 & 1.86987 & 1.92791 & 3.10 & 8 + 9 & 2s@xmath92p@xmath67 & @xmath9p@xmath13 & 2.63792 & 2.80123 & 6.19 & 9 + 10 & 2p@xmath123s & @xmath67s@xmath14 & 2.82384 & 2.81852 & 0.19 & 10 + 11 & 2p@xmath123s & @xmath12s@xmath13 & 2.91087 & 2.91399 & 0.11 & 11 + 12 & 2p@xmath123p & @xmath67p@xmath13 & 3.17481 & 3.15680 & 0.57 & 12 + 13 & 2p@xmath123p & @xmath67p@xmath14 & 3.17509 & 3.15720 & 0.56 & 13 + 14 & 2p@xmath123p & @xmath67p@xmath68 & 3.17558 & 3.15781 & 0.56 & 14 + 15 & 2p@xmath123s & @xmath12d@xmath68 & 3.21799 & 3.21957 & 0.05 & 17 + 16 & 2p@xmath123s & @xmath12d@xmath14 & 3.21825 & 3.21940 & 0.04 & 16 + 17 & 2p@xmath123s & @xmath12d@xmath13 & 3.21844 & 3.21934 & 0.03 & 15 + 18 & 2p@xmath123p & @xmath12p@xmath13 & 3.25096 & 3.26061 & 0.30 & 18 + 19 & 2p@xmath123p & @xmath12p@xmath10 & 3.25105 & 3.26065 & 0.30 & 19 + 20 & 2p@xmath123p & @xmath12p@xmath14 & 3.25106 & 3.26075 & 0.30 & 20 + 21 & 2p@xmath123s & @xmath9d@xmath14 & 3.26169 & 3.26701 & 0.16 & 21 + 22 & 2p@xmath123s & @xmath12p@xmath13 & 3.41259 & 3.37052 & 1.23 & 23 + 23 & 2p@xmath123s & @xmath12p@xmath10 & 3.41260 & 3.37048 & 1.23 & 22 + 24 & 2p@xmath123s & @xmath12p@xmath14 & 3.41266 & 3.37064 & 1.23 & 24 + 25 & 2p@xmath123s & @xmath9p@xmath13 & 3.46123 & 3.42538 & 1.04 & 25 + 26 & 2p@xmath123p & @xmath9p@xmath13 & 3.53537 & 3.51826 & 0.48 & 26 + 27 & 2p@xmath123p & @xmath12d@xmath13 & 3.54523 & 3.52981 & 0.43 & 27 + 28 & 2p@xmath123p & @xmath12d@xmath14 & 3.54533 & 3.53017 & 0.43 & 28 + 29 & 2p@xmath123p & @xmath12d@xmath68 & 3.54596 & 3.53137 & 0.41 & 29 + 30 & 2p@xmath123p & @xmath12f@xmath14 & 3.56670 & 3.55916 & 0.21 & 30 + 31 & 2p@xmath123p & @xmath12f@xmath68 & 3.56684 & 3.55965 & 0.20 & 31 + 32 & 2p@xmath123p & @xmath12f@xmath69 & 3.56703 & 3.56029 & 0.19 & 32 + 33 & 2p@xmath123p & @xmath9f@xmath68 & 3.57947 & 3.57865 & 0.02 & 33 + @xmath70 @xmath71 + @xmath72 absolute percentage difference relative to nist values + the advantage of applying the icft method is that the r - matrix inner region is essentially carried out in ls coupling , including only the mass - velocity and darwin terms . since we are primarily focusing on line ratios between transitions within the ground state complex , we used only 20 continuum basis functions for each continuum angular momenta to keep the dimensions of the hamiltonian matrix to a more manageable size . the inclusion of the @xmath73 , @xmath74 and @xmath75 as spectroscopic orbitals did increase the extent of the r - matrix box to 22.88 a.u , an increase of over 50@xmath61 in comparison to the breit - pauli value . the scattering calculations were then performed with our set of parallel @xmath76-matrix programs @xcite , which are modified versions of the serial rmatrix i programs @xcite . partial waves with total angular momentum from @xmath77 were calculated , with the burgess - tully top - up method used to account for higher partial wave contributions . however , considering the temperatures of astronomical interest here this should be more than sufficient to converge our excitation rates . in the outer region , the calculation of the cross sections was split into two energy regions . region 1 spans the energy range from the first excitation threshold up to the ionization threshold , whereas region 2 spans the energy range from from the ionization limit to four times the threshold . in region 1 , for both the bp and icft models , we used a fine energy mesh of 40,000 points to resolve the fine rydberg resonance structure . to account for the limited contribution of the higher partial waves to the low energy cross sections , we interpolated a coarse mesh of 200 energy points and added this to the fine mesh cross section . in region 2 , above the ionisation limit , where a coarse mesh is applicable , only 200 energy points were also used for both the exchange and non - exchange calculations . the resulting total collision strengths were subsequently maxwellian averaged across a range of electron temperatures and the effective collision strengths @xcite archived for modelling . for comparison with the 56 level bp calculation , the collision strength file of the 554 level icft model was reduced to include only those transitions between the lowest 56 levels . in fig . 3 , a representative set of collision strengths for transitions within the ground state @xmath38 multiplet is given . it has been shown in previous studies , such as those of griffin _ et al . _ @xcite and munoz _ et al . _ @xcite that both icft and bp calculations for atomic ions should provide essentially identical results , provided that the number of terms / levels included in the configuration - interaction and close - coupling expansions are the same . however , this is evidently not the case for the bp calculation shown in fig . 3 , as although the background collision strengths between the icft and the bp are similar , there are spurious resonances in the bp calculation , especially at approximately 10 rydbergs . in fact , perhaps less obviously , for the bp model there should not be any resonance structure above 3.2 rydbergs , as that is the last threshold included in that model . such spurious resonances in the bp calculations would need to be removed if reliable rates are to be produced . lastly , it should also be noted that the bp background collision strength is higher than icft . the cumulative effect of these differences is reflected in the corresponding effective collision strengths @xcite shown in fig . 4 . over a wide range of electron temperatures , the 56 level bp effective collision strengths are consistently higher than the benchmark 554 level icft values . however , in terms of line ratios , a low density model using the coronal approximation involves the ratio of the effective collision strengths , which can lead to smaller than expected differences in the predicted line ratios , see the results in section 4 . figure 5 shows the 2s@xmath652p@xmath123s(@xmath12s@xmath78)@xmath11 2s@xmath652p@xmath123p(@xmath12p@xmath79 ) collision strengths as a function of electron energy . in this case , although small differences can be seen , the 554 levels icft and 56 levels bp calculations are in reasonable accord with each other . this level of agreement ( i.e. , @xmath8015@xmath1825% ) was expected and typical for the collision strengths calculated as more target states are included in the scattering calculation . as anticipated , in fig . 6 , the corresponding effective collision strengths are indeed in good agreement between the two calculations . temperature and density sensitive energy intensity ratios between transitions among the ground state @xmath5 multiplet have been studied for ar iii @xcite . here we examine the corresponding energy intensity ratios for transitions among the ground state @xmath38 multiplet of ne iii . the effective collision strengths are processed through the adas suite of codes ( http://www.adas.ac.uk ) with the first 5 levels in both the 56 level bp and 554 level icft calculation , moved to nist energies and nist einstein a - coefficients where available . the nist values are taken primarily from the work of froese fischer and tachiev @xcite and kramida and nave @xcite . the wavelengths for the various transitions used in the emissivity modelling are given in figure 4 . for the temperatures and densities of interest , cascades from higher levels are not significant . therefore , although both the icft and bp collisional rates include those involving the first 56 levels , only the transitions within the ground state complex are important for the following line ratios . both datasets used the same a - values for transitions within the ground configuration . we first examine the temperature sensitive energy intensity ratio @xmath81 with @xmath82 the population in energy level @xmath83 , @xmath84 the einstein a coefficient for a transition between energy levels @xmath83 and @xmath85 and @xmath86 the corresponding wavelength for the transition . note that the index numbers for the levels are given in table 1 . the bp and icft results are shown in fig . the bp and icft results are in good agreement with each other even though the bp effective collision strengths for these transitions were higher than the icft results . this is due to a cancellation of errors in the ratio . due to the fact that the bp effective collision strengths for transitions within the ground configuration are in general higher than the icft ones , the bp model has increased emissivities for all of the transitions . while this does not result in a large change in the line ratios , one would see a difference when looking at absolute intensities of individual spectral lines . the results from the bp model provides larger emissivities which can affect measurements of the ne elemental abundances , since such measurements are obtained using the absolute emissivity , or the ratio with the emissivity of a completely different ion / element . the measurements based on the bp results would lead to a lower ne abundance . next we consider the density sensitive energy intensity ratio @xmath87 with the bp and icft results shown in fig . the @xmath88 ratio for the bp and icft results agree closely with each other and show a strong density dependence between electron densities of @xmath89 and @xmath90 @xmath91 . as for the @xmath92 ratio , the @xmath88 ratio for ne iii shows a wider range of variation than the @xmath88 ratio for ar iii . therefore , although systematically the two systems are both excellent density diagnostics , the ne iii is the marginally better option . a large scale electron - impact excitation calculation of this o - like ion was carried out to investigate high energy irregularities in the cross sections of an earlier r - matrix calculation and the subsequent impact on line ratios . we have resolved these issues , and have benchmarked effective collision strengths for use in the modelling of laboratory and astrophysical plasmas . as an illustration of an application of these effective collision strengths , a study was undertaken into the viability of applying the same ar iii line ratios @xcite to the ne iii case . overall , the o - like ion , ne iii , exhibits great promise in diagnosing the temperature and density of ne iii in planetary nebulae plasmas , being more sensitive than the equivalent ar iii line ratios . our effective collisions strengths for the 554 level icft r - matrix calculation shall be made available on the oak ridge national laboratory atomic data website ( http://www-cfadc.phy.ornl.gov/data_and_codes ) and in the adas database ( http://www.adas.ac.uk ) . it is a pleasure to thank professor a dalgarno frs , for drawing our attention to this problem and the need for accurate atomic data required for the modelling of planetary nebulae . t - g lee , c p ballance and j a ludlow were supported by us department of energy ( doe ) grants through auburn university . e landi acknowledges support from nasa grants nnx10am17 g and nnx11ac20 g . b m mclaughlin acknowledges support by the us national science foundation through a grant to itamp at the harvard - smithsonian center for astrophysics . the computational work was carried out at the national energy research scientific computing center in oakland , ca , usa and on the tera - grid at the national institute for computational sciences ( nics ) in knoxville , tn , usa , which is supported in part by the us national science foundation . 99 de robertis m m , dufour j r and hunt r w 1987 j. roy . astron . soc * 81 * 195 ( colour online ) an overview of energy levels ( levels given in @xmath93 ) illustrating the transitions of interest in the planetary nebulae ngc 3918 @xcite for ne iii . the wavelength of the astrophysically observed transitions are labelled in units of angstroms . for multiplets , the wavelength ( in ) of the strongest observed lines singlet levels are omitted for clarity . ] ( colour online ) collision strengths for the transitions ( a ) @xmath94 , ( b ) @xmath95 , ( c ) @xmath96 and ( d ) @xmath97 as a function of the incident electron energy . the solid black line indicates the 554 level icft results and the dashed red line indicates the 56 level breit - pauli results . ] ( colour online ) effective collision strengths obtained by averaging the collision strengths over a maxwellian temperature distribution for the electrons . results are shown for the transitions ( a ) @xmath94 , ( b ) @xmath95 , ( c ) @xmath96 and ( d ) @xmath97 as a function of the electron temperature ( k ) . the solid black line indicates the 554 level icft results and the dashed red line indicates the 56 level breit - pauli results . the wavelengths in angstroms for the various transitions are also included for completeness ] ( colour online ) collision strengths for the transitions ( a ) @xmath98 , ( b ) @xmath99 , ( c ) @xmath100 as a function of the incident electron energy . the solid black line indicates the 554 level icft results and the dashed red line indicates the 56 level breit - pauli results . ] ( colour online ) effective collision strengths obtained by averaging the collision strengths over a maxwellian temperature distribution for the electrons . results are shown for the transitions ( a ) @xmath98 , ( b ) @xmath99 , ( c ) @xmath100 as a function of the electron temperature ( k ) . the solid black line indicates the 554 level icft results and the dashed red line indicates the 56 level breit - pauli results . ] ( colour online ) @xmath92 line ratio as a function of electron temperature for an electron density of @xmath101cm@xmath12 . the solid black line shows the results using the 554 level icft r - matrix calculation and the dot - dashed blue line shows the result using the 56 level breit - pauli r - matrix calculation , with the energies and transition rates for the first 5 levels in both r - matrix calculations shifted to nist energies and transition rates@xcite for the modelling calculations . ] ( colour online ) @xmath88 line ratio as a function of electron density for electron temperatures of ( a ) 5000k ( b ) 8000k ( c ) 15000k , and ( d ) 30000k . the solid black line shows the results using the 554 level icft r - matrix calculation and the dot - dashed blue line shows the result using the 56 level breit - pauli r - matrix calculation . ]	the five j@xmath0 levels within a @xmath1 or @xmath2 ground state complex provide an excellent testing ground for the comparison of theoretical line ratios with astrophysically observed values , in addition to providing valuable electron temperature and density diagnostics . the low temperature nature of the line ratios ensure that the theoretically derived values are sensitive to the underlying atomic structure and electron - impact excitation rates . previous r - matrix calculations for the o - like ne ion , ne@xmath3 , exhibit spurious structure in the cross sections at higher electron energies , which may affect maxwellian averaged rates even at low temperatures . furthermore , there is an absence of comprehensive excitation data between the excited states that may provide newer diagnostics to compliment the more established lines discussed in this paper . to resolve these issues , we present both a small - scale 56-level breit - pauli ( bp ) calculation and a large - scale 554 levels r - matrix intermediate coupling frame transformation ( icft ) calculation that extends the scope and validity of earlier jajom calculations both in terms of the atomic structure and scattering cross sections . our results provide a comprehensive electron - impact excitation data set for all transitions to higher @xmath4 shells . the fundamental atomic data for this o - like ion is subsequently used within a collisional radiative framework to provide the intensity line ratios across a range of electron temperatures and densities of interest in astrophysical observations . url # 1#1urlprefix [ 2][]#2 short title : electron impact excitation of ne@xmath3 ions + revised : j. phys . b : at . mol . & opt . phys :
electron dynamics in a rigid uniform neutralizing background is a well - studied problem ( jellium in quantum mechanics @xcite , one component plasma in classical mechanics @xcite ) . more realistically , point ions lead to a polarization of the electron density ( e.g. , in a hydrogen plasma ) and the dynamics of the non - uniformly distributed electrons is radically changed . the objective here is to provide a practical theory for the description of equilibrium time correlation functions for electrons in the simplest case of a single point ion of charge number @xmath0 . if the charge is positive , essential quantum diffraction effects must be accounted for even at high temperatures and low densities to avoid the electron - ion coulomb singularity . a classical hamiltonian description is used here , with a regularized electron - ion interaction that accounts for such effects filinov . this study is an outgrowth of recent investigations based on molecular dynamics simulations for this system @xcite . the qualitative features observed for the electron field autocorrelation function from simulation were captured by a simple mean field kinetic theory . such a kinetic theory is obtained here from the asymptotically exact short time limit for the generator of the dynamics , providing both context and a generalization of the analysis in reference 4 to strong electron - electron coupling conditions . the kinetic theory is solved exactly to express the correlation functions in terms of effective single electron trajectories about the ion and collective excitations via a dielectric function for the non - uniform electron fluid . for @xmath1 the results reduce to the familiar random phase approximation ( rpa ) with local field effects ( the generalized vlasov approximation of reference 5 ; see also reference 6 for a related nonlinear kinetic equation for dusty plasmas ) . more generally it constitutes a generalization of the rpa to a non - uniform electron gas , with both ion - electron and electron - electron interactions renormalized by correlations ( the vlasov equation for an electron gas in a periodic potential is discussed in reference 7 ) . the only required input is the time independent correlations for one or two electrons and the ion . for the calculations here the hypernetted chain approximation ( hnc ) integral equations are used for these static correlations . the correlation functions are further decomposed into contributions from the bound and free ( positive and negative energy ) states of the effective single particle dynamics in section [ sec3 ] . as a special case , the electric field autocorrelation function is considered in section [ sec4 ] with the objective of providing a clear interpretation for the @xmath0 dependence observed in simulations . for increasing @xmath0 this dependence includes 1 ) an increasing covariance of the field ( initial value of the correlation function ) , 2 ) a decreasing correlation time , and 3 ) the development of a strong domain of anti - correlation at intermediate times @xcite . it is shown here that all three features can be attributed to an increasing contribution from the bound states of the single particle effective dynamics representing actual metastable trapped trajectories of the @xmath2 particle dynamics . with this understanding of the active mechanisms , a simple analytic and accurate model for the bound and free state contributions is proposed and tested . the dynamics is restricted to circular and straight line trajectories , and the electron - ion charge correlation is represented by a nonlinear debye distribution . somewhat surprisingly , the model reproduces all of the above @xmath0 dependencies with remarkable accuracy . this provides the basis for a practical representation of more general correlation functions , such as the dynamic structure factor , and more complex state conditions required for plasma spectroscopy in hot , dense matter @xcite . to illustrate the practical utility of the model , collective excitations are explored briefly in section [ sec6 ] using the model to evaluate the dielectric function for this nonuniform electron distribution about the ion . for weakly nonuniform conditions ( small @xmath0 ) the results are suggestive of a local density approximation whereby the modes are similar to those of a uniform electron gas , but with the density replaced by the actual local density near the ion . however , this simple approximation fails for larger @xmath0 where the bound states dominate and long wavelength plasmons are replaced by local excitations at the circular orbit frequencies . finally , some future directions are discussed in the last section . consider a system of @xmath3 electrons of charge @xmath4 , an infinitely massive positive ion of charge @xmath5 placed at the origin , and a rigid uniform positive background for overall charge neutrality contained in a large volume @xmath6 . the hamiltonian is @xmath7where @xmath8 and @xmath9 are the position and velocity of electron @xmath10 . the coulomb interaction between electrons @xmath11 and @xmath12 is denoted by @xmath13 where @xmath14 . also , @xmath15 is the electron - ion interaction for electron @xmath10 , and @xmath16 is the coulomb interaction for electron @xmath10 with the uniform neutralizing background . in a quantum description , @xmath17 is also a coulomb interaction but in the classical case the short range attractive divergence must be regularized " within a distance @xmath18 of the order of the de broglie wavelength @xcite . the simplest such form is @xcite@xmath19 in the remainder of this presentation such a semi - classical description is assumed . comments on the corresponding quantum analysis are given in the final discussion section . the typical response functions characterizing dynamical excitations in a plasma are the charge density or current autocorrelation functions , which are sums of single particle functions . more generally , the correlation functions of this type are defined by @xmath20where @xmath21 is a point in the @xmath22 dimensional phase space , and @xmath23,@xmath24 denotes a point in the phase space of particle @xmath25 . the notation @xmath26 denotes the evolution of the point @xmath27 to a time @xmath28 later under the dynamics generated by the hamiltonian of ( 2.1 ) . the role of the central fixed ion is suppressed in this notation , and it acts as an external potential for the electrons . the phase functions @xmath29 and @xmath30 denote some observables of interest , composed of sums of single particle functions . @xmath31finally , the average is over an equilibrium ensemble ( e.g. , gibbs ) , @xmath32 . because of the special form ( [ 2.3 ] ) , the @xmath2 particle average can be reduced to a corresponding average in the single electron subspace , by partial integration over @xmath33 electron degrees of freedom ( see appendix [ appb ] ) @xmath34here , @xmath35 is the equilibrium number density for electrons at a distance @xmath36 from the ion ( the precise definition as a partial integral of @xmath32 is given in appendix a ) , and @xmath37 is the maxwell - boltzmann velocity distribution . the function @xmath38 at @xmath39 is linearly related to the single particle phase function @xmath40 in ( [ 2.3])@xmath41the correlation function @xmath42 is related to the joint number density @xmath43 for two electrons at @xmath44 and @xmath45 with the ion at the origin by@xmath46the precise definition for @xmath43 as a partial integral of @xmath47 is given in appendix [ appa ] . the time evolution of @xmath48 in the single particle phase space is governed by a linear equation of the form @xmath49all of the results up to this point are still exact . the difficult many - body problem is encountered in the determination of @xmath50 . weak coupling and perturbation expansions are not appropriate for high @xmath0 ions or conditions for strongly coupled electrons so instead a markovian approximation is proposed , @xmath51this approximation assumes that the exact generator for the initial dynamics persists as the dominant form for later times as well . in this way the exact initial correlations among electrons and with the ion are included . the detailed form for @xmath52 is obtained in appendix [ appb ] with the result @xmath53where @xmath54 and @xmath55 are renormalized " electron - ion and electron - electron interactions @xmath56the direct correlation function @xmath57 is defined in terms of @xmath58 by @xmath59at @xmath1 this becomes the usual ornstein - zernicke equation @xcite . to interpret ( [ 2.9 ] ) , substitute this approximation into ( [ 2.7 ] ) to get the markovian linear kinetic equation for @xmath48@xmath60at weak electron - electron coupling @xmath61 and at weak electron - ion coupling @xmath62 and ( [ 2.12 ] ) is recognized as the linear vlasov equation . more generally , the markov approximation ( [ 2.12 ] ) upgrades this mean field result to include the effects of equilibrium correlations on all interaction potentials . thus it is suitable for a discussion of the strong coupling conditions that occur for @xmath63 . the left side of ( [ 2.12 ] ) describes single electron motion about the ion in the effective potential @xmath64 , while the right side describes dynamical screening of this motion . in summary , the description of electron dynamical correlations and fluctuations has been reduced in the markovian approximation to @xmath65where @xmath66 is the operator whose kernel is ( [ 2.9 ] ) . this operator requires as input the equilibrium electron density @xmath67 and the equilibrium direct correlation function . the kinetic equation can be solved exactly in terms of the single particle trajectories about the ion and dielectric function for an inhomogeneous electron gas , describing the dynamical screening due to interactions among the electrons in the presence of the ion . the details are carried out in appendix appc , and the correlation functions are obtained from that solution in appendix [ appd ] . for the class of correlation functions for which @xmath68 ( i.e. , is independent of the velocity ) the laplace transform of ( [ 2.13 ] ) takes the simpler form@xmath69the dynamics is governed by the resolvent operator @xmath70the generator for the dynamics , @xmath71 is seen to be that for a single electron interacting with the ion via the effective mean field potential @xmath54 . the function @xmath72 is the given function @xmath73 , modified by dynamical screening@xmath74where @xmath75 is the dielectric function " for the electrons in the presence of the ion dielectric@xmath76and @xmath77 is @xmath78 an important example is the autocorrelation function for the electron density near the ion . in the absence of the ion this is referred to as the dynamic structure factor and that terminology will be used here in the presence of the ion as well . the correlation function @xmath79 is constructed from the local densities of ( [ 2.3 ] ) with @xmath80 and @xmath81 . then ( [ 2.14 ] ) becomes@xmath82here @xmath83 is the static structure factor@xmath84representing the exact initial correlations . the last equality of ( 2.19a ) is proved in appendix [ appd ] . equation ( [ 2.19 ] ) is the exact short time ( markovian ) form for the dynamic structure factor . for @xmath1 it becomes the usual random phase approximation ( rpa ) with local field corrections " ; this means that the bare electron - electron potential has been replaced by @xmath85 ( the corresponding direct correlation function ) to account for exact initial correlations . the @xmath1 case has been studied in detail for the hydrogen plasma , where this approximation is found to be very good up to moderate plasma coupling strengths over a wide range of space and time scales @xcite . equation ( [ 2.15 ] ) extends this approximation to include the presence of the ion for @xmath86 , corresponding to an inhomogeneous rpa . a second important example is the autocorrelation function @xmath87 for the electron electric field at the ion , where@xmath88this correlation function also is obtained from the dynamic structure factor @xmath89 by integration@xmath90so @xmath91interestingly , one of the fields is dynamically screened while the other is statically screened,@xmath92 the trajectories of the mean field generator @xmath93 do not represent the dynamics of any given electron , but rather their effective collective representation . thus , the interactions among many electrons appears in the mean field theory only through their modification of the potentials . the bound states of the effective potential @xmath94 are representations of real metastable states in the md simulation . at weak coupling there are few such metastable states and their lifetimes are short compared to the correlation time for the field autocorrelation function . as @xmath0 increases the stronger coupling gives rise to more metastable states with longer lifetimes . there is a crossover of these lifetimes to values larger than the correlation time at which point they behave essentially as bound states for the relevant time scales . to isolate the effects of such bound states it is useful to divide the phase space integral of ( [ 2.13 ] ) into contributions from bound and free parts of that phase space . the decomposition is defined by the negative and positive energy states for the effective potential @xmath95 . for a given position @xmath96 there is a maximum velocity @xmath97 above which the total energy is positive@xmath98the single particle equilibrium density for the position and velocity can therefore be divided into two contributions @xmath99for example , integration over the velocity gives the relative contribution of bound states to the density @xmath35 @xmath100where @xmath101 is the thermal velocity of the electron and @xmath102 is the error function . since @xmath97 is proportional to @xmath103 , @xmath104 is an increasing function of @xmath0 for all @xmath96 . the decomposition ( [ 2.26 ] ) provides the identification of contributions to the correlation functions from bound and free states @xmath105the analysis of the following sections shows that the interesting @xmath0 dependence of electron dynamics can be understood in terms of the relative sizes of these two contributions . the dielectric function @xmath106 describes a crossover from no screening at short times ( large @xmath107 ) to static screening at large times ( @xmath108 ) . in the next two sections , the short time form with @xmath109 will be considered . in that case the correlation functions in ( [ 2.14 ] ) become the effective single particle functions @xmath110which are asymptotically exact at short times . the effects of dynamical screening at longer times are discussed briefly in section [ sec6 ] . the electric field autocorrelation function is particularly instructive since the field is sensitive to configurations closest to the ion . also , its time integral determines the dominant contribution to the half width of spectral line widths broadened by electrons in many practical caseslewis . the short time form of ( [ 2.22a ] ) is @xmath111it is shown in appendix [ appa ] that the statically screened field of ( [ 2.22 ] ) simplifies further to@xmath112thus , the number density @xmath113 determines all of the ingredients needed for calculation of @xmath87 . it is calculated here in the hnc approximation described in appendix [ appa ] . the electron - electron coupling strength is measured by the dimensionless ratio @xmath114 where @xmath115 is the average distance between electrons , determined from the density by @xmath116 . the ion - electron coupling is measured by @xmath117 . the results presented here are for @xmath118 , and @xmath119 for values of @xmath120 . the corresponding quantum regularization length is @xmath121 . the electron - electron coupling is therefore weak , but the ion - electron coupling can be very strong , @xmath122 . these conditions were chosen because previous molecular dynamics studies have been performed at these values @xcite . it is useful to anticipate the increasing role of bound states with increasing @xmath0 by considering first the time independent covariance @xmath123 . this is shown in figure [ fig1 ] . the sharp increase above @xmath124 is seen to be entirely due to the appearance of the bound states . similar strong effects on dynamical structure are observed . as a function of @xmath0.,title="fig : " ] . and @xmath125,title="fig : " ] . figure [ fig2 ] shows the results for @xmath87 calculated from ( [ 3.2 ] ) for @xmath126 . the development of a strong anti - correlation and the decreasing initial correlation time with increasing @xmath0 is evident . these are the effects noted above , first observed in md simulations @xcite . the interpretation of this two - fold dependence on @xmath0 is provided by figures [ fig3 ] and [ fig4 ] showing the contributions from bound and free state contributions for @xmath127 . for @xmath128 the dominant contribution is from free states , which have a monotonic positive decay . in contrast , for @xmath129 the dominant contribution is from bound states which provide the negative anti - correlation as the sign of the field changes along each trajectory when it passes through apsidal distances . the time for this change can be estimated by half the period for a circular orbit at position @xmath130 , which is proportional to @xmath131 . this is consistent with the observed decrease in correlation time in figure fig2 . further elaboration and explanation is provided by the simple model of the next section . for @xmath128.,title="fig : " ] . for @xmath132.,title="fig : " ] . calculated from kinetic theory and from the simple model for @xmath128 and @xmath133.,title="fig : " ] . the results of this section demonstrate the utility of the kinetic theory for conditions of strong ion - electron coupling . although only weak electron - electron coupling was considered , the theory is applicable to strong coupling among electrons as well . also , while attention in this section has been limited to the electric field autocorrelation function it is clear that the analysis applies with equal ease to the dynamic structure factor as well , with only the additional complication of more parameters ( i.e. , two position vectors ) characterizing that function . the decomposition of the correlation function into bound and free contributions demonstrates that the interesting features associated with increasing @xmath0 can be attributed entirely to the increasing contribution from bound states . consider again the electric field auto correlation function given by ( 3.2 ) explicitly decomposed into its bound and free contributions@xmath134@xmath135the objective here is to capture the qualitative features of the bound and free contributions in a very simple model that allows further elaboration of their relative roles and the mechanisms involved . this is accomplished by assuming circular trajectories for the bound states and straight line trajectories for the free states , @xmath136@xmath137here @xmath138 and @xmath139 for circular obits . in this case the velocity must be orthogonal to @xmath44 with the specified magnitude@xmath140for consistency with newton s equations . the maxwellian in @xmath141 must therefore be replaced by this restriction on the velocities @xmath142where @xmath143 is the component of @xmath144 orthogonal to @xmath145 . the factor @xmath104 is given by ( [ 2.27 ] ) and is required by the correct normalization on integration over all velocities . equation ( [ 4.4 ] ) then becomes @xmath146where @xmath147 . note that these modifications of the trajectories do not affect the initial value , @xmath123 , which is still exact . one further simplification is made to complete the model . the effective potential @xmath54 is replaced by its weak coupling debye form@xmath148where @xmath149 is the debye length and @xmath150 is the quantum regularization length of ( [ 2.1a ] ) . the corresponding electron density is now the non - linear debye form@xmath151this is exact in the weak coupling limit , with @xmath152 . more generally , @xmath153 is chosen to give the correct value of @xmath123 using ( [ 4.8 ] ) and ( [ 4.9 ] ) in the exact equation for @xmath123 and adjusting @xmath153 to fit the values obtained from hnc@xmath154with@xmath155table 1 gives the values obtained using @xmath156 from the hnc approximation , for the case @xmath118 and @xmath119 . [ cols="^,^,^",options="header " , ] table 1 : effective charge number @xmath153 for the debye form figure 5 shows again the results of figures 3 and 4 for @xmath128 and @xmath133 , now including as well the results from the simple model of this section . remarkably , the use of the debye form with the straight line and circular trajectories gives an accurate representation of the kinetic theory results . this provides the basis for a practical tool for use in more complex conditions , as discussed in the last section , and for other correlation functions , as illustrated in the next section . the last two sections addressed the short time form ( [ 3.1 ] ) of the correlation functions for which the dielectric function behaves as @xmath157 . for very long times ( small @xmath158 ) @xmath159 crosses over to represent static screening . at intermediate times there are contributions from collective excitations . their description is more complex than for the uniform electron gas for which the space dependence of the dielectric function occurs only through @xmath160*. * to simplify the discussion here consider the weak electron - electron limit ( but possibly strong ion - electron coupling ) for which @xmath161 and define the partial transform@xmath162collective excitations for the non - uniform system , @xmath163 , are defined by @xmath164 . for zero charge number on the ion , @xmath1 , @xmath165 reduces to the familiar rpa dielectric function of the uniform electron gas@xmath166identifying the excitation spectrum @xmath167 . for @xmath86 the modes , @xmath168 , depend on @xmath169 due to the inhomogeneity caused by the ion . as an example , consider the solutions with very large @xmath158 . expanding @xmath170 to order @xmath171 gives @xmath172where @xmath173 is the component of @xmath174 along @xmath44 . for small @xmath175 the first term goes to the square of the local plasma frequency@xmath176and ( [ 5.4 ] ) simplifies to@xmath177here @xmath178 is the corresponding local debye length . therefore , if @xmath179 the system supports plasmons with frequencies defined in terms of the local density . this is suggestive of a more general local density approximation " @xcite where the dielectric function for the uniform electron gas is modified by replacing the uniform density with the actual non - uniform density @xmath180however , this is not correct in general as is evident from ( [ 5.4 ] ) for @xmath181 and the following . to continue with the evaluation of ( [ 5.1 ] ) separate into bound and free contributions @xmath182 . \label{5.7}\end{aligned}\]]next , introduce the approximate trajectories of the last section . the analysis is straightforward but lengthy so only the final result is given here@xmath183the first term of the brackets , proportional to @xmath184 , is the contribution from free states , while the second term proportional to @xmath104 is that from bound states ( recall ( [ 2.27 ] ) for the bound state contribution @xmath104 ) . the functions @xmath185 and @xmath186 are @xmath187@xmath188 . \label{5.10}\end{aligned}\]]the bessel functions @xmath189 and @xmath190 are @xmath191the free state contribution depends only on the magnitudes @xmath192 while the bound state contribution depends on the directions as well , where @xmath193 and @xmath194 are the components of @xmath195 parallel and perpendicular to @xmath44 . consider the small @xmath175 , long wavelength limit of this expression for @xmath196 . retaining the leading order contributions to ( [ 5.9 ] ) and ( [ 5.10 ] ) gives@xmath197the limiting forms for these long wavelength excitations as functions of the ion charge number @xmath0 are@xmath198the local plasmons are recovered for small charge numbers , while for large charge numbers the local circular frequencies @xmath199 dominate . the complex coefficient in ( [ 5.13 ] ) implies that these excitations are damped . the dielectric function @xmath165 is quite complex , but it is seen that the simple model of straight line and circular orbits simplifies this considerably . further analysis of the collective modes for the inhomogeneous electron gas will be given elsewhere . a very general description of time correlation functions for electron properties near a positive ion has been given by the markovian approximation ( [ 2.14 ] ) . the time dependence has two contributions , an effective single particle dynamics that dominates for short and intermediate times , and a modification of that dynamics due to collective modes . the single particle dynamics has a strong dependence on the ion charge number @xmath0 , which is due to the growing dominance of bound states for large @xmath0 . the description is valid for such strong coupling conditions , since the markovian approximation preserves the exact equilibrium electron - ion correlations in the effective potential governing the single particle dynamics . similarly the electron - electron potential is renormalized by the exact equilibrium electron - electron correlations . the description has been illustrated here for the special case of the electric field autocorrelation function under the same conditions as have been studied by md simulations @xcite . as the time scales are short , only the effective single particle dynamics has been considered . it remains to explore other correlation functions for which the collective modes are expected to be more important , and to provide a detailed characterization of those modes via the inhomogeneous electron gas dielectric function . the promise for progress in this direction is provided by the success of a simple model for the bound and free state dynamics . there are several avenues for future directions based on this work : 1 . the analysis here is based on a semi - classical description using a `` regularized '' electron - ion potential . the quality of this type of description can be benchmarked by comparison with a corresponding quantum description of the same effective single particle dynamics . 2 . a generalization of the markovian approximation to a two component electron - ion plasma is straightforward . in that case the interest is in the electron dynamics in the vicinity of one of the ions . an additional feature is the effect of the dynamics of the ions . this constitutes a kinetic theory for the two particle ( electron and ion ) distribution function . the spectral line shapes from charged radiators are an important diagnostic tool in laser fusion studies . a recent formulation of this problem including all plasma charge correlations is expressed in terms of constrained equilibrium time correlation functions @xcite . the constraint arises from a specified value of the total ion electric field during the dynamical broadening by the electrons . the simple model described here provides the potential for practical evaluation of these constrained time correlation functions under the demanding conditions of hot , dense matter . the role of charge correlations in plasma spectroscopy also has been discussed recently in reference 8 . a corresponding identification of the markov limit in a fully quantum analysis is straightforward but the resulting renormalization of the ion - electron and electron - electron interactions by initial correlations is more complicated . still , the structure obtained here of single electron dynamics in the presence of the ion modified by collective modes of the dielectric function remains the same . the attractive distortion of the electron density by the ions , particularly for the bound states , is a type of electron confinement . the analysis here can be applied to real traps for charged particle confinement ( e.g. , dusty plasmas near an electrode , ultra - cold plasmas in a laser trap , valence electrons in metallic clusters , electrons in quantum dots ) . significant differences include complete confinement and relaxing charge neutrality . the research was supported by the nsf / doe partnership in basic plasma science and engineering under the department of energy award de - fg02 - 07er54946 . consider a point ion of charge number @xmath0 in an electron gas of average density @xmath200 with a positive uniform neutralizing background of density @xmath201 . the equilibrium structure of the electrons in the presence of the ion is given by the one and two particle distribution functions , defined for the equilibrium ensemble by@xmath202@xmath203where @xmath204 is the position of an electron at @xmath205 relative to the ion at @xmath206 . this position dependence reflects the fluid symmetry ( rotational invariance ) about the ion . the distribution function @xmath207 obeys the equilibrium bbgky hierarchy equation@xmath208the last term on the right side is due to the interaction of the ion and electron with the uniform neutralizing background whose density is @xmath201 . also , @xmath209 is the force of the electron on the ion , @xmath210 is the reaction force of the ion on the electron , and @xmath211 is the force between electrons . these forces are derived from corresponding potentials@xmath212the solutions to ( [ a.4 ] ) have the forms @xmath213@xmath214here @xmath215 and @xmath216 are the maxwellians for the ion and electron@xmath217and @xmath218 @xmath219 are the one and two particle electron number densities ( relative to the ion ) normalized to @xmath220 and @xmath221 , respectively . use of these forms in ( [ a.4 ] ) gives directly@xmath222 \mathbf{\nabla } _ { 2}v_{ei}\left ( r_{20}\right ) \bigg ) \nonumber \\ & & + \,\mathbf{v}_{1}\cdot \bigg(\mathbf{\nabla } _ { 1}n\left ( r_{10}\right ) + \beta n\left ( r_{10}\right ) \mathbf{\nabla } _ { 1}v_{ei}\left ( r_{10}\right ) \nonumber \\ & & \qquad \qquad \qquad \qquad \qquad \qquad -\beta \int d\mathbf{r}_{2}\left [ n\left ( \mathbf{r}_{10},\mathbf{r}_{20}\right ) -n\left ( r_{10}\right ) n_{e}% \right ] \mathbf{\nabla } _ { 2}v_{ee}\left ( r_{21}\right ) \bigg ) \label{a.7}\end{aligned}\]]since the velocities of the ion and electron are independent , the following two equations hold@xmath223 \mathbf{\nabla } _ { 2}v_{ei}\left ( r_{20}\right ) \label{a.8}\]]@xmath224 \mathbf{\nabla } _ { 2}v_{ee}\left ( r_{21}\right ) \label{a.9}\]]this provides two , seemingly independent , equations for the same electron density around the ion @xmath225 . their equivalence implies@xmath226 = 0 \label{a.10}\]]which means that the total external force on the system of two selected particles , the ion and the one electron , is zero at equilibrium . in other words , the distortion of @xmath227 is just such as to enforce this condition . both equations ( [ a.8 ] ) and ( [ a.9 ] ) are useful , as illustrated in the following two subsections . in the remainder of the appendices and in the text , only the special case of a massive ion fixed at the origin is considered . the electric field due to one electron at the ion is defined by @xmath228 the electric field autocorrelation function of ( [ 2.22a ] ) depends on the associated statically screened field@xmath229this dependence on the two electron density @xmath230 can be eliminated using ( [ a.8 ] ) to get @xmath231an effective potential is defined in terms of the density @xmath113 in ( [ 2.10])@xmath232so the screened electric field is given in terms of the gradient of this effective potential@xmath233 a simple and accurate method to determine the density is given by the hypernetted chain ( hnc ) integral equations @xcite . they can be obtained from the usual form for a three component plasma of electrons , protons , and ions of charge number @xmath0 . then the limit is taken of uniform proton distribution and dilute concentration for the ions of charge @xmath0 . instead , it is useful here to state the result as an approximation to ( a.9 ) . consider the mean field limit of no electron - electron correlations @xmath234 \mathbf{\nabla } % _ { 2}v_{ee}\left ( r_{21}\right ) \rightarrow \int d\mathbf{r}_{2}\left [ \frac{% n\left ( r_{1}\right ) n\left ( r_{2}\right ) } { n\left ( r_{1}\right ) n_{e}}-1% \right ] \mathbf{\nabla } _ { 2}v_{ee}\left ( r_{21}\right ) . \label{a.16}\ ] ] then equation ( [ a.9 ] ) simplifies to@xmath235 = 0 , \label{a.17}\]]or@xmath236an arbitrary constant has been used to assure the limit @xmath237 when @xmath238 . equation ( [ a.18 ] ) is an integral form of the boltzmann - poisson equation . the hnc approximation is similar , but retains electron - electron correlations in the absence of the ion @xmath239 \mathbf{\nabla } % _ { 2}v_{ee}\left ( r_{21}\right ) \rightarrow \int d\mathbf{r}_{2}\left [ \frac{% n\left ( r_{1}\right ) n\left ( r_{2}\right ) } { n\left ( r_{1}\right ) n_{e}}-1% \right ] \mathbf{\nabla } _ { 2}c_{ee}\left ( r_{21}\right ) . \label{a.19}\]]the function @xmath240 is the electron direct correlation function defined in terms of the electron - electron pair correlation function ( without the ion ) , @xmath241 , by the ornstein - zernicke equation@xmath242the approximation ( [ a.19 ] ) in ( [ a.9 ] ) gives the hnc approximation for @xmath113 @xmath243this is the same as ( [ a.18 ] ) except that @xmath244 has been replaced by @xmath245 . the electron - electron direct correlation function is determined independently from the ornstein - zernicke equation ( [ a.20 ] ) and the hnc approximation@xmath246equations ( [ a.20 ] ) - ( [ a.22 ] ) are the hnc equations used for the numerical calculations presented here . the dynamics of @xmath248 is conveniently expressed in terms of the fundamental correlation function @xmath249@xmath250@xmath251the initial value @xmath252 is easily calculated , with the result @xmath253 in the last equality the two electron correlation function @xmath254 has been identified from ( [ 2.6 ] ) . comparison with ( [ 2.5 ] ) shows that @xmath255 can be written@xmath256this leads to the representation ( [ 2.4 ] ) for @xmath248@xmath257with@xmath258@xmath259the inverse of @xmath252 has been introduced by the definition@xmath260it is verified that@xmath261if @xmath262 obeys the equation@xmath263which is a generalization of the ornstein - zernicke equation @xcite . the formal equation for @xmath48 , ( [ 2.7 ] ) , follows from differentiation of ( [ a.11 ] ) with respect to time and the identification@xmath264this provides the desired result for identifying the markovian approximation@xmath265where the property @xmath266 has been used . equation ( [ a.12 ] ) gives finally @xmath267 it only remains to calculate the initial derivative @xmath268 to determine @xmath269 . to simplify the notation it is useful to denote the force on the electron due to both the ion and the uniform positive background by @xmath270 @xmath271 \nonumber \\ & & -\int dx_{1}dx_{2}\delta \left ( x_{1}-x\right ) m^{-1}\mathbf{f}_{ee}\left ( \mathbf{r}_{12}\right ) \cdot \nabla _ { \mathbf{v}_{1 } } \nonumber \\ & & \quad \times \bigg[\left ( \delta \left ( x^{\prime } -x_{1}\right ) + \delta \left ( x^{\prime } -x_{2}\right ) \right ) \phi \left ( v_{1}\right ) \phi \left ( v_{2}\right ) n\left ( \mathbf{r}_{1},\mathbf{r}_{2}\right ) \nonumber \\ & & \quad + \int dx_{3}\delta \left ( x^{\prime } -x_{3}\right ) \phi \left ( v_{1}\right ) \phi \left ( v_{2}\right ) \phi \left ( v_{3}\right ) \left ( n\left ( \mathbf{r}_{1},\mathbf{r}_{2},\mathbf{r}_{3}\right ) -n\left ( \mathbf{% r}_{1},\mathbf{r}_{2}\right ) n\left ( r^{\prime } \right ) \right ) \bigg ] \nonumber \\ & = & -\left ( \mathbf{v}\cdot \nabla _ { \mathbf{r}}+m^{-1}\mathbf{f}_{0}\left ( \mathbf{r}\right ) \cdot \nabla _ { \mathbf{v}}\right ) g(x , x^{\prime } ; 0 ) \nonumber \\ & & \ -\nabla _ { \mathbf{v}}\cdot \delta \left ( x^{\prime } -x\right ) \phi \left ( v\right ) m^{-1}\int d\mathbf{r}_{2}\mathbf{f}_{ee}\left ( \left\vert \mathbf{r}-\mathbf{r}_{2}\right\vert \right ) n\left ( \mathbf{r},\mathbf{r}% _ { 2}\right ) \nonumber \\ & & \quad + \beta \phi \left ( v\right ) \phi \left ( v^{\prime } \right ) \mathbf{v}% \cdot \bigg[\mathbf{f}_{ee}\left ( \left\vert \mathbf{r}-\mathbf{r}^{\prime } \right\vert \right ) n\left ( \mathbf{r},\mathbf{r}^{\prime } \right ) \nonumber \\ & & \qquad \qquad \quad + \int d\mathbf{r}_{2}\mathbf{f}_{ee}\left ( \left\vert \mathbf{r}-\mathbf{r}_{2}\right\vert \right ) \left ( n\left ( \mathbf{r}_{1},% \mathbf{r}_{2},\mathbf{r}^{\prime } \right ) -n\left ( \mathbf{r}_{1},\mathbf{r}% _ { 2}\right ) n\left ( r^{\prime } \right ) \right ) \bigg ] \label{a.19}\end{aligned}\]]the two integrals on the right can be performed using the hierarchy equation ( [ a.9 ] ) for @xmath113 and the corresponding next order hierarchy equation for @xmath272 . in the current notation these are @xmath273@xmath274then ( [ a.19 ] ) becomes @xmath275next , eliminate the delta function using ( [ a.8 ] ) @xmath276substitution of this result into ( [ a.18 ] ) gives the generator for initial dynamics @xmath277 \phi \left ( v^{\prime } \right ) n\left ( r^{\prime } \right ) . \label{a.22}\end{aligned}\]]finally , using ( [ a.15 ] ) the result ( [ 2.9 ] ) is obtained@xmath278 the solution to ( [ 2.12 ] ) can be obtained in terms of an effective single electron dynamics by direct integration @xmath279where the generator for the effective single particle dynamics is@xmath280and the source term @xmath281 is @xmath282the initial condition @xmath255 is given by ( [ 2.5 ] ) . it is convenient at this point to introduce the corresponding laplace transform@xmath283then laplace transformation of ( [ b.1 ] ) gives the equation @xmath284 @xmath285@xmath286 an equation for @xmath287 follows from substitution of ( [ b.5 ] ) into ( [ b.6])@xmath288where @xmath77 is @xmath289and @xmath290this is an integral equation for @xmath281 which can be written @xmath291the dielectric function @xmath106 is defined by @xmath292 with these results ( [ b.5 ] ) becomes@xmath293 . \label{b.12}\]]the inverse dielectric function is defined by@xmath294equation ( [ b.12 ] ) is the desired solution to the kinetic equation , in terms of the single particle dynamics of @xmath295 , since all terms on the right side are now explicit . the low frequency limit @xmath296 of @xmath297 has a simple form in terms of the electron correlations . first write @xmath106 as @xmath298 where the definition of @xmath299 in ( [ b.8 ] ) and @xmath295 in ( [ b.6 ] ) have been used . then taking the real and imaginary parts of @xmath158 going to zero gives@xmath300it follows from the generalized ornstein - zernicke equation ( [ a.15 ] ) that the inverse of @xmath301 is @xmath302where the static structure factor is defined by@xmath303 the high frequency limit @xmath304 of @xmath305 also has a simple form@xmath306this implies no screening at asymptotically short times . the laplace transform of the correlation function ( [ 2.4 ] ) is @xmath307substitution of the solution ( [ b.12 ] ) gives @xmath308 . \label{c.2}\]]considerable simplification occurs for the special case where @xmath309 , i.e. it is independent of the velocity,@xmath310\end{aligned}\]]@xmath311where use has been made of the definition ( [ b.8 ] ) for @xmath312 . it follows from ( [ b.11 ] ) that@xmath313so ( [ c.3 ] ) becomes@xmath314 \widetilde{i}_{0}(% \mathbf{r}^{\prime } , z ) \nonumber\]]@xmath315where ( [ b.9 ] ) for @xmath316 has been made explicit , and the dummy labels @xmath317 have been interchanged . finally , this can be put in the simple form @xmath318where@xmath319use has been made of the fact that @xmath320 commutes with @xmath93 . the dynamic structure factor is the autocorrelation function for the electron density , corresponding to @xmath321the correlation function ( [ c.6 ] ) in this case is@xmath322where @xmath83 is the static structure factor of ( [ b.17 ] ) . b. talin , a. calisti , j. dufty , i. pogorelov , phys . rev . e 77 , 036410 ( 2008 ) ; j. w. dufty , i. pogorelov , b. talin , and a. calisti , j. phys . a * 36 * , 6057 ( 2003 ) ; j. dufty , b. talin , and a. calisti , in _ theory of energy deposition _ , adv . quant . chem . * 46 * , 293 ( 2004 ) . strictly speaking the dielectric function is defined in terms of the response function and is different in general from that given here . they agree only in the weak electron - electron limit . however , it is a convenient terminology as the function considered here does determine the collective modes .	a theoretical description of time correlation functions for electron properties in the presence of a positive ion of charge number @xmath0 is given . the simplest case of an electron gas distorted by a single ion is considered . a semi - classical representation with a regularized electron - ion potential is used to obtain a linear kinetic theory that is asymptotically exact at short times . this markovian approximation includes all initial ( equilibrium ) electron - electron and electron - ion correlations through renormalized pair potentials . the kinetic theory is solved in terms of single particle trajectories of the electron - ion potential and a dielectric function for the inhomogeneous electron gas . the results are illustrated by a calculation of the autocorrelation function for the electron field at the ion . the dependence on charge number @xmath0 is shown to be dominated by the bound states of the effective electron - ion potential . on this basis , a very simple practical representation of the trajectories is proposed and shown to be accurate over a wide range including strong electron - ion coupling . this simple representation is then used for a brief analysis of the dielectric function for the inhomogeneous electron gas .
the parameters used in the line fittings for fig . [ largesystemsims ] are shown in table [ scaling ] . .fitting parameters of curves in fig . [ scaling ] . [ cols="^,^,^",options="header " , ] for a rectangular network of the type shown in fig . [ network - pt ] , the hamiltonian reads @xmath111 here , the onsite energy @xmath112 and quantum hopping strength @xmath113 need not be all identical , but we assume that @xmath114 for all nearest neighbour terms . if all sites on the perimeter are exit sites ( yellow sites in fig . [ network - pt ] ) , i.e. , coupled to the reaction centre , a locked state is an eigenstate of the hamiltonian with zero amplitude on the whole perimeter . in general , in the single - excitation subspace the locked state can be written as @xmath115 where the amplitude @xmath116 for all @xmath117 . here , the set @xmath118 contains the four corner sites , and the set @xmath119 contains all other sites on the perimeter . furthermore , we use @xmath120 to denote sites on an imagined inner perimeter ( blue sites in fig . [ network - pt ] ) , all of these are coupled to perimeter sites . since , @xmath121 is an eigenstate , @xmath122 , where @xmath123 is the eigenenergy of the locked state . for perimeter sites in @xmath119 , @xmath124 , where the site @xmath125 is the only off - perimeter site coupled to the perimeter site-@xmath4 . however , from @xmath126 and @xmath127 it follows that @xmath128 . therefore , for a state to be locked , not only does its amplitude on the perimeter have to be zero , its amplitude on the inner perimeter must also vanish . letting the inner perimeter now take the role of the exit sites and repeating the previous discussion , one finds by induction that a locked state possesses zero amplitude on the whole network . in other words , there is no locked state for such a network with the whole perimeter filled by exit sites . m. santha , quantum walk based search algorithms , in theory and applications of models of computation , lecture notes in computer science , edited by m. agrawal , d. z. du , z. h. duan , and a. s. li ( springer , berlin , 2008 ) , vol . 4978 , p. 31 c. derrico , m. moratti , e. lucioni , l. tanzi , b. deissler , m. inguscio , g. modugno , m. b. plenio , and f. caruso , special issue article in focus on the frontiers of disorder physics , new . j. phys . * 15 * , 045007 ( 2013 ) .	an important challenge in quantum science is to fully understand the efficiency of energy flow in networks . here we present a simple and intuitive explanation for the intriguing observation that optimally efficient networks are not purely quantum , but are assisted by some interaction with a ` noisy ' classical environment . by considering the system s dynamics in both the site - basis and the momentum - basis , we show that the effect of classical noise is to sustain a broad momentum distribution , countering the depletion of high mobility terms which occurs as energy exits from the network . this picture predicts that the optimal level of classical noise is reciprocally related to the linear dimension of the lattice ; our numerical simulations verify this prediction to high accuracy for regular 1d and 2d networks over a range of sizes up to thousands of sites . this insight leads to the discovery that dramatic further improvements in performance occur when a driving field targets noise at the low mobility components . + + the simulation code which we wrote for this study has been made openly available at http://figshare.com/articles/quantum_classical_hybrid_transport_simulations/1050158[figshare ] . the study of energy transfer in quantum networks is a broad field , ranging from abstract theoretical studies @xcite through to experimentally observed transport dynamics of real networks , for instance in light - harvesting complexes @xcite . an important observation is that while a purely quantum mechanical energy transfer process ( i.e. a quantum random walk ) is inherently faster than the classical equivalent , nevertheless when one measures the efficiency of a network in terms of the time needed for a unit of energy to completely traverse it , then it is optimal to temper the purely quantum dynamics with a degree of classical ` noise ' . this noise may be , for example , dephasing of the quantum state or a spontaneous hopping process ; in each case the ideal level of noise is non - zero , and for the latter case this has recently been verified for networks of arbitrary topology and sizes up to thousands of sites @xcite . various explanations for this phenomenon of classically - assisted quantum transport have been discussed @xcite . in networks with disorder ( e.g. irregularities in the site couplings ) the quantum state can become localised . injecting classical noise can break this localisation , as for instance recently experimentally observed in ultra - cold atomic systems on optical lattices @xcite . however , classical noise has been found to be advantageous even in perfectly ordered networks . for such systems it is generally argued that quantum networks suffer from a kind of locking effect , where destructive interference occurs between the different possible pathways to the exit site . following the notion of invariant subspaces introduced in ref . @xcite , an equivalent statement of this effect is that some of the spatial eigenstates of the system ( which are in general highly non - local ) may have zero amplitude on a given site if such a site happened to be the exit , then any part of the initial wavefunction associated with such an eigenstate would never be able to leave the network . classical noise disrupts such eigenstates , thus alleviating the problem . in this paper we will say that a system exhibits `` quantum locking '' if there is a finite probability of energy remaining on the network as @xmath0 ( for zero noise ) . some of the regular network topologies considered here : ( a ) linear ( b ) ring , and ( c ) rectangular lattices . each circle represents a site , i.e. a molecule or other entity which can be excited by a quantum of energy ; the excitation can hop from one molecule to the nearest neighbouring molecule and leave the network via the exit site . , width=272 ] as interesting and intuitive as the quantum locking effect is , nevertheless it is not comprehensive explanation : there are a number of basic cases where classical noise does assist quantum transport , and yet provably there is no quantum locking effect present . the simplest example ( fig . [ fig : topologies]a ) is the 1d chain with an exit site at one end and an initial excitation site within the chain @xcite . as we presently discuss , many examples also exist in the more complicated landscape of 2d networks . therefore while quantum locking is certainly an important effect , we should continue to seek an explanation for the general phenomenon . in this paper , we will argue that classically - assisted quantum transport in regular networks is best understood by viewing the dynamics through a combination of the site - basis and the momentum - basis pictures . we begin by explaining the basic principle , before developing the idea more formally and finally checking the predictions versus intensive numerical simulations . our analysis in this paper is restricted to the case where there is a single quantum of energy in the network , although we anticipate that our argument translates straightforwardly to multiple energy quanta . ( we note in passing that in certain photosynthetic systems , the single excitation model is certainly appropriate since fewer than 10 photons are absorbed per molecule per second @xcite , while the coupling strength of molecules @xmath1 , and thus the transport process is @xmath2 times faster than the absorption process . ) we identify the site basis with states written as @xmath3 , which corresponds to the energy being definitely located at site @xmath4 , and we will assign the index @xmath5 to the ` exit ' site . we will presently define the momentum basis as the standard canonical complement to the site basis . to explain the process let us first consider a rather contrived model for classical noise : we will subject our quantum system to a periodic series of instantaneous events where the state is completely dephased in the site basis . we initialise our system by injecting energy at some randomly chosen location , i.e. we select some random initial state @xmath3 , and there follows a period of evolution before the first dephasing event . state @xmath3 will of course correspond to a broad superposition of states in the momentum picture . the components with a high group velocity will rapidly transit the network and will be the first to impinge on the exit site ( to ` hit ' that site in the terminology of random walks ) . there will then be a finite probability per unit time of the energy exciting the system ; over time if the energy does _ not _ exit , the wavefunction will skew further and further toward lower phase velocity terms . suppose that this has occurred for some period , and then the first of our classical noise events occurs . this dephasing event is equivalent to measuring the system in the site basis , and then forgetting the outcome . in effect we are reinitialising the system to some state @xmath6 ( although of course this choice is not purely random , since a given site @xmath7 is more likely if it is closer to the original site @xmath4 ) . but regardless of which @xmath6 we select , the important point is that the momentum distribution is now once again broad and includes elements with a high group velocity . thus our periodic classical noise process repeatedly reinvigorates , or rejuvenates , the momentum distribution and so counters the skew towards low mobility elements . in - between noise events the high group velocity components will reach the exit and be removed . clearly there is some optimal rate of classical noise : if it is too strong , i.e. the frequency of the events is too high , then after one event another will occur before the more rapidly propagating components can exit this would merely reduce the quantum random walk to a classical one without any advantage . conversely , too weak a level of classical noise will mean that we miss the opportunity to rejuvenate the momentum distribution . this line of thought leads one to conclude that the optimal frequency of the classical noise should depend on the lattice size . in a larger lattice , the high momentum components have further to go to reach the exit , and so a longer period should be allowed between the classical noise events . one might expect a @xmath8 dependence , where @xmath9 is the linear dimension of the lattice ( so that a square lattice has @xmath10 sites ) . in the following derivation , this form is indeed predicted , and the prediction is verified to a high degree of accuracy by our numerical simulations . the transport of energy in a regular network with identical sites can be modelled as ( @xmath11 ) @xmath12 + \mathcal{l}\rho + \mathcal{l}_x\rho ~,\end{aligned}\ ] ] where the coherent transport is given by the hamiltonian @xmath13 the noisy classical process is either classical hopping ( ch ) @xmath14 with @xmath15 or pure dephasing ( pd ) @xmath16 and the excitation leaves the network via the exit site-@xmath5 according to the process @xmath17 here , @xmath18 is the state of the network , @xmath19 and @xmath20 are ladder operators describing transitions between the ground state @xmath21 and the excited state @xmath22 of the site-@xmath4 , @xmath23 denotes two connected sites , @xmath24 is the coupling strength , @xmath25 and @xmath26 are the relative weightings of the quantum and classical processes ( @xmath27 ) @xcite , the constant @xmath28 for one - dimensional chain lattices and two - dimensional square lattices , respectively , and @xmath29 is the strength of the exit coupling from the network . note that we do not need to explicitly include the reaction centres depicted in fig . [ fig : topologies ] in our model . our single - excitation subspace is spanned by @xmath30 , where the state @xmath31 and @xmath32 denotes the overall ground state of the entire system . when @xmath33 , the transport process is purely quantum mechanical ; when @xmath34 , the transport becomes a completely classical random walk in the ch model , and will be switched off entirely in the pd model . here our analysis will be largely focused on the ch model , but we note that the same basic argument will apply to pd . in order to measure the transport efficiency , we look at the probability that at time @xmath35 the energy quantum has failed to exit the network , @xmath36 $ ] . this is the ` population ' remaining on the network . in cases where there is no quantum locking all population eventually vanishes ; one such system is the 1d chain with the exit at one end of the chain , as shown in fig . [ fig : topologies](a ) and fig . [ chain](a ) . moreover , it will always vanish when we have finite classical noise . therefore we can gauge the transport efficiency , or rather the inefficiency , by finding the average dwelling time in the network which we define as @xmath37 . we regard a network as optimised when this quantity has been minimised . sites with an ` exit ' site at one end [ c.f . fig . [ fig : topologies](a ) ] , and a range of classical hopping rates @xmath26 , we plot ( a ) the probability @xmath38 that the energy quantum is still on the network , and ( b ) the average dwelling time @xmath39 . the energy is initially located at a randomly chosen site , so that curves are computed using an initial @xmath40 . we set the exit coupling to be @xmath41 . in ( a ) , the green line corresponds to the pure coherent transport case @xmath33 , the red line corresponds to the pure ch case @xmath34 , and black lines represents @xmath42 from bottom to top , respectively . in ( b ) , one can find that the average dwelling time is minimised at a finite ch rate @xmath43 . , width=8 ] for the one - dimensional chain , the average dwelling time for different ch rates @xmath26 is shown in fig . [ chain](b ) . we note that the average dwelling time is minimized at @xmath43 for the chain with lattice size @xmath44 when we take the exit to be site @xmath9 . in the single - excitation subspace of the 1d chain , the momentum eigenstate with wave vector @xmath45 is @xmath46 where @xmath47 , @xmath48 , and @xmath49 is the lattice constant . the eigenenergy of the state @xmath50 is @xmath51 , and the corresponding group velocity is @xmath52 . therefore , a wavefunction formed from states in the middle of the band has a high group velocity @xmath53 , while a superposition of states at the edge of the band has a low group velocity @xmath54 . suppose that at @xmath55 our excitation is at site @xmath4 , then inverting eq . [ momentumbasis ] we see that in the momentum basis the state has a probability of @xmath56 associated with the eigenstate @xmath50 . hence , the site - localised excitation will have significant population in both the high- and the low - velocity states . as we have discussed , the higher - velocity components leave the system first , leaving behind an ever more slowly falling population remnant . suppose that a finite level of classical hopping ( ch ) is present . even though our noise models ( eqs . [ hopeqn ] and [ pdeqn ] ) are continuous , we can map this onto discrete events whereby the excitation is effectively reinitialized during the transport process . for a short time @xmath57 , the state evolves as @xmath58 ~.\end{aligned}\ ] ] in the single - excitation subspace , the effect of ch can be expressed as @xmath59 where @xmath60 hence , a ch event @xmath61 corresponds to a measurement of the position of the excitation at site @xmath7 followed by moving the excitation to a randomly chosen neighbouring molecule @xmath4 , i.e. , effectively reinitialising the excitation to one of the neighbouring sites . the presence of dephasing ( pd ) will have essentially the same effect , of course without the final ` hop ' ; it can be expressed as @xmath62 where @xmath63 thus for both our models of continuous classical noise , we may equivalently think in terms of an occasional reinitialisation process where part of the population in low - velocity states will be promoted to high - velocity states . to optimize the transport efficiency , the high - velocity component has to leave the network before the next ch or pd event happens . therefore , the optimal level of classical noise @xmath26 decreases with the time required to leave the network , i.e. , the lattice size . notice that this implies that in the limit @xmath64 , the optimal rate @xmath65 . in other words , for very large ordered lattices we expect that the classical noise mechanism described here will no longer be advantageous . in fig . [ chainm ] we show how the momentum distribution @xmath66 evolves over time , starting from a broad initial distribution corresponding to a site eigenstate @xmath3 . for the purely quantum limit @xmath33 one finds that populations in high - velocity states , @xmath67 , vanish much faster than those in low - velocity states , so our initially even distribution becomes highly skewed as time passes . for the pure ch case , populations with different momenta vanish with almost the same rate , but this rate is much slower than the coherent transport ( note the units of the time axis are ten times greater ) . in the ` best of both worlds ' case a modest level of classical hopping serves to rejuvenate the distribution , i.e. to keep it relatively even . the high - velocity components can still escape from the network quite efficiently , while at the same time the lingering wings of the distribution are depleted as we transfer population from low - velocity states to high - velocity states via ch events . as a result , although population in high - velocity states persists for longer than it does in the purely quantum case , nevertheless the overall transport efficiency is increased . the analysis so far has been presented in terms of the 1d lattice , but the exactly the same argument applies to higher dimensional regular lattices . we now present the results of numerical simulations which we have performed on 1d and 2d arrays in order to test the predicted @xmath8 scaling of the optimal noise level , where @xmath9 is the linear dimension of the array . to be specific , we should anticipate that the quantity @xmath68 will scale in this way , according to the following reasoning : the effective rate at which we reinitialise the momentum distribution goes with @xmath69 , but the rate at which the quantum coherent evolution occurs is proportional to @xmath70 , see eq . [ generalh ] . thus the width which the excitation s spatial distribution can reach between reinitialisation events varies with @xmath71 , and we expect that this width should be proportional to the linear lattice size @xmath9 ( and therefore the average distance between a randomly chosen site and the exit site ) . thus we expect to see @xmath72 where @xmath73 is a constant of proportionality and we have also added an adjustment @xmath74 to allow for finite size effects ( which should be significant only for small arrays ) . we do not predict the specific values of these constants , which will depend for example on the location of the exit site ( corner , edge or internal to the lattice ) . however , we expect that the above expression will be accurately obeyed if @xmath73 and @xmath74 are treated as free fitting parameters . in fig . [ largesystemsims ] we display the results of a series of numerical experiments where we test our hypothesis . we find that the proposed function does indeed fit the data well . corresponding fitting parameters and confidence intervals are shown in appendix i. the optimal classical hopping ( ch ) rate @xmath75 of ( a ) one - dimensional chain lattices and ring lattices with @xmath9 sites , ( b ) two - dimensional square lattices , torus lattices with @xmath76 sites , and rectangle lattices with @xmath77 sites . the optimal ch rate depends on the position of the exit site . the exit site is placed at the end on chain lattices and a corner on square and rectangle lattices as examples . optimal ch rates given by the average remaining population for all possible positions of the exit site are also considered for chain and square lattices . the curves are obtained by fitting the function @xmath78 , and fitting parameters are shown in table [ fitting ] . we have supposed that @xmath41 and the energy is initially located at a randomly chosen site . , width=8 ] it is worth noting that one can obtain a reasonable estimate of the optimal rate of classical noise merely by applying our simple discretised picture : fully dephasing events occur intermittently , in place of the real continuous noise . we suppose that the single - excitation state evolves coherently , i.e. , @xmath33 , and is completely reinitialized with the period @xmath79 . a complete reinitialization reads @xmath80 which means the excitation is initialized at a randomly chosen site . if @xmath38 denotes the population in the network at time @xmath35 when @xmath33 ( i.e. the green curve in fig . [ chain ] ) , the population after @xmath81 occurrences of the reinitialization event with period @xmath79 is @xmath82 . therefore , roughly speaking , the population in the network decays with a rate @xmath83 . for the chain with the size @xmath44 shown in fig . [ chain ] , we extract a maximum of the decay rate @xmath84 at @xmath85 . comparing the frequency of complete reinitializations with the ch rate , @xmath86 , then this optimal period corresponds to the ch rate @xmath87 , which is remarkably close to the observed optimal ch rate @xmath88 . how does the role of noise in overcoming quantum locking compare to its role in this momentum rejuvenation picture ? we know that quantum locking is a significant effect in many systems . for example in a regular 2d square array there will always be finite quantum locking in the absence of noise , and therefore the assistance provided by noise is , in part , to remove this locking effect . however there are also many cases where there is no quantum locking in the zero - noise limit . our numerical simulations indicate that such cases include the following ( for any location of the initial state and exit , unless otherwise noted ) : * a 1d array with the exit at one end , and the initial excitation site within the chain . * a 2d array @xmath89 with @xmath90 and the exit in a corner . * any 2d array with dimensions @xmath91 , where @xmath92 and @xmath93 are two different primes . * any 2d array with multiple exit sites forming the perimeter ( shown analytically in appendix ii ) . * any 2d array with ( arbitrarily small ) random perturbations to the coupling strengths or the on - site energies . for all these cases we would expect that the momentum rejuvenation picture still applies , since it has no special reliance on symmetries in the system . and indeed in all these cases we find that there is a finite value of classical noise which optimises the network . therefore we conjecture that the momentum rejuvenation picture is a general explanation for noise - assisted transport in regular arrays ; we see the role of noise in disrupting quantum locking as an important additional factor in many systems . in support of this view , we note that fig . [ largesystemsims ] ( b ) contains one instance of a network where locking is present ( the square array ) and one for which it is not ( rectangular with exit on the corner ) . for small networks we note that the optimal amount of noise is higher for the square lattice compared to the rectangular one . in fact , this is not unexpected since here the noise is advantageous for overcoming the locking effect as well as maintaining a broad momentum distribution . and the higher excited state @xmath94 . because of the hopping coupling , a ` conduction ' band with the width @xmath95 is formed around the energy of the state @xmath22 . ( b ) colour map showing the populations associated with momentum eigenstates @xmath96 on a network with driving fields ; as in fig . [ chainm ] ( which shares the same colour scale ) , a one - dimensional chain lattice with @xmath44 sites is considered as an example . here , the rabi frequency @xmath97 and the decay rate @xmath98 . comparison with fig . [ chainm ] shows that the population remaining on the network at time @xmath99 is significantly smaller than for the passive optimal quantum - classical hybrid network ( 0.0024 versus 0.0473 ) . ] one might ask if the random classical noise from a network s environment is an ideal means of maximising the transport efficiency , or whether other mechanisms might lead to even higher efficiency . according to the picture we have presented , classical noise effectively rejuvenates the whole momentum distribution indiscriminately . it would presumably be even more beneficial if one could somehow preferentially rejuvenate the low velocity states , leaving the high velocity state unperturbed . we now present a simple model showing this is possible when a global driving field is applied . let us assume that each lattice site possesses an additional excited state @xmath94 [ see fig . [ chain2ectdrivem ] ( a ) ] . these higher excited levels form a second excited tier of the network , which will in general experience different hopping strengths @xmath100 . for simplicity let us assume for now that no classical noise is present and that the coupling strength @xmath100 is negligible compared to @xmath24 . ( note that only @xmath101 is a requirement for the control protocol described here , but the present assumption makes the following argument particularly straightforward . ) as a result , we have purely quantum transport in the lower excited manifold spanned by the @xmath102 levels , governed by the hamiltonian [ generalh ] with @xmath33 . let @xmath103 be the energy difference between @xmath22 and @xmath94 ; the application of global driving fields with frequencies @xmath104 will then drive transitions between low velocity @xmath22 states [ indicated by red circles in fig . [ chain2ectdrivem ] ( a ) ] and @xmath94 states . at the same time , high - velocity transitions are detuned and thus suppressed . the driving hamiltonian is given by @xmath105 ~,\end{aligned}\ ] ] where the rabi frequency @xmath106 is proportional to the intensity of the applied field , and @xmath107 and @xmath108 are ladder operators describing transitions between the state @xmath22 and the state @xmath94 of the site @xmath4 . the only other process we require is dissipative relaxation from the second excited tier back into the lower excited manifold , e.g. phonon - assisted transitions which randomise the momentum of the electronic state . these can be described by the following lindblad superoperator @xmath109 where @xmath84 is the decay rate from the state @xmath94 to the state @xmath22 . under these circumstances , the population is preferentially pumped from low - velocity states to the excited state , from which it decays to any momentum state ( as shown in fig . [ chain2ectdrivem ] ( b ) ) . thus by targeting the low - velocity state we achieve a much faster transfer ( i.e. lower population remaining on the network ) than for the optimal quantum - classical hybrid network shown in fig . [ chainm ] ( b ) . in conclusion , we offer an intuitive explanation for the observation that classically - assisted quantum transport can be more efficient than pure quantum evolution , even in systems where quantum locking is negligible . we predict that the optimal level of classical noise will scale inversely with the linear dimension of the array , and our numerical simulations have confirmed this behaviour for systems of up to @xmath110 sites . the picture we have presented is one in which the classical environment acts to continually rejuvenate the momentum distribution of the quantum particle as it traverses the network . we use this intuition to show that the use of global driving fields can be far more efficient than simple random noise . we thank viv kendon and shane mansfield for helpful conversations . this work was supported by the epsrc platform grant ` molecular quantum devices ' ( ep / j015067/1 ) and by the national research foundation and ministry of education , singapore . the work of f.c . has been supported by eu fp7 marie - curie programme ( career integration grant ) and by miur - firb grant ( project no . rbfr10m3sb ) .
[ sec : intro ] in the studies of necessary conditions for singular minimizers containing surfaces of gradient discontinuity various local jump conditions have been proposed . a partial list of such conditions include weierstrass - erdmann relations ( traction continuity and maxwell condition ) @xcite , quasi - convexity on phase boundary @xcite , grinfeld instability condition @xcite and roughening instability condition @xcite . while some of these conditions have been known for a long time , a systematic study of their _ interdependence _ have not been conducted , and a full understanding of which conditions are primary and which are derivative is still missing . the absence of hierarchy is mostly due to the fact that strong and weak local minima have to be treated differently and that variations leading to some of the known necessary conditions represent an intricate _ combination _ of strong and weak perturbations . in particular , if the goal is to find local necessary conditions of a _ strong local minimum _ , the use of weak variations gives rise to redundant information . for instance , euler - lagrange equations in the weak form should not be a part of the minimal ( essential ) local description of strong local minima . in this paper , we study strong local minimizers and our goal is to derive an _ irreducible _ set of necessary conditions at a point of discontinuity by using only `` purely '' strong variations of the interface that are complementary to the known strong variations at nonsingular points @xcite . more specifically , our main theorem states that all known local conditions associated with gradient discontinuities follow from quasi - convexity on both sides of the discontinuity plus a single interface inequality which we call the _ interchange stability _ condition . while this condition is fully explicit and deceptively simple , to the knowledge of the authors , it has not been specifically singled out , except for a cursory mention by r. hill @xcite of the corresponding _ equality _ which we call _ elastic normality condition_. to emphasize a relation between this condition and strong variations we show that it is responsible for gteaux differentiability of the energy functional along special multiscale `` directions '' . we call them _ material interchange _ variations and show that they are devoid of any weak components . we also explain why the elastic normality condition , which r. hill associated exclusively with weak variations , plays such an important role in the study of strong local minima . the paper is organized as follows . in section [ sec : prelim ] we introduce the interchange stability condition and formulate our main result . in section [ sec : norm ] we link interchange stability with a strong variation which we interpret as material exchange . we then interpolate between this strong variation and a special weak variation which independently produces the normality condition . our main theorem is proved in section [ sec : proof ] , where we also establish the inter - dependencies between the known local necessary conditions of strong local minimum . an illustrative example of locally stable interfaces associated with simple laminates is discussed in detail in section [ sec : ex ] . in section [ sec : normality ] we build a link between the notions of elastic and plastic normality and show in which sense the elastic normality condition can be interpreted as the actual orthogonality with respect to an appropriately defined `` yield '' surface . we then illustrate the general construction by studying the case of an anti - plane shear in isotropic material with a double - well energy . [ sec : prelim ] consider the variational functional most readily associated with continuum elasticity theory @xmath0 here @xmath1 is an open subset of @xmath2 , and @xmath3 is the neumann part of the boundary . we can absorb the boundary integral into the volume integral by finding a divergence - free @xmath4 matrix field @xmath5 , such that @xmath6 on @xmath3 which suggests that the variational functional @xmath7 can be used in place of ( [ energy ] ) . we assume that @xmath8 is a continuous and bounded from below function on @xmath9 , where @xmath10 is the set of all @xmath4 matrices . we use the following definition of strong local minimum : [ def : slm ] the lipschitz function @xmath11 satisfying boundary conditions is a strong local minimizer . in the study of local necessary conditions in the interior this difference is irrelevant . ] if there exists @xmath12 so that for every @xmath13 for which @xmath14 , we have @xmath15 . in this paper we focus on special singular local minimizers containing jump discontinuity of @xmath16 across a @xmath17 surface @xmath18 . then for every point @xmath19 there exist @xmath4 matrices @xmath20 and @xmath21 such that for any @xmath22 @xmath23 where @xmath24 is the unit normal '' . ] to @xmath25 . we further assume that @xmath26 which imposes kinematic compatibility constraint on the jump of the deformation gradient @xcite : @xmath27 where @xmath28 and @xmath29 is a called a shear vector . material stability of the deformation @xmath30 at point @xmath31 is understood as stability with respect to local variations of the form @xmath32 where @xmath33 is the unit ball can be supported in any bounded domain of @xmath2 , see @xcite ] . the corresponding energy variation @xmath34 is defined by @xmath35 the condition of material stability can be written in different forms for points where @xmath16 is continuous and for points on jump discontinuity where @xmath16 satisfies ( [ fpmdef ] ) . the condition of material stability in the regular point is obtained by changing variables @xmath36 in ( [ strvar ] ) , @xcite @xmath37 to be closer to standard notations we redefine @xmath38 and write the necessary condition of material stability in the form of quasi - convexity condition @xcite @xmath39 where @xmath40 denotes the average over @xmath33 . we say that @xmath41 is _ strongly locally stable _ if @xmath42 for any @xmath43 . when the point @xmath31 lies at the jump discontinuity we can again change variables @xmath36 and write @xcite @xmath44 where @xmath45 is defined in ( [ fpmdef ] ) . the associated necessary condition can be then written in the form of quasiconvexity on the surface of jump discontinuity condition @xcite @xmath46 where @xmath47 . we say that the pair @xmath48 satisfying ( [ kincomp ] ) determines a _ strongly locally stable interface _ @xmath49 if @xmath50 for any @xmath43 . it is clear that the strong local stability of the interface @xmath51 implies strong local stability ( [ qcx ] ) of @xmath52 and @xmath53 . it will be convenient to reformulate conditions of strong local stability in terms of the properties of global minimizers of localized variational problems . thus , according to ( [ qcx ] ) , @xmath54 is strongly locally stable if and only if @xmath55 is a minimizer in the localized variational problem @xmath56 the value of the infimum in ( [ homloc ] ) coincides with the quasiconvex envelope @xmath57 of @xmath58 , i.e. the largest quasiconvex function that does not exceed @xmath59 @xcite . it is then clear that @xmath41 is strongly locally stable if and only if @xmath60 . similarly we say that the pair @xmath61 satisfying ( [ kincomp ] ) determines a strongly locally stable interface if @xmath62 solves the localized variational problem @xmath63 where we defined a lipschitz continuous function @xmath64 we are now in a position to formulate our main claim that strong local stability ( [ qcx ] ) of @xmath61 together with a single additional condition , which we call interchange stability , implies strong local stability of the interface : [ th : main ] let @xmath58 be a continuous , bounded from below function that is of class @xmath65 in a neighborhood of @xmath66 . assume that the pair @xmath61 satisfies the kinematic compatibility condition @xmath67 for some @xmath68 and @xmath69 . then the surface of jump discontinuity @xmath70 is strongly locally stable if and only if the following conditions are satisfied : * material stability in the bulk : @xmath71 , * interchange stability : @xmath72 . before we turn to the proof of theorem [ th : main ] it is instructive to look closely at the meaning of the scalar quantity @xmath73 entering the algebraic condition ( @xmath74 ) [ sec : norm ] while it is natural that condition ( [ jdqcx ] ) of strong local stability of the interface implies strong local stability of each individual deformation gradient @xmath52 and @xmath53 , a less obvious claim of theorem [ th : main ] is that the only _ joint _ stability constraint on the kinematically compatible pair @xmath75 is provided by condition ( @xmath74 ) . a natural challenge is then to identify the variation producing this condition . we observe that conventional variations , linking both sides of a jump discontinuity and leading to maxwell condition @xcite or roughening instability condition @xcite , represent combinations of weak and strong variations . this creates unnecessary coupling and obscures the strong character of the minimizer under consideration . physically it is clear that if materials on both sides of the interface are stable and if we can interchange one material by another without increasing the energy , then the whole configuration should be stable . between the original and perturbed normals is plotted as a function of length in the tangential direction . ] the idea of material interchange is illustrated schematically in fig . [ fig : interchange ] where the two adjacent rectangular domains are flipped and then translated . at @xmath76 this construction can be viewed as a interface generalization of the weierstrass `` needle variation '' since neither the fields are modified , except on a set of zero surface area . as we show in fig . [ fig : normalnuc ] this variation can be also interpreted as a strong variation of the interface normal . notice that if taken literally , the schematics of perturbed field shown in fig . [ fig : interchange ] is incompatible with a gradient of any admissible variation . to fix this technical problem we present below an explicit construction of the variation whose gradient differs significantly from the one shown in fig . [ fig : interchange ] only on a set of an infinitesimal measure . we define a family of lipschitz cut - off functions @xmath77 on @xmath78 such that @xmath79 , when @xmath80 , while @xmath81 , when @xmath82 . let @xmath83 be another lipschitz cut - off function with @xmath84 , when @xmath85 and @xmath86 , when @xmath87 . suppose that @xmath88 is a unit vector in @xmath2 , such that @xmath89 . we then define the test function @xmath90 , to be used in ( [ jdstab ] ) , as follows @xmath91 where @xmath92 3 in .,title="fig : " ] 3 in .,title="fig : " ] observe that @xmath93 , where the graph of @xmath94 is given in figure [ fig : interchange2 ] . we remark that the variation ( [ wdipole ] ) belongs to the class of multiscale variations proposed in @xcite : it uses a small scale @xmath95 and another small scale @xmath96 from ( [ strong ] ) . the interpretation of the function @xmath73 as an `` interchange driving force '' is immediately clear from the following theorem : [ th : wdipole0 ] suppose @xmath66 satisfy ( [ kincomp ] ) . let @xmath97 be given by ( [ wdipole ] ) . then @xmath98 where @xmath62 is given by ( [ ybardef ] ) and @xmath99 is the @xmath100 dimensional volume of the unit ball in @xmath101 . in order to compute the energy increment @xmath102 we use the weierstrass function @xmath103 we can then rewrite the energy increment @xmath104 as @xmath105 we easily compute @xmath106 observe that @xmath107 is non - zero only on @xmath108 , while @xmath109 is non - zero only on @xmath110 . therefore , @xmath111 in order to estimate the right - hand side , we identify 3 regions where @xmath112 ( see figure [ fig : regions ] ) : @xmath113 @xmath114 and @xmath115 in order to estimate @xmath116 we write @xmath117 , where @xmath118 it is easy to see that @xmath119 and @xmath120 in regions @xmath121 , @xmath122 in the region @xmath123 . thus @xmath124 thus , @xmath125 in the region @xmath123 and @xmath126 in region @xmath127 . we also see that @xmath128 , @xmath129 , while @xmath130 . thus , we estimate @xmath131 while @xmath132 we conclude that @xmath133 combining ( [ farfield ] ) and ( [ locfield ] ) we obtain ( [ wdipolevar ] ) . while the theorem [ th : wdipole0 ] associates the interchange stability ( @xmath74 ) with strong variations , the function @xmath73 is also known to be linked with stability with respect to weak variations . indeed , after being projected onto the shear vector @xmath134 , the traction continuity condition @xmath135 , which can be viewed as a weak form of euler - lagrange equations , gives the following _ normality condition _ @xcite @xmath136 to understand the origin of ( [ norm ] ) consider the energy increment corresponding to classical weak variations @xmath137 we obtain @xmath138 the formula ( [ tracvar ] ) shows that if @xmath139 then the vanishing of the first variation implies the normality condition ( [ norm ] ) . the crucial observation is that our strong variation @xmath97 given by ( [ wdipole ] ) is also a scalar multiple of @xmath134 . this suggests the idea that our both weak and strong variations can be regarded as two limits of a single continuum of variations @xmath140\}$ ] interpolating between them . indeed , it is easy to see that if @xmath141 for fixed @xmath95 , the two - parameter family of functions @xmath142 converges to the weak variation ( [ weak ] ) , and when @xmath143 at @xmath144 , we obtain the strong variation ( [ wdipole ] ) . to compute the corresponding asymptotics of the energy increment we can use the method used in the proof of theorem [ th : wdipole0 ] which is applicable for any @xmath145 . let @xmath146 rewriting the energy increment in terms of the weierstrass function we obtain @xmath147 thus , @xmath148 when @xmath145 is fixed we repeat the steps in the proof of theorem [ th : wdipole0 ] and obtain @xmath149 hence , using the formula ( [ farfield ] ) we obtain @xmath150 and @xmath151 from ( [ hthent ] ) we obtain @xmath152 which shows that the @xmath95 and @xmath153 limits commute . the existence of an explicit interpolation between weak and strong variations suggests to examine the behavior of the normalized energy along the connecting path . to this end , consider the expression @xmath154 on the interval @xmath155 $ ] . to ensure the symmetry of the two limits , we consider the special case @xmath156 . ) and the interchange ( @xmath144 ) variation.,title="fig : " ] ) and the interchange ( @xmath144 ) variation.,title="fig : " ] it is clear that the fine structure of the energy landscape along such a path is not universal and depends sensitively on the function @xmath58 . for the purpose of illustration , let us consider the energy density @xmath157 assuming kinematic compatibility ( [ kincomp ] ) and normality ( [ norm ] ) we obtain @xmath158 where @xmath159 with @xmath160 satisfying @xmath161 one can see that if the function @xmath162 has a `` double - well '' structure ( see figure [ fig : ws - interch](a ) ) the graph of @xmath163 looks like @xmath164 , see figure [ fig : ws - interch](b ) . the presence of `` energy barrier '' indicates that any `` combination '' of the interchange variation and the weak variation ( [ weak ] ) produces a cruder test of stability than either of the pure variations , incapable of detecting the existing instability . this result confirms our intuition that the realms of weak and strong variations are well separated and that the energy landscapes in the strong and weak topologies can be regarded as unrelated ( unless all non - trivial features are removed by assuming uniform convexity or quasiconvexity ) . we conclude this section by proving an important property of the interchange driving force @xmath73 . more specifically , we show that if the deformation gradients @xmath61 are strongly locally stable and if they are linked only by the kinematic compatibility condition ( [ kincomp ] ) , then the interchange driving force @xmath73 is non - negative . we first recall the definition of the maxwell driving force @xcite @xmath165 where @xmath166 . [ th : pospres ] assume that both @xmath52 and @xmath53 are strongly locally stable and satisfy the kinematic compatibility condition ( [ kincomp ] ) . then @xmath167 in particular , @xmath168 . the theorem is an immediate consequence of lemma [ lem : qcxnorm ] below , that shows that the algebraic inequality ( [ normest ] ) is a consequence of the `` weierstrass condition '' stated in the next lemma . [ lem : weier ] suppose @xmath54 is strongly locally stable . then the weierstrass condition holds @xmath169 the proof of the lemma can be found in @xcite . [ lem : qcxnorm ] suppose that both @xmath52 and @xmath53 satisfy the weierstrass condition ( [ arest ] ) and the kinematic compatibility condition ( [ kincomp ] ) . then the inequality ( [ normest ] ) holds . setting @xmath170 and @xmath171 in ( [ arest ] ) , we obtain @xmath172 writing @xmath173 , where @xmath174 , we obtain @xmath175 the inequality ( [ normest ] ) follows . theorem [ th : pospres ] , whose proof is now straightforward , quantifies to what extend conditions of stability of surfaces of jump discontinuity are stronger than conditions of strong local stability of each individual phase . [ sec : proof ] we are now in a position to prove theorem [ th : main ] . the necessity of ( s ) was already observed in section [ sec : prelim ] , and the necessity of ( i ) , even with equality , was shown in section [ sec : norm ] . the proof of sufficiency will be split into a sequence of lemmas . our first step will be to recover the known interface jump conditions . in order to prove these algebraic relations only the weierstrass condition ( [ arest ] ) will be needed . [ lem : jc ] assume that the pair @xmath61 satisfies the following three conditions * kinematic compatibility : @xmath67 for some @xmath68 and @xmath69 , * interchange stability of the interface : @xmath176 , * weierstrass condition : @xmath177 for all @xmath178 and @xmath179 . then the following interface conditions must hold : * the maxwell jump condition @xmath180 * traction continuity @xmath181 * interface roughening condition @xcite @xmath182 combining lemma [ lem : qcxnorm ] with ( i ) we conclude that @xmath156 , and hence , by ( [ normest ] ) , the maxwell condition ( [ maxwell ] ) holds . in order to prove the remaining equalities we set @xmath183 and @xmath184 in the weierstrass condition ( w ) , where @xmath134 and @xmath185 are as in ( [ kincomp ] ) and @xmath186 and @xmath88 are small parameters . then we obtain a pair of inequalities @xmath187 that hold for all @xmath188 and all @xmath189 . under our smoothness assumptions on @xmath58 the functions @xmath190 are of class @xmath65 in the neighborhood of @xmath191 in the @xmath192-space . the taylor expansion up to first order in @xmath192 gives @xmath193 then the inequalities ( [ feps ] ) , together with @xmath156 and ( [ maxwell ] ) imply ( [ phbequil ] ) and ( [ pta ] ) . next we prove a differentiability lemma that guarantees the existence of rank-1 directional derivatives of quasiconvex and rank-1 convex envelopes at `` marginally stable '' deformation gradients @xcite . this result does not require any additional growth conditions , as in the envelope regularity theorems from @xcite . [ lem : r1cx ] let @xmath194 be a rank - one convex function such that @xmath195 . let @xmath196 where @xmath197 is an open subset of @xmath10 on which @xmath58 is of class @xmath17 . then for every @xmath198 and every @xmath178 , @xmath179 @xmath199 in particular , @xmath200 by our assumption @xmath201 is convex on @xmath202 . recall from the theory of convex functions that @xmath203 is monotone increasing on each of the intervals @xmath204 , @xmath205 . therefore , the limits @xmath206 exist . moreover , the convexity of @xmath207 implies that @xmath208 . let @xmath209 . by assumption @xmath210 . we also have @xmath211 , since @xmath198 . when @xmath145 @xmath212 therefore , @xmath213 . similarly , when @xmath87 we obtain @xmath214 . thus , @xmath215 we conclude that the limit on the left - hand side of ( [ deriv ] ) exists and is equal to @xmath216 thus , the convex function @xmath207 is differentiable at @xmath217 and @xmath218 is a tangent line to its graph at @xmath217 . convexity of @xmath207 then implies that @xmath219 for all @xmath220 . in general , one does not expect explicit formulas for the values of the quasiconvex envelope @xmath221 in terms of @xmath59 . in that respect lemma [ lem : rwformula ] below provides a nice exception to the rule . [ lem : rwformula ] assume that the pair @xmath61 satisfies all conditions of theorem [ th : main ] . then @xmath222 for all @xmath223 $ ] , where @xmath224 is the rank-1 convex envelope of @xmath58 @xcite . by assumption ( s ) we have @xmath225 . therefore , @xmath226 by rank - one convexity of @xmath57 @xcite and assumptions ( k ) and ( s ) we have @xmath227 to prove the opposite inequality we apply lemma [ lem : r1cx ] and obtain @xmath228 thus , lemma [ lem : jc ] , ( [ rwineq1 ] ) and ( [ rwineq2 ] ) result in the formula for @xmath229 from ( [ rwformula ] ) . we also have , in view of the assumption ( k ) and ( [ qcxr1 ] ) , that @xmath230 formula ( [ rwformula ] ) is now proved . we are now ready to establish the inequality ( [ jdqcx ] ) , proving theorem [ th : main ] . let @xmath231 be an arbitrary test function . let @xmath232 be the cube with side 2 centered at the origin and having a face with normal @xmath185 . consider a @xmath232-periodic function @xmath233 on @xmath2 given on its period @xmath232 by @xmath234 assumption ( k ) implies that @xmath235 where @xmath236 is a 2-periodic saw - tooth function such that that @xmath237 , @xmath238 $ ] . let @xmath239 be @xmath232-periodic function such that @xmath240 this function is lipschitz continuous , since @xmath241 and @xmath242 . the function @xmath57 is quasiconvex and the function @xmath243 is @xmath232-periodic . therefore ( see @xcite ) , @xmath244 by lemma [ lem : rwformula ] @xmath245 hence @xmath246 the inequality ( [ jdqcx ] ) is proved , since @xmath247 is supported on @xmath33 . we remark that theorem [ th : main ] answers the question studied in @xcite by giving a complete characterization of all possible pairs of deformation gradient values @xmath61 that can occur on a stable phase boundary . global minimality of @xmath62 , given by ( [ ybardef ] ) also implies that any other interface conditions , like , for example , local grinfeld condition @xcite or roughening stability inequality ( * ? ? ? * remark 4.2 ) , must be consequences of ( k ) , ( i ) and ( s ) . [ sec : ex ] in this section we establish a relation between _ particular _ solutions of the variational problems ( [ homloc ] ) and ( [ jdloc ] ) which elucidates the role played in the theory by the normality condition . consider the set @xmath248 of all @xmath41 that are not strongly locally stable ; we called this set the `` elastic binodal '' in @xcite . for such @xmath54 the infimum in the variational problem ( [ homloc ] ) may be reachable only by minimizing sequences characterized by their young measures @xcite . suppose that for some @xmath249 the young measure solution of ( [ homloc ] ) has the form of a simple laminate @xcite : @xmath250 the set @xmath251 of all such @xmath249 will be called the _ simple laminate region_. there is a direct connection between the simple laminate region @xmath251 and locally stable interfaces . [ th : link ] a strongly locally stable interface determined by @xmath61 corresponds to a straight line segment @xmath252 , so that the laminate young measure ( [ ymsol ] ) solves ( [ homloc ] ) with @xmath253 . conversely , every point @xmath254 , corresponding to a laminate young measure ( [ ymsol ] ) determines a strongly locally stable interface . if @xmath70 is a strongly locally stable interface determined by @xmath52 and @xmath53 then the pair @xmath61 satisfies conditions ( k ) , ( i ) and ( s ) . by lemma [ lem : rwformula ] the gradient young measure ( [ ymsol ] ) attains the minimum in ( [ homloc ] ) for @xmath253 , @xmath255 , and thus @xmath254 . if @xmath251 has a non - empty interior then formula ( [ rwformula ] ) says that the graph of the quasiconvex envelope @xmath57 over @xmath251 is formed by straight line segments joining @xmath256 and @xmath257 . in other words the graph of @xmath57 is a ruled surface . conversely , if the gradient young measure ( [ ymsol ] ) attains the minimum in ( [ homloc ] ) , then @xmath61 satisfy the kinematic compatibility condition ( [ kincomp ] ) and @xmath258 the difference between ( [ single ] ) and ( [ rwformula ] ) is that ( [ single ] ) is assumed to hold for a single fixed value of @xmath255 . therefore , both material stability ( s ) at @xmath61 and interchange stability ( i ) need to be established . [ lem : cxty ] assume that the pair @xmath61 satisfies ( [ kincomp ] ) and that ( [ single ] ) holds for some @xmath259 . then @xmath71 and @xmath156 . the proof is based on the following general property of convex functions . [ lem : cx ] let @xmath260 be a convex function on @xmath155 $ ] . suppose that @xmath261 for some @xmath259 . then @xmath262 for all @xmath223 $ ] . by convexity @xmath263.\ ] ] if @xmath264 , then @xmath265 therefore , by the assumption of the lemma and convexity of @xmath266 @xmath267 it follows that @xmath268 this inequality in combination with ( [ cxdef ] ) establishes ( [ aff ] ) for @xmath264 . the proof of ( [ aff ] ) for @xmath269 is similar . to prove lemma [ lem : cxty ] we recall that @xmath270 for all @xmath41 . by ( [ single ] ) and rank-1 convexity of @xmath57 we have @xmath271 which is possible if and only if @xmath71 . then , defining @xmath272 and applying lemma [ lem : cx ] we obtain ( [ rwformula ] ) . we can also apply lemma [ lem : r1cx ] with @xmath170 , @xmath273 . the formula ( [ deriv ] ) allows us to differentiate ( [ rwformula ] ) at @xmath217 and @xmath144 : @xmath274 subtracting the two equalities we obtain @xmath156 . thus , we have shown that every @xmath54 in the simple laminate region @xmath251 gives rise to the pair @xmath61 satisfying all conditions of theorem [ th : main ] . theorem [ th : main ] then implies that the interface @xmath51 determined by @xmath61 is strongly locally stable . theorem [ th : link ] is now proved . [ rem : marg ] the system of algebraic equations ( [ kincomp ] ) , ( [ maxwell ] ) , ( [ phbequil ] ) and ( [ pta ] ) defines a co - dimension 1 surface @xmath275 called the `` jump set '' . we have shown in @xcite that under some non - degeneracy assumptions the jump set must lie in the closure of the binodal region @xmath248 . in fact all points on the jump set are `` marginally stable '' and detectable through the nucleation of an infinite layer in an infinite space @xcite . it follows that the existence of a strongly locally stable interface has significant consequences for the geometry of @xmath248 . the presence of stable interfaces implies that a part of the jump set must coincide with a part of the `` binodal '' , the boundary of @xmath248 . the rank-1 lines joining @xmath52 and @xmath53 , both of which lie on the binodal , cover the simple lamination region @xmath276 . [ sec : normality ] in this section we show that the algebraic equation @xmath277 interpreted above as condition of interchange equilibrium , is conceptually similar to the well known _ normality condition _ in plasticity theory @xcite . to build a link between the two frameworks we now show that a microstructure in elasticity theory plays the role of a `` mechanism '' in plasticity theory . consider a loading program with affine dirichlet boundary conditions @xmath278 . suppose that @xmath279 for an interval of values of the loading parameter @xmath153 . then , for every @xmath153 the deformation gradient @xmath280 will be accommodated by a laminate ( [ ymsol ] ) , so that @xmath281 we now interpret the representation ( [ ldprog ] ) from the point of view of plasticity theory . while the deformations associated with the change of @xmath52 and @xmath53 in each layer of the laminate are elastic , the deformation associated with the change of parameter @xmath153 , affecting the microstructure and modifying the young measure @xmath282 , can be regarded as `` inelastic '' . in fact , it is similar to lattice invariant shear characterizing elementary slip in crystal plasticity theory . to be more specific , we can decompose the strain rate @xmath283 as follows @xmath284 where @xmath285 is the elastic strain rate and @xmath286 is the `` plastic '' strain rate . next we notice that in equilibrium the `` inelastic '' strain rate @xmath287 defines an _ affine _ direction along the quasiconvex envelope of the energy ( see lemma [ lem : rwformula ] ) . this suggests that there is a stress plateau with which one can associate a notion of the `` yield '' stress @xcite . to find an equation for the corresponding `` yield surface '' we choose a special loading path where the elastic fields in the layers do not change @xmath288 . then , differentiating ( [ rwformula ] ) in @xmath153 , we find that the total stress field @xmath289 lies on the hyperplane @xmath290 which we interpret as the `` yield surface '' associated with `` plastic '' mechanism @xmath291 . if we now rewrite our elastic normality condition @xmath156 in the form @xmath292 it becomes apparent that the `` plastic '' strain rate @xmath286 is orthogonal to the yield surface @xmath293 . to strengthen the analogy we observe that in plasticity theory the yield surface marks the set of minimally stable elastic states @xcite . in elastic framework the states @xmath61 adjacent to the jump discontinuity are also only marginally stable , see remark [ rem : marg ] . the fact that in elasticity setting the normality condition appears as a part of energy _ minimization _ while in plasticity theory it is usually derived by maximizing plastic _ dissipation _ , is secondary in view of the implied rate independent nature of plastic dissipation @xcite . the analogy between elastic and plastic normality conditions becomes more transparent if we consider a simple example . suppose that our material is isotropic and the deformation is anti - plane shear . take the energy density in the form @xmath294 where @xmath295and the shear moduli of the `` phases '' @xmath296 are positive . in this scalar example the quasiconvex and convex envelopes of the energy density coincide , and hence we can write ( see fig . [ fig : aps ] ) @xmath297 \dfrac{\mu_{-}}{2}\|\bf\|^{2}+w_{-},&\text { if } \|\bf\|\ge{\varepsilon}_{-}\\[3ex ] \text { if } { \varepsilon}_{+}\le\|\bf\|\le{\varepsilon}_{-}. \end{array}\right.,\ ] ] where @xmath298 observe also that the binodal region @xmath299 coincides with the simple laminate region @xmath251 since for @xmath300 the gradient young measures @xmath301 attain the infimum in ( [ homloc ] ) . by fixing @xmath52 on the circle @xmath302 we then obtain the unique @xmath303 which furnishes the `` plastic '' mechanism @xmath291 . the associated `` yield plane '' @xmath293 can be written explicitly @xmath304 we observe that as @xmath52 is varied over the circle @xmath305 , the yield lines @xmath293 form an envelope of the circle @xmath306 in stress space ( see fig . [ fig : env ] ) , which is the image of the binodal @xmath307 under the map @xmath308 . since the stress in each phase of the laminate is always the same @xmath309 we can write @xmath310 thus , in an arbitrary loading program the total stress @xmath311 will be confined to the yield surface envelope @xmath312 , provided @xmath313 . two cautionary notes are in order . first , in contrast with conventional plasticity theory , the regions of stress space both inside and outside of the `` yield '' surface @xmath312 are elastic . this distinguishes our `` transformational plasticity '' where hysteresis is infinitely narrow , from the classical plasticity where hysteresis is essential . such geometric picture continue to hold as long as @xmath314 , in particular , it holds for all scalar problems ( @xmath315 ) . the second observation is that for @xmath316 , our `` hardening free '' plastic analogy breaks down because the total stress in an arbitrary loading program is no longer confined to any surface . in this case the `` plastic '' mechanism operates on a set of full measure and the proposed analogy requires a generalization . this material is based upon work supported by the national science foundation under grant no . 1008092 and the french anr grant evocrit ( 2008 - 2012 ) . j. d. eshelby . energy relations and energy momentum tensor in continuum mechanics . in m. kanninen , w. adler , a. rosenfeld , and r. jaffee , editors , _ inelastic behavior of solids _ , pages 77114 . mcgraw - hill , new york , 1970 . l. tartar . tude des oscillations dans les quations aux drives partielles non linaires . in _ trends and applications of pure mathematics to mechanics ( palaiseau , 1983 ) _ , pages 384412 . springer - verlag , berlin , 1984 .	strong local minimizers with surfaces of gradient discontinuity appear in variational problems when the energy density function is not rank - one convex . in this paper we show that stability of such surfaces is related to stability outside the surface via a single jump relation that can be regarded as interchange stability condition . although this relation appears in the setting of equilibrium elasticity theory , it is remarkably similar to the well known _ normality _ condition which plays a central role in the classical plasticity theory .
many ( most ? ) of the massive stars are formed in clusters and the observations reveal that many cluster members are binary components ( to illustrate , see sana et al . , 2007 ) . the effect of binaries on population studies has been discussed many times by different research groups ( the brussels group was one of them ) but the following points may be worth repeating : * binaries with initial primary mass larger than 40 @xmath4 and with a period that is large enough so that lbv mass loss happens before the roche lobe overflow ( rlof ) would start , avoid rlof ( the lbv scenario as it was introduced in vanbeveren , 1991 ) ; this may not be true for case a binaries * most of the primaries of binaries with orbital period as large as 10 yrs and with initial mass smaller than 40 @xmath4 lose most of their hydrogen rich layers by rlof ; the evolution of these stars is therefore quite different from the evolution of single stars with the same mass * the rlof in late case b and case c binaries may be accompanied by the common envelope process that may result in the merger of the two components ; it can be expected that this merger will be a rapid rotator with peculiar chemical abundances * the rlof in case a and early case b binaries implies mass and angular momentum transfer towards the gainer ; together with the merger process discussed above , this process is a natural way to make rapid rotators and therefore massive close binaries may be natural progenitors of long gamma ray bursts ( cantiello et al . , 2007 ) * the favored model for short gamma ray bursts is the binary merger of two relativistic compact stars and it can not be excluded that such mergers are important sites of galactic r - process enrichment ( de donder and vanbeveren , 2003a ) * the observed massive binary frequency is not necessarily the binary frequency at birth . this is due to the fact that a significant fraction of all massive binaries become single stars during their evolution ( many binaries with small mass ratio merge , late case b and case c binaries evolve through a common envelope phase and may merge as well , the supernova explosion of one of the components disrupts the binary ) , e.g. many single stars may have a binary past with an evolution which is distinctly different from stars that are born as single stars . the proceedings of the conference on massive stars in interacting binaries ( eds . a. moffat and n. st .- louis ) were published a few months ago . they can be considered as an update of the monograph the brightest binaries published in 1998 ( vanbeveren et al . 1998c ) and we recommend careful reading to anyone with an open mind willing to admit that many theoretical population studies , where the effects of binaries are ignored , have mainly an academic value . in section 2 we list a few extra references that may be interesting in order to learn more about the effects of massive binaries on population studies . many massive objects ( single or binary ) are born in dense ( embedded ) clusters , containing a few 10s up to a few 1000s massive stars and even more ( e.g. , the clusters in the carina association , the arches and quintuplet clusters in the galactic bulge , the orion nebula cluster , r136 in the lmc , mgg-11 in m82 , etc . ) . in these clusters the evolution of each object may be affected by the presence of all the others , or at least by the closest neighbors . in sections 3 and 4 we discuss a few peculiar massive objects that may be the result of the dynamical evolution of stars in dense clusters , we like to call them unconventionally formed objects ( ufos ) . * _ the wr population _ : in the sixties and seventies , before it was realized that lbv and/or rsg stellar wind mass loss rates are large enough to remove most of the hydrogen rich layers of a massive star , it was generally accepted that most of the wr stars are formed through binary mass loss processes . it may sound as a surprise , but the study of vanbeveren and conti ( 1980 ) was probably among the first where convincing arguments were presented that no more than 40 - 50% of all galactic wr stars are binary members , indicating that massive single stars have to evolve into wr stars as well . this binary percentage still holds today . the effect of binaries on the wr population was described in detail in vanbeveren et al . ( 1998a , b , c ) , vanbeveren et al . ( 2007 ) ( see also j. eldridge in the present proceedings ) . since all primaries of interacting binaries become hydrogen deficient stars at the beginning of core helium burning , all these primaries become wr - like stars . therefore , a crutial parameter that affects theoretical wr - binary population studies is the minimum mass a binary component should have so that the wr - like star will be observed as a real wr star * _ the supernova rates _ : the effect of binaries on supernova rates ( types ia and ib / c ) has been investigated in much detail by tutukov et al . ( 1992 ) , belczynski et al . ( 2002 ) , also in the framework of a galactic evolutionary model ( de donder and vanbeveren , 2003b , 2004 ) . more recent studies confirm this earlier work , although proper referencing is sometimes missing . * _ the be stars _ : many be stars are the optical component of be - x ray binaries , which may indicate that they became be stars via the binary evolutionary channel ( the binary mass transfer process ) . accounting for the fact that the supernova explosion of a binary component disrupts most of the binaries , the large number of be - x ray binaries may indicate that there are a much larger number of be single stars which have a similar evolutionary past as the be stars in the x - ray binaries . more information on how binaries affect the be - star population is given in pols et al . ( 1991 ) , pols and marinus ( 1994 ) , van bever and vanbeveren ( 1997 ) . in the latter paper it was argued that the rich population of be stars in some clusters can not be explained by binary evolution alone . * _ binaries and the spectral synthesis of massive starbursts _ : belkus et al . ( 2003 ) and van bever and vanbeveren ( 2000 , 2003 ) studied the effect of binaries on the spectral synthesis of massive starbursts . note that binaries play a very important role when the starburst is at least 4 - 5 myr old . * _ binaries and galactic chemical evolution _ : many stars are formed in binaries and it therefore looks quite strange that most of the galactic chemical evolutionary studies account for single star chemical yields only . in brussels we have a project where the effect of binaries on galactic evolution is studied in detail . results have been published in an extended review by de donder and vanbeveren ( 2004 ) and in references quoted therein . notice that as far as the c , n , o , ... fe elements and r - process elements are concerned , binary yields are very different from single star yields and the effect on a galactic model is at least as large as the effect of fast rotation . but including binaries in a galactic code is not something as simple as replacing one set of yields by another . the x - ray luminosity in ultra - luminous x - ray sources ( ulxs ) may be as high as 10@xmath5 erg / s ( ptak and colbert , 2004 ) . many ( most ) of these ulxs are found in young dense star clusters ( e.g. , the ulx in the cluster mgg-11 in m82 ) , preferentially outside the cluster core . to understand these high luminosities by means of sub - eddington mass accretion onto a relativistic object , one has to accept that a black hole ( bh ) component is present with a mass as high as 1000 @xmath4 ( the term intermediate mass bh or imbh is used ) . however , note that there exists an alternative model where the high luminosities are explained by supra - eddington accretion onto a bh with a much smaller mass of 50 - 100 @xmath4 ( soria , 2007 and references therein ) . the dynamical evolution of dense massive clusters ( like mgg-11 ) obviously depends on the initial star density , but , most interestingly , is characterized by the processes of mass segregation and core collapse , which happen on a timescale similar to the evolutionary timescale of the most massive stars in the cluster . core collapse is accompanied by real physical stellar collisions between the most massive stars and the possible formation of an intermediate mass object with a mass of 1000 @xmath4 or larger . it is therefore tempting to link the core collapse process and the existence of ulxs . this has been investigated in many papers by the team working with the direct n - body starlab software ( e.g. , portegies zwart et al , 2007 , and references therein ) and working with the statistical monte - carlo software ( e.g. , fregeau and rasio , 2007 , and references therein ) . the link with ulxs has two major uncertainties : core collapse is a fact and the successive collision of the most massive stars in the core ( the term runaway merger is used ) is a fact as well , but , when this merger process stops , it is at present unclear what happens with that merger object ? does it form a star with a mass similar to the mass of the merger object , and , when a very massive star is formed , how does it evolve ? it is clear that when it forms , its evolution will be critically affected by stellar wind mass loss during core hydrogen burning ( chb ) and during core helium burning ( cheb ) . belkus et al . ( 2007 ) investigated the evolution of very massive stars with an initial mass up to 1000 mo , with a metallicity z between 0.001 and 0.04 , using the chb stellar wind mass loss rate formalism of kudritzki ( 2002 ) and the cheb wind formalism proposed by vanbeveren et al . ( 1998a , b , c ) , which is very similar to the one proposed by nugis and lamers ( 2000 ) . belkus et al . concluded that very massive stars with solar or super solar metallicity end their life as a bh with a mass not larger than 70 - 80 @xmath4 . yungelson et al . ( 2008 ) recently published their study on the evolution of very massive stars with solar metallicity . although they use ad hoc ( the term used by the authors ) stellar wind mass loss rate formalisms , they essentially arrive at the same conclusion as belkus et al . belkus et al . presented an easy evolutionary recipe for very massive stars . the latter is combined with our own direct n - body code and with our massive star evolution handler to simulate the early evolution of mgg-11 . we adopt a starburst model with 3000 massive stars with solar type metallicity and a king density distribution with a half mass radius of 0.5 pc ( more details are given in vanbeveren et al . , 2008 and in belkus et al . , 2008 ) . the main conclusion is that stellar wind mass loss during chb does not prevent the formation of a very massive object , but when the merger process stops and when this object becomes a very massive star , stellar wind mass loss determines its further evolution and an imbh is not formed . this argues against the sub - eddington model of the ulx in mgg-11 . when a ( small or large ) number of massive objects ( single stars or binaries ) form together in a cluster with small radius , the evolution of each member may be affected by the presence of the others . the dynamical evolution of such clusters implies close encounters leading to physical collisions and stellar mergers . the collision process of two massive stars ( two 88 @xmath4 stars and a 88 @xmath4 star with a 28 @xmath4 star ) has been investigated by suzuki et al . ( 2007 ) by using sph and the following conclusions are striking : * the merging lasts a few days and during the merging process 10 - 12 @xmath4 are lost * the merger product is a mixed star where the degree of homogenization is larger in case both stars have similar masses . _ @xmath0 car _ : it is tempting to link the collision process discussed above and the loss of about 10 @xmath4 during the 19th century event of this object . if the @xmath0 car progenitor experienced a dynamical encounter with a binary or with a single star that resulted in a stellar merger , the sudden very large mass loss and the very large eccentricity of the binary at present can be explained in a natural way . _ @xmath2@xmath6 velorum _ : the formation of wr+ob binaries in young dense stellar systems may be quite different from the conventional binary evolutionary scenario . mass segregation in dense clusters happens on a timescale of one or a few million years which is comparable to the evolutionary timescale of a massive star . within the lifetime of a massive star , close encounters may therefore happen very frequently . when we observe a wr+ob binary in a dense cluster of stars , its progenitor evolution may be very hard to predict . our simulations predict the following ufo - scenario of wr+ob binaries . after 4 million years the first wr stars are formed , either single or binary . due to mass segregation , this happens most likely when the star is in the starburst core . dynamical interaction with another massive object becomes probable , especially when the other object is a binary . we encountered a situation where a wr star ( a single wc - type with a mass = 10 mo ) interacts with a 16 @xmath4 + 14 @xmath4 circularized binary with a period p = 6 days . we explored the result of such an encounter by using the fewbody software of fregeau et al . the details of all these simulations will be discussed elsewhere ( belkus et al . , 2008 ) . one of the results is the following : the two binary components merge and the 30 @xmath4 merger forms a binary with the wc star with a period of 80 days and an eccentricity e = 0.3 . this binary very well resembles the wr+ob binary @xmath2@xmath6-velorum but it is clear that conventional binary evolution has not played any role in its formation . _ @xmath1 pup _ : @xmath1 pup , @xmath7 cep and bd+43@xmath83654 are 3 massive runaways with a runaway velocity between 40 km / s and 70 km / s . their location in the hr diagram suggests that they belong to the most massive star sample of the solar neighborhood ( vanbeveren et al . , 1998b , c ; hoogerwerf et al . , 2001 ; comeron & pasquali , 2007 ) . runaways can be formed by the binary - sn scenario ( blaauw , 1961 ) , where the original massive primary ( the mass loser when the roche lobe overflow process happens ) explodes and eventually disrupts the binary , leaving a neutron star remnant and a runaway secondary ( the mass gainer when the roche lobe overflow happens ) . such a scenario for @xmath1 pup was presented by vanbeveren et al . ( 1998b , c ) . to explain the significant surface helium enrichment of the star , its rapid rotation and its runaway velocity (= 70 km / s ) , the mass transfer phase and the accretion process must be accompanied by spinning - up and quasi - homogenization of the mass gainer ( the full mixing model as it was introduced by vanbeveren and de loore , 1994 ) whereas the overall evolution should have resulted in a pre - sn binary with a period of the order of 4 days . the latter requires some fine - tuning . to illustrate that the dynamical ejection mechanism is a very valuable alternative , the fewbody software of fregeau et al . ( 2004 ) was used to reproduce the observed properties of @xmath1 pup . we performed over 1 million single star - binary and binary - binary scattering experiments . the details of these experiments will also be given in belkus et al . we explored the effects of different masses and different binary periods and eccentricities and , obviously , many experiments reproduce @xmath1 pup , but to obtain a runaway velocity as observed , the binaries participating in the scattering process always have to be very close ( periods smaller than 100 days ) . most interestingly , in all our experiments , _ @xmath1 pup turns out to be a merger of 2 or 3 stars . _ _ wr 140 _ : this wc5 + o4 - 5 binary has a period of 7.9 yrs , an eccentricity e = 0.85 and minimum component masses = 23 + 62 @xmath4 . the o4 - 5 star is much younger than the wc5 star . accounting for the very large period of the binary and the very large masses of both components , a typical binary rejuvenation process during rlof seems unlikely . the only alternative then is a dynamical process , e.g. single - binary or binary - binary . interestingly , a dynamical process explains the eccentricity in a natural way and it is tempting to link wr 140 and @xmath0 car . about 10% of all the o - type stars in the solar neighborhood may be runaway stars with a peculiar space velocity @xmath9 30 - 40 km / s ( gies and bolton , 1986 ) . runaways can be formed by the binary - sn scenario ( blaauw , 1961 ) or by two - body dynamical interactions in star clusters where at least one body is a binary ( leonard and duncan , 1990 ) . it is clear that the number of o - type runaways predicted by both processes depends on the adopted binary frequency . we performed the following experiment : _ calculate the o - type runaway frequency as function of the primordial binary cluster frequency accounting for cluster stellar dynamics and for the sn explosion in binaries , using typical initial parameters of solar neighborhood type clusters ( like the orion nebula cluster ) _ preliminary results indicate that a 10% o - type runaway frequency can be obtained only by a model where the primordial ( interacting ) massive binary frequency in the solar neighborhood is larger than 60% . belczynski , k. , kalogera , v. , bulik , t. , 2002 , apj 572 , 407 belkus , h. , van bever , j. , vanbeveren , d. , 2007 , apj 659 , 1576 belkus , h. , van bever , j. , vanbeveren , d. , 2008 , in preparation belkus , h. , van bever , j. , vanbeveren , d. , van rensbergen , w. , 2003 , a&a 400 , 429 blaauw , a. , 1961 , bull . netherlands 15 , 265 cantiello , m. , yoon , s .- c . , langer , n. , livio , m. 2007 , a&a 465 , 29 comeron , f. , pasquali , a. , 2007 , a&a 467 , l23 de donder , e. , vanbeveren , d. , 2003a , newa 9 , 1 de donder , e. , vanbeveren , d. , 2003b , newa 8 , 817 de donder , e. , vanbeveren , d. , 2004 , newa rev 48 , 864 fregeau , j.m . , cheung , p. , portegies zwart , s.f . , rasio , f.a . , 2004 , mnras 352 , 1 fregeau , j.m . , rasio , f.a . , 2007 , apj 658 , 1047 gies , d.r . , bolton , c.t . , 1986 , apjs 61 , 419 hoogerwerf , r. , de bruijne , j.h.j . , de zeeuw , p.t . , 2001 , a&a 365 , 49 kudritzki , r.p . , 2002 , apj 577 , 389 leonard , p.j.t . , duncan , m.j . , 1990 , j. , 99 , 608 nugis , t. , lamers , h. j. g. l. m. , 2000 , a&a 360 , 227 pols , o.r . , cote , j. , waters , l.b.f.m . , heise , j. , 1991 , a&a 241 , 419 pols , o.r . , marinus , m. , 1994 , a&a 288 , 475 portegies zwart , s.f . , mcmillan , s.l.w . , makino , j. , 2007 , mnras 374 , 95 ptak , a , colbert , e. , 2004 , apj 606 , 291 sana , h. , gosset , e. , naze , y. , rauw , g. , linder , n. , 2007 , mnras ( in press ) , ( 2007arxiv0707.2847s ) soria , r. , 2007 , astrophys . space sci 311 , 213 suzuki , t.k . et al . , , 2007 , apj 668 , 435 tutukov , a. , yungelson , l.r . , iber , i. , 1982 , apj 386 , 197 van bever , j. , vanbeveren , d. , 1997 , a&a 332 , 116 van bever , j. , vanbeveren , d. , 2000 , a&a 358 , 462 van bever , j. , vanbeveren , d. , 2003 , a&a 400 , 63 vanbeveren , d. , 1991 , a&a 252 , 159 vanbeveren , d. , belkus , h. , van bever , j. , mennekens , n. , 2008 , astrophys . space sci . ( in press ) vanbeveren , d. , conti , p.s . , 1980 , a&a 88 , 230 vanbeveren , d. , de loore , c. , 1994 , a&a 290 , 129 vanbeveren , d. , de loore , c. , van rensbergen , w. 1998b , a&a review 9 , 63 vanbeveren , d. , et al . , . 1998a , newa 3 , 443 vanbeveren , d. , van bever , j. , belkus , h. , 2007 , apj 662 , 107 vanbeveren , d. , van rensbergen , w. , de loore , c. , 1998c , the brightest binaries , kluwer academic publishers . yungelson , l.r . , et al . , 2008 , a&a 477 , 223	we first summarize work that has been done on the effects of binaries on theoretical population synthesis of stars and stellar phenomena . next , we highlight the influence of stellar dynamics in young clusters by discussing a few candidate ufos ( unconventionally formed objects ) like intermediate mass black holes , @xmath0 car , @xmath1 pup , @xmath2@xmath3 velorum and wr 140 .
in @xcite a two - slit experiment supposedly showing the violation of the priciple of complementarity is realized . authors suggest that a which - way information is obtained from the position of incident photon / light while at the same time the existence of interference is testified . in several articles this experiment is being criticised . these critiques are based usually on explanations dwelling deep inside the basics of quantum mechanics . here we present numerical results showing that there is no need to go so deep to show that the principle of complementarity hold also in this case . in @xcite a simple analytic analysis is shown to support our arguments and numerical simulation . the claim of afshar , that the law of conservation of linear momentum compels us to accept that a photon in a particular spot on the photosensitive surface must have originated from the corresponding pinhole , is indeed correct . while having only one slit open the momentum of the photon forces it to end in a state from which an information about pinhole can be obtained . however having both slits open transverse momentum of photon ( as a quantum - mechanical object ) is zero and at the end must stay zero . the final state of the photon is thus symmetric and holds no evidence about the slits . in other words the two - peaked distribution is an interference pattern and the photon behaves as a wave and exhibits no particle properties until it hits the plate . as a result a which - way information can never be obtained in this way . tu support previous arguments we performed numerical simmulations following set - up used in @xcite . in the simulation process we utilized huygens - fresnel principle in the following form . the wave function on a chosen plane ( surface ) can be expressed in the form @xmath0 where @xmath1 represents the surface of the previous plane from which the wave originates , e.g. for the wave function on the first interference plane @xmath1 represents the surface of the two pinholes ; @xmath2 represents the distance from the point of origin and @xmath3 is the initial wave - function ( on the surface of the slits being constant ) . to succesfully simulate the propagation of the wave we alsohave to know , how to implement the lens light passing through a lens changes its phase depending on the distance @xmath4 from the center of the lens , since the thickness of the lens depends on this distance . this phase - shift ( up to the unsignificant constant ) can be expressed as : @xmath5 where @xmath6 is the focal length and @xmath7 is the circular wave - number of the wave , with @xmath8 being its wave - length . including this in the simulation is done by changing the exponent in the intetgral eq.([integral ] ) to @xmath9 and @xmath4 is the position on the originating plane lens . according to previous scheme we obtained following results . first we simulated the intensity of light on the first photosensitive surface ( which is in the experiment afterwards taken away ) . if the light is passing only throug one slit ( fig . [ fig01 ] ) , we see that the resulting intensity of light in quantum mechanical sense the probability density function of the incident photon has only one peak . opening both slits creates an interference pattern that is clearly observable and corresponds also quantitatively to afshar s results . by replacing the photosensitive surface by a convex lens ( in experiment the lens was positioned before this point , however this has no great implication for the results ) we can now obtain the distribution of the intensity / photons behind lens in the position where the geometrical optics predicts focused image of slits . the results are depicted on the fig . [ fig02 ] . when only one slit is open , it is clear that the incident photon originates from the open slit ( upper slit thus corresponds to the lower peak and lower slit corresponds to the upper slit ) . when both slits are open , afshar claims , that previous results still hold and he can thus obtain the which - way information . this is however false because having both slits open is a different experiment from the one with a single open slit . numerical results confirm this , since to obtain the double - peaked interference pattern in fig . [ fig02 ] corresponding to the focused image of both slits , we used only wave properties of the light . the resulting image is thus an interference pattern that does not contain any which - way information . to further support this claim we present results of simulation when the wires are put on the places where the interference pattern from the first simulation had its minima now it is clear that this can not change results obtained without wires ( see fig . [ fig03 ] ) for both slits open , since the minima in the interference pattern mean that the probability of finding photon in thet place is zero and thus no photon is intercepted . situation of course changes when only one slit stays open . since there is a non zero probability to find a photon in the positiion of the wires , thes intercept some of the light so the obtained image has lower peak intensity than that obtained without wires . this decrease in intensity is as large as 10% for collected data . in conclusion , we have numerically simulated experiment from @xcite . since in the simulation only wave properties of light were used and the results correspond to those obtained by afshar , we conclude that his claim , that he can obtain the which - way information from the position of the photon incident on the photosensitive surface placed behind the lens , is false . the which - way information in the case of both slits open is thus unobtainable and the same probably holds also for afshar s second experiment carried out in @xcite .	results of the experiments carried out in @xcite and @xcite are reviewed and their interpretation by the authors is questioned . arguments are supported by numerical simulations .
quantum teleportation @xcite allows one party to send an arbitrary quantum state to another using only a classical channel , local operations , and a preshared entanglement resource . such teleportation protocols can be classified as either discrete or continuous variable ( cv ) based on the nature of the state that is to be transferred . discrete teleportation deals with quantum states in a finite dimensional hilbert space @xcite , such as the polarization of a photon @xcite , or the energy level of an atom @xcite . continuous variable teleportation deals with states in an infinite dimensional hilbert space @xcite , such as the quadratures of a single mode of the em field , or the position of an atom in a harmonic potential . quantum teleportation has proven to be a useful module in the field of quantum computation for both algorithm development as well as assisting in carrying out actual computations @xcite . continuous variable teleportation has the advantage that it can be carried out using relatively simple states , transformations and detectors @xcite . the conventional method of cv teleportation relies on a two - mode squeezed vacuum state which serves as an entanglement resource between two parties . unfortunately , the teleportation fidelity is highly dependent on the amount of two - mode squeezing , and using state - of - the - art technology the highest fidelity measured to date is only @xmath0 @xcite . many important quantum protocols require about 10 db of two - mode squeezing @xcite , however achieving high fidelity is limited not only by technical details but it is not possible even in theory to attain 100% fidelity as this would require infinite squeezing , which is unphysical . in a recent paper andersen and ralph explore a new approach to the teleportation of continuous variable states @xcite . in this protocol , one replaces the standard two - mode squeezed state resource with a set of maximally entangled states on two dimensional hilbert spaces in order to attain teleportation fidelities close to 100% with modest resources . in this paper we explore an extension of these ideas to hilbert spaces of arbitrary dimension , as well as discuss the possibility of using alternative intermediary teleportation protocols . the advantage of this new scheme depends on the ability to teleport an optical state containing up to @xmath1 photons defining a qudit . there are a number of different propositions for such generalized quantum scissor protocols , some of which will be discussed in later sections . the advantage of moving to a higher dimension is that one will need to implement less teleportation modules , and if the modules are sufficiently simple this may result in a more robust protocol . this paper is organized as follows , first we state the formalism used in the conventional teleportation of a state on a finite dimensional hilbert space . then we consider a new scheme ( * ii * ) where a cv state is teleported using a set of maximally entangled qudits , this is an extension of the scheme ( * i * ) developed in ref . @xcite where the qudits are taken to be of dimension two . next we demonstrate how this new protocol implements entanglement swapping , as well as considering sources of noise and loss . finally , we provide a discussion detailing how this new scheme might be implemented , how it compares to other protocols , and what advantages it might offer . in the conventional cv teleportation protocol , there is a continuous degree of freedom in both the squeezing , and thus entanglement , of the shared resource as well as the nature of the local recovery operation that needs to be performed by the recipient . it should not be surprising that by threading our state through a set of finite dimensional hilbert spaces something must be lost . specifically , all of the higher dimensional terms will be `` cut - off '' when teleporting a state by using an entangled resource of a lower dimension @xcite . so in order to teleport a cv state in this manner with high fidelity , we require extra information about the state to be teleported ; we need to know the average photon number in order to ensure not too much information is lost upon truncation . let us review the traditional teleportation procedure on a finite dimensional hilbert space @xcite . this is of interest because it can be used as a building block in the process of teleporting a cv quantum state . suppose alice wishes to transfer the quantum state of an arbitrary qudit @xmath2 with @xmath3 . alice and bob share an entangled quantum state of the form @xmath4 where bob has system 3 while alice has systems 1 and 2 . define unitary operations consisting of @xmath5 where @xmath6 and the symbols @xmath7 refer to addition and subtraction modulo @xmath1 respectively . to teleport a state @xmath8 alice applies the xor operator on her systems as follows : @xmath9 where @xmath10 notice that the @xmath11 operators of the form @xmath12 are in one to one correspondence with the joint states of the form @xmath13 , and thus a projection onto these orthonormal states with outcome corresponding to @xmath14 would project system 3 onto the normalized state @xmath15 . alice is then able to send bob the classical information @xmath16 , from which he can apply a correction operation corresponding to @xmath17 . if the quantum state is ideal and maximally entangled to begin with , the @xmath18 states correspond to the quantum fourier transformed basis states , a well known and studied unitary operation @xcite . in order to use a set of maximally entangled states of the form @xmath19 to teleport a cv state , we must first divide the input state into a number of intermediate states ; for the purposes of this paper we will consider the input state to be a mode of the em field . perhaps the most intuitive approach is to evenly split the input mode into a set of @xmath20 modes by way of an array of @xmath21 asymmetric beamsplitters called an @xmath20-splitter @xcite . it will be convenient to first study how such a protocol will transform a coherent state , and then to use this information to deduce a generalization . a coherent state can be expressed in terms of the fock basis as @xmath22 where @xmath23 denotes the vacuum state . coherent states are often thought of as semi - classical states , and it should not be surprising that a coherent state will transform through an @xmath20-splitter as @xmath24 . in other words , the state is evenly split amongst the @xmath20 intermediate modes . although each of these modes @xmath25 is still a coherent state in an infinite dimensional hilbert space , the average number of photons @xmath26 is reduced by a factor @xmath20 . equivalently , the coefficients associated with high photon numbers in ( [ eq : coherent ] ) have smaller amplitudes , and thus less information will be lost if we were to truncate the state at some suitable dimension . consider a truncation of the state after it is passed through the @xmath20-splitter given by @xmath27^{\otimes n}\end{gathered}\ ] ] where we have dropped all terms corresponding to photon number larger than @xmath1 . this result corresponds to what we would be left with if we were to implement some form of teleportation for each mode on a hilbert space of dimension @xmath1 + 1 . the scheme developed by andersen and ralph ( * i * ) ends this truncation at @xmath28 , corresponding to keeping only the vacuum and one - photon state , the scheme we propose ( * ii * ) is a generalization to arbitrary dimensions . the final step of the teleportation protocol is to recombine the @xmath20 modes on an inverted @xmath20-splitter as shown in . modes on an @xmath20-splitter . ] if we condition on the outcome that all of the modes @xmath29 contain no photons , then each of the annihilation operators in ( [ eq : expansion ] ) simply pick up another factor of @xmath30 and become @xmath31 . thus we are left with the state @xmath32 where @xmath33 is some normalization constant . it will be useful to explicitly write @xmath34 in the fock basis as @xmath35 for some set of coefficients @xmath36 . the coefficient @xmath36 is given by : @xmath37 where @xmath38 results from a counting argument ; see . using this we can write the final output as @xmath39 where @xmath40 is the probability of detecting no photons in any of the @xmath41-modes . observe that an arbitrary fock state is transformed through the protocol as @xmath42 so we have that an arbitrary state @xmath43 transforms as @xmath44 where @xmath45 ensures normalization . in the case that we limit ourselves to a two - dimensional hilbert space , of only the vacuum and one photon states , we recover the result that @xmath46 . in other words , if one can only pick @xmath47 the problem of finding the number of ways to obtain @xmath48 from ( [ eq : phi ] ) reduces to the amount of ways that one can choose @xmath49 1 s from @xmath20 boxes . this result is then in agreement with the findings of @xcite . generally speaking , in order for the resulting fidelity to be high we require that the average photon number satisfy the constraint @xmath50 . if this is true , then the success probability @xmath45 will also be high , and the need to actually detect photons in the @xmath41-modes and condition on the empty outcome can be neglected . is fixed at a value of 20 . it is readily seen that the amplitudes associated with the teleported number states , given by ( [ eq : trans ] ) , remain higher for larger @xmath1 . ] a simple way of observing how this procedure affects a state is to see how the coefficients in ( [ eq : trans ] ) scale as a function of the number of modes @xmath20 and the dimension of the maximally entangled states @xmath1 . in an ideal procedure we would have that for any @xmath49 the coefficient would be @xmath51 , the actual results can be seen in . we now turn to the specific case of teleporting one mode of an epr state given by @xcite @xmath52 where @xmath53 , with @xmath54 being the squeezing parameter . the teleportation of one mode of an epr pair models the teleportation of a gaussian ensemble of different pure states , and thus this is an important case to study @xcite . by applying equations ( [ eq : tele ] ) and ( [ eq : psuc ] ) , we see that an epr state transforms as @xmath55 where @xmath45 is the probability that no photons are detected in the @xmath41-modes , and @xmath56 is the fidelity between the input and output states . . fidelity is plotted against the size of the @xmath20-splitter for a variety of different hilbert space dimensions @xmath1 associated with each mode . ] . success probability is plotted against the size of the @xmath20-splitter for a variety of different hilbert space dimensions @xmath1 associated with each mode . ] one measure of the amount of entanglement in an epr pair is related to the degree of two - mode squeezing given by @xmath57 , we can use this measure to observe how the fidelity and success probability vary as the number of entangled pairs is changed in and . by using larger hilbert spaces , of dimension @xmath58 , to teleport our cv state we are able to achieve comparable fidelities with fewer modes , as should be expected . the usefulness of this result hinges on the ability to create maximally entangled states of a dimension greater than two , and while this problem has been studied @xcite its practical implementation will likely require more experimental effort in this field . the original result of andersen and ralph , the two - dimensional case , has the advantage of being practically feasible with today s technology since maximally entangled states in two - dimensions can be generated with a single photon source and an ideal 50/50 beamsplitter . however , by moving to a higher dimension as proposed in ( * ii * ) one can obtain a large increase in fidelity for modest values of @xmath59 , even while using less modes @xmath60 in the teleportation . in general , it is found that when the input state has a low average energy this protocol is capable of outperforming the standard cv teleportation procedure . another interesting possibility is to use an alternative teleportation procedure for each mode after the first @xmath20-splitter . for example , port - based teleportation @xcite has found application to programmable quantum processors , non - local quantum computation , and position - based cryptography @xcite . as such , the ability to teleport a cv state without the need to apply a recovery operation directly on the system may prove to be a useful primitive in cv quantum computation . an important consideration is that of the loss and imperfection inevitably associated with any physical realization of this protocol . one cause for concern may arise from the fact that the quantum channel between alice and bob , to be used for the qudit teleportation , may not be maximally entangled . if we model the deviation from being a pure maximally entangled state by the action of a depolarizing channel we obtain the state @xmath61 , where @xmath62 is as given in ( [ eq : psi ] ) . in this case it has been shown @xcite that the channel fidelity is given by @xmath63 . this statement is equivalent to saying the fidelity is given by @xmath64 , where @xmath65 is the maximal singlet fraction attainable through trace preserving local operations and classical communication . one method to create such maximally entangled states involves distilling entanglement out of squeezed states @xcite . in general , the types of loss will depend on which scheme is utilized to generate a maximally entangled state . in this procedure , a detailed discussion of the types of noise is given by duan et al . it is interesting to note that some forms of noise depend in one way or another on the dimension @xmath1 of the maximally entangled state involved , while others are independent of this factor . the authors show that for the case of small preparation noise , the probability of distilling a maximally entangled state of dimension @xmath1 picks up an extra factor of @xmath66 where the @xmath67 s are cavity loss factors and @xmath68 is a transmission time . so preparation noise makes it less likely to obtain a maximally entangled state of higher dimension , but in the end one does obtain a pure state . there are also other proposed methods of truncating cv states through teleportation @xcite that sidestep the requirement of starting with a maximally entangled state . another source of loss comes from imperfect detection as well as inaccurate photon counting . photon counting in a mode @xmath69 by an imperfect detector can be modelled as a positive - operator - valued - measure ( povm ) described by @xcite @xmath70 where the elements sum to the identity . the number of photons registered by the detector is given by @xmath20 whereas @xmath71 corresponds to the number of photons that actually enters the detector . the detection efficiency is given by @xmath67 and the mean dark count rate is described by @xmath72 . conventional avalanche photo - diodes can be described by a povm @xmath73 where the detector is only capable of distinguishing between no photons , and at least one photon . single photon resolving detectors are similarly described by a povm with elements corresponding to @xmath74 , and this generalizes in a straightforward manner to detectors capable of distinguishing up to an arbitrary number of photons . generally speaking the protocol will perform better when less terms are lost in the truncation , however this will typically come at a cost . either generation of maximally entangled states of higher dimension will succeed with lower probability or will result in a lower fidelity . to consider the possible advantages and disadvantages of this new scheme , it will be useful to compare it with other possibilities of teleporting a cv quantum state . the traditional approach to cv teleportation is to use a squeezed vacuum state of the form given by ( [ eq : epr ] ) . however , the resulting fidelity generally scales exponentially with the amount of squeezing . for example , the fidelity for teleportation of a coherent state is given by @xmath75 where @xmath76 is the amount of squeezing . in this case , the fidelity approaches unity only in the limit of an unphysical amount of squeezing , and thus this scheme does not scale well . in an ideal case , both the scheme proposed by andersen and ralph ( * i * ) as well as the scheme in this paper ( * ii * ) will approach perfect fidelity with finite resources . however , the possible advantages or pitfalls will lie in the physical implementations , and by taking account of imperfections . perhaps the most obvious source of loss is that of finite detector efficiency , so we will consider how the schemes perform while taking this into account . unfortunately , it becomes difficult to compare the two schemes on a level playing field since ( * i * ) deals with qubits which can be implemented with linear optics and ( * ii * ) requires nonlinearities . specifically , one must be able to implement the xor gate ( [ eq : xor ] ) , and one must also be able to make measurements that distinguish between different photon numbers . one possible implementation of the xor gate is based on the kerr effect @xcite , and phonton number resolving ( pnr ) detectors fulfil the second requirement . however , the relatively low nonlinear susceptibility constants @xmath77 available in state - of - the - art materials leads to impractically long interaction times , and conventional qudit teleportation of this sort is not currently possible in the lab . if we suppose that such a xor gate could be implemented in an ideal fashion then we could compare ( * i * ) and ( * ii * ) by simply comparing the detection efficiencies . in the case of ( * i * ) the overall success rate would scale as @xmath78 , where @xmath20 is the number of modes , if one were restricted to linear optics , since only two of the four bell states can be distinguished . to make a fair comparison to the qudit case , we assume that all of the generalized bell states can be deterministically detected , in this case the detection efficiency of ( * i * ) is given by @xmath79 , where one factor of @xmath67 comes from the teleportation while the other comes from conditioning on no photons in the auxiliary modes . similarly , the detection efficiency of ( * ii * ) would be given by @xmath80 , where @xmath81 is the effective detection efficiency of a pnr detector when distinguishing between vacuum through @xmath82 photons . unfortunately , for any of the above to be a meaningful comparison we would have to have values for @xmath67 and @xmath83 , as well as being able to implement a xor gate . fortunately , this is not the only proposed method for teleporting a system on a finite dimensional hilbert space . for example , we can consider the multimode interferometer approach found in @xcite . to take a specific case , consider teleporting a four dimensional qudit , in this case one can show that the probability of success can be made at least as high as @xmath84 . if one desires a fidelity of approximately @xmath85 using scheme ( * ii * ) one could use either @xmath86 qubits or @xmath87 quartits , as can be seen in . the success probability using the qubits scales as @xmath88 whereas it scales as @xmath89 for the quartits ; the factor of @xmath84 coming from the success probability of the interferometer post - selection . one can distinguish between zero , one , and more than one photons using a single photon resolving detector , such as a visual light photon counter ( vlpc ) @xcite , and this type of detector is sufficient to teleport a quartit state using the interferometer approach . it is also possible to use this approach with a conventional photodetector , however one suffers a loss of fidelity . an important remark is that for a reasonable range of values for @xmath67 and @xmath90 the proposed scheme ( * ii * ) offers an advantage over the two - level scheme ( * i * ) . a general advantage of the new work detailed in scheme ( * ii * ) is that one does not need to work with as many modes to achieve a larger , yet there are still unique difficulties in extending the dimension to include higher photon numbers . another potential method to teleport each mode after the @xmath20-splitter is to use a continuous variable epr state to teleport a discrete variable state using gain tuning @xcite . this approach has the advantage of being deterministic , so although the fidelity will be degraded due to finite squeezing , the efficiency of the protocol may still be higher than the interferometer approach discussed above . this approach has already been used to demonstrate teleportation of a dual - rail qubit with fidelity on the order of @xmath91 @xcite . to conclude , we have demonstrated an extension to the cv teleportation scheme proposed by andersen and ralph @xcite . this protocol splits an input cv state equally among @xmath20 modes , and then teleports each mode using a finite dimensional hilbert space with @xmath92 . the reduction of cv states to a @xmath1-dimensional space results in a truncation of the fock basis , and corresponds to a loss in fidelity . however , by choosing an appropriate number of modes @xmath20 and dimension @xmath1 , one is able to achieve arbitrary fidelity with high probability . the ability to create and distribute maximally entangled qudits of dimension @xmath1 is a technique required in order for this protocol to be practically feasible . an interesting possibility opened up by this approach is to use generalized teleportation protocols for each mode after the input is split . one is then able to harness the results for systems of qudits and attempt to apply them to the field of continuous variable quantum information . to obtain the coefficient of a number state @xmath48 in the expansion of the state @xmath93 consider all of the ways one can get a term @xmath94 to the power @xmath49 , for a fixed but arbitrary @xmath49 . in expanding equation ( [ eq : a1 ] ) , we take one term from each of the @xmath20 summations that are being multiplied . for example , one term in the expansion corresponds to the set of values @xmath95 where @xmath96 for all @xmath97 . @xmath98 to find the coefficient @xmath36 of the number state @xmath48 , where @xmath35 , we consider the sum of all such terms with the constraint that @xmath99 . recalling that @xmath100 we have that @xmath101 which can be calculated numerically .	the need for high fidelity quantum teleportation arises in a variety of quantum algorithms and protocols . unfortunately , conventional continuous variable teleportation schemes rely on epr states that yield a fidelity that approaches unity only in the limit of an unphysical amount of squeezing . a new method , which utilizes an ensemble of single photon entangled states to teleport continuous variable states with fidelity approaching unity with finite resources was recently proposed by andersen and ralph [ phys . rev . lett . 111 , 050504 ( 2013 ) ] . we extend these ideas to consider the general case of using maximally entangled states of an arbitrary dimension to teleport a continuous variable state and discuss how the corresponding results are affected .
consider a convex polygon @xmath3 in @xmath4 whose vertices lie in the integer lattice @xmath5 . it determines a vector space @xmath6 ( over a filed @xmath7 ) of polynomials @xmath8 whose monomials correspond to the lattice points in @xmath3 : @xmath9 consider a finite field @xmath10 . the _ toric surface code _ @xmath0 , first introduced by hansen in @xcite , is defined by evaluating the polynomials in @xmath11 at all the points @xmath12 in the algebraic torus @xmath13 . to be more precise , @xmath0 is a linear code whose codewords are the strings @xmath14 for @xmath15 . it is convenient to assume that @xmath3 is contained in the square @xmath16 ^ 2 $ ] so that all the monomials in @xmath11 are linearly independent over @xmath10 . thus @xmath0 has block length @xmath17 and dimension equal to the number of the lattice points in @xmath3 . note that the weight of each non - zero codeword in @xmath0 is the number of points @xmath18 where the corresponding polynomial does not vanish . therefore , the minimum distance of @xmath0 ( which is the minimum weight for linear codes ) equals @xmath19 where @xmath20 is the number of zeroes ( i.e. points of vanishing ) in @xmath13 of @xmath21 . the name _ toric surface code _ comes from the fact that @xmath3 defines a toric surface @xmath22 over @xmath23 ( strictly speaking the fan that defines @xmath22 is a refinement of the normal fan of @xmath3 ) , where @xmath24 can be identified with the space of global sections of a semiample divisor on @xmath22 ( see for example @xcite ) . this allows to exploit algebraic geometric techniques to produce results about the minimum distance of @xmath0 . in particular , little and schenck in @xcite used intersection theory on toric surfaces to come up with the following general idea : if @xmath25 is sufficiently large then polynomials @xmath15 with more absolutely irreducible factors will necessarily have more zeroes in @xmath13 ( @xcite , proposition 5.2 ) . in this paper we expand this idea to produce explicit bounds for the minimum distance of @xmath0 in terms of certain geometric invariant @xmath2 , which we call the full minkowski length of @xmath3 . essentially @xmath2 tells you the largest possible number of absolutely irreducible factors a polynomial @xmath15 can have , but it derives it from the geometry of the polygon @xmath3 ( see definition [ d : maximal ] ) . the number @xmath2 is easily computable we give a simple algorithm which is polynomial in the number of lattice points in @xmath3 . moreover we obtain a description of the factorization @xmath26 for @xmath15 with the largest number of factors . more precisely , in proposition [ p : hwl ] we show that the newton polygon @xmath27 ( which is the convex hull of the exponents of the monomials in @xmath28 ) is either a primitive segment , a unit simplex , or a triangle with exactly 1 interior and 3 boundary lattice points , called an _ exceptional triangle_. this description enables us to prove the following bound : let @xmath29 be a lattice polygon with area @xmath30 and full minkowski length @xmath31 . then for @xmath32 , where @xmath33 , the minimum distance of the toric surface code @xmath0 satisfies @xmath34 the condition that no factorization @xmath26 contains an exceptional triangle ( as the newton polygon of one of the factors ) is geometric and can be easily checked for any given @xmath3 ( we provide a simple algorithm for this which is polynomial in the number of lattice points in @xmath3 ) . in this case we have a better bound for the minimum distance of the toric surface code : let @xmath29 be a lattice polygon with area @xmath30 and full minkowski length @xmath31 . under the above condition on @xmath3 , for @xmath35 , where @xmath36 , the minimum distance of the toric surface code @xmath0 satisfies @xmath37 we remark that our thresholds for @xmath25 where the bounds begin to hold are much smaller than the ones in little and schenck s result ( @xcite , proposition 5.2 ) . although , as mentioned above , the minimum distance problem for toric codes is tightly connected to toric varieties , all our methods are geometric and combinatorial and do not use algebraic geometry ( except for the hasse weil bound , see section [ s : prelim ] ) . in section [ s : length ] we define the full minkowski length @xmath2 and establish combinatorial properties of polygons with @xmath38 . in section [ s : main ] we give a proof of theorem 1 and theorem 2 . section [ s : alg ] is devoted to the above mentioned algorithms for computing @xmath2 and determining the presence of an exceptional triangle . finally , in section [ s : ex ] we give a detailed analysis of three toric surface codes which illustrates our methods . we thank leah gold and felipe martins for helpful discussions on coding theory . let @xmath3 and @xmath39 be convex polytopes in @xmath40 . their _ minkowski sum _ is @xmath41 which is again a convex polytope . figure [ f : sum ] shows the minkowski sum of a triangle and a square . let @xmath21 be a laurent polynomial in @xmath42 $ ] ( for some field @xmath7 ) . then its _ newton polytope _ @xmath43 is the convex hull of the exponent vectors of the monomials appearing in @xmath21 . thus @xmath43 is a _ lattice polytope _ as its vertices belong to the integer lattice @xmath44 . note that if @xmath45 $ ] then the newton polytope of their product @xmath46 is the minkowski sum @xmath47 . a _ primitive lattice segment _ @xmath48 is a line segment whose only lattice points are its endpoints . the difference of the endpoints is a vector @xmath49 whose coordinates are relatively prime ( @xmath49 is defined up to sign ) . a polytope which is the minkowski sum of primitive lattice segments is called a _ ( lattice ) zonotope_. the automorphism group of the lattice is the group of affine unimodular transformations , denoted by @xmath50 , which consists of translations by an integer vector and linear transformations in @xmath51 . affine unimodular transformations correspond to monomial changes of variables in @xmath42 $ ] and preserve the zero set of @xmath21 in the algebraic torus @xmath52 . let @xmath3 be a lattice polytope in @xmath40 . consider a minkowski decomposition @xmath53 into lattice polytopes @xmath54 of positive dimension . clearly , there are only finitely many such decompositions . we let @xmath55 be the largest number of summands in such decompositions of @xmath3 , and call it the _ minkowski length _ of @xmath3 . [ d : maximal ] the _ full minkowski length _ of @xmath3 is the maximum of the minkowski lengths of all subpolytopes @xmath39 in @xmath3 , @xmath56 a subpolytope @xmath57 is called _ maximal _ for @xmath3 if @xmath58 . a minkowski decomposition of @xmath39 into @xmath2 summands of positive dimension will be referred to as a _ maximal ( minkowski ) decomposition in p_. here are a few simple properties of @xmath2 and maximal subpolytopes . [ p : simple ] let @xmath3 , @xmath59 , @xmath60 , and @xmath39 be lattice polytopes in @xmath40 . 1 . @xmath2 is @xmath50-invariant . 2 . @xmath61 if and only if @xmath62 . 3 . if @xmath63 then @xmath64 . 4 . if @xmath39 is maximal for @xmath3 then @xmath39 contains a zonotope @xmath65 maximal for @xmath3 . the first three statements are trivial . for the forth one , note that if @xmath66 is a maximal minkowski decomposition in @xmath3 then by replacing each @xmath67 with one of its edges we obtain a zonotope @xmath68 with @xmath69 . but @xmath70 , so @xmath71 . notice that the summands of every maximal decomposition in @xmath3 are polytopes of full minkowski length 1 . it seems to be a hard problem to describe polytopes of full minkowski length 1 in general . however , in dimensions 1 and 2 we do have a simple description for such polytopes ( theorem [ t : indecomp ] ) . a lattice polytope @xmath3 is _ strongly indecomposable _ if its full minkowski length @xmath2 is 1 . in other words , no subpolytope @xmath57 is a minkowski sum of lattice polytopes of positive dimensions . clearly , primitive segments are strongly indecomposable and they are the only 1-dimensional strongly indecomposable polytopes . let @xmath72 be the standard 2-simplex and @xmath73 be the triangle with vertices @xmath74 , @xmath75 and @xmath76 ( see figure [ f : indecomp ] ) . it is easy to see that the they both are strongly indecomposable . next theorem shows that these are essentially the only strongly indecomposable polygons . in the proof of this theorem and frequently later in the paper we will use pick s formula : let @xmath3 be a lattice polygon in @xmath4 . then the area of @xmath3 equals @xmath77 where @xmath78 is the number of interior lattice points in @xmath3 and @xmath79 is the number of boundary points in @xmath3 . the proof of this formula can be found for example in @xcite . [ t : indecomp ] let @xmath3 be a strongly indecomposable polygon . then @xmath3 is @xmath80-equivalent to either the standard 2-simplex @xmath72 or the triangle @xmath73 above . first , note that @xmath3 can not contain more than 4 lattice points . indeed , suppose @xmath81 and @xmath82 lie in @xmath83 . if @xmath84 , for @xmath85 , then the segment @xmath86 $ ] lies in @xmath3 and is not primitive , hence , @xmath87 . since there are only 4 possible pairs of remainders mod @xmath88 , @xmath3 has at most 4 lattice points . suppose @xmath3 is a triangle , then its sides must be primitive and either @xmath3 has no interior lattice points or it has exactly one interior lattice point . in the first case @xmath3 has area @xmath89 ( by pick s formula ) and so is @xmath80-equivalent to @xmath72 . in the second case @xmath3 has area @xmath90 ( by pick s formula ) and hence any two of its sides generate a parallelogram of area 3 . every such triangle is @xmath80-equivalent to @xmath73 . now suppose @xmath3 is a quadrilateral . then it has no interior lattice points and so its area is 1 ( by pick s formula ) . every such quadrilateral is @xmath80-equivalent to the unit square . but the unit square is obviously decomposable . a lattice polygon is called a _ unit triangle _ if it is @xmath80-equivalent to @xmath72 , and an _ exceptional triangle _ if it is @xmath80-equivalent to @xmath73 . the following theorem describes maximal minkowski decompositions for a given lattice polygon @xmath3 . [ t : decomp ] let @xmath3 be a lattice polygon in @xmath4 with full minkowski length @xmath2 . consider a maximal minkowski decomposition in @xmath3 : @xmath91 for some @xmath57 . then one of the following holds : 1 . every @xmath67 is either a primitive segment or a unit triangle ; 2 . after an @xmath80-transformation and reordering of the summands the decomposition is @xmath92+m_2[0,e_2]+m_3[0,e_1+e_2],\ ] ] where @xmath93 are non - negative integers such that @xmath94 and the @xmath95 are the standard basis vectors . since every @xmath67 must be strongly indecomposable , by theorem [ t : indecomp ] it is a primitive segment , a unit triangle , or an exceptional triangle . we claim that if one of the @xmath67 is an exceptional triangle then the other summands are primitive segments in only three possible directions . this follows from the two lemmas below . [ l : ml3 ] consider two primitive segments @xmath96 in @xmath5 and let @xmath97 , @xmath98 be the corresponding vectors . if @xmath99 then @xmath100 . we can assume that @xmath101 and @xmath102 with @xmath103 and @xmath104 . cases when @xmath105 are easily checked by hand . for @xmath106 we can use the same argument as in the proof of theorem [ t : indecomp ] to show that @xmath107 contains a segment of lattice length 3 . indeed , the area of @xmath108 equals @xmath106 . by pick s formula @xmath108 has at least 10 lattice points . but then there exist @xmath81 and @xmath82 in @xmath108 such that @xmath109 mod 3 , for @xmath85 . therefore the segment @xmath86 $ ] is contained in @xmath108 and has lattice length 3 . [ l : triangles ] let @xmath1 be strongly indecomposable . then @xmath110 unless @xmath3 is a primitive segment in the direction of @xmath111 , @xmath112 or @xmath113 . let @xmath114 be an edge of @xmath73 and @xmath115 an edge of @xmath3 and let @xmath116 be the corresponding vectors . if @xmath99 then by lemma [ l : ml3 ] @xmath100 and since @xmath117 we also have @xmath110 . so we suppose that @xmath118 for all edges @xmath114 in @xmath73 . then we have the following linear inequalities for @xmath119 : @xmath120 clearly , the only integer solutions ( up to central symmetry ) are @xmath101 , @xmath75 , and @xmath121 . now if @xmath3 contains at least 2 edges in these directions then it must also contain ( up to a translation ) either @xmath122 or @xmath123 . but in both cases the sum @xmath124 contains a @xmath125 rectangle which has minkowski length three . therefore , @xmath110 . [ r : invtzero ] notice that in lemma [ l : triangles ] the special directions @xmath111 , @xmath112 or @xmath113 have an easy @xmath80-invariant description : they are obtained by connecting the interior lattice point in @xmath73 to the vertices . while classifying polygons of every given full minkowski length does not seem feasible , we will make a few statements about polygons of full minkowski length 2 , which we will use later . [ p : length2 ] suppose @xmath126 . then 1 . @xmath3 has at most 3 interior lattice points , i.e. @xmath127 ; 2 . if @xmath128 then @xmath3 is @xmath80-equivalent to one of the polygons depicted in figure [ f : length2 ] ; 3 . if @xmath128 then @xmath129 . \(1 ) the proof is somewhat technical so we will sketch its major steps . assume @xmath3 has 4 or more interior lattice points . first , it is not hard to show that one can choose 4 interior lattice points in @xmath3 so that after an @xmath80-transformation they form either a unit square : @xmath130 or a base 2 isosceles triangle : @xmath131 . in the first case , note that @xmath3 must include a lattice point which is distance one from the square and lies on one of the lines containing the sides of the square . by symmetry we can assume it is @xmath132 . in figure [ f : square ] on the left , the solid dots represent the 5 points that now belong to @xmath3 , the crosses represent the points that can not belong to @xmath3 ( otherwise its length would be greater than 2 ) . now if point @xmath133 does not belong to @xmath3 ( the middle picture in figure [ f : square ] ) then either @xmath134 or @xmath135 does . but in either case the four points of the unit square can not all lie in the interior of @xmath3 . if point @xmath133 does belong to @xmath3 then it produces more forbidden points ( the rightmost picture in figure [ f : square ] ) . then again , it is not hard to see that no such @xmath3 can exist . playing the same game one can show that no @xmath3 exists in the second case as well . \(2 ) first one can show that the three interior lattice points can not be collinear . thus we can assume that they are @xmath136 . our first case is when @xmath121 also lies in @xmath3 . since this must be a boundary point and there are no more interior points in @xmath3 we see that @xmath134 and @xmath133 are the only possible boundary points of @xmath3 on the line @xmath137 . similarly , @xmath132 and @xmath138 are the only possible boundary points of @xmath3 on the line @xmath139 . since both @xmath134 and @xmath138 can not belong two @xmath3 , using symmetry we arrive at two possibilities for the boundary piece of @xmath3 containing @xmath121 , depicted in figure [ f:3points ] on the left . as in part ( 1 ) we crossed out the points which can not appear in @xmath3 since @xmath126 . then it becomes clear that the only @xmath3 ( up to symmetry ) containing @xmath136 and @xmath121 are @xmath59 and @xmath60 in figure [ f : length2 ] . in the second case , when @xmath121 does not lie in @xmath3 we can assume that @xmath140 and @xmath141 do not lie in @xmath3 as well , otherwise we can reduce it to the previous case by a unimodular transformation . also , both @xmath138 and @xmath134 can not lie in @xmath3 , so by symmetry we can assume that @xmath138 does not . as before crossing out forbidden points we obtain the rightmost picture in figure [ f:3points ] . now it is easy to see that the only @xmath3 containing the 3 points in the interior is @xmath142 in figure [ f : length2 ] . \(3 ) by ( 2 ) it is enough to check that @xmath143 for every @xmath144 and any exceptional triangle @xmath145 . we first look at @xmath59 . by lemma [ l : triangles ] and remark [ r : invtzero ] we have @xmath146 for any primitive segment @xmath48 except for the three special segments @xmath114 , @xmath115 , @xmath147 that connect the interior lattice point of @xmath145 to its vertices . if @xmath148 then one of @xmath149 $ ] , @xmath150 $ ] , @xmath151 $ ] is not among the @xmath152 . but @xmath59 contains the segments @xmath153 , 2[0,e_2]$ ] , and @xmath154 $ ] . if , say , @xmath149 $ ] is not among the @xmath152 then @xmath155+t)\geq 4 $ ] and hence @xmath156 . it remains to show that @xmath157 which can easily be checked by hand . a similar argument works for @xmath142 . we only need to replace @xmath73 with @xmath158 , the triangle with vertices @xmath159 and @xmath134 . its special segments @xmath149 $ ] , @xmath150 $ ] , @xmath160 $ ] are contained in @xmath3 with multiplicity 2 . finally , since @xmath161 we do not need to do any extra work for @xmath60 . fix a finite filed @xmath10 where @xmath25 is prime power . for any convex lattice polygon @xmath3 in @xmath4 we associate a @xmath10-vector space of bivariate polynomials whose monomials have exponent vectors in @xmath83 : @xmath162 if @xmath3 is contained in the square @xmath163 ^ 2 $ ] then the monomials @xmath164 are linearly independent over @xmath10 and so @xmath165 . in what follows we will always assume that @xmath166 . the _ toric surface code _ @xmath0 is a linear code whose codewords are the strings of values of @xmath167 at all points of the algebraic torus @xmath13 : @xmath168 this is a linear code of block length @xmath17 and dimension @xmath169 . the weight of each non - trivial codeword equals the number of points @xmath170 where the corresponding polynomial does not vanish . let @xmath20 denote the number of points in @xmath13 where @xmath21 vanishes . then the minimum distance @xmath171 , which is also the minimum weight , equals @xmath172 consider @xmath167 . its _ newton polygon _ @xmath43 is the convex hull of the lattice points in @xmath4 corresponding to the monomials in @xmath21 , so @xmath173 let @xmath22 be a toric variety over @xmath23 defined by a fan @xmath174 which is a refinement of the normal fan of @xmath43 . then @xmath21 can be identified with a global section of a semiample divisor on @xmath22 . if @xmath21 is absolutely irreducible then it defines an irreducible curve @xmath175 on @xmath22 whose number of @xmath10-rational points @xmath176 satisfies the hasse weil bound ( see for example @xcite ) : @xmath177 where @xmath178 is the arithmetic genus of @xmath175 . it is a standard fact from the theory of toric varieties that the genus @xmath178 equals the number @xmath179 of interior lattice points in @xmath43 ( see @xcite ) . let @xmath180 be the invariant divisor at `` infinity '' , i.e. @xmath181 . some of the @xmath182-rational points of @xmath175 may lie on @xmath183 , we will denote their number by @xmath184 . then we have the following bound for the number of @xmath10-rational points of @xmath175 in the torus @xmath13 , i.e the number of zeroes of @xmath21 in @xmath13 : @xmath185 the divisor @xmath183 is the disjoint union of zero- and one - dimensional orbits in @xmath22 . the one - dimensional orbits @xmath186 are isomorphic to @xmath187 and correspond to the rays of @xmath174 . since @xmath174 is a refinement of the normal fan of @xmath43 , some of the orbits correspond to the edges of @xmath43 . let @xmath48 be an edge of @xmath43 and @xmath188 the corresponding orbit in @xmath22 , and consider the `` restriction '' of @xmath21 to @xmath48 , i.e. a univariate polynomial @xmath189 whose coefficients are @xmath190 for @xmath191 , ordered counterclockwise . then the intersection number @xmath192 equals the number of zeroes of @xmath193 in @xmath187 ( see for example @xcite ) . in particular if @xmath48 is primitive then @xmath193 is a binomial which has one @xmath182-rational zero on @xmath188 . therefore , @xmath184 is greater than or equal to the number of primitive edges of @xmath43 . we obtain the following proposition . [ p : hw ] let @xmath167 be absolutely irreducible and @xmath43 its newton polygon . 1 . if @xmath43 is an exceptional triangle then @xmath194 . 2 . if @xmath195 then @xmath196 unless @xmath43 is twice a unit triangle in which case @xmath197 . \(1 ) follows immediately from ( [ e : hw0 ] ) and the above discussion . for ( 2 ) we use the classification of polygons with no interior lattice points ( see for example @xcite ) : @xmath43 is @xmath80-equivalent to either ( a ) @xmath198 or ( b ) a trapezoid ( see figure [ f : empty ] ) where @xmath199 ( this includes primitive segments when @xmath200 and unit triangles when @xmath201 , @xmath202 ) . in the first case @xmath197 by ( [ e : hw0 ] ) . in the second case @xmath43 has at least 2 primitive edges , so @xmath203 , again by ( [ e : hw0 ] ) . let @xmath0 be the toric surface code defined by a lattice polygon @xmath3 in @xmath204 . in this section we prove bounds for the minimum distance of @xmath0 in terms of the full minkowski length @xmath2 of the polygon @xmath3 . here is our first application of the results of the previous section . [ p : hwl ] let @xmath167 be a polynomial with the largest number of absolutely irreducible factors , @xmath205 . then 1 . @xmath206 and every @xmath27 is either a primitive segment , a unit triangle , or an exceptional triangle ; 2 . the number of zeroes of @xmath21 in @xmath13 satisfies @xmath207 3 . if @xmath27 is not an exceptional triangle for any @xmath208 then @xmath209 \(1 ) follows directly from theorem [ t : decomp ] . moreover , the theorem implies that either ( a ) all @xmath54 are primitive segments or unit triangles or ( b ) one of the @xmath54 is an exceptional triangle and the others are primitive segments . in the first case every @xmath28 has at most @xmath210 zeroes in @xmath13 by proposition [ p : hw ] . not accounting for possible common zeroes of the @xmath28 we obtain the bound in ( 3 ) . in the second case one of the @xmath28 has at most @xmath211 zeroes and the others have at most @xmath210 zeroes , again by proposition [ p : hw ] . as before , disregarding possible common zeroes of the @xmath28 we get the bound in ( 2 ) . the next proposition deals with polynomials @xmath21 whose number of absolutely irreducible factors is @xmath212 . [ p : hwl-1 ] let @xmath3 have full minkowski length @xmath31 and let @xmath167 have @xmath213 absolutely irreducible factors . then @xmath214 as before let @xmath215 be the decomposition of @xmath21 into absolutely irreducible factors and let @xmath54 be the newton polygon of @xmath28 . first , by proposition [ p : simple ] @xmath216 hence , up to renumbering , @xmath217 and @xmath218 for @xmath219 . assume @xmath220 . then every @xmath54 is either a strongly indecomposable triangle or a lattice segment . we claim that at most 3 of the @xmath54 are exceptional triangles , and so the statement follows from proposition [ p : hw ] . indeed , if , say , @xmath221 are exceptional triangles then by lemma [ l : triangles ] @xmath222 . applying proposition [ p : simple ] again we get @xmath223 a contradiction . now assume @xmath224 . according to proposition [ p : length2 ] , ( 1 ) we have @xmath225 . also since @xmath224 , at most one of the other @xmath54 is an exceptional triangle . this follows from lemma [ l : triangles ] using similar to the previous case arguments . we now have three subcases . if @xmath226 then we have @xmath227 if @xmath228 then @xmath59 has at least one primitive edge which we prove in the lemma below . therefore by ( [ e : hw0 ] ) and proposition [ p : hw ] we have @xmath229 finally , if @xmath230 then none of the other @xmath54 is an exceptional triangle . this follows from proposition [ p : length2 ] , ( 3 ) and the above arguments . in this case @xmath59 has at least 2 primitive edges by proposition [ p : length2 ] , ( 2 ) . therefore using ( [ e : hw0 ] ) we have @xmath231 if @xmath126 and @xmath232 then @xmath3 has a primitive edge . since @xmath126 no edge can have more than 3 lattice points . if @xmath3 has 4 or more edges none of which is primitive then @xmath3 has at least 8 boundary lattice points and , hence , at least 10 lattice points total . but then @xmath3 contains a lattice segment of lattice length 3 ( see the proof of lemma [ l : ml3 ] ) , which contradicts the assumption @xmath126 . it remains to show that triangles with no primitive edges , 2 interior lattice points , and 6 boundary lattice points do not exist . let @xmath145 be such a triangle and let @xmath233 , @xmath234 be two of its edges , where @xmath114 and @xmath115 are primitive . then @xmath114 , @xmath115 form a triangle @xmath235 of area @xmath236 . on the other hand , by pick s formula @xmath237 , and hence @xmath238 . this implies that up to an @xmath80-transformation @xmath239 $ ] and @xmath240 $ ] , but then @xmath241 , a contradiction . now we are ready for the main result of this section . [ t:2-bound ] let @xmath242 be a lattice polygon with area @xmath243 and full minkowski length @xmath206 . then 1 . for @xmath32 , where @xmath33 , every polynomial @xmath167 has at most @xmath244 zeroes in @xmath13 . consequently , the minimum distance for the toric surface code @xmath0 satisfies @xmath34 2 . if no maximal decomposition in @xmath3 contains an exceptional triangle then for @xmath35 , where @xmath36 , every polynomial @xmath167 has at most @xmath245 zeroes in @xmath13 . consequently , the minimum distance for the toric surface code @xmath0 satisfies @xmath37 \(1 ) as we have seen in proposition [ p : hwl ] ( 2 ) the bound holds for the polynomials with the largest number of irreducible factors . we are going to show that for large enough @xmath25 every polynomial with fewer irreducible factors will have no greater than @xmath246 zeroes in @xmath13 . let @xmath167 have @xmath247 absolutely irreducible factors @xmath248 and let @xmath54 be the newton polygon of @xmath28 . if @xmath249 we can use the bound in proposition [ p : hwl-1 ] : @xmath250 the latter is at most @xmath246 for all @xmath251 . now suppose @xmath252 . first assume @xmath253 for all @xmath254 . then by proposition [ p : hw ] ( 2 ) , @xmath255 where @xmath256 is the number of twice unit triangles among the @xmath54 . since the sum of the full minkowski lengths of the @xmath54 can not exceed @xmath31 we have @xmath257 , i.e. @xmath258 . using this inequality along with @xmath259 we obtain @xmath260 the latter is at most @xmath245 for all @xmath261 and the bounds follow . suppose @xmath262 for at least one of the @xmath54 . then , as we will show in lemma [ l : others ] , @xmath263 now the right hand side will be at most @xmath244 whenever @xmath25 satisfies @xmath264 before proceeding we introduce the following notation : @xmath265 , @xmath266 . then ( [ e : quadratic ] ) becomes @xmath267 since this is a quadratic inequality in @xmath268 , it will hold if @xmath269 since @xmath270 it is enough to choose @xmath271 . finally , if @xmath272 then @xmath273 , since @xmath270 , and it is enough to choose @xmath274 if @xmath275 then @xmath276 and it is enough to choose @xmath277 . \(2 ) the proof of the second statement is completely analogous . first , if @xmath21 has @xmath31 irreducible factors the bound holds by proposition [ p : hwl ] , ( 3 ) . second , if @xmath21 has fewer than @xmath31 factors we choose @xmath25 large enough so that the right hand sides of ( [ e : ourbound1 ] ) and ( [ e : ourbound2 ] ) are no greater than @xmath245 . the same arguments as before show that it is enough to choose @xmath278 it remains to prove the following lemma . [ l : others ] let @xmath248 , for @xmath279 , and @xmath262 for at least one @xmath280 . then @xmath281 we order the @xmath54 so that for @xmath282 every @xmath54 either has interior lattice points or is twice a unit triangle . then according to ( [ e : hw0 ] ) and proposition [ p : hw ] we have @xmath283 now we want to get a bound for @xmath284 . recall that given two polytopes @xmath285 and @xmath286 in @xmath4 , their normalized _ mixed volume _ ( 2-dimensional ) is @xmath287 the mixed volume is symmetric ; bilinear with respect to minkowski addition ; monotone increasing ( i.e. if @xmath288 then @xmath289 ) ; and @xmath80-invariant ( see , for example @xcite , page 138 ) . this implies that @xmath290 indeed , by monotonicity it is enough to show that @xmath291 for any lattice segment @xmath48 , and by @xmath80-invariance we can assume that @xmath48 is horizontal . it follows readily from the definition that @xmath292 , where @xmath293 is the length of the horizontal projection of @xmath54 ( the height of @xmath54 ) and @xmath294 is the length of @xmath48 . clearly , @xmath295 and @xmath296 if @xmath54 has at least one interior lattice point or is twice a unit triangle . using ( [ e : mixed ] ) and bilinearity of the mixed volume , by induction we obtain @xmath297 now , by pick s formula @xmath298 since @xmath299 , the number of boundary lattice points , is at least 3 . therefore @xmath300 substituting this into ( [ e : before ] ) and simplifying we obtain @xmath301 it remains to note that the maximum of the right hand side of ( [ e : after ] ) is attained at @xmath302 , provided @xmath303 and @xmath304 , and that establishes the required inequality . given a polytope @xmath3 , to make use of our bound in theorem [ t:2-bound ] it remains to understand 1 . how to find @xmath2 , the full minkowski length of @xmath3 , 2 . how to determine whether there is a maximal minkowski decomposition in @xmath3 one of whose summands is an exceptional triangle . here we provide algorithms that answer these questions in polynomial time in @xmath169 . recall that a zonotope @xmath305 is called _ maximal for @xmath3 _ if @xmath306 , the number of non - trivial minkowski summands ( counting their multiplicities ) , is equal to @xmath2 . it follows from proposition [ p : simple ] that a maximal zonotope always exists although it is usually not unique . it turns out that any maximal zonotope of @xmath3 has at most four distinct summands and among them there are maximal zonotopes with a particularly easy description . [ p:3segments ] let @xmath3 be a lattice polygon . then 1 . any zonotope @xmath65 maximal for @xmath3 has at most 4 different summands . 2 . there exists a zonotope @xmath65 maximal for @xmath3 with at most 3 different summands . moreover up to an @xmath80-transformation these summands are @xmath149 $ ] , @xmath150 $ ] , and @xmath151 $ ] . let @xmath307 be a zonotope maximal for @xmath3 and let @xmath308 be the vector of @xmath309 . according to lemma [ l : ml3 ] , @xmath310 for any @xmath311 . the case when all @xmath312 are the same is trivial . suppose there are exactly two different summands , i.e. @xmath313 for some positive integers @xmath314 and @xmath315 . if @xmath316 then we can transform @xmath317 to the standard basis @xmath318 and ( 2 ) follows . if @xmath319 then we can assume that @xmath320 and @xmath321 . but @xmath322 contains @xmath323 $ ] , so we can pass to @xmath324 + 2m_2[0,e_2]$ ] . clearly @xmath325 and @xmath326 is maximal . now suppose that @xmath65 has at least three different summands . first , assume @xmath327 for all @xmath328 . as before , without loss of generality , @xmath320 and @xmath321 . consider @xmath329 . by looking at the determinants @xmath330 for @xmath85 we have @xmath331 and @xmath332 . this implies that @xmath333 is not primitive , a contradiction . therefore , @xmath334 for some @xmath328 and we can assume that @xmath320 and @xmath335 . again , we let @xmath329 and look at the determinants @xmath330 for @xmath85 . we see that the only vectors @xmath333 ( up to central symmetry ) that may appear are @xmath121 , @xmath140 , @xmath336 , @xmath138 , @xmath135 , @xmath337 . no two out of the last four vectors can appear together as they generate parallelograms of area at least 3 . for the same reason @xmath121 can not appear with @xmath138 or @xmath337 , and @xmath140 can not appear with @xmath336 or @xmath135 . we have three possible combinations : * @xmath338 * @xmath339 and @xmath340 or @xmath341 * @xmath342 and @xmath343 or @xmath344 we have proved our first claim . to prove the second , note that we can actually reduce the number of distinct segments @xmath309 . in case ( a ) @xmath345 and we will be able to get rid of either @xmath147 or @xmath346 by replacing @xmath347 with @xmath233 . in either case the remaining segments are @xmath80-equivalent to @xmath149 $ ] , @xmath150 $ ] , and @xmath151 $ ] . in case ( b ) we can assume that @xmath340 . since @xmath348 we will be able to get rid of either @xmath114 or @xmath346 and the remaining segments are @xmath80-equivalent to @xmath149 $ ] , @xmath150 $ ] , and @xmath151 $ ] . case ( c ) is obtained from ( b ) by flipping the second coordinate . to find @xmath2 we only need to look at all the zonotopes @xmath349 with at most three different summands @xmath80-equivalent to @xmath149 $ ] , @xmath150 $ ] , and @xmath151 $ ] and find the one that has the largest number of summands ( counting multiplicities ) . [ t : alg1 ] let @xmath3 be a lattice polygon and let @xmath169 be the number of lattice points in @xmath3 . then the full minkowski length @xmath2 can be found in polynomial time in @xmath169 . the case when @xmath3 is 1-dimensional is trivial so we will be assuming that @xmath3 has dimension 2 . for every triple of points @xmath350 , where it is important which point goes first and the order of the other two does not matter , we check if @xmath351 $ ] and @xmath352 $ ] generate a parallelogram of area one . if so , we want to construct various zonotopes whose summands are @xmath114 , @xmath115 and @xmath353 $ ] . we do this in the most straightforward way . first , for every @xmath144 , we find @xmath354 , the largest integer such that a lattice translate of @xmath355 is contained in @xmath3 . for this we find the maximum number of lattice points in the linear sections of @xmath3 with lines in the direction of @xmath152 ( there are finitely many such lines with at least one lattice point of @xmath3 ) . second , for each triple of integers @xmath356 where @xmath357 , we check if some lattice translate of the zonotope @xmath358 is contained in @xmath3 ( we run through lattice points @xmath183 in @xmath3 to check if @xmath359 is contained in @xmath3 ) . for all such zonotopes that fit into @xmath3 we look at @xmath360 and find the maximal possible value @xmath361 of this sum . finally , the largest such sum @xmath361 over all choices of @xmath350 is @xmath2 , by proposition [ p:3segments ] . clearly , this algorithm is polynomial in @xmath169 . notice that in the above we have taken care of the maximal zonotopes that are possibly multiples of a single segment . indeed , if @xmath362 $ ] is a primitive segment connecting two lattice points in @xmath3 then unless @xmath3 is 1-dimensional there is a lattice point @xmath363 in @xmath3 such that @xmath362 $ ] and @xmath364 $ ] generate a parallelogram of area one . we can assume that @xmath30 is the origin and @xmath365 . let @xmath366 be a lattice point in @xmath3 with smallest positive @xmath367 ( flip @xmath3 with respect to the @xmath368-axis if necessary ) . by the minimality of @xmath367 the triangle @xmath369 has no lattice points except its vertices . by pick s formula , its area is @xmath89 and we have found the required third vertex @xmath363 . [ t : alg2 ] let @xmath3 be a lattice polygon in @xmath40 . then we can decide in polynomial time in @xmath169 if there is a maximal minkowski decomposition in @xmath3 one of whose summands is an exceptional triangle . we first run the algorithm from theorem [ t : alg1 ] to find @xmath2 . next for each triple of points @xmath370 we check if the triangle @xmath371 has exactly four lattice points the three vertices @xmath372 and one point @xmath183 strictly inside the triangle . if so , this triangle is exceptional . if this triangle is a summand in some maximal minkowski decomposition in @xmath3 then the other summands that may appear in this decomposition are the primitive segments @xmath114 , @xmath115 , and @xmath147 connecting @xmath183 to the vertices @xmath373 ( see remark [ r : invtzero ] ) . now it remains to look at all minkowski sums @xmath374 with @xmath94 and check if any of them fits into @xmath3 . if this indeed happens for some @xmath371 , there is a maximal decomposition in @xmath3 with an exceptional triangle . otherwise any maximal decomposition is a sum of primitive segments and unit triangles . clearly , this algorithm is polynomial in @xmath169 . in this section we illustrate our methods with three examples . example [ ex:2 ] was given by joyner in @xcite . example [ ex:3 ] appears in the little and schenck s paper @xcite . one can easily check that @xmath379 and there is a maximal decomposition in @xmath3 containing @xmath73 . in fact , @xmath3 contains @xmath380+[0,e_1+e_2]$ ] . it defines a toric surface code of dimension @xmath381 . to apply theorem [ t:2-bound ] we compute @xmath382 , so @xmath383 . therefore , @xmath384 for all @xmath385 . in this particular example we can establish a better lower bound for @xmath25 , namely @xmath251 . indeed , we have already seen in the proof of theorem [ t:2-bound ] that every @xmath21 with 2 absolutely irreducible factors will have at most @xmath386 for all @xmath251 ( see ( [ e : ourbound1 ] ) ) . if @xmath21 is absolutely irreducible we use ( [ e : hw0 ] ) . then it has at most @xmath387 zeroes since @xmath388 has at most 5 interior lattice points in which case it will have at least 2 primitive edges . but @xmath389 for all @xmath251 . this example is similar to the previous one . we also have @xmath379 , @xmath382 , but the dimension of the corresponding toric surface code is slightly smaller , @xmath392 . however in this case @xmath3 has no exceptional triangles in any maximal decomposition . therefore , theorem [ t:2-bound ] provides a better bound for the minimum distance : @xmath393 which holds for all @xmath394 . as before , this can be improved to @xmath395 using ( [ e : hw0 ] ) and the fact that @xmath396 . note that @xmath397 , for @xmath398 distinct , has exactly @xmath399 zeroes in @xmath13 , hence for @xmath395 the above bound is exact : @xmath400 for @xmath401 this was previously established by joyner @xcite . also ( [ e : exact ] ) follows from little and schenck s result @xcite for all @xmath402 . we have @xmath379 , @xmath404 and @xmath0 has dimension @xmath405 . also @xmath3 has no maximal decomposition with an exceptional triangle . therefore , theorem [ t:2-bound ] implies @xmath393 for all @xmath395 . little and schenck s result @xcite proves this bound for @xmath406 . in fact we can show more in this example : for all @xmath407 @xmath408 to see this , first note that @xmath409 , for @xmath398 distinct , has exactly @xmath410 zeroes in @xmath13 . furthermore , every maximal decomposition in @xmath3 is of the form @xmath411 , where @xmath152 is a primitive segment in the direction of @xmath111 , @xmath112 , or @xmath113 . this implies that every polynomial @xmath21 with the largest number of absolutely irreducible factors ( three ) will have at most @xmath410 zeroes in @xmath13 ( here we take into account the intersections of the irreducible curves defined by the factors of @xmath21 ) . now we claim that for @xmath407 polynomials with fewer factors ( one or two ) will have at most @xmath410 zeroes in @xmath13 as well . indeed , decompositions with 2 summands in @xmath3 can have at most one exceptional triangle , hence , @xmath412 for every @xmath21 with 2 irreducible factors . this will be no greater than @xmath410 for @xmath413 . if @xmath21 is absolutely irreducible then by ( [ e : hw0 ] ) @xmath414 , which is no greater than @xmath410 starting with @xmath415 .	in this paper we prove new lower bounds for the minimum distance of a toric surface code @xmath0 defined by a convex lattice polygon @xmath1 . the bounds involve a geometric invariant @xmath2 , called the full minkowski length of @xmath3 which can be easily computed for any given @xmath3 .
the electromagnetic form factor of pion is an interesting quantity to investigate in lattice qcd . experimentally it has been measured in a set of experiments @xcite . together with the nucleon form factors , it provides the first important test case of our understanding of hadron structure that arises from the quark content . from lattice qcd point of view , form factor calculations represent one of the first steps going beyond static quantities like the mass spectrum which require only two - point functions . the pion form factor is a natural first choice in this direction since usually pion green s functions are statistically the most stable quantities in lattice qcd measurements . an interesting point with the pion form factor @xmath10 is its slope at the origin as a function of the momentum transfer squared @xmath11 , _ i.e. , _ the pion charge radius defined by @xmath12 it has been known for some time from chiral perturbation theory analysis @xcite that this quantity diverges logarithmically with vanishing pion mass squared . quantitative confirmation of such a behavior would provide an important check on the control of chiral behavior in lattice qcd simulations toward the physical point . the pioneering lattice qcd calculations of the pion form factor appeared more than 20 years ago @xcite , and a number of studies were carried out over the years . recently , with the development of simulations with dynamical quarks , several groups have attempted calculations with @xmath13 @xcite and @xmath0 @xcite dynamical flavors using a variety of quark actions . the @xmath13 calculations employed wilson - clover @xcite , twisted mass @xcite or overlap @xcite quark action , and explored the pion mass region down to @xmath6 mev . the values for @xmath14 from those simulations are significantly smaller than the experimental value , and nnlo fits of su(2 ) chiral perturbation theory were needed to find consistency with it at the physical pion mass . for @xmath0 dynamical flavors , there has been a single calculation employing domain - wall quark action @xcite , which made measurements at a single pion mass of @xmath6 mev . carrying out nlo analyses in su(2 ) and su(3 ) chiral perturbation theory , this work found @xmath14 to be consistent with the experimental value at the upper edge of a 10% error band . in this paper we present our calculation of the electromagnetic form factor of pion in @xmath0 dynamical flavor qcd using the wilson - clover quark action . for measurements we employ the @xmath0 pacs - cs gauge configurations generated on a @xmath15 lattice using the iwasaki gauge action and the wilson - clover action at a lattice spacing estimated to be @xmath2 fm at the physical point @xcite . since the pion mass on this gauge configuration set covers the range from @xmath16mev down to @xmath17 mev , we are able to examine both the known range above @xmath6mev and a novel range below toward the physical pion mass . the paper is organized as follows . in sec . 2 we present our method to calculate the pion form factor . in order to access the region of small momentum transfer , we use the method of partially twisted boundary condition @xcite , and in order to fight increasing computational cost for smaller pion mass , we apply the method of random wall source @xcite . in addition we make use of a judicious choice of momenta carried by the incoming and outgoing quarks off the electromagnetic vertex , which helps in reducing statistical fluctuations in the form factor measurements . in sec . 3 we present the results of pion form factor measurements , and in sec . 4 we analyze the data as a function of the momentum transfer squared and pion mass . in particular we examine consistency with the predictions of chiral perturbation theory . finally , in sec . 5 , we discuss our findings closer to the physical point including the results of our test calculation on a @xmath8 lattice with the quark masses tuned to the neighbour of the physical point . we end this work with conclusions in sec . a preliminary report of this work was presented in @xcite . the electromagnetic pion form factor @xmath18 is defined by @xmath19 where @xmath20 is the four - momentum transfer , and @xmath21 is the electromagnetic current given in @xmath0 qcd by @xmath22 in the limit of vanishing four - momentum transfer @xmath23 , the form factor equals unity due to the charge conservation . we extract the form factor from a suitable ratio of the pion two- and three - point functions . we use a ratio , which has the advantage of simultaneously reducing fluctuations and renormalizing the current , defined as @xmath24 which converges as @xmath25 for large @xmath26 and @xmath27 . @xmath28 denotes pion energy for spatial momentum @xmath29 , @xmath30 is the three - point function with momenta @xmath29 at the source and @xmath31 at the sink @xmath27 , @xmath32 and @xmath33 is the two - point function , @xmath34 after contraction of quark fields , the three - point function consists of the connected and disconnected contributions . the latter vanishes after the gauge field average due to charge conjugation invariance , and hence need not be calculated . since we assume degeneracy of up and down quark masses in the present calculation , the connected contribution is equal to @xmath35\right>.\ ] ] this contribution can be calculated by the traditional source method @xcite . the ratio ( [ eq : ratio ] ) makes use of two- and three - point functions in an appropriate combination to extract the form factor for the renormalized current . the presence of ratios guarantee that statistical fluctuations are suppressed . nonetheless , making simple choices such as @xmath36 and @xmath37 , we have observed an increasingly larger fluctuation of the ratio as pion mass is reduced , and this trend worsens for larger momenta . with an interesting choice of momenta , @xmath38 but @xmath39 , the ratio ( [ eq : ratio ] ) simplifies to @xmath40 since the two - point functions as well as the ratio of energies drop out , leaving just the ratio of three - point functions , we expect this choice to yield better signals than those choices for which all factors are present . furthermore , one can choose 6 permutations in momentum directions while keeping @xmath39 , gaining more statistics . the minimum non - zero quark momentum @xmath41 for the periodic boundary condition on a @xmath15 lattice with a 2 gev inverse lattice spacing is about 0.4 gev . to probe the region of smaller momentum transfer as well as to improve the resolution of four - momentum transfer , we apply the method of partially twisted boundary condition @xcite in which valence quark fields are subjected to twisted boundary condition while periodic boundary condition is kept for sea quark fields . if one imposes the boundary condition given by @xmath42 on a valence quark field , the spatial momentum of that quark is quantized according to @xmath43 where @xmath44 denotes the spatial lattice size , @xmath45 the unit vector in the spatial @xmath46-th direction and @xmath47 real parameter . in this way one can explore arbitrarily small momentum on the lattice by adjusting the value of twist @xmath47 . for the meson two - point function consisting of quark and anti - quark propagators , we apply the twist only to quark and not to antiquark or vice versa . similarly , for the three - point function , we twist only one or two out of the three quark propagators . in other words , we pretend that each valence quark line in the two- and three - point function quark diagrams carry a different flavor and select the appropriate flavor to apply twisting . in this way we can avoid a twist of a quark line cancelled by the opposite twist of the antiquark line carrying the same flavor @xcite . this procedure and the twisting of only valence quarks mean that we deal with partially quenched qcd with a different flavor symmetry content in the valence and sea quark sectors . as was discussed in detail in @xcite using chiral perturbation theory , the associated effects are expected to appear as finite - size effects exponentially small in spatial volume for channels which do not have final - state interactions such as three - point functions for form factor calculations . since terms of such magnitude are also present in unitary theory with periodic valence and sea quarks , we ignore this issue in the present work . the twisted boundary condition can be imposed on a periodic quark field configuration by the following transformation @xmath48 in practice we transfer the twist from the quark sector to the gluon sector by an @xmath49 transformation on the spatial gluon link fields given by @xmath50 thus valence quark propagators are solved with the periodic boundary condition but on the pacs - cs gauge configurations twisted by the u(1 ) transformation above . in order to check that the term @xmath51 acts as true physical momentum , we carried out a test of the energy - momentum dispersion relation of the pion on some pacs - cs configurations . of the two valence quarks inside the pion , we twisted one quark with a twist angle @xmath52 and left the other untwisted . in fig . [ fig : checktbc13700](a ) we plot the effective energy for the ground state , two values of the twist angle and their combination with the first integer momenta at the hopping parameters @xmath53 , @xmath54 where @xmath55 mev . the propagator is fitted over @xmath56 to extract the energy @xmath57 . errors are estimated by the jackknife method with the bin size of 100 trajectories . the results are plotted in fig . [ fig : checktbc13700](b ) , together with the expected behavior , @xmath58 which demonstrates clearly that the term @xmath51 acts as true physical momentum . the two data points on the right represent combinations of an integer momentum @xmath59 and a twist . the energy - momentum relation is correctly reproduced in this case as well . at light quark masses the computing cost for inversion of dirac operator becomes very expensive . thus we have employed some improvements for obtaining the form factor with acceptable statistical errors at reasonable computing time . the first improvement is to utilize the random wall source . this method has a long history and has been applied to two - point functions in a variety of contexts . more recently , applications to three - point functions have shown their effectiveness for form factor calculations @xcite . we consider the use of @xmath60 random noisy source as introduced in @xcite . consider a set of random sources whose real and imaginary components are randomly chosen from @xmath61 for each site , color and spin , @xmath62 this set has the property that in the limit @xmath63 @xmath64 to use this kind of source in calculating correlators , one can choose it to be a set of random wall source located at @xmath65 , @xmath66 @xmath67 making use of ( [ eq : impr5 ] ) to rewrite the pion two - point function at zero momentum as , @xmath68 d^{-1\dagger}_{c\kappa , a\alpha}(\vec{y},t ; \vec{z},t_0 ) \big)\\ & = & \sum_{\vec{y } } \left < \psi^{(n)}(\vec{y},t\|t_0 ) \psi^{\dagger(n)}(\vec{y},t\|t_0 ) \right>_n,\end{aligned}\ ] ] where @xmath69 is the solution vector of the dirac equation , @xmath70 + with a random @xmath60 wall source the solution for quark propagator needs only single inversion instead of @xmath71 corresponding to color and dirac components required for a point source or smeared source . when the number of configurations in the ensemble is large enough , even if one uses a single random source for each configuration , ( [ eq : impr5 ] ) is expected to hold in the ensemble average . one may expect to obtain meson correlators of a similar statistical quality as with the traditional point source with only 1/12 of computing time . in fig . [ fig : c2ptpointvsz2 ] , we compare the effective pion mass plot calculated on a set of 10 configurations at @xmath72 ( @xmath73mev ) from the pacs - cs ensemble using ( a ) point source , ( b ) single random @xmath60 wall source , and ( c ) smeared source . we observe that the signal with the random @xmath60 wall source is somewhat better than that for point source , while the signal for smeared source is better than that with the random @xmath60 wall source . using 4 random wall sources for each configuration , we observed that the quality of signal becomes comparable to that for smeared source . since the computing time is still @xmath74 for the random wall source , we employ the method of random wall source with 4 sets of random wall sources in our measurements . in addition we repeat measurements with the source located at @xmath75 since the time extent of our lattice is 64 . we apply our calculational setup to a subset of the pacs - cs gauge configurations @xcite corresponding to the degenerate up - down hopping parameter in the set @xmath76 , @xmath77 , @xmath78 , @xmath79 . the hopping parameter of strange quark is fixed at @xmath80 . the first set of measurements , which we call data set i , is made with an exponentially smeared source and local sink , setting the final pion at zero momentum @xmath81 and varying that of the initial pion @xmath29 in the three - point function . the fixed sink time @xmath27 in the ratio ( [ eq : ratio ] ) needs to be chosen large enough to eliminate excited states contributions . however , statistical fluctuations increase as @xmath27 increases , and examining measurement results , we choose @xmath82 to balance the two opposite features . the twist technique is applied to the quark running from the source to the current . two values are chosen for the twist angle @xmath83 such that the smallest four - momentum transfer of the current takes the value @xmath84 or @xmath85 . adding integer momenta , we then collect data for @xmath86 in the range @xmath87 . the statistics of data set i is given in table [ table : stat_dataset1 ] together with pion and kaon mass . results of data set i have been previously reported in @xcite . in order to extract the form factor , we fit the plateau of the ratio @xmath88 by a constant . the fitting range should be chosen around the symmetry point between the source and the sink , with additional considerations on the time interval required for the pion state to become dominant . since we employ an exponential smeared source and a point sink , we shift the fitting range one time unit closer to the source than the symmetric point @xmath89 . in fig . [ fig : plateau_pff](a ) we plot the ratio @xmath88 at various momentum transfer for the pion mass @xmath55 mev . at this pion mass we have good signals for all 7 values of the four - momentum transfer . there is a good plateau from @xmath90 to 15 for every momentum transfer . thus at this pion mass we can choose the fitting range from @xmath90 to 15 to extract the form factor . however , as the pion mass decreases , the plateau signal becomes worse as exhibited in fig . [ fig : plateau_pff_296](a ) for the lightest case of @xmath91 mev where the ratio @xmath88 at the two smallest momentum transfers , @xmath84 and @xmath85 , is shown . we then choose larger values for the starting point of the fitting range for better suppression of excited states at lighter pion masses . the error is estimated by the jackknife method using 10 configurations corresponding to 50 hybrid molecular dynamics time units as the bin size after checking saturation of the magnitude of error as function of bin size . fit results for the pion form factor are listed in table [ table : pff_dataset1 ] . .statistics of data set i. [ cols="^,^,^,^,^,^",options="header " , ] for checking the convergence at nnlo , we carry out fits employing both @xmath92 and @xmath93 as the expansion parameter . for the former fit we use ( [ eq : decayconstant ] ) to reexpand the expression for the form factor to the necessary order , and use the value of @xmath94 obtained in @xcite . besides the pion decay constant the chpt formula of the form factor to nnlo depends on 5 other lecs : @xmath95 . it is very difficult to find a stable fit in the 5-dimension parameter space . therefore we fix @xmath96 at the phenomenology values since they were calculated with small error bar from experimental data@xcite . for @xmath97 , which is only required in the formulation with @xmath98 , we apply the value obtained by an nlo fit of data in the range @xmath99 mev by the pacs - cs collaboration@xcite . we find that stable fits are difficult to obtain unless we utilize data at more than a single pion mass in the fit procedure . the fit result obtained with data for the two pion masses @xmath100 and 411 mev , which were used for the nlo analysis , is listed in table [ table : pff_chpt_su2_nnlo_2lightest ] and shown in figs . [ fig : chpt_su2_nnlo_2lightest - f ] and [ fig : chpt_su2_nnlo_2lightest - fpi ] . combining the data at the two pion masses is acceptable since strange quark mass does not vary much , @xmath101 and 92.2(1.3 ) mev at @xmath100 and 411 mev , respectively @xcite . we observe that the results for @xmath102 are consistent between the two fits within the error of 1015% and so are the predictions for the squared charge radius at the physical point , indicating that chpt reasonably converges at nnlo up to @xmath103 mev and @xmath104 gev@xmath105 . this is also seen by plotting the nlo and nnlo contributions separately as shown in figs . [ fig : chpt_su2_nnlo_2lightest - f ] and [ fig : chpt_su2_nnlo_2lightest - fpi ] . the squared charge radius predicted at the physical point is close to the experimental value and is consistent within statistics errors of 10% . while chpt to nnlo yields a reasonable result for the physical pion charge radius , the estimated error of 10% is quite large . we feel that for a convincing understanding of the physical pion charge radius one needs to explore the region of pion mass closer to the physical point than the value @xmath6 mev analyzed so far . the pacs - cs gauge configurations has one more set corresponding to @xmath106 . we tried to calculate the form factor on this set , and found that the pion two- and three - point correlators exhibit very large fluctuations , to the extent that taking a meaningful statistical average is difficult . this trend becomes more pronounced as the twist carried by quarks becomes larger . since @xmath107 at this pion mass for @xmath108 , we suspect that this phenomenon is caused by a small size of the lattice relative to the pion mass , and consequent increase of large fluctuations . a natural remedy to this difficulty is to employ larger lattices as one moves toward the physical pion mass . pacs - cs collaboration has been pushing a simulation on the physical point on a @xmath8 lattice as a continuation of the work on a @xmath15 lattice . the hopping parameter of the run is adjusted to the best estimate of the physical point @xmath109 . we have used a subset of those configurations to calculate the pion form factor on a @xmath8 lattice . this requires much computer time , and hence we have only 4 configurations measured so far . we used the same setup as for the data set ii , namely , ( i ) the incoming and outgoing pions in the three - point function carry momenta of the same magnitude @xmath39 but point in different directions , ( ii ) 4 random @xmath60 wall sources located at @xmath75 are employed , ( iii ) the twist technique is applied to the two quarks running from the source to the current and from the current to the sink , and ( iv ) four values are chosen for the twist angle @xmath110 and its permutations such that the four - momentum transfer of the current takes the value @xmath111 . the fixed sink time @xmath27 is chosen to be 28 . with 4 source points located at @xmath75 and 4 random sources for each source location . center lines exhibit pacs - cs estimation of pion mass calculated from larger statistics of 53 configurations . ] since we use 4 random sources for each of the 4 locations of the source in time , our measurement on 4 configurations gives 64 two - point functions . the pion effective mass from our measurement together with the pacs - cs estimate of pion mass calculated from larger statistics of 53 configurations is shown in fig . [ fig : c2ptdata_64latt ] . although only 4 configurations have been used , one can already observe a plateau - like behavior for pion effective mass in this figure . the central value from the pacs - cs estimate corresponds to @xmath112 mev . this is somewhat smaller than the charged pion mass , and significantly smaller than @xmath113 mev considered earlier on a @xmath15 lattice where we encountered problem of large fluctuations . mev with measurement taken on 4 configurations of a @xmath8 lattice . black line is a fit of two data points closest to @xmath23 to the monopole ansatz ( [ eq : mono ] ) . ] in fig . [ fig : pff_64latt ] we plot the ratio @xmath114 obtained from the 4 configurations . making a constant fit over @xmath115 yields the result for the form factor displayed in fig . [ fig : pff_q2_64latt ] . our data seems reasonable up to @xmath116gev@xmath105 . estimating the slope at the origin by a monopole fit to the two points closest to @xmath23 , we obtain @xmath117 . while the error is too large to seriously discuss consistency with experiment , it is certainly encouraging that the value is larger than those obtained at @xmath118 of about 300 mev , and that the physical point simulation appears possible for the pion electromagnetic form factor on a @xmath8 lattice . we have presented a lattice calculation of the pion electromagnetic form factor in 2 + 1 dynamical flavor qcd with the o(a)-improved wilson - clover quark action and iwasaki gauge action . in order to obtain data with reasonable error at light quark masses close to the physical point , we have utilized some improved techniques besides traditional methods for the form factor calculation . we have shown that , choosing momenta of the incoming and outgoing pions to have the same magnitude but different directions , the ratio for extracting the pion form factor becomes statistically much better behaved . we have confirmed the validity of the twisted boundary condition and employed it to explore the form factor in the region of small four - momentum transfer . application of the random @xmath60 wall source has helped us to save computing time considerably . our data for the pion mean - square charge radius @xmath119 agree with recent data of other groups , and show that @xmath119 increases toward the physical value as @xmath118 decreases . nevertheless , on a @xmath1 lattice , we could extract reasonable data only down to @xmath120 mev . chpt analysis of the form factor utilizing nlo su(3 ) or su(2 ) formula lead to the squared charge radius which is smaller than experiment by 1520% . employing chpt to nnlo improves the agreement . in fact our nnlo su(2 ) fit using @xmath93 as the expansion parameter yields @xmath121 where the first error is statistical and the second error due to the error in the lattice spacing . we feel that a complete explanation of the behavior of the squared charge radius would require successful calculation of the form factor below @xmath6 mev . our experience points toward the necessity of enlarging the lattice size sufficiently . a sample calculation on a @xmath8 lattice indicates that the @xmath8 lattice with @xmath9 probably satisfies the requirement . we leave further exploration of the form factor calculation on such a lattice as future work . numerical calculations for the present work have been carried out under the `` interdisciplinary computational science program '' of center for computational sciences , university of tsukuba . we thank the members of the pacs - cs collaboration for discussions . this work is supported in part by grants - in - aid for scientific research from the ministry of education , culture , sports , science and technology ( nos . 16740147 , 18104005 , 20740123 , 20740139 ) .	we present a calculation of the electromagnetic form factor of the pion in @xmath0 flavor lattice qcd . calculations are made on the pacs - cs gauge field configurations generated using iwasaki gauge action and wilson - clover quark action on a @xmath1 lattice volume with the lattice spacing estimated as @xmath2 fm at the physical point . measurements of the form factor are made using the technique of partially twisted boundary condition to reach small momentum transfer as well as periodic boundary condition with integer momenta . additional improvements including random wall source techniques and a judicious choice of momenta carried by the incoming and outgoing quarks are employed for error reduction . analyzing the form factor data for the pion mass at @xmath3 mev and 296 mev , we find that the nnlo su(2 ) chiral perturbation theory fit yields @xmath4 for the pion charge radius at the physical pion mass . albeit the error is quite large , this is consistent with the experimental value of @xmath5 . below @xmath6 mev , we find that statistical fluctuations in the pion two- and three - point functions become too large to extract statistically meaningful averages on a @xmath7 spatial volume . we carry out a sample calculation on a @xmath8 lattice with the quark masses close to the physical point , which suggests that form factor calculations at the physical point become feasible by enlarging lattice sizes to @xmath9 . ut - ccs-60 +
in most of the textbooks the circle of apollonius is discussed in conjunction with the angle bisector theorem : _ the angle bisector in a triangle divides the opposite sides into a ratio equal to the ratio of the adjacent sides . " _ once one realizes that the statement can be equally applied to the exterior angle bisector , then the circle of apollonius appears naturally ( figure [ locus1 ] ) , since the two angle bisectors are perpendicular . ] for instance , an easy exercise in algebra shows that the circle of equation @xmath0 is equivalent to @xmath1 taking @xmath2 , @xmath3 , @xmath4 and @xmath5 . similar calculations can be employed to treat the general situation , i.e. , taking @xmath6 , @xmath7 and @xmath8 with real positive numbers @xmath9 , @xmath10 and @xmath11 such that @xmath12 . then we can state the well - known result : [ apollonius ] given points @xmath13 , @xmath14 and @xmath15 as above , the set of points @xmath16 in the plane characterized by the equality @xmath17 is \(i ) the line of equation @xmath18 if @xmath19 ; \(ii ) the circle of equation @xmath20 , if @xmath21 and @xmath22 . let us observe that the statement of this theorem does not reduce the generality since the coordinate of point @xmath23 can be shown to be @xmath24 and one can take the origin of coordinates to be the midpoint of @xmath25 . this turns out to happen precisely when @xmath22 . what is the equivalent of this result in hyperbolic geometry we are going to use the half - plane model , @xmath26 , to formulate our answer to this question ( see anderson @xcite for the terminology and notation used ) . without loss of generality we may assume as before that the three points are on the @xmath27-axis : @xmath6 , @xmath7 and @xmath8 , with real positive numbers @xmath9 , @xmath10 and @xmath11 such that @xmath12 . [ theorem1 ] given points @xmath13 , @xmath14 and @xmath15 as above , the set of points @xmath16 in the half - plane @xmath26 , characterized by the equality @xmath17 in @xmath26 is the curve given in polar coordinates by @xmath28 moreover , \(i ) if @xmath29 , this curve is half of the hyperbola of equation @xmath30 \(ii ) if @xmath31 this curve is the semi - circle @xmath32 , @xmath33 , \(iii ) if @xmath34 , this curve is half of the lemniscate of equation @xmath35 it is interesting that all these particular cases in theorem [ theorem1 ] , can be accomplished using integer values of @xmath9 , @xmath10 and @xmath11 . this is not surprising for the diophantine equation @xmath22 since one can play with the prime decomposition of @xmath9 and @xmath11 to get @xmath36 a perfect square . for the equation @xmath37 one can take a pythagorean triple and set @xmath38 , @xmath39 and @xmath40 for @xmath41 . perhaps it is quite intriguing for some readers that the last diophantine equation @xmath42 is satisfied by the product of some quadratic forms , namely @xmath43 another surprising fact is that theorem [ theorem1 ] appears as a more general result than theorem [ apollonius ] because of part ( ii ) and the observation that we can always assume that the axes of coordinates are chosen with the origin to be the center of the apollonius circle . @xmath44{figure3.eps}}$ ] @xmath45{figure4.eps}}$ ] @xmath46{figure5.eps}}$ ] @xmath47{figure6.eps}}$ ] @xmath48{figure7.eps}}$ ] @xmath49{figure8.eps}}$ ] in figure [ locus2 ] , we included all of the possible shapes of the locus in theorem [ theorem1 ] except for the case @xmath50 which is similar to the case @xmath51 . we notice a certain symmetry of these cases showing that the hyperbola ( @xmath29 ) is nothing else but a lemniscate in hyperbolic geometry . with this identification , it seems like the curves we get , resemble all possible shapes of the intersection of a plane with a torus . in the next section we will prove the above theorem and in the last section we will give the motivation of our work . let us consider a point @xmath52 of coordinates @xmath53 with the given property as in figure [ locus3 ] which is not on the line @xmath54 , ( @xmath55 ) . then the hyperbolic lines determined by @xmath52 and the three points @xmath13 , @xmath14 and @xmath15 are circles orthogonal on the @xmath56-axis . we denote their centers by @xmath57 , @xmath58 and @xmath59 . and the lines determined by it with @xmath13 , @xmath14 and @xmath15 ] the point @xmath61 can be obtained as the intersection of the perpendicular bisector of @xmath62 and the @xmath56-axis . similarly we obtain the other two points @xmath63 and @xmath64 . the equation of the perpendicular bisector of @xmath62 is @xmath65 and so @xmath66 . similar expressions are then obtained for @xmath67 and @xmath68 , i.e. , @xmath69 and @xmath70 . which shows that the order of the points @xmath61 , @xmath63 and @xmath64 is reversed ( @xmath71 ) . the angle between the hyperbolic lines @xmath72 and @xmath72 is defined by the angle between the tangent lines to the two circles at @xmath52 , which is clearly equal to the angle between the radii corresponding to @xmath52 in each of the two circles . so , @xmath73 and @xmath74 . this equality is characterized by the proportionality given by the angle bisector theorem in the triangle @xmath75 : @xmath76 using polar coordinates , @xmath77 and @xmath78 , we observe that @xmath79 and @xmath80 . hence the above equality is equivalent to @xmath81 one can check that a factor of @xmath82 can be simplified out and in the end we obtain ( [ eq1 ] ) . our interest in this locus was motivated by the problem 11915 in this monthly ( ) . this problem stated : _ given four ( distinct ) points @xmath13 , @xmath14 , @xmath15 and @xmath23 in ( this ) order on a line in euclidean space , under what conditions will there be a point @xmath52 off the line such that the angles @xmath83 , @xmath84 , and @xmath85 have equal measure ? _ it is not difficult to show , using two apollonius circles , that the existence of such a point @xmath52 is characterized by the inequality involving the cross - ratio @xmath86=\frac{\frac{bc}{ba}}{\frac{dc}{da}}<3.\ ] ] we were interested in finding a similar description for the same question in hyperbolic space . one can think using the same idea of the locus that replaces the apollonius circle in euclidean geometry and that is why we looked into finding what this locus is . having the description of this locus one can see that the intersection of two curves as in theorem [ theorem1 ] is difficult to predict . fortunately , we can use the calculation done in the proof of theorem [ theorem1 ] and formulate a possible answer in the new setting . given four points @xmath13 , @xmath14 , @xmath15 and @xmath23 in ( this ) order on a line in the hyperbolic space , we can use an isometry to transform them on the line @xmath87 and having coordinates @xmath6 , @xmath7 , @xmath8 and @xmath88 with @xmath89 . the the existence of a point @xmath52 off the line @xmath87 , such that the angles @xmath83 , @xmath84 , and @xmath85 have equal measure in the hyperbolic space is equivalent to the existence of @xmath52 in euclidean space corresponding to the points @xmath61 , @xmath63 , @xmath64 and @xmath90 as constructed in the proof of theorem [ theorem1 ] . therefore the answer is in terms of a similar inequality @xmath91=\frac{\frac{b'c'}{b'a'}}{\frac{d'c'}{d'a'}}<3 \leftrightarrow \frac{(b^2-c^2)(a^2-d^2)}{(a^2-b^2)(c^2-d^2)}<3.\ ] ] to have a different take of what ( [ eq2 ] ) means we will translate it into a geometric probability which is not difficult to compute : _ if two points are randomly selected ( uniform distribution ) on the segment @xmath92 , then the probability that a point @xmath52 off the line @xmath92 such that the angles @xmath83 , @xmath84 , and @xmath85 have equal measure exists ( where the two points are denoted by @xmath14 and @xmath15 , @xmath14 being the closest to @xmath13 ) , is equal to @xmath93 _ the inequality ( [ eq3 ] ) gives us the similar probability in the hyperbolic space : @xmath94 where the uniform distribution here means , it is calculated with respect to the measure @xmath95 along the @xmath27-axis . our both problems must have an even more interesting answer in the setting of spherical geometry . due to the infinite nature of both euclidean and hyperbolic spaces the geometric probability question makes sense only in limiting situations . in this case we can simply ask : * problem 2 : * _ given a line in spherical geometry and four points on it , chosen at random with uniform distribution , what is the probability that the points look equidistant from a point on the sphere that is not on that line ? _ we find the problems more fascinating because there is no clear order in this geometry . given three collinear points , any of them can be thought as between the other two . so , we may expect three different curves as a result . if three points @xmath13 , @xmath14 and @xmath15 are already equidistant on a line @xmath96 , the three lines perpendicular to @xmath96 through these points satisfy the locus requirement and they are concurrent . is there such a point ( in fact at least two if one exists ) of concurrency , for a general position of three points on a line ? perhaps not , if the three points are close to one another ; but there must be quite a variety of situations when it is possible . also , the probability in question , we anticipate to be somewhat bigger .	in euclidean geometry the circle of apollonious is the locus of points in the plane from which two collinear adjacent segments are perceived as having the same length . in hyperbolic geometry , the analog of this locus is an algebraic curve of degree four which can be bounded or unbounded " . we study this locus and give a simple description of this curve using the half - plane model . in the end , we give the motivation of our investigation and calculate the probability that three collinear adjacent segments can be seen as of the same positive length under some natural assumptions about the setting of the randomness considered . [ section ] [ theorem ] proposition [ theorem ] corollary [ theorem ] lemma [ theorem ] definition [ theorem ] problem [ theorem ] conjecture [ theorem ] example [ theorem ] remark
recent progress in building devices for quantum information processing ( qip ) has lead to renewed interest in studying the relaxation times of electron spins bound to impurities and in quantum dots to be used for encoding quantum information @xcite . electron spins bound to shallow donors in si are characterized by particularly long relaxation times , since the spin orbit interaction is weak and isotopically enriched @xmath0si is available @xcite . however a weak interdonor spin coupling represents a challenge for the controlled interaction of quantum bits . the interstitial lithium donor in silicon is unique amongst shallow donors because it has an inverted level structure and near degeneracy of the 1@xmath1 ground state manifold @xcite . the latter gives rise to a strong long - range elastic - dipole coupling of the neutral li donor _ orbital _ states @xcite , similar to that obtained earlier for acceptors in si @xcite but absent in the case of other donors . in @xcite a quantum computing architecture was proposed based on the stress - defined orbital qubits and the above interaction was used to enable 2-qubit gates . at the same time the near degeneracy of 1@xmath1 manifold in si : li also gives rise to a nontrivial coupling between the spin and orbital degrees of freedom first investigated by watkins and ham @xcite . one can expect that the inverted level structure and near degeneracy in si : li will lead to a very short electron spin relaxation times which can also strongly affect orbital relaxation and be detrimental for qip . somewhat surprisingly , in view of extensive studies of spin relaxation of other shallow donors in silicon since the early sixties , both experimental @xcite and theoretical @xcite , the question about li spin relaxation has not been addressed . in this work we propose a new relaxation mechanism for coupled donor electron spin and orbital degrees of freedom . it is based on ( i ) our new finding of the chiral symmetry of the donor states that persists in prsence of the random starin and magnetic field of certain orientations and ( ii ) an interplay between the umklapp phonon processes and a short - range electric dipole interaction that only weakly break this symmetry and enable spin relaxation . as a result li electron spin and orbital relaxation times become very long in the presence of modest external stress ( @xmath2 10@xmath3 pa ) at temperatures below 0.3 k. we also report pulsed esr measurements of the longitudinal spin relaxation time t@xmath4 in @xmath5si : li at temperatures 1.8 to 4.5 k. we find that in the absence of external stress these times are shorter than our instrumental resolution ( @xmath6 100 ns ) at temperatures as low as 1.8 k. this rapid relaxation is caused by the degeneracy in the ground state manifold . however , a modest applied stress lifts the degeneracy , and leads to a much longer relaxation , as will be discussed below . the results of the theoretical analysis of the selection rules and spin - orbital relaxation mechanisms are in excellent quantitative agrement with the experiment as shown in fig [ fig : comparison ] . energy levels at @xmath7 . ( b ) energy levels at finite stress and @xmath8 . solid lines 2,3 indicate the relatively fast one - phonon transitions connecting the states with the _ same _ chirality . dashed line 1 indicates the slow spin - flip transition between the `` left '' and `` right '' states ; ( c ) comparison of experimental and theoretical kinetics of magnetization recovery at 2.1 k and 4.5 k. the theoretical curves ( thin solid lines ) correspond to @xmath9 mev and @xmath10 pa . the lower inset displays the theoretically calculated temperature dependence of the spin lifetime @xmath11 , an inverse of the lowest eigenvalue of relaxation operator @xmath12 , eq . it is characterized by two activation energies @xmath13 and @xmath14 ( the corresponding activation processes are shown in fig . [ fig : comparison](b ) with dashed lines ) [ fig : comparison ] ] the ground state @xmath15 manifold of an interstitial lithium donor in silicon can be described by the hamiltonian in a twelve dimensional hilbert space formed by a tensor product of electron spin states @xmath16 and six 1@xmath1 orbitals corresponding to the six conduction band minima @xmath17 @xcite @xmath18 the term @xmath19 is the spin - diagonal , effective mass orbital 1@xmath1 hamiltonian also describing the valley - orbit splitting due to the short range tetrahedral potential . the rest of individual terms in @xmath20 to be considered below describe the coupling of the donor electron to magnetic field ( @xmath21 ) , long - wavelength strain field ( @xmath22 ) , spin - orbit interaction ( @xmath23 ) , and a small off - center displacement of li core from the symmetry site ( @xmath24 ) . for each of the two spin orientations the six eigenstates of @xmath19 are linear superpositions of the six silicon valley orbitals , @xmath25 . here @xmath26 and @xmath27 are , respectively , the hydrogenic envelope and the bloch function for the j - th conduction band minima @xmath28 ( @xmath29=@xmath30 or @xmath31 ) . @xmath32 belongs to the irreducible representation @xmath33 of the tetrahedral group characterized by the valley - orbit coefficients @xmath34 . in contrast to the group v substitutional donors the ground state of @xmath19 is five - fold degenerate for each spin orientation . it is composed of a triplet of states @xmath35 which are antisymmetric combinations of just two opposite valleys , @xmath36 , and a doublet @xmath37 that are symmetric combinations of valleys , @xmath38 , with a second - order zero at the origin . a fully symmetric singlet state @xmath39 lies at @xmath40= 1.76 mev above the @xmath41 orbital ground state @xcite , so that matrix elements @xmath42 , where @xmath43 is 2@xmath442 unit matrix operating in spin subspace . electron - phonon interaction @xmath45 within 1@xmath46+@xmath47 manifold can be described in terms of deformation potentials because the states there decay via emission of long - wavelength acoustic phonons with @xmath48@xmath4910@xmath50 . a matrix element between a pair of donor states @xmath51 and @xmath52 of @xmath45 with phonon wavevector @xmath48 and polarization @xmath53 obeys parity selection rules @xcite and can be written in the form @xmath54 , \label{umklapp}\ ] ] where @xmath55 . here @xmath56 is the energy shift of valleys @xmath57 due to the phonon mode @xmath58 with polarization @xmath53 . in can be expressed in terms of the deformation potential constants and angular functions @xcite . @xmath59 and @xmath60 are zero for the states @xmath61 with the same and opposite parities , respectively . in the latter case the transition amplitude ( [ umklapp ] ) contains a very small quantity @xmath62 where @xmath63 is determined by the overlap in @xmath64-space of the envelope functions localized near the different conduction band minima separated by umklapp vector @xmath65 @xmath66 @xcite . matrix elements @xmath67 depend on a shear deformation potential @xmath68 that couples the long - wavelength strain tensor components @xmath69 to the donor states @xcite . for most experimental situations the strain is caused by a superposition of an external uniaxial and much weaker random biaxial stress that completely removes the orbital degeneracy . a compressive uniaxial ( 001 ) stress @xmath70 only partially splits the 1@xmath71 manifold @xcite as shown in fig . [ fig : comparison ] . the ground state is @xmath72 and the first excited state is a superposition of @xmath73 and @xmath39 orbitals with dominant @xmath73 character . the third level is triple - degenerate , corresponding to states @xmath74 as shown in fig . [ fig : comparison ] . for @xmath75@xmath4410@xmath76 pa the splitting of @xmath72 and @xmath73 levels is @xmath77 0.2 mev ; much larger than the random - stress induced level broadening in @xmath0si @xcite . we note that due to the symmetry of an interaction involving deformation potentials @xcite an _ arbitrary _ biaxial stress mixes only even - parity states , @xmath78 and @xmath79 and therefore preserves the above parity - selection rules . matrix elements of the zeeman hamiltonian equal @xmath80 where @xmath81 is the bohr magneton , @xmath82 and @xmath83 are , respectively , the 3 cartesian components of the electron spin operator and magnetic field . also @xmath84 , and @xmath85 are the components of the g - tensor @xcite . there exists a small spin - orbit interaction in the ground state of li donor induced by the host silicon lattice @xcite . it is typically smaller then the strain and zeeman terms . the spin - orbit hamiltonian @xmath23 reads @xcite : @xmath86 , and small constants @xmath87 mev and @xmath88 mev determine the strength of the spin - orbit coupling within the triplet and between the triplet and singlet , respectively @xcite . spin - orbit hamiltonian @xmath23 mixes the orbital states of opposite parities and , as one might expect , it violates the above parity selection rules for the transition rates . however , we show below that the time - reversal symmetry in the absence of a magnetic field leads to a factorization of the total hamiltonian and emergence of the new selection rules that persist for finite magnetic fields of certain orientations . we assume that the magnetic field @xmath89 . then the eigenstates of the hamiltonian @xmath90 form two non - overlapping subspaces with an element in each subspace being a linear superposition of the following 6 basis vectors @xmath91 : @xmath92 where @xmath93 , @xmath94 is the pauli matrix , and @xmath95 is a subspace index . the eignstates @xmath96 and their time - reversed counterparts @xmath97 are also the eigenstates of the `` chirality '' operator @xmath98 , where @xmath99 is the operator of rotation through the angle @xmath100 around @xmath101-axis . indeed , the operator @xmath102 commutes with the hamiltonian @xmath90 and @xmath103 . in the absence of the magnetic field pairs of similar eigenstates with opposite chiralities form kramers doublets with the same energy . a finite magnetic field @xmath104 lifts the kramers degeneracy . however for @xmath105 with @xmath89 the chirality still remains a good quantum number and the subspaces do not mix despite an arbitrary biaxial strain present in @xmath90 . similar subspaces can be constructed for @xmath106 and @xmath107 . watkins and ham postulated a small displacement @xmath108 of the li atom from a tetrahedral interstitial site @xcite , preferentially in the direction of the uniaxial stress @xmath109 to generate the correct number of lines in the si : li esr spectrum . it gives rise to a short - range electric dipole potential @xmath110 ( [ htot ] ) with the strength @xmath111 and matrix elements @xmath112 connecting states of opposite parities and the same spin orientation . using ( [ subspaces ] ) , ( [ dip ] ) and coefficients @xmath113 one gets @xmath114 where @xmath115 , @xmath116 or @xmath117 . however , we will assume a small random misalignment , @xmath118 ( perhaps , due to a random biaxial stress in addition to a strong uniaxial stress @xmath119=@xmath120 ) . it will lead to a slight mixing of the `` left '' and `` right '' states @xmath121 . consider @xmath122@xmath123@xmath119@xmath123@xmath124 ( @xmath125 ) . then the matrix elements of @xmath126 between a pair of exact eigenstates , @xmath127 , that are predominantly `` left '' and `` right '' have the form : @xmath128 . the first term is due to an umklapp process ( @xmath129@xmath4910@xmath130 ) and the second term is due to a small misalignment @xmath131=@xmath132 of @xmath108 and @xmath124 . both terms are small and . as shown below , under the appropriate conditions each of the chiral subspaces can be made nearly decoherence free " . an important application of the above rule is a suppression of the spin - flip transitions between the states with dominant characters @xmath133 that differ only in spin orientation . those states always have opposite chiralities t ; ( b ) calculation of the lifetimes of the first two excited states versus compressive 001 stress at @xmath134=0.1 k , @xmath135 t , and @xmath136 . solid lines @xmath137 and dashed lines @xmath138 . inset shows magnified region of the avoided crossing for the magnetic field @xmath139 ; ( c ) calculation of the lifetimes of the first two excited states versus compressive 001 stress at @xmath134=0.1 k , @xmath135 t , and @xmath140 mev . solid lines @xmath137 and dashed lines @xmath138 . ] we shall label each eigenstates after its dominant character and consider @xmath141 , @xmath142 . if the compressive stress @xmath70 is sufficiently large the zeeman splitting @xmath143 becomes smaller than a stress - induced splitting between @xmath144 and @xmath145 ( see fig . [ fig : energy ] ) . then at low temperature the state @xmath146 decays directly into the state @xmath147 of the opposite chirality . there is a competition between umklapp and dipole - induced processes . for moderate uniaxial stress and magnetic field the dipole contribution to the lifetime @xmath148 dominates @xmath149 here @xmath150 and @xmath151 where @xmath152 are elastic compliance moduli . prefactor in ( [ tzup1 ] ) corresponds to the lifetime @xmath1531 psec of the states @xmath154 that decay within the same subspaces to @xmath144 and @xmath155 . for greater stress and/or magnetic field the umklapp process dominates the decay , giving @xmath156 here @xmath157 is the longitudinal bohr radius , @xmath158 is the transverse sound velocity , and @xmath159 depends on the overlap factor @xmath160 given above via @xmath161 . we calculated transition rates @xmath162 between each pair of the 12 eigenstates @xmath163 by numerical diagonalization of @xmath20 ( [ htot ] ) and employing the electron - phonon interaction ( [ umklapp ] ) . the stress dependence of the lifetimes of the first two excited states of si : li donors at low temperatures is shown in fig . [ fig : energy ] for @xmath137 and @xmath164 . these states can be classified as either orbital ( predominantly @xmath165 ) or spin ( predominantly @xmath166 ) excitations . if the stress - induced splitting ( sis ) is smaller than the zeeman splitting ( zs ) the first and second excited states correspond to the orbital and spin excitations respectively . as soon as sis overcomes zs the two states switch their characters . the orbital and spin excitations have distinctly different stress dependencies of their lifetimes . the lifetime of the orbital exitation is decreasing with the stress increase while the spin excitation behaves quite oppositely . the stress increase leads to a large separation between @xmath165 and @xmath167 states . consequently the probability of @xmath168 phonon decay grows as the phonon density of states or faster . on the contrary , the stress does not affect the zeeman doublet @xmath166@xmath167 but rather separates it from other states of the manifold making the spin excitation @xmath166 more stable . the lifetime decrease of the orbital excitation is more pronounced for an ideal position of the li - donor , i.e. for @xmath169 ( see fig . [ fig : energy](b ) ) and for @xmath164 . in this case the orbital excitation and the ground state have different chiralities and the former may decay through the umklapp process only . this adds additional stress dependence due to the matrix element of the umklapp process . the magnitude and the stress dependence of @xmath165-lifetime shown in fig . [ fig : energy](b ) ( dashed line ) is very similar to our previous result @xcite . the dipole term @xmath131 reduces both the magnitude and the stress dependence of @xmath165-lifetime , which is weakly decreasing with stress near the value of 10@xmath130 s ( see fig . [ fig : energy](c ) ) . we conducted a pulsed esr study of the li donor spin relaxation under stress . an isotopically - enriched @xmath0si crystal ( 800 ppm residual @xmath170si ) was @xmath3li doped by implanting with a variable energy , up to 360 kev , and a total dose 6@xmath17110@xmath172 of @xmath3li/@xmath173 followed by a 30 min anneal at 800 @xmath174c to ensure a homogeneous distribution of the lithium . the estimated li concentration is @xmath49 10@xmath175 @xmath176 . uniform , ( 110 ) , in - palne tensile stress was applied to the si by gluing fused silica slides ( 1 mm thick ) to the two ( 001 ) faces with apiezon n vacuum grease . this `` sandwich '' was then frozen to liquid helium temperatures . the thermal expansion coefficient of fused silica is substantially smaller than that of silicon in the temperature range 4 - 300 k @xcite , and therefore freezing the sandwich results in an uniform in - plane tensile strain , equivalent to the action of a compressive uniaxial stress @xmath70 along the ( 001 ) axis of the silicon crystal ( perpendicular to the faces ) . the spin resonance experiments were done at x - band ( @xmath490.34 t ) . under @xmath70 the esr spectrum is a single line as described by watkins and ham @xcite though our linewidth ( 8 @xmath33 t ) is considerably narrower than in their natural si , since inhomogeneous broadening from the @xmath170si is eliminated in the isotopically enriched @xmath0si @xcite . the electron magnetization decay measured at 2.1 and 4.5 k with a conventional inversion recovery pulse sequence @xcite is shown in fig . [ fig : comparison ] . the stress @xmath70 in the sandwiched samples was estimated by performing fourier transform infrared spectroscopy ( ftir ) with a bruker ifs 66v / s spectrometer ( resolution , 0.14 @xmath66 ) on a similarly prepared stressed samples of li - doped natural float zone silicon ( a hole with a diameter 3 mm in the silica slides allowed fir transmission through the stressed sample ) . the stress - induced splitting 0.202 mev was observed between the @xmath177 line at 21.287 mev and @xmath178 line at 21.489 mev . using this splitting we estimate @xmath179 pa based on the previous measurements with calibrated stresses @xcite and expressions for the stress - induced line shifts @xcite . to explain the magnetization recovery data in fig . [ fig : comparison](b ) we studied the multilevel kinetics , @xmath180 where @xmath181 is a relaxation operator and @xmath182 are individual state populations . we used the stress value @xmath10 pa estimated from the above ftir experiment . the only unknown parameter in the expressions for the decay rates @xmath183 is @xmath131 , or rather @xmath184 . we were able to fit the two experimental time traces at @xmath185 k and @xmath186 k using @xmath187 mev ( @xmath188 ) as shown in fig . [ fig : comparison ] . this value corresponds to a statistical average @xmath189 over all li donors . we note the multi - exponential character of the decay at 4.5 k. the temperature dependence of the long - time magnetization kinetics is determined by that of the smallest eigenvalue of the relaxation operator @xmath190 . the magnetization decay proceeds via thermal activation to rapidly decaying excited states as shown in fig . [ fig : comparison](b ) . for @xmath134 @xmath6 0.1 k the spin life - time @xmath191 saturates at large value consistent with those shown in fig . [ fig : energy ] . to conclude , we introduced a new spin - orbital relaxation mechanism for li donors in si and predicted theoretically that the longitudinal spin relaxation @xmath192 time of a li donor in si under the uniaxial compressive stress and magnetic field aligned along the @xmath193,@xmath194 or @xmath195 axis can be very large ( hours ) for @xmath134 @xmath6 0.3 k despite the inverted level structure of interstitial li donor as compared to group v substitutional donors in silicon . the orbital relaxation time is 0.1 msec at that temperature . by virtue of the chirality selection rules for the electron - phonon interaction , the donor electron spin can decay only via inter - valley umklapp processes or a very weak direct phonon process caused by a small li off - site displacement misalignment . at higher temperatures @xmath192 is reduced to a @xmath33s range due to thermal population of the lower excited states which in turn promotes fast transitions within the same chiral subspace allowed by the parity - selection rules . we obtained a very close ( within few per cent ) agreement with pulsed esr measurements of @xmath192 times at @xmath196 k ( 12 @xmath33s ) and at @xmath197 k ( 1.5 @xmath33s ) that corroborates our prediction of extremely long - lived spin states of li donors in si at lower temperatures . since the chiralty is a universal property of the silicon host ur study can be applied to substitutional donors in si with similar near - degeneracy of the 1@xmath71 coupled to the ground state via stray electric fields and umklapp processes . our results of the very long li spin lifetime can also stimulate the analysis of the possibility of li spin - based qip applications with direct long - range elastic - dipole interaction between the li donor spins similar to that considered in @xcite . this research was supported by us nsa , nasa grant nnx07al35a , onr grant n00014 - 06 - 1 - 0616 , and doe contract no . de - ac02 - 05ch11231 . 99 d. loss , d.p . divincenzo , phys . a * 57 * , 120 ( 1998 ) . kane , nature * 393 * , 133 ( 1998 ) . r. vrijen , et al . a * 62 * , 012306 ( 2000 ) . m. friesen , et al . b * 67 * , 121301 ( 2003 ) . j.r . petta , et al . , science * 309 * , 2180 ( 1985 ) . gordon , k.d . bowers , phys . * 1 * , 368 ( 1958 ) . a.m. tyryshkin , et al . b * 68 * , 193207 ( 2003 ) . r.l . aggarwal , et al . , phys . rev . * 138 * , a 882 ( 1965 ) . c. jagannath , a. ramdas , phys . b * 23 * , 4426 ( 1981 ) . smelyanskiy , a.g . petukhov and v.v . osipov , phys . b * 72 * , r081304 ( 2005 ) . b. golding and m.i . dykman , cond - mat/0309147 . watkins , f.s . ham , phys . b * 1 * , 4071 ( 1970 ) . g. feher , phys . rev . * 114 * , 1219 ( 1959 ) . t.g . castner , phys . rev . * 139 * , 58 ( 1963 ) ; phys . rev . lett . * 8 * , 13 ( 1962 ) ; phys . rev . * 155 * , 816 ( 1967 ) . h. hasegawa , phys . rev . * 118 * , 1523 ( 1960 ) . roth , phys . rev . * 118 * , 1534 ( 1960 ) . a. yang , et al . , physica b * 376 - 377 * , 54 ( 2006 ) . d. karaiskaj , et al . lett . * 90 * , 186402 ( 2003 ) . li line is @xmath49100 times stronger than li@xmath198o line . a. nyilas , aip conf . proc . * 711 * , 151 ( 2004 ) . a. schweiger and g. jeschke , _ principles of pulse electron paramagnetic resonance _ ( oxford university press , oxford , 2001 ) .	we report theoretical and experimental studies of the longitudinal electron spin and orbital relaxation time of interstitial li donors in @xmath0si . we predict that despite the near - degeneracy of the ground - state manifold the spin relaxation times are extremely long for the temperatures below 0.3 k. this prediction is based on a new finding of the chiral symmetry of the donor states , which presists in the presence of random strains and magnetic fields parallel to one of the cubic axes . experimentally observed kinetics of magnetization reversal at 2.1 k and 4.5 k are in a very close agreement with the theory . to explain these kinetics we introduced a new mechanism of spin decoherence based on a combination of a small off - site displacement of the li atom and an umklapp phonon process . both these factors weakly break chiral symmetry and enable the long - term spin relaxation .
the concept of _ entanglement _ has played a crucial role in the development of quantum physics . in the early days entanglement was mainly perceived as the qualitative feature of quantum theory that most strikingly distinguishes it from our classical intuition . the subsequent development of bell s inequalities has made this distinction quantitative , and therefore rendered the non - local features of quantum theory accessible to experimental verification @xcite . bell s inequalities may indeed be viewed as an early attempt to quantify the quantum correlations that are responsible for the counterintuitive features of quantum mechanically entangled states . at the time it was almost unimaginable that such quantum correlations could be created in well controlled environments between distinct quantum systems . however , the technological progress of the last few decades means that we are now able to coherently prepare , manipulate , and measure individual quantum systems , as well as create controllable quantum correlations . in parallel with these developments , quantum correlations have come to be recognized as a novel resource that may be used to perform tasks that are either impossible or very inefficient in the classical realm . these developments have provided the seed for the development of modern quantum information science . given the new found status of entanglement as a resource it is quite natural and important to discover the mathematical structures underlying its theoretical description . we will see that such a description aims to provide answers to three questions about entanglement , namely ( 1 ) its characterisation , ( 2 ) its manipulation and , ( 3 ) its quantification . in the following we aim to provide a tutorial overview summarizing results that have been obtained in connection with these three questions . we will place particular emphasis on developments concerning the _ quantification _ of entanglement , which is essentially the theory of _ entanglement measures_. we will discuss the motivation for studying entanglement measures , and present their implications for the study of quantum information science . we present the basic principles underlying the theory and main results including many useful entanglement monotones and measures as well as explicit useful formulae . we do not , however , present detailed technical derivations . the majority of our review will be concerned with entanglement in bipartite systems with finite and infinite dimensional constituents , for which the most complete understanding has been obtained so far . the multi - party setting will be discussed in less detail as our understanding of this area is still far from satisfactory . it is our hope that this work will give the reader a good first impression of the subject , and will enable them to tackle the extensive literature on this topic . we have endeavoured to be as comprehensive as possible in both covering known results and also in providing extensive references . of course , as in any such work , it is inevitable that we will have made several oversights in this process , and so we encourage the interested reader to study various other interesting review articles ( e.g. @xcite ) and of course the original literature . * _ what is entanglement ? _ * any study of entanglement measures must begin with a discussion of what entanglement _ is _ , and how we actually _ use _ it . in the following we will adopt a highly operational point of view . then the usefulness of entanglement emerges because it allows us to overcome a particular constraint that we will call the _ locc constraint _ - a term that we will shortly explain . this restriction has both technological and fundamental motivations , and arises naturally in many explicit physical settings involving quantum communication across a distance . we will consider these motivations in some detail , starting with the technological ones . in any quantum communication experiment we would like to be able to distribute quantum particles across distantly separated laboratories . perfect quantum communication is essentially equivalent to perfect entanglement distribution . if we can transport a qubit without any decoherence , then any entanglement shared by that qubit will also be distributed perfectly . conversely , if we can distribute entangled states perfectly then with a small amount of classical communication we may use teleportation @xcite to perfectly transmit quantum states . however , in any forseeable experiment involving these processes , the effects of noise will inevitably impair our ability to send quantum states over long distances . one way of trying to overcome this problem is to distribute quantum states by using the noisy quantum channels that are available , but then to try and combat the effects of this noise using higher quality local quantum processes in the distantly separated labs . such local quantum operations ( ` lo ' ) will be much closer to ideal , as they can be performed in well - controlled environments without the decoherence induced by communication over long - distances . however , there is no reason to make the operations of separated labs totally independent . classical communication ( ` cc ' ) can essentially be performed perfectly using standard telecom technologies , and so we may also use such communication to coordinate the quantum actions of the different labs ( see fig . [ fig2 ] ) . it turns out that the ability to perform classical communication is vital for many quantum information protocols - a prominent example being teleportation . these considerations are the technological reasons for the key status of the _ local operations and classical communication _ ` locc ' paradigm , and are a major motivation for their study . in a standard quantum communication setting two parties alice and bob may perform any generalized measurement that is localized to their laboratory and communicate classically . the brick wall indicates that no quantum particles may be exchanged coherently between alice and bob . this set of operations is generally referred to as locc.,width=302 ] however , for the purposes of this article , the fundamental motivations of the locc paradigm are perhaps more important than these technological considerations . we have loosely described entanglement as the _ quantum correlations _ that can occur in many - party quantum states . this leads to the question - how do we define quantum correlations , and what differentiates them from _ classical correlations _ ? the distinction between ` quantum ' effects and ` classical ' effects is frequently a cause of heated debate . however , in the context of quantum information a precise way to define classical correlations is via locc operations . classical correlations can be defined as those that can be generated by locc operations . if we observe a quantum system and find correlations that can not be simulated classically , then we usually attribute them to quantum effects , and hence label them _ quantum correlations _ so suppose that we have a noisy quantum state , and we process it using locc operations . if in this process we obtain a state that can be used for some task that can not be simulated by classical correlations , such as violating a bell inequality , then we must not attribute these effects to the locc processing that we have performed , but to quantum correlations that were _ already present _ in the initial state , even if the initial state was quite noisy . this is an extremely important point that is at the heart of the study of entanglement . it is the constraint to locc - operations that elevates entanglement to the status of a resource . using locc - operations as the only other tool , the inherent quantum correlations of entanglement are required to implement general , and therefore nonlocal , quantum operations on two or more parties @xcite . as locc - operations alone are insufficient to achieve these transformations , we conclude that entanglement may be defined as the sort of correlations that may not be created by locc alone . allowing classical communication in the set of locc operations means that they are not completely local , and can actually have quite a complicated structure . in order to understand this structure more fully , we must first take a closer look at the notion of general quantum operations and their formal description . * _ quantum operations _ * in quantum information science much use is made of so - called ` generalised measurements ' ( see @xcite for a more detailed account of the following basic principles ) . it should be emphasized that such generalised measurements do not go beyond standard quantum mechanics . in the usual approach to quantum evolution , a system is evolved according to unitary operators , or through collapse caused by projective measurements . however , one may consider a more general setting where a system evolves through interactions with other quantum particles in a sequence of three steps : ( 1 ) first we first add ancilla particles , ( 2 ) then we perform joint unitaries and measurements on both the system and ancillae , and finally ( 3 ) we discard some particles on the basis of the measurement outcomes . if the ancillae used in this process are originally uncorrelated with the system , then the evolution can be described by so - called _ kraus operators_. if one retains total knowledge of the outcomes obtained during any measurements , then the state corresponding to measurement outcomes @xmath0 occurs with probability @xmath1 and is given by @xmath2 where @xmath3 is the initial state and the @xmath4 are matrices known as _ operators ( see part ( a ) of fig . [ fig1 ] for illustration ) . the normalisation of probabilities implies that kraus operators must satisfy @xmath5 . in some situations , for example when a system is interacting with an environment , all or part of the measurement outcomes might not be accessible . in the most extreme case this corresponds to the situation where the ancilla particles are being traced out . then the map is given by @xmath6 which is illustrated in part ( b ) of fig . ( [ fig2 ] ) . schematic picture of the action of quantum operations with and without sub - selection ( eqs . ( [ eq1 ] ) and ( [ eq2 ] ) respectively ) shown in part ( a ) and part ( b ) respectively . , width=283 ] such a map is often referred to as a _ trace preserving _ quantum operation , whereas operations in which measurement outcomes are retained are sometimes referred to as _ measuring _ quantum operations ( or sometimes also _ selective _ quantum operations , or _ stochastic _ quantum operations , depending upon the context ) . conversely , it can be shown ( see e.g. @xcite ) that for _ any _ set of linear operators @xmath7 satisfying @xmath8 we can find a process , composed of the addition of ancillae , joint unitary evolution , and von - neumann measurements , that leads to eq . ( [ eq1 ] ) . in trace preserving operations the @xmath7 should strictly all be matrices of the same dimensions , however , if knowledge of outcomes is retained , then different @xmath7 may have different dimensions . having summarized the basic ingredients of generalised quantum operations , we are in a position to consider approaches that may be taken to determine which operations are implementable by locc . the locc constraint is illustrated in figure [ fig2 ] . in general this set of operations is quite complicated . alice and bob may communicate classically before or after any given round of local actions , and hence in any given round their actions may depend upon the outcomes of previous measuring operations . as a consequence of this complexity , there is no known simple characterisation of the locc operations . this has motivated the development of larger classes of operations that can be more easily characterised , while still retaining a considerable element of locc - ality . one of the most important such classes is the set of _ separable operations_. these are the operations that can be written in terms of kraus operators with a _ product _ decomposition : @xmath9 such that @xmath10 . clearly , any locc operation can be cast in the form of separable operation , as the local kraus operators corresponding to the individual actions of alice and bob can be joined into product kraus operators . however , it is remarkable that the converse is _ not _ true . this was first demonstrated in @xcite , where an example task of a separable operation is presented that can not be implemented using locc actions - the example presented there requires a finite amount of quantum communication to implement it , even though the operation is itself separable . it is nevertheless convenient from a mathematical point of view to work with separable operations , as optimising a given task using separable operations provides strong bounds on what may be achieved using locc . sometimes this process can even lead to tight results - one may try to show whether the optimal separable operation may in fact be also implemented using locc , and this can often , but not always , be guaranteed in the presence of symmetries ( see e.g. @xcite and refs . therein ) . even more general classes of operations such as positive partial transpose preserving operations ( ppt ) such that @xmath11 is also completely positive , where @xmath12 corresponds to transposition of _ all _ of bob s particles , including ancillas . one can also consider transposition only of those particles belonging to bob that undergo the operation @xmath13 . however , we believe that this does not affect the definition . it is also irrelevant whether the transposition is taken over alice or bob , and so one may simply assert that @xmath14 must be completely positive , where @xmath15 is the transposition of one party . it can be shown that the ppt operations are precisely those operations that preserve the set of ppt states . hence the set of non - ppt operations includes any operation that creates a free ( non - bound ) entangled state out of one that is ppt . hence ppt operations correspond to some notion of locality , and in contrast to separable operations it is relatively easy to check whether a quantum operation is ppt @xcite . ] may also be used in the study of entanglement as they have the advantage of a very compact mathematical characterization @xcite . after this initial discussion of quantum operations and the locc constraint we are now in a position to consider in more detail the basic properties of entanglement . * _ basic properties of entanglement _ * following our discussion of quantum operations and their natural constraint to local operations and classical communication , we are now in a position to establish some basic facts and definitions regarding entangled states . given the wide range of tasks that exploit entanglement one might try to define entanglement as ` that property which is exploited in such protocols ' . however , there is a whole range of such tasks , with a whole range of possible measures of success . this means that situations will almost certainly arise where a state @xmath16 is better than another state @xmath17 for achieving one task , but for achieving a different task @xmath17 is better than @xmath16 . consequently using a task - based approach to quantifying entanglement will certainly not lead to a single unified perspective . however , despite this problem , it is possible to assert some general statements which are valid regardless of what your favourite use of entanglement is , as long as the key set of ` allowed ' operations is the locc class . this will serve us a guide as to how to approach the quantification of entanglement , and so we will discuss some of these statements in detail : @xmath18 _ separable states contain no entanglement . _ a state @xmath19 of many parties @xmath20 is said to be _ separable _ @xcite , if it can be written in the form @xmath21 where @xmath22 is a probability distribution . these states can trivially be created by locc - alice samples from the distribution @xmath22 , informs all other parties of the outcome @xmath0 , and then each party @xmath23 locally creates @xmath24 and discards the information about the outcome @xmath0 . as these states can be created from scratch by locc they trivially satisfy a local hidden variables model and all their correlations can be described classically . hence , it is quite reasonable to state that separable states contain no entanglement . @xmath18 _ all non - separable states allow some tasks to be achieved better than by locc alone , hence all non - separable states are entangled . _ for a long time the quantum information community has used a ` negative ' characterization of the term entanglement essentially defining entangled states as those that can not be created by locc alone . on the other hand , it can be shown that a quantum state @xmath25 may be generated perfectly using locc if and only if it is separable . of course this is a task that becomes trivially possible by locc when the state @xmath25 has been provided as a non - local resource in the first place . more interestingly , it has been shown recently that for any non - separable state @xmath25 , one can find another state @xmath26 whose teleportation fidelity may be enhanced if @xmath25 is also present @xcite . this is interesting as it allows us to positively characterize non - separable states as those possessing a useful resource that is not present in separable states . this hence justifies the synonymous use of the terms _ non - separable _ and _ entangled_. @xmath18 _ the entanglement of states does not increase under locc transformations . _ given that by locc we can only create separable , ie non - entangled states , this immediately implies the statement that locc can not create entanglement from an unentangled state . indeed , we even have the following stronger fact . suppose that we know that a quantum state @xmath25 can be transformed with certainty to another quantum state @xmath26 using locc operations . then anything that we can do with @xmath26 and locc operations we can also achieve with @xmath25 and locc operations . hence the utility of quantum states can not increase under locc operations @xcite , and one can rightfully state that @xmath25 is at least as entangled as @xmath26 . @xmath18 _ entanglement does not change under local unitary operations . _ this property follows from the previous one because local unitaries can be inverted by local unitaries . hence , by the non - increase of entanglement under locc , two states related by local unitaries have an equal amount of entanglement . @xmath18 _ there are maximally entangled states . _ now we have a notion of which states are entangled and are also able , in some cases , to assert that one state is more entangled than another . this naturally raises the question whether there is a _ maximally entangled state _ , i.e. one that is more entangled than all others . indeed , at least in two - party systems consisting of two fixed @xmath27-dimensional sub - systems ( sometimes called qudits ) , such states exist . it turns out that any pure state that is local unitarily equivalent to @xmath28 is maximally entangled . this is well justified , because as we shall see in the next subsection , any pure or mixed state of two @xmath27-dimensional systems can be prepared from such states with certainty using only locc operations . we shall later also see that the non - existence of an equivalent statement in multi - particle systems is one of the reasons for the difficulty in establishing a theory of multi - particle entanglement . + the above considerations have given us the extremes of entanglement - as long as we consider locc as our set of available operations , separable states contain zero entanglement , and we can identify certain states that have maximal entanglement . they also suggest that we can impose some form of ordering - we may say that state @xmath25 is more entangled than a state @xmath26 if we can perform the transformation @xmath29 using locc operations . a key question is whether this method of ordering gives a partial or total order ? to answer this question we must try and find out when one quantum state may be transformed to another using locc operations . before we move on to the discussion of entanglement measures we will consider this question in more detail in the next part . note that the notion that ` _ entanglement does not increase under locc _ ' is implicitly related to our restriction of quantum operations to locc operations - if other restrictions apply , weaker or stronger , then our notion of ` more entangled ' is likely to also change . _ * manipulation of single bi - partite states * _ in the previous section we indicated that for bi - partite systems there is a notion of maximally entangled states that is independent of the specific quantification of entanglement . this is so because there are so - called _ maximally entangled states _ from which all others can be created by locc only ( at least for bipartite systems of fixed maximal dimension ) . we we will show this explicitly here for the case of two qubits and leave the generalization as an exercise to the reader . in the case of two qubits , we will see that the maximally entangled states are those that are local - unitarily equivalent to the state @xmath30 our aim is now to justify this statement by showing that for any bipartite pure state written in a schmidt decomposed form ( see discussion around equation ( [ schmidt ] ) for an explanation of the schmidt decomposition ) : @xmath31 we can find a locc map that takes @xmath32 to @xmath33 with certainty . to this end we simply need to write down the kraus operators ( see eq . ( [ eq1 ] ) of a valid quantum operation . it is easy to show that the kraus operators defined by @xmath34 satisfy @xmath35 and @xmath36 , so that @xmath37 . it is instructive to see how one can construct this operation physically using only locc transformations . let us first add an ancilla in state @xmath38 to alice which results in the state @xmath39 if we then perform the local unitary operation @xmath40 on alice s two particles , we arrive at @xmath41 finally , a local measurement on alice s ancilla particle now yields two outcomes . if alice finds @xmath38 then bob is informed and does not need to carry out any further operation ; if alice finds @xmath42 then bob needs to apply a @xmath43 operation to his particle . in both cases this results in the desired state @xmath44 . given that we can obtain with certainty any arbitrary pure state starting from @xmath32 , we can also obtain any mixed state @xmath25 . this is because any mixed state @xmath25 can always be written in terms of its eigenvectors as @xmath45 for some set of unitaries @xmath46 and @xmath47 ( this in turn is simply a consequence of the schmidt decomposition ) . it is an easy exercise , left to the reader , to construct the operation that takes @xmath32 to @xmath25 . a natural generalisation of this observation would be to consider locc transformations between general pure states of two parties @xcite . although this question is a little more difficult , a complete solution has been developed using the mathematical framework of the theory of _ majorization_. the results that have been obtained not only provide necessary and sufficient conditions for the possibility of the locc interconversion between two pure states , they are also constructive as they lead to explicit protocols that achieve the task @xcite . these conditions may be expressed most naturally in terms of the _ schmidt coefficients _ @xcite of the states involved . it is a useful fact that any bi - partite pure quantum state @xmath48 may be written in the form @xmath49 where the positive real numbers @xmath50 are the _ schmidt - coefficients _ of the state @xmath48 .. the amplitudes @xmath51 can be considered as the matrix elements of a matrix @xmath52 . this matrix hence completely represents the state ( as long as a local basis is specified ) . if we perform the local unitary transformation @xmath53 then the matrix @xmath52 gets transformed as @xmath54 . it is a well established result of matrix analysis - the _ singular value decomposition _ @xcite - that any matrix @xmath52 can be _ diagonalised _ into the form @xmath55 by a suitable choice of ( @xmath56 ) , even if @xmath52 is not square . the coefficients @xmath57 are the so - called _ singular values _ of @xmath52 , and correspond to the schmidt coefficients . ] the local unitaries do not affect the entanglement properties , which is why we now write the initial state vector @xmath58 and final state vector @xmath59 in their schmidt - bases , @xmath60 where @xmath61 denotes the dimension of each of the quantum systems . we can take the schmidt coefficients to be given in decreasing order , i.e. , @xmath62 and @xmath63 . the question of the interconvertibility between the states can then be decided from the knowledge of the real schmidt coefficients only , as any two pure states with the same schmidt coefficients may be interconverted straightforwardly by local unitary operations . in @xcite it has been shown that a locc transformation converting @xmath64 to @xmath65 with unit probability exists if and only if the @xmath66 are _ majorized _ by @xmath67 , i.e. if for all @xmath68 we have that @xmath69 and @xmath70 , where @xmath61 denotes the number of nonzero schmidt - coefficients @xcite . various refinements of this result have been found that provide the largest success probabilities for the interconversion between two states by locc , together with the optimal protocol ( according to certain figures of merit ) where such a deterministic interconversion is not possible @xcite . these results allow us in principle to decide any question concerning the locc interconversion of pure states by employing techniques from linear programming @xcite . it is a direct consequence of the above structures that there are _ incomparable _ states , i.e. pairs of states such that neither can be converted into the other with certainty . these states are called incomparable as neither can be viewed as more entangled than the other . note that borrowed entanglement can make some pairs of incomparable states comparable again . indeed , there are known examples where the locc transformation of @xmath71 is not possible with probability one , but where given a suitable entangled state @xmath72 the locc transformation of @xmath73 is possible with certainty @xcite . this phenomenon is now called _ entanglement catalysis _ , as the state @xmath72 is returned unchanged after the transformation , and acts much like a catalyst . the majorization condition also reveals another disadvantageous feature of the single copy setting - there can be _ discontinuities_. for instance , it can be shown that the maximal probability of success for the locc transformation from @xmath74 to @xmath75 is unity , while the probability for the transformation @xmath74 to @xmath76 , i.e. even if the target states in the two examples are arbitrarily close . that the probability of success for the later transformation is zero can also be concluded easily from the fact that the schmidt - number , i.e. the number of non - vanishing schmidt - coefficients , can not be increased in an locc protocol , even probabilistically . the key problem is that we are being too restrictive in asking for _ exact _ state transformations . physically , we should be perfectly happy if we can come very close to a desired state . indeed , admitting a small but finite value of @xmath77 there will be a finite probability to achieve the transformation . this removes the above discontinuity @xcite , but the success probability will now depend on the size of the imprecision that we allow . the following subsection will serve to overcome this problem for pure states by presenting a natural definition of interconvertibility in the presence of vanishing imprecisions , a definition that will constitute our first entanglement measure . _ * state manipulation in the asymptotic limit * _ the study of the locc transformation of pure states has so far enabled us to justify the concept of maximally entangled states and has also permitted us , in some cases , to assert that one state is more entangled than another . however , we know that exact locc transformations can only induce a partial order on the set of quantum states . the situation is even more complex for _ mixed _ states , where even the question of when it is possible to locc transform one state into another is a difficult problem with no transparent solution at the time of writing . all this means that if we want to give a definite answer as to whether one state is more entangled than another for any pair of states , it will be necessary to consider a more general setting . in this context a very natural way to compare and quantify entanglement is to study locc transformations of states in the so called _ asymptotic regime_. instead of asking whether for a single pair of particles the initial state @xmath25 may be transformed to a final state @xmath26 by locc operations , we may ask whether for some large integers @xmath78 we can implement the ` wholesale ' transformation @xmath79 . the largest ratio @xmath80 for which one may achieve this would then indicate the relative entanglement content of these two states . considering the many - copy setting allows each party to perform collective operations on ( their shares of ) many copies of the states in question . such a many copy regime provides many more degrees of freedom , and in fact paves part of the way to a full classification of pure entangled states . to pave the rest of the route we will also need to discuss what kind of approximations we might admit for the output of the transformations . there are two basic approaches to this problem - we can consider either _ exact _ or _ asymptotically exact _ transformations . the distinction between these two approaches is important , as they lead to different scenarios that yield different answers . in the _ exact _ regime we allow no errors at all - we must determine whether the transformation @xmath79 can be achieved perfectly and with @xmath81 success probability for a given value of @xmath82 and @xmath61 . the supremum of all achievable rates @xmath83 is denoted by @xmath84 , and carries significance as a measure of the exact locc ` exchange rate ' between states @xmath85 . this quantity may be explored and gives some interesting results @xcite . however , from a physical point of view one may feel that the restriction to exact transformations is too stringent . after all , it should be quite acceptable to consider approximate transformations @xcite that become arbitrarily precise when going to the asymptotic limit . asymptotically vanishing imperfections , as quantified by the trace norm ( i.e. tr@xmath86 ) , will lead to vanishingly small changes in measurements of bounded observables on the output . this leads to the second approach to approximate state transformations , namely that of _ asymptotically exact _ state transformations , and this is the setting that we will consider for the remainder of this work . in this setting we consider imperfect transformations between large blocks of states , such that in the limit of large block sizes the imperfections vanish . for example , for a large number @xmath61 of copies of @xmath25 , one transforms @xmath87 to an output state @xmath88 that approximates @xmath89 very well for some large @xmath82 . if , in the limit of @xmath90 and for fixed @xmath83 , the approximation of @xmath89 by @xmath88 becomes arbitrarily good , then the rate @xmath91 is said to be _ achievable_. one can use the optimal ( supremal ) achievable rate @xmath92 as a measure of the relative entanglement content of @xmath85 in the asymptotic setting . this situation is reminiscent of shannon compression in classical information theory - where the compression process loses all imperfections in the limit of infinite block sizes as long as the compression rate is below a threshold @xcite . clearly the asymptotically exact regime is less strongly constrained than the exact regime , so that @xmath93 . given that we are considering two limiting processes it is not clear whether the two quantities are actually equal and it can be rigorously demonstrated that they are different in general , see e.g. @xcite . such an asymptotic approach will alleviate some of the problems that we encountered in the previous section . it turns out that the asymptotic setting yields a unique total order on bi - partite pure states , and as a consequence , leads to a very natural measure of entanglement that is essentially unique . to this end let us start by defining our first entanglement measure , which happens also to be one of the most important measures - the _ entanglement cost _ , @xmath94 . for a given state @xmath25 this measure quantifies the maximal possible rate @xmath91 at which one can convert blocks of _ 2-qubit _ maximally entangled states @xcite into output states that approximate many copies of @xmath25 , such that the approximations become vanishingly small in the limit of large block sizes . if we denote a general trace preserving locc operation by @xmath95 , and write @xmath96 for the density operator corresponding to the maximally entangled state vector in @xmath97 dimensions , i.e.@xmath98 , then the entanglement cost may be defined as @xmath99 = 0 \right\}\ ] ] where d@xmath100 is a suitable measure of distance @xcite . a variety of possible distance measures may be considered . it has been shown that the definition of entanglement cost is independent of the choice of distance function , as long as these functions are equivalent to the trace norm in a way that is sufficiently independent of dimension ( see @xcite for further explanation ) . hence we will fix the trace norm distance , i.e. @xmath101 , as our canonical choice of distance function . it may trouble the reader that in the definition of @xmath102 we have not actually taken input states that are blocks of @xmath103 copies of 2-qubit maximally entangled states , but instead have chosen as inputs single maximally entangled states between subsystems of increasing dimensions @xmath104 . however , these two approaches are equivalent because ( for integral @xmath103 ) @xmath105 is local unitarily equivalent to @xmath106 . the entanglement cost is an important measure because it quantifies a wholesale ` exchange rate ' for converting maximally entangled states to @xmath25 by locc alone . maximally entangled states are in essence the ` gold standard currency ' with which one would like to compare all quantum states . although computing @xmath102 is extremely difficult , we will later discuss its important implications for the study of channel capacities , in particular via another important and closely related entanglement measure known as the _ entanglement of formation _ , @xmath107 . just as @xmath102 measures how many maximally entangled states are required to create copies of @xmath25 by locc alone , we can ask about the reverse process : at what rate may we obtain maximally entangled states ( of two qubits ) from an input supply of states of the form @xmath25 . this process is known in the literature either as _ entanglement distillation _ , or as _ entanglement concentration _ ( usually reserved for the pure state case ) . the efficiency with which we can achieve this process defines another important basic asymptotic entanglement measure which is the _ distillable entanglement _ , @xmath108 . again we allow the output of the procedure to _ approximate _ many copies of a maximally entangled state , as the exact transformation from @xmath109 to even one pure maximally entangled state is in general impossible @xcite . in analogy to the definition of @xmath102 , we can then make the precise mathematical definition of @xmath108 as @xmath110 = 0 \right\}.\ ] ] @xmath108 is an important measure because if entanglement is used in a two party quantum information protocol , then it is usually required in the form of maximally entangled states . so @xmath108 tells us the rate at which noisy mixed states may be converted back into the ` gold standard ' singlet state by locc . in defining @xmath108 we have ignored a couple of important issues . firstly , our locc protocols are always taken to be trace preserving . however , one could conceivably allow probabilistic protocols that have varying degrees of success depending upon various measurement outcomes . fortunately , a thorough analysis by rains @xcite shows that taking into account a wide diversity of possible success measures still leads to the same notion of distillable entanglement . secondly , we have always used 2-qubit maximally entangled states as our ` gold standard ' . if we use other entangled _ pure _ states , perhaps even on higher dimensional hilbert spaces , do we arrive at significantly altered definitions ? we will very shortly see that this is not the case so there is no loss of generality in taking singlet states as our target . given these two entanglement measures it is natural to ask whether @xmath111 , i.e. whether entanglement transformations become _ reversible _ in the asymptotic limit . this is indeed the case for pure state transformations where @xmath108 and @xmath102 are identical and equal to the _ entropy of entanglement _ @xcite . the entropy of entanglement for a pure state @xmath48 is defined as @xmath112 where @xmath113 denotes the von - neumann entropy @xmath114 $ ] , and @xmath115 denotes the partial trace over subsystem b. this reversibility means that in the asymptotic regime we may immediately write down the optimal rate of transformation between _ any _ two pure states @xmath58 and @xmath59 . given a large number @xmath116 of copies of @xmath117 , we can first distill @xmath118 singlet states and then create from those singlets @xmath119 copies of @xmath120 . in the infinite limit these approximations become exact , and as a consequence @xmath121 is the optimal asymptotic conversion rate from @xmath122 to @xmath123 . it is the reversibility of pure state transformations that enables us to define @xmath108 and @xmath102 in terms of transformations to or from singlet states - the use of any other entangled pure state ( in any other dimensions ) simply leads to a constant factor multiplied in front of these quantities . following these basic considerations we are now in a position to formulate a more rigorous and axiomatic approach to entanglement measures that captures the lessons that have been learned in the previous sections . in the final section we will then review several entanglement measures , presenting useful formulae and results and discuss the significance of these measures for various topics in quantum information . in the previous section we considered the quantification of entanglement from the perspective of locc transformations in the asymptotic limit . this approach is interesting because it can be solved completely for pure states . it demonstrates that locc entanglement manipulation is reversible in this setting , therefore imposing a unique order on pure entangled states via the entropy of entanglement . however , for mixed states and locc operations the situation is more complicated and reversibility is lost @xcite . the concomitant loss of a total ordering of quantum states ( in terms of rates of locc entanglement interconversions ) implies that in general an locc based classification of entanglement would be extremely complicated . however , one can try to salvage the situation by taking a more axiomatic approach . one can _ define _ real valued functions that satisfy the basic properties of entanglement that we outlined earlier , and use these functions to attempt to _ quantify _ the amount of entanglement in a given quantum state . this is essentially the process that is followed in the definition of most entanglement measures . various such quantities have been proposed over the years , such as the entanglement of distillation @xcite , the entanglement cost @xcite , the relative entropy of entanglement @xcite and the squashed entanglement @xcite . some of these measures have direct physical significance , whereas others are purely axiomatic . initially these measures were used to give a physically motivated classification of entanglement that is simple to understand , and can even be used to assess the quality of entangled states produced in experiments . however , subsequently they have also been developed into powerful mathematical tools , with great significance for open questions such as the additivity of quantum channel capacities @xcite , quantifying quantum correlations in quantum - many - body systems @xcite , and bounding quantum computing fault tolerance thresholds @xcite to name a few . in this section we will now discuss and present a few basic axioms that any measure of entanglement should satisfy . this will allow us to define further quantities that go beyond the two important mixed state measures ( @xmath102 and @xmath124 ) that we have already introduced . so what exactly are the properties that a good entanglement measure should possess ? an entanglement measure is a mathematical quantity that should capture the essential features that we associate with entanglement , and ideally should be related to some operational procedure . depending upon your aims , this can lead to a variety of possible desirable properties . the following is a list of possible postulates for entanglement measures , some of which are not satisfied by all proposed quantities @xcite : 1 . a _ bipartite _ entanglement measure @xmath125 is a mapping from density matrices into positive real numbers : @xmath126 defined for states of arbitrary bipartite systems . a normalisation factor is also usually included such that the maximally entangled state @xmath127 of two qudits has @xmath128 . @xmath129 if the state @xmath25 is separable . @xmath130 does not increase on average under locc , i.e. , @xmath131 where the @xmath7 are the kraus operators describing some locc protocol and the probability of obtaining outcome @xmath0 is given by @xmath132 ( see fig [ fig1 ] ) . 4 . for pure state @xmath133 the measure reduces to the entropy of entanglement @xmath134 we will call any function @xmath130 satisfying the first three conditions an _ entanglement monotone_. the term _ entanglement measure _ will be used for any quantity that satisfies axioms 1,2 and 4 , and also does not increase under _ deterministic _ locc transformations . in the literature these terms are often used interchangeably . note that the conditions ( 1)-(4 ) may be replaced by an equivalent set of slightly more abstract conditions which will be explained below eq . ( [ continuity20 ] ) . frequently , some authors also impose additional requirements for entanglement measures : @xmath18 _ convexity _ one common example for an additional property required from an entanglement measure is the concept of convexity which means that we require @xmath135 requiring this mathematically very convenient property is sometimes justified as capturing the notion of the loss of information , i.e. describing the process of going from a selection of identifiable states @xmath136 that appear with rates @xmath22 to a mixture of these states of the form @xmath137 . we would like to stress , however , that some care has to be taken in this context . indeed , in the first situation , when the states are locally identifiable , the whole ensemble can be described by the quantum state @xmath138 where the @xmath139 denote some orthonormal basis for a particle belonging to one of the two parties . clearly a measurement of the marker particle @xmath140 reveals the identity of the state of parties @xmath52 and @xmath141 . losing the association between @xmath142 and state @xmath143 then correctly describes the process of the forgetting , a process which is then described by tracing out the marker particle @xmath140 to obtain @xmath137 @xcite . as this is a local operation we may then require that @xmath144 , which is , of course , already required by condition 3 above . hence there is no strict need to require convexity , except for the mathematical simplicity that it might bring . a compelling example of the technical simplicity that convexity can bring is the very simple test for entanglement monotonicity of a convex function @xmath145 . indeed , a convex function @xmath145 does not increase under locc if and only if it satisfies ( i ) @xmath146 for all local unitary @xmath147 and ( ii ) @xmath148 and @xmath149 form local , orthogonal bases @xcite . @xmath18 _ additivity _ given an entanglement measure and a state @xmath26 one may ask for the condition @xmath150 to be satisfied for all integer @xmath61 . a measure satisfying this property is said to be _ additive_. unfortunately , there are some significant entanglement measures that do not satisfy this condition , and for this reason we have not included additivity as a basic postulate . however , given any measure @xmath130 that is not additive there is a straightforward way of removing this deficiency . we may define the _ regularized _ , or _ asymptotic _ version : @xmath151 which is a measure that then automatically satisfies additivity . a much stronger requirement could be to demand _ full additivity _ , by which we mean that for any pair of states @xmath26 and @xmath25 we have @xmath152 . this is a very strong requirement and in fact it may be too strong to be satisfied by all quantities that otherwise satisfy the four basic properties stated above . indeed , even such basic measures as the distillable entanglement may not satisfy this property @xcite . for these reason we have not included the full additivity in our set of properties . however , additivity can be a very useful mathematical property , and we will discuss it further in the context of specific measures . @xmath18 _ continuity _ conditions ( 1 - 3 ) listed above seem quite natural - the first two conditions are little more than setting the scale , and the third condition is a generalisation of the idea that entanglement can only decrease under locc operations to incorporate probabilistic transformations . the fourth condition appears considerably stronger and perhaps arbitrary at first sight . however , it turns out that it is also quite a natural condition to impose . in fact we know that @xmath153 represents the reversible rate of conversion between pure states in the asymptotic regime which strongly suggests that it is the appropriate measure of entanglement for pure states . this is reinforced by the following nontrivial observation : it turns out that _ any _ entanglement monotone that is ( a ) additive on pure states , and ( b ) `` sufficiently continuous '' must _ equal _ @xmath153 on all pure states . before we see what sufficiently continuous means we present a very rough argument for this statement . we know from the asymptotic pure state distillation protocol that from @xmath61 copies of a pure state @xmath33 we can obtain a state @xmath154 that closely approximates the state @xmath155 to within @xmath77 , where @xmath156 is the entropy of entanglement of @xmath33 . suppose therefore that we have an entanglement monotone @xmath157 that is _ additive _ on pure states . then we may write @xmath158 where the inequality is due to condition 3 for entanglement monotones . if the monotone @xmath157 is `` sufficiently continuous '' , then @xmath159 , where @xmath160 will be small . then we obtain : @xmath161 if the function @xmath157 is sufficiently continuous as the dimension increases , i.e. @xmath162 when @xmath90 , then we obtain @xmath163 . invoking the fact that the entanglement cost for pure states is also given by the entropy of entanglement gives the reverse inequality @xmath164 using similar arguments . hence sufficiently continuous monotones that are additive on pure states will naturally satisfy @xmath165 . of course these arguments are not rigorous , as we have not undertaken a detailed analysis of how @xmath166 or @xmath77 grow with @xmath61 . a rigorous analysis is presented in @xcite , where it is also shown that our assumptions may be slightly relaxed . the result of this rigorous analysis is that a function is equivalent to the entropy of entanglement on pure states _ if and only if _ it is ( a ) normalised on the singlet state , ( b ) additive on pure states , ( c ) non - increasing on _ deterministic _ pure state to pure state locc transformations , and ( d ) _ asymptotically continuous _ on pure states . the term _ asymptotically continuous _ is defined as the property @xmath167 whenever the trace norm @xmath168 on a sequence of hilbert spaces @xmath169 tends to 0 as @xmath170 . it is interesting to notice that these constraints only concern pure state properties of @xmath157 , and that they are _ necessary and sufficient_. as a consequence of the above discussion we can conclude that we could have redefined the set of axiomatic requirements ( 1)-(4 ) , without changing the set of admissible measures . we could have replaced axiom ( 4 ) with two separate requirements of ( 4a ) additivity on pure states and ( 4b ) asymptotic continuity on pure states . together with axiom ( 3 ) this would automatically force any entanglement measure to coincide with the entropy of entanglement on pure states . it is furthermore interesting to note that the failure of an entanglement measure to satisfy asymptotic continuity is strongly related to the counterintuitive effect of _ lockability _ @xcite . the basic question behind lockability is : how much can entanglement of any bi- or multipartite system change when one qubit is discarded ? a measure of entanglement is said to be _ lockable _ if the removal of a single qubit from a system can reduce the entanglement by an arbitrarily large amount . this qubit hence acts as a ` key ' which once removed ` locks ' the remaining entanglement . so which entanglement measures are lockable ? the remarkable answer is this effect can occur for several entanglement measures , including the entanglement cost . on the other hand another class of measures that will be described later , _ relative entropies of entanglement _ , are not lockable @xcite . it can also be proven that any measure that is convex and is _ not _ asymptotically continuous is lockable @xcite . _ extremal entanglement measures _ in the discussions so far we have formulated several requirements on entanglement measures and suggested that various measures exist that satisfy those criteria . it is now interesting to bound the range of such entanglement measures . one may in fact show that the entanglement cost @xmath102 and the distillable entanglement @xmath108 are in some sense _ extremal _ measures @xcite , in that they are upper and lower bounds for many ` _ wholesale _ ' entanglement measures . to be precise , suppose that we have a quantity @xmath171 satisfying conditions ( 1 ) - ( 3 ) above , that is also asymptotically continuous on mixed states , and also has a _ regularisation _ @xmath172 then it can be shown that @xmath173 in fact the conditions under which this statement is true are slightly more general than the ones that we have listed - for more details see @xcite . _ entanglement ordering _ the above considerations have allowed us to impose quite a great deal of structure on entangled states and the next section will make this even more clear . it should be noted however that the axioms 1 - 4 as formulated above are _ not _ sufficient to give a _ unique _ total ordering in terms of the entanglement of the set of states . one can show that any two entanglement measures satisfying axiom 4 can only impose the same ordering on the set of entangled states if they are actually exactly the same . more precisely , if for two measures @xmath174 and @xmath175 and _ any _ pair of states @xmath176 and @xmath177 we have that @xmath178 implies @xmath179 , then if both measures satisfy axiom 4 it must be the case that @xmath180 @xcite ( see @xcite for ordering results for other entanglement quantities ) . given that the entanglement cost and the distillable entanglement are strictly different on all entangled mixed states @xcite this implies that there is not a unique order , in terms of entanglement , on the set of entangled states . this suggests one of several viewpoints . we may for example have neglected to take account of the resources in entanglement manipulation with sufficient care , and doing so might lead to the notion of a unique total order and therefore a unique entanglement measure @xcite . alternatively , it may be possible that the setting of locc operations is too restrictive , and a unique total order and entanglement measure might emerge when considering more general sets of operations @xcite . both approaches have received some attention but neither has succeeded completely at the time of writing this article . in the following we will simply accept the non - uniqueness of entanglement measures as an expression of the fact that they correspond to different operational tasks under which different forms of entanglement may have different degrees of usefulness . in this section we discuss a variety of bipartite entanglement measures and monotones that have been proposed in the literature . all the following quantities are entanglement _ monotones _ , in that they can not increase under locc . hence when they can be calculated they can be used to determine whether certain ( finite or asymptotic ) locc transformations are possible . however , some measures have a wider significance that we will discuss as they are introduced . before we continue , we consider some features of the distillable entanglement , particularly with regard to its computation , as this will be important for some of our later discussion . _ the distillable entanglement _ the distillable entanglement , @xmath108 , provides us with the rate at which noisy mixed states @xmath25 may be converted into the ` gold standard ' singlet state by locc alone . it formal definition is @xmath110 = 0 \right\}.\ ] ] the complexity of this variational definition has the unfortunate consequence that despite the importance of the distillable entanglement as an entanglement measure , very little progress has been made in terms of its computation . it is known for pure states ( where it equals the entropy of entanglement ) , and for some simple but very special states @xcite ( see the end of this paragraph ) . to obtain such results and to gain insight into the amount of distillable entanglement it is particularly important to be able to provide bounds on its value . _ upper bounds _ can , by virtue of eq . ( [ upperlower ] ) and requirement 3 for entanglement monotones , be provided by any other entanglement monotone and measure but non - monotonic bounds are also of interest ( see the remainder of this section on entanglement measures ) . calculating _ lower bounds _ is more challenging . some lower bounds can be obtained by the construction of explicit entanglement purification procedures @xcite in particular for bell diagonal states @xcite . as every state can be reduced to a bell diagonal state by random bi - local rotations of the form @xmath181 ( a process known as _ twirling _ ) , these methods result in general lower bounds applicable to all states . improving these bounds is very difficult as it generally requires the explicit construction of complex purification procedures in the asymptotic limit of many copies . in this context it is of considerable interest to study the _ conditional entropy _ , which is defined as @xmath182 for a bipartite state @xmath183 . it was known for some time that @xmath184 gives a lower bound for both the entanglement cost and another important measure known as the _ relative entropy of entanglement _ however , this bound was also recently shown to be true for the one way distillable entanglement : @xmath185 where @xmath186 is the distillable entanglement under the restriction that the classical communication may only go one way from alice to bob @xcite . this bound is known as the _ hashing inequality _ @xcite , and is significant as it is a computable , non - trivial , _ lower bound _ to @xmath108 , and hence supplies a non - trivial lower bound to many other entanglement measures . while this bound is generally not tight , it should be noted that there are examples for which it equals the distillable entanglement , these include bell diagonal states of rank @xmath187 @xcite and some other special classes of state such as @xmath188 for which relative entropy of entanglement ( ie an upper bound to @xmath189 ) can be computed @xcite and is found to equal the hashing inequality . the following subsection will present a variety of other entanglement measures and quantities that provide upper bounds on the distillable entanglement . @xmath18 _ entanglement cost _ for a given state @xmath25 the entanglement cost quantifies the maximal possible rate @xmath91 at which one can convert blocks of _ 2-qubit _ maximally entangled states into output states that approximate many copies of @xmath25 , such that the approximations become vanishingly small in the limit of large block sizes . if we denote a general trace preserving locc operation by @xmath95 , and write @xmath96 for the density operator corresponding to the maximally entangled state vector in @xmath97 dimensions , i.e. @xmath98 , then the entanglement cost may be defined as @xmath190 = 0 \right\}\ ] ] this quantity is again very difficult to compute indeed . it is known to equal the entropy of entanglement for pure bi - partite states . it can also be computed for trivial mixed states @xmath191 where the states @xmath192 may be discriminated locally perfectly without destroying the states . a simple example is @xmath193 and @xmath194 . fortunately , a closely related measure of entanglement , namely the entanglement of formation , provides some hope as it may actually equal the entanglement cost . therefore , we move on to discuss its properties in slightly more detail . @xmath18 _ entanglement of formation _ for a mixed state @xmath25 this measure is defined as @xmath195 given that this measure represents the minimal possible average entanglement over all pure state decompositions of @xmath25 , where @xmath196 is taken as the measure of entanglement for pure states , it can be expected to be closely related to the entanglement cost of @xmath25 . note however that the entanglement cost is an asymptotic quantity concerning @xmath87 in the limit @xmath90 . it is not self - evident and in fact unproven that the entanglement of formation accounts for that correctly . note however , that the _ regularised _ or _ asymptotic _ version of the entanglement of formation , which is defined as @xmath197 can be proven rigorously to equal the entanglement cost @xcite , i.e. @xmath198 obviously , the computation of either , the entanglement cost or the asymptotic entanglement of formation , are extraordinarily difficult tasks . however , there are indications , though no general proof , that the entanglement of formation is additive , i.e. @xmath199 , a result that would simplify the computation of @xmath102 significantly if it could be proven . further to some numerical evidence for the correctness of this property it is also known that the entanglement of formation is additive for maximally correlated states in @xmath200 dimensions , ie states @xmath201 @xcite . more generally it is a _ major _ open question in quantum information to decide whether @xmath202 is a fully additive quantity , i.e. whether @xmath203 schematic picture of the situation described by eq . ( [ subadd ] ) . the entanglement of formation of an arbitrary four particle state @xmath48 , with particles held by parties @xmath52 and @xmath141 is given is given on the left hand side of eq . ( [ subadd ] ) . the right hand side of eq . ( [ subadd ] ) is the sum of the entanglement of formation of the states @xmath204 and @xmath205 obtained by tracing out the lower upper half of the system . ] this problem is known to be equivalent to the _ strong superadditivity _ of @xmath202 @xmath206 where the indices @xmath207 and @xmath187 refer to two pairs or entangled particles while @xmath52 and @xmath141 denote the different parties ( see fig . [ strongsubadd ] ) . the importance of these additivity problems is twofold . firstly , additivity would imply that @xmath208 leading to a considerable simplification of the computation of the entanglement cost . secondly , the entanglement of formation is closely related to the classical capacity of a quantum channel which is given by the holevo capacity @xcite , and it can be shown that the additivity of @xmath202 is also _ equivalent _ to the additivity of the classical communication capacity of quantum channels @xcite ! the variational problem that defines @xmath202 is extremely difficult to solve in general and at present one must either resort to numerical techniques for general states @xcite , or restrict attention to cases with high symmetry ( e.g. @xcite ) , or consider only cases of low dimensionality . quite remarkably a closed form solution is known for bi - partite qubit states @xcite that we present here . this exact formula is based on the often used two - qubit _ concurrence _ which is defined as @xmath209 where the @xmath57 are , in decreasing order , the square roots of the eigenvalues of the matrix @xmath210 where @xmath211 is the elementwise complex conjugate of @xmath25 . for general bi - partite qubit states it has been shown that @xcite @xmath212 with @xmath213 the two - qubit @xmath107 and the two - qubit concurrence are monotonically related which explains why some authors prefer to characterise entanglement using only the concurrence rather than the @xmath202 . it should be emphasised however that it is only the entanglement of formation that is an entanglement measure , and that the concurrence obtains its meaning via its relation to the entanglement of formation and not vice versa . for higher dimensional systems this connection breaks down - in fact there is not even a unique definition of the concurrence . therefore , the use of the entanglement of formation even in the two - qubit setting , is preferable . @xmath18 _ entanglement measures from convex roof constructions _ the entanglement of formation @xmath202 is an important example of the general concept of a _ convex roof _ construction . the convex roof @xmath214 of a function @xmath145 is defined as the largest convex function that is for all arguments bounded from above by the function @xmath145 . a simple example in one variable is given by @xmath215 and its convex roof @xmath216 fig . [ convexroof ] illustrates this idea graphically with an example for the convex roof for a function of a single variable . generally , for a function @xmath145 defined on a convex subset of @xmath217 , the convex roof @xmath214 can be constructed via the variational problem @xmath218 where the infimum is taken over all possible probability distributions @xmath22 and choices of @xmath219 such that @xmath220 . it is easy to see that @xmath214 is convex , that @xmath221 and that any other convex function @xmath222 that is smaller than @xmath145 also satisfies @xmath223 . the importance of the convex roof method is based on the fact that it can be used to construct entanglement monotones from any unitarily invariant and concave function of density matrices @xcite . as this construction is very elegant we will discuss how it works in some detail . suppose that we already have a function @xmath130 of _ pure _ states , that is known to be an entanglement monotone on _ pure _ states . this means that for an locc transformation from an input pure state @xmath48 to output pure states @xmath192 with probability @xmath22 , we have that : @xmath224 such pure state entanglement monotones are very well understood , as it can be shown that a function is a pure state monotone iff it is a unitarily invariant concave function of the single - site reduced density matrices @xcite . let us consider the convex - roof extension @xmath225 of such a pure state monotone @xmath130 to mixed states . a general locc operation can be written as a sequence of operations by alice and bob . suppose that alice goes first , then she will perform an operation that given outcome @xmath226 performs the transformation : @xmath227 where the @xmath228 are alice s local kraus operators corresponding to outcome @xmath226 , and @xmath229=tr@xmath230 is the probability of getting outcome @xmath226 . if @xmath231 for any particular outcome , then alice s operation is _ impure _ , in that an input pure state may be taken to a mixed output . however , any such locc _ impure _ operation may be implemented by first performing a locc _ pure _ operation , where alice and bob retain information about all @xmath232 , followed by ` forgetting ' the values of @xmath232 at the end . it can be shown quite straightforwardly that if an entanglement measure is convex , then the process of ` forgetting ' can not increase the average output entanglement beyond the average output entanglement of the intermediate pure operation . hence if one shows that a convex quantity is an entanglement monotone for _ pure _ locc operations , then it will be an entanglement monotone in general . this means that we need only prove that @xmath225 is an entanglement monotone for _ pure _ operations acting upon input mixed states . this can be done as follows @xcite . let @xmath25 be an input state with optimal decomposition @xmath233 , i.e. @xmath234 suppose that we act upon this state with a measuring locc operation , where outcome @xmath226 signifies that we have implemented the ( not trace - preserving ) _ pure _ map @xmath235 ( i.e. corresponding to a single kraus - operator ) . let us define : @xmath236 it is clear that @xmath237 , as required by the standard probabilistic interpretation of ensembles . hence given outcome @xmath226 the state @xmath25 transforms to : @xmath238 hence by the convexity of @xmath225 we have that : @xmath239 and because @xmath225 is a monotone for operations from pure to pure states , and as each @xmath240 is pure by assertion , we find that : @xmath241 hence it can be seen that the convex - roof of any pure state entanglement monotone is automatically an entanglement monotone for locc transformations from mixed states to mixed states . together with the result that a function of pure states is an entanglement monotone iff it is a unitarily invariant concave function of the single - site density matrices @xcite , this provides a very elegant way of constructing many convex - roof entanglement monotones . it is interesting to note that although this method can also be used to construct monotones under separable operations , it does not work for constructing monotones under the set of ppt transformations , as unlike the case of locc/ separable operations , an _ impure _ ppt operation _ can not _ always be equated to a _ pure ppt _ operation plus _ forgetting _ @xcite . a schematic picture of the convex roof construction in one dimension . the non - convex function @xmath242 is given by the solid line . the dotted curve is a convex function smaller than @xmath145 and the convex roof , the largest convex function that is smaller than @xmath145 , is drawn as a dashed curved ( it coincides in large parts with @xmath145 ) . ] @xmath18 _ relative entropy of entanglement _ so far we discussed the extremal entanglement measures , entanglement cost and entanglement of distillation . for some time it was unclear whether they were equal or whether there are any entanglement measures that lie between these two . the regularised version of the _ relative entropy of entanglement _ provides an example of a measure that lies between @xmath243 and @xmath189 . one way of understanding the motivation for its definition is by considering total correlations . these are measured by the quantum mutual information @xcite @xmath244 employing the _ quantum relative entropy _ @xmath245 which is a measure of distinguishability between quantum states one may then rewrite the quantum mutual information as @xmath246 if the total correlations are quantified by a comparison of the state @xmath183 with the uncorrelated state @xmath247 then it is intuitive to try and measure the quantum part of these correlations by a comparison of @xmath183 with the closest separable state - a classically correlated state devoid of quantum correlations . this approach gives rise to the general definition of the relative entropy of entanglement @xcite with respect to a set @xmath23 as @xmath248 this definition leads to a class of entanglement measures known as the _ relative entropies of entanglement _ ( see @xcite for a possible operational interpretation ) . in the bipartite setting the set @xmath23 can be taken as the set of separable states , states with positive partial transpose , or non - distillable states , depending upon what you are regarding as ` free ' states . in the multiparty setting there are even more possibilities @xcite but for each such choice a valid entanglement measure is obtained as long as the set @xmath23 is mapped onto itself under locc ( one may even consider more general classes of operations as long as @xmath23 is mapped onto itself ) . the relative entropy of entanglement is defined as the smallest relative entropy distance from the state @xmath25 to states @xmath26 taken from the set @xmath23 . the set @xmath23 may be defined as the set of separable states , non - distillable states or any other set that is mapped onto itself by locc . ] employing the properties of the quantum relative entropy it is then possible to prove that it is a convex entanglement measure satisfying all the conditions 1 - 4 @xcite which is also asymptotically continuous @xcite . the bipartite relative entropies have been used to compute tight upper bounds to the distillable entanglement of certain states @xcite , and as an invariant to help decide the asymptotic interconvertibility of multipartite states @xcite . the relative entropy of entanglement is bounded from below by the conditional entropy @xmath249 which can be obtained from the fact that for any bi - partite non - distillable state @xmath25 we have @xmath250 the relative entropy measures are generally not additive , as bipartite states can be found where @xmath251 the regularized relative entropy of entanglement @xmath252 is therefore of some interest . in various cases exhibiting a high degree of symmetry the regularised versions of some relative entropy measures can be calculated employing ideas from semi - definite programming and optimization theory @xcite . these cases include the _ werner states _ , i.e. states that are invariant under the action of unitaries of the form @xmath181 , and which take the form @xmath253 , where @xmath254 $ ] and @xmath255 ( @xmath256 ) are states proportional to the projectors onto the anti - symmetric ( symmetric ) subspace . it can be shown that @xcite @xmath257 where @xmath258 . it is notable that while this expression is continuous in @xmath259 it is not differentiable for @xmath260 . these results can be extended to the more general class of states that is invariant under the action of @xmath261 , where @xmath262 is an orthogonal transformation @xcite . _ other distance based measures _ in eq . ( [ distance ] ) one may consider replacing the quantum relative entropy by different distance measures to quantify how far a particular state is from a chosen set of disentangled states . many interesting examples of other functions that can be used for this purpose may be found in the literature ( see e.g. @xcite ) . it is also worth noting that the relative entropy functional is _ asymmetric _ , in that @xmath263 . this is connected with asymmetries that can occur in the discrimination of probability distributions @xcite . one can consider reversing the arguments and tentatively define an locc monotone @xmath264 . the resulting function has the advantage of being additive , but unfortunately it has the problem that it can be infinite on pure states @xcite . an additive measure that does not suffer from this deficiency will be presented later on in the form of the ` squashed ' entanglement . @xmath18 _ the distillable secret key _ the distillable secret key , @xmath265 , quantifies the asymptotic rate at which alice and bob may distill secret classical bits from many copies of a shared quantum state . alice and bob may use a shared quantum state to distribute a classical bit of information - for instance if they share a state @xmath266 , then they may measure it in the @xmath267 basis to obtain an identical classical bit @xmath268 , which could form the basis of a cryptographic protocol such as one - time pad ( see e.g. @xcite for a description of one - time pad ) . however , if we think of a given bipartite mixed state @xmath183 as the reduction of a pure state held between alice , bob , and a malicious third party eve , then it is possible that eve could obtain information about the secret bit from measurements on her subsystem . in defining @xmath269 it is assumed that each copy of @xmath183 is purified _ independently _ of the other copies . if we reconsider the example of the state @xmath270 , we can easily see that it is not secure . for instance , it could actually be a reduction of a ghz state @xmath271 held between alice , bob and eve , in which case eve could also have complete information about the ` secret ' bit . the quantity @xmath269 is hence zero for this state , and is in fact zero for all separable states . one way of getting around the problem of eve is to use entanglement distillation . if alice and bob distill bipartite pure states , then because pure states must be uncorrelated with any environment , any measurements on those pure states will be uncorrelated with eve . moreover , if the distilled pure states are epr pairs , then because each local outcome @xmath267 occurs with equal probability , each epr pair may be used to distribute exactly 1 secret bit of information . this means that @xmath272 . however , entanglement distillation is not the only means by which a secret key can be distributed , it examples of ppt states are known where @xmath273 , even though @xmath274 for all ppt states @xcite . it has also been shown that the regularized relative entropy with respect to separable states is an upper bound to the distillable secret key , @xmath275 @xcite . @xmath18 _ logarithmic negativity _ the partial transposition with respect to party @xmath141 of a bipartite state @xmath183 expanded in a given local orthonormal basis as @xmath276 is defined as @xmath277 the spectrum of the partial transposition of a density matrix is independent of the choice of local basis , and is independent of whether the partial transposition is taken over party @xmath52 or party @xmath141 . the positivity of the partial transpose of a state is a necessary condition for separability , and is sufficient to prove that @xmath278 for a given state @xcite . the quantity known as the _ negativity _ @xcite , @xmath279 , is an entanglement monotone @xcite that attempts to quantify the negativity in the spectrum of the partial transpose . we will define the negativity as @xmath280 where @xmath281tr@xmath282 is the trace norm . while being a convex entanglement monotone , the negativity suffers the deficiency that it is not additive . a more suitable choice for an entanglement monotone may therefore be the so called _ logarithmic negativity _ which is defined as @xmath283 the monotonicity of the negativity immediately implies that @xmath284 is an entanglement monotone that can not increase under the more restrictive class of deterministic locc operations , ie @xmath285 . while this is not sufficient to qualify as an entanglement monotone it can also be proven that it is a monotone under probabilistic locc transformations @xcite . it is additive by construction but fails to be convex . although @xmath284 is manifestly continuous , it is not asymptotically continuous , and hence does not reduce to the entropy of entanglement on all pure states . the major practical advantage of @xmath284 is that it can be calculated very easily . in addition it also has various operational interpretations as an upper bound to @xmath108 , a bound on teleportation capacity @xcite , and an asymptotic entanglement cost for exact preparation under the set of ppt operations @xcite . @xmath18 _ the rains bound _ the logarithmic negativity , @xmath284 , can also been combined with a relative entropy concept to give another monotone known as the _ rains bound _ @xcite , which is defined as @xmath286\ , .\ ] ] the function @xmath287 that is to be minimized is not convex which suggests the existence of local minima making the numerical minimization infeasible . nevertheless , this quantity is of considerable interest as one can observe immediately that @xmath288 is a lower bound to @xmath289 as @xmath290 vanishes for states @xmath26 that have a positive partial transpose . it can also be shown that @xmath288 is an upper bound to the distillable entanglement . it is interesting to observe that for werner states @xmath288 happens to be equal to @xmath291 @xcite , a connection that has been explored in more detail in @xcite . @xmath18 _ squashed entanglement _ another interesting entanglement measure is the squashed entanglement @xcite ( see also @xcite ) which is defined as @xmath292 \nonumber\\ & ~ & { \rm where : } \nonumber \\ & ~ & i(\rho_{abe } ) : = s(\rho_{ae})+s(\rho_{be})-s(\rho_{abe})-s(\rho_{e})\ , . \nonumber\end{aligned}\ ] ] in this definition @xmath293 is the _ quantum conditional mutual information _ , which is often also denoted as @xmath294 . the motivation behind @xmath295 comes from related quantities in classical cryptography that determine correlations between two communicating parties and an eavesdropper . the squashed entanglement is a convex entanglement monotone that is a lower bound to @xmath107 and an upper bound to @xmath108 , and is hence automatically equal to @xmath153 on pure states . it is also additive on tensor products , and is hence a useful non - trivial lower bound to @xmath102 . it has furthermore been proven that the squashed entanglement is continuous @xcite , which is a non - trivial statement because in principle the minimization must be carried out over _ all _ possible extensions , including infinite dimensional ones . note that despite the complexity of the minimization task one may find upper bounds on the squashed entanglement from explicit guesses which can be surprisingly sharp . for the totally anti - symmetric state @xmath255 for two qutrits one obtains immediately ( see example 9 in @xcite ) that @xmath296 which is very close to the sharpest known upper bound on the distillable entanglement for this state which is @xmath297 @xcite . the squashed entanglement is also known to be lockable @xcite , and is an upper bound to the secret distillable key @xcite . @xmath18 _ robustness quantities and norm based monotones _ this paragraph discusses various other approaches to entanglement measures and then moves on to demonstrate that they and some of the measures discussed previously can actually be placed on the same footing . _ robustness of entanglement _ another approach to quantifying entanglement is to ask how much noise must be mixed in with a particular quantum state before it becomes separable . for example @xmath298 measures the minimal amount of _ global _ state @xmath26 that must be mixed in to make @xmath25 separable . despite the intuitive significance of equation ( [ lam ] ) , for mathematical reasons it is more convenient to parameterize this noise in a different way : @xmath299 this quantity , @xmath300 , is known as the _ global robustness _ of entanglement @xcite , and is monotonically related to @xmath301 by the identity @xmath302 . however , the advantage of using @xmath303 rather than @xmath301 is that the first quantity has very natural mathematical properties that we shall shortly discuss . the global robustness mixes in arbitrary noise @xmath26 to reach a separable state , however , one can also consider noise of different forms , leading to other forms of robustness quantity . for instance the earliest such quantity to be defined , which is simply called the _ robustness _ , @xmath304 , is defined exactly as @xmath300 except that the noise @xmath26 must be drawn from the set of separable states @xcite . one can also replace the set of separable states in the above definitions with the set of ppt states , or the set of non - distillable states . the robustness monotones can often be calculated or at least bounded non - trivially , and have found applications in areas such as bounding fault tolerance @xcite . _ best separable approximation _ rather than mixing in quantum states to destroy entanglement one may also consider the question of how much of a separable state is contained in an entangled state . the ensuing monotone is known as the _ best separable approximation _ @xcite , which we define as @xmath305 this measure is not easy to compute analytically or numerically . note however , that replacing the set sep by the set ppt allows us to write this problem as a semidefinite programme @xcite for which efficient algorithms are known . _ one shape fits all _ it turns out that the robustness quantities , the best separable approximation as well as the negativity are all part of a general family of entanglement monotones . such connections were first observed in @xcite , where it was noted that the negativity and robustness are part of a general family of monotones that can be constructed via a concept known as a _ base norm _ @xcite . we will explain this connection in the following . however , our discussion will deviate a little from the arguments presented in @xcite , as this will allow us to include a wider family of entanglement monotones such as @xmath303 and @xmath306 . to construct this family of monotones we require two sets @xmath307 of operators satisfying the following conditions : ( a ) @xmath307 are closed under locc operations ( even measuring ones ) , ( b ) @xmath307 are convex cones ( i.e. also closed under multiplication by non - negative scalars ) , ( c ) each member of @xmath308 can be written in the form @xmath309positive - semidefinite operator , where @xmath310 are fixed real constants , and ( d ) any hermitian operator @xmath311 may be expanded as : @xmath312 where @xmath313 are normalised to have trace @xmath314 respectively , and @xmath315 . given two such sets @xmath307 and any state @xmath25 we may define an entanglement monotone as follows : @xmath316 note that if @xmath317 are also constrained to be quantum states ( i.e. @xmath318 ) , then we may rewrite this equation : @xmath319 hence equation ( [ rdef ] ) defines a whole family of quantities that have a similar structure to robustness quantities . in the more general case where @xmath320 , the quantities @xmath321 will not be robustness measures , but they will still be entanglement monotones . this can be shown as follows , where we will suppress the subscripts @xmath322 for clarity . suppose that a locc operation acts on @xmath25 to give output @xmath323 with probability @xmath324 . suppose also that the optimum expansion of the initial state @xmath25 is : @xmath325 then the output ensemble can be written as : @xmath326 where @xmath327 now because of the structure of each operator in @xmath307 , we have that @xmath328 , and hence for each outcome @xmath0 the expansion in ( [ expansion ] ) is a valid decomposition . this means that the average output entanglement satisfies : @xmath329 and hence the @xmath330 give entanglement monotones . it is also not difficult to show that the @xmath330 are convex functions . in the case that the two sets @xmath23 and @xmath331 are identical , then the quantity @xmath332 can be shown to be a norm , and in fact it is a norm of the so - called _ base norm _ kind . as @xmath333 can be written as a simple function of the corresponding @xmath334 , this gives the robustness quantities a further interesting mathematical structure . all the monotones mentioned at the beginning of this subsection fit into this family - the ` _ robustness _ ' arises when both @xmath307 are the set of separable operators ; the ` _ best separable approximation _ ' arises when @xmath23 is the set of separable operators , @xmath331 is the set @xmath335positive semi - definite operators@xmath336 ; the global robustness arises when @xmath23 is the set of separable operators , @xmath331 is the set of all positive semidefinite operators @xcite ; the negativity arises when @xmath337 where both @xmath307 are the set of normalised hermitian matrices with positive partial transposition . note that the ` _ random robustness _ ' is not a monotone and so does not fit into this scheme , for definition and proof of non - monotonicity see @xcite . _ the greatest cross norm monotone _ another form of norm based entanglement monotone is the _ cross norm _ monotone proposed in @xcite . the _ greatest cross norm _ of an operator @xmath52 is defined as : @xmath338 \label{rud}\ ] ] where @xmath339 is the trace norm , and the infimum is taken over all decompositions of @xmath52 into finite sums of product operators . for finite dimensions it can be shown that a density matrix @xmath183 is separable iff @xmath340=1 , and that the quantity : @xmath341 is an entanglement monotone @xcite . as it is expressed as a complicated variational expression , @xmath342 can be difficult to calculate . however , for pure states and cases of high symmetry it may often be computed exactly . although @xmath342 does not fit precisely into the family of base norm monotones discussed above , there is a relationship . if the sum in ( [ rud ] ) is restricted to _ hermitian _ @xmath343 and @xmath344 ( which is of course only allowed if @xmath52 is hermitian ) , then we recover precisely the base norm @xmath345 , where @xmath307 are taken as the set of separable states . hence @xmath346 is an upper bound to the robustness @xcite . @xmath18 _ entanglement witness monotones _ entanglement witnesses are tools used to try to determine whether a state is separable or not . a hermitian operator @xmath347 is defined as an entanglement witness if : @xmath348 hence @xmath347 acts as a linear hyperplane separating some entangled states from the convex set of separable ones . an entanglement witness is a hermitean operator defining a hyperplane in the space of positive operators such that for all separable states we have @xmath349 and there is a @xmath25 for which @xmath350 . ] many entanglement witnesses are known , and in fact the chsh inequalities are well known examples . one can take a suitable entanglement witness ( ew ) and use the amount of ` violation ' @xmath351 as a measure of the non - separability of a given state . many entanglement monotones can be constructed by choosing ( bounded ) sets of of ews and defining monotones as the minimal violation over all witnesses taken from the chosen set - see e.g. @xcite . it turns out that this approach also offers another unified way of understanding the robustness and negativity measures discussed in the previous item @xcite . this concludes our short survey of basic entanglement measures . our review has mostly been formulated for two - party systems with finite dimensional constituents . in the remaining two subsections we will briefly summarize the problems that we are faced with in more general settings - where we are faced with more parties and infinite dimensional systems . we will present some of the results that have been obtained so far , and highlight some unanswered questions . in the preceding sections we have explicitly considered only finite dimensional systems . however , one may also develop a theory of entanglement for the infinite dimensional setting . this setting is often also referred to as the _ continuous variable _ regime , as infinite dimensional pure states are usually considered as wavefunctions in continuous position or momentum variables . the quantum harmonic oscillator is an important example of a physical system that needs to be described in an infinite dimensional hilbert space , as it is realized in many experimental settings , e.g. as modes of quantized light . _ general states _ a naive approach to infinite dimensional systems encounters several complications , in particular with regards to continuity . firstly , we will need to make some minimal requirements on the hilbert space , namely that the system has the property that tr@xmath352 to avoid pathological behaviour due to limit points in the spectrum @xcite . the harmonic oscillator is an example of a system satisfying this constraint . even so , without further constraints , entanglement measures can not be continuous because by direct construction one may demonstrate that in any arbitrarily small neighborhood of a pure product state , there exist pure states with _ arbitrarily strong _ entanglement as measured by the entropy of entanglement @xcite . the following example makes this explicit . chose @xmath353 where @xmath354 defined by @xmath355 where @xmath356 and @xmath357 are orthonormal bases . then @xmath358 converges to @xmath359 in trace - norm , i.e. , @xmath360 while @xmath361 . obviously , @xmath130 is not continuous around the state @xmath359 . however , this perhaps surprising feature can only occur if the mean energy of the states @xmath362 grows unlimited in @xmath232 . if one imposes additional constraints such as restricting attention to states with bounded mean energy then one finds that the continuity of entanglement measures can be recovered @xcite . more precisely , given the hamiltonian @xmath363 and the set @xmath364\le m\}$ ] where @xmath365 is the set of all density matrices , then we find for example that for @xmath366 , @xmath367 , being a pure state that is supported on a finite - dimensional subspace of @xmath368 , and @xmath369 , @xmath370 , being a sequence of states satisfying @xmath371 then @xmath372 similar statements hold true for the entropy of entanglement and the relative entropy of entanglement . the technical details can be found in @xcite . even with this constraint however , the description of entanglement and its quantification is extraordinarily difficult , although some concrete statements can be made @xcite . note however , that for continuous entanglement measures that are strongly super - additive ( in the sense of eq . ( [ subadd ] ) in the situation given in fig . [ strongsubadd ] ) one can provide lower bounds on entanglement measures in terms of a simpler class of state , the gaussian states @xcite . this motivates the consideration of more constrained sets of states . * _ gaussian states _ * a further simplification that can be made is to consider only the set of _ gaussian _ quantum states . this set of states is important because not only do they play a key role in several fields of theoretical and experimental physics , but they also have some attractive mathematical features that enable many interesting problems to be tackled using basic tools from linear algebra . we will concentrate on this class of states , as they have been subject to the most progress . the systems that are being considered possess @xmath61 canonical degrees of freedom representing for example @xmath61 harmonic oscillators , or @xmath61 field modes of light . these canonical operators are usually arranged in vector form @xmath373 then the canonical commutation relations take the form @xmath374=i \sigma_{j , k}$ ] , where we define the _ symplectic _ matrix as follows : @xmath375.\ ] ] states @xmath25 may now also be characterized by functions that are defined on phase space . given a vector @xmath376 , the weyl or glauber operator is defined as : @xmath377 these operators generate displacements in phase space , and are used to define the _ characteristic function _ of @xmath25 : @xmath378.\ ] ] this can be inverted by the transformation @xcite : @xmath379 and hence the characteristic function uniquely specifies the state . gaussian states are now defined as those states whose characteristic function is a gaussian @xcite , i.e. , @xmath380 where @xmath15 is a @xmath381-matrix and @xmath382 is a vector . in defining gaussian states in this way it is easy to see that the reduced density matrix of any gaussian state is also gaussian - to compute the characteristic function of a reduced density matrix we simply set to zero any components of @xmath383 corresponding to the modes being traced out . as a consequence of the above definition , a gaussian characteristic function can be characterized via its first and second moments only , such that a gaussian state of @xmath61 modes requires only @xmath384 real parameters for its full description , which is polynomial rather than exponential in @xmath61 . the first moments form the displacement vector @xmath385 $ ] @xmath386 which is linked to the above @xmath387 by @xmath388 . they can be made zero by means of a unitary translation in the phase space of individual oscillators and carry no information about the entanglement properties of the state . the second moments of a quantum state are defined as the expectation values @xmath389 . because of the canonical commutation relationships the value of @xmath390 is fixed by the value of @xmath389 ( the operators @xmath391,@xmath392 either commute , or their commutator is proportional to the identity ) , and so all second moments can be embodied in the real symmetric @xmath381 covariance matrix @xmath393 which is defined as @xmath394 \nonumber\\ & = & \mbox{tr}\left [ \rho \left ( \{o_j , o_k\ } - 2 \langle o_j\rangle_{\rho } \langle o_k\rangle_{\rho } \right ) \right]\end{aligned}\ ] ] where @xmath395 denotes the anticommutator . the link to the above matrix @xmath15 is @xmath396 . with this convention , the covariance matrix of the @xmath61-mode vacuum is simply @xmath397 . clearly , not all real symmetric @xmath381-matrix represent quantum states as these must obey the heisenberg uncertainty relation . in terms of the second moments the ` uncertainty principle ' can be written as the matrix inequality @xmath398 note that for one mode this uncertainty principle is actually _ stronger _ than the usual heisenberg uncertainty principle presented in textbooks , and in fact equation ( [ res ] ) is the strongest uncertainty relationship that may be imposed on the 2nd - moments @xmath389 . this is because it turns out that _ any _ real symmetric matrix @xmath393 satisfying the uncertainty principle ( [ res ] ) corresponds to a valid quantum state . proving equation ( [ res ] ) is actually not too difficult @xcite - we start with a @xmath399 component vector of complex numbers @xmath400 , and define an operator @xmath401 . then the positivity of @xmath25 implies that tr@xmath402 . a little algebra , and use of the canonical commutation relationships , shows that tr@xmath403 . this observation has quite significant implications concerning the separability of two - mode gaussian states shared by two parties . indeed , a necessary condition for the separability of gaussian states can be formulated on the basis of the partial transposition , or more precisely partial time reversal , expressed on the level of covariance matrices . in a system with canonical degrees of freedom time reversal is characterized by the transformation that leaves the positions invariant but reverses the relevant momenta @xmath404 . a two - party gaussian state is then separable exactly if the covariance matrix corresponding to the partially transposed state again satisfies the uncertainty relations @xcite . more advanced questions concerning the interconvertibility of pairs of states under local operations can also often be answered fully in terms of the elements of the covariance matrix @xcite . in particular , the question of the interconvertability of pure bi - partite gaussian states of an arbitrary number of modes can be decided in full generality @xcite . _ gaussian operations _ the development of the theory of entanglement of gaussian states requires also the definition of the concept of gaussian operations . gaussian operations may be defined as those operations that map _ all _ gaussian input states onto a gaussian output state . this definition is not constructive but fortunately more useful characterizations exist . physically useful is the fact that gaussian operations correspond exactly to those operations that can be implemented by means of optical elements such as beam splitters , phase shifts and squeezers together with homodyne measurements @xcite . the most general real linear transformation s which implements the mapping @xmath405 will have to preserve the canonical commutation relations @xmath406=i\sigma_{jk}{\openone}$ ] which is exactly the case if @xmath113 satisfies @xmath407 this condition is satisfied by the real @xmath381 matrices @xmath113 that form the so - called real symplectic group @xmath408 . its elements are called symplectic or canonical transformations . it is useful to know that any orthogonal transformation is symplectic . to any symplectic transformation @xmath113 also @xmath409 are symplectic . the inverse of @xmath113 is given by @xmath410 and the determinant of every symplectic matrix is @xmath411 = 1 $ ] @xcite . given a real symplectic transformation @xmath113 there exists a unique unitary transformation @xmath412 acting on the state space such that the weyl operators satisfy @xmath413 for all @xmath414 . on the level of covariance matrices @xmath393 of an @xmath61-mode system a symplectic transformation @xmath113 is reflected by a congruence @xmath415 generalized gaussian quantum operations may also be defined analogously to the finite dimensional setting , ie by appending gaussian state ancillas , performing joint gaussian unitary evolution followed by tracing out the ancillas or performing homodyne detection on them @xcite . _ normal forms _ given a group of transformations on a set of matrices it is always of great importance to identify normal forms for matrices that can be achieved under this group of transformations . of further interest and importance are invariants under the group transformations . for the set of hermitean matrices and the full unitary group these correspond to the concepts of diagonalization and eigenvalues . in the setting of covariance matrices and the symplectic group we are led to the williamson normal forms and the concept of symplectic eigenvalues . indeed , williamson @xcite ( see @xcite for a more easily accessible reference ) proved that for any covariance matrix @xmath15 on @xmath61 harmonic oscillators there exists a symplectic transformation @xmath113 such that @xmath416 the diagonal elements @xmath417 are the so - called _ symplectic eigenvalues _ of a covariance matrix @xmath15 which are the invariants under the action of the symplectic group . the set @xmath418 is usually referred to as the _ symplectic spectrum_. the symplectic spectrum can be obtained directly from the absolute values of the eigenvalues of @xmath419 . the transformation to the williamson normal form implements a normal mode decomposition thereby reducing any computational problem , such as the computation of the entropy , to that for individual uncoupled modes . each block in the williamson normal form represents a thermal state for which the evaluation of most physical quantities is straightforward . * _ entanglement quantification _ * equipped with these tools we may now proceed to discuss the quantification of entanglement in the gaussian continuous variable arena . despite all the above technical tools the quantification of entanglement for gaussian states is complicated and only very few measures may be defined let alone computed . @xmath18 _ entropy of entanglement : _ on the level of pure state we may again employ the entropy of entanglement which we may now express in terms of the covariance matrix . assume alice and bob are in possession of @xmath420 harmonic oscillators in a gaussian state described by the covariance matrix @xmath15 and alice holds @xmath421 of these oscillators . then it can be shown that the entropy of entanglement is given by @xmath422 where the @xmath417 are the _ symplectic _ eigenvalues of alice s reduced state described by the covariance matrix @xmath423 which is simply the submatrix of @xmath15 referring to the system pertaining to alice . these symplectic eigenvalues are , as remarked above , the positive eigenvalues of @xmath424 . the proof of the above formula is obtained by transforming the covariance to its williamson normal form and subsequently determine the entropy of the single mode states . note that on the set of gaussian states the entropy is evidently continuous and it can be shown that this remains the case for the set of states with bounded mean energy @xcite . @xmath18 _ entanglement of formation : _ in the finite dimensional setting the defintion of the entanglement of formation has been unambiguous . in the gaussian state setting however this is no longer the case . one may define the entanglement of formation of a gaussian state either ( i ) with respect to decompositions in pure gaussian states or ( ii ) with respect to decompositions in arbitrary pure states . in case ( i ) it has been proven that the so - defined entanglement of formation is an entanglement monotone under gaussian operations and that it can be computed explicitly in the case where both parties hold a single harmonic oscillator each . remarkably , this entanglement of formation is even additive for symmetric two - mode states @xcite . for the case of a single copy of a mixed symmetric gaussian two mode state it can also be demonstrated that the definition ( i ) coincides with definition ( ii ) @xcite . the entanglement of formation can be shown to be continuous for systems with energy constraint @xcite . @xmath18 _ distillable entanglement : _ the distillable entanglement in the continuous variable setting is , as expected , extremely difficult to compute . furthermore , its definition is not unambiguous as one may define distillation with respect to ( i ) gaussian operations only , or ( ii ) general quantum operations . it is remarkable that it has been proven that the setting ( i ) does not actually permit entanglement distillation at all @xcite . therefore , non - gaussian operations need to be considered . then , in setting ( ii ) , for gaussian states it can be shown to be continuous and interestingly it can also be demonstrated that for any @xmath25 there exists a gaussian state @xmath425 with the same first and second moments such that @xmath426 . finding explicit procedures implementing distillation protocols is very difficult which makes it very difficult to determine lower bounds on the distillable entanglement . various other measures of entanglement , such as those described below , may be used to find upper bounds on the distillable entanglement . @xmath18 _ relative entropy of entanglement : _ as for the entanglement of formation there are now various possible definitions of the relative entropy of entanglement all of which are at least as difficult to compute as in the finite dimensional setting . if the relative entropy of entanglement should serve as a provable upper bound on the distillable entanglement under general locc , then it will have to be computed with respect to the set of separable general continuous variable states . this is obviously a very involved quantity and only known on pure states where it equals the entropy of entanglement . if one considers the relative entropy of entanglement of a state with bounded mean energy with respect to the unrestricted set of separable states , then it can be shown that the relative entropy of entanglement is continuous @xcite . a more tractable setting is that of the relative entropy of entanglement with respect to the set of gaussian separable states but in this case its interpretation is unclear . @xmath18 _ logarithmic negativity : _ as in the finite dimensional setting , most entanglement measures are exceedingly difficult to compute . the exception is again the logarithmic negativity which is an entanglement monotone @xcite but differs , on pure states , from the entropy of entanglement . for a system of @xmath427 harmonic oscillators in a gaussian state described by the covariance matrix @xmath15 , the logarithmic negativity can again be expressed in terms of symplectic eigenvalues . indeed , considering the covariance matrix @xmath428 of the partially transposed state we find @xmath429\ ] ] where the @xmath430 form the symplectic spectrum for the partially transposed state described by covariance matrix @xmath428 , ie the symplectic eigenvalues . this formula is again proven by applying a normal mode decomposition , this time to the partially transposed covariance matrix , reducing the problem to a single mode question . it is interesting to note that on gaussian states the logarithmic negativity also possesses an interpretation as a special type of entanglement cost @xcite . the tools for the manipulation and quantification are used in the assessment of the quality of practical optical entanglement manipulation protocols . it should also be noted that these tools have been used successfully to study entanglement properties of quasi - free fields on lattices ( i.e. lattices of harmonic oscillators ) initiating the study of the scaling behaviour of entanglement between contiguous blocks in the ground state of interacting quantum systems @xcite . the above methods and quantities permitted the rigorous proofs of the scaling of entanglement between contiguous blocks in the ground state of a linear harmonic chain with a hamiltonian that is quadratic in position and momentum@xcite and a rigorous connection between the entanglement of an arbitrary set of harmonic oscillators and its surrounding with the boundary area @xcite . this illustrates the usefulness of the results that have been obtained in continuous variable entanglement theory over the last years . * multi - particle entanglement * although the two - party setting has provided many interesting examples of quantum entanglement , the multiparty setting allows us to explore a much wider range of effects . phenomena such as quantum computation especially when based on cluster states @xcite , entanglement enhanced measurements @xcite , multi - user quantum communication @xcite and the ghz paradox all require consideration of systems with more than two particles . for this reason it is important to investigate entanglement in the multi - party setting . we will proceed along similar lines to the bi - partite setting , first discussing briefly basic properties of states and operations and then describing various approaches to the quantification of multi - particle entanglement . _ states and operations _ in the following we are going to concentrate again on local operations and classical communication whose definition extend straightforwardly to the multi - party setting . some remarks will also be made concerning ppt operations which are here defined as operations that preserve ppt - ness of states across _ all _ possible bi - partite splits . that is , any three - party state shared between @xmath52 , @xmath141 and @xmath431 that remains positive under partial transposition of particle @xmath52 or @xmath141 or @xmath431 is mapped again onto a state with this property . in the bi - partite setting we initiated our discussions with the identification of some general properties of multi - party entangled states such as the identification of disentangled states and maximally entangled states . at this stage crucial differences between the two - party and the multi - party setting become apparent . let us begin by trying to identify the equivalent of the two - party maximally entangled states . in the bi - partite setting we already identified qubit states of the form @xmath74 maximally entangled because every other qubit state can be obtained from it with certainty using locc only . one natural choice for a state with this property could be the ghz - state @xmath432 this state has the appealing property that its entanglement across any bi - partite cut e.g. party a versus parties b and c assume the largest possible value of 1 ebit . also , a local measurement in the @xmath433 basis for example on party a allows us to create deterministically a maximally entangled two - party state of parties @xmath141 and @xmath431 . then we can obtain any other two - party entangled state for parties @xmath141 and @xmath431 by locc . unfortunately , however there are tri - partite entangled states that can not be obtained from the ghz state using locc alone . one such example is the w - state @xcite @xmath434 note however that locc operations applied to a ghz - state allow us to approximate the w - state as closely as we like , albeit with decreasing success probability . in the four party setting however it can be shown that there are pairs of pure states that can not even be obtained from each other approximately employing locc alone @xcite . this clearly shows that on the single - copy level it is not possible to establish a generic notion of a maximally entangled state . of course we have already learnt in the bi - partite setting that the requirement of exact transformations on single copies can lead to phenomena such as incomparable states and does not yield a simple and unified picture of entanglement . in the bi - partite setting such a unified picture for pure state entanglement emerges however in the asymptotic setting of arbitrarily many identically prepared states . one might therefore wonder whether a similar approach will be successful in multi - partite systems . these hopes will be dashed in the following . in the asymptotic setting we would need to establish the possibility for the reversible interconversion in the asymtptotic setting . if that were possible we could rightfully claim that all tri - partite entanglement is essentially equivalent and only appears in different concentrations that we could then quantify unambiguously . the simplest situation that one may consider to explore this possibility is the interconversion between ghz and the epr pairs across parties @xmath435 , @xmath436 and @xmath437 , ie in the limit @xmath438 we would like to see @xmath439 to decide this question one needs to identify sufficiently many entanglement monotones . in the case of reversibility these entanglement monotones will remain constant . the local entropies represent such a monotone . these are not enough to decide the question but it turns out that @xmath440 , ie the sume of the relative entropy of entanglement of the reduction to two parties and the entanglement between these two parties and the third , is also an entanglement monotone in this setting . this is then sufficient to prove that the above process can not be achieved reversibly @xcite . this result suggests that as opposed to the bi - partite setting there is not such a simple and unique concept of a maximally entangled state in the multi - partite setting . one may however try and and make progress by generalizing the idea of a single entangled state from which all other states can be obtained reversibly in the asymptotic setting . instead one may consider a set of states from which all other state may be obtained asymptotically reversibly . the smallest such set is usually referred to as an mregs which stands for minimal reversible entanglement generating set @xcite . it was natural to try and see whether the set @xmath441 @xcite . it is noteworthy that ppt operations also overcome the constraint that is imposed by the non - increase of the schmidt - number under locc . indeed , ppt operations ( and also the use of locc with bound entanglement as a free resource ) allow us to increase the schmidt number . this result already implicit in @xcite was made explicit in @xcite . it was hoped for that this strong increase in probabilities and the vanishing of the schmidt number constraint would lead to reversibility in the multi - partite setting , ie a finite mregs under ppt operations . this question is however still remains open @xcite . up until now we have restricted attention to pure multi - party entangled states . now let us consider the definition of separable multi - particle states . the most natural definition for disentangled states arises from the idea that we call a state disentangled if we can create it from a pure product state by the action of locc only . this implies that separable states are of the form @xmath442 where the @xmath443 label different parties . however , one can go beyond this definition . indeed , the state @xmath444 is clearly entangled and therefore not separable in the above sense . however , it also does not exhibit three - party entanglement as the third party @xmath431 is uncorrelated from the other two . therefore may call this tri - partite state 2-entangled . one may now try and generalize this idea to mixed states . for example we could define as the set of 2-entangled states any @xmath25 that may be written in the form @xmath445 with positive @xmath446 and @xmath447 . then , for @xmath116 parties one may then define k - entangled states as a natural generalization of the above 3-party definition . while this definition appears natural it encounters problems when we consider several identical copies of states of the form given above . in that case one can obtain a 3-entangled state by locc acting on two copies of the above 2-entangled state . as a simple example consider a three party state where alice has two qubits and bob and charlie each hold one . then a state of the form : @xmath448 $ ] is only 2-entangled . however , given two copies of this state the ` classical flag ' particle @xmath449 can enable alice to obtain ( with some probability ) one epr pair with bob , and one with charlie . she can then use these epr pairs and teleportation to distribute any three party entangled state she chooses . states of three qubits displaying a similar phenomenon can also be constructed . hence we are faced with a subtle dilemma - either this notion of ` k - entanglement ' is not closed under locc , or it is not closed under taking many copies of states . note however that these states may still have relevance for example in the study of fault - tolerant quantum computation @xcite . _ quantifying multi - partite entanglement _ already in the bi - partite setting it was realized that there are many non - equivalent ways to quantify entanglement @xcite . this concerned mainly the mixed state case , while in the pure state case the entropy of entanglement is a distinguished measure of entanglement . in the multipartite setting this situation changes . as was discussed above it appears difficult to establish a common currency of multipartite entanglement even for pure states due to the lack of asymptotically reversible interconversion of quantum states . the possibility to define k - entangled states and the ensuing ambiguities lead to additional difficulties in the definition of entanglement measures in multi - partite systems . owing to this there are many ways to go about quantifying multipartite entanglement . some of these measures will be natural generalizations from the bi - partite setting while others will be specific to the multi - partite setting . these measures and their known properties will be the subject of the remainder of this section . _ entanglement cost and distillable entanglement _ in the bi - partite setting it was possible to define unambiguously the entanglement of pure states establishing a common currency for entanglement . this then formed the basis for unique definitions of the entanglement cost and the distillable entanglement . the distillable entanglement determined the largest rate , in the asymptotic limit , at which one may obtain pure maximally entangled states from an initial supply of mixed entangled states using locc only . however , in the multi - particle setting there is no unique target state that one may aim for . one may of course provide a target state specific definition of distillable entanglement , for example the largest rate at which one may prepare ghz states @xcite , cluster states @xcite or any other class that one is interested in . as these individual resources are not asymptotically equivalent each of these measures will capture different properties of the state in question . one encounters similar problems when attempting to define the entanglement cost . again , one may use singlet states as the resource from which to construct the state by locc but one may also consider other resources such as ghz or w states . for each of these settings one may then ask for the best rate at which one can create a target state using locc in the asymptotic limit . therefore we obtain a variety of possible definitions of entanglement costs . while the interpretation of each of these measures is clear it is equally evident that it is not possible to arrive at a unique picture from abstract considerations alone . the operational point of view becomes much more important as different resources may be readily available in different experimental settings and then motivating different definitions of the entanglement cost and the distillable entanglement . @xmath18 _ relative entropic measures . distance measures _ in the bipartite setting we have discussed various distance based measures in which one minimizes the distance of a state with respect to a set of states that does not increase in size under locc . one such set was that of separable states and a particularly important distant measure is the relative entropy of entanglement . this lead to the relative entropy of entanglement . as we discussed in the first part of this section the most natural extension of the definition of separable states in the multipartite setting is given by @xmath442 where the @xmath443 label different parties . in analogy with the bipartite definition one can hence define a multipartite relative entropy measure : @xmath450 where @xmath23 is now the set of multipartite separable states . as in the bipartite case the resulting quantity is an entanglement monotone which , for pure states , coincides with the entropy of entanglement . therefore , on pure states , this measure is additive while it is known to be sub - additive on mixed states . remarkably , the multipartite relative entropy of entanglement is _ not _ even additive for pure states - a counterexample is provided by the totally anti - symmetric state @xmath451 where @xmath452 is the totally anti - symmetric tensor @xcite . one can also compute the relative entropy of entanglement for some other tri - partite states . examples of particular importance in this respect are the w - state for which we find @xmath453 and the states @xmath454 more examples can be found quite easily . also in our discussion of multi - partite entanglement we introduced the notion of k - entangled states . let us denote the set of k - entangled state of an n - partite system by @xmath455 . if ew explicitly consider the single copy setting , then it is clear that that the set @xmath455 does not increase under locc . as a consequence it can be used as the basis for generalizations of the relative entropy of entanglement simply replacing the set @xmath23 above by @xmath455 . we have learnt however that the set @xmath455 may grow when allowing for two or more copies of the state . this immediately implies that the so constructed measure will exhibit sub - additivity again . given that even the standard definition for the multi - partite relative entropy of entanglement is sub - additive this should not be regarded as a deficiency . indeed , this subadditivity may be viewed as a strength as it could lead to particularly strong bounds on the associated distillable entanglement . exactly the same principle may be used to extend any of the distance based entanglement quantifiers to multi - party systems - one simply picks a suitable definition of the ` unentangled ' set @xmath23 ( i.e. a set which is closed under locc operations , and complies with some notion of locality ) , and then defines the minimal distance from this set as the entanglement measure . as stated earlier , one may also replace the class of separable states with other classes of limited entanglement - e.g. states containing only bipartite entanglement . such classes are _ not _ in general closed under locc in the many copy setting and so the resulting quantities may exhibit strong subadditivity and their entanglement monotonicity needs to be verified carefully . @xmath18 _ robustness measures . norm based measures . _ the robustness measures discussed in the bipartite case extend straightforwardly to the multiparty case . in the bipartite case we constructed the robustness monotones from two sets of operators @xmath307 that were closed under locc operations , and in addition satisfied certain convexity and ` basis ' properties . to define analogous monotones in the multiparty case we must choose sets of multiparty operators that have these properties . one could for example choose the sets @xmath307 to be the set of k - separable positive operators , for any integer @xmath232 . @xmath18 _ entanglement of assistance . localizable entanglement . collaborative localizable entanglement . _ one way of characterizing the entanglement present in a multiparty state is to understand how local actions by the parties may generate entanglement between two distinguished parties . for example , in a ghz state @xmath456 of three parties , it is possible to generate an epr pair between any two parties using only locc operations - if one party measures in the @xmath457 basis , then there will be a residual epr pair between the remaining two parties . this is the case even though the reduced state of the two parties is by itself unentangled . the first attempt to quantify this phenomenon was the _ entanglement of assistance _ proposed by @xcite . the _ entanglement of assistance _ is a property of 3-party states , and quantifies the maximal bipartite entanglement that can be generated on average between two parties @xmath458 if party c measures her particle and communicates the result to @xmath458 . a related measure known as the _ localizable entanglement _ was proposed and investigated in @xcite for the general multiparty case - this is defined as the maximum entanglement that can be generated between two parties if all _ remaining _ @xmath61 parties act using locc on the particles that they possess @xcite . both these measures require an underlying measure of bipartite entanglement to quantify the the entanglement between the two singled - out parties . in the original articles @xcite the pure state entropy of entanglement was used , however , one can envisage the use of other entanglement measures @xcite . the localizable entanglement has been shown to have interesting relations to correlation functions in condensed matter systems @xcite . as multiparty entanglement quantifiers , both the entanglement of assistance and the localizable entanglement have the drawback that they can deterministically _ increase _ under locc operations between all parties @xcite . this phenomenon occurs because these measures are defined under the restriction that alice and bob can not be involved in classical communication with any other parties - it turns out that in some situations allowing this communication can increase the entanglement that can be obtained between alice and bob @xcite . this observation lead the authors of @xcite to define the _ collaborative _ localizable entanglement as the maximal bipartite entanglement ( according to some chosen measure ) that may be obtained ( on average ) between alice and bob using locc operations involving _ all _ parties . it is clear that by definition these collaborative entanglement measures are entanglement monotones . it is interesting to note that although the bare localizable entanglement is not a monotone , its regularised version _ is _ a monotone for multiparty pure states @xcite . in @xcite it is shown that the regularised version of the localizable entanglement reduces to the minimal entropy of entanglement across any bipartite cut that divides alice and bob , which is clearly a locc monotonous quantity by the previous discussion of bipartite entanglement measures . @xmath18 _ geometric measure . _ in the case of pure multiparty states one could try to quantify the ` distance ' from the set of separable states by considering various functions of the maximal overlap with a product state @xcite . one interesting choice of function is the logarithm . this was used in @xcite to define the following entanglement quantifier : @xmath459 where the supremum is taken over all pure product states . this quantity is non - negative , equals zero iff the state @xmath48 is separable , and is manifestly invariant under local unitaries . one can extend this quantity to mixed states using a convex roof construction . however @xmath460 is not an entanglement monotone , and it is _ not _ additive for multiparty pure states @xcite . nevertheless , @xmath460 is worthy of investigation as it has useful connections to other entanglement measures , and also has an interesting relationship with the question of channel capacity additivity @xcite . we could also have described @xmath460 as a norm based measure , as the quantity @xmath461 is a norm ( of vectors ) known to mathematicians as the _ injective tensor norm _ @xcite . @xmath18 _ ` tangles ' and related quantities . entanglement quantification by local invariants . _ an interesting property of bipartite entanglement is that it tends to be _ monogamous _ , in the sense that if three parties @xmath462 have the same dimensions , and if two of the parties @xmath52 and @xmath141 are very entangled , then a third party @xmath431 can only be weakly entangled with either @xmath52 or @xmath141 . if @xmath435 are in a singlet state then they can not be entangled with c at all . in @xcite this idea was put into the form of a rigorous inequality for three qubit states using a entanglement quantifier known as the _ tangle _ , @xmath463 . for a @xmath464 dimensional systems the tangle is defined as @xmath465 where @xmath466 is the square of the concurrence of pure state @xmath48 and the infimum is taken over all pure state decompositions . the concurrence can be used in this way as any pure state of a @xmath467 system is equivalent to a two qubit pure state . it has been shown that @xmath468 satisfies the inequality @xcite @xmath469 where the notation @xmath470 means that @xmath471 is computed across the bipartite splitting between party @xmath52 and parties @xmath472 . this shows that the amount of bipartite entanglement between party @xmath52 and several individual parties @xmath473 is bounded from above by the amount of bipartite entanglement between party @xmath52 and parties @xmath474 collectively . in the case of three qubit pure states the _ residual tangle _ @xmath475 is a local - unitary invariant that is independent of which qubit is selected as party @xmath52 , and might be proposed as a ` quantifier ' of three party entanglement for pure states of 3-qubits . however , there are states with genuine three party entanglement for which the residual tangle can be zero ( the w - state serves as an example @xcite ) . however , the residual tangle can only be non - zero if there is genuine tripartite entanglement , and hence can be used as a indicator of three party entanglement . another way to construct multiparty entanglement measures for multi - qubit _ pure _ systems is simply to single out one qubit , compute the entanglement between that qubit and the rest of the system , and then average over all possible choices of the singled out qubit . as any _ pure _ bipartite system of dimensions @xmath476 can be written in terms of two schmidt coefficients , one can apply all the formalism of two - qubit entanglement . this approach has been taken , for example , in the paper by meyer and wallach @xcite . that the quantity proposed in @xcite is essentially only a measure of the bipartite entanglement across various splittings was shown by brennen @xcite . extensions of this approach are presented in @xcite . _ local unitary invariants : _ the residual tangle is only one of many _ local _ unitary invariants that have been developed for multiparty systems . such local invariants are very important for understanding the structure of entanglement , and have also been used to construct prototype entanglement measures . examples of local invariants that we have already mentioned are the schmidt coefficients and the geometric measure . in the multiparty case we may define the _ local _ invariants as those functions that are invariant under a _ local _ group transformation of fixed dimensions . if each particle is assumed for simplicity to have the same dimension @xmath27 , then these local groups are of the form @xmath477 where @xmath443 are taken from a particular @xmath27-dimensional group representation such as the unitary group @xmath478 or the group of invertible matrices @xmath479 . the physical significance of the local @xmath479 invariants is that if two states have different values for such an invariant then they can not even be inter - converted probabilistically using stochastic locc ( ` slocc ' ) operations . in the case of local unitary groups one typically only need consider invariants that are _ polynomial _ functions of the density matrix elements - this is because it can be shown that two states are related by a local unitary iff they have the same values on the set of polynomial invariants @xcite . for more general groups a complete set of polynomial invariants can not always be constructed , and one must also consider local invariants that are not polynomial functions of states - one example is a local @xmath480 invariant called the ` schmidt rank ' , which is the minimal number of product state - vector terms in which a given multiparty pure state may be coherently expanded . it can be shown that one can construct an entanglement monotone ( the ` schmidt measure ' ) as the convex - roof of the logarithm of this quantity @xcite . finding non - trivial local invariants is quite challenging in general and can require some sophisticated mathematics . however , for pure states of some dimensions it is possible to use such invariants to construct a variety of entanglement quantifiers in a similar fashion to the tangle . these quantifiers are useful for identifying different types of multiparty entanglement . we refer the reader to articles @xcite and references therein for further details . quantum entanglement is a rich field of research . in recent years considerable effort has been expended on the characterization , manipulation and quantification of entanglement . the results and techniques that have been obtained in this research are now being applied not only to the quantification of entanglement in experiments but also , for example , for the assessment of the role of entanglement in quantum many body systems and lattice field theories . in this article we have surveyed many results from entanglement theory with an emphasis on the quantification of entanglement and basic theoretical tools and concepts . proofs have been omitted but useful results and formulae have been provided in the hope that they prove useful for researchers in the quantum information community and beyond . it is the hope that this article will be useful for future research in quantum information processing , entanglement theory and its implications for other areas such as statistical physics . _ multiparty entanglement : _ the general characterisation of multiparty entanglement is a major open problem , and yet it is particularly significant for the study of quantum computation and the links between quantum information and many - body physics . particular unresolved questions include : * _ finiteness of mregs for three qubit states _ in an attempt to achieve a notion of reversibility in the multi - partite setting , the concept of mregs was introduced @xcite . this was a set of n - partite states for fixed local dimension from which all other such states may be obtained asymptotically reversibly . it was hoped for that such a set may contain only a finite number of states . however , there are suggestions @xcite that this may not the case . * _ distillation results for specific target states _ in the bi - partite setting the uniqueness of maximally entangled states led to clear definitions for the distillable entanglement . as outlined above this is not so in the multi - party setting . given a specific interesting multiparty target state ( e.g. ghz states , cluster states etc . ) , or set of multiparty target states , what are the best possible distillation protocols that we can construct ? are there good bounds that can be derived using multiparty entanglement measures ? some specific examples have been considered @xcite but more general results are still missing . _ additivity questions : _ of all additivity problems , deciding whether the entanglement of formation @xmath202 is additive is perhaps the most important unresolved question . if @xmath202 is additive this would greatly simplify the evaluation of the entanglement cost . it would furthermore imply the additivity of the classical capacity of a quantum channel @xcite . related to the additivity question is the question of the monotonicity of the entanglement cost under general locc . this may be proven reasonably straightforwardly if the entanglement cost itself is fully additive . however , without this assumption no proof is known to the authors , and in fact a recent argument seems to show that full additivity of the entanglement cost is equivalent to its monotonicity @xcite . in addition to @xmath202 , there are many other measures for which additivity is unknown . examples include the distillable entanglement and the distillable key . _ distillable entanglement _ distillable entanglement is a well motivated entanglement measure of significant importance . its computation is however supremely difficult in general and even the determination of the distillability of a state is difficult . indeed , good techniques or algorithms for deciding whether a bipartite state is distillable or not , and for bounding the distillable entanglement , are still largely missing . * _ are there npt bound entangled states ? _ in the bi - partite setting there are currently three known distinct classes of states in terms of their entanglement properties under locc . these are the separable states , the non - separable states with positive partial transpose ( which are also non - distillable ) , and finally the distillable states . some evidence exists that there is another class of states that do not possess a positive partial transpose but are nevertheless non - distillable @xcite . * _ bounds on the distillable entanglement . _ any entanglement measure provides an upper bound on the distillable entanglement . various bounds have been provided such as the squashed entanglement @xcite , the rains bound @xcite and asymptotic relative entropy of entanglement @xcite . the last two of these coincide for werner states @xcite and it is an open question whether they always coincide , and whether they are larger or smaller than the squashed entanglement . _ entanglement measures _ the present article has presented a host of entanglement measures . many of their properties are known but crucial issues remain to be resolved . amongst these are the following . * _ operational interpretation of the relative entropy of entanglement _ while the entanglement cost and the distillable entanglement possess evident operational interpretations no such clear interpretation is known for the relative entropy of entanglement . a possible interpretation in terms of the distillation of local information has been conjectured and partially proven in @xcite . * _ calculation of various entanglement measures _ there are very few measures of entanglement that can be computed exactly and possess or are expected to possess an operational interpretation . a notable exception is the entanglement of formation for which a formula exists for the two qubit case @xcite . is it possible to compute , or at least derive better bounds , for the other variational entanglement measures ? one interesting possibility is the 2-qubit case - in analogy to @xmath202 , is there a closed form for the relative entropy of entanglement or the squashed entanglement ? * _ squashed entanglement _ as an additive , convex , and asymptotically continuous entanglement monotone the squashed entanglement is known to possess almost all potentially desirable properties as an entanglement measure . nevertheless , there are a number of open interesting questions - in particular : ( 1 ) is the squashed entanglement strictly non - zero on inseparable states , and ( 2 ) can the squashed entanglement be formulated as a finite dimensional optimisation problem ( with eve s system of bounded dimension ) ? * _ asymptotic continuity and lockability questions _ it is unknown whether measures such as the distillable key , the distillable entanglement , and the entanglement cost are asymptotically continuous , and it is unknown whether the distillable entanglement or distillable key are lockable @xcite . this is important to know as lockability quantifies ` continuity under tensor products ' , and so is a physically important property - if a system is susceptible to loss of particles , then any characteristic quantified by a lockable measure will tend to be very fragile in the presence of such noise . _ entanglement manipulation _ entanglement can be manipulated under various sets of operations , including locc and ppt operations . while some understanding of what is possible and impossible has been obtained , a complete understanding has not been reached yet . * _ characterization of entanglement catalysis _ for a single copy of bi - partite pure state entanglement the locc transformations are fully characterized by the theory of majorization @xcite . it was discovered that there are transformations @xmath481 such that its success probability under locc is @xmath482 but for which an entangled state @xmath72 exists such that @xmath483 can be achieved with certainty under locc @xcite . a complete characterization for states admitting entanglement catalysis is currently not known . * _ other classes of non - global operation . reversibility under ppt operations _ it is well established that even in the asymptotic limit locc entanglement transformations of mixed states are irreversible . however in @xcite it was shown that that the antisymmetric werner state may be reversibly interconverted into singlet states under ppt operations @xcite . it is an open question whether this result may be extended to all werner states or even to all possible states . in addition to questions concerning ppt operations , are there other classed of non - global operation that can be useful ? if reversibility under ppt operations does not hold , do any other classes of non - global operations exhibit reversibility ? more open problems in quantum information science can be found in the braunschweig webpage of open problems @xcite . we hope that this list will stimulate some of the readers of this article into attacking some of these open problems and perhaps report solutions , even partial ones . _ acknowledgments _ we have benefitted greatly from discussions on this topic with numerous researchers over several years . in relation to this specific article , we must thank k. audenaert , f. brando , m. horodecki , o. rudolph , and a. serafini for helping us to clarify a number of issues , as well as g.o . myhr , p. hyllus , and a. feito - boirac for careful reading and helpful suggestions . we would also like to thank m. christandl , j. eisert , j. oppenheim , a. winter , and r.f . werner for sharing with us their thoughts on open problems . this work is part of the qip - irc ( www.qipirc.org ) supported by epsrc ( gr / s82176/0 ) , the eu integrated project qubit applications ( qap ) funded by the ist directorate as contract no . 015848 , the eu thematic network quprodis ( ist-2001 - 38877 ) , the leverhulme trust , the royal commission for the exhibition of 1851 and the royal society . 99 m. bell , k. gottfried and m. veltmann , _ john s. bell on the foundations of quantum mechanics _ , world scientific publishing , singapore . bell , physica * 1 * , 195 ( 1964 ) . l. hardy , contemp . phys . * 39 * , 419 ( 1998 ) . plenio and v. vedral , contemp . phys . * 39 * , 431 ( 1998 ) . b. schumacher and m. d. westmoreland , e - print arxiv quant - ph/0004045 . m. horodecki , quant . inf . comp . * 1 * , 3 ( 2001 ) . m. christandl , phd thesis , quant - ph/0604183 . p. horodecki and r. horodecki , quant . inf . comp . * 1 * , 45 ( 2001 ) . j. eisert and m. b. plenio , int . j. quant . inf . * 1 * , 479 ( 2003 ) . m. nielsen and i. chuang , _ quantum information and computation _ c.u.p . 2000 however , it is important to realise that strong correlations can be obtained by non - local theories that in other respects might be considered quite classical @xcite . hence it requires much more discussion than we will present here to decide whether certain correlations are really _ quantum _ , or _ classically non - local_. n. j. cerf , n. gisin , s. massar , and s. popescu , phys . lett . * 94 * , 220403 ( 2005 ) . j. eisert , k. jacobs , p. papadopoulos and m.b . plenio , phys . rev . a * 62 * , 052317 ( 2000 ) . d. collins , n. linden and s. popescu , phys . a * 64 * , 032302 ( 2001 ) . s. virmani and m.b . plenio , phys . a * 67 * , 062308 ( 2003 ) . c. h. bennett , d. p. divincenzo , c. a. fuchs , t. mor , e. rains , p. w. shor , j. smolin , and w. k. wootters , phys . a * 59 * , 1070 ( 1999 ) . rains , ieee trans . theory * 47 * , 2921 ( 2001 ) . t. eggeling , k. g. h. vollbrecht , r. f. werner , and m. m. wolf , phys . * 87 * , 257902 ( 2001 ) . k. audenaert , m. b. plenio , and j. eisert , phys . lett . * 90 * , 027901 ( 2003 ) . werner , phys . a * 40 * , 4277 ( 1989 ) . l. masanes , phys . * 96 * , 150501 ( 2006 ) . brando , e - print arxiv quant - ph/0510078 l. masanes , e - print arxiv quant - ph/0510188 c. h. bennett , h. bernstein , s. popescu and b. schumacher , phys . a * 53 * , 2046 ( 1996 ) . v. vedral , m.b . plenio , m.a . rippin , and p.l . knight , phys . lett . * 78 * , 2275 ( 1997 ) . v. vedral and m.b . plenio , phys . a * 57 * , 1619 ( 1998 ) . lo and s. popescu , phys . a * 63 * , 022301 ( 2001 ) . nielsen , phys . lett . * 83 * , 436 ( 1999 ) . g. vidal , phys . lett . * 83 * , 1046 ( 1999 ) . d. jonathan and m.b . plenio , phys . lett . * 83 * , 1455 ( 1999 ) . l. hardy , phys . a * 60 * , 1912 ( 1999 ) . r. bhatia , d. jonathan and m.b . plenio , phys . lett . * 83 * , 3566 ( 1999 ) . g. vidal , d. jonathan and m. a. nielsen , phys . rev . a * 62 * , 012304 ( 2000 ) . t.m . cover and j.a . thomas , _ elements of information theory _ ( wiley interscience , new york , 1991 ) common synonyms for two - qubit maximally entangled states include _ ` singlet states ' , ` bell pairs ' _ or _ ` epr pairs'_. even though these terms strictly mean different things , we will follow this widespread abuse of terminology . rains , e - print arxiv quant - ph/9707002 p. hayden , m. horodecki , and b.m . terhal , , 6891 ( 2001 ) . a. kent , phys . lett . * 81 * , 2839 ( 1998 ) . e. rains , phys . a * 60 * , 173 ( 1999 ) ; ibid * 60 * , 179 ( 1999 ) . g. vidal and j.i . cirac , phys . 86 * , 5803 ( 2001 ) . m. horodecki , a. sen(de ) and u. sen , phys . a * 67 * , 062314 ( 2003 ) . bennett , d.p . divincenzo , j.a . smolin , and w.k . wootters , , 3824 ( 1996 ) . wootters , , 2245 ( 1998 ) . wotters , quant . inf . comp . * 1 * , 27 ( 2001 ) . v. vedral , m.b . plenio , k. jacobs and p.l . knight , phys . a * 56 * , 4452 ( 1997 ) . m. christandl and a. winter , j. math . phys * 45 * , 829 ( 2004 ) . shor , comm . phys . * 246 * , 453 ( 2004 ) . k.m.r . audenaert and s.l . braunstein , comm . phys . * 246 * , 443 ( 2003 ) . a. a. pomeransky , phys . a * 68 * , 032317 ( 2003 ) . j. smolin , f. verstraete and a. winter , phys . a * 72 * , 052317 ( 2005 ) f. verstraete , m .- a . martin - delgado and j.i . cirac , phys . 92 * , 087201 ( 2004 ) f. verstraete , m. popp , and j. i. cirac , phys . . lett . * 92 * , 027901 ( 2004 ) m. popp , f. verstraete , m. a. martin - delgado and j. cirac , phys . a. * 71 * , 042306 ( 2005 ) . pachos and m.b . plenio , phys . lett . * 93 * , 056402 ( 2004 ) a. key , d.k.k . lee , j.k . pachos , m.b . plenio , m. e. reuter , and e. rico , optics and spectroscopy * 99 * , 339 ( 2005 ) . a. harrow and m. nielsen , phys . a * 68 * , 012308 ( 2003 ) s. virmani , s.f . huelga , and m.b . plenio , phys . a. * 71 * , 042328 ( 2005 ) m. murao , m.b . plenio , s. popescu , v. vedral and p.l . knight , phys . a * 57 * , 4075 ( 1998 ) . h. aschauer , w. dr and h.j . briegel , phys . a * 71 * , 012319 ( 2005 ) . k. goyal , a. mccauley and r. raussendorf , e - print arxiv quant - ph/0605228 k. chen and h - k lo , quant - ph/0404133 . donald , m. horodecki , and o. rudolph , j. math . phys . , 4252 ( 2002 ) . plenio and v. vitelli , contemp . phys . * 42 * , 25 ( 2001 ) . plenio , phys . lett . * 95 * , 090503 ( 2005 ) . m. horodecki , open syst . * 12 * , 231 ( 2005 ) . shor , j.a . smolin and b.m . terhal , phys . lett . * 86 * , 2681 ( 2001 ) . d. divincenzo , m. horodecki , d. leung , j. smolin , and b. terhal phys . lett . * 92 * , 067902 ( 2004 ) . k. horodecki , m. horodecki , p. horodecki and j. oppenheim , phys . lett . * 94 * , 200501 ( 2005 ) . b. synak - radtke and m. horodecki , e - print arxiv quant - ph/0507126 m. horodecki , p. horodecki , and r. horodecki , phys . rev.lett . * 84 * , 2014 ( 2000 ) . s. virmani and m.b . plenio , phys . a * 288 * , 62 ( 2000 ) . j. eisert and m.b . plenio , j. mod . opt . * 46 * , 145 ( 1999 ) . k. zyczkowski and i. bengtsson , ann . phys . * 295 * , 115 ( 2002 ) . a. miranowicz and a. grudka , j. opt . b : quantum semiclass . optics * 6 * , 542 ( 2004 ) . d. yang , m. horodecki , r. horodecki and b. synak - radtke , phys . lett . * 95 * , 190501 ( 2005 ) . m. horodecki , p. horodecki , r. horodecki , j. oppenheim , a. sen(de ) , u. sen and b. synak , phys . a * 71 * , 062307 ( 2005 ) . plenio , s. virmani , and p. papadopoulos , j. phys . a * 33 * , l193 ( 2000 ) . k.g.h . vollbrecht and f. verstraete , phys . a * 71 * , 062325 ( 2005 ) . i. devetak and a. winter , proc . a * 461 * , 207 ( 2005 ) . a. s. holevo , ieee trans.info.theor . * 44 * , 269 ( 1998 ) . k. audenaert , f. verstraete and b. de moor , phys . a * 64 * , 052304 ( 2001 ) . vollbrecht and r.f . werner , phys . a * 64 * , 062307 ( 2001 ) . j. eisert , t. felbinger , p. papadopoulos , m.b . plenio and m. wilkens , phys . * 84 * , 1611 ( 2000 ) . b. terhal and k.g.h . vollbrecht , phys . * 85 * , 2625 ( 2000 ) . g. vidal , j. mod . opt . * 47 * , 355 ( 2000 ) . plenio and v. vedral , j. phys . a * 34 * , 6997 ( 2001 ) . m.j . donald and m. horodecki , physics letters a * 264 * , 257 ( 1999 ) . k. audenaert , j. eisert , e. jan , m.b . plenio , s. virmani , and b. demoor , phys . lett . * 87 * , 217902 ( 2001 ) . n. linden , s. popescu , b. schumacher and m. westmoreland , e - print arxiv quant - ph/9912039 e. galvo , m.b . plenio and s. virmani , j. phys . a * 33 * , 8809 ( 2000 ) . s. wu and y. zhang , phys . a * 63 * , 012308 ( 2001 ) . s. ishizaka , phys . lett . * 93 * , 190501 ( 2004 ) . s. ishizaka and m.b . plenio , phys . a * 71 * , 052303 ( 2005 ) s. ishizaka and m.b . plenio , phys . a * 72 * , 042325 ( 2005 ) . s. boyd and l. vandenberghe , _ convex optimization _ , cambridge university press 2004 . k. audenaert , b. demoor , k.g.h . vollbrecht and r.f.werner , phys . rev a * 66 * , 032310 ( 2002 ) . t. wei and p. goldbart , phys . a. * 68 * , 042307 ( 2003 ) . j. eisert , k. audenaert , and m.b . plenio , j. phys . a * 36 * , 5605 ( 2003 ) . k. horodecki , m. horodecki and p. horodecki , and j. oppenheim , phys . 94 * , 160502 ( 2005 ) . a. peres , phys . lett . * 77 * , 1413 ( 1996 ) . horodecki , phys . a * 232 * , 333 ( 1997 ) . m. horodecki , p. horodecki and r. horodecki , phys . lett . * 80 * , 5239 ( 1998 ) . k. zyczkowski , p. horodecki , a. sanpera , and m. lewenstein , phys . a * 58 * , 883 ( 1998 ) . j. lee , m.s . kim , y.j . park , and s. lee , j. mod . opt . * 47 * , 2151 ( 2000 ) . j. eisert , g. vidal and r.f . werner , , 32314 ( 2002 ) . r. alicki and m. fannes , e - print arxiv quantum - ph/0312081 s. ishizaka , phys . rev . a * 69 * , 020301(r ) ( 2004 ) . r. r. tucci , quant - ph/9909041 ; r. r. tucci , quant - ph/0202144 . g. vidal and r. tarrach , phys . a. * 59 * , 141 ( 1999 ) . m. steiner , phys . a * 67 * , 054305 ( 2003 ) . m. lewenstein and a. sanpera * 80 * , 2261 ( 1998 ) . r. nagel ` order unit and base norm spaces ' , in _ foundations of quantum mechanics and ordered linear spaces _ , eds . a. hartkmper and h. neumann , springer - verlag ( 1974 ) . o. rudolph , j. phys . a : math gen * 33 * , 3951 ( 2000 ) . o. rudolph , j. math . phys . * 42 * , 2507 ( 2001 ) . o. rudolph , quant - ph/0202121 . brando , phys . a * 72 * , 040303(r ) ( 2005 ) . this excludes for example the hydrogen atom from our considerations as it has a limit point in the spectrum at the ionization level . j. eisert , ch . simon and m.b . plenio , j. phys . a * 35 * , 3911 ( 2002 ) s. parker , s. bose and m.b . plenio , phys . a * 61 * , 032305 ( 2000 ) . wolf , g. giedke and j.i . cirac , phys . lett . * 96 * , 080502 ( 2006 ) . k. cahill and r. glauber , phys . rev . * 177 * , 1857 ( 1969 ) . see e.g. h. p. robertson , phys . rev . * 46 * , 794 ( 1934 ) ; a. serafini , phys . rev . lett . * 96 * , 110402 ( 2006 ) . r. simon , n. mukunda , b. dutta , phys . a. * 49 * , 1567 ( 1994 ) r. simon , phys . lett . * 84 * , 2726 ( 2000 ) . duan , g. giedke , j.i . cirac and p. zoller , lett . * 84 * , 2722 ( 2000 ) r.f . werner and m.m . wolf , phys . 86 * , 3658 ( 2001 ) . g. giedke , b. kraus , m. lewenstein , and j.i . cirac , phys . lett . * 87 * , 167904 ( 2001 ) g. giedke , l .- m . duan , j.i . cirac and p. zoller , quantum . * 1 * , 79 ( 2001 ) . j. eisert and m.b . plenio , phys . lett . * 89 * , 097901 ( 2002 ) . j. eisert , s. scheel , and m.b . plenio , phys . lett . * 89 * , 137903 ( 2002 ) . g. giedke and j. i. cirac , phys . a * 66 * , 032316 ( 2002 ) . the no - go theorem for gaussian entanglement distillation is actually quite subtle , as even defining gaussian distillation is non - trivial . consider for example the version of the no - go theorem proven in @xcite - there entanglement distillation is defined as a process which allows you to reach a two - mode squeezed state with arbirarily high squeezing . although @xcite show that this is not possible , from an operational perspective one can use gaussian operations to improve the quality of the entanglement present . for instance , consider a bi - partite four - mode state @xmath25 which consists of a two - mode highly entangled state tensored with a product pure state . given two copies of @xmath25 , alice and bob can merely throw away the product components and are left with a single four - mode state that possesses greater utility for quantum information processing . hence according to an operational definition of entanglement distillation ( say improvement of teleportation fidelity ) , it is possible to distill gaussian entanglement using gaussian operations from gaussian states of 2 @xmath484 2 modes . this apparent contradiction arises because there are different ways in which one can quantify the increase of entanglement of quantum states - in principle this subtle distinction may also be an issue in the finite dimensional regime . g. giedke , j. eisert , j.i . cirac and m.b . plenio , quantum . comput . * 3 * , 211 ( 2003 ) . j. fiurek , phys . lett . * 89 * , 137904 ( 2002 ) . r. simon , e.c.g . sudarshan , and n. mukunda , phys . a * 36 * , 3868 ( 1987 ) . arvind , b. dutta , n. mukunda , and r. simon , pramana * 45 * , 471 ( 1995 ) . j. williamson , am . j. math . * 58 * , 141 ( 1936 ) . g. giedke , m.m . wolf , o. krger , r.f . werner , and j.i . cirac , phys . lett . * 91 * , 107901 ( 2003 ) m.m . wolf , g. giedke , o. krger , r.f . werner and j.i . cirac , phys . a * 69 * , 052320 ( 2004 ) . shirokov , e - print quant - ph/0411091 . k. audenaert , j. eisert , m.b . plenio and r.f . werner , phys . a * 66 * , 042327 ( 2002 ) . plenio , j. eisert , j. dreiig , m. cramer , phys . lett . * 94 * , 060503 ( 2005 ) . m. cramer , j. eisert , m.b . plenio and j. dreissig , phys . a * 73 * , 012309 ( 2006 ) r. raussendorf and h .- j . briegel , phys . lett . * 86 * , 5188 ( 2001 ) . d. j.wineland , j. j. bollinger , w.m . itano , f. l. moore , and d. j. heinzen , phys . a * 46 * , r6797 ( 1992 ) . s.f huelga , c. macchiavello , t. pellizzari , a.k . ekert , m. b. plenio , j.i . cirac , phys . lett . * 79 * , 3865 ( 1997 ) . m. murao , d. jonathan , m.b . plenio and v. vedral , phys . a * 59 * , 156 ( 1999 ) . m. hillery , v. buzek , and a. berthiaume , phys . a * 59 * , 1829 ( 1999 ) . a. karlsson , m. koashi , and n. imoto , phys . a * 59 * , 162 ( 1999 ) m. murao , d. jonathan , m.b . plenio and v. vedral , phys . a * 61 * , 032311 ( 2000 ) . w. dr , g. vidal , and j. i. cirac phys . a * 62 * , 062314 ( 2000 ) . f. verstraete , j. dehaene , b. de moor , and h. verschelde , phys . a * 65 * , 052112 ( 2002 ) . m. horodecki , j. oppenheim , and a. winter , nature * 436 * , 673 ( 2005 ) . t. wei , m. ericsson , p. goldbart and w. munro , quant . inf . comp . * 4 * , 252 ( 2004 ) . r. werner and a. holevo , j. math . phys . * 43 * , 4353 ( 2002 ) . see e.g. a. defant and k. floret , ` _ tensor norms and operator ideals _ ' , north - holland ( 1993 ) . v. coffman , j. kundu , and w. wootters , phys . a. * 61 * , 052306 ( 2000 ) . t. osborne and f. verstraete , quant - ph/0502176 . the web site at the universitt braunschweig _ http://www.imaph.tu-bs.de/qi/problems/3.html _ contains an interesting review of this problem and further references . j. eisert and h. j. briegel , phys . a * 64 * , 022306 ( 2001 ) . a. miyake , phys . a. * 67 * , 012108 ( 2003 ) . p. levay , journal of physics a * 38 * , 9075 ( 2005 ) . m. horodecki , private communication . divincenzo , r. jozsa , p.w . shor , j.a . smolin , b.m . terhal , and a.v . thapliyal , phys . a * 61 * , 062312 ( 2000 ) . w. dr , j.i . cirac , m. lewenstein and d. bru , phys . a * 61 * , 062313 ( 2000 ) .	we review the theory of entanglement measures , concentrating mostly on the finite dimensional two - party case . topics covered include : single - copy and asymptotic entanglement manipulation ; the entanglement of formation ; the entanglement cost ; the distillable entanglement ; the relative entropic measures ; the squashed entanglement ; log - negativity ; the robustness monotones ; the greatest cross - norm ; uniqueness and extremality theorems . infinite dimensional systems and multi - party settings will be discussed briefly .
the soft x - ray transients ( sxts ) , sometimes referred to as `` x - ray novae '' , are low - mass x - ray binary systems in which a late - type star loses material via an accretion disc to a neutron star or black hole . they undergo brief outbursts , typically lasting a few months with a recurrence time of the order of decades ( see tanaka & shibazaki 1996 and van paradijs & mcclintock 1995 for reviews of their x - ray and optical properties ) . the x - ray transient v616 mon ( = a062000 ) was discovered in 1975 ( elvis et al . , 1975 ) . when the system faded into quiescence the binary nature of the system was established ; the orbital period was determined to be 7.75 h and the secondary star was found to be a k dwarf ( oke 1977 ; mcclintock et al . mcclintock & remillard ( 1986 ) then went on to measure the orbital radial velocity curve of the secondary star , thereby obtaining a firm _ minimum _ mass for the compact object ( i.e. the mass function ) of 3.08 . this result immediately placed v616 mon amongst the best black hole candidates ( see van paradijs & mcclintock 1995 ) . high resolution optical spectroscopy can provide further constraints on the system masses , as a measurement of the rotational broadening of the secondary star absorption features leads directly to the _ ratio _ of the component masses ( mrw ) . but the binary inclination , can only be determined by exploiting the ellipsoidal modulation of the secondary star , i.e. the variations caused by observing the differing aspects of the gravitationally distorted star as it orbits the compact object ( see shahbaz , naylor & charles 1993 and references within ) . in v616 mon these variations have been observed at both optical ( mcclintock & remillard 1986 ; haswell et al . 1993 ) and infrared ( ir ) wavelengths ( shahbaz , naylor & charles 1994 ; hereafter snc ) . snc determined the binary inclination for v616 mon by modelling the ir ellipsoidal modulation of the secondary star . they obtained @xmath2 , which when combined with the value for the binary mass function and mass ratio resulted in a most probable mass for the compact object of @xmath310 ( 5.117.1 ; 90 percent confidence ) . there are several complications in such a determination of the binary inclination ; inevitably the optical light from the mass donor is diluted by the flux from the quiescent accretion disc . also , the ellipsoidal light curve may be distorted by other variable contributions to the observed flux ; such as the bright spot , star spots and variable disc emission including the superhump phenomenon . underestimating the diluting flux leads to an underestimate in the binary inclination . at optical wavelengths it is clear that the accretion disc contributes a significant amount of flux to the optical light curves , which makes any interpretation of these light curves somewhat dubious . this can be seen from the scatter and asymmetric maxima observed in the optical light curves of v616 mon by mcclintock & remillard ( 1986 ) . the veiling by the continuum emission from the accretion disc has been measured by mrw . they find that 94 percent or more of the optical flux close to h@xmath4 comes from the secondary star , a fraction which rises with increasing wavelength , while the veiling ratio falls . therefore one would expect the veiling by the accretion disc to be very small or even negligible in the @xmath0 band . haswell et al . ( 1993 ) attempted to deduce the binary inclination from modelling their optical multi - colour light curves with a combination of an ellipsoidal variation and a grazing eclipse , assuming a large circular disk . the inclination derived from the grazing eclipse constraint is in serious disagreement with that determined from the ellipsoidal variations in the ir light curves ( snc ) . however , it is known that in low - mass x - ray binaries and cataclysmic variables the outer , cool regions of the accretion disc can be a strong source of ir radiation ( e.g. beall et al . 1984 ; berriman , szkody & capps 1985 ) . consequently , this introduces doubts on the snc assumption that the observed ir flux has very little contamination by the accretion disc . this would then imply that any interpretation of the ellipsoidal variations of the secondary star in the ir might be misleading . nevertheless , for the sxt v404 cyg we have already shown directly that the ir accretion disc contamination is small ; it only affects the implied mass of the black hole by at most 2 ( shahbaz et al . , 1996 ) . we therefore decided to obtain through ir spectroscopy a direct measurement of any contamination of the ir flux by the accretion disc in v616 mon . from this we can determine the effect it has on the interpretation of the ellipsoidal variation of the secondary star , and hence on the component masses . we obtained @xmath0-band ( 2.02.5@xmath5 ) spectra of v616 mon using the cooled grating spectrometer ( cgs4 ) on the united kingdom 3.8-m infrared telescope on mauna kea , during the night of 1997 november 13 . the 40 l / mm grating was used with the 150 mm camera and the 256@xmath6256 pixel insb array . the bright star bs2233 was also observed through - out the night in order to remove telluric atmospheric features . a journal of observations is presented in table 1 . in order to minimise the effects of bad pixels , the standard procedure of oversampling was used . the spectra were sampled over two pixels by mechanically shifting the array in 0.5 pixel steps in the dispersion direction , giving a full width half maximum resolution of 47 (@xmath3610 at 2.31@xmath5 ) . we employed the non - destructive readout mode of the detector in order to reduce the readout noise . the slit width was 1.2 arcseconds which corresponds to 2 pixels on the detector . in order to compensate for the fluctuating atmospheric emission lines we took relatively short exposures and nodded the telescope primary so that the object spectrum switched between two different spatial positions on the detector . throughout the observing run the slit orientation was north to south in the spatial direction . the cgs4 data reduction system performs the initial reduction of the 2d images . these steps include the application of the bad pixel mask , bias and dark subtraction , flat field division , interlacing integrations taken at different detector positions , and co - adding and subtracting the nodded images ( see daly & beard 1994 ) . extraction of the 1d spectra , wavelength calibration , and removal of the telluric atmospheric features was then performed using iraf . a more detailed description of the data reduction procedure is provided in shahbaz et al . ( 1996 ) . first the individual spectra of v616 mon were averaged using the variance in the spectrum as the weight . this resulted in a final summed spectrum which had a signal - to - noise ratio of @xmath3 30 . we then cross correlated all the summed v616 mon spectrum with the template star spectrum in order to determine the velocity shift . given the poor resolution of the data this was not done on the individual spectra of v616 mon before they were averaged . we applied the appropriate velocity shift and then binned all the spectra onto a uniform velocity scale . figure 1 shows the summed spectrum of v616 mon and the k3v template star spectrum , normalised and shifted for clarity . the v616 mon spectrum shows the cai triplet and @xmath7co bands in absorption and doubled - peaked br@xmath8 in emission . the peak separation of the br@xmath8 emission line arising from the accretion disc is 1204@xmath9156 km s@xmath10 . note that this is comparable with the peak - to - peak separation in the h@xmath4 and h@xmath11 emission lines ( mrw ) . the equivalent width ( ew ) of br@xmath8 in the k3v star spectrum is [email protected] whereas in the residual spectrum [ i.e. the spectrum of the accretion disc ; see figure 1 and section 3 ] it is [email protected] . the @xmath7co bands are the strongest features and will be used in the next section to determine the fraction of light arising from the secondary star . we first binned the v616 mon and template star spectra onto a logarithmic wavelength scale using a @xmath12 interpolation scheme to minimize data smoothing ( stover et al . we then optimally subtracted the template star from the v616 mon spectra using the standard procedure ( see mrw ) as follows . taking the accretion disc contribution as a flat continuum , we fit the v616 mon spectra with this plus a variable fraction of the template star . the fraction of light from the template star , ( @xmath13 ) , is adjusted to minimise the residual scatter between the spectra . the scatter is measured by carrying out the subtraction and then computing the between the residual spectrum and a smoothed version of the residual spectrum . this is done to remove any large - scale structure . the for different fractions of the template star is computed and fitted with a parabola , and the fraction at the minimum value of is taken to be the best fit . the above analysis was performed only on the 2.292@xmath5 @xmath7co(2,0 ) bandhead in the summed spectrum of v616 mon . other bandheads were not used because of poor atmospheric subtraction and bad pixels . figure 1 shows the summed spectra of v616 mon along with the template k3v star . the residual light spectrum ( i.e. the accretion disc light ) is also shown . in addition to using hd42606 ( k3v ) we also used template stars of other spectral types [ hd42606 ( k0v ) , hd42606 ( k5v ) and hd42606 ( k7v ) ] in order to determine the sensitivity of @xmath13 to the assumed spectral type . we find that for the summed spectra of v616 mon , the minimum value for ( 1.01 , a good fit ) is obtained using the k3v secondary star , which contributes 75@xmath917 percent of the flux in the ir ( the disc fraction is 25 per cent ) . we also fitted the region used in the v616 mon spectrum for the optimal subtraction with a constant and obtained a much worse fit with a of 2.7 thereby showing that the absorption feature is real as the 99.9 per cent level . the large uncertainties we obtain for the accretion disc light suggest that our result is consistent with the secondary star contributing all the light to the observed flux . in order to determine a lower limit to the fraction of light from the secondary , we simulated a spectrum in which _ all _ the light arises from the secondary star . this was done using the k3v template star and adding noise to the data such that the signal - to - noise was the same as the v616 mon spectrum . we then performed the optimal subtraction and determined the 2-@xmath1 per cent lower limit to the fraction of light arising from the secondary to be 73 per cent . this value is comparable with the fraction of light arising from the secondary star deduced from our spectrum of v616 mon , suggesting v616 mon most probably contains a companion star which contributes all the light in the ir . the quiescent optical light curves of v616 mon show a double - humped ellipsoidal modulation due to the changing projected area of the roche - lobe filling mass donor star . maxima occur at quadrature , corresponding to maximum projected area , and are hence expected to be symmetric . minima occur at orbital conjunctions ; in general they have differing depths due to limb- and gravity - darkening . the light from the secondary star is diluted by the flux from the accretion disc . also the ellipsoidal light curve may be distorted by other variable contributions to the total light ; the bright spot associated with the impact of the mass transfer stream on the edge of the disc ( mcclintock & remillard 1990 ) ; star spots on the secondary ; and variable disc emission , including the superhump phenomenon ( warner 1995 ) . recently leibowitz , herman & orio ( 1998 ) have collated the optical light curves of v616 mon obtained over the last 7 years . they find that the depth of the maxima and minima of the light curves vary with time . they observe a long term photometric behaviour of a few hundred days with a peak to peak amplitude of 0.3 mag . they suggest that the minimum in the light curve corresponding to when the red dwarf lies between the observer and the compact object ( orbital phase 0.0 ) changes depth with time . since in the standard precessing disc model one does not expect this minimum to change depth significantly ( as at this phase the secondary star is least affected by x - ray illumination ) they conclude that they can not explain the variations in terms of a simple geometrical precessing accretion disc model ( heemskerk & van paradijs 1989 ) . however , it should be noted that the main thrust of their conclusion lies in the interpretation of the varying component in the light curves . in the precessing disc model the depth of either minimum can vary with time , depending on the tilt of the accretion disc ( heemskerk and van paradijs 1989 ) . also they have assumed that the maximum in the light curve ( orbital phase 0.75 ) with respect to which the minima is measured remains constant . this is probably not the case as the observed optical light near this orbital phase is contaminated by the bright spot . optical spectroscopy of v616 mon shows the presence of a bright spot ( mrw ) . bright spots have been seen in the optical light curves of cataclysmic variables with similar mass ratios such as z cha and oy car ( wood et al . 1986 , 1989 ) where it is observed between orbital phase 0.6 and 1.1 . variations in the optical flux emitted by the bright spot as the result of clumpy mass transfer from the secondary star could easily give rise to variability in the optical flux observed at this orbital phase . in the ir the effects of the variability discussed above are much less ( snc ) . the ir light curve of v616 mon shows equal maxima ; this is what is expected if the ir variations are due solely to the ellipsoidal modulation of the secondary star . various authors have pointed out that the double - peaked emission line profiles can be interpreted as arising from an accretion disk viewed at high inclination . however , it should be noted that a double - peaked emission line profile can also arise from a system with an inclination as low as 15@xmath14 ( see the models of horne & marsh 1986 ) . assuming that the double - peaked lines arise entirely from the disk , we can estimate the binary inclination of v616 mon by measuring the separation of the br@xmath8 emission - line peaks . the keplerian velocity of the outer edge of the disk ( @xmath15 ) is given by @xmath16 . combining this with kepler s third law , paczynski s ( 1971 ) formula for the roche lobe radius , and using the fact that the accretion disk fills 50 per cent of the compact object s roche - lobe ( mrw ) , gives @xmath17 km s@xmath10 , where @xmath18 is the mass of the black hole ( in solar masses ) , and @xmath19 is the orbital period ( in hours ) . the separation of the br@xmath8 emission - line peaks measured from the summed spectrum of v616 mon is 1204 km s@xmath10 , implying a projected velocity of the outer edge of the accretion disk of @xmath20 602 km s@xmath10 . using the above formula with @xmath19=7.75 hrs ( mrw ) and @xmath21 ( snc ) gives a @xmath22 , which agrees well with that obtained by snc . in cataclysmic variables there is observational evidence that the accretion disc contamination in the @xmath0-band is significant . the eclipse light curves of the dwarf nova oy car , for example show that during quiescence the accretion disc can contribute about 30 percent of the flux at 2.2@xmath5 ( sherrington et al . ir spectra of cataclysmic variables also show emission lines arising from the optically thin gas in the accretion disc ( ramseyer et al . 1993 ; dhillon & marsh 1995 ) , such as hei ( 2.0587@xmath5 ) and br@xmath8 ( 2.1655@xmath5 ) . in contrast , the x - ray transients v404 cyg ( shahbaz et al . 1996 ) and v616 mon show only br@xmath8 in emission . it should also be noted that the mass accretion rate during quiescence in the x - ray transients is a factor of 10 lower than that in dwarf novae . one expects the ew of the emission lines arising from the accretion disc to decrease as the orbital period of the binary increases . this is simply because larger systems will have larger , cooler accretion discs . if one looks at the h@xmath4 ew in the sxts , then one can find some evidence for this type of correlation ; for nova per 1992 the h@xmath4 ew is 205 , whereas for the larger systems such as nova mus 1991 and v404 cyg it is 50 and 40 rrespectively . also note that in all the sxts the disc contamination near h@xmath4 is in the range 6 - 16 percent , i.e. it is small despite the large h@xmath4 ew . in v404 cyg the br-@xmath8 ew of the accretion disc is 2.7 . one expects the br-@xmath8 ew of the accretion disc in the much shorter system v616 mon to be higher ; this is what is observed ( see section 2 ) . snc obtained an ir light curve of v616 mon which showed a double humped feature characteristic of the ellipsoidal variations of the secondary star . they modelled these variations assuming all the ir flux was arising from the secondary star , and determined the most probable mass of the compact object to be 10 . justification for this assumption comes from the analysis of the ir light curve of the transient cen x4 ( shahbaz et al . 1993 ) . the mass of the compact object in cen x4 is consistent with that of a canonical neutron star ; which the compact object must be because of the type i x - ray bursts observed during outburst ( matsuoka et al . this provides indirect evidence that the contribution of the accretion disc to the observed ir flux is very small , at least from cen x4 . this may also be the case for v616 mon ; an upper limit to the accretion disc contribution to the ir flux being 27 percent ( section 3 ) . the effects of any accretion disc contamination to the observed ir flux will be to dilute the actual ellipsoidal modulation , making the observed modulation smaller than the _ true _ value . since the amplitude of the ellipsoidal modulation is correlated with the binary inclination ( large amplitudes imply a high binary inclination ) , this means that modelling a contaminated light curve will underestimate @xmath23 . we have modelled the amplitude of the ellipsoidal variations as a function of @xmath23 . we used the same parameters as snc : t@xmath24=4000 k , @xmath25=14.9 ( mrw ) , a gravity darkening exponent of 0.08 ( lucy 1967 ) , and the limb darkening coefficient from al - naimiy ( 1978 ) . figure 2 shows the effect of differing amounts of contamination in the ir light curves on the binary inclination . if we take the 2-@xmath1 limit to the disc contamination of 27 percent , we find that @xmath23 increases by 7 degrees and the mass of the black hole decreases by 3.6 ( 2-@xmath1 limit ) . note that this extreme value for @xmath23 is still lower than that obtained by haswell et al . ( 1993 ) , and the implied mass of the compact object ( @xmath3 6.4 ) still substantially exceeds the canonical maximum mass of a neutron star ( 3.2 ; rhoades & ruffini 1974 ) . we have determined the contamination of the observed ir flux by the accretion disk in v616 mon . we obtained ir spectra of v616 mon which when optimally subtracted from the secondary star spectrum ( k3v star ) allows us to determine the 2-@xmath1 upper limit to the accretion disc contribution to be less than 27 percent . we find that this only increases the determination of the binary inclination by less than 7 degrees ; the mass of the black hole is decreased by less than 3.6 . we would like to thank t. geballe for useful discussions . the data reduction was carried out using the iraf and ark software packages . the united kingdom infrared telescope is operated by the royal observatories on behalf of the uk particle physics and astronomy research council .	we have obtained for the first time @xmath0-band infrared spectra of the soft x - ray transient v616 mon ( = a062000 ) . we determine the 2-@xmath1 upper limit to the fraction of light arising from the accretion disc to be 27 percent . the effect this has on the binary inclination , determined from modelling the infrared ellipsoidal variations is to increase it by less than 7 degrees and decrease the mass of the black hole by less than 3.6 .
synchronization phenomena have been intensely studied for decades , in part because of the roles such phenomena play in chemical systems , laser arrays , cellular biology models , and neural networks to name just a few ( see refs . @xcite for extensive reviews ) . one of the most extensively studied models is that proposed by kuramoto in 1975 @xcite , a model that has become paradigmatic for the description of many synchronization phenomena . originally the model was applied to an interacting population of oscillators with randomly distributed frequencies . when the interaction is sufficiently strong , most of the units in the array synchronize their dynamics to a single frequency which may differ from the natural frequency of any one of the synchronized oscillators , and also to equal phases . many variants of the original model have been introduced over the years to study different effects in different physical and biological systems , too many to list here ( for an extensive review , see ref . we specifically mention the inclusion of fluctuations , because of their central role in our studies . noise leads to disorder , so in the presence of noise the interactions that in its absence may be strong enough to lead to frequency or to phase synchronization must in general be stronger for synchronization to occur . in all of these models , the form and range of the interactions has varied greatly in the literature . beyond the kuramoto model , many different models for synchronization have been proposed , ranging from arrays of continuous oscillatory and excitable units to discrete models . for instance , over the past decade coupled maps have attracted a great deal of attention @xcite . recently , arrays of coupled stochastic units each with a discrete set of states but , in contrast with maps , with continuous time have increased in popularity as a simpler paradigm for synchronization @xcite . even though these discrete - state oscillator models may be motivated by discrete processes ( for example , protein degradation @xcite ) , it has been claimed that they can also be used to model a coarse - grained phase space of continuous noisy oscillators . for instance , prager et al . @xcite established a link between a globally coupled ensemble of excitable units described by the fitzhugh - nagumo equations with additive white noise , and a coupled array of 3-state non - markovian stochastic oscillators . our own work has focused on arrays of 2-state and of 3-state stochastic oscillators . the transitions between the states of individual units are governed by a rate process . this rate process might be markovian @xcite or might involve distributed delays ( such as , for instance , a refractory period ) @xcite . interactions among units in our model appear as a dependence of the transition rates of a particular unit on the states of the other units to which it is coupled . the goal of the work presented herein is to address the following two questions : ( 1 ) under what conditions can we describe the dynamics of kuramoto - like coupled noisy oscillators as periodic continuous - time markov chains ? in other words , when can we model continuous - phase stochastic dynamics as discrete - phase models in which the transitions between the discrete states are governed by memoryless rate processes ? ( 2 ) is there a lower limit to the discretization of the continous noisy oscillators ? in other words , how many discrete states are necessary to capture the essential synchronization features of the continuous system ? the popularity of three - state models leads us to explore whether the synchronization properties of coupled three - state markovian units in any way capture those of the continuous oscillator system . to arrive at some answers to these questions , in sec . [ sec : review ] we present the continuous phase model that is the starting point of our analysis . it is an array of kuramato - like oscillators with additive noise and a generalized nonlinear interaction . we start with the full amplitude equations , but will always work in the limit where the phase equations alone provide a valid description of the important dynamics . for the sake of simplicity we consider a globally coupled ensemble of identical oscillators ( all with the same natural frequency ) , and thus focus on the phase synchronization phenomenon . in sec . [ sec : coarsegraining ] we perform the coarse - graining of the phase space and discuss the conditions under which the dynamics can be modeled as a periodic markov chain . here we also discuss the questions associated with the three - state systems . finally , in sec . [ sec : summary ] we present our concluding remarks . we also include two appendices with technical details of our calculations . our starting point is an ensemble of @xmath0 identical noisy oscillators described by the complex time - dependent dimensionless amplitudes @xmath1 , with @xmath2 . these amplitudes are governed by the equations of motion @xmath3 the overdot indicates a derivative with respect to time . @xmath4 is a real positive parameter that governs the internal dynamics of each oscillator . for the function @xmath5 , which describes this internal dynamics , we take the normal form of a supercritical hopf bifurcation , @xmath6 for simplicity , we take all parameters to be real and have scaled out irrelevant constants . the oscillators are identical , and we have removed the natural frequency of oscillation of each unit , that is , we are working in a moving framework . in the usual language of the kuramoto model , the frequency distribution of the oscillators is @xmath7 , where the @xmath8-function is appropriate for the continuous variable @xmath9 ( and below also for the continuous time @xmath10 ) . therefore , the internal dynamics of each oscillator tends to set @xmath11 , with an arbitrary phase . the second term on the right hand side of eq . ( [ amplitudeosc ] ) accounts for the interaction between the oscillators . the coupling strength is quantified by @xmath12 . the interaction is assumed to be global ( all - to - all interaction ) , with the customary kuramoto order parameter given by the average amplitude as a function of time , @xmath13 the original kuramoto model @xcite is recovered if we set @xmath14 equal to @xmath15 , so that the global interaction is given by @xmath16 . the function @xmath17 accounts for a nonlinear interaction between the oscillators via @xmath18 . the advantage of including a general nonlinear function @xmath17 in the interaction will be clear when we subsequently perform the coarse - graining operations . the third term on the right hand side of eq . ( [ amplitudeosc ] ) is a complex additive noise of intensity @xmath19 . this term models the fluctuations . the noise is of the form @xmath20 where @xmath21 and @xmath22 are independent real gaussian white noises of zero mean and correlation functions @xmath23 @xmath24 here @xmath25 is the kronecker delta appropriate for the discrete variable @xmath26 . we note that the form of eq . ( [ amplitudeosc ] ) respects the phase invariance , that is , the equation is invariant under the transformation @xmath27 with @xmath28 constant , but the equation is otherwise quite general . we consider the parameter range @xmath29 and @xmath30 , so that the time scale of the internal dynamics of each oscillator dominates over ( i.e. is shorter than ) that of the interactions between the oscillators . then , after a fast transient defined by the internal dynamics , we have that @xmath31 . after that , the phase of each oscillator varies as a function of time on a slower time scale defined by the interactions ( albeit with very rapid fluctuations ) . on this longer time scale we can write @xmath32 . the dynamics specified by eq . ( [ amplitudeosc ] ) can then be reduced to the phase equation @xcite @xmath33 here we have defined a kuramoto order parameter @xmath34 which follows directly from eq . ( [ order1 ] ) , @xmath35 from which we extract the real phase variable @xmath36 , and @xmath37 where @xmath38 $ ] and @xmath39 $ ] are also real . the noise @xmath40 is again gaussian and white , with zero mean and correlation function that follows directly from eqs . ( [ zeta1 ] ) and ( [ zeta2 ] ) , @xmath41 the order parameter @xmath34 can be written as @xmath42 where we have introduced the density of oscillators with phase @xmath43 , @xmath44.\ ] ] in the thermodynamic limit @xmath45 , @xmath46 where @xmath47 is the probability that the phase of an oscillator lies in the interval @xmath48 $ ] at time @xmath10 . in the thermodynamic limit , the stochastic phase equation ( [ phaseosc ] ) ( which is an equation of the langevin form ) can be replaced by a nonlinear fokker - planck equation ( see ref . @xcite for a detailed derivation using the path integral formalism ) , @xmath49\right\ } , \label{nlfpe}\ ] ] where the second derivative term on the right is the diffusion term , and where the drift contains @xmath50 = f(r\left[\rho\right])\sin\left(\psi\left[\rho\right ] - \phi\right ) , \label{5b}\ ] ] with @xmath51e^{i\psi\left[\rho\right ] } \equiv \int_{0}^{2\pi } \rho\left(\phi , t\right)e^{i\phi } d \phi.\ ] ] since all of our oscillators have the same frequency ( @xmath52 in the moving frame ) , synchronization in this framework corresponds to the tendency of the oscillators to have the same phase . the desynchronized state corresponds to a uniform distribution of phases , @xmath53 this choice obeys the normalization condition @xmath54 the dynamics described by the phase equation ( [ phaseosc ] ) contains two competing trends : the fluctuations , which tend to desynchronize the system and stabilize the state described in eq . ( [ eq : desync ] ) , and the attractive interactions that tend to synchronize the system . when the coupling strength is weaker than a critical value ( @xmath55 ) , the desynchronized state is stable , while an interaction stronger than this critical value ( @xmath56 ) destabilizes the desynchronized state , leading to a phase transition to synchronization . calculations of the critical point have been widely documented in the literature @xcite , and we see no point in repeating them here . instead , here we consider the continuous limit of the coarse - graining procedure presented in sec . [ sec : coarsegraining ] and detailed in appendix a. this limit is in turn detailed in appendix b , where we show that this critical value is given by @xmath57 as would be expected , stronger fluctuations require a stronger interaction for synchronzation to occur , that is , the critical value of the coupling parameter increases with increasing noise intensity @xmath19 . in the thermodynamic limit the absence of synchronization is associated with a vanishing order parameter , @xmath58 . away from the thermodynamic limit @xmath34 fluctuates to values away from zero because of the finite number of oscillators , but remains small as long as @xmath0 is not exceedingly small . the occurrence of self - organization is characterized by a non - vanishing order parameter . near the critical point ( [ criticalk-1 ] ) , the mean - field evolution of the order parameter can be approached via the normal form ( see refs.@xcite and appendix b ) @xmath59 with @xmath60 where @xmath61 . for the original kuramoto model @xcite , @xmath62 , the transition to synchronization is always supercritical because @xmath63 . the presence of a nonlinear interaction @xmath64 might induce a change in sign of @xmath65 and might therefore lead to a subcritical transition . the function @xmath64 affects the critical coupling strength in eq . ( [ criticalk-1 ] ) simply as a renormalization of this critical value . the main role of @xmath64 is to increase or decrease the dispersion of the phases around a common value in the synchronized regime . the qualitative effect of including the function @xmath17 can be seen in the interaction term in the phase equation ( [ phaseosc ] ) . in the synchronized phase a function @xmath64 that increases with increasing @xmath66 increases the attractive interactions among the oscillators , leading to a decrease of the dispersion of phases around a common value . in contrast , if @xmath64 decreases with increasing @xmath66 , this tendency will be weakened and , since fluctuations lead to a greater dispersion of the phases , the synchronization will be coarser . a decreasing function might thus be a better candidate for a successful increasingly coarse coarse - graining procedure . + for numerical simulations we must choose a particular form for the function @xmath64 . as stated above , our goal of coarse - graining the phase space is better achieved if we choose a decreasing function of @xmath66 . we have tested a number of different functions and have determined that the particular form and parameters in it are not as important as is simply the choice of almost any decreasing function of @xmath66 . we use the exponentially decreasing function @xmath67 with @xmath68 a positive parameter . note that when @xmath69 , @xmath70 and we recover the standard kuramoto model . in the upper panel of fig.[fig01 ] we show , after a transient , the state of an ensemble of @xmath71 interacting oscillators governed by eq . ( [ amplitudeosc ] ) . the points represent the complex amplitude of each oscillator in the complex plane . as expected for large @xmath4 ( @xmath72 in all our simulations ) , after a transient the amplitudes settle around a circle of unit radius ( red line ) . we have chosen a coupling strength @xmath73 ( see caption ) so that the array of oscillators is in the synchronized regime . the system exhibits coarse phase clustering , that is , the synchronization is not perfect . this is attributed to the fluctuations that tend to disperse the phases , and to the rapid decrease of the function ( [ fmodel-1 ] ) . the bottom panel of fig.[fig01 ] displays the structure of a phase cluster . the figure shows the histogram of the phases of the oscillators in a cluster obtained from a numerical simulation of eq . ( [ amplitudeosc ] ) . the red continuous curve is a numerical computation of the steady state of the nonlinear fokker - planck equation ( [ nlfpe ] ) . this steady state fits the simulation results very well even though the nonlinear fokker - planck equation relies on two approximations : the reduction of the amplitude and phase equation eq . ( [ amplitudeosc ] ) to eq . ( [ phaseosc ] ) for only the phases , and the mean field assumption that @xmath74 although we have a finite number of oscillators . the dashed curve shows an analytic estimation of the steady state solution of the nonlinear fokker - planck equation , obtained from the normal form approach ( see appendix b ) : @xmath75 here the phase @xmath36 is a constant that defines the phase - cluster location in the unit circle and is fixed by the initial condition and the particular realization of the fluctuations . for finite @xmath0 , the location of this phase in the unit circle is random . this random phase is not captured by the steady state solution of eq . ( [ nlfpe ] ) because in the mean field approximation ( [ mfapprox ] ) the limit @xmath45 is applied before the limit @xmath76 . in general these limits do not conmute , and finite size effects are lost in the mean field equation eq . ( [ nlfpe ] ) . in any case , sufficiently large @xmath0 allows for the formation of a phase - cluster which is associated with a non - vanishing value of the order parameter @xmath34 . this fact is independent of the order of the limits . instead of the continuous phase @xmath43 , we move on to a discrete set of @xmath77 groups of phases discretely and equidistantly centered around the circle of unit radius seen in the top panel of fig . [ fig01 ] . that is , @xmath78~ \rightarrow ~ \phi \in \left\{j\delta\phi\right\}_{j=0}^{m-1},\ ] ] where @xmath79 we then convert the nonlinear continuous fokker - planck equation ( [ nlfpe ] ) to the finite difference equation @xmath80 with periodic boundary conditions @xmath81 and @xmath82 . here @xmath83 represents the probability of finding a unit in the phase group @xmath84 ( henceforth called state @xmath85 ) , at time @xmath10 , and @xmath86 with @xmath87 this is the associated version of the mean field order parameter . note that in our new discrete state space the normalization of the probability is [ cf . ( [ normcont ] ) ] @xmath88 the desynchronized state now is @xmath89 for all @xmath85 . equation ( [ nlme-1 ] ) can be written in the form @xmath90 where we define the transition rates @xmath91 equation ( [ nlme-2 ] ) is a master equation for a discrete state system , where the transitions between the @xmath77 states are governed by the rates ( [ rate ] ) . only nearest - neighbor transitions ( @xmath92 ) are allowed . we stress that eq . ( [ nlme-2 ] ) is nonlinear because the rates ( [ rate ] ) depend on the @xmath93 via @xmath94 [ see eqs . ( [ bound ] ) and([opmf ] ) ] . we have arrived at the following point in this development : the stochastic dynamics of a single kuramoto oscillator described by the continuum equation eq . ( [ amplitudeosc ] ) may be converted to the problem of a continuous - time markov chain of @xmath77 states , with periodic boundary conditions . these states model the internal structure of each stochastic oscillator . the interaction between the oscillators is contained in the dependence of the rates ( [ rate ] ) on the global state of the system . an ensemble of @xmath0 of these @xmath77-state oscillators is characterized by the densities @xmath95 where @xmath96 is the number of oscillators in state @xmath97 at time @xmath10 . the rates ( [ rate ] ) depend on the global state @xmath98 via the @xmath94 , which depend on the finite size order parameter @xmath99 in the thermodynamic limit , @xmath100 and the order parameter ( [ opn ] ) converges to the mean - field version ( [ opmf ] ) . in this limit , the infinite size ensemble of @xmath77-state units is described by the nonlinear master equation ( [ nlme-2 ] ) , in the same spirit as the nonlinear fokker - planck equation ( [ nlfpe ] ) description of the continuum units . if @xmath0 is finite , we need to work with a set of coupled langevin equations , as in ref . however , except for our numerical simulations , finite size effects are beyond the scope of this paper ; here we focus on the mean - field theory . we note that first taking the limit @xmath45 , and then the limit @xmath101 to study the steady states of ( [ nlme-2 ] ) in general does not commute with taking these two limits in the opposite order @xcite . the continuous - time markov - chain modeling scheme rests on the assumption that the rates ( [ rate ] ) are positive , which is not trivially satisfied . this requirement imposes constraints : @xmath102 is positive for any initial condition and realization of the fluctuations only if @xmath103 where @xmath104 is the maximum value of @xmath105 for any @xmath85 in any realization of the evolution of the system . an evident bound is obtained from eq . ( [ bound ] ) by noting that the maximum value of the sine is unity , @xmath106 } \left\{f(r)\right\}. \label{omegamax}\ ] ] fixing the noise intensity and using the coupling strength @xmath107 as the control parameter , we impose the condition @xmath108 equations ( [ kmax ] ) and ( [ omegamax ] ) ensure that condition ( [ kmax ] ) is satisfied . in the limit @xmath109 ( continuous phase space ) , there is no upper limit on the coupling ( @xmath110 ) , as expected from our earlier analysis . the desynchronized state ( [ uniform2 ] ) becomes unstable when the coupling strength @xmath107 exceeds the critical value ( see appendix a ) @xmath111 this converges to the value ( [ criticalk-1 ] ) when @xmath109 ( continuous state space limit ) . at the opposite extreme , when @xmath112 ( which corresponds to only two states on the circle , ( @xmath113 ) , eq . ( [ criticalk-2 ] ) gives @xmath114 , so that no instability to a synchronized state is observed . to observe synchronization within the framework of the continuous - time markov chain picture , we must impose the condition @xmath115 which implies [ see eqs . ( [ dphi ] ) and ( [ kmax ] ) ] @xmath116 for the standard kuramoto model , @xmath117 and @xmath118 , that is , @xmath119 . coarse - graining for this model requires that @xmath120 , that is , @xmath121 . discretization with smaller @xmath77 within the kuramoto model can not be interpreted as a markov chain ( the resulting transition rates " may not be positive ) . coarser graining can be obtained by moving away from the kuramoto model and considering ratios @xmath122 . for the model function ( [ fmodel-1 ] ) , maximizing the coupling @xmath123 [ cf . ( [ gfunction ] ) ] yields @xmath124 this allows us to arbitrarily decrease the number of states in each unit . the limiting case @xmath125 can be reached by choosing @xmath126 . although we can arbitrarily manipulate the number of states by using the function @xmath64 , the limiting case @xmath125 exhibits anomalies in the bifurcation structure that deserve a separate analysis . in the next subsection we discuss the 3-state case in detail . for @xmath127 the transition to synchronization is described by the normal form for the mean field order parameter ( see appendix a ) @xmath128 where @xmath129 . \label{forapb}\end{aligned}\ ] ] clearly , as @xmath130 or equivalently @xmath131 , @xmath132 and @xmath133 as defined in eq . ( [ alphabeta ] ) . moreover , the bifurcation picture is quite similar to that of the continuous phase oscillator . for decreasing @xmath17-functions the bifurcation is supercritical , while increasing @xmath17-functions may induce subcriticality . figure [ fig02 ] displays a comparison between the direct numerical simulation of eq . ( [ amplitudeosc ] ) presented as a five - bar histogram , the steady state solution of the fokker - planck equation ( [ nlfpe ] ) obtained numerically ( red line ) , and the direct numerical simulation of the continuous - time markov chain model for @xmath134 ( black squares ) . the 5-state periodic markov chain gives a surprisingly good picture of the dynamics displayed by the model eq . ( [ amplitudeosc ] ) . the case @xmath125 requires a separate treatment because its behavior is completely different from those of the discrete @xmath127 oscillators . in the @xmath125 case the mean field order parameter near the critical point and at the lowest nonlinear order obeys the normal form ( see appendix a ) @xmath135 where @xmath136 is the complex conjugate of @xmath34 , and @xmath137 which corresponds to a transcritical bifurcation . separating magnitude and phase , that is , @xmath138 , we obtain @xmath139 the desynchronized state @xmath140 with an arbitrary phase @xmath36 is stable for @xmath55 and unstable for @xmath56 . also , here we have the unstable fixed points @xmath141 these are hyperbolic points in a two - dimensional phase space that undergo a phase shift from one of the angles in @xmath142 to one in @xmath143 when they cross the critical point ( at the critical point @xmath144 , @xmath145 because @xmath146 ) . the unstable manifold of @xmath147 is unstable to perturbations of the modulus @xmath66 , but stable to phase perturbations . in contrast , @xmath148 is stable to perturbations of the modulus @xmath66 , but unstable to phase perturbations . the nonlinear saturation of the instability occurs with the inclusion of higher nonlinear orders , which are not captured in the weakly nonlinear analysis used to deduce eq . ( [ normalform-3 ] ) . we analyze it numerically and again use the model function ( [ fmodel-1 ] ) for @xmath64 . [ fig03 ] displays the nullclines of the dynamical system @xmath149 where the nonlinear functions @xmath150 and @xmath151 are obtained from the mean field master equation ( [ nlme-2 ] ) , with @xmath125 and the normalization condition @xmath152 . the nullclines correspond to the curves @xmath153 and @xmath154 in the bidimensional phase space @xmath155 . moreover , due to the fact that @xmath155 are normalized probabilities , the physically accessible phase space is restricted to the right triangle defined by the region bounded by the lines @xmath156 , @xmath157 , and @xmath158 . the fixed points correspond to the intersections of the two nullclines . therefore , the bifurcation scenario is the following : for low enough coupling strength , the only stable equilibrium is the desynchronized state , @xmath140 ( fig . [ figa ] ) . at some point , say @xmath159 , six new fixed points appear simultaneously by saddle - node bifurcation , three of them stable and three of them unstable . in the region @xmath160 , there are four stable fixed points ( fig . [ figb ] ) . at the critical point @xmath144 ( fig.[figc ] ) , the three unstable fixed points collide with the desynchronized state , destabilizing it . for @xmath56 , there remain only three stable fixed points , related to one another by synchronization with a phase shift @xmath161 ( fig . [ figd ] ) . the normal form ( [ normalform-3 ] ) only describes the collision between the desynchronized state and the three hyperbolic points . a neater representation of the bifurcation scenario is shown in fig . [ fig04 ] , where we have plotted the equilibrium order parameter modulus of @xmath66 as a function of the coupling strength . the nonvanishing-@xmath66 branches represent three steady states with different phases @xmath36 . for the parameters in this section , @xmath162 and @xmath163 . note that for @xmath125 ( the three - state system ) the phase invariance occurs in discrete steps @xmath161 . this should be quite general and , from the structure of the expansion performed in appendix a , one can see that , @xmath164 where , in general , the coefficient @xmath165 should be computed for higher orders ( perhaps thus revealing the existence of an extra time scale , slower than the one used for the expansion ) . in any case , it is intuitively rather clear that the distributions obtained from the discrete model will be invariant only to phase transitions occurring at discrete steps @xmath161 . we have considered a noisy version of a kuramoto - like model of identical continuous phase coupled oscillators that exhibit a transition to phase synchronization . we have addressed the following question : under what condition can we model this continuous dynamics as a discrete periodic continuous - time markov chain , that is , as a discrete - phase model where the transitions between these discrete states are governed by a memoryless rate process ? the states in the discrete chain represent a phase cluster of higher and higher density as the number of states in the chain decreases . the markov chain model provides a surprisingly good description of the phase synchronization exhibited by the continuous model , even for chains of only five states ( as confirmed with our numerical simulations ) , and we surmise that the same is true with at least four states . reduction down to seven discrete states is possible with the interaction structure of the original kuramoto model , that is , an interaction linear in the order parameter . further reduction to six , five , four , and three states requires a generalization of the interaction to a nonlinear dependence on the order parameter . however , we have shown that reduction to three - state exhibits more complex behavior in the bifurcation structure ( that we analyzed in detail in sec . [ sec : threestate ] ) which are not present in the continuum model or in the discrete state models down to four states . discrete stochastic models for synchronization phenomena have been increasing in popularity as a simple paradigm of synchronization @xcite . of course , this simplicity is related precisely to the relative ease of dealing with only a few states . in most of these discrete models there are only three states . our results have shown that caution must be exercised when assuming too close a correspondence with kuramoto - like continuous models and even with coarse - grained versions of the latter down to four states . we are currently analyzing the behavior of continuous - time continuous - space oscillator arrays as well as markov chains of four or five states for finite numbers of units . this introduces fluctuations that have not been considered in this work . it will be interesting to compare the effects of finite - size fluctuations in these various arrays . we are also considering arrays with negative coupling parameter @xmath107 , a situation that we have analyzed for three - state markov chains @xcite and that leads to interesting dynamical structures . de thanks fondecyt project no . 1140128 for financial support . ar . acknowledges capes for its support ( grant no . 99999.000296/2015 - 05 ) . kl acknowledges the support of the u.s . office of naval research ( onr ) under grant no . n00014 - 13 - 1 - 0205 . rt acknowledges feder ( eu ) and mineco ( spain ) under grant esotecos fis2015 - 63628-c2-r . in this appendix we detail steps that lead to three important equations in the coarse - graining process , namely , the critical coupling strength for synchronization [ eq . ( [ criticalk-2 ] ) ] , the normal form of the evolution equation for the mean field order parameter that describes synchronization for @xmath127 [ eq . ( [ normalform-2 ] ) ] , and the corresponding result for @xmath125 [ eq . ( [ normalform-3 ] ) ] . to arrive at eq . ( [ criticalk-2 ] ) , we begin by analyzing the dynamics defined by eq . ( [ nlme-1 ] ) , which is of the form @xmath166 where we define the operators @xmath167 and @xmath168 acting on any vector @xmath169 with components @xmath170 and periodic boundary conditions ( @xmath171 and @xmath172 ) as @xmath173 the coefficients in eq . ( [ eq : a1 ] ) are @xmath174 the array of oscillators is desynchronized and every state is equally likely , @xmath175 , if the coupling @xmath107 is sufficiently weak . to find the critical coupling for synchronization we consider a perturbation of this state , @xmath176,\ ] ] with @xmath177 the inequality requires the perturbation to be small , and the sum ensures that the normalization of the @xmath93 is preserved . we introduce this perturbation into eq . ( [ eq : a1 ] ) and retain terms up to first order in the perturbation : @xmath178 here @xmath179 . from eqs . ( [ bound ] ) and ( [ opmf ] ) , @xmath180 and @xmath181 we define the operator @xmath182 as follows : @xmath183 we can then write the evolution equation for @xmath184 : @xmath185 where the linear operator @xmath186 is given by @xmath187 we next introduce the discrete orthogonal fourier basis @xmath188 \equiv c_m e^{-ijm\delta\phi } + c^{}_m e^{ijm\delta\phi}. \label{fourier}\ ] ] the operator ( [ olinear1 ] ) is hermitian and diagonal in this basis under the inner product between two arbitrary @xmath77-element vectors , say @xmath189 and @xmath190 : @xmath191 the associated eigenvalues are @xmath192 + \delta_{1m}\tilde{k}\sin\delta\phi . \label{eigenvalues}\ ] ] the eigenvalues for @xmath193 are all negative . the eigenvalue @xmath194 is associated with the conservation of probability and has no dynamical consequences . synchronization requires that at least one of the eigenvalues be positive . only @xmath195 can be positive , so @xmath196 is the critical mode and the critical point occurrs when @xmath197 , that is , when @xmath198 + \tilde{k}_c\sin\delta\phi = 0 . \label{eigenvaluem1}\ ] ] using eq . ( [ tilde ] ) , this gives the critical coupling parameter given in eq . ( [ criticalk-2 ] ) . therefore , the dynamics near criticality evolve on two time scales , a fast one associated with the relaxation of the modes with @xmath193 , and a slow one related with the critical mode @xmath196 . our next step is to derive the normal form of the evolution equation for the mean field order parameter that describes synchronization for @xmath127 [ eq . ( [ normalform-2 ] ) ] . the order parameter in eq . ( [ opmf ] ) can be written as @xmath199 . then , expanding @xmath93 in the basis ( [ fourier ] ) , that is @xmath200,\ ] ] we find that @xmath201 therefore , the order parameter @xmath34 is associated with the amplitude of the critical mode . that is , @xmath202 is related to the slow time scale ( central manifold ) . we next introduce the ansatz @xmath203 } \left[r(t)\right]\right ) . \label{nlexpansion}\ ] ] where the @xmath204}$ ] represents the contribution of order @xmath205 in @xmath34 and @xmath206 . note that , the conservation of probability implies that , for all @xmath205 , @xmath207}\left\|\psi_{0j } \right\rangle = 0 . \label{conserprob}\ ] ] @xmath208 evolves on the slow time scale @xmath209 because we are assuming that the fast contributions have already relaxed . that is , @xmath210 which is small near criticality . for the order parameter we assume the pitchfork bifurcation scaling @xmath221 and check the consistency of this assumption _ a posteriori_. assuming ( [ scalingpitch ] ) and ( [ w1 ] ) , the second order , @xmath222 , leads to @xmath223}=v_{j}^{\left[2\right ] } = -\tilde{k}_c \sin\left(2\delta\phi\right)\psi_{2j}\left[r^2\right].\label{v2}\ ] ] for @xmath224 , eq . ( [ v2 ] ) does not have solvability problems , leading to @xmath225 } = \psi_{2j}\left[\frac{2\tan\left(\delta\phi/2\right)}{\tan\left(\delta\phi\right)}r^2\right].\ ] ] the third order , @xmath226 , has solvability problems , and the solvability condition ( [ sc1 ] ) leads to the equation @xmath227\left\|r\right\|^2 r & = 0 \nonumber.\end{aligned}\ ] ] using eq . ( [ tilde ] ) , the above solvability condition takes the form given in eq . ( [ normalform-2 ] ) . for @xmath125 , eq . ( [ v2 ] ) has no solution , since in this case @xmath228 , which implies @xmath229 = \psi_{1j}\left[(r^{})^2\right].\ ] ] here the scaling assumption ( [ scalingpitch ] ) does not allow us to impose a suitable solvability condition at second order . hence , in order to ensure the consistency of the expansion ( [ nlexpansion ] ) , we must modify our scaling assumption and adopt the transcritical scaling @xmath230 this scaling allows us to write the solvability condition ( [ sc1 ] ) for eq . ( [ o2 ] ) in the form @xmath231 using eq . ( [ tilde ] ) , the above solvability condition leads to the normal form eq . ( [ normalform-3 ] ) . critical point calculations and normal forms near the transition to synchronization for the continuos - phase kuramoto model and its variants have been extensively documented in the literature @xcite ( see ref . @xcite for an extensive review of a number of approaches ) . here we simply point out that in the limit @xmath109 some of the results of appendix a reduce to those appropriate for the continuos - phase oscillators . in particular , in this limit , @xmath232 that is , we obtain the continuous critical point ( [ criticalk-1 ] ) . moreover , from eq . ( [ forapb ] ) we find @xmath233 , \nonumber\end{aligned}\ ] ] which leads to the normal form given in eq . ( [ normalform-1 ] ) . when comparing our results to those reported in the literature , in addition to the limit @xmath109 we stress that here we are working with identical oscillators [ @xmath234 and that in the literature on the kuramoto model the function @xmath117 . to obtain the analytic estimate ( [ nfapprox ] ) of the steady state distribution , we note that for small @xmath161 , @xmath235 then , retaining the lowest order of the expansion ( [ nlexpansion ] ) , @xmath236}\right),\ ] ] next , we take the continuos limit @xmath109 or @xmath237 , which implies @xmath238 , with the solution ( [ w1 ] ) . the function @xmath239 $ ] is obtained from the definition ( [ fourier ] ) . therefore , at the steady state , @xmath240 \right),\ ] ] where the steady state value @xmath241 of the order parameter is estimated from the equilibrium value predicted by the normal form ( [ normalform-1 ] ) for @xmath56 . that is , @xmath242 where @xmath36 is an arbitrary phase constant . using the model function ( [ fmodel-1 ] ) for @xmath64 , we find that @xmath243 therefore we obtain eq . ( [ nfapprox ] ) .	the theoretical description of synchronization phenomena often relies on coupled units of continuous time noisy markov chains with a small number of states in each unit . it is frequently assumed , either explicitly or implicitly , that coupled discrete - state noisy markov units can be used to model mathematically more complex coupled noisy continuous phase oscillators . in this work we explore conditions that justify this assumption by coarse - graining continuous phase units . in particular , we determine the minimum number of states necessary to justify this correspondence for kuramoto - like oscillators .
unveiling the evolution of dusty debris disks as a function of stellar age was a major focus of studies of main sequence stars with the spitzer space telescope . rebull et al ( 2008 ) and gaspar et al ( 2009 ) each presented tables listing stellar associations and clusters of known age and the fraction of these stars with spitzer - detected excess infrared emission . in a recent paper , we added the ab doradus , tucana / horologium , and argus associations to such tabulations ( zuckerman et al 2011 , table 11 ) . the @xmath0 persei cluster ( hereafter @xmath0 per ) is the only nearby ( 182@xmath211 pc , jackson & jefferies 2010 ; 191@xmath27 pc , robinchon et al . 1999 ; 183 pc , makarov 2006 ) , rich ( hundreds of members ) , open cluster not surveyed by spitzer ; indeed , at an age of @xmath160 myr ( section 3.2 ) it is the youngest rich open cluster in the solar vicinity . to partially rectify this omission we correlated mid - infrared sources in the first and final wise data releases with members of @xmath0 per that have spectral classes between m- and b - type , but with most emphasis on g - type and earlier . because one of our goals was determination of the fraction of cluster members with wise - detected mid - ir emission above the stellar photosphere , it was important to identify stars with a high probability of membership . toward this end , we evaluated results from various published papers to compose lists of likely and possible cluster members ( tables 1 and 2 , respectively ) . the list in table 1 should be mostly complete for stars of spectral type g and earlier and located within 3 degrees of the @xmath0 persei cluster center which we took to be 03h26.0 m + 49d07 ( kharchenko et al 2005 ) . in the following sections we outline our search procedure for dusty @xmath0 per stars , discuss a few of the dusty stars that appear in table 3 , and consider where @xmath0 per stands relative to other young open clusters and associations as regards the presence of dusty debris disks . in compiling tables 1 and 2 we had occasion to consult many of the important papers written about the @xmath0 persei cluster . these and additional important papers are cited in section 3.2 where we consider the relative ages of @xmath0 persei and other nearby stellar clusters and associations . our identification of @xmath0 per members with dusty debris disks involved three principal steps . first we produced a list of likely cluster members . we correlated this list with the wise catalog , and then searched for stars with infrared emission above the stellar photosphere . to maintain as wide a breadth as possible in the early stages of our search , we employed three cluster input catalogs . the first was taken from the webda online service ( http://www.univie.ac.at/webda/ ) , another was from makarov ( 2006 ) , and the third was generated by ourselves from a focused search of the ucac3 catalog ( zacharias et al 2009 ) . all sources listed in the webda @xmath0 persei catalog ( regardless of membership probability ) were searched for infrared excess emission ( see section 2.2 ) , as were all 139 stars listed in table 1 of makarov ( 2006 ) and identified by him as `` high fidelity '' members . because makarov s list is mostly complete only to spectral type mid - g and earlier , our search of the ucac3 catalog was designed to reveal late - type stars with exceptionally large mid - ir excess emission similar to and in addition to that of v488 per ( section 3.1 , figure 2 ) , but none were found . the ucac3 input sample was generated as follows . we searched the ucac3 catalog within 3 arc degrees of the @xmath0 persei cluster center at 03h26.0 m + 49d07. to restrict the size of the search output , we further required that a star have proper motion errors of less than 10 mas / yr in each of ra and dec , that pmra lie between 7 and 37 mas / yr , and that pmdec lie between -40 and -10 mas / yr . the proper motion of stars meeting these criteria were compared to the mean @xmath0 persei proper motion values ( [email protected] mas / yr , [email protected] mas / yr ; kharchenko et al . if the ucac3 proper motions of a queried source agreed with the mean cluster proper motion values to within 3-sigma of their respective uncertainties , then the source was tentatively labeled as a cluster member . all tentative cluster members were then searched for infrared excess emission . the webda service collects data products from papers that discuss open clusters . the service makes no effort to vet these data products , and merely adds them as - is to their repository . this means that if , for example , a paper were to report imaging photometry of sources in the field around @xmath0 persei , all such sources would be added to the webda @xmath0 persei cluster database regardless of actual membership status . in this way the webda service collected @xmath33000 candidate @xmath0 persei cluster members . to be as conservative as possible , and to enable as much infrared discovery as possible , we started with all 3000 + candidates given by wedba for the @xmath0 persei cluster . these candidates come from 14 sources , three of which are cited in the present paper in other contexts . to compile excess statistics , cluster members both with and without infrared excess emission needed to be culled from our larger input catalogs . proper motion information , photometric distance estimates , and radial velocity measurements ( when available ) were used to identify members and to remove non - members from our input samples . in this way the majority of the webda input sample was rejected . following our own search , as outlined in the preceding four paragraphs , we then turned to the @xmath0 persei cluster literature . in addition to makarov ( 2006 ) , we principally utilized six other papers ( tables 1 and 2 , see the six right hand columns and the table notes ) to identify likely and possible cluster members . in addition to proper motion , hipparcos and photometric distances , and radial velocities , lithium abundance and x - ray fluxes have been utilized to establish cluster membership in these various papers . makarov s ( 2006 ) selection criteria are described in the first two sentences of his abstract : `` a kinematical study of the nearby open cluster @xmath0 persei is presented based on astrometric proper motions and positions in the tycho-2 catalog and second usno ccd astrographic catalog ( ucac2 ) . using the astrometric data and photometry from the tycho-2 and ground - based catalogs , 139 probable members of the cluster are selected , 18 of them new . '' as noted two paragraphs previous , we and others have used criteria in addition to those employed by makarov . the result is that 117 of makarov s 139 suggested members appear in our table 1 and 20 appear in our table 2 . as may be seen by inspection of table 2 , the principal reasons why 20 stars from makarov s list are deemed by us to be only possible cluster members is because either other authors do not agree with makarov or because no data other than kinematics and photometry are available . by contrast , almost all stars listed in table 1 are regarded as members in at least two papers . in the few cases where an `` n '' ( nonmember ) or `` n ? '' appears in table 1 , such characterization is often superceded in later references , sometimes by the same authors . while table 1 should be a mostly complete list of members of spectral type g and earlier and located within 3 degrees of the cluster center , the table only sparsely samples later - type members ( e.g. , ap members listed by prosser 1992 ) . because of wise sensitivity limits that would require huge fractional excess ir emission for cluster members later than spectral type g , we have not made an effort to identify such stars . therefore , many previously suggested late - type cluster members do not appear in tables 1 and 2 . also , although we considered x - ray data when compiling tables 1 and 2 ( e.g. randich et al 1996 ) , we did not do a thorough search of all published x - ray papers that might be relevant to identification of cluster members . each input catalog of @xmath0 persei stars was initially cross - correlated with the preliminary ( 2011 ) wise catalog . given the wise angular resolution of 6.1 `` , 6.4 '' , 6.5 `` and 12.0 '' in its four bands of 3.35 , 4.60 , 11.56 , and 22.09 @xmath4 m ( wright et al 2010 ) , we used a search radius of 10 " to identify wise counterparts of input catalog stars . for conversion from wise magnitudes to the flux densities given in table 3 we used 0 ( vega - magnitude ) flux densities of 306.68 , 170.66 , 29.045 , and 8.284 jy and color corrected these as appropriate for each star . spectral energy distributions ( seds ) for each matched source were generated using 2mass photometry , wise data , and any other available photometry ( e.g. , tycho-2 bv magnitudes ) . a fully automated sed - fitting technique employing a theoretical atmospheric model ( hauschildt et al 1999 ) was used to predict stellar photospheric fluxes . a detailed description of our photospheric fitting procedure is given in section 2 of rhee et al ( 2007 ) , so we do not repeat it here . in section 3 of our recent paper ( zuckerman et al 2011 ) we describe how we checked this theoretical model against actual mips measurements of stars that show no evidence for an infrared excess . the result was an upward correction of the model photospheric flux densities at 24 um by a factor of 1.03 . therefore , in table 3 the listed 22 @xmath4 m photospheric flux densities are those produced from the theoretical model , multiplied by a factor of 1.03 . from a sample of 149 likely cluster members ( table 1 ) , mostly b- , a- , f- and g - type stars , in the 2011 wise catalog we initially identified a total of 11 @xmath0 per cluster members with excess mid - ir emission , probably due to orbiting dust particles ( table 3 ; figures 1 - 2 ) . the final ( 2012 ) version of the wise catalog appeared shortly before this paper was finalized and the same 11 stars were identified ; all table 3 flux densities are from the final wise catalog . these 11 include v488 per which is the only star with excess emission in each of the four wise filters ( see discussion of v488 per in section 3.1 ) . for each of the 11 stars in table 3 , the 22 @xmath4 m excess is at least 5 times the listed wise uncertainty . in addition to these 11 stars , hd 21855 might have a 22 @xmath4 m excess ( see discussion in section 3.1 ) . v488 per was serendipitously observed with the infrared array camera ( irac ; fazio et al . 2004 ) on spitzer through a program searching for substellar companions to low mass stars in @xmath0 per ( aorkey : 18853632 ) . it was observed in all four irac channels on 9 - 26 - 2006 . irac post basic calibrated data ( pbcd ) mosaic files were obtained from the spitzer heritage archive . source extraction was performed on the pbcd mosaics utilizing the astronomical point source extraction ( apex ) package , and the irac data points are added to the sed of v488 per to further constrain the dust fitting ( figure 2 ) . the spitzer irac observations support the large l@xmath5/l * indicated by wise for this object . we identified 11 stars with excess emission above the stellar photosphere at 22 @xmath4 m and sometimes at other wavelengths ( table 3 , figures 1 and 2 ) that can definitely or probably be attributed to an orbiting dusty debris disk . while most are of early - type for which the detected dust emission is usually the wien tail of emission at longer wavelengths generated by cool dust particles one remarkable k - type star , v488 per , appears in tables 1 and 3 and figure 2 . the distribution of spectral types in table 3 is five b , three a , two f and one k. some additional b - type stars display excess emission as measured by wise , but these excesses are more likely due to free - free radiation in an outflowing ionized stellar wind rather than from an orbiting debris disk . characteristically , when emission is due to free - free transitions , then ir flux densities decline toward longer wavelengths . one prominent table 1 example of such a star is @xmath6 per , strongly detected with iras many years ago . another such b - type star is hr 1037 (= hd 21362 ; compare the 24 and 70 @xmath4 m mips measurements reported by su et al ( 2006 ) with the wise catalog flux densities ) . below we consider a few stars in table 3 and also the stars 29 per and hd 21855 . these are followed by a discussion of the evolution of dusty debris disks . 29 per : this b3 star , whose sed is shown in figure 1 , might be expected to display an outflowing ionized wind as mentioned just above . however , the shape of the sed of 29 per is very different from that of other b - type stars . the excess emission at 22 @xmath4 m is weak , whereas iras measured strong excess emission at 60 @xmath4 m ( figure 1 ) . we therefore inspected the wise 22 @xmath4 m images and also used the iras `` scanpi '' function to characterize the spatial distribution of the 60 @xmath4 m emission in the vicinity of 29 per . the wise image displayed in figure 3 illustrates the extended nature of the 22 @xmath4 m emission . our conclusion is that the excess emission is carried by heated interstellar dust ( `` cirrus '' ) at substantial distances from 29 per and not to an orbiting debris disk . thus , 29 per is included in figure 1 to illustrate its sed , but not to suggest the presence of a debris disk . 0.1 in v488 per : this solar - type star has a larger fractional infrared luminosity , defined as the ratio of excess infrared luminosity to bolometric luminosity ( @xmath7 ) , than any other known main sequence star . from the sed in figure 2 , @xmath7 @xmath116% , to be compared to the previous main sequence champion bd+20 307 with @xmath7 @xmath13.3% . ( song et al 2005 ; weinberger et al 2011 ) . most main sequence stars with strong mid - ir emission from warm dust particles show no evidence for substantial quantities of cool dust ( e.g .. melis et al 2010 ; melis et al 2012 ; weinberger et al 2011 ; and references therein ) . however , as may be seen from blackbody fits to its sed , substantial quantities of both warm and cool dust particles orbit v488 per . much of the dust that orbits v488 per is at a temperature of @xmath1800 k. while comparably warm dust with @xmath7 orders of magnitude smaller than at v488 per may orbit some main sequence stars , such small quantities of dust could easily be generated by asteroidal collisions near these stars . by contrast , the large @xmath7 at v488 per indicates a recent collision of planet - mass objects ( melis et al 2010 ) . if these warm grains radiate like blackbodies , then they lie only @xmath10.06 au from the star . this semimajor axis is comparable to that characteristic of many transiting planets discovered by nasa s kepler satellite . based on the data given in various published papers these include photometric distance , proper motion , radial velocity , lithium abundance , and chromospheric activity there seems to be little doubt that v488 per is a bona fide member of the @xmath0 per cluster and it has been so classified in three of the papers cited in table 1 . in addition we can add new determinations of radial velocity , proper motion , and lithium abundance that buttress previous work on this star . concerning radial velocity , archival keck / hires observations of v488 per obtained by dr . i. n. reid on ut 12 december 1997 were acquired from the keck observatory archive . we analyzed this high quality echelle spectrum and found a heliocentric velocity of [email protected] km / s which agrees reasonably well with previous radial velocity determinations by mermilliod et al ( 2008 ; [email protected] km / s ) and prosser ( 1992 ; [email protected] km / s ) . all of these measured radial velocities are characteristic of stars in @xmath0 per ( see , e.g. , table 4 in prosser 1992 ) . from the hires spectrum we measured the equivalent width ( ew ) of the 6708 lithium absorption line to be @xmath1200 m . this is the ew anticipated for an early k - type star of age @xmath160 myr ; in figure 3 of zuckerman & song ( 2004 ) this ew lies just below the ew of most of the plotted 30 myr old members of the tucana association and near the top of the range of ew for @xmath1100 myr old pleiades members . we computed the proper motion of v488 per by comparing positions in the 2mass and wise catalogs that are separated by epochs 10.1 years apart . we derive proper motions of + 15.7+/- 6.6 and @xmath824.9+/-9.1 mas / yr in r.a . and decl . , respectively . these values compare well with those given in the usno - b1 ( 16 , @xmath824 ) and the ppmxl ( 15.9 , @xmath826.4 ) catalogs . because @xmath7 is so large for v488 per compared to all other known dusty main sequence stars , one must consider the possibility of contamination by radiation from a background object . the best spatial resolution image of the field is likely that obtained with hst by patience et al ( 2002 ) . the hst / nicmos camera and f140w filter revealed no source of 1.4 @xmath4 m emission other than the star . to further confirm that the large ir excess is from the star , at our request , dr . charles beichman and collaborators ( see caption to figure 4 ) obtained a seeing - limited image of v488 per with a 3.8 @xmath4 m ( lp ) filter and the nirc2 camera on the keck ii telescope at mauna kea observatory . this image shows only a single source of fwhm @xmath10.6 " and approximately circular shape ( figure 4 ) . thus , there can be little doubt that the strong ir excess emission is indeed associated with v488 per . an additional concern is whether v488 per might be a dusty , first ascent `` phoenix giant '' star ( melis 2009 ; melis et al 2009 ) well to the background of the @xmath0 per cluster . some phoenix giants have very large @xmath7 and hot and cool dust with seds similar to that of v488 per . if v488 per is actually a phoenix giant , then the fact that the photometric distance , radial velocity , and proper motion of v488 per are all consistent with membership in @xmath0 per would have to be regarded as simply a strange coincidence . as outlined in melis et al ( 2009 ) , we use line ratio diagnostics presented in strassmeier & fekel ( 1990 ) and the keck / hires spectrum to evaluate whether v488 per is a luminosity class iii or class v star . we find that v488 per is a luminosity class v star . all k - type phoenix giants exhibit luminosity class iii line ratios when similar analysis is performed for them . as an aftermath of collisions of planetary embryos , solar - type stars are most likely to display large quantities of warm dust particles during the last stages of the assembly of rocky planets in the terrestrial planet zone ; this would be at ages between 30 and 100 myr ( melis et al 2010 ) . the age of v488 per lies in this range and , so , this very dusty star can be added to the small sample of similar stars discussed by melis et al . 0.1 in hd 21375 : this is a single - line , 31 day , spectroscopic binary according to morrell & abt ( 1992 ) . 0.1 in hd 21641 : this is a weak emission , shell , be star ( e.g. , briot 1986 ) . therefore we can not be sure that the observed excess ir emission ( table 3 , figure 2 ) is due to a dusty debris disk rather than an outflowing ionized wind . if the latter , then the excess emission becomes apparent only at much longer wavelengths than is seen toward be stars @xmath6 per and hr 1037 ( both mentioned above ) . future observations at far - ir wavelengths may distinguish between the two potential emission mechanisms . 0.1 in hd 21855 : this a - type may have weak excess emission at 22 @xmath4 m ( w4 ) and perhaps also at 11.5 @xmath4 m ( w3 ) . at w4 , the photosphere is expected to be 6.4 mjy . the measured wise flux density is [email protected] mjy , so excess emission at the 4.4 @xmath9 level may be present . in the w3 band , the expected photosphere is 22.6 mjy and the wise measured flux density is [email protected] mjy , for a possible 2.3 @xmath9 excess . table 11 in zuckerman et al ( 2011 ) presented infrared excess fractions at wavelengths of 24 and 70 @xmath4 m for various open clusters and nearby moving groups . table 4 of the present paper repeats the 24 @xmath4 m column from their table 11 , but now with the addition of the 22 @xmath4 m excess fraction in the @xmath0 per cluster . before comparing @xmath0 per to other clusters we consider some techniques for estimation of ages of stars in clusters and associations . the two principal techniques for derivation of ages of young clusters are matching cluster members of various spectral types to theoretical pre - main sequence isochrones and identification of the location of the lithium depletion boundary . for a given cluster with age comparable to that of @xmath0 per and for which both age - dating methods have been applied , the lithium method generally yields an older age . therefore , it is important to take into account how ages have been deduced when comparing the age of @xmath0 per to those of other clusters in table 4 . figure 5 presents color magnitude diagrams for @xmath0 per and for the pleiades . the latter is clearly the older cluster . a quantitative measure of this age difference is suggested from the li depletion analysis of barrado y navascues et al ( 2004 ) who derive ages of 50@xmath25 , 85@xmath210 , and 130@xmath220 myr for ic 2391 , @xmath0 per , and the pleiades , respectively . in comparison , isochrone ages for these clusters have been given as 35 - 40 myr ( ic 2391 , barrado y navascues et al 2004 ; torres et al 2008 ) , 50 - 80 myr ( @xmath0 per , makarov 2006 ; prosser 1992 ) , and @xmath1100 myr ( pleiades , see references listed in luhman et al 2005 ) . discussions of the age of @xmath0 per by makarov ( 2006 ) and by prosser ( 1992 ) illustrate that ages derived from main sequence fitting can be based on the upper main sequence , the lower main sequence , or both , and that the former is subject to uncertainties in the degree of convection overshooting . the age of the ab dor moving group has been a subject of considerable discussion ( luhman et al 2005 ; janson et al 2007 ; close et al 2007 ; ortega et al 2007 ) with ages ranging from 50 myr to the age of the pleiades ( @xmath10120 myr ) . this discussion is outlined in section 8 of torres et al ( 2008 ) who present their own color - magnitude diagram of ab dor members and who prefer an age of 70 myr . for m47 rojo arellano et al ( 1997 ) use main sequence fitting to derive an age of @xmath1100 myr , while for ngc 2451 an isochronal age of 60 myr is deduced ( references listed in balog et al 2009 ) . based on the discussion in the preceding paragraphs , one may see that the ages in table 4 are weighted more toward isochrone fitting than li depletion . this is in part because li ages have been derived for only some of the listed clusters and associations . however , whatever the true absolute cluster ages , the relative ages as given in table 4 should be quite reliable . as noted in the introduction , excepting the @xmath0 persei cluster , spitzer surveyed stars in all the clusters and associations listed in table 4 . the listed 24 @xmath4 m spitzer / mips excess statistics are usually dominated by f , g and k type stars . in the zuckerman et al ( 2011 ) paper , a ratio of mips measured 24 @xmath4 m flux density to expected photospheric flux of 1.16 or greater was deemed sufficient to establish the presence of an excess . this ratio was not uncharacteristic of excess criteria employed in other references listed in table 4 . unfortunately , given the size of the wise catalog errors in flux density at 22 @xmath4 m and the greater distance from earth of the @xmath0 per cluster compared to a majority of the regions listed in table 4 , this level of sensitivity to excess emission was obtained only for most of the b - type stars in @xmath0 per . even for an @xmath0 per a - type star , the excess emission would have to be at a level @xmath160% of the photosphere for it to satisfy the excess criterion set out in section 2.2 . as a result , it is not possible at this time to sensibly compare excess fractions in @xmath0 per and other table 4 clusters . that is , the excess statistics listed for the @xmath0 per cluster pertain only to b- and a - type members and , to qualify as an excess star , the measured 22 @xmath4 m flux density must often be as large as 1.5 times the photospheric flux . in contrast , the statistics for other listed regions in table 4 usually are dominated by f , g and k - type stars and represent excesses that need only be @xmath11.2 times the photospheric flux . since many stars in , for example , the zuckerman et al ( 2011 ) lists have 24 @xmath4 m flux density between 1.2 and 1.5 times that of the photosphere , there can be little doubt that some a - type stars in the @xmath0 per cluster have excess emission not far below our defined level of excess and will have been missed with our wise survey . such caveats apply even more strongly to the many f- and g - type stars listed in tables 1 and 2 . we may summarize what might reasonably be inferred from table 4 as follows : although the @xmath0 persei cluster may be deficient in dusty debris disks relative to other table 4 clusters and moving groups of comparable age , because ( 1 ) the @xmath0 per statistics pertain only to b- and a - type stars whereas the statistics for the other groups are dominated by later - type stars , and ( 2 ) the wise sensitivity to dusty debris at these @xmath0 per stars was not quite as good as spitzer sensitivities in these other groups , it would be premature to draw any firm conclusions . additional infrared measurements at 22 or 24 @xmath4 m are unlikely to come before the james webb space telescope is operational . before then , perhaps the best way to determine if debris disks in @xmath0 per are relatively scarce would be to confirm or deny whether the a- and b - type stars in table 2 are actual cluster members ( none have 22 @xmath4 m excess emission ) . if most are members , then as indicated by statistics given in the note to table 4 , the implication of a real relative deficiency of debris disks in the @xmath0 per cluster would be suggestive , if not totally convincing . @xmath0 per is at the same or smaller absolute galactic latitude than all table 4 clusters and moving groups excepting m47 . therefore , in the event that background cirrus contamination could introduce an error in the disk fraction , this error would most likely be in the sense to increase the apparent number of dusty stars in @xmath0 per relative to other clusters and associations in table 4 . as remarked in the notes to table 4 , the tabulated excess fraction of the @xmath0 per cluster is for a wavelength of 22 @xmath4 m , but at 24 @xmath4 m for all other entries . dusty debris disks studied by iras and spitzer at far - ir wavelengths have typical temperatures of @xmath180 k. for this temperature the flux density will be @xmath11.5 times larger at 24 @xmath4 m than at 22 @xmath4 m . given our 5 @xmath9 wise detection threshold at 22 @xmath4 m ( see section 2.2 ) , we checked among table 1 and 2 stars , but not listed in table 3 , for any b- or a - type stars with at least 3.3 @xmath9 of apparent excess emission ; that is , for stars for which `` w4 meas '' is greater by at least 3.3 @xmath9 than `` w4 photo '' ( see the header to table 3 ) . other than hd 21855 with 4.4 @xmath9 of possible excess emission ( see section 3.1 ) , no such stars were found . thus , the difference between 22 and 24 @xmath4 m should not alter the ratio of percentage of excess stars in @xmath0 per compared to excess percentages in other listed clusters dukes & krumholz ( 2011 ; see also references cited therein ) suggest that disk disruption events are no more likely in massive clusters than in low mass clusters . they cite observations that would indicate that , for most clusters , 90% of the stars disperse into the field within 10 myr . the stellar binary fraction in @xmath0 per and other rich nearby clusters ( pleiades , hyades , praesepe ) have been considered in various papers . patience et al ( 2002 ) note that the distribution of semimajor axes of binary stars in these four clusters peaks at 5 au , `` a significantly smaller value '' than for stars in the solar neighborhood . morrell & abt ( 1992 ) comment on the relatively small number of spectroscopic binaries among b- and a - type stars in @xmath0 per and in the pleiades . makarov ( 2006 ) notes that counting all types of binaries in @xmath0 per , only @xmath120% of members are binaries which is `` modest even compared with field g - dwarfs and much smaller than the binarity rate in the hyades and pleiades . '' calculations utilizing the known characteristics of clusters and associations listed in table 4 of the ( relative ) frequency of close stellar encounters might be of interest . we assembled what likely is the most complete listing of probable and possible b - type through g - type members of the nearby @xmath0 persei cluster . we then correlated sources in the final wise catalog with cluster members and found 11 , or possibly 12 , with excess emission above the photosphere at 22 @xmath4 m ( and sometimes also at other wise wavelengths ) , that we interpret as due to a surrounding dusty debris disk . in addition , a few b - type stars display excess emission consistent with production in an outflowing wind . of the probable b- and a - type cluster members , 14% display excess emission ( from an orbiting dusty debris disk ) at a wavelength of 22 @xmath4 m . if b- and a - type stars that are possible members are included in the statistics , the percentage of excess stars could be as small as 11.6% . for later - type stars , the wise sensitivity is inadequate to detect many ( most ? ) of the debris disks that likely orbit stars in the cluster . v488 per is the most interesting dusty star in the @xmath0 persei cluster . the luminosity of orbiting dust grains is at least 16% of the bolometric luminosity of this k - type star ; this is a far larger excess luminosity percentage than that previously known at any main sequence star . although there are numerous reasons to include v488 per as a bona fide member of the cluster , for confirmation , a measurement of its trigonometric parallax would be worthwhile . recently , melis et al ( 2012b ) identified a young star with , initially , a fractional excess luminosity at mid - ir wavelengths almost as large as that at v488 per . however , the excess at this young star recently disappeared in the space of only a few years . thus , the excess infrared emission at v488 per bears careful watching during the coming years . previously we considered the final era of the accretion of rocky planets around solar mass stars in a region analogous to that where the four rocky planets of the solar system reside ( melis et al 2010 ) . by assembling a list of stars known to contain large quantities of warm dust grains , we found that this era occurs at stellar ages from 30 to 100 myr . with its very strong mid - ir excess emission and its @xmath160 myr age , v488 per can now be included among those stars that are orbited by large rocky bodies in the terrestrial planet zone . by contrast , melis et al ( 2012a ) find that the dominant era of rocky planet formation around stars with masses a few times that of the sun is over by about 30 myr ; our study of the @xmath160 myr old @xmath0 persei cluster is consistent with this finding . specifically , among 70 late - b to early - f type @xmath0 per cluster members , wise reveals none with the massive amounts of warm dust that would herald the aftermath of a collision of planetary embryos . future observations at far infrared and/or submillimeter will be necessary to characterize the properties of the @xmath0 per debris disks . for v488 per such observations will be especially important because , unlike most other main sequence stars that are orbited by large quantities of warm dust particles but little or no cold dust , v488 per appears to be orbited by large amounts of both warm and cold dust . 0.2 in we are grateful to chas beichman , chris gelino , greg mace , joel aycock , and randy campbell for obtaining the data that went into the image in figure 4 ( just before v488 per set for the evening ) , and to alan stockton for attempting a similar measurement for us . we thank john stauffer for donation of figure 5 and for helpful conversations and the referee for a constructive report that substantially improved this paper . this research was funded in part by nasa grants to ucla and the university of georgia . c.m . acknowledges support from the national science foundation under award ast-1003318 . this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration . this research has made use of the keck observatory archives ( koa ) which is operated by the w.m . keck observatory and the nasa exoplanet science institute ( nexsci ) under contract with the national aeronautics and space administration . vol 5 , ed . b. reipurth , 757 + star name & ra & dec & pmra & pmdec & spectral & he & pr ? & ma ? & st ? & me ? + & ( j2000 ) & ( j2000 ) & ( mas / yr ) & ( mas / yr ) & type & & & & & & & + + bd+48 851 & 03 07 49.8 & + 49 06 23 & [email protected] & @[email protected] & f6v & 12 & y & y & & y & & + tyc 3315 1159 1 & 03 11 16.8 & + 48 10 37 & [email protected] & @[email protected] & f9v & 94 & y & y & & & & + hd 19655 & 03 11 41.1 & + 48 03 15 & [email protected] & @[email protected] & f2vn & 104 & y & y & & & & + hd 19624 & 03 11 42.9 & + 52 09 48 & [email protected] & @[email protected] & b5 & 145 & y ? & y & & & & + bd+49 868 & 03 11 50.0 & + 50 22 47 & [email protected] & @[email protected] & f5v & 135 & y & y & & y & y & + hd 19767 & 03 12 43.3 & + 47 50 19 & [email protected] & @[email protected] & f0vn & 151 & y & y & & & y & + hd 19805 & 03 13 05.2 & + 49 00 34 & [email protected] & @[email protected] & b9.5v & 167 & y & y & & & y & + tyc 3319 446 1 & 03 13 07.4 & + 49 34 04 & [email protected] & @[email protected] & g & 174 & y & y & y ? & y & & + hd 19893 & 03 13 50.3 & + 49 34 08 & [email protected] & @[email protected] & b9v & 212 & y & y & & & y & + bd+48 871 & 03 15 23.6 & + 49 26 25 & [email protected] & @[email protected] & f7v & 270 & y & y & & y & y & + tyc 3319 9 1 & 03 15 58.9 & + 50 24 19 & [email protected] & @[email protected] & f7 & 299 & y & y & & y & & + bd+49 889 & 03 16 23.2 & + 49 37 33 & [email protected] & @[email protected] & f5v & 309 & y & y & & y & & + hd 20191 & 03 16 49.1 & + 51 13 05 & [email protected] & @[email protected] & b9 & 333 & y & y & & & y & + bd+49 892 & 03 16 59.4 & + 49 55 36 & [email protected] & @[email protected] & f7v & 334 & y & y & y & y & & + bd+48 876 & 03 17 20.5 & + 49 30 01 & [email protected] & @[email protected] & f7v & 338 & y & y & & y & & + ap 119 & 03 17 31.4 & + 48 51 51 & [email protected] & @[email protected] & k2 & & y & & & & & + tyc 3319 306 1 & 03 17 36.9 & + 48 50 08 & [email protected] & @[email protected] & g3 & 350 & y & y & & y & & y + ap 121 & 03 17 42.1 & + 49 01 46 & [email protected] & @[email protected] & g5 & & yy & & y & ? & & + hd 20282 & 03 17 43.2 & + 50 21 40 & [email protected] & @[email protected] & a0 & 357 & & y & & & & + bd+49 896 & 03 18 01.7 & + 49 38 39 & [email protected] & @[email protected] & f4v & 361 & y & y & & y & & + bd+49 897 & 03 18 05.2 & + 49 54 22 & [email protected] & @[email protected] & f6v & 365 & y & y & & & y & + hd 20344 & 03 18 23.9 & + 50 33 21 & [email protected] & @[email protected] & a0 & 379 & y ? & y & & & y & + v522 per & 03 18 27.4 & + 47 21 15 & [email protected] & @[email protected] & g3 & 373 & y & y & & & & + 29 per & 03 18 37.7 & + 50 13 20 & [email protected] & @[email protected] & b3v & 383 & y & y & & & y & + hd 20391 & 03 18 44.8 & + 49 46 12 & [email protected] & @[email protected] & a2v & 386 & y & y & & & y & + tyc 3319 589 1 & 03 18 50.3 & + 49 43 52 & [email protected] & @[email protected] & g0 & 389 & y & y & & & & + v524 per & 03 18 59.3 & + 48 50 35 & [email protected] & @[email protected] & k5 & & y & & & & & + 31 per & 03 19 07.6 & + 50 05 42 & [email protected] & @[email protected] & b5v & 401 & y & y & & & & + hd 20475 & 03 19 42.2 & + 48 54 49 & [email protected] & @[email protected] & f2v & 421 & y & y & & & y & y + hd 20487 & 03 19 47.2 & + 48 37 41 & [email protected] & @[email protected] & a0vn & 423 & y & y & & & y & + ap 126 & 03 19 57.3 & + 49 04 21 & [email protected] & @[email protected] & m4 & & y & & & & & + hd 20510 & 03 20 06.3 & + 50 58 07 & [email protected] & @[email protected] & b9v & 441 & y ? & y & & & y & + hd 20537 & 03 20 23.7 & + 51 37 06 & [email protected] & @[email protected] & b9 & 450 & & y & & & y & + tyc 3315 1781 1 & 03 20 39.3 & + 47 29 22 & [email protected] & @[email protected] & f8v & 453 & y ? & y & y & y & & y ? + ap 97 & 03 20 41.9 & + 48 24 38 & [email protected] & @[email protected] & g6.5 & & y@xmath11 & & y & y & & y + ap 131 & 03 20 56.5 & + 49 20 43 & [email protected] & @[email protected] & m4 & & y & & & & & + v625 per & 03 21 06.5 & + 48 26 13 & [email protected] & @[email protected] & g9 & & y@xmath11 & & y & y & & y + ap 134 & 03 21 20.5 & + 47 53 15 & [email protected] & @[email protected] & m3 & & y & & & & & + bd+47 808 & 03 21 30.2 & + 48 29 38 & [email protected] & @[email protected] & f1ivn & 481 & y & y & & & y & + bd+48 892 & 03 21 40.2 & + 49 07 13 & [email protected] & @[email protected] & f3iv - v & 490 & y & y & & y & & + v459 per & 03 21 58.6 & + 49 12 53 & [email protected] & @[email protected] & f0iv & 501 & ? & y & & & & + v529 per & 03 22 06.8 & + 47 34 07 & [email protected] & @[email protected] & k2 & & y & & & & & y + v484 per & 03 22 21.9 & + 49 08 28 & [email protected] & @[email protected] & g6v & 520 & y & y & & & & y + hd 20714 & 03 22 26.3 & + 51 39 39 & [email protected] & @[email protected] & a7vn & 522 & ? & y & & & & + v575 per & 03 23 13.2 & + 49 12 48 & [email protected] & @[email protected] & b5v & 557 & y & y & & & y & + bd+47 815 & 03 23 40.3 & + 47 57 30 & [email protected] & @[email protected] & f3v & 577 & ? & y & y ? & & & y ? + hd 20842 & 03 23 43.1 & + 51 46 13 & [email protected] & @[email protected] & a0v & 575 & y & y & & & & + hd 20863 & 03 23 47.3 & + 48 36 16 & [email protected] & @[email protected] & b9v & 581 & y & y & & & y & y + bd+ 49 914 & 03 23 55.1 & + 50 18 24 & [email protected] & @[email protected] & f5v & 588 & y & y & & & & + hd 20903 & 03 24 06.5 & + 46 17 19 & [email protected] & @[email protected] & a2 & & & y & & & & + ap 109 & 03 24 06.7 & + 49 24 52 & [email protected] & @[email protected] & m3 & & y@xmath11 & & & & & y + melotte 20 601 & 03 24 17.1 & + 49 39 00 & [email protected] & @[email protected] & g6 & 601 & n & & y & y & & + v461 per & 03 24 19.2 & + 49 13 16 & [email protected] & @[email protected] & a8v & 606 & y ? & y & & & & y ? + @xmath0 per & 03 24 19.4 & + 49 51 40 & [email protected] & @[email protected] & f5iab & 605 & y & y & & & y & + ap 14 & 03 24 19.9 & + 48 47 20 & [email protected] & @[email protected] & g4 & & y & & & y & & y + v485 per & 03 24 25.1 & + 48 48 21 & [email protected] & @[email protected] & k5 & & & & & & & y + bd+49 918 & 03 24 25.6 & + 50 19 34 & [email protected] & @[email protected] & f0v & 609 & y & y & & & & + hd 20931 & 03 24 30.0 & + 49 08 24 & [email protected] & @[email protected] & a1v & 612 & y & y & & & y & + bd+47 816 & 03 24 47.1 & + 48 24 42 & [email protected] & @[email protected] & f4v & 621 & y & y & & y & y & + v531 per & 03 24 49.7 & + 48 52 18 & [email protected] & @[email protected] & g5 & 622 & y & y & y & & & y + hd 20961 & 03 24 52.1 & + 47 54 54 & [email protected] & @[email protected] & b9.5v & 625 & y & y & & & & + bd+46 745 & 03 24 54.6 & + 47 24 54 & [email protected] & @[email protected] & f4v & 632 & y & y & & & y & + hd 20969 & 03 25 04.4 & + 49 47 43 & [email protected] & @[email protected] & a8v & 635 & y & y & & & & + hd 20986 & 03 25 10.0 & + 49 15 06 & [email protected] & @[email protected] & a3vn & 639 & y & y & & & & + ap 25 & 03 25 16.2 & + 48 22 24 & [email protected] & @[email protected] & k0 & & y & & & y & & y + hd 21005 & 03 25 20.7 & + 49 18 58 & [email protected] & @[email protected] & a5vn & 651 & y & y & & & & + tyc 3320 1768 1 & 03 25 37.6 & + 50 19 18 & [email protected] & @[email protected] & f5v & 660 & y ? & y & y ? & & & + hd 21046 & 03 25 37.7 & + 47 01 14 & [email protected] & @[email protected] & a7v & 665 & y & y & & & & + v576 per & 03 25 57.4 & + 49 07 15 & [email protected] & @[email protected] & b7v & 675 & y & y & & & y & + v688 per & 03 26 04.2 & + 48 48 07 & [email protected] & @[email protected] & f9v & 684 & y & y & y & y & & y + hd 21091 & 03 26 10.8 & + 48 23 03 & [email protected] & @[email protected] & b9.5v & 692 & y ? & y & & & y & + ap 38 & 03 26 19.3 & + 49 13 32 & [email protected] & @[email protected] & g3 & 696 & y & & y & y & & y + v532 per & 03 26 22.2 & + 49 25 37 & [email protected] & @[email protected] & g2 - 3v & 699 & y & y & & & & y + v628 per & 03 26 25.3 & + 48 20 07 & [email protected] & @[email protected] & g5 & & y & & & y & & y + hd 21122 & 03 26 32.6 & + 47 15 59 & [email protected] & @[email protected] & a0 & 710 & & y & & & y & + ap 158 & 03 26 33.7 & + 50 13 54 & [email protected] & @[email protected] & k0 & & y & & y & y & & + hd 21117 & 03 26 39.4 & + 50 50 47 & [email protected] & @[email protected] & b8 & 703 & & y & & & y & + bd+47 825 & 03 26 39.2 & + 47 52 56 & [email protected] & @[email protected] & f2vn & 721 & ? & y & & & & + tyc 3320 1715 1 & 03 26 40.7 & + 48 46 37 & [email protected] & @[email protected] & f4v & 715 & y & y & & & & y + tyc 3320 818 1 & 03 26 43.9 & + 49 54 34 & [email protected] & @[email protected] & g0v & 709 & y & y & & & & y + ap 51 & 03 26 50.7 & + 48 47 31 & [email protected] & @[email protected] & f7v & 727 & y & y & & & & y + bd+48 916 & 03 27 03.2 & + 48 47 13 & [email protected] & @[email protected] & f6v & 733 & y ? & y & & y & & y ? + hd 21181 & 03 27 05.1 & + 48 12 20 & [email protected] & @[email protected] & bvn & 735 & y & y & & & y & + ap 161 & 03 27 18.9 & + 47 25 24 & [email protected] & @[email protected] & m3 & & y & & & & & + hd 21239 & 03 27 37.6 & + 48 16 23 & [email protected] & @[email protected] & a3vn & 756 & ? & y & & & & + ap 58 & 03 27 37.8 & + 48 59 29 & [email protected] & @[email protected] & f9v & 750 & y & y & & & & y + hd 21238 & 03 27 38.9 & + 49 36 00 & [email protected] & @[email protected] & b9v & 747 & & y & & & y & + v534 per & 03 27 51.0 & + 49 12 10 & [email protected] & @[email protected] & k2 & & y & & & & & y + tyc 3320 2239 1 & 03 27 55.0 & + 49 45 37 & [email protected] & @[email protected] & f9v & 767 & y & y & n ? & y & & + hd 21279 & 03 27 55.8 & + 47 44 09 & [email protected] & @[email protected] & b8.5v & 775 & y & y & & & + hr 1034 & 03 28 03.1 & + 49 03 47 & [email protected] & @[email protected] & b5v & 774 & y & y & & & y & + hd 21302 & 03 28 18.6 & + 49 57 10 & [email protected] & @[email protected] & a1vn & 780 & y & y & & & & y + v488 per & 03 28 18.7 & + 48 39 48 & [email protected] & @[email protected] & k0 & & y & & & y & & y + bd+48 923 & 03 28 31.5 & + 48 56 27 & [email protected] & @[email protected] & f4v & 799 & y & y & & & & y + bd+49 939 & 03 28 34.7 & + 50 16 01 & [email protected] & @[email protected] & f3iv - v & 794 & y ? & y & & y & & + hd 21345 & 03 28 38.0 & + 49 23 15 & [email protected] & @[email protected] & a5vn & 802 & ? & y & & & & + hr 1037 & 03 28 52.3 & + 49 50 54 & [email protected] & @[email protected] & b6vn & 810 & y & y & & & y & + hd 21375 & 03 28 53.6 & + 49 04 13 & [email protected] & @[email protected] & a1v & 817 & y & y & & & & y + melotte 20 828 & 03 28 59.6 & + 48 14 08 & [email protected] & @[email protected] & f8 & 828 & y & & ? & y & & y + hd 21398 & 03 29 07.6 & + 48 18 10 & [email protected] & @[email protected] & b9v & 831 & y & y & & & & + tyc 3316 904 1 & 03 29 08.3 & + 48 10 51 & [email protected] & @[email protected] & f6 & 833 & y & & & y & & + ap 176 & 03 29 19.0 & + 46 07 27 & [email protected] & @[email protected] & m4.2 & & y & & & & & + hd 21428 & 03 29 22.0 & + 49 30 32 & [email protected] & @[email protected] & b3v & 835 & y & y & & & & y + tyc 3320 1057 1 & 03 29 24.9 & + 48 57 45 & [email protected] & @[email protected] & f7v & 841 & y & y & & & & y + bd+47 837 & 03 29 26.2 & + 48 12 12 & [email protected] & @[email protected] & f9v & 848 & y & y & y & y & & y + bd+48 931 & 03 29 46.9 & + 49 00 34 & [email protected] & @[email protected] & f6v & 863 & n & y & & y & & y + bd+47 839 & 03 29 46.9 & + 47 34 58 & [email protected] & @[email protected] & f0 & 876 & y & y & & & & y + hd 21480 & 03 29 47.0 & + 49 09 13 & [email protected] & @[email protected] & a7v & 862 & y & y & & & & + hd 21481 & 03 29 50.0 & + 47 58 37 & [email protected] & @[email protected] & a0vn & 875 & n ? & y & & & & y + hd 21479 & 03 29 51.8 & + 49 12 49 & [email protected] & @[email protected] & a1ivn & 868 & y & y & & & & + hd 21527 & 03 30 19.3 & + 48 29 57 & [email protected] & @[email protected] & a7iv & 885 & y & y & & & y & y + v465 per & 03 30 34.0 & + 47 37 41 & [email protected] & @[email protected] & a6vn & 906 & y & y & & & & + hr 1051 & 03 30 36.9 & + 48 06 13 & [email protected] & @[email protected] & b8v & 904 & y & y & & & y & + melotte 20 917 & 03 30 47.6 & + 47 53 22 & [email protected] & @[email protected] & f4 & 917 & y & & & y & & y + ap 183 & 03 30 56.8 & + 50 00 51 & [email protected] & @[email protected] & m3 & & y & & & & & + hd 21600 & 03 31 14.6 & + 49 42 22 & [email protected] & @[email protected] & a6vn & 921 & y & y & & & y & + bd+48 937 & 03 31 29.0 & + 48 59 28 & [email protected] & @[email protected] & f9.5v & 935 & y & y & & & & y + hd 21619 & 03 31 30.2 & + 49 54 07 & [email protected] & @[email protected] & a6v & 931 & y & y & & & y & + hd 21641 & 03 31 33.1 & + 47 51 45 & [email protected] & @[email protected] & b8.5v & 955 & y & y & & & y & + bd+49 957 & 03 31 44.2 & + 49 32 04 & [email protected] & @[email protected] & f3v & 944 & y & y & & & & y + ap 189 & 03 31 44.9 & + 49 33 04 & [email protected] & @[email protected] & k3 & & y?y & & & & & y + hd 21672 & 03 31 53.9 & + 48 44 06 & [email protected] & @[email protected] & b8v & 965 & y & y & & & & y + tyc 3316 956 1 & 03 31 54.2 & + 48 31 38 & [email protected] & @[email protected] & f8v & 968 & y & y & y & y & & y + bd+48 944 & 03 31 55.8 & + 48 35 02 & [email protected] & @[email protected] & a4v & 970 & y & y & & & & + bd+49 958 & 03 31 59.5 & + 49 52 10 & [email protected] & @[email protected] & f1v & 958 & n ? & y & & & y & + v396 per & 03 32 08.6 & + 48 01 24 & [email protected] & @[email protected] & b8iii & 985 & y & y & & & y & + v689 per & 03 32 10.2 & + 49 08 29 & [email protected] & @[email protected] & g5v & & y & & & & & y + ap 196 & 03 32 19.3 & + 47 04 27 & [email protected] & @[email protected] & k3 & & y & & & y & & + v537 per & 03 32 30.7 & + 49 10 35 & [email protected] & @[email protected] & g8 & & y & & & & & y + hd 232804 & 03 32 31.9 & + 51 29 22 & [email protected] & @[email protected] & f5 & 972 & y & y & & & & + hd 21855 & 03 33 22.2 & + 47 25 19 & [email protected] & @[email protected] & a0 & 1056 & ? & y & & & y & + bd+49 967 & 03 33 54.4 & + 50 17 48 & [email protected] & @[email protected] & a6me & 1050 & y ? & y & & & y & + bd+50 784 & 03 33 58.9 & + 50 52 56 & [email protected] & @[email protected] & f6 & 1045 & y & y & & & & + hd 21931 & 03 34 12.9 & + 48 37 03 & [email protected] & @[email protected] & b9v & 1082 & y & y & & & & + bd+48 950 & 03 34 21.6 & + 48 39 36 & [email protected] & @[email protected] & a2 & 1084 & y & y & & & & + tyc 3325 239 1 & 03 35 05.0 & + 50 54 45 & [email protected] & @[email protected] & g0 : & 1086 & yy & y & y & y & & + tyc 3321 1655 1 & 03 35 08.7 & + 49 44 39 & [email protected] & @[email protected] & g4 & 1101 & y & y & & y & & + hd 22136 & 03 35 58.5 & + 47 05 28 & [email protected] & @[email protected] & b8v & 1153 & y & y & & & y & + v540 per & 03 36 22.0 & + 49 09 21 & [email protected] & @[email protected] & k0 & 1143 & y & & & & & + @xmath6 per & 03 36 29.4 & + 48 11 33 & [email protected] & @[email protected] & b5ve & 1164 & n & y & & & y & + tyc 3317 1064 1 & 03 36 31.8 & + 48 39 17 & [email protected] & @[email protected] & f7 & 1160 & y & y & & & & + tyc 3321 187 1 & 03 36 55.1 & + 48 49 43 & [email protected] & @[email protected] & f7 & 1180 & y & y & & y & & + tyc 3317 16 1 & 03 36 57.7 & + 48 44 46 & [email protected] & @[email protected] & f7 & 1185 & yy & y & & y & & + ap 229 & 03 37 27.5 & + 47 33 44 & [email protected] & @[email protected] & k8 & & y & & & & & + hd 22401 & 03 38 15.6 & + 47 34 37 & [email protected] & @[email protected] & a0v & 1259 & y & y & & & y & + hd 22440 & 03 38 35.1 & + 48 35 37 & [email protected] & @[email protected] & a2 & 1260 & y & y & & & y & + tyc 3313 1551 1 & 03 38 51.0 & + 46 36 12 & [email protected] & @[email protected] & g1 & & & y & & & & + ap 248 & 03 41 04.8 & + 49 09 32 & [email protected] & @[email protected] & m3 & & y & & & & & + hd 232823 & 03 41 40.9 & + 51 16 35 & [email protected] & @[email protected] & f2 & 1349 & y ? & y & & & & + note @xmath8 source names are from simbad or vizier . ra and dec are from 2mass . most listed proper motions are from the tycho-2 or ucac3 catalog . he = heckmann et al 1956 ; pr = prosser 1992 , if * is present , then also prosser 1994 ; ma = makarov 2006 ; st = stauffer et al 1993 ; me = mermilliod et al 2008 ; rob = robinchon et al 1999 ; ran = randich et al 1996 . @xmath11 = cluster member from prosser et al 1998 . in the cited papers , y = yes , member ; y ? = probably member ; ? membership uncertain ; n ? = probably nonmember ; n = nonmember . these evaluations are taken directly from the cited papers and have not been reevaluated by the authors of the current study . + star name & ra & dec & pmra & pmdec & spectral & he & pr ? & ma ? & st ? & me ? & rob ? + & ( j2000 ) & ( j2000 ) & ( mas / yr ) & ( mas / yr ) & type & & & & & & & + + hd 17744 & 02 52 39.9 & + 48 48 59 & [email protected] & @[email protected] & a0 & & & y & & & & + hd 18280 & 02 57 54.1 & + 48 52 43 & [email protected] & @[email protected] & a2 & & & y & & & & + bd+50 703 & 03 07 04.1 & + 51 17 40 & [email protected] & @[email protected] & f2v & 7 & ? & & & & & + tyc 3318 1509 1 & 03 08 14.5 & + 49 50 21 & [email protected] & @[email protected] & k0 & & & y & & & & + bd+47 772 & 03 08 23.7 & + 47 36 16 & [email protected] & @[email protected] & f7 & 13 & & & & & & + hd 19458 & 03 09 24.2 & + 45 22 58 & [email protected] & @[email protected] & a0 & & & y & & & & + hip 14697 & 03 09 51.5 & + 48 28 18 & [email protected] & @[email protected] & g3v & 56 & yn & y & n & & y & + v572 per & 03 15 48.7 & + 50 57 21 & [email protected] & @[email protected] & a0 & 295 & & y & & & & + hd 20135 & 03 16 01.5 & + 48 01 34 & [email protected] & @[email protected] & a0p & 285 & y & & & & & + tyc 3319 915 1 & 03 17 24.0 & + 49 21 14 & [email protected] & @[email protected] & g0 & 340 & y?y & & y & n & & + tyc 3319 219 1 & 03 18 43.7 & + 50 23 11 & [email protected] & @[email protected] & f6iv - v & 387 & y ? & & & & & + tyc 3703 665 1 & 03 20 39.6 & + 52 43 35 & [email protected] & @[email protected] & g3 & & & y & & & & + hd 232778 & 03 21 00.1 & + 50 59 35 & [email protected] & @[email protected] & f8 & 463 & & & & & & + hd 20701 & 03 22 03.5 & + 47 56 05 & [email protected] & @[email protected] & a1v & 507 & n & y & & & & + hd 21152 & 03 26 50.2 & + 47 54 58 & [email protected] & @[email protected] & b9v & 729 & n ? & & & & & + hd 232793 & 03 27 14.1 & + 50 52 44 & [email protected] & @[email protected] & f5v & 732 & ? & y & & & & + tyc 3316 677 1 & 03 27 36.7 & + 47 23 18 & [email protected] & @[email protected] & g1 & 761 & & & & & & + ap 68 & 03 28 13.6 & + 49 13 13 & [email protected] & @[email protected] & k3 & & & & & & & + ap 73 & 03 28 24.4 & + 48 53 46 & [email protected] & @[email protected] & g3 & & & & & & & + tyc 3320 778 1 & 03 29 03.7 & + 50 21 12 & [email protected] & @[email protected] & g5 & 818 & & & & & & + hd 21490 & 03 29 32.1 & + 43 13 50 & [email protected] & @[email protected] & f2 & & & y & & & & + ap 85 & 03 30 22.0 & + 48 57 18 & [email protected] & @[email protected] & k5 & & & & & & & + melotte 20 1102 & 03 34 42.8 & + 47 53 14 & [email protected] & @[email protected] & g1v & 1102 & n ? & y & & & & ? + bd+46 780 & 03 37 17.6 & + 47 20 54 & [email protected] & @[email protected] & f3iv & 1218 & y ? & & & & & + bd+51 756 & 03 39 02.9 & + 51 36 37 & [email protected] & @[email protected] & g2 & 1234 & yy & y & & ? & & + tyc 3313 2091 1 & 03 41 05.7 & + 45 47 38 & [email protected] & @[email protected] & g1 & & & y & & & & + hd 23219 & 03 45 27.5 & + 47 39 37 & [email protected] & @[email protected] & b9v & & & y & & & & + hd 23287 & 03 45 53.2 & + 45 36 00 & [email protected] & @[email protected] & a0 & & & y & & & & + hd 23255 & 03 45 54.7 & + 50 25 09 & [email protected] & @[email protected] & a5 & & & y & & & & + hd 23690 & 03 49 01.0 & + 46 48 10 & [email protected] & @[email protected] & a0 & & & y & & & & + hd 24260 & 03 53 39.5 & + 45 45 29 & [email protected] & @[email protected] & a0 & & & y & & & & + hd 24980 & 04 00 20.5 & + 47 34 19 & [email protected] & @[email protected] & a2 & & & y & & & & + hd 25109 & 04 01 50.0 & + 50 38 17 & [email protected] & @[email protected] & a0 & & & y & & & & + note @xmath8 source names are from simbad or vizier . ra and dec are from 2mass . proper motions are from the tycho-2 or ucac3 catalog . he = heckmann et al 1956 ; pr = prosser 1992 , if is present , then also prosser 1994 ; ma = makarov 2006 ; st = stauffer et al 1993 ; me = mermilliod et al 2008 ; rob = robinchon et al 1999 ; ran = randich et al 1996 . in the cited papers , y = yes , member ; y ? = probably member ; ? membership uncertain ; n ? = probably nonmember ; n = nonmember . these evaluations are taken directly from the cited papers and have not been reevaluated by the authors of the current study . @lccccccccccccccccc star & spectral & & w1 & ( mjy ) & & & w2 & ( mjy ) & & & w3 & ( mjy ) & & & w4 & ( mjy ) & + & type & photo & meas & err & ex & photo & meas & err & ex & photo & meas & err & ex & photo & meas & err & ex + hd 19624 & b5 & 570 & 578 & 19.7 & & 318 & 317 & 6.1 & & 54.1 & 55.8 & 0.88 & & 15.1 & 45.8 & 1.5 & 30.7 + hd 19893 & b9 & 466 & 480 & 14.6 & & 261 & 262 & 4.6 & & 44.4 & 56.1 & 0.83 & 11.7 & 12.5 & 35.4 & 1.6 & 22.9 + v459 per & f0 & 144 & 147 & 3.3 & & 81 & 80 & 1.5 & & 13.8 & 15.2 & 0.27 & 1.4 & 3.9 & 9.0 & 0.87 & 5.1 + hd 21091 & b9.5 & 339 & 335 & 8.7 & & 190 & 182 & 3.5 & & 32.2 & 34.3 & 0.54 & 2.1 ? & 9.1 & 18.9 & 0.94 & 9.8 + hd 21117 & b8 & 348 & 353 & 9.1 & & 195 & 191 & 3.5 & & 33.3 & 37.2 & 0.55 & 3.9 & 9.4 & 17.8 & 1.0 & 8.4 + hd 21122 & a0 & 252 & 261 & 6.0 & & 142 & 142 & 2.9 & & 24.1 & 25.8 & 0.4 & 1.7 ? & 6.8 & 13.7 & 1.0 & 6.9 + v 488 per & k0 & 21.9 & 51.0 & 1.1 & 29.1 & 9.3 & 55.4 & 1.1 & 46.1 & 2.1 & 40.9 & 0.6 & 38.8 & 0.57 & 75.7 & 2.2 & 75.1 + hd 21375 & a1 & 441 & 450 & 14.1 & & 247 & 247 & 4.6 & & 42.1 & 41.3 & 0.6 & & 11.9 & 19.2 & 1.1 & 7.3 + hd 21480 & a7 & 234 & 239 & 5.5 & & 132 & 132 & 2.6 & & 22.4 & 24.1 & 0.4 & 1.7 ? & 6.4 & 14.7 & 0.96 & 8.3 + hd 21641 & b8.5 & 609 & 609 & 21.3 & & 340 & 343 & 6.6 & & 57.7 & 70.7 & 1.0 & 13.0 & 16.2 & 39.9 & 1.3 & 23.7 + bd+50 784 & f6 & 111 & 116 & 2.5 & & 60.4 & 62.2 & 1.2 & & 10.8 & 12.9 & 0.24 & 2.1 & 3.1 & 8.54 & 0.9 & 5.4 + note @xmath8 in addition to the above 11 stars , also hd 21855 may have weak excess emission in the w4 band . details may be found in section 3.1 . see also discussion of hd 21641 , v 488 per , and hd 21375 in section 3.1 . the four wise wavebands w1 , w2 , w3 and w4 are 3.35 , 4.60 , 11.56 and 22.09 @xmath4 m , respectively . @lccc cluster/ & age & 24 @xmath4 m excess & reference + association & ( myr ) & ( # ) ( % ) & + @xmath12 cha & 6 & 9/16 56 & rebull et al ( 2008 ) + tw hya assoc . & 8 & 7/23 30 & rebull et al ( 2008 ) + ucl / lcc & 10 & 10/35 34 & rebull et al . ( 2008 ) + @xmath13 pic mg & 12 & 7/30 23 & rebull et al ( 2008 ) + ngc 2547 & 30 & 16/38 42 & gaspar et al ( 2009 ) + tuc / hor / columba & 30 & 27/62 43.5 & zuckerman et al ( 2011 ) + ic 2391 & 40 & 6/26 23 & rebull et al ( 2008 ) & gaspar et al ( 2009 ) + argus assoc . & 40 & 5/8 62.5 & zuckerman et al ( 2011 ) + ngc 2451 & 60 & 12/38 32 & balog et al ( 2009 ) + @xmath0 persei & 60 & 8/56@xmath11 14@xmath11 & this paper + ab dor mg & 70 & 12/47 25.5 & zuckerman et al ( 2011 ) + pleiades & 100 & 10/73 14 & gorlova et al ( 2006 ) + m47 & 100 & 8/63 13 & rebull et al ( 2008 ) + hyades & 650 & 2/78 2.5 & gaspar et al ( 2009 ) + praesepe & 750 & 1/135 0.7 & gaspar et al ( 2009 ) + note @xmath8 @xmath11for @xmath0 persei , these statistics pertain to stars in table 1 of spectral types b and a and at a wavelength of 22 rather than 24 @xmath4 m . should the excess ir emission at hd 21641 be due to an ionized outflowing wind rather than a dusty debris disk ( see section 3.1 ) , then the entries for @xmath0 per in the third and fourth columns would read 7/56 and 12.5% , respectively . if the 12 a - type stars and b - star hd 23219 in table 2 were to be added to the table 1 stars , then entries in the third and fourth columns would read 8/69 and 11.6% , respectively . three b - type stars from table 1 with mid - ir excess emission ( @xmath6 per , 29 per , and hr 1037 , see section 3 ) are excluded from all entries in table 4 and in this caption . spectral energy distributions ( sed ) for five @xmath0 persei cluster members in table 3 with excess infrared emission and , in addition , the sed of 29 per . near infrared jhk data points are from the 2mass catalog . the four diamonds are from wise . circles between 12 and 100 @xmath4 m for 29 per are from iras . the excess ir emission toward 29 per is due to `` cirrus '' and not to an orbiting dusty debris disk ( see section 3.1 ) . long wavelength blackbodies have been added to the various panels for illustrative purposes . in all cases the blackbody temperature is 120 k , except for hd 19624 ( 80 k ) and 29 per ( 60 k).,width=529 ] same as for figure 1 . the square data points for v488 per between 3.5 and 8 @xmath4 m are from the irac camera on spitzer . the blackbodies indicated on the v488 per panel have temperatures of 820 and 120 k. the excess ir emission toward hd 21641 may not be due to a dusty debris disk ( see section 3.1 ) . long wavelength blackbodies have been added to the various panels for illustrative purposes . in all cases the blackbody temperature is 120 k.,width=529 ] wise 22 @xmath4 m image of the area surrounding and including 29 per ( center star in figure ) . north is up and east is left . the other sources are unresolved and give an idea of the point spread function . the image is presented with a logarithmic stretch . 29 per is surrounded by an extended nebulosity . the scale bar is 1 arcminute.,width=529 ] seeing - limited image of v488 per obtained with the lp filter and nirc2 camera on the keck telescope on 31 march 2012 ( ut ) . the pixel scale is 0.01 `` . north is up , east to the left . the displayed image is a combination of three 20 second exposures ( 0.2 sec times 100 coadds per exposure ) . the images were obtained at an elevation of @xmath11.6 air masses just before the telescope hit a hard limit , so there was not time to well - center v488 per . as a result , different noise levels are evident in various regions in the combined image which uses an inverted linear display . since the infrared excess emission at v488 per at this wavelength is about twice as bright as the photosphere , if the excess were coming from a somewhat offset background object then this image would look like a double star . the scalebar on the image indicates one arcsec ; the wise point spread function is about 6 '' fwhp at this wavelength . data courtesy of c. beichman , c. gelino , g. mace , j. aycock , and r. campbell.,width=529 ] comparison of color magnitude diagrams for the @xmath0 persei and pleiades clusters ( from b. kamai et al 2012 , in preparation ) . the figure has not been cleaned of multiple star systems . the data points for the pleiades are from stauffer et al ( 2007 ) and kamai et al ( 2012 , in preparation ) and for @xmath0 per from prosser ( 1992 and 1994 ) and stauffer et al ( 1985 and 1989 ) . the relative placement of the late - type stars shows that @xmath0 per is younger than the pleiades.,width=529 ]	because of proximity to the galactic plane , reliable identification of members of the @xmath0 persei cluster is often problematic . based primarily on membership evaluations contained in six published papers , we constructed a mostly complete list of high - fidelity members of spectral type g and earlier that lie within 3 arc degrees of the cluster center . @xmath0 persei was the one nearby , rich , young open cluster not surveyed with the spitzer space telescope . we examined the first and final data releases of the wide field infrared survey explorer ( wise ) and found 11 , or perhaps 12 , @xmath0 per cluster members that have excess mid - infrared emission above the stellar photosphere attributable to an orbiting dusty debris disk . the most unusual of these is v488 per , a k - type star with an excess ir luminosity 16% ( or more ) of the stellar luminosity ; this is a larger excess fraction than that of any other known dusty main sequence star . much of the dust that orbits v488 per is at a temperature of @xmath1800 k ; if these grains radiate like blackbodies , then they lie only @xmath10.06 au from the star . the dust is probably the aftermath of a collision of two planetary embryos or planets with small semimajor axes ; such orbital radii are similar to those of many of the transiting planets discovered by the kepler satellite .
in this article the results of earlier work on the transition from potential flow to turbulence at velocity amplitudes above the critical velocity @xmath1 ( where @xmath8 = 0.0997 mm@xmath9/s is the circulation quantum and @xmath3 is the oscillation frequency ) observed with a sphere oscillating in superfluid @xmath0he are reviewed and extended . our sphere has a radius of 0.1 mm and it is oscillating in a cell of cylindrical shape of 2 mm diameter and 1 mm height . the surface roughness of the sphere is estimated to be of the order of few micrometer and , therefore , the sphere is rough on the scale of the vortex core of 0.1 nm . at temperatures above 0.5 k a hysteretic behavior in the velocity amplitude vs. driving force amplitude was first reported in 1995 @xcite . more recent data are shown in fig . 1 @xcite . what one generally observes is a linear increase of the velocity amplitude with increasing driving force . the linear drag force is due to thermally excited quasiparticles , i.e. , ballistic phonons at low temperatures and normal phase drag ( stokes ) in the hydrodynamic regime above 1 k. at larger drive a strong nonlinear drag is observed when turbulence ( vortices ) are created . there is a hysteresis : only when reducing the drive in the turbulent regime the critical velocity @xmath4 can be determined , because in the initial up - sweep potential flow breaks down only at velocities @xmath10 larger than @xmath4 . m sphere oscillating at 236 hz in superfluid @xmath0he at 1.90 k. red circles : data taken with increasing drive ; blue x : decreasing drive . the initial breakdown of the potential flow occurs at @xmath10 = 82 mm / s . because of the strong hysteresis @xmath4 = 46 mm / s can only be determined with the down sweep.,scaledwidth=60.0% ] of the oscillating sphere ( frequency 114 hz ) as a function of the driving force @xmath11 at @xmath12 . at small drives we have potential flow where the linear drag is determined by ballistic phonon scattering . the shaded area above the critical velocity @xmath13 20 mm / s indicates the unstable regime where the flow switches intermittently between potential flow and turbulence ( as shown in fig . 3 ) . at larger driving forces we observe apparently stable nonlinear turbulent drag where @xmath14 ( when the small laminar drag is subtracted ) . ] in a set of careful experiments with a vibrating wire the osaka group @xcite showed in 2007 that no transition to turbulence occurred up to very large velocity amplitudes of ca . 1 m / s ( corresponding to oscillation amplitudes of 100 @xmath15 m ) if the superfluid was prepared in a state without any remanent vortices . obviously , some initial vortices must exist in the fluid within the range of the oscillation amplitude for the vibrating object to produce turbulence . the hysteresis which can be observed almost up to the lambda transition changes into an instability of the flow pattern below ca . 0.5 k where intermittent switching between potential flow and turbulent flow around an oscillating sphere was observed at velocity amplitudes slightly above @xmath4 , see fig . 2 and fig . 3 ( experimental details are described in @xcite . ) corresponds to turbulent drag while the increase of the velocity amplitude occurs during potential flow , occasionally reaching the maximum value @xmath16 given by ballistic phonon drag.,scaledwidth=80.0% ] during a turbulent phase the drag is given by the large turbulent drag force and , hence , the velocity amplitude is low . ( the statistical properties of the turbulent phases were analyzed recently in some detail , see @xcite ) . when turbulence breaks down the velocity amplitude begins to grow due to the much smaller phonon drag . because the velocity amplitudes are above @xmath4 these phases of potential flow break down at some velocity @xmath5 and a new turbulent phase is observed . the distribution of the maximum of the excess velocities @xmath7 above the turbulent level is analyzed . this is done by evaluating the probability @xmath17 that a given @xmath18 is exceeded ( reliability function " ) , see fig . 4a . from a fit to the data we find that @xmath19 is a rayleigh distribution and the fitting parameter @xmath20 is independent of temperature and driving force , see fig . 4b . it is a property of the rayleigh distribution that @xmath20 is the rms value of the distribution . is a rayleigh distribution . b ) the fitting parameter @xmath20 is independent of temperature and driving force ( phonon drag subtracted).,scaledwidth=80.0% ] from the rayleigh distribution we obtain the failure rate @xmath21 which is the conditional probability that the potential flow decays in a small interval just above @xmath18 , provided it has not decayed until @xmath18 : @xmath22 we note that the failure rate is proportional to @xmath18 and , hence , to the increase of the oscillation amplitude @xmath23 . this result will be of significance when we consider the effect of the remanent vortex density @xmath6 on the critical velocity , see section 4 . another general property of this distribution can be found when changing the variable from @xmath18 to the lifetime @xmath24 of the potential flow : @xmath25 where in our case @xmath26 is due to the relaxation to the maximum amplitude @xmath16 given by ballistic phonon scattering ( see upper trace in fig . 3 ) : @xmath27 where @xmath28 = @xmath29 and @xmath30 = 2m/@xmath31 is the time constant , @xmath31 is the friction coefficient of phonon scattering and m is the mass of the sphere ( 27 @xmath15 g ) . initially we have a linear increase of @xmath26 and therefore @xmath32 . at a time @xmath24 = @xmath33 the failure rate has a maximum . and , if the maximum level @xmath28 is approached for large @xmath24 , we find @xmath34 . thus the flow becomes stable although the velocity @xmath16 is clearly above @xmath4 . nevertheless , these metastable states of potential flow ultimately will also break down , in our experiments after a mean lifetime of 25 min , because natural radioactivity occasionally produces vorticity in the fluid , see @xcite . below 100 mk the phonon drag is negligibly small , only the residual damping of the empty cell @xmath35 = 4.4 @xmath36 kg / s @xcite determines the oscillation amplitude . equivalent is a time constant @xmath30 = 2m/@xmath35 = 1200 s. therefore , the exponential function in ( 3 ) can be expanded : during potential flow is described by eq.4 , see text . the lifetimes of the turbulent phases can not be resolved ( not even on an expanded time scale).,scaledwidth=80.0% ] + @xmath37 in fig . 5 the time series recorded at 27 mk shows a stochastic saw tooth pattern . the slope of the linear increase of the velocity amplitude during potential flow amounts to 6.9 @xmath38 m / s@xmath9 . inserting in ( 4 ) the relevant numbers for @xmath11 = 4.6 pn , @xmath4 = 19 mm / s , @xmath39 = 27 @xmath15 g , and @xmath35 = 4.4 @xmath36 kg / s , we find a slope of 7.0 @xmath38 m / s@xmath9 , in agreement with the experimental value . from ( 4 ) follows that the lifetimes of the phases of potential flow now also have a rayleigh distribution . from the measured rms value @xmath20 = 6.5 mm / s of the @xmath18 distribution we obtain the rms lifetime @xmath40 = 93 s. the second term in the brackets of ( 4 ) which is due to the residual damping , is here 18% of the first one and decreases further at larger drives . ultimately , only the first term will dominate , i.e. , we may neglect damping and approach the case of an undamped oscillator : @xmath41 the rms lifetime @xmath42 of the potential flow is then proportional to @xmath43 : @xmath44 and the failure rate ( 2 ) would simply be given by @xmath45 for six different frequencies . each data point is obtained from the average of several individual time series recorded at different temperatures and driving forces . the error bars represent the standard deviations . a power law fit that takes into account the error bars indicates a slope of @xmath46.,scaledwidth=80.0% ] we observe a weak frequency dependence of @xmath20 , see fig . a power law fit to the data indicates a slope of @xmath47 . this may be interpreted as an @xmath48 law . however , our frequency range is rather limited ( @xmath49 factor 2 ) and there is some scatter . therefore , this dependence is not firmly established . on the other hand , the data point at 212 hz is the mean of 14 individual @xmath20 data measured at different temperatures ( 28 mk and 300 mk ) and different driving forces and , therefore , can not simply be neglected . a theoretical interpretation of this result is not available as long as we do not have a theory of the transition to turbulence in oscillating superflows . however , a strong support of this frequency dependence of @xmath20 comes from the failure rate ( 7 ) . in that case it follows that @xmath50 , i.e. , the failure rate is proportional to the number of the completed cycles . this is what one would naturally assume . the analysis of our data allows us to obtain informations concerning the properties of the instable phases of potential flow above the critical velocity . the failure rate ( 1 ) is proportional to the increase of the oscillation amplitude beyond @xmath4 , and it is independent of temperature and driving force . the final breakdown of potential flow at @xmath5 is known to be affected by the remanent vorticity and occurs statistically during the switching process at mk temperatures . moreover , the stability of the phases of potential flow when @xmath16 is reached ( see upper trace in fig . 3 ) leads us to assume that at breakdown the critical oscillation amplitude is determined by the intervortex spacing @xmath51 of the remanent vorticity , which in this case can not be reached by the sphere : @xmath52 40 @xmath15 m that is less than half of the radius of the sphere . introducing also at @xmath4 the intervortex spacing @xmath53 we assume , without having a rigorous theory , that the critical oscillation amplitudes are proportional to the intervortex spacings : from the hysteresis shown in fig . 1 we find from the velocities @xmath10 = 82 mm / s and @xmath4 = 46 mm / s a ratio of 0.56 for the spacings and hence a ratio @xmath55 = 0.31 . defining @xmath56 we obtain from ( 8) the relation inserting @xmath20 = 6.5 mm / s and @xmath4 = 19 mm / s ( see figs . 3 and 5 ) we get @xmath61 = 0.34 and thus an average @xmath62 = 1.34 @xmath63 or an average @xmath64 = 0.56 . this means that on the average 56 % of the vortex density at @xmath4 have survived the breakdown of the turbulent phase . but as long as @xmath65 is unknown for oscillatory flow we can not determine absolute values of the vortex densities . also , since our sphere radius may be larger than the intervortex distance it is rather peculiar that the transition to turbulence is not triggered by the vortices likely to be attached to the sphere . however , it is possible that such vortices are absent , and in the laminar state the surface of the sphere becomes free from vorticity or is only covered by very tiny loops which require a large velocity in order to expand . the transition to turbulence can then occur at oscillation amplitudes that are smaller than the radius of the sphere , see above . this would happen when the sphere becomes again in contact with the tangle that still remains in the vicinity of the sphere . the correct physical picture requires experiments that could visualize the vortices around the sphere or more realistic computer simulations . hysteresis and switching of the flow have been observed also with vibrating tuning forks and wires @xcite and the significance of remanent vorticity for the critical velocity @xmath10 at breakdown was demonstrated @xcite . in our work we are relating @xmath10 directly to the intervortex spacing of the remanent vortex density . discussing the distribution of the lifetimes of the potential flow we note that because of the nonlinear dependence of @xmath66 in ( 3 ) the distribution is very complicated . however , the situation becomes much simpler at low temperatures , see section 2.2 , where @xmath18 is proportional to the lifetime , see ( 4 ) . the lifetimes are now also following a rayleigh distribution and the rms lifetime decreases with increasing driving force . this may be compared with the recent experimental result of the lancaster group on intermittent switching of the flow of around a vibrating tuning fork at mk temperatures @xcite . these authors found that the average lifetimes of potential flow decrease towards larger flow velocities , in qualitative agreement with our rms value @xmath42 ( 6 ) . we are grateful to jan jger and hubert kerscher for their co - operation . w.s . acknowledges discussions with matti krusius ( aalto university , finland ) and shaun fisher ( lancaster university , uk ) . r.h is supported by the academy of finland .	the onset of turbulent flow around an oscillating sphere in superfluid @xmath0he is known to occur at a critical velocity @xmath1 where @xmath2 is the circulation quantum and @xmath3 is the oscillation frequency . but it is also well known that initially in a first up - sweep of the oscillation amplitude , @xmath4 can be considerably exceeded before the transition occurs , thus leading to a strong hysteresis in the velocity sweeps . the velocity amplitude @xmath5 where the transition finally occurs is related to the density @xmath6 of the remanent vortices in the superfluid . moreover , at temperatures below ca . 0.5 k and in a small interval of velocity amplitudes between @xmath4 and a velocity that is about 2% larger , the flow pattern is found to be unstable , switching intermittently between potential flow and turbulence . from time series recorded at constant temperature and driving force the distribution of the excess velocities @xmath7 is obtained and from that the failure rate . below 0.1 k we also can determine the distribution of the lifetimes of the phases of potential flow . finally , the frequency dependence of these results is discussed .
since the first systematizations of their chemistry and basic properties by , e.g. , kahn in 1993 @xcite , molecular magnets are now developing into a promising class of magnetic nanomaterials with a great potential in applications in areas such as magnetic storage , quantum computing , or magneto - optical devices . an up - to - date review on microscopic properties ( spin density , exchange interactions , magnetic anisotropy ) of molecular magnets accessible in experiment and from first - principles calculations , is going to appear in ref . its major part is summarized as an internet publication @xcite . in the present work we address the issue of extracting the interatomic exchange interaction parameters entering the heisenberg model form the density functional theory ( dft ) . we discuss two chemical systems of different complexity . the actual calculations are done by two different methods , siesta @xcite using compact and strictly confined atom - centered basis functions ( see ref . @xcite for details ) , and the fleur @xcite code , a realization of the highly accurate full - potential linearized augmented plane wave ( flapw ) method . both calculations used the generalized gradient approximation to the exchange - correlation functionals . we skip the subtleties in the formulation of exchange parameters within the dft ; a detailed discussion can be found in ref . basically , we refer to total energies in different specially prepared magnetic configurations , calculated from first principles . [ sec : fe3 ] covers the extraction of meaningful results for a fe - trinuclear system by applying the fixed spin moment formalism . [ sec : fe2 ] addresses a model fe - binuclear system and , specifically , the importance of intraatomic correlation , treated beyond the conventional dft formalism . one of the ways to arrive at the values of interaction parameters in a magnetic system whose properties are well decribed by the heisenberg model is to compare total energies for different magnetic configurations . this is rather straightforward if , say , ferromagnetic ( fm ) and antiferromagnetic ( afm ) states are well defined and at least metastable . a more complicated case is a frustrated system , an example of which is made of three fe atoms incorporated into organic molecule . we consider specifically fe@xmath1(oac)@xmath1l@xmath1 , a synthesized by boskovic _ _ @xcite derivate of the schiff base h@xmath2l = salicylidene-2-ethanolamine . the molecular unit used in the electronic structure calculation by the siesta method is shown in fig . 2 ( a ) of ref . the magnetization measurements reported in ref . @xcite indicate an ( almost afm ) ground state with the total spin @xmath3=1/2 , coming about from the interaction of three @xmath4=5/2 spins of fe(iii ) ions . experimentally , the interaction parameters @xmath5 of the heisenberg model , introduced in ref . @xcite as @xmath6 , were reported to be necessarily different for an acceptable fitting of data , even as the molecule nearly maintains a trifold symmetry axis : @xmath7 = @xmath815.1 k , @xmath813.6 k , and @xmath812.4 k. we proposed earlier @xcite to use the fixed spin moment ( fsm ) method @xcite for selecting a magnetic configuration of a molecular magnet ( a six - center `` ferric wheel '' ) , improving simultaneously the stability of the electronic structure calculation . for the present system this idea becomes even more important , because it is no more obvious how to simulate a ( nearly ) afm trial magnetic configuration in the dft run , nor how to enforce an inversion of a selected local magnetic moment . on the contrary , by running a sequence of fsm calculations from the maximal nominal value of the total spin @xmath3=15/2 ( and higher ) down to zero , one can scan the variety of all possible magnetic solutions within the dft . ( oac)@xmath1l@xmath1 ] . see main text and explanations @xcite for details . ] the results of our fsm calculation are shown in fig . [ fig : fe3_fsm ] . the total energy ( left panel ) has several local minima , the lowest one being for the fsm value of 5 @xmath10 . the right panel shows that this situation corresponds to two fe magnetic moments aligned in parallel and the third one aligned in the opposite direction . a competitive energy minimum is found for fsm = 15 @xmath10 , i.e. , the ferromagnetic configuration . the solution with fsm = 1 @xmath10 has higher energy than the latter one by 74 mev , per fe atom . it should be noted that , whereas the ground state of the system is experimentally found to have @xmath3=1/2 , it must have a mixed nature over many quantum states , and hence does not correspond to any single - determinant dft result . nevertheless , the tendency for antiparallel coupling of fe magnetic moments is correctly provided by our calculation ; moreover the order of magnitude of the interaction constants can be correctly recovered : consistently with the above formula for the heisenberg hamiltonian , @xmath11 amounts to 57 k. the error ( overestimation by a factor of about 4 ) in the dft calculation is nearly the same as we found earlier for the `` ferric wheel '' @xcite . a possible reason for such systematic discrepancy can be the underestimation of the intraatomic coulomb correlation , to be discussed in the next section . two issues might need further discussion . the homo - lumo gaps shown in the middle panel of fig . [ fig : fe3_fsm ] are different in the fsm formalism for two spin directions , because the fixing of the total spin amounts to an imposition of an external magnetic field , moving apart the chemical potentials in two spin channels . normally this would require the consideration of a zeeman term when dealing with total energies . however , for a system with a _ common _ band gap in both spin channels a unique chemical potential can be found for both spin directions , therefore the fsm scheme would simply fix one or another of the metastable magnetic configurations at no additional energy cost . one can see that , indeed , a common band gap can be found in all situations relevant for our discussion , i.e. with fsm = 1 , 5 , and 15 @xmath10 , thus justifying the extraction of the @xmath5 value above . another observation concerns the local magnetic moments at fe sites : as the fsm value is changing , and even as the saturation of magnetization ( at fsm = 15 @xmath10 ) is achieved , the local fe magnetic moments merely flip , essentially maintaining their magnitudes of 4 @xmath10 , and _ not _ 5 @xmath10 . the nominal value of @xmath4=5/2 , associated with a single fe atom , comes about due to magnetic polarization of ligands . this situation is identical to that described earlier for `` ferric wheels '' @xcite . in spite of the very different structure of organic ligands , both systems have similarities in the nearest neighbourhood and in the charge state of fe atoms ( o@xmath12 twisted octahedral coordination in `` ferric wheels '' vs. slightly distorted octahedral o@xmath13n coordination in the present system ) . we turn now to the discussion of the effect of on - site _ intraatomic _ coulomb correlation on _ interatomic _ exchange parameters . our objective was to find a compact and yet realistic system , for which an electronic structure calculation could be done with the use of an ultimately accurate method within the dft , that is , the flapw method , and to apply a phenomenological correction by the lda+@xmath0 formalism @xcite on top of it . specifically , we apply the realization of the method in the fleur code @xcite . we derive a simple model system from the fe - binuclear complex described by real _ @xcite being subject to previous studies @xcite . we neglect for the moment the spin crossover property , which makes this system particularly interesting , and make our calculation for the high - spin configuration on both fe centers . the chemical formula of the substance is [ fe(bt)(ncs)@xmath2]@xmath2-bpym , with bt = 2,@xmath15-bithazoline and bpym = 2,@xmath15-bipyrimidine , and the molecular unit as shown on the left of fig . [ fig : fe2_molec ] . we also performed electronic structure calculation for this substance in its molecular and crystallized form by the siesta method , allowing structure relaxation , the results of this study will be reported elsewhere . for the present simulation , as the molecule is yet too large and too low - symmetric for an affordable flapw calculation , we simplify the molecular units , substituting the bt fragments by bpym that is realistic , as this is encountered in related compounds see , e.g. , @xcite . moreover we cut and connect the ( ncs ) tails into n = c = n chains that makes a dense 3-dimensional lattice out of planar fe bipyrimidine groups ( fig . [ fig : fe2_molec ] , right panel ) . the nearest neighbourhood of each fe center remains a slightly distorted n@xmath12 octahedron , as was the case in the original molecular unit . an idealization is that all n fe n bonds make now right angles , and the repeated bipyrimidine fragments are coplanar . ] @xmath2bpym described in ref . @xcite ( left ) and a simplified model thereof , used in the calculation ( right ) . the fe atom is in the middle of a slightly distorted n@xmath12 octahedron . ] , for parallel orientation of fe magnetic moments . left panel : lda calculation , right panel : lda+@xmath0 calculation with @xmath0=4 ev . ] the calculated densities of states ( per unit cell and per one fe site ) in fig . [ fig : fe2_dos ] clearly indicate a hybridization of fe@xmath16 with the @xmath17 states of neighbouring nitrogen atoms . since the minority - spin fe @xmath16 states are not completely empty , the local magnetic moment is less than 5 @xmath10 . as for the previous trinuclear system , the magnetic moment induced on nitrogen neighbours substantially contributes to the net value , associated to each fe site . this resolves , at least in part , a controversy between the spin value @xmath4=5/2 , expected at the fe site from formal valence considerations , and the local magnetic moment of our calculation . the cumulative magnetic moment , distributed over the fe site and its neighbours , behaves like a `` rigid '' spin in the sense of the heisenberg model . however , one can expect that a stronger localization of this distributed magnetic moment at the fe site will be enforced by an inclusion of intraatomic coulomb correlation , underestimated in conventional dft calculations . in order to check this possible effect , we brought in an additional coulomb correlation applying a semi - empirical `` lda+@xmath0 '' scheme @xcite . we choose the coulomb interaction constant @xmath0 = 4 ev , which seems reasonable for correlated fe compounds @xcite . one sees in the dos of fig . [ fig : fe2_dos ] ( right panel ) the known effect of the `` + @xmath0 '' correction to lower the occupied states in energy whereas shifting unoccupied ones upwards . the values of magnetic moments and exchange parameters @xmath5 , determined by the formula @xmath18 , are listed in the table . it is noteworthy that , whereas the total magnetic moment per fe site is increased in the lda+@xmath0 calculation , the _ local _ fe magnetic moment still does not exceed 4 @xmath10 . on can conclude that the spatial distribution of magnetic density is not confined to the fe atom even as the intraatomic coulomb correlation is enhanced ; instead , the magnetization maintains its complex spatial distribution , spilling onto neighbouring n atoms . .local magnetic moments at the fe site ( in @xmath10 , within muffin - tin sphere of 2.3 bohr ) ; total magnetic moments _ per _ fe center , total energy difference between fm and afm configurations , and corresponding extracted values of the heisenberg exchange parameter @xmath5 , assuming nominal spin value @xmath4=5/2 . [ cols="^,^,^,^,^,^ " , ] as the intraatomic coulomb correlation shifts apart on the energy scale the centers of gravity of occupied and unoccupied states , it reduces the interatomic exchange parameter , because the latter ( in one of its formulations , see @xcite ) has corresponding energy differences in the denominator . the same trend prevails in our case even as we estimate the values of @xmath5 by comparing the total energies of two magnetic configurations . the effect of coulomb correlation , included on top of a straightforward dft calculation , is therefore always towards lowering the absolute values of heisenberg exchange parameters . the same conclusion has been drawn by boukhvalov _ _ in their calculations for `` mn@xmath19 '' @xcite and `` v@xmath20 '' @xcite molecular magnets . comparing with the results of the previous section , one can conclude that the estimation of exchange parameters in molecular magnets , that is already qualitatively correct ( in sign and order of magnitude ) in a conventional dft calculation , can be brought closer to the experimental estimates by reducing their numerical values due to additionally included intraatomic coulomb correlation . the authors thank deutsche forschungsgemeinschaft for financial support ( priority program ` molecular magnetism ' ) . a.v.p . thanks colette boskovic and jos real for providing crystallographic information on substances synthesized by them , and danil boukhvalov for making accessible his manuscript on v@xmath20 prior to publication . j. kortus , a. v. postnikov , molecular nanomagnets , to be published in : m. rieth , w. schommers ( eds . ) , handbook of theoretical and computational nanotechnology , vol . 10 of encyclopedia of nanoscience and nanotechnology , american scientific publishers ( 2004 ) http://aspbs.com/html/a0000tcn.htm a. v. postnikov , j. kortus , m. r. pederson , density functional studies of molecular magnets , scientific highlight of the month february 2004 , newsletter 61 of the @xmath21-network , http://psi-k.dl.ac.uk/newsletters/news_61/highlight_61.pdf . it should be explained that the fe - trinuclear complex reported in ref . @xcite crystallizes with two molecular units per primitive cell , which are slightly different . in particular , there is a formal disproportionality in the nominal electron number between these two fragments . as a consequence , a single molecular unit we took for our calculation possesses an odd number of electrons . therefore only odd fsm values correspond to integer number of electrons in both majority- and minority - spin channels and may lead to an appearance of homo - lumo gaps , shown in the middle panel of fig . [ fig : fe3_fsm ] . however , the even fsm values can be formally applied in the calculation as well , notwithstanding that they do not possess the homo - lumo gap . technically we applied an artificial broadening of molecular orbitals eigenvalues by 400 k. some of the even fsm numbers allow an easily converging calculation , with the total energy and local moments which follow the general trend . such results are added in the left and right panels of fig . [ fig : fe3_fsm ] .	the calculation of interatomic magnetic exchange interactions entering the heisenberg model from the standpoint of the density functional theory ( dft ) is outlined for two fe - based molecular magnets : a trinuclear complex with a schiff base ligand , which makes an antiferromagnetically coupled frustrated system , and a model bipyrimidine - connected planar network of fe ions . first - principles electronic structure calculations are performed using the real - space method siesta and the full - potential linearized augmented plane wave flapw method fleur , correspondingly . we discuss the application of fixed spin moment technique for preparing the system in a given magnetic configuration , and the effect of intraatomic coulomb correlation , approximated by the lda+@xmath0 technique , on the values of interaction parameters . _ ab initio _ calculations , single - molecule magnets , magnetic interactions 31.15.ew , 31.70.ks , 71.15.nc , 71.20.rv , 71.70.gm , 75.50.xx

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.